The ^Aristotelian Ethics

A STU D Y O F T H E R E L A T IO N S H IP
BETW EEN TH E E U D E M IA N AND
N IC O M A C H EA N E T H IC S O F A R IS T O T L E

ANTHONY KEN N Y

CLAREN D O N PRESS · O XFO RD

1978
 Oxford University Press, Walton Street, Oxford 0X2 6dp
OXFORD LONDON GLASGOW NEW YO RK
TORONTO M ELBO URNE W E LLIN G TO N CAPE TOWN
IBAD AN N A IRO BI DAR ES SALAAM LUSAKA PREFACE
KU ALA LU M PU R SIN GAPORE JA KA RTA HONG KO NG TOKYO
D ELHI BO M BA Y CALCUTTA MADRAS KARACHI

© Oxford University Press ig jtt T h is book is an attempt to solve a long-standing problem o f

AH rights reserved. No part oj this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, «r
otherwise, without the prior permission o f Oxford University Press
B r it is h L ib r a r y C a t a lo g in g in P u b lic a tio n D a ta
K enny, Anthony Joh n Patrick
T he Aristotelian ethics,
i. Aristotle— Ethics
I. T itle
170 B 4 9 1.E 7
IS B N 0 -19 -8 2 4 5 5 4 -8
Aristotelian scholarship on the basis o f historical and philosophical
arguments and a statistical study o f features o f style. T o be fully
qualified to undertake such a task a man must be a professional
philosopher, classicist, and statistician. I can claim to be pro­
fessionally qualified only as a philosopher: I am a very amateur
classicist and a complete novice in statistics. M y excuse for being
undeterred by this is the fact that most o f those working in the field
o f literary statistics are also, in one or other respect, novices, or, as
77-30640 they would no doubt prefer to put it, pioneers. When the statistical
study o f literary texts has grown into a systematically organized
discipline, monographs such as the present will no doubt look very
amateurish. But any discipline in its earliest days must depend
upon the contributions o f untaught and inexpert amateurs.
T h e amateur who ventures into a field on the borderline o f
several disciplines must, however, be more than usually conscious
o f the debt he owes to his colleagues in the adjacent fields in which
he is himself unskilled. In writing the present work I have been at
every step dependent on the learning and generosity o f the
philologists, historians, statisticians, and computer experts who have
patiently striven to remove my misunderstandings and fill up the
gaps in my knowledge. In a field so well trodden as that o f
Aristotelian studies it would no doubt be rash to claim even that my
mistakes are original : but it is more than usually true to say that the
errors which remain in the book are in no way the responsibility o f
the kind friends who have taken pains to see that others were
expunged.
I must express my gratitude to my classical colleagues in Balliol
who were often the immediate target for my brain-picking forays ;
in particular D r. Oswyn M urray who gave me much useful
historical information, and M r. Jasper Griffin who read the whole
book in typescript and tactfully deleted a number o f solecisms. I am
Printed in Great Britain by very much indebted to M r. Paul Griffith, statistical officer at the
William Clowes Sons Ltd. Oxford Computing Laboratory, who checked the statistical
London, Beccles and Colchester
argument o f Chapter 4, to M r. Frank Pettit, who first taught me
VI Preface Preface VII

how to use a computer, and to M rs. Susan Hockey who taught me Barnes, Professor G . E . L . Owen, Richard Sorabji, Pamela H uby,
everything that I know about the use o f the computer for literary and M ichael Woods.
studies. I wish also to thank Professor Thedore Brunner and the Finally I must express my gratitude to four scholars whom I have
staff o f the Thesaurus Linguae Graecae at Irvine, and Stephan never met but to whose writings I owe a great deal. Professor J . D .
Gruen who as a graduate student prepared for me a computer- M onan’s M oral Knowledge and its Methodology in Aristotle first
readable text o f the Eudemian Ethics, on which I carried out my made me aware o f the flimsiness o f the arguments on which the
preliminary studies before the tapes prepared by the Thesaurus accepted chronology o f Aristotle’s ethical writings rested. T h e
Linguae Graecae became available. monumental commentaries o f Gauthier and Jo lif and o f Dirlm eier
D rafts o f chapters o f the book were read as papers to the Bs club are classical expressions o f the theories which are the target o f the
in Cam bridge, to the Philological Society in O xford, to the present work : but I am aware that at every step I draw upon their
Am erican Philological Society in W ashington, and to the M oral erudition and industry for the very information on which my
Sciences Club in Cam bridge; and also to societies and seminars at criticism o f the reigning orthodoxy o f Aristotelian scholarship is
K in g ’s College, London, at the U niversity o f Texas at Austin, at based.
Swarthmore College, at the University o f Pennsylvania, at the I am indebted to the Trustees o f the Craven Fund for a travel
U niversity o f M ichigan, and at Bristol University. I am most grant which enabled me to consult manuscripts o f the Ethics and of
grateful to those who took part in the discussions at those places for Aspasius in Florence and in Rome.
many helpful suggestions and criticisms.
T h e Revd. Andrew Q. M orton o f Culross, and D r. Alan Jones o f
the Oriental Institute in Oxford were both extremely helpful and
encouraging to me when first I became interested in the statistical
study o f classical texts. Professor Charles K ahn o f the University o f
Pennsylvania and an anonymous reader for the Archiv für
Geschichte der Philosophie made searching and valuable criticism
o f an early draft o f the central argument o f the book.
D r. C . J . Rowe, whose study o f the relationship between the
Eudemian and Nicomachean Ethics in 19 7 1 awoke my own interest in
the problem, assisted me, in discussion and in correspondence, at
every stage o f my work— bent as I was on the destruction o f the
theory he had so ably defended— with the most admirable
impartiality, generosity, and friendliness. Professor D . J . A llan—
who has for decades stood out, almost alone in the republic o f
classical letters, against the universal acceptance o f the fashions set
by W erner Jaeger, and has given courage to others less learned than
him self to do likewise— has placed me greatly in his debt by his
constant encouragement and willingness to share with me the
unparalleled store o f learning he has acquired in more than twenty
years’ study o f the Eudemian Ethics. T o these and to the other
Aristotelian scholars who have taken a friendly interest in my work
I am most grateful, in particular to Professor J . L . Ackrill, Jonathan
 CONTENTS

L ist o f Tables x

1 . T he Aristotelian Ethics in Antiquity i

2. Cross-references in the Ethics 50

3. T h e Style o f the Treatise on Justice 60

4. Particles and Connectives in the Aristotelian Ethics 70

5. Prepositions, Adverbs, and Pronouns in the Ethical

T reatises 113

6. The U se o f Technical Term s in the Ethical Treatises 137

7. Wisdom in the Aristotelian Ethics 16 1

8. Happiness in the Aristotelian Ethics 190

9. T he Dating o f the Aristotelian Ethical Treatises 2 15

Bibliography 240

Index 247
 List o f Tables xi

5·2 Adverbial Modifiers of Degree in Individual Books of the

N E, AE, and EE 116
5-3 απλώ ς and μόνον in Individual Books of the N E, A E, and
LIST OF TABLES EE 117
5-4 Expressions of Doubt and Certainty in Individual Books of
the NE, AE, and EE I 18
Table page 5-5 Scattergram of Occurrences of Expressions of Doubt and
1.1 Parallels to the Magna Moralia in the N E and EE II Certainty in A E and EE 120
1.2 The Ordering of the Virtues in Peripatetic Ethical Treatises 23 5-6 Scattergram of Occurrences of Expressions of Doubt and
4-1 Occurrences of Thirty-six Particles and Connectives in the Certainty in N E and EE 121
NE, EE, and A E 74 5-7 Pronouns and Pronominal Adjectives in Individual Books of
4-2 Ranking of Particles and Connectives in the N E, EE, and A E the N E, A E, and EE 124
78
4-3 Scattergram of Occurrences of Particles and Connectives in 5-8 Nicomachean Favourites in Seventeen Samples of the A E :
the N E and A E 80 Part One 132
4-4 Scattergram of Occurrences of Particles and Connectives in 5-9 Eudemian Favourites in Seventeen Samples of the A E : Part
the E E and A E 81 One 133
4-5 Frequencies of Particles and Connectives in the N E, EE, and 6. i Logical Terms in the NE, AE, and E E 140
AE 83 6.2 Metaphysical Technical Terms in Individual Books of the
4.6 Agreements and Disagreements between A E , N E, and E E in N E, AE, and EE 142
respect of Particle Use 86 6-3 Scattergram of Occurrences of Metaphysical Technical
4-7 Significance of the Difference between Frequencies of Terms in N E and A E 143
Particles in the A E, N E, and E E 88 6.4 Scattergram of Occurrences of Metaphysical Technical
4.8 Occurrences of el in the EE Terms in E E and A E 144
91
4-9 The Calculation of Chi-squared 92 6-5 Expressions for Volitional States in Individual Books of the
4.10 Particles and Connectives in Individual Books of the N E NE, AE, and EE 146
93
6.6 Evaluative Expressions in Individual Books of the NE, AE,
4-11 Particles and Connectives in Individual Books of the E E 94
4.12 Particles and Connectives in Individual Books of the A E and EE ISO
95 Scattergram of Occurrences of Evaluative Expressions in the
6.7
4 1 3 Comparison of Two Contexts for the A E in respect of EE and A E
Particle Use Ï 51
97
6.8 Scattergram of Occurrences of Evaluative Expressions in the
4 r4 (a-d) Bar Charts of the Use of e-nei and ώ στε in the
Traditional N E and E E N E and A E 152
98
6.9 Nicomachean Favourites in Seventeen Samples of the A E :
4-15 Frequencies o f Particles and Connectives in Individual Part Two
Books of the N E 100 155
4.16 Frequencies o f Particles and Connectives in Individual 6.10 Eudemian Favourites in Seventeen Samples of the A E : Part
Books of the E E 102 Two 1 5&
9-1 Nicomachean and Eudemian Favourite Expressions in
4-17 Frequencies o f Particles and Connectives in Individual
Books of the A E 104 the Two Treatments of Pleasure 234
4.18 Particles: Books as Samples from the Traditional N E as a
Single Population 107
4.19 Particles: Books as Samples from the Traditional E E as a
Single Population 108
4.20 Particles and Connectives in Seventeen Samples of the A E no
5 1 Nineteen Common Prepositions in the N E, AE, and EE. 114
 CH APTER I

The Aristotelian Ethics

in Antiquity

F or many centuries the Nicomachean Ethics have been regarded as

the Ethics o f Aristotle. From the Byzantine period twenty
manuscripts o f the Nicomachean Ethics survive; of the Eudemian
Ethics only two (Harlfinger, 19 7 1, 27). Since the M iddle Ages
commentaries on the Nicomachean Ethics have appeared about once
a decade; the Eudemian Ethics has received only four com­
mentaries in its whole history (M aurus, 1668; Fritzsche, 18 5 1;
Dirlmeier, 1969; von Fragstein, 1974). Nineteenth-century
scholars, following the lead o f Schleiermacher (1835) and Spengel
(18 4 1) almost unanimously regarded the Eudemian Ethics as
spurious : they were so confident o f its inauthenticity that Susemihl
could print on the title-page o f his still-indispensable edition o f the
text iEudemii Rhodii Ethica’ . Early in the present century scholarly
opinion, in the wake o f von der M iihll (1909), K app (19 12 ), and
above all Jaeger (1923) swung in favour o f accepting the work as
authentic; but it has never been accorded anything like parity o f
esteem with the Nicomachean treatise. Twentieth-century scholars
have treated it as a product o f the comparatively young Aristotle,
still under the stiflingly metaphysical influence o f Plato. Once again
the most recent scholarly monograph on the topic (Rowe, 19 7 1)
comes to the conclusion that the Nicomachean Ethics is the
definitive statement o f Aristotle’s ethical system, reorganizing an
earlier excessively Academic draft in the Eudemian. T h e third o f the
traditional Aristotelian ethical treatises, the Magna M oralia, was
rejected as post-Aristotelian by the school o f Jaeger (Walzer, 1929 ;
Brink, 1933). T h is rejection has been hotly contested (von Arnim,
1924, 1927, 19 29 ; Gohlke, 1944; Dirlm eier, 19 58; Düring, 1966;
Cooper, 1973). Despite these defences it is still probably the case
that most scholars continue to reject the Magna M oralia : but while
Jaeger’s position here has been under attack, his judgement on the
2 The Aristotelian Ethics
Eudemian Ethics has been so widely accepted and so little
questioned (exeptions are Allan, 1958 and 1966; M onan, 1968) as to
have become a dogma o f Aristotelian scholarship. It is the purpose
o f the present work to attack, and, I hope, to demolish this dogma.
T h e arguments used to relegate the Eudemian Ethics to an
inferior and provisional position in comparison with the
Nicomachean have been predominantly based on internal evidence.
Scholars have disagreed whether the Eudemian Ethics was the work
o f the immature Aristotle or o f a posthumous admirer ; but they have
agreed that it was in various ways unworthy o f the master in his
years o f greatness : it was too unworldly, too pious, too formalistic,
too incoherent, too chaotic, too systematic. Evidence has been
collected in the Eudem ian books for a Platonic, purely theoretical,
sense o f the word phronësis, for a Platonic view o f ethical method, and
a Platonic ideal o f the philosophical life. Scholars have argued that
in style and content the Eudemian Ethics is closer than the
Nicomachean Ethics to the currently accepted reconstructions o f the
lost text o f Aristotle’s juvenile Protrepticus. Similarities o f thought
and vocabulary have been traced between the Eudemian Ethics and
the Plato o f the Philebus and the Laws.
T h e especial esteem o f tradition for the Nicomachean Ethics is
self-fortifying in a way which makes it difficult to examine the
internal evidence with an impartial eye. Stock put the problem in a
vivid if provincial light in his introduction to the Oxford translation
o f the M agna M oralia and o f the Eudemian Ethics (1925, iii).
To an Oxford man . . . who has been nurtured on the Nicomachean Ethics,
and to whom the treatise has become, mentally speaking, ‘bone of his bone
and flesh of his flesh’ it seems too self-evident to require discussion that the
Nicomachean Ethics is the substance of which the others are the shadow.
But this confidence may be born of prejudice, and it is possible that, if the
same person had had the Eudemian Ethics equally carefully instilled into
him in youth, he might on making acquaintance with the Nicomachean find
nothing more in that than a less literary rearrangement of the Eudemians.
M ore seriously, the impoverished manuscript tradition o f the
Eudemian Ethics means that when reading important parts o f it we
must struggle with a corrupt text which either does not make sense
or does so only by benefit o f editors’ conjectures. These
conjectures in their turn cannot help being influenced by the
editors’ over-all views o f the nature o f the treatise and its
relationship to Aristotle’s other works.
The Aristotelian Ethics in Antiquity 3
Despite this, the internal case for the priority and inferiority o f
the Eudemian Ethics crumbles on close inspection. Row e, whose
careful study supports the main lines o f Jaeger’s thesis, none the
less admits ‘T h e kernel o f Jaeger’s discussion o f the ethics, his
account o f the history o f the term phronësis . . . has been shown
conclusively to rest on a misinterpretation o f the texts’ (Rowe, 19 7 1,
1 1). Criteria for tracing the development o f Aristotle’s thought that
have been offered by scholars more recent and less panoramic than
Jaeger— N uyens’s criterion o f the gradual application o f hy-
lomorphic theory, and Owen’s criterion o f the growing realization
o f the power o f the theory o f focal meaning (Nuyens, 1948; Owen,
i960)— suggest if valid that the Eudemian Ethics is earlier than the
De Anima and the developed position o f Metaphysics Γ E Z , but
they offer no clear grounds for assigning a chronological priority to
one Ethics over the others.1 Scholars have seen that in some
respects— notably in placing happiness essentially in the activity o f
the separable nous rather than in the exercise o f the virtues of man as
a whole— the Nicomachean Ethics is closer to Platonic ideals than
the Eudemian is (Monan, 1968). In particular the explicit criticisms
o f Plato in E E 1.8 seem to be made from a greater distance than
those of N E 1.6 (Cf. Verbeke, 19 51). T h e non-ethical works o f
Aristotle which in virtue o f their content stand closest to the
Eudemian Ethics (Metaphysics Λ and the Politics, especially books 1,
7, and 8, cf. Bendixen, 1856) are themselves books whose dating is a
matter o f keen scholarly debate:2 but they are both, as it happens,
works whose early placing by Jaeger has been rejected by some
recent scholars (cf. Monan, 1968; Nuyens, 1948). Finally, the
reliability o f our reconstructed texts o f the Protrepticus has been
seriously called into question (Rabinowitz, 1957). U ntil the
criticisms o f the reconstruction have been met, it seems unwise to
rely on them for the dating o f the Eudemian Ethics.
When we turn from the date to the quality o f the Eudemian Ethics
matters become yet more complicated. What counts as internal
evidence for the superiority or inferiority o f a work o f philosophy
naturally depends on the standards o f evaluation within philosophy
1. N uyens’s system has been severely criticized by Hardie (1964) and Block (19 6 1) and
seems to be now generally rejected by writers in English, despite the magisterial endorsement
o f Gauthier and J o lif (1958, 1970).
2. D üring (1966, 333) places E E before Plato’s death\J*olitics i, 7, 8 in the years o f
Aristotle’s travels; Met. Λ very early, before E E \ the rest o f Politics in Aristotle’s second
sojourn in Athens.
4 The Aristotelian Ethics
itself. An analytical philosopher in the latter half o f the twentieth
century is not likely to have the same canons o f philosophical value
to apply to an ancient work as those applied in neo-Platonic Athens,
Byzantine monasteries, or Hegelian or positivist Germ an uni­
versities o f a bygone age. As it happens, many o f the features which
scholars have noted as characteristic o f the Eudemian books in
contrast to the Nicom achean— a greater interest in the rigorous
presentation o f argument and a lesser interest in the dramatic
portrayal of character ; a preoccupation with the specific character­
istics o f practical reasoning which set it o ff from theoretical
deduction; an identification o f pleasure with pleasurable activity
rather than with a supervenient phenomenon— these are features in
which contemporary analytic fashion accords more closely with the
interests and positions o f the Eudemian than with the Nicomachean
version o f Aristotle’s system. N o conclusion, o f course, can be
drawn from this except that it is unwise to rely too heavily on
philosophical value judgements as evidence for chronological
conclusions.
M any o f those who have attempted to assess the internal
evidence for the relationship of the three Aristotelian Ethics seem to
me to have made a serious methodological error. T h ey have
compared the undisputed Nicomachean books with the undisputed
Eudemian books, and on the basis o f pre-established criteria have
attempted to settle the priority between them. T h ey have then gone
on— often almost as an afterthought— to discuss, on the basis of
their results, the provenance o f the three books which the
manuscript tradition attributes to both the Nicomachean and
Eudemian treatises. T h is is a mistaken procedure because, as I shall
show, there is far more evidence both internal and external to
establish the provenance o f these disputed or common books than
there is to enable us to decide by independent means the
relationship between the two sets o f undisputed books.1 Th us, to
tackle the question in the customary order is to reverse the correct
procedure o f approaching what is more dubious on the basis o f what
is more certain.
T h e greater part o f the present work will consist o f an attempt to
settle in a definitive manner the original context o f the disputed
i. T h e faulty method is most palpable in Jaeger, who having dated the Eudemian Ethics on
the basis o f a theory which takes for granted the Nicomachean origin o f the disputed books,
mentions the dispute itself only in a closing footnote (Jaeger, 1948, 258).
The Aristotelian Ethics in Antiquity 5
books. In order to have an untendentious mode o f reference to the
Aristotelian treatises, I shall use the abbreviation N E for the
undisputed Nicomachean books, which I shall refer to by Arabic
numerals; and I shall use, with Rom an numerals, the abbreviation
E E for the undisputed Eudemian books. For the common books I
shall use the abbreviation A E (Aristotelian Ethics) and the letters A,
B , and C. T h us A E K = N E 5 = E E IV ; A E B = N E 6 = E E V ; A E
C = N E 7 = E E V I. Having, as I hope, established the provenance
o f the disputed books I shall then, in the light o f my results, turn to
the question o f chronology and endeavour to show that the ortho­
dox theory o f Aristotle’s ethical development is devoid o f all support.
In the first part o f the work I shall limit m yself to a consideration
o f the external evidence for the relationship between the two ethics
and the provenance o f the disputed books. T h e evidence is scanty : a
mere score o f authors can be cited who discuss or explicitly utilize
either o f the treatises in the first five centuries after their
composition. B u t what evidence there is points in a single,
surprising, direction. Ancient authors up to the second century a .d .
do not regard the Nicomachean Ethics as having that primacy over
the Eudemian Ethics which has been taken for granted during the
last millennium and more. It is the Eudemian treatise which is more
likely to be used and quoted as canonical, as the Ethics o f Aristotle.
When we reach the period o f Alexander o f Aphrodisias, this
situation has clearly come to an end. T h e turning-point appears to
be the second-century commentary on Aristotle’s ethical writing by
Aspasius, which has come down to us only in fragmentary form.
Th at commentary itself, as we shall see, contains material which
rightly understood provides inescapable evidence o f the original
position o f the disputed books in the Aristotelian corpus. T h e
ancient testimonia to be discussed are, in order: Aristotle himself,
two pseudo-Aristotelian writers, Theophrastus, Andronicus o f
Rhodes, Cicero, Arius Didym us, Xenarchus, Nicolaus of
Damascus, Plutarch, Favorinus o f Arles, Aspasius, Atticus,
Diogenes Laertius, Alexander of Aphrodisias, Clement of
Alexandria, the anonymous scholiast on N E 2 -4, and the unknown
authors o f the oldest lists o f Aristotle’s surviving works.
A R IS T O T L E
Aristotle’s Ethics are quoted seven times in his other works : six
times in the Politics and once in the Metaphysics. T h e majority of
6 The Aristotelian Ethics
these quotations are from the common books o f the A E . T h us, in
Metaphysics A i, <)8rb25 we read ‘We have said in the Ethics what
the difference is between art and science and the other kindred
faculties’— a clear reference to A E B 3 ( 1 1 39bi 5 ff. : the parallel
passage in M M ii9 0 b36 ff. makes a different distinction). Three
passages in the Politics refer back to the Ethics in ways which make it
most likely that the book on Justice {A E A) is meant. T h u s Politics
B 2, 1261*30 f. says ‘Wherefore the principle o f reciprocal
proportion, as I have already remarked in the Ethics, is the salvation
o f states.’ T h is seems to allude to A E A 8, i i 3 2 b3 i f f .: ‘In
associations for exchange this sort o f justice does hold men
together— reciprocity in accordance with a proportion and not on
the basis o f precisely equal return. For it is by proportionate
requital that the state holds together.’ 1 Again, Politics Γ 9,
1 280“ 1 6 ff., ‘Ju stice implies a relation to persons as well as to things,
and a just distribution, as I have already said in the Ethics, implies
the same ratio between the persons and between the things.’ T h is is
a reference to A E A 5, i i 3 i ai 4 f f .:
Since the equal is a mean, the just will be a mean. Now equality implies at
least two things. The just, then, must be both a mean and equal and
relative (i.e. for certain persons). And qua mean it must be between certain
things (which are respectively greater and less); qua equal it involves two
things ; qua just it is for certain people. The just, therefore, involves at least
four terms ; for the persons for whom it is just are two, and the things in
which it is manifested, the objects distributed, are two.2
Connected with this is the passage in Politics Γ 12 , I28 2bi8 ff. ‘All
men think justice to be a sort o f equality ; and to a certain extent
they agree in the philosophical distinctions which have been laid
down by us about Ethics. F or they admit that justice is a thing and
1. Pol. B 2, I 2 0 ia3 0 - i : δ ιό π ε ρ το ίσ ο ν τό ά ν τ ιπ ε π ο ν θ ό ς σ φ ζ ε ι π α ς π ό λ ε ις , ώσττ€ρ εν
τ ο ΐς Ή θ ι κ ο ΐ ς ε ιρ η τ α ι π ρ ο τ ε ρ ο ν . N E 5.8, i i 3-2b3 1—3 * ά λλ ’ εν μ ε ν τ α ΐς κ ο ιν ω ν ία ις τ α ΐς
ά λ λ α κ τ ικ α ΐς σ ν ν έ χ ε ι τ ο t o l o u t o v δ ίκ α ιο ν, το ά ν τ ιπ ε π ο ν θ ό ς κ α τ ' α ν α λ ο γ ία ν κ α ι μ η
κ α τ ' ισ ό τ η τ α , τ ώ α ν τ ιπ ο ΐ€ ΐν γ α ρ ά ν ά λ ο γ ο ν σ ν μ μ ε ν ε ι ή π ό λ ις . T here is a prima-facie
discrepancy here which is resolved by the context.
2. €7ret 5 e τό ισ ο ν μ έ σ ο ν , τ ο δ ίκ α ιο ν μ έ σ ο ν τ ι α ν ε ΐη . έ σ τ ι δ ε το Ισ ον ε ν ε λ α χ ίσ τ ο ις
δ υ σ ίν . ά ν ά γ κ η τ ο ίν υ ν το δ ίκ α ιο ν μ έ σ ο ν τ€ κ α ί 'ίσον ε ΐν α ι κ α ί π ρ ό ς τ ι κ α ί τ ισ ίν , κ α ί η
μ ε ν μ έ σ ο ν , τ ιν ώ ν (τ α ΰ τ α δ 'ε σ τ ί π λ ε ΐο ν κ α ί ε λ α τ τ ο ν ), η δ ’Γσον, δ υο ιν, $ δ έ δ ίκ α ιο ν ,
τ ισ ίν . α ν α γ κ η α ρ α τ ο δ ίκ α ιο ν ε ν έ λ α χ ίσ τ ο ις ε ΐν α ι τ έ τ τ α ρ σ ι ν ο ϊς τ€ γ ά ρ δ ίκ α ιο ν
τ υ γ χ ά ν € ΐ ον, δύο ε σ τ ί, κ α ι ε ν ο??, τα π ρ ά γ μ α τ α , δυο ( N E 5.6, 1 1 i 3 i ai 4- 2 o)· ώ σ τ ’ ε π ε ι
τ ο δ ίκ α ιο ν τ ι σ ί ν , κ α ι δ ιη ρ η τ α ι τ ο ν α ν τ ό ν τ ρ ό π ο ν ε π ί τ ε τ ώ ν π ρ α γ μ ά τ ω ν , κ α ι ο ΐς,
κ α θ ά π ε ρ ε ιρ η τ α ι π ρ ό τ € ρ ο ν εν τ ο ΐς Ή θ ι κ ο ΐ ς (Pol. Γ 9, i28oai6 - i8 ).
The Aristotelian Ethics in Antiquity 7
has a relation to persons, and that equals ought to have equality.’ 1
T h e reference seems to be to the same passage in the Ethics, and in
particular to the passages which immediately precede and succeed
that just quoted.
T h e passages so far discussed, being references which fit with a
greater or lesser degree of exactitude sections o f the common books
o f the A E , do not help to settle any question o f precedence between
the N E and the E E . T h e remaining quotations from the Politics are
more interesting. ‘ I f what was said in the Ethics is true, that the
happy life is the life according to virtue lived without impediment,
and that virtue is a mean, then the life which is in a mean, and
a mean attainable by every one, must be the best,’ we are told in Δ
i i , i29 5a35 ff-2 T h e expression‘without impediment’ recalls A E C .
13, 1 1 53b9 ff. ‘Perhaps it is even necessary, if each disposition has
unimpeded activities, that whether the activity (if unimpeded) o f all
our dispositions, or that o f some one o f them is happiness, this
should be the thing most worthy o f choice.’ T h e allusion is once
again to a passage o f the common books, but this time to a section of
A E C which deals with pleasure, which even those who regard most
of the disputed books as belonging to the Nicomachean version are
prepared to assign to the Eudemian one. T h e idea that happiness is
the exercise o f virtues concerned with means is also closer to the
Eudemian ideal o f happiness as the exercise of all virtues than the
Nicomachean one according to which happiness is essentially the
exercise of an intellectual virtue which has nothing to do with a
mean.
T h e final two quotations from the Politics are more decisively
Eudemian. In chapter 13 o f Politics H (1332*7 ff.) happiness is said
to have been defined in the Ethics as ‘the complete activity and use
o f virtue, not conditionally but absolutely’ . Though the definition
cannot be found in these precise terms in either Ethics, the notion o f
the ‘ use’ o f virtue is one which is characteristic o f the Eudemian
version, and the identification o f happiness with the activity of
1 . δ ο κ ε ΐ δ έ π ά σ ι ν ίσ ο ν τ ι τ ό δ ί κ α ι ο ν ε ΐν α ι, κ α ί μ έ χ ρ ι γ έ τ ί ν ο ς ό μ ο λ ο γ ο ΰ σ ι τ ο ΐ ς κ α τ α
φ ιλ ο σ ο φ ία ν λ ό γ ο ι ς , cv ο ΐς δ ι ώ ρ ι σ τ α ι π ε ρ ί τ ώ ν η θ ι κ ώ ν (τ ι γ ά ρ κ α ί τ ι σ ί το δ ίκ α ιο ν , κ α ί
δίΓν τ ο ΐ ς ί σ ο ι ς ί’ σ ον € ΐ ν α ί φασιν).
2. €ΐ γ ά ρ κ α λ ώ ς èv τ ο ΐ ς Ή θ ι κ ο ΐ ς € ΐ ρ η τ α ι το τ ο ν € υ δ α ίμ ο ν α β ί ο ν €ΐνα ι τ ο ν κ α τ '
αρετήν άνεμπόδιστον, μ εσ ό τη τα τ η ν α ρ ε τ ή ν , το ν μ έ σ ο ν d v a y K a îo v e fv a i β ίο ν
β έ λ τ ι σ τ ο ν (Pot. Δ 1 1 , i 295a35 - ^)· δέ κ α ί ά ναγκαΐο ν, είπ ερ εκά σ της ε ξ ε ώ ς είσ ιν
ε ν έ ρ γ ε ι α ι α ν ε μ π ό δ ι σ τ ο ι , εΐθ ' η π α σ ώ ν ε ν έ ρ γ ε ι ά ε σ τ ι ν ε υ δ α ι μ ο ν ί α ε ί τ ε η τ ι ν ό ς α υ τ ώ ν ,
άν fi ά ν ε μ π ό δ ι σ τ ό ς , α ί ρ ε τ ω τ ά τ η ν ε ί ν α ι (N E 7 · ϊ 3> I I 53b9 " 12 )·
8 The Aristotelian Ethics
The Aristotelian Ethics in Antiquity 9
virtue (as opposed to ‘activity o f the soul in accordance with virtue’ )
Problems is not by Aristotle him self but by a follower who is
is likewise Eudem ian.1 Later in the same chapter o f the Politics we
drawing on his work, it is clear from a number o f points that the
read : ‘ T h is definition too was given in our ethical writings— that a
work he is drawing on is the Eudemian and not the Nicomachean
good man is the sort o f man for whom, on account o f his virtue, the
Ethics. Only the E E draws the distinction between the pleasures o f
things that are good in the abstract are good.’ T h is is the clearest o f
the tongue and the pleasures o f the throat ; and only the E E gives us
the Eudemian references, to the final book (i2 4 8 b26 f.): ‘ a good
the name o f the man who prayed to have the throat o f a crane.1
man is the sort o f man for whom things good by nature are good’ .2
O f the seven references to the Ethics, then, in Aristotle’ s other P S E U D O -A R IS T O T L E : T H E M A G N A M O R A L IA
works, five are to the common books o f the A E , one is undoubtedly
It has been recognized since the time o f Schleiermacher that the
to the E E , and the remaining one, while fitting neither exactly, is
M M and the E E are much closer to each other than either o f them is
closer to the Eudemian than to the Nicomachean version.
to the N E . For the nineteenth-century scholars who held the
Eudemian Ethics to be the work o f Aristotle’s discipline Eudemus
P S E U D O -A R IS T O T L E : THE PRO BLEM S
this presented no problem : the Magna M oralia merely represented
Books 2 7 -30 o f the Aristotelian Problems deal with topics covered a further stage in the decline o f the Peripatos from the standards set
in the Ethics: courage, temperance, justice, and wisdom re­ by the master. In the twentieth century, however, the M M set a
spectively. T h e following passage comes in the treatment of problem for the reigning orthodoxy. According to Jaeger the
temperance : chronological order of the three treatises was Eudemian,
Nicomachean, Magna M oralia : it was strange that the earliest and
Why are those who exceed in the pleasures of touch and taste called
the latest ethics should resemble each other more than either o f
incontinent? For those who exceed in sexual lust and in the enjoyment of
food are called intemperate. In some of the enjoyments connected with them resembled the intermediate one. T h is point gave support to
food the pleasure is in the tongue, in others in the throat; that is why the partisans o f von Arnim (who shared with the Jaegerians the
Philoxenus prayed to have the throat of a crane. But those whose pleasures unquestioning acceptance of the priority o f the E E to the N E ). T h e
are in sight and hearing are not so described. Is it because the pleasures order offered by von Arnim — Magna M oralia, Eudemian,
from those senses are common to us and the other animals?3 Nicomachean— was, in its way, as logical as that o f Spengel and
Fritzsche in the nineteenth century. In support o f Jaeger, Brink
T h e restriction o f the scope o f intemperance to the pleasures o f
(1933) attempted to show that despite appearances the Magna
taste and o f touch, and indeed in the case o f taste to the tactile
M oralia, in its deep structure, was closer to the Nicomachean than to
pleasures o f the gourmand, is made both in the E E (i2 3 0 b30 -
the Eudemian Ethics. This, the theme of the second half o f his
1 2 3 1 “37) and in the N E ( u i 8 ai ff.). But if this passage in the
monograph, did not win that scholarly support which has been
1. φ α μ ε ν δ ε (κ α ι δ ιω ρ ίσ μ ε θ α εν τ ο ΐς Ή θ ι κ ο ί ς , ε ι τ ι τ ω ν λ ό γ ω ν ε κ ε ίν ω ν όφ ελος)
ε ν ε ρ γ ε ια ν ftV at [sc. τ η ν ε υ δ α ιμ ο ν ία ν ] κ α ι χ ρ ή σ ιν α ρ ε τ ή ς τ ε λ ε ία ν , κ α ί τ α ύ τ η ν ο ύ κ ε ξ commanded by the first half in which he studied the peculiarities of
ύ π ο θ ε σ ε ω ς άλλ’ ά π λ ώ ς (P o l. Η 13 , 1332a7—10 ) ί δ ή λ ο ν . . . τ η ν τ η ς α ρ ε τ ή ς ε ν ε ρ γ ε ια ν the style o f the M M .2 Festugière, in his monograph on Aristotle’s
τ η ς φ ν χ η ς ά ρ ισ τ ο ν ε ΐν α ι. ή ν δ ε κ α ι ή ε υ δ α ιμ ο ν ία τ ο ά ρ ισ τ ο ν (E E II. ι, i 2 i 9 a28 - 34 )· On theory o f pleasure, reached a conclusion which, though based on a
the ‘use’ o f virtue see E E I I. 1, I 2 i9 ei8 , E E V I I I . 1 , passim, and below, p. 68. On the
difficulty o f fitting the definition to any Nicomachean text, see Newman, 18 9 1, 3 4 1, and 575. particular study o f the sections o f the Ethics devoted to this topic, is
2. κ α ί γ α ρ τ ο ύ τ ο δ ιώ ρ ισ τ α ι κ α τ ά το υ ? η θ ικ ο ύ ς λ ό γ ο υ ς , ό τ ι τ ο ιο ΰ τ ό ς ε σ τ ιν <3 true in general o f almost the whole o f the M M : ‘T h e common
σ π ο υ δ α ίο ς , ώ δ ιά τ η ν ά ρ ε τ ή ν [ τ ά ] ά γ α θ ά ε σ τ ι τ ά ά π λ ώ ς ά γ α θ ά (P ol. Η 13, 13 3 2 * 2 1-3 ) ;
a y a ö d s μ ε ν ο υ ν ε σ τ ιν ω τ ά φ ύ σ ε ι ά γ α θ ά ε σ τ ιν ά γ α θ ά ( E E V I I I .3 , X248 b20 - 7 ); τά φ ύ σ ε ι 1. In Problems 30.5, 955 b 37 science is described as an ‘ instrument o f the understanding’ in
ά γ α θ ά ~ τ α ά π λ ώ ς ά γ α θ ά (i24 9 a*_ i2). a manner paralleled in the works o f Aristotle only by E E V I I I .2, i248 a29. I am indebted to
3- δ ια τ ί ο ι κ α τ ά τ η ν τ η ς ά φ ή ς η γ ε ύ σ ε ω ς η δ ο ν ή ν , οΰ αν ύ π ε ρ β ά λ λ ω σ ι ά κ ρ α τ ε ις Professor D . J . Allan for drawing my attention to this further point o f contact between the
λ έ γ ο ν τ α ι; ο ΐ τ ε γ α ρ π ε ρ ι τ α ά φ ρ ο δ ίσ ια ά κ ό λ α σ τ ο ι, ο ι τ ε π ε ρ ι τ ά ς τ η ς τρ ο φ ή ς Problems and the E E .
ά π ο λ α ν σ ε ις . τ ω ν δ ε κ α τ ά τ η ν τ ρ ο φ ή ν α π ’ ε ν ίω ν μ ε ν εν τ ή γ λ ώ τ τ η τό ή δ ύ , α π ’ ε ν ίω ν 2. B rink’s claim that the N E was the model for the M M depended on a serious
δε εν τ ω λ ά ρ υ γ γ ι, διο κ α ί Φ ιλ ό ξ ε ν ο ς γ ε ρ ά ν ο υ φ ά ρ υ γ γ α ε ύ χ ε τ ο ε χ ε ιν . κ τ λ (9 5 °3 ι ~~7)· misunderstanding o f the structure o f E E I. T h e real structure o f that book has been made
clear by Allan (19 6 1) and Rowe (19 71 ) and is discussed below at pp. 19 1- 8 .
10 The Aristotelian Ethics The Aristotelian Ethics in Antiquity II

model o f the M M is the E E : the borrowings from the N E are o f the N E is unparalleled elsewhere in the M M . There is little if
entirely secondary. . . . In our section too the E E is the author’s any verbatim quotation o f the E E : but its ideas and structure are
textbook; he treats the N E simply as a kind o f commentary.’ 1 all-pervasive. I f scholars had not been so obstinately convinced that
T h e verdict o f Festugière seems the only possible one for those the N E is later than the E E , they might have been struck by the fact
who do not regard the M M as being the work o f Aristotle himself. that the Nicomachean cento occurs in the M M at a point in the
In my opinion the arguments o f Jaeger, Walzer, and Brink for the argument where the E E itself, in a mysterious phrase which has
inauthenticity o f the treatise have not been overthrown by the become a notorious crux, makes reference to an appendix.1 I f the
counterattacks o f von Arnim , Dirlm eier, Düring, and Cooper. Eudemian treatment were later than the Nicomachean it might well
Whether the Jaegerians were right to assign the work to a post- have carried with it an appendix o f Nicomachean material : in that
Aristotelian date is another matter, which we shall investigate in the case, the M M by incorporating it at the point suggested by the E E
final chapter o f this book. T h e weakness o f the view that the M M is would be following rather than violating its usual practice. But at
an early work o f Aristotle him self (the view o f von Arnim ’s the present stage o f the argument, that is a piece o f premature
followers, though not o f von Arnim himself) is shown above all by speculation.
the need felt by partisans o f this theory to postulate a double What is not speculation is the general closeness o f the M A I to the
redaction o f the work— a manifestly ad hoc device to escape the E E . T h e Nicomachean cento is one of half a dozen places where the
evidence for a comparatively late terminus post quem given by the M M deviates from the E E as a model, as can be seen from T able i . i .
historical allusions. T h is is a question to which we shall return. For
T able 1.1 Parallels to the M agna Moralia in the N E and E E
the moment there is no need to enter into the question o f the date of
the composition o f the M M : what is significant for our present MM book and chapter Bekker pages* Closest Aristo­
purpose is that the author, whoever he was and for whatever telian parallel
purpose he compiled his treatise, regarded the Eudemian Ethics, and
not the Nicomachean, as the appropriate Aristotelian work on which A i a (Introduction) 1 18 1 b I 182” (2) None
to base his own.2 A i b (Idea of Good) Il82b- ■i83b (3) EE
A 2 (Divisions of ‘Good') I 184“ (I) None
T h is is significant because the author undoubtedly knew the
A 3-4 (Happiness) I 184b I I 8 sb (2) EE
Nicomachean Ethics and from time to time makes use of parts o f it.

OC
t:
A 5-9 (Virtue) 1 185”- (4) N E (the Allan
D . J . Allan has recently drawn attention to a cento o f quotations cento)f
from the N E occurring in the fifth to the ninth chapters o f the first EE
i i93b (12)

00
A 10 32 (Will, Virtues)
book, beginning with a citation which is introduced by the words A 33 (Justice) • 193°- -1196" (6) AE
‘this can be seen from the Ethics' (i 1 8 sb1 5).3 As Allan says, this use A 34-B 3 (Wisdom) I iç6b 1199“ (5) AE
1. . . . le modèle ordinaire de la G .M . est VÉth. End. L e s emprunts à VÉth. N ie. sont tous
B 3 (5 Aporiai ) 11991 - I200a (2) None
secondaires. E n ce qui regarde notre section, le phénomène est manifeste et bien B 4-7 (Akrasia, Pleasure) I200b--1 2o6b (12) AE
significatif. . . Y É th.E u d. demeure, ici encore, sonlextbook. E iû n e x o it enVÉth. N ie. qu'une B 7-10 (Topics of EE VIII) i2o6b--I2o8b (4) EE
sorte de commentaire destiné à expliciter, ou, parfois, à corriger le livre qui lui sert de base’ B 11 - 12 (Friendship) i2o8b--1 2 12a (9) EE
(Festugière, 1946, lvi).
2. Theiler, 1934, says that it leaps to the eye that M M is closer t o E E ; to save the Jaegerian B 13 -14 (Self-love) I2 I2a--I2 I2b (2) NE
position he postulates a lost M iddle Ethics {M E ) between E E and NE> from which M M is B 1 5—17 (Friendship) I2 I2b--I2I3b (2) EE
derived. Among those who deny the authenticity o f the M M opinions as to its date vary from
the Theophrastean peripatos to the early first century b.c.: there is a neat summary of * T h e number in brackets is the approximate number o f Bekker columns im olved.
positions in Allan, 1957. f E\en within this section, there are passages where the ΛίΛί is dependent rather on the
3. Allan, 19 57, 1966. M ost scholars have refused to see these words as a citation, τ ο υ τ ' E E : e.g. 1 i86a2 0 -b4.
iS tt v €Ot îv €κ τ ω ν ’ Η θ ι κ ώ ν is an unusual way o f quoting, but the text is most naturally
interpreted in A llan’s way, as witness the scholars who have felt obliged to emend the text to
avoid having so to interpret it (e.g. Cooper, 1973). 1. iv τ ο ΐς ά π η λ λ α γ μ έ ν ο ίς , i220bi 1. See D irlm eier, 1962; Allan, 1966.
12 The Aristotelian Ethics
It will be seen that the M M is closer to the E E for approximately
thirty-two Bekker columns, and closer to the N E for only six.
Spengel summed up the situation thus:
At the beginning the M M looks back to both, and contributes a certain
amount of its own material. But from the tenth chapter of the first book
until the thirty-fourth there is a manifest attachment to the EE, although
here too individual expressions can be found which can only have been
taken from the NE. . . . From A 34 until B 7 the M M follows the N E, and
immediately afterwards goes back to the E E (1841, 515).
B y saying that from A 34 until B 7 the M M follows the N E and then
returns to the E E what o f course Spengel really meant was that from
A 34 until B 7 the M M followed the common books o f the A E : he
took it for granted that the disputed books belonged to the N E
because he regarded the E E as spurious. But surely it is on the face
of it unlikely that the author of the M M , having followed the
structure o f t h e £ £ faithfully from 118 7 “ to U 9 3b, should switch for
twenty-five Bekker columns to following the N E , and then switch
back, by an extraordinary coincidence, just when the disputed
books end, to follow once again the lead o f the E E . I f the disputed
books are taken as belonging with the E E , then everything runs
more smoothly : for twenty-four Bekker pages the M M lets itself be
guided, with minor variations in order and occasional small
interruptions, by the E E . The structure o f the M M is itself, then,
substantial evidence o f a Eudemian origin for the disputed books.
And its over-all fondness for the E E , despite its acquaintance with
parts at least o f the N E , is a striking indication o f the relative
standing o f the two treatises in peripatetic circles at whatever time it
was written. T h e whole topic will occupy us further in our final
chapter.
TH EO PH RA STU S
T here have survived two quotations o f the Ethics by
Theophrastus. One occurs in a passage o f Stobaeus’ anthology
taken by scholars as part o f Arius D idym us’ sum mary of
peripatetic thought.
The mean with regard to us is best; as, for instance, Theophrastus says,
one man spends a lot of time in idle chatter, another says very little and not
The Aristotelian Ethics in Antiquity !3
even what is essential, while a third chooses the right time to say just what
he ought. This mean is a mean with regard to us, because it is fixed by us
by reasoning. Thus virtue is a disposition for choice placed in a mean with
respect to us, determined by reasoning and the way that the wise man
would determine it. Then, setting out a number of triads in the footsteps of
his master, examining each in turn he endeavoured to complete an
induction as follows : for the sake of example the following were chosen :
temperance, intemperance, insensibility; gentleness, irascibility, in­
sensitivity; courage, rashness, cowardice; justice; liberality, prodigality,
meanness ; magnanimity, pusillanimity, vanity ; magnificence, shabbiness,
extravagance. Of these dispositions some are bad because they are
excessive or deficient with regard to the passions ; others are good, clearly
because they are means.1
T h e passage then proceeds to a consideration of each o f the triads in
turn. Scholars disagree about the extent and the degree of
literalness o f the quotation from Theophrastus. Von Arnim , who
first focused the attention o f the learned world on the importance o f
the passage, regarded it as an extensive, verbatim quotation. He
pointed out (1924, 12 4 -4 1) that the table o f virtues and vices given
here corresponds in several respects to the one given in the E E
(i2 2 0 b38“ i 2 2 i ai2) rather than to the one mentioned in the N E
(1107*34) so far as the latter can be reconstructed from the text.
(For example : the two lists are introduced by the same phrase ; they
both give the defect corresponding to gentleness as insensitivity
rather than spiritlessness; both mention justice with the major
moral virtues.)2 Som e details o f the subsequent development o f the
discussion, likewise, recall the Eudemian discussion o f the table:
1. T o ovv π ρ ο ς η μ ά ς μ έ σ ο ν α ρ ισ τ ο ν , ο ίο ν , φ η σ ιν ό Θ ε ό φ ρ α σ τ ο ς, ε ν τ α ΐς ε ν τ ν χ ια ις
όδι μ ε ν π ο λ λ ά δ ιε λ θ ώ ν κ α ι μ α κ ρ ώ ς ά δ ο λ ε σ χ ή σ α ς , ό δ ι δ ’ ο λ ίγ α κ α ι ο ύ δ ε τ ά ν α γ κ α ΐα ,
ο ΰ τ ο ζ Ôè α υ τ ά α ε δ ε ι μ ό ν α το ν κ α ιρ ό ν ε λ α β ε ν . Λ ύ τ η ( η ) μ ε σ ό τ η ς π ρ ο ς η μ ά ς
( ά ρ ίσ τ η ) , α ν τ η γ α ρ ύφ ' η μ ώ ν ώ ρ ισ τ α ι τ ώ λ ο γ ω . Δ ιο ε σ τ ιν η α ρ ε τ ή ‘ 'έξις
π ρ ο α ιρ ε τ ικ ή , ε ν μ ε σ ό τ η τ ι ο ν σ α τ η π ρ ο ς η μ ά ς , ώ ρ ισ μ ε ν η λ ο γ ω κ α ι ώ ς αν ό φ ρ ό ν ιμ ο ς
ό ρ ίσ ε ιε ν \ ε ΐτ α π α ρ α θ ε μ ε ν ο ς τ ιν α ς σ υ ζ υ γ ία ς , α κο λο υθ ώ ? τ ώ ύ φ η γ η τ η σ κ ο π ώ ν
ε π ε ιτ α κ α θ ' ε κ α σ τ α ε π ά γ ε ιν ε π ε ιρ ά θ η τ ο ν τ ρ ό π ο ν τ ο ύ τ ο ν · ε λ ή φ θ η σ α ν 8 ε π α ρ α δ ε ι­
γ μ ά τ ω ν χ ά ρ ιν α ΐδ ε · σ ω φ ρ ο σ ύ ν η , α κ ο λ α σ ία , α ν α ισ θ η σ ία ■ π ρ α ο τ η ς , ό ρ γ ιλ ό τ η ς ,
α ν α λ γ η σ ία · α ν δ ρ ε ία , θ ρ α σ ύ τ η ς , δ ε ιλ ία - δ ικ α ιο σ ύ ν η * * * ε λ ε υ θ ε ρ ω τ ή ς , ά σ ω τ ία ,
α ν ε λ ε υ θ ε ρ ία · μ ε γ α λ ο ψ υ χ ία , μ ικ ρ ο φ υ χ ία , χ α υ ν ό τ η ς · μ ε γ α λ ο π ρ ε π ε ια , μ ικ ρ ο π ρ έ π ε ια ,
σ α λ α κ ω ν ία . Τ ο ύ τ ω ν δη τ ώ ν ε ξ ε ω ν α ί μ ε ν τ ώ ύ π ε ρ β ά λ λ ε ιν η ε λ λ ε ίπ ε ιν π ε ρ ι π α θ η
φ α ύ λ α ί ε ίσ ιν , α ί δ« σ π ο ν δ αΓαι, τ ώ μ ε σ ό τ η τ ε ς ε ίν α ι δ η λ ο ν ό τ ι (Wachsmuth, ii. 140 6 ft.).
2. [T h eop h rastus?]: ελ ή φ θ η σ α ν δ ε π α ρ α δ ε ιγ μ ά τ ω ν χ ά ρ ιν α ΐδ ε. Ε Ε : ε ίλ ή φ θ ω δη
π α ρ α δ ε ίγ μ α τ ο ς χ ά ρ ιν . E E and Theophrastus have α ν α λ γ η σ ία where N E has ά ο ρ γ η σ ία .
Som e o f von A rnim ’s conclusions perhaps need qualification. On the order o f the list, sec
below, p. 23.
H The Aristotelian Ethics
for instance, the comparison o f the insensible man to a stone (Stob.
Ü .I4I.7, E E I2 2 Ia23).'
It is difficult to settle with any certainty the boundaries between
the nested quotations from Arius, Theophrastus, and Aristotle.
Von Arnim thought all the passage set out above was from
Theophrastus except the words ‘Theophrastus says’ and ‘Then,
setting out a number o f triads . . . as follows’ . T h is would include in
the Theophrastean passage a verbatim quotation of the definition o f
virtue from the N E ( 1 io 7 ai), and the words ‘for the sake o f example
the following were chosen’ , which would thus lead into a verbatim
quotation by Theophrastus from Aristotle. M ore recently M oraux
has argued convincingly that the Nicomachean quotation, impor­
tant features o f which are irrelevant to the context, cannot be part o f
the Theophrastean passage; he is inclined to think that the
quotation from Theophrastus is interrupted by a number o f
remarks by Areios (who is certainly the spokesman in the sentence
about setting out triads in the footsteps o f his master) and is not
resumed until the triads themselves are set out.2 T h e most natural
way o f reading the passage is to see the quotation as being resumed
immediately after ‘as follows’ (ton tropon tout on ) ; it is then difficult
not to agree with von Arnim that the clause ‘for the sake of example
the following were chosen’ is a quotation by Theophrastus o f the
familiar E E passage beginning ‘ for the sake o f example let there be
chosen . . .’ and continuing with the list o f triads. Having set forth
the E E text, Theophrastus then goes on to comment on it : there is
no reason to be surprised that not everything in the commentary
1. C f. Von der M ühl], 1909, 28. Nor everything in the discussion o f the table can be traced
to the E E : von Arnim concludes that it was the M M Theophrastus had before him. B ut the
M M neither contains nor mentions a table o f virtues, and von Arnim is obliged to suppose a
lacuna at the crucial point. M ansion’s judgement is surely sound : ‘ Pour affirmer qu’ il dépend
du passage correspondant du M M que nous ne possédons plus, on en est réduit en somme à
se baser sur le contenu conjectural d ’un texte perdu. C ’est bien peu de chose’ (19 27, 449).
2. M oraux 19 7 3 ,3 8 3 ff. M oraux concludes that the quotation must be very free: ‘ Kr zitiert
nicht wörtlich, sondern gibt in grossen Zügen das Wesentliche aus Theophrasts
Ausführungen wieder. W enn es sich so verhält, so verlieren die gelegentlichen Berührungen
mit E E oder M M beträchtlich an Bedeutung, denn es ist nicht sicher, ob sie bereits in
Theophrasts O riginalschrift gestanden haben oder von deren Epitom ator herrühren’ (390). I
do not see why A reios' method o f quoting Theophrastus should differ from his method of
quoting Aristotle when he does so at first hand : i.e. not verbatim , throughout, but preserving
many o f the actual turns o f phrase o f the original. I f the quotation o f Theophrastus is o f this
sort, then quite enough turns o f phrase and features o f structure are preserved to make it clear
that what Theophrastus was commenting on was the table o f virtues in the E E . I f von Arnim
exaggerates the amount o f information to be obtained from this text, M oraux undervalues it
to an equal degree.
The Aristotelian Ethics in Antiquity 15
can be traced to the E E itself; commentaries would be superfluous
if nothing was contained in a commentary on a text except material
from the text.
Aspasius’ commentary on Aristotle’s Ethics ( C A G x ix .15 6 .17 )
preserves another fragment in which Theophrastus made a literal
quotation from Aristotle’s Ethics : this is the sentence ‘ any pleasure,
whether it happens to be the contrary o f the pain involved or not,
has the effect, provided it be strong enough, of driving out that
pain.’ T h is comes from the section o f A E C devoted to pleasure.
T h e anonymous scholiast on A E A tells us that the quotation from
Theognis occurring at A E A, ii2()b30 is used also twice by
Theophrastus, though attributed by him to Phocylides.1
( : I ( I RO·’
A passage in De Finibus V .12 ( r . 45 b . c .) is often quoted as the
earliest evidence for the title o f the Nicomachean Ethics. Cicero
1. On Theophrastus' use o f the E ttuts, see also Regenbogen, R J: Suppl. vii, 1354 ft.;
W chrli, 1972, 4<)o ff.
2. T h e reader may be surprised to find no section dexoted to Kpicurus' use o f Aristotle.
T h e reason for the omission is that I am uneon\inced that we have any evidence that
Epicurus knew any o f Aristotle’s sur\ i\ ing ethical treatises at first hand : the negativ e, if not
the positive, side o f the thesis o f Bignone ( 11)36) seems to me to have survived the criticism o f
later scholars. Certainly, Kpicurus during his Athenian studies had the opportunity to
become familiar with Aristotle’s doctrines, and two of his major interests— pleasure and
the relation between freedom and necessity- are discussed at length in the N E and the h h ;
but nothing survives to show first-hand knowledge o f the relevant texts. Merlan ( tyfto, 1 37)
has made an interesting comparison between the two treatments o f pleasure in the N E and
Λ -Eand the theory o f K picurus; but he attempts no textual rapprochement. Furley ( 1 <jf>7, l*)l
256) compares the treatment o f voluntary action in Aristotle and Kpicurus. H e begins with
the candid admission: ‘ I have not made a systematic attempt to prove that Kpicurus could
have read or did read Aristotle’s school treatises’ ; he simply assumes in Kpicurus familiarity
with the N E . But if one is to assume without proof acquaintance with Aristotle's Iùliii s, the
E E is at least as likely, on the basis o f Fu rle y ’s comparison, to have been Kpicurus'
Aristotelian source. T h e closest textual comparison he offers is between N E 3.5, 1 1 1 3b 17 tt.
and Kpicurus referring to voluntary agents as internally caused agents —ίχ ο ν τ α ς κ α ι ei·
έ α υ τ ο ίς τ η ν α ιτ ία ν (On Nature 3 1 .27.3 <), Arringhetti). Kqually good parallels would be E E
11.6 , 1222’’ 15 2 3a20, and S, t224b2 29. On voluntariness Furley writes: ‘T h e relation
between Aristotle’s two Ethics on this problem is particularly interesting but a thorough
examination o f it would take too long now. M y own view is that the treatment in the
Eudemian Ethics is consistent with its being intermediate between Plato’s views and that of
the Nicomachean E thu s, and that the Nicomachean Etlncs (including the controversial books
K H) rather than the Eudemian Ethics is likely to have influenced Kpicurus' (p. 225). It is
unfortunate that Furley did not have time to give reasons for this judgement: in fact he goes
on to assert that on the principal point -offering a positive rather than a negative criterion ot
voluntariness— the E E is more explicit. Rist ( 1970, 101 - 2) argues that the distinctions drawn
in A E C 11 15 ‘form at least part o f the background’ for Kpicurus’ distinctions between
ι6 The Aristotelian Ethics
complains that Theophrastus in his treatise on the happy life made
happiness too dependent on good fortune. ‘L e t us therefore follow
Aristotle and his son Nicomachus, whose carefully written books on
ethics are said to be the work o f Aristotle, though I do not see why
the son could not have been similar to the father.’ 1 It is commonly
taken that Cicero is referring here exclusively to the Nicomachean
Ethics : his attribution o f the work to Nicom achus is taken not very
seriously, as a mere conjecture based on the title. But the natural
way to read the passage is that Cicero is here referring to two
authorities : first, an unnamed but undisputed work o f Aristotle’s,
and secondly an ethics with the title ‘Nicomachean’ which is
attributed to the father but which Cicero is prepared to attribute to
the son.2 T h e important point is that the ethical work which Cicero
regards as undoubtedly Aristotle’s is not the Nicomachean Ethics.3
G iven the authority which Theophrastus and the author o f the
Magna M oralia attach to the Eudemian Ethics, it is tempting to see
here a reference to that treatise; but we cannot be certain that
Cicero is not referring to the Magna M oralia or an exoteric work
like the Protrepticus.* It has often been pointed out that the doctrine
Cicero here attributes to Aristotle and Nicom achus— that virtue
and wisdom suffice for happiness without fortune— does not fit the
teaching o f the N E ; but the Eudemian Ethics also teaches, less
insistently, that gifts o f fortune are a sine qua non o f happiness, and
the common books o f the A E include an attack on the thesis that the
just man is happy on the rack.5
Cicero twice makes reference to Aristotle’s comparison o f the life
katastematic and kinetic pleasures. Su ch highly qualified phrases seem to me the appropriate
ones for expressing the relationships between Aristotle’s doctrines and those o f Epicurus. On
the alleged use o f Aristotle’s Ethics by M enander and the comic poets, see the judicious
remarks o f Bodéus, 1973, 463-4.
1. ‘Quare teneamus Aristotelem et eius filium Nicom achum, cuius accurate scripti de
moribus libri dicuntur illi quidem esse Aristoteli, sed non video, cur non potuerit patri
similis esse filius.’
2. I f only one work were in question here, Cicero should surely have written ‘Aristotelem
V e I eius filium Nicom achum ’ .
3. O f all the commentators I have read on this passage, only T itze (1826, 38 f.) has seen
this important point: he concludes the reference is to the E E .
4. On the other hand it is possible that the work he attributes to Nicomachus is the M M
and not the N E : we know from Elias (C A G xviii, 32) that the Magna M oraita was known in
antiquity as the ‘ Great Nicomachean Ethics’ , the N E as the ‘Lesser Nicomachean Ethics’ .
5. See e.g. E E I, 1 2 Γ41516 ; V I I I , i249ai4 ; A E C , 1 i5 3 big.
The Aristotelian Ethics in Antiquity 17
o f a Sardanapallus to that o f a bu ll,1 which occurs in E E (1.5,
I 2 i6 ai- i6 ) and not in N E . But he does so in a way which makes it
probable that he is drawing on a different source, which may or may
not have been, as Jaeger conjectured, the Protrepticus.2
In De Fmihus 11.6 Cicero attributes to Aristotle a definition of
happiness as the exercise o f virtue accompanied by the prosperity of
a complete life. T h is definition is perhaps closer to the one given in
the Magna M oralia than to anything in either o f the authentic
Ethics.3
Cicero, then, is a witness that in his time the N E did not yet enjoy
that primacy as the canonical expression o f Aristotle’s ethical
thought which it held for later generations : but it is unclear whether
his own knowledge o f Aristotle’s ethical teaching, in so far as it
depends at all on first-hand acquaintance with his writings, derives
from the E E , the M M , or the exoteric writings.4
T H E E D IT IO N O F A N D R O N IC U S
Some time in the first century B . C . 5 Andronicus o f Rhodes
undertook an edition of Aristotle’s works : he grouped and ordered
the treatises, it appears, in the form w hich the manuscript tradition
was later to hand them down to us. H e drew up a pina.x, or catalogue
raisonné, o f Aristotle’s works. T h is has not survived, but scholars
believe that it can be reconstructed from the list o f Aristotle’s works
found in the L ife o f Aristotle by Ptolem y— which is known only
from late Arabic sources: Ibn al-Qifti (i2 th -i3 th cent.) and Ibn
Abi Usaibia ( 1 3th cent.) both of w hom offer closely resembling lists
o f the w ritings o f the Aristotelian corpus which they attribute to a
1. Tusculan Disputations, v .3 5 .1 0 1 ; De Finibus ii.32.106.
2. Jaeger, 1948, 25 4 -5. On Cicero’s possible use o f the Protrepticus in his lost Hortensius,
see Diels, r888; Chroust, 1973, 98 ff.; Rabinowitz, 1957.
3. ‘Virtutis usum cum vitae perfectae prosperitate’ . C f. M M A 3 -4 , 1 1 8-|.b 15 - 39. Closer
than any Aristotelian source is the version in Areios Didym os, ap. Stobaeus ii.5 1.12
W achsmuth: χ ρ ή σ ις α ρ ε τ ή ς T iÀ eiaç kv β ίω reÀ eio j π ρ ο η γ ο ύ μ ε ν η .
4. Cicero him self makes the distinction between A ristotle's exoteric writings and those
written in a more austere professional style {limatius). H e could have found the distinction in
the N E (1 i02a27) the A E (1140*3), and most clearly in the E E ( 1 2 1 ^ 2 2 ) , though it is also to
be found in the Politics. Dirlm eier (19 6 9 ,116 ) is surely correct in thinking that D üring (1957,
442) goes too far in saying that this passage proves knowledge o f the E E . Elsewhere (1957,
428) Düring him self is more cautious on the same topic.
5. Scholars disagree on the date o f Andronicus’ edition: D üring (1968, 195) places his
activity between 40 and 20 B .c.; M oraux, accepting the tradition o f Elias (C A G xviii. 113 ),
thinks he became head o f the peripatetic school c. 80-78 (19 73, 58).
ι8 The Aristotelian Ethics
Ptolem y-el-G arib.1 Clearly, a reconstruction o f Andronicus’
catalogue based on a tradition passing through so many in­
termediaries (Ptolemy, his Arabic translator, and the two medieval
writers) cannot be treated with complete confidence. None the less,
it is o f great interest that our best evidence for the contents o f the
edition o f Andronicus suggests unequivocally that it did not contain
the Nicomachean Ethics. T h e only ethical entries in the list are:
35, a book entitled M ajor treatises on ethics, in Greek ëthikôn
megalön, two books;
36, a book entitled M inor treatises on ethics, which he dedicated to
Eudem us, in Greek ëthikon Eudëmeiôn, eight books.2
Not only is the omission o f the N E significant : so also is the fact that
the E E is listed as having eight books. T h is means, if the list in the
Arabic sources really goes back to Andronicus’ pinax, that the
disputed books were included in his edition with the E E . A s has
been said, the complicated history o f the transmission o f the pinax
makes it difficult to be certain o f its content : but changes made by
editors in the course o f its transmission would surely be likely to be
in the direction o f making it accord with the state o f the Aristotelian
corpus at the time o f editing. It is, then, very striking that during a
period o f centuries during which the N E became regarded as the
Ethics par excellence, and during which the disputed books were
commented on as part o f the Nicomachean treatise, this evidence
for a deutero-canonical status o f the N E and for a Eudemian
provenance o f the disputed books should have been so faithfully
preserved.
1. N o one has succeeded in identifying, or even dating within a century or two, this
Ptolemy to the satisfaction o f other scholars. See M oraux, 19 5 1, 289-94; D üring, τ968, ι68.
D üring regards it as probable that he is identical with the fourth-century Platonist, cited by
Iamblichus and Proclus.
2. T h e translation is that o f D üring, 1957, 224. M any commentators cannot believe their
eyes at this point, and emend the text (e.g. M oraux, 19 5 1, 138). A recent example o f such
incredulity is Bodéus, who having decided that the omission o f the N E is ‘un accident dans
la transm ission’ goes on to conclude— on the sole ground that Andronicus and Eudem us both
hailed from Rhodes— that it was Andronicus who put together the E E ‘parachevant son
édition en la baptisant sur le modèle dé YÉihtque à Nicomaque' (19 73, 462). ‘ L ’attitude
d ’A ndronicus’, he continues, ‘ éditant sur cette base un nouveau cursus revenait, en somme, à
épingler l’Ethique à Nicomaque en m odèle; les commentateurs de l’ère chrétienne ont
respecté cette échelle des valeurs, en négligeant à son profit YÉthique à Eudème et les Grandes
M o r a l e s It would be difficult to find— even among Aristotelian scholars— a case o f such
weighty conclusions being drawn from a non-existent piece o f evidence.
The Aristotelian Ethics in Antiquity 19
XENARCH US
D uring the latter part o f the first century B .C . the scholarch o f the
Peripatos was Xenarchus. Sim plicius has preserved, in his
commentary on Aristotle’s De Caelo (C A G vii.5 5 .2 5 -3 1) a
fragment o f his writing. D iscussing the principle that everything
has one and only one contrary or opposite, he observes that
Xenarchus gave as a counter-example the fact that every virtue has
two opposites. ‘ In our ethical writings,’ he says, ‘we say that to each
o f the virtues there are two contraries : for instance wisdom has the
two contraries cunning and stupidity, courage has the two
contraries rashness and cowardice, and so in other cases.’ 1 The
examples show that the writings he is alluding to were following the
E E , not the N E : the listing of wisdom as a mean between cunning
and stupidity occurs only in the table o f virtues in E E I I .3,
i 2 2 i ai2 .2 N o passage in the N E suggests anything o f the kind, and
it would be hard to pick a single brief quote from the E E which
would more clearly betray its source.3
A R IU S D ID Y M U S
It is commonly agreed among scholars that the outline o f
peripatetic ethics preserved in the second book o f Stobaeus’
anthology is one prepared by Arius Didym us, the court philo­
sopher o f the emperor Augustus. T h e outline, which occupies some
thirty-six pages o f Wachsmuth’s edition Stobaeus (116 -4 9 ), is
entitled ‘T h e opinions o f Aristotle and the other Peripatetics about
Ethics’ . It does not profess, therefore, to be a summary o f
Aristotle’s views: and in part it bears apparent marks o f Stoic
influence. Von Arnim , the author o f the most extended twentieth-
century study o f the work (1926), strove to show that the Stoic
influence was no more than apparent, and that the non-Aristotelian
elements need not derive from any source substantially later than
1. λ ε γ ο μ ε ν 6e, φ η σ ί [<Ξ ε ν ά ρ χ ο ς ] κ α ι ε v τ ο ΐς π ε ρ ι τ ω ν -ηθών λ ό γ ο ις > έ κ α σ τ η τ ω ν
α ρ ε τ ώ ν δυο ε ίν α ι τ ά ε ν α ν τ ία , ώ ς φ ρ ο ν ή σ ε ι μ ε ν π α ν ο υ ρ γ ία ν κ α ί ε ν ή θ ε ια ν , α ν δ ρ ε ία δε
θ ρ α σ ν τ η τ α τ ε κ α ί δ ε ιλ ία ν κ α ί ε π ί τ ώ ν ά λλ ω ν ο μ ο ίω ς .
2. Even in the E E the appearance o f wisdom as a mean is surprising, with the apparent
corollary that wisdom is a moral virtue. For this reason some scholars, without manuscript
warrant, have excised i 2 2 i ai2 (von Arnim, 1924, 127). See below, pp. 2 1 3 - 1 4 .
3. See M oraux, 19 7 3, 1.208. O f Xenarchus1 date, M oraux says: \ .. werden wir kaum
fehlgehen, wenn sir sein Leben zwischen 80/75 v · C hr. und der Zeitwende ansetzen’ (ibid.
197)-
20 The Aristotelian Ethics
Theophrastus. H is thesis has not secured universal assen t;1 for our
present purposes it is fortunately not necessary to decide whether
the non-Aristotelian parts o f the epitome derive from Theophrastus
or from later sources. What is important, and what is established
beyond doubt by von Arnim ’s extensive and painstaking
examination, is that in its Aristotelian sections it is much closer
to the Eudemian Ethics and to the M agna M oralia than it is to the
Nicomachean Ethics— as indeed is apparent to anyone who looks up
the parallels marked in the footnotes to W achsmuth’s edition.2 T h e
reader is referred to von Arnim for the detailed comparison: but
here are a few parallel texts by way o f illustration.
Arius, 1 28.11 ff. : Excellence is the name of the best disposition, of the
disposition which makes its possessor in the best condition. This is clear
from induction: for excellence in a shoemaker is the condition which
makes him best able to make a good shoe. . . .
E E II. i, I2i8b37“ i9ai : Let it be assumed as to excellence that it is the best
disposition or state or power of each thing that has a use or output. This is
clear from induction . . . (i2 i9 aig-23) . . . we say that something is the
output of a thing and its excellence, though not in the same way ; as a shoe
is the output of the act of shoemaking and of the art of shoemaking ; so that
if there is an excellence of shoemaking and of the good shoemaker, its
output is a good shoe.3
Arius, 147.22: The virtue which is a synthesis of all the moral virtues is
called kalokagathia, and it is perfect excellence; it makes expedient goods
into noble goods, and chooses noble things for their own sake.4
In most respects, this passage could hardly be bettered as a
summary o f the first part o f the final chapter of the E E , the key
passage dealing with the union o f all excellence in an all-round
virtue o f kalokagathia— a notion quite lacking in the N E .
1. F o r a recent study and a bibliography, see M ingay, 1973.
2. So too the most recent account, M oraux, 1973, 4 35 : . . dabei sind die Berührungen
mit den M M und auch die der E E enger und zahlreicher als mit der N E .'
3. ά ρ ζ τ ή ν δ ’ ώ ν ο μ ά σ θ at τ η ν ά ρ ίσ τ η ν δ ιά θ ζ σ ιν η κ α θ ' η ν ά ρ ισ τ α δ ιά κ € ΐτ α ι το έχο ν.
τ ο ΰ τ ο δ’ €κ τ η ς έ π α γ ω γ ή ς δ ή λ ο ν σ κ υ τ ο τ ό μ ο υ γ ά ρ ά ρ € τη ν λ έ γ € σ θ α ι καθ'1 η ν
ά π ο τ ζ λ ζ ΐν ά ρ ισ τ ο ν υ π ό δ η μ α δ ύ ν α τ α ι (Areios, 1 2 8 .11 ff.), τ α ν τ α δη ο ύ τ ω ς ν π ο κ € ΐσ θ ω
κ α ί 7T€pl ά ρ € τ η ς , ο τ ι έ σ τ ιν ή β ζ λ τ ίσ τ η δ ιά 0€ σ ις η έ ξ ις ή δ ν ν α μ ις έ κ ά σ τ ω ν , ό σ ω ν έ σ τ ί
τ ι ς χ ρ ή σ ις η 'έργον. δ ή λ ο ν δ ’ έ κ τ η ς έ π α γ ω γ ή ς . . . λ έ γ ο μ € ν ο τ ι το 'έργον το ν
π ρ ά γ μ α τ ο ς κ α ι τ η ς ά ρ € τ ή ς, ά λλ’ ο υ χ ω σ α ύ τ ω ς , ο ϊο ν σ κ ν τ ο τ ο μ ικ ή ς κ α ι σ κ υ τ ε ύ σ € ω ς
υ π ό δ η μ α ■ €ΐ δ ή τ ι ς έ σ τ ιν ά ρ € τ ή σ κ υ τ ικ ή ς κ α ι σ π ο υ δ α ίο υ σ κ υ τ έ ω ς , το έ ρ γ ο ν έ σ τ ι
σ π ο υ δ α ΐο ν υ π ό δ η μ α (E E I I . i, I 2 i8 b 37 ff·)·
4. τ η ν δβ έ κ π α σ ώ ν τ ώ ν η θ ικ ώ ν ά ρ € τ η ν σ υ ν ς ο τ η κ υ ια ν λ έ γ € σ θ α ι μ έ ν κ α λ ο κ ά γ α θ -
ίαν> T eX eiav δ ’ ά ρ € τ η ν e îv a i, τ α τ€ ά γ α θ ά ω φ έ λ ιμ α κ α ί κ α λ ά π ο ιο ύ σ α ν τ α r e κα λ α δ ι
α υ τ ά a Ιρ ο υ μ έν η ν .
The Aristotelian Ethics in Antiquity 21
I have chosen these two passages because they illustrate
dependences on the E E which must be direct and not mediated via
the M M . T h e similarities with that work are even more striking
than the similarities with the E E : for instance the thirteen main
moral excellences are listed in exactly the same order in Arius (p.
146) as they are in M M A 20-34, 1 i90b9 - i iç 6 b3. Almost certainly
Arius knew also the N E , since a number o f points to be found in his
outline occur only in that work and not in the E E or M M . 1 This
makes it the more significant that he should prefer to follow the
structure o f the E E and M M and that on a number o f points where
the E E and N E diverge he should follow the Eudemian version.2
In Stobaeus’ anthology, prior to the outline o f Stoic and
peripatetic philosophy, there occurs a section o f prolegomena. This
passage contains the following remark: ‘Aristotle in the tenth book
o f the Nicomachean Ethics thinks that the astronomer Eudoxus held
the doctrine that the end is pleasure’ (Wachsmuth 5 2 .10 ).3 Are
these prolegomena, as many scholars assume, drawn from the
epitome o f Arius ? I f they are, then the passage just quoted is of
great importance : it becomes the earliest unambiguous evidence for
the existence o f the Nicomachean Ethics— for an Ethics, moreover,
in ten books, which must therefore have included the disputed
books. But is the statement really by Arius?
T h e reason for believing that the summary o f peripatetic ethics
in Stobaeus is by Arius is that the anonymous epitome contains a
passage ( 12 9 .19 - 13 0 .12 ) which appears also in the fourth volume of
the anthology (9 18 .16 -9 19 .6 ) with the rubric ‘ffom the epitome of
D idym us’.4 T h is makes it likely that the whole epitome also belongs
to Arius, and it is a reasonable assumption that so too does the
preceding epitome o f Stoic philosophy constructed on the same
1. e.g. the expression ev β ίω τ € λ € ΐω on p. 13 0 ; the parallel to N E 8 .12 at 14 8 .15 ; the
reference to the proverb about one swallow not making a sum m erat 13 2 .5 ; the description o f
the good man in adversity ( 13 2 .10 ) ; the definition o f virtue discussed above, p. 14.
2. T h e following are among the points on which Arius follows Eudemian by preference to
Nicomachean doctrine: the identification o f the blessed and the happy ( 1 3 2 .1 1 ) ; the
restriction o f happiness to those who are awake (the divergence here— 1 3 3 . 1 1 — is one o f
terminology rather than o f substance, cf. von Arnim 19 2 6 ,4 1) ; the triple sense o f ‘T h e G ood’
(Areios 134.8 ff. = E E I, 1 2 1 7b 1 ff.); the attitude to riches (cf. von Arnim , 60); the theory o f
natural virtue (von Arnim , 7 1 ) ; the presence o f kalokagathia, as illustrated in the passage
cited, and the absence o f an account o f theoretical happiness on the model o f N E 10.
3. 1 Α ρ ισ τ ο τ έ λ η ς έν τ ώ δ € κ α τ ω τ ώ ν Ν ικ ο μ α χ ε ίω ν Ε ϋ δ ο ξ ο ν τ ο ν ά σ τ ρ ό λ ο γ ο ν
ο ’ί ε τ α ι τ έ λ ο ς δ ο γ μ α τ ίζ € ΐν τ η ν ηδο νή ν.
4· T his was first pointed out by M einecke (18 5 9 ); his interpretation has won general
acceptance especially since D iels, 1879, 7 1 ff.
22 The Aristotelian Ethics The Aristotelian Ethics in Antiquity 23

model. But it provides no grounds for believing that every reference
to peripatetic philosophy in Stobaeus is a direct quotation from
Arius. T h e passage which refers to N E 10 is preceded by a
discussion o f Aristotle’s definition o f happiness (5 0 .11 - 5 1 .1 7 ) and
o f possible goals ( 5 1.18 - 5 2 .9 ) ; a passage marked by an obsession
with triads, which in part duplicates and in part conflicts with the
treatment o f Aristotle’s views on happiness presented in the
epitome itself.1 T h e explicit quotation of, and reference to, book 10
o f the N E is totally without parallel in the epitome and is quite
unlike Arius’ manner o f utilizing his texts.2 Altogether, it is much
more likely that the reference to the Eudoxus passage is by Stobaeus
himself, and dates therefore from the fifth rather than the first
century a . d .
though not the work o f Nicolaus this central section may well be the
oldest surviving commentary on the Aristotelian Ethics}
It is regrettable that the date o f the treatise is so uncertain
because in the context o f the relationship between the N E and the
E E its interest is considerable. A large part o f the discussion is
devoted to the ethical virtues, and both in general structure and in
detail the treatment follows the E E and the M M more closely than
it follows the N E , although from time to time Nicomachean
material is utilized.
T h e order in which the major moral virtues are treated is closer to
the E E and M M than to the N E . (No great importance can be
attached to the order o f treatment : but it may be o f interest to set
T able 1 .2 The Ordering o f the Virtues m Peripatetic Ethical Treatises

N IC O L A U S O F D A M A S C U S Virtue NE NE EE EE EE MM Theo. Arius Fez

Som e twenty years ago there was discovered in a mosque at Fez 2 -7 3 -4 list 2 -3 3

an Arabic manuscript containing, along with an Arabian version o f Courage I I 2 2 I I I

3 I
parts o f the Nicomachean Ethics, a translation o f a Greek ethical T emperance 2 2 2 2 I 2 2
4 3
treatise attributed by the copyist to one Nicolaus. Scholars Liberality 3 3 7 10 4 4 5 4 4
conjectured that it might be the work o f Nicolaus o f Damascus, the Magnificence 4 4 !3 I I 6 6 7 6 5
adviser and court historian of Herod the Great, who wrote a Magnanimity 5 5 12 9 5 5 6 5 6
number o f peripatetic treatises. T h is attribution was endorsed by (Ambition)* 6 6 - - — — — — — —

R . Walzer in The Encyclopedia o f Islam .3 Gentleness 7 7 I I 3 3 2 3 3

Unless the treatise has been very heavily interpolated it cannot in Candour 8 9 8 5 I I 12 — 12 9
Wittiness 9 10 — 12 10 — 10 12
fact be the work o f Nicolaus, because it refers to Plotinus and
Friendliness ίο (8)* 9 6 9 II — [ I 7
mentions other proper names from late antiquity. However, the
Dignity — — 10 7 10 8 — 8 8
central section o f the treatise, which breaks the continuity o f the Shame I I I I — —
3 8 9 9 I I
work as a whole, may well be o f earlier date than the surrounding Nemesis 12 10
5 5 J3 7 7 — 7
material: it is a summary o f Aristotelian ethics which follows Justice 12
13 6 4 ‘ 3 >3 4 l3 r3
sources in Aristotle very closely. T h e only indication o f its date is a Hardiness ---- — I I 8 — — — —
critical reference to the Christians. Scholars have suggested that Wisdom --- — H 12 — — — — —
* H ere this virtue is unnamed.
1. Several o f these divergencies are noted by M oraux (19 73, 274) who concludes ‘D aran
geht deutlich hervor, dass Areios aus zwei verschieden Quellen, die eine jeweils anderer i. See M oraux, 1973, 442. Partly because o f the anti-Christian remarks Lyon s suggests
Deutung desselben D efinition bote, geschöpft hat.’ It is surely Stobaeus, not A rius, who is that the Nicolaus o f the treatise may be Nicolaus o f Laodicea, a contemporary o f Julian the
using two incompatible sources. Apostate. B ut there is nothing in the central section which is incompatible with a
2. T h is too is noticed by M oraux. From A rius’ general method o f using his sources, considerably earlier date. It is known that Nicolaus o f Damascus did write an introduction to
M oraux concludes that he may well not have read them at first hand; it is the more the Ethics: it is quoted by the twelfth-century Arabic philosopher Ibn A l-M atran (Peters,
astonishing that he should ascribe this precisely located reference to A rius rather than Aristotle and the Arabs, 270). T he surviving genuine works o f Nicolaus make use o f
Stobaeus. (Compare M oraux, 19 73, 3 13 and 436.) Aristotelian ethical ideas (e.g. that man is an ά ρ χ ή π ρ ά ξ€ ω ς ) but not in a manner which
3. On the Fez manuscript see A rberry, 19 5 5 ; L yon s, Oriens, 1 3 .5 5 ; Paret, 1959/60. enables one to tell which treatise is being drawn upon. See Drossaart Lulofs, 1969.
24 The Aristotelian Ethics
out the ordering o f the virtues in the treatises we have so far
considered (Table 1.2).) M ore important, the inclusions and
exclusions in the list agree with the E E and M M against the N E .
T h e principal differences between the treatment o f the virtues in
N E 3 -4 and that in E E I I I is that the N E contains seven major
virtues and three minor ones (the distinction between the two is
drawn at 1 io8a9) whereas the E E contains six virtues and six ‘means
o f passion’ (the distinction between the two being drawn at
i2 3 4 a24~34). T h e six virtues o f the E E are the seven virtues listed
as major in the N E , less the unnamed one which is a proper degree
o f ambition ; the six passional means o f the E E are the three minor
virtues o f the N E , plus shame, nemesis, and dignity. (The first two o f
these are mentioned in the N E in the preliminary discussion in book
2 o f means o f passion, but only one, shame, is actually discussed in
the full treatment in book 4 ; dignity is not mentioned in the N E at
all.) In these respects the Fez manuscript, like the M M and Areios,
follows the E E and not the N E : it does not include the unnamed
mean o f ambition and therefore has six and not seven major virtues ;
it includes dignity in its list, and it has a discussion o f nemesis or
righteous indignation which is based on the E E text. T o be sure, in
these respects the Fez manuscript resembles the M M also; but a
number o f details indicate direct dependence on the E E : for
instance, the naming o f the gourmand who wanted a crane’s gullet
as the son o f Eryxis (cf. E E III , i 2 3 i ai8). In the discussions o f the
particular virtues the text is sometimes closer to the Eudemian
version (as in accounts o f courage1 and temperance) and
sometimes closer to the Nicomachean (e.g. in the distinction
between possession and use in the treatment o f liberality, and in the
attribution o f wittiness to Socrates as at N E i i 2 7 b5).
P H IL O O F A L E X A N D R IA
Philo (d. A .D . 45) shows acquaintance with a number o f
Aristotelian ethical themes, such as the doctrine of the mean and the
peripatetic definition o f happiness. T h e way in which Philo applies
the doctrine o f the mean to particular virtues makes it more likely, if
he is drawing on Aristotle directly, that he was acquainted with the
Eudem ian than the Nicomachean version o f the doctrine;2 but the
1. Especially in the accounts o f the five types o f unreal courage.
2. W hile most o f the virtues and vices to which Philo applies the doctrine o f the mean
occur in both the E E and N E lists (e.g. α λ α ζ ο ν ε ία , θ ρ α σ ύ τ η ς , δ ε ιλ ία , α ν δ ρ εία ,
The Aristotelian Ethics in Antiquity 25
freedom with which he applies it— e.g. to the triad im piety-piety-
superstition (Immut. xxxiv. 16 3)— leaves it open that he is improvis­
ing upon a theme known to him at second hand. Philo divides
virtues into those that are inborn, those that are acquired by
training, and those that come by treaching, in a way which has
reminded some scholars o f N E 10, i i7 9 b2o; but it is equally
reminiscent o f Plato’s Meno (70ai ff.) as is his frequent use of the
concept o f divine destiny.1 Philo frequently uses the threefold
division o f goods into external goods, goods o f the body and goods
o f the soul : this division is to be found in N E 1, i098bi2 and other
Aristotelian contexts, but it is also very frequent in Plato.2 Philo
defines happiness as the use o f complete virtue in a complete life:
this is close to the E E definition, but closer to the M M and closer
still to the epitome o f Areios D idym us.3 Like the A E , Philo regards
happiness as a combination o f learning and wisdom ; but unlike the
A E he regards learning as being concerned with the service o f God
(Praem. xiv.81). T h e service o f G od, he says, in a phrase which
echoes the final section o f the E E , is the beginning and end of
happiness: G od is the standard (horos) o f happiness (De Spec. Leg.,
ed. Cohn, v.594). Philo’s use o f Aristotelian ethical material is too
free for us to be able to say that he actually quotes the ethical
treatises, or to decide with confidence which of them he is using: we
can only say he is in general slightly closer to the E E than to the
N E .A Certainly there is no passage which conclusively proves first­
hand knowledge o f any Aristotelian ethical treatise.5
δ α π α ν η ρ ία , σ ω φ ρ ο σ ύ ν η --s e e Aitgr. xxvi.147 and Immul. xxiv.16 3 4) φ ρ ό νη σ ές and
π α ν ο υ ρ γ ία occur only in the Eudemian treatment.
1. See e.g. Somn. i.x xv ii.16 4 ; on θ ε ία μ ο ίρ α see Wolfson, 1962, 165 ft.
2. T he relevant texts in Plato are collected in Gauthier and Jo lif, 1959, 6 2 ; for Philo see
Cohn’s edition i.259, v.325, etc.
3. Philo, ed. Cohn, i.2 7 2 ; see p. 17 above. Philo regards happiness as a combination of
practical and theoretical life, like E E and the Politics, but he uses this division in a way
somewhat different from Aristotle (Praem. 2 .11) .
4. Wolfson, 1962, 165 ff., collects together all the texts in which he thinks Philo is
following Aristotle: none o f them amounts to a verbatim quotation.
5. T h e same observation holds for Seneca, some years later than Philo. Som e scholars have
thought that the clementia o f Seneca is a Romanization o f Aristotle’s ε π ιε ίκ ε ια . G riffin
(1976, 159 ff.) has shown this opinion to be without foundation, and M urray, who once
(1965, 1 7 6 ff.) discussed whether the Aristotelian theory o f equity gained widespread
acceptance towards the end o f the Roman republic, has decided that it did not (1967, 355 f.).
Even i f Seneca was acquainted with Aristotle’s notion o f equity, he can have derived his
knowledge o f it from a reading o f the Rhetoric (1374*26 ff.), a work with which the De Ira
suggests familiarity.
20 The Aristotelian Ethics
PLU TARCH
A moralist as voluminous as Plutarch might be expected to have
had ample occasion to quote from Aristotle’s ethical writings. A
glance at Helmbold and O ’N eill’s index Plutarch's Quotations seems
to confirm this expectation, for it lists seventeen reminiscences or
quotations to the N E and three to the E E . Inspection o f the three
E E passages mentioned show that the corresponding passages in
Plutarch are simply common quotations o f third parties (Herodotus
and Bion) and so not any evidence of knowledge of the E E . This
fact, plus the prima facie striking fact that not one o f the seventeen
passages o f the N E listed in the index is from the A E , might seem
to lend support to the hypothesis that the N E did not originally
contain them: for the hypothesis would explain why Plutarch,
apparently a keen jV£-quoter and a n on-££-quoter, makes no use
o f them. But, alas, close inspection shows that o f the seventeen
‘ reminiscences’ o f the N E all except one are souvenirs so faint as to
amount to no more than the common employment o f a com­
monplace. T h e single passage in which Aristotle is explicitly quoted
is at 704 E , where Plutarch criticizes him for saying that there is no
incontinence in respect o f the pleasures o f sight and hearing on the
grounds that these are peculiar to humans.1 T h e reference could be
either to the N E or the E E ; if it has to be to one or other, the E E is
the more likely source because o f details o f language. But more
likely than either as a source for Plutarch is the passage o f the
Problems considered earlier.2 Only there is the question o f the
sharing o f pleasures with animals brought into conjunction with the
topic o f incontinence rather than intemperance.
F A V O R IN U S O F A R L E S
In the Life o f Aristotle by Diogenes Laertius there is a list o f
Aristotle’s sayings which concludes thus: ‘Favorinus in the second
book o f his Memorabilia mentions as one o f his habitual sayings “ H e
who has many friends has no friend” . T h is can be found also in the
seventh book o f the Ethics.’ 3 Though the sentiment quoted by
1. 8o k€î 0€ μ ο ι μ η δ ' 'Α ρ ισ τ ο τ έ λ η ς α ιτ ία δ ίκ α ια τ ά ς π €p i θ έ α ν κ α ι α κ ρ ό α σ ιν
€ ν π α θ € ΐα ς ά π ο λ ν ε ιν ά κ ρ α σ ία ς , ώ ς μ ό ν α ς ά ν θ ρ ω π ίν α ς ο ν σ α ς, τ α ΐς δ ’ άλΛ αι? κ α ι τα
θ η ρ ία φ ύ σ ιν ε χ ο ν τ α χ ρ ή σ θ α ι κ α ι κ ο ιν ω ν ε ΐν .
2. ρ. 8.
φ η σ ι δ ε Φ α β ω ρ ιν ο ς i v τ ω δ ε υ τ ε ρ ω τ ω ν ''Α π ο μ ν η μ ο ν ε υ μ ά τ ω ν ώ ? έ κ ά σ τ ο τ ε
λ έ γ ο ι, ‘ ω φ ίλ ο ι, ο ύ δ ε ίς φ ίλ ο ς \ άλλα κ α ι ev τ ω ε β δ ό μ ω τ ω ν ’Η θ ι κ ώ ν ε σ τ ι.
The Aristotelian Ethics in Antiquity 27
Favorinus can be traced in both Ethics, in the epigrammatic form
quoted it occurs only in the E E ( V I I .12 , i24 5b2 i ) .1 T h is is made
explicit in the attached comment, which must be a reference to the
Eudemian version since the corresponding passage in the N E
( 1 1 7 i a1 5—17) occurs in the ninth book.2 T h e comment is evidence
also that the version o f the E E referred to must have contained the
disputed books o f the A E ; otherwise the book on friendship would
be the fourth and not the seventh book.3 It is striking that the E E is
cited so firmly as the Ethics o f Aristotle.
It would be pleasant to know for certain who was responsible for
the citation. It is possible, though hardly likely, that it was attached
to the epigram by Favorinus. It may be by Diogenes himself, or by
some earlier writer from whom he has drawn the collection of
Aristotelian sayings. Other scholars have thought that the citation
was attached to the epigram before Favorinus and originated in
some source o f his. Once the edition o f Andronicus had established
the Nicomachean Ethics in a position o f pre-eminence, argued von
der Miihll (1909, 26), no one could refer to the E E as ‘the Ethics'·,
therefore the quotation must be by some writer o f the Ptolemaic era
who included a list o f apophthegms in a life o f Aristotle.4 D üring
went so far as to identify this ancient source as H erm ippus.5
In fact, as we have seen, there is no reason to believe that
Andronicus’ edition even included the Nicomachean Ethics, let
alone that it conferred on it a pre-eminent status. T h e allusion to
the E E as Aristotle’s Ethics is consistent with all the ancient
allusions we have so far considered. T o whatever date prior to
1. ο ύ θ ε ίς φ ίλ ο ς ω π ο λ λ ο ί φ ίλ ο ι.
2. ο ι Sè π ο λ ν φ ιλ ο ι κ α ι π ά σ ιν ο ικ ε ίω ς έ ν τ ν γ χ ά ν ο ν τ ε ς ούδξν'ι δ ο κ ο ν σ ιν e îv a i φ ίλ ο ι,
π λ ή ν ττο λ ιτικ ώ ς .
3· Consequently those who have, implausibly, maintained that the E E is complete in five
books have had to explain away this passage in D iogenes as a late gloss (e.g. T itze : see Rowe,
19 7 1, 80).
4* ‘U t Diogenem vel Favorinum , quem Diogenes excerpsit, Eudem ia perlegisse putem,
vix mihi persuadebis, praesertim cum editione Androniaca pervulgata Nicomachea talia
dignitati Eudemiis praecellerent, ut Ëthikôn vocabulo semper ad Nicomachea usus
scriptorum revocaret. Itaque laudationem illam neque Diogenis neque Favorini sapientiae
thesauros suppeditasse apparet, sed Ptolemaeorum saeculo a viro docto in vita Aristotelis
cum apophthegmate collocata est, unde Favorinus eam recepit.’
5- H is comment on the passage is puzzling, seeming to support two inconsistent
interpretations: ‘T h en follows . . . a note from Phavorinus, including an annotation from
D iogenes him self “ this is found in the seventh book o f the Ethics, too” . T h is note proves that
H erm ippus (or the unknown author) had access to or knew an edition o f the Eudemian Ethics
which included the books which it has in common with the Nicomachean E th ics' I f the note
is by Diogenes, it surely proves nothing about H erm ippus’ library (Düring, 1957, 67).
28 The Aristotelian Ethics The Aristotelian Ethics in Antiquity 29

Diogenes we assign the present passage, we have further evidence T w o other points elsewhere in Diogenes are more indicative.
for the ancient preference of the E E over the N E . Among Aristotle’s sayings he gives the definition o f friendship as a
single soul dwelling in two bodies.1 L ik e the saying in Favorinus,
D IO G E N E S L A E R T IU S this too seems to be an allusion to the E E . T h e N E does not figure,
in the same way, in the Life o f Aristotle. Indeed, Diogenes, like
In addition to the apophthegm from Favorinus the L ife o f Cicero, seems to have believed that the Nicomachean Ethics was the
Aristotle contains other material relevant to our problem. Diogenes’ work o f Aristotle’s son : for in his Life o f Eudoxus (viii.88) we read
summary o f the philosopher’s doctrines contains a section ( v .3 1- ‘Nicomachus, the son o f Aristotle, states that he declared pleasure
52) on his ethical teaching. P. M oraux (1949) made a detailed study to be the good’ ; 2 a reference to N E 10.2, i i 7 2 b9 ff.
o f this summary, along with the rest o f the outline in Diogenes, with Diogenes’ most valuable service is his preservation o f the ancient
a view to establishing its immediate sources. After comparing it list o f Aristotle’s writings. This, which raises more problems about
with the Aristotelian texts he came to the conclusion that it must the Ethics than it clearly solves, will be considered later on its own.
have been written by an unknown author between the time of
Andronicus and that o f Diogenes. But there is no reason to believe
A S P A S IU S
that in constructing his ethical summary Diogenes looked anywhere
else than to the epitome o f Arius D idym us. In structure and in From the study we have so far undertaken o f twelve ancient
detail o f vocabulary Diogenes’ summary follows A riu s1 and in its writers, the following conclusions emerge. Every writer, from
whole length there is nothing which demands any different source Aristotle him self until the second century A .D ., who shows a first­
except for three insignificant details. These details will not help us hand knowledge o f the Aristotelian ethical writings also shows a
to decide whether Diogenes (or any source he had other than preference for the E E over the N E : either in the sense that he refers
Arius) followed by preference the N E or the E E : for they are to it as The Ethics, sans phrase, or that he quotes it exclusively, or
details which occur in none of the treatises o f the Aristotelian that he prefers its doctrine, or its terminology, or its systematic
corpus.2 structure. N o author (with the exception o f the author of the Magna
M oraha) quotes the Nicomachean Ethics as ‘T h e Ethics’ ; and the
only two authors who show awareness o f an Ethics with that title
i. I will illustrate this from D iogenes’ first sentence:
both regard it as the work o f Aristotle’s son.
Diogenes Arius
τ έ λ ο ς δε εν ε ζ ε θ ε τ ο χ ρ ή σ ιv α ρ ε τ ή ς εν ε υ δ α ιμ ο ν ία ν δ 'ε ν α ι χ ρ ή σ ιν ά ρ ε τ ή ς τ ε λ ε ία ς
T h e comparative silence of antiquity about the N E comes to an
β ία ) τελ εία ), εφ η δ ε κ α ί τ ή ν ε υ δ α ­ εν β ίω τ ε λ ε ιω (130 , 19)· end with Aspasius, the author o f a line-by-line commentary on the
ιμ ο ν ία ν σ υ μ π λ ή ρ ω μ α ε κ τ ρ ιώ ν α γ α ­ It is a whole whose parts are σ υ μ π λ η ρ ω τ ικ ά Ethics which is the earliest o f all the surviving Aristotelian
θ ώ ν εΐνα ι- τ ώ ν π ε ρ ί ψ υ χ ή ν , ά δ ή κ α ί (U0-9)· commentaries.3 In Aspasius’ writing we find the situation with
π ρ ώ τ α τ ή δ υ ν ά μ ε ι κ α λ ε ι· ε κ δεύ ­ ε τ ι τ ώ ν ά γ α θ ώ ν τ ά μ ε ν ε ΐν α ι π ε ρ ί ψ υ χ ή ν , τα
τ ε ρ ω ν δ ε τ ώ ν π ε ρ ί σ ώ μ α , ύ γ ιε ία ς δε π ε ρ ί σ ώ μ α , τά δ’ εκτός, π ε ρ ί ψ υχήν μ ε ν , which we have been familiar for centuries : the Nicomachean Ethics
κ α ί ισ χ ύ ο ς κ α ι κ ά λ λ ο υ ς κ α ι τ ώ ν ο ίο ν ε υ φ υ ΐα ν τ ε κ α ί τ έ χ ν η ν κ α ί α ρ ε τ ή ν κ α ι is the undoubted treatise o f Aristotle, the Eudemian Ethics is the
π α ρ α π λ ή σ ιω ν ε κ τ ρ ίτ ω ν δε τ ώ ν σ ο φ ία ν κ α ίφ ρ ό ν η σ ιν κ α ίή δ ο ν ή ν π ε ρ ί σ ώ μ α
€/CTO?, π λ ο ύ τ ο υ κ α ί ε ύ γ ε ν ε ία ς κ α ί δ ε ύ γ ίε ια ν κ α ί ε υ α ισ θ η σ ία ν κ α ί κ ά λ λ ο ς κ α ί
problematic treatise whose attribution fluctuates, regarded now as
δ ό ξ η ς κ α ι τ ώ ν ό μ ο ιω ν . ίσ χ ύ ν κ α ί ά ρ τ ιό τ η τ α κ α ι π ά ν τ α τ α μ ό ρ ια authentic Aristotle, now as the work o f his disciple Eudemus. B u t if
σ υν τ α ΐς δ υ ν ά μ ε σ ι κ α ί ε ν ε ρ γ ε ία ις · έ κ τ ο ς δε Aspasius departs from the earlier tradition in his ranking o f the two
π λ ο ύ τ ο ν κ α ί δο^αν κ α ί ε ύ γ ε ν ε ια ν κ α ί δ υ ν α σ ­
τ ε ία ν κ α ί φ ίλ ο υ ς κ α ί σ υ γ γ ε ν ε ίς 1 κ α ι π α τ ­
1. μ ία ψ υ χ ή δυο σ ώ μ α σ ιν ε ν ο ικ ο ΰ σ α (cf. E E \2<\ο°2\ see D üring, 1 957 ^ 67 )·
ρ ίδ α . (13 6 .9 -16 ). 2. φ η σ ι δ ’ α υτο ί' Ν ικ ό μ α χ ο ς ό Ά ρ ισ τ ο τ έ λ ο υ ς τ ή ν ή δ ο ν ή ν λ ε γ ε ιν το α γ α θ ό ν .
2. T h e three points are ( i ) the statement that the goods o f the soul are π ρ ώ τ α τ ή δ υ ν ά μ ε ι, 3· T h e date o f Aspasius is uncertain: his floruit is generally placed in the first half o f the
(2) the definition o f friendship as equality o f reciprocal goodwill ( ισ ό τ η τ α ε ύ ν ο ια ς second century. It may well be that his commentary is earlier in date than the passages from
ά ν τ ισ τ ρ ό φ ο υ ), (3) the statement that the virtues are not mutually interdependent, which is Diogenes cited above, p. 28. Diogenes himself, according to R . D . H icks (19 25, xvi) ‘ has
in plain contradition to A E C , 114 4*30 ff. been variously dated in every century a . d . from the first to the fourth1.
30 The Aristotelian Ethics
Ethics, he is at one with it— as we shall see— in regarding the
disputed books as belonging essentially to the Eudemian version.
Unfortunately, all that survives o f Aspasius’ writing is com­
mentaries on N E i “ 4, part o f A E C , and part o f N E 8. These
commentaries are, o f course, in our editions treated as com­
mentaries on books 1- 4 , 7, and 8 o f the Nicomachean Ethics : but we
cannot be certain without further ado o f the original context o f his
commentary on C . I f the disputed books belonged originally with
the Eudemian Ethics, we cannot rule out in advance the possibility
that Aspasius’ commentary covered both ethical treatises, and that
we have four and a half books surving o f his commentary on the N E ,
and half a book surviving from his commentary on the E E . That is,
indeed, the hypothesis that some o f the evidence from his
commentary would suggest, though it may not be the one which,
taking the evidence as a whole, is the most plausible over-all
explanation. L et us examine the evidence in detail.
T h e most important evidence is contained in Aspasius’ com­
mentary on the following well-known passage in A E C.
I f certain pleasures are bad, that does not prevent the chief good from
being some pleasure, just as the chief good may be some form of
knowledge though certain kinds of knowledge are bad. Perhaps it is even
necessary, if each disposition has unimpeded activities, that, whether the
activity (if unimpeded) of all of our dispositions or that of some one of
them is happiness, this should be the thing most worthy of our choice; and
this activity is pleasure. Thus the chief good would be some pleasure,
though most pleasures might perhaps be bad without qualification.
( H 5 3 V 1 3 , trans· R o ss )
Commenting on the second o f these sentences, Aspasius writes :
These words make it look as if he was proving that the good and pleasure
are one and the same; but that is not so: what he is doing is—in a
dialectical manner against those who say that pleasure is a process or that
some pleasures are bad and draw the conclusion that it is not the good—to
try to show that it is possible to say that it is the supreme good. For in the
Nicomachean Ethics it is discussed, and there Aristotle says clearly about
pleasure that it is not the same thing as happiness but accompanies it like
the bloom on the cheek of youth. It is a sign that the present passage is not
by Aristotle but by Eudemus that in th e ........... he talks about pleasure as
if it had not yet been discussed ; but whether this text is by Eudemus or by
Aristotle, what is said is dialectical ; the reason that the greatest good is
The Aristotelian Ethics in Antiquity 31
said to be pleasure is that pleasure goes with the greatest good and cannot
be separated from it. (C A G xix. 15 1.18 -2 7 )1
T h is important passage has many puzzling features. It is commonly
taken as showing that Aspasius is in doubt whether the passage he is
commenting on— and, by implication, the disputed books in
general— belong to the Eudemian or to the Nicomachean version.
And this is certainly suggested by his reference to the fact that in the
tenth book o f the N E Aristotle talks o f pleasure as if he were
discussing it for the first time: this is a premiss which many people
since Aspasius have reasonably used as evidence that book seven is
Eudemian. But a closer reading o f the text suggests that his doubt is
not whether the present book belongs to the Eudemian Ethics, but
whether the Eudemian Ethics, to which in his view the present book
belongs, is by Aristotle or by Eudem us.2 When he says that
Aristotle discussed pleasure ‘in the Nicomachean Ethics'1 it is barely
credible that he takes himself to be at that moment commenting on
a book o f the Nicomachean Ethics', if he did so, surely he would have
to say ‘in the tenth book’ or ‘later’ . Aspasius is obviously
embarrassed by what Aristotle here says about pleasure, and is torn
between several ways o f dealing with it and resolving the apparent
discrepancy between the present passage and the passage that we
know as the tenth book o f the N E . One solution he investigates is
that there is no real inconsistency because the present passage is
1. δ ία μ € ν o v v τ ο ύ τ ω ν δ ο κ € Ϊ τ α ν τ ό ν ά π ο φ α ίν ζ σ θ α ι τ ά γ α θ ό ν κ α ί τ η ν η δ ο ν ή ν ο ν μ η ν
ό ν τ ω ς ê^€t, άλλα π ρ ο ς τ ο υ ς λ έ γ ο ν τ α ς γ ίν € σ ιν e i v a i ή φ α ύ λ α ς τ ιν ά ς τ ώ ν η δ ο ν ώ ν , ο ΐς
κ α ι δ ι’ α υ τό τ ό μ ή € ΐν α ι α υ τ ή ν τό α γ α θ ό ν € π ιγ ίν € τ α ι[ κ α ΐ] , € π ιχ € ΐp e f ^ δ ο £ ω ? ώ ? kvov
α ν τ η ν το ά ρ ισ τ ο ν Aeyety, έ π ζ ι 'kv y e τ ο ΐς Ν ικ ο μ α χ € ΐο ις , ένθ α δ ΐ€ ΐλ η π τ α ι κ α ι π € ρ ι
η δ ο ν ή ς ’Α ρ ισ τ ο τ έ λ η ς σ α φ ώ ς € ΐρη κ€ν α υ τ ή ν μ ή τ α ύ τ ό ν ζ ιν α ι τ ή € υ δ α ιμ ο ν ία αλλά
7τα ρ α κολο υ0 €Ϊν ‘ ώ σ π € ρ τ ο ΐς ά κ μ α ίο ις τ ή ν ώ ρ α ν \ σ η μ € ΐο ν 6k τ ο ύ μ ή € iv a i t o u t '
Ά ρ ισ τ σ τ έ λ ο υ ς α λλ ’ Ε ύ δ ή μ ο υ το kv τ ώ ............ λ έ γ € ΐν π € ρ ί ή δ ο ν ή ς ώ ς ο ύ δ έ π ω π ζ ρ ί
α υ τ ή ς δ ΐ€ ΐλ € γ μ έ ν ο υ ' π λ ή ν €ΐτ€ Ε ύ δ ή μ ο υ τ α ν τ ά k a n v eiVe Ά ρ ι σ τ ο τ ί λ ο υ ς , ζ ν δ ό ξ ω ς
ζ ϊρ η τ α ι· δ ιά τ ο ΰ τ ο λ έ γ € τ α ι τό ά ρ ισ τ ο ν η δ ο ν ή , δ τ ι σ νν τ ώ ά ρ ίσ τ ω κ α ί ά χ ώ ρ ισ τ ο ν
α ν τ ο ΰ . T h e lacuna o f approximately six letters where one would expect the number o f the
N E book which treats o f pleasure is interesting. T h e number which we would expect, and
which a later tradition and H eylbut obviously did expect, is ‘tenth’ . T h e lacuna may mean
that the number which was actually in the manuscripts was not this : which would support the
thesis that the original place o f the disputed books was not in the context o f the N E .
2. O f course the ‘t o u t ’ ’ which he is prepared to consider as being not by Aristotle but by
Eudemus need not be the whole disputed books o f the A E , but a more limited context o f the
quotation: e.g. the treatment o f pleasure. B ut (a) within the text o f Aspasius him self there is
no indication o f a break between the treatment o f pleasure and the rest o f A E book C ; (b)
there is no indication, independent o f the present passage, o f a Eudemian origin for the
treatment o f pleasure here other than the fact that the whole o f the A E is by the manuscript
tradition regarded as belonging also to a work which some scholars have regarded as written
by Eudemus rather than by Aristotle.
 The Aristotelian Ethics
32
only dialectical; another is that the inconsistency is not surprising
because one treatment is by Aristotle and the other is by Eudemius.
But having offered the fact that the two treatments o f pleasure
ignore each other as independent evidence that the current one is by
Eudemus he then goes on to say that even if it is by Eudem us it is
consistent with the Aristotelian treatment because it is only
dialectical.
I f we reject as incredible the traditional suggestion that Aspasius,
at the moment when he talks o f ‘ the Nicomachean Ethics’’ as a
different work, was commenting on a text o f the N E which included
the disputed books, there are two solutions which can be offerred of
the problems set by this difficult passage. One possibility is that
Aspasius commented on both the Nicomachean and Eudemian
Ethics, and that the partial commentary on book seven is a fragment
o f his commentary on the E E just as his partial commentary on book
eight is a fragment o f his commentary on the N E . T h e weak point of
this hypothesis is the following: if Aspasius regarded the two
treatments o f pleasure as clearly belonging to two different works,
why should he be surprised to find no cross-reference? Other
parallel treatments o f topics in the two Ethics are not signalized in
either Ethics, so why should the absence o f a reference here be taken
as an argument that the E E is by Eudem us?
There is another solution which seems to me the more plausible
one. Th at is, that Aspasius was writing a commentary on a text of
the N E which did not contain the disputed books, but which
contained, as our texts o f N E 1 - 4 do, forward references to
treatments o f topics such as justice, wisdom, and continence. There
existed in his time— as is surely most likely from the whole history
we have hitherto investigated— texts o f the Eudemian Ethics
containing the common books ; but their title gave Aspasius some
reason to believe that they were not by Aristotle but by Eudemus,
just as the title o f the Nicomachean Ethics gave Diogenes and
perhaps Cicero reason to believe that the N E was not by Aristotle
but by Nicomachus. I conjecture that in these circumstances
Aspasius, having commented on books 1 - 4 o f the N E , proceeded to
comment on books I V - V I o f the Eudemian Ethics, in order to have a
full commentary on Aristotle’s ethical system, in spite o f the lacuna
in the Nicomachean version.
T h is can be no more than a conjecture: but it would explain all
the features o f the puzzling passage o f Aspasius in a very simple
The Aristotelian Ethics in Antiquity 33
manner. F ir s t,,it explains why he refers to the N E as if it was a
different work from the one he is currently commenting on.
Second, it explains why, in spite o f this, he comments on the lack o f
cross-reference between the two treatments o f pleasure as surpris­
ing. F o r if the only books o f the E E he commented on were the
disputed books, the passage about pleasure, being the only clear
doublet with the N E books, would be the most obvious place to
remark on the fact that the N E and E E apparently make no cross-
references to each other. T h is is indeed surprising, given the
general frequency in Aristotle of cross-references to treatments o f
similar topics in other works (e.g. to the Analytics)·,1 it is indeed
something from which a scholar might be tempted to draw
conclusions about the authenticity or spuriousness o f a work.
Obiously, this is no more than conjecture, as has been said. I f
Aspasius deliberately moved the three disputed books into the
Nicomachean Ethics from the Eudemian for purposes o f commenting
on a unified and comprehensive treatment o f Aristotle’s ethics, the
place where he would have to announce this would be the beginning
o f his commentary on the first disputed book, N E 5 = E E IV .2 But
his commentary on book 5 is unfortunately lost. But there are
several other features o f the surviving parts o f the commentary
which go some way to confirm our conjecture.
First, a passage in the commentary on N E 8 places it beyond
doubt that Aspasius regarded the N E as containing a lacuna that
needed filling. Aristotle is critizing those who argue that there is
only one kind o f friendship on the ground that friendship admits o f
degrees. ‘Things different in kind also admit o f degrees,’ he says,
‘But these matters have been discussed before’ ( 1 1 35b 13 16).
1. It has been fashionable to treat the cross-references as untrustworthy and possibly
interpolated. D üring (1968, 19 1) is rightly critical o f this fashion: ‘D ie zahlreichen Vor-,
Querund Riichverweise in seinen Schriften stimmen erstaunlich gut; höchstens ein Zehntel
der Verweise sind problematisch. Ein später Redaktor hätte nie ein so feinmaschiges Netz
von Verbindungen zwischen den einzelnen Schriften hersteilen können.’
2. Earlier scholars who attributed the A E to Eudem us thought they could detect the
suture in Aristotle’s own text. Grant regarded as spurious the last two sentences o f N E 4.
(‘Continence is not a virtue either, but an intermediate state : T h is will be explained later. We
must now discuss justice.’ ) ‘Aristotle’s M S . o f the fourth book having ended abruptly at the
word ε π ιε ικ έ ς Nicomachus, or the editor, whoever he was, in all probability added these
clauses in order to give the book a seeming unity with the three Eudemian books which were
now to be grafted on’ (1885, ii.94). Aspasius’ commentary on N E 4 ends a column earlier at
i i 2 8 a3 3 ; the anonymous second-century commentary has a curious comment at this point
(C A G xix.126, X X . 204; see below, p. 37). B ut there is no manuscript support for the
excision.
34 The Aristotelian Ethics
Aspasius comments: ‘ “ T h ey have been discussed” he says
“ before” . It looks as if they have been discussed in the lost parts o f
the Nicomachean Ethics' (C A G x ix .16 1) .1 N ow o f course we cannot
be certain that by ‘ the lost parts o f the Nicomachean Ethics' Aspasius
means the lacuna it displays if it is deprived o f the common books.
But such is certainly the simplest explanation o f his remark : that
lacuna is the only lacuna for which there is independent evidence in
the anomalous nature o f the manuscript tradition.2 I f this is the
lacuna that Aspasius means, we cannot tell whether he knew as a
fact that there were once other Nicomachean books where the A E
books now stand, or whether he merely conjectured that there must
have been because o f the forward and backward references in the
N E itself.
Secondly, an examination o f Aspasius’ method o f citation from
one part o f the Ethics to another suggests that he regarded the A E as
forming a block separate from that o f the undisputed N E . When he
makes a forward or back reference within a single book, or from one
undisputed N E book to another, he uses an expression such as ‘ as
was said before’ or ‘later’ .3 So too when he refers, within his
commentary on book C, to the disputed books.4 But when he quotes
the disputed books from the commentary on the N E , he uses
different modes o f expression: ‘in the book on justice’ (16 0 .11)
‘elsewhere’ (178.22, a reference to book A ; 18.6, a reference to a
number o f texts including book B as well as N E 1 ).5 Sim ilarly, when
referring from the disputed book C to the undisputed Nicomachean
books, Aspasius says ‘elsewhere’ (138.22, reference to N E 4) ‘ in the
Nicomachean Ethics' ( 15 1.2 2 ) .6 T h e discrepancy is sufficiently
systematic to suggest that it is deliberate.7 ‘Elsew here’ , the
1. ε ϊρ η τ α ι δ ε , φ η σ ί, π ε ρ ί α υ τ ώ ν έ μ π ρ ο σ θ ε ν , ε ο ικ ε δ ε €ΐρΎ}σθαι εν τ ο ΐς ε κ π ε π τ ω -
κ ό σ ι τ ώ ν Ν ικ ο μ α χ ε ω ν .
2. I agree in this, against Rose ( 18 7 1, 107), with M oraux and Harlfinger (Harlfinger, 1972,
45 - 7)·
3. π ρ ό τ ε ρ ο ν is used thus at 19 .2 5 ; 1 2 1 . 1 5 ; 119 .2 7 ; 83.4; 88.2; 1 16 .3 ; 86.9; 1 0 5 .14 ; 53.4 ;
10 1.8 ; 10 6 .15 ; 168.4; 174.2. ύ σ τ ε ρ ο ν is used in a similar way at 19 .2 ; 23.29.
4. So for references within b o okC , and in the two references back from C to B , 13 6 .3 1 and
140.2 and 30.
5. ev ά λλ ο ι? ( 17 8 .2 2 ; i8 .6 ); ev τ ο ΐς π ε ρ ι δ ικ α ιο σ ύ ν η ς ( ιό ο .ιι) .
6. ε ν ά λλ ο ι? (138 .22) ε ν τ ο ΐς Ν ικ ο μ α χ ε ίο ις (15 1.2 2 ).
7· O f thirty-odd cross-references in Aspasius I have found only one that is a real exception
to the rule that internal citations are introduced differently from citations across the N E - A E
boundary. T h at is 8.30, where, discussing an aporia in book N E 1 Aspasius says that Aristotle
solves it ττροϊώ ν. T h e reference appears to be to a passage in B ( 1 1 45s 10). T h is, unlike the
other cross-rcfcrences in Aspasius, suggests that he read B as part o f a text continuous with
the N E .
The Aristotelian Ethics in Antiquity 35
expression used to refer across the boundary between the disputed
books and the rest o f the N E , is also used in a case where, from the
commentary on the N E , Aspasius refers to the E E }
T hirdly, with regard to both the A E and the E E Aspasius is
undecided whether to ascribe authorship to Aristotle or to
Eudemus. In the case of the A E we have already looked at the
famous passage in which he assigns at least the section on pleasure
to Eudemus ; at 2.24 on the other hand Aristotle is given credit for
the definition o f technëin A E A and at 18.6 the psychology o f book
B (and elsewhere) is attributed to Aristotle, and at 178.22 in the
commentary on N E 8 Aristotle is referred to as the author o f the
treatise on justice, which is presumably A E A .2 Sim ilarly, there is
vacillation about the authorship o f the Eudemian Ethics. At 11.5
Aristotle is cited as the author of the Eudemian view o f riches, and
at 132.3 (in a section o f C where Aristotle is treated without
question as the author) the Eudemian account o f the insensible man
is quoted.3 B u t in the commentary on N E 8 a quotation from the
seventh book o f the E E is attributed to Eudem us.4
T o summarize, therefore, the evidence o f Aspasius. It seems
clear that Aspasius regarded the disputed books as belonging to the
E E for the following reasons : (1) In his commentary on C he refers
to the N E as a different work. (2) H e is certain that the N E is by
Aristotle, but is prepared to consider the possibility that C is by
Eudemus, just as he is unsure whether the E E is by Aristotle or
Eudemus. (3) With one exception, he quotes the A E when
commenting on the N E , and the N E when commenting on the A E ,
1. At 1 1.5 Aspasius says that the remark in the N E that riches are desirable άλλου χ ά ρ ιν
does not fit his remarks εν α λ λ ο ίς that they are good in themselves. T h is appears to be a
reference to the E E , where several times ( i 2 i 4 b8, 12 17 *3 7 , Ι22 7“ · 4ι Ι2 32ΐ>1°> i 248b28) riches
are given as an example o f an ultimate end.
2. And o f course throughout the text o f C itself is attributed to Aristotle (13 3 .3 0 ; 136 .7;
13 8 .19 ; 150-5)·
3. ό δ ε ε λ λ ε ίπ α ιν π ε ρ ί τ ά ς ή δ ο ν ά ς κ α ί μ η τ α ΐς ά ν α γ κ α ία ις χ ρ ώ μ ε ν ο ς ό
ά ν τ ικ ε ιμ ε ν ο ς τ ω ά κ ο λ ά σ τ ψ , ον εν τ ο ΐς κ α τ ’ ά ρ χ ά ς λ ό γ ο ις ά ν α ίσ θ η τ ο ν ε λ ε γ ε . T his
reference is clearly to E E I I I .2, 12 31*26 ff. : α ν α ίσ θ η τ ο ς μ ί ν οΰν . . . ό ο ύ τ ω ς ε χ ω ν ώ σ τ ε
κ α ι ε λ λ ε ιπ ε ιν ο σ ω ν α ν α γ κ η κ ο ιν ω ν ε ΐν ώ ς επ'ι τ ο π ο λ ν π ά ν τ α ς κ α ι χ α ίρ ε ιν . T h e N E
does not link the insensible man with the notion o f necessary pleasures, and is indeed
sceptical o f his very existence ( n o 7 b28; 1114 * 10 ). T h is quotation is very interesting, for it is
introduced in C by the expression εν τ ο ΐς κ α τ ' ά ρ χ ά ς λ ό γ ο ις which is used by Aspasius
(162.3) and by Aristotle ( n o 4 a2 ; ii4 g b27) to refer to earlier parts o f the same work. It is thus
another piece o f evidence that Aspasius regarded the A E as belonging to the E E .
4. λ έ γ ε ι δε κ α ί Ε ύ δ η μ ο ς κ α ί Θ εό φ ρ α σ το ς , ό τ ι κ α ί α ΐ κ α θ ' υ π ε ρ ο χ ή ν φ ιλ ία ι ε ν τ ο ΐς
α ύ τ ο ΐς γ ίν ο ν τ α ι, ή δ ι' ή δ ο ν η ν ή δ ια το χ ρ ή σ ιμ ο ν ή δ ι' α ρ ε τ ή ν . T h e reference is to E E
V II.4 , 1239*1 ff. (C A G x ix .178.3).
 36 The Aristotelian Ethics
with the style o f quotation he uses for introducing quotations from
other works, rather than that which he uses for forward and
backward references in the same work. (4) From the A E he quotes
the E E with a form o f words he uses for backward reference in the
same work. (5) H e speaks o f the existence o f a lacuna in the N E as a
familiar fact.
We shall see that after the time o f Aspasius the common books
are normally taken with the N E : the earlier tradition leaves as its
only trace— other than the presence o f the A E in the manuscripts of
the E E — a continuing confusion about the numbering o f the books
o f the N E . I f we are right in our conjecture that the Aspasian
evidence is best explained by postulating that he commented on the
three E E - books in the Nicomachean context in order to fill a lacuna
in the N E , then it may very well have been Aspasius’ own activity as
a commentator which inaugurated the tradition o f counting the
disputed books with the Nicomachean treatise. I f that is so, the
Nicomachean Ethics as we know them— N E plus A E — might be
more aptly termed the ‘Aspasian Ethics’ .
A T T IC U S
Eusebius (Praeparatio Evangelica X V .4) quotes a comment by
the Platonist Atticus on the worthlessness o f Aristotle’s three
ethical treatises: the Eudemian, Nicomachean, and Great Ethics.1
Atticus was active in the time of M arcus Aurelius (16 1-8 0 ): he is
thus the earliest testimony to the three traditional titles o f the
treatises o f the Aristotelian corpus. H e is also useful in the present
context as determining the date o f the anonymous scholia on books
2 -5 of the N E , which we shall next consider.
T H E S C H O L IA S T
T h e scholiast on book 5 of the Nicomachean Ethics ( C A G xx.
248.24-6) is discussing the idea that some things are indifferent
(adiaphora) : he mentions its proponents ‘among whom previously
Aristonymus belonged, but now even some who pretend to be
Platonists, while surreptitiously introducing this doctrine, among
whom also Atticus appears to belong’ . T h is passage, noticed by
I. α ί γ ο νν Ά ρ ισ τ ο τ έ λ ο υ ς π ε ρ ί ταντα πρα γματείαν Ε ύ δ ή μ ε ιο ίτ ε κα ί Ν ικ ο μ ά χ ειο ι
κ α ι Μ ε γ ά λ ω ν Η θ ι κ ώ ν ε τ τ ιγ ρ α φ ο μ ε ν α ι μ ι κ ρ ό ν τ ι κ α ι τ α π ε ι ν ό ν κ α ί δ η μ ώ δ ε ς π ε ρ ί
α ρ ε τ ή ς φ ρ ο ν ο ν σ ι.
The Aristotelian Ethics in Antiquity 37
M ercken,1 indicates that the writer o f the scholia was a con­
temporary o f Atticus and should therefore be assigned to the second
half o f the second century. T h e commentary on N E 5 appears
continuous with those on N E 2 -4 and is written in very much the
same style : the writer clearly regarded N E 5 as belonging with N E
1- 4 , and refers in the first line o f his commentary on 5 to the
discussions o f akribeia preceding in the N E . He refers forward to B
from an earlier book, and cites it by its Nicomachean number ( C A G
x x .12 5 .3 0 ); on the other hand, at the end of book 4 he appears to
promise a discussion o f continence by Aristotle in the book after
next, rather than as we would expect after two intervening books.2
In general, this writer witnesses to the Nicomachean tradition as
familiar in later centuries.3
A L E X A N D E R O F A P H R O D IS IA S
In the writings o f Alexander, who began lecturing at Athens
c. 198, there is no doubt o f the existence o f the ten-book
Nicomachean Ethics with which we are familiar. Alexander did not
write a line-by-line commentary like Aspasius; but both in his
commentary on the Metaphysics and in his minor ethical writings he
refers to the disputed books at belonging to the Nicomachean
Ethics,4
'· ' 973, 15*·
2. ύ σ τερ ο ν δ ε ε π ί π λ έ ο ν έ ρ ε ΐ π ε ρ ί έ γ κ ρ α τ ε ία ς κ α ί κ α ρ τ ε ρ ία ς κ α ί μ α λ α κ ία ς μ ε τ α το
έ ξ η ς β ιβ λ ίο ν ( C A G χχ,204)·
3· We can make a guess at the identity o f this scholiast. We know from Athenaeus
(X V . 673 e) that Adrastus o f Aphrodisias » rote five books on Theophrastus’ ethics and six (or
a sixth) on Aristotle’s. S ix books must have been the extent o f the second-century scholia on
the N E , since the popular composite medieval commentary published in C A G xx appears to
have been built up on the basis o f the scholiast’s work, replacing particular books by works of
more prestigious Christian writers as these became available, as in the case ofE ustratius for
book i. (T he trumping process can actually be observed in the N ew College manuscript of
the commentaries in which both the anonymous scholia on A E A and the preferred
commentary o f M ichael o f Ephesus have been preserved.) T he composite commentary,
however, has to rely on a late anonymous author for book 7, suggesting that the original
scholiast had here neither commented nor been trumped. Som e may find it too much of a
coincidence to have both a second-century six-book commentator on the Ethics whose
commentary has been lost and a second-century six-book commentary on the Ethics whose
author has disappeared. In addition to the reference to Atticus the internal evidence o f the
scholia suggests a second-century date. T h e latest author quoted is L u cian ; there is no
evidence o f Christian interest with the possible exception o f the three words ο ύ τ ω ς κ α ι
Ά β ε σ α λ ώ μ (229.22) at the end o f the story in book 5 o f Phoenix’s affair with his father’s
concubine. M ercken (1973, 15 *) is probably right to see these words as a late gloss, added at
the very end o f a paragraph.
4. In his commentary on the Prior Analytics ( C A G ii. 1.8 .3 1) he refers 10 hook 1 o f the N E
and then refers to there being ten books in that work. In commenting on the Metaphysics
38 The Aristotelian Ethics
C L E M E N T O F A L E X A N D R IA
In his ‘miscellanies’ , which are commonly dated to the very
earliest years o f the third century, Clement draws freely on both the
Eudemian and the Nicomachean Ethics, quoting the Eudemians
without hesitation as Aristotle’s work. His method o f utilizing his
sources makes it difficult to draw any conclusions about the position
o f the disputed books in the texts available to him .1
LATER CO M M ENTATO RS
N o commentary on Aristotle’s Ethics survives from the period
between Alexander and the Byzantine renaissance. In the second
h alf o f the third century Porphyry wrote a commentary which is
known only through Arabic sources. Scholars assume that it was a
commentary on the Nicomachean Ethics, but there is no decisive
evidence for this; all we know is that the commentary was on a
twelve-book ethics (or was itself in twelve books).2
A study o f citations o f the Ethics by commentators on Aristotle
(C A G 1.7 .13) he refers to B as book 6 o f the Nicomachean Ethics ; the commentary on the
Sophistici Elenchi in his name ( I I .3.172.20) refers to A as book 5 o f the N E . In his
commentary on the Topics he quotes the N E a dozen times, but without specific citations. H e
does not appear to have quoted the E E . T h e Ethical Questions attributed to him on doubtful
authority discuss passages o f the N E at length, and refer twice (Suppl. A r. II. 12 7 .2 9 ; 134.30)
to book C as N E 7. T h e picture is not entirely clear, however, for in A lexander’s second
treatise De Anima there are two passages which suggest an eleven-book Nicomachean Ethics
(Suppl. A r. II). A t 1 5 1 . 1 1 a reference to N E 8 is described as occurring kv τ ώ iv a r w , nine
lines further on a reference to N E 9 is described as èv τ ώ δ € κ α τ ώ , and in 15 3 .14 a quotation
o f E E 10 .3, i i7 5 aio is introduced with the words A eyet Sk kv τ ώ κ τ ώ ν Ν ικ ο μ α χ ε ίω ν .
T h is work may be however considerably later than Alexander (see Gauthier, 1970, 10 1).
1 . T hu s he makes use o f the quotation which introduces the E E and occurs later in the N E y
at Strom . 7 .1 2 ; he refers to the etymology o f σ ω φ ρ ο σ ύ ν η from the common book B at 7.3,
and the reference to the stubbornness o f m ere belief in book C at 7 .16 . T here is a long
development, without explicit acknowledgement o f Aristotle, o f the passage in N E 3
concerning ignorance o f circumstances.
A quotation from the E E is introduced with an explicit mention o f Aristotle in the
following passage: κ α ν ο ’Α ρ ισ τ ο τ έ λ η ς ά ν α λ ο γ η > τ ό μ € ν π ο ιε ΐν κ α ι k n i τ ώ ν α λ ό γ ω ν
ζ ώ ω ν τ ά σ σ € σ θ α ι κ α ι €7τι α ψ ύ χ ω ν δ ιδ ά σ κ ω ν , τ ό δ έ ττρά ττ€ ΐν ά ν θ ρ ώ π ω ν e iv a t μ ό ν ω ν
(cf. Ε Ε 2.6, I2 2 2 b20; 2.8, i2 2 4 a28).
2. Porphyry’s commentary is mentioned by the Arab bibliographer Ibn al-Nadim ( a .d .
987); see M iiller, 1873, an^ Walzer, Greek into A rabic, 220 if., 240. T h e commentary
certainly included discussions o f N E 1 and N E 10, as we know from the quotations in other
Arabic writers: see G hourab, in Stern, H ourani, and Brown, 1972, 78, on al’A m iri. From
al’ A m iri we know also that Them istius, the secretary o f Ju lian the Apostate, commented on
the passages on practical reasoning in book B o f the A E (ibid. 83-7).
The Aristotelian Ethics in Antiquity 39
between the third and tenth centuries yields rather inconclusive
results.1
A late witness to the uncertainty about the authorship and
content o f the Nicomachean Ethics is contained in the entry in the
Suda on Nicomachus, which is said to be likely to derive from
Hesychius (i. 5th cent.). There we are told that Nicomachus wrote
six books o f Ethics as well as a commentary on his father’s
Physics. Some commentators regard this as a confusion between
Nicomachus and Eudemus, who did write a commentary on the
Physics, and whose name is associated with the E E (some
manuscripts o f which lump books V II and V I I I together). Possibly
so: but equally possibly the six-book Ethics might be the
Nicomachean Ethics less the disputed books; there is one book short
in either case.2
T H E C A T A L O G U E S O F A R IS T O T L E ’ S W R IT IN G S
Three ancient lists o f Aristotle’s works have come down to us.
T h e first, in the Life by Diogenes Laertius (\\22--7) is commonly
held to have been made by Herm ippus, librarian o f the library o f
Alexandria, who wrote a life of Aristotle in about 200 B .C . T h e
second is in the Life o f Aristotle by Hesychius (the Vita Menagiana)
o f the fifth century. T h e third is the list o f Ptolemy which is
preserved in two Arabic versions in the Lives o f Aristotle by al-Qifti
(d. 1248) and by Usaibia (d. 1270).3 We have already referred to the
third list in discussing Andronicus : we must now consider the other
two.
T H E L I S T O F H E R M IP P U S
T h e thirty-eighth item in the Hermippus list is an Ethics in five
books ; no other Ethics is listed. M any scholars have seen this five-
1 . Philoponus in the fifth century quotes both the N E (C A G xv i.3.19 ) and the E E (C A G
xvi.237.26, 335.22) as sim ply ‘the Ethics'. H e quotes 1139*8, from A E B , as being from the
Nicomachean Ethics (xiv.3.84.15). Sim plicus in the sixth century quotes the N E and E E both
by name (v iii.4 .2 7,6 .14 , 12 .8 ,17 0 .6 ). A to n e point he attributes to the N E a distinction made
only in the E E in our manuscripts (x.303.32 ff.).
2. έ γ ρ α φ ε ν ’ Η θ ι κ ώ ν β ιβ λ ία ϊ ' κ α ι π ε ρ ί τ η ς φ υ σ ικ ή ς ά κ ρ ο ά σ € ω ς τ ο υ π α τ ρ ό ς α ν το ΰ .
T h e passage is printed in D üring, 1957, 265, who conjectures that it has been transferred
from a notice o f Eudemus.
3. M oraux attributes the list in Diogenes to Ariston, scholarch o f the Peripatos in the last
quarter o f the second century ( 19 5 1, 2 37 ); but this attribution has not been generally
accepted: see D üring, 1956, 1 1 - 2 1 ; K earney, 1972. T h e list o f H esychius is printed in
Düring, 1957, 8 3 - 9 ; the most accessible reproduction o f the Ptolemy list is likewise in
Düring, 2 2 1 - 3 1 ; see also M oraux, 1951, 2 8 9 -3 2 1.
40 The Aristotelian Ethics
The Aristotelian Ethics in Antiquity 41
book Ethics as being the E E without the disputed books.1 It may be Ethical works appear to be misplaced. One would expect them to be
so, and if it is so, it is further confirmation o f the general thesis that gathered together in the section devoted to practical philosophy,
the E E , during the centuries immediately succeeding Aristotle’s immediately before the political works, say between items 73 and 74
death, was better known than the N E . It would, however, suggest o f the list. In fact they are scattered among the logical and
that the disputed books did not belong with the E E , and this would dialectical works in pairs, so that the first part o f the list presents the
clash with the evidence we have seen and the evidence to be following appearance:
presented later in this book. Other scholars have stressed that the
division o f the Ethics into books may well be later than the catalogue 1- 2 2 Popular and Platonic works
o f Herm ippus, so that the five books may not correspond to any o f 23 Economicus
our present books. T h us Margueritte, reviewing Jaeger in 1930 24 On Friendship, one book
(p. 104) suggested that the title in the catalogue represented the 2 5 -3 6 Logical and dialectical works
complete Eudemian Ethics in five books made up from E E I (1), E E 37 On the passions o f anger, one book
II and I I I (2), A E A (3), A E B and C and E E V I I I (4), E E V II (5) : 38 Ethics, five books
the topics o f the five books being happiness, virtue, justice, wisdom, 39-52 Logical and dialectical works
friendship respectively.2 53 On what is better, one book
A close inspection o f Diogenes’ list permits a conjecture about 54 On the idea (of the good), one book
the five-book Ethics which is simpler than M argueritte’s, and which 55-67 Logical and dialectical works
is more in accord with the evidence o f Aspasius, and the internal 68 On the voluntary, one book
evidence we shall shortly examine, than the common opinion which 69 On the noble, one book
sees in Diogenes’ list a reference to the E E without the disputed 70-3 Dialectical works1
books. From 74 the practical works begin, and the disturbed pattern ends.
Before proposing it, I must draw attention to some features o f the T h e fact that the ethical works occur thus spaced in pairs, at more or
ordering of the titles in the list.3 T h e list begins with some two less regular intervals (from the beginning the intervals are 2 x 1 1
dozen popular and Platonic works. There follow about four dozen titles; 12 titles; 13 titles; 12 titles) suggests that they originally
mainly logical and dialectical works. N ext come sixteen works on belonged together and have been misplaced either owing to an
practical philosophy (politics, rhetoric, poetry). Then, twenty accident in the manuscript tradition or to some feature o f the library
works on physics (including ten on the history of physics) and on whose catalogue is most likely at the basis o f Herm ippus’ list.2
biology, plus half a dozen works on mathematical topics including I f this is so, then inter alia the three titles O n friendship’, O n the
optics and musical theory. T h e remaining thirty titles are passions o f anger’, and ‘Ethics, five books’ may be taken together.
‘hypomnematic’ works, collectanea, letters, and poetry. T h e most
interesting feature o f the ordering, in the present context, is that the 1. T h e Greek titles o f the Ethical works are: Ο ικ ο ν ο μ ικ ο ί α, π ε ρ ι φ ιλ ία ς a, π ε ρ ί
π α θ ώ ν ο ρ γ ή ς â , Η θ ι κ ώ ν α β γ ά ί, π ε ρ ί τ ο ν β ε λ τ ίο ν ο ς α, π ε ρ ί τ ή ς ιδ έ α ς ä, π ε ρ ι
εκ ο υ σ ίο υ <χ, π ε ρ ί κ α λ ο ύ ä. Several points in m y list call for comment, (t) 1 have not counted
1 . T h u s Jaeger, 1948 ; R oss, 19 23 ; M oraux, 19 5 1. D üring, in the apparatus to his edition
34— π ρ ο τ ά σ ε ις π ε ρ ί α ρ ε τ ή ς ä ß — as an ethical title, because π ρ ο τ ά σ ε ις on all topics seem
o f the list (19 57, 43) says : ‘Potest esse editio Ethicae Eudemeae, sine libris D E Z et cum libro
to be regarded by the compiler as dialectical (see 4 5 -7 ,6 6 -7 ). (2) In counting I have followed
H in duos libros ( 1 - 5 , 6 - 1 5 ) distributo, sed mera coniectura est.’ Titze, 1825, thought that
D üring ( 19 5 7 ,4 4 -5 ) in regarding 45-6 and 66-7 as making a single title. (3) I assume that the
the five-book Ethics was the first five books o f the N E ; he sought the remaining books o f that
Idea mentioned in 54 is the idea o f the Good. What other idea deserves the title o f ‘ The Idea’ ?
Ethics under other titles, and located the E E as items 20 (v e p i τ ά γ α θ ο ν α 'β 'γ ) and item 24
(So Titze, 1825, 38.) (4) It is not clear whether the practical section o f the list should be
(7τ ςρ ϊ φ ιλ ία ς·, ά) o f D iogenes’ list.
regarded as commencing at 68 (π ε ρ ί ε κ ο ν σ ίο ν ) or at 74 (π ο λ ιτ ικ ά β I. T h e reason is that
2. In support o f this M argueritte appeals to ‘l’annonce que fait V E E au début du L iv re I
7 0 -3 are all θ έ σ ε ις and therefore dialectical; but are, in at least three o f the cases, about
des deux principaux sujets à traiter, Γ ά ρ € τή et le φ ρ ό ν η ο ις , et aux réferences ou aux
more or less ethical topics. See below, p. 45.
transitions que nous lisons ailleurs’ (1930, 104).
2. M oraux observed that the ethical titles 24, 37, and 38 were out o f place (19 5 1, 190); he
3. On the ordering o f the list, see M oraux 19 5 1, 190, and D uring 1957, 68-9.
has an ingenious reconstruction o f the pinax to explain this and other displacements.
42 The Aristotelian Ethics
In the light o f this, I conjecture that the five-book Ethics is neither
the truncated E E minus the A E , nor the hypothetically recon­
structed Eudemian Ethics o f M argueritte, but the first five books o f
the Eudemians as they are in the manuscript tradition, i.e. E E I —I I I
followed by A E A and A E B. These books read as a consecutive
whble, each leading naturally into the next and forming as a totality
a treatise on the virtues. Book C o f the common books makes a fresh
start and applies a new scheme for the classification o f character to
the material discussed in earlier books. W hile its doctrine is
consistent with, and is at many points linked with, the earlier
Eudem ian books, it is capable o f being studied separately as a
treatise on continence and incontinence and pleasure, and in
general the influence o f passion on action. Book V II o f the E E (with
or without the chapters which in some manuscripts are treated as a
separate book V III) is also capable o f standing alone as a treatise on
friendship.
I f we are right to see the five-book Ethics as E E I V , it is natural
to look for signs o f E E V I and V I I elsewhere in the list.1 N ext to it in
the list is the title O n the passions o f anger’ . T h e plural ‘passions’ is
very strange, and all editors emend the text. Rose altered the text to
O n the passions or on anger’ ; D üring thought that two titles O n
the passions’ and ‘on anger’ had been telescoped together.2 A much
simpler emendation than either o f these is possible. T h e change o f a
single letter in the Greek would make the title O n the influence of
the passions’ .3 T h is would be an appropriate title for E E V I (the
common book C) which is mainly devoted to incontinence.4 N ext to
this title, if the ethical works were originally grouped together
without the intervening dialectical dozens, we read O n
friendship’— the obvious title for E E V II. Th us, these three
adjacent titles would yield our Eudemian Ethics complete. N ext to
them is the Economicus. I f this is the first book o f our Oeconomica, as
many scholars believe, it would go naturally with the E E as a link
1 . T h e several books o f the Topics apparently appear scattered in the list under separate
titles; cf. M oraux, 19 5 1, 54 ff.
2. R o se: περι, π α θ ώ ν ( ή π ε ρ ι) ο ρ γ ή ς ; D ü rin g: π ε ρ ι π α θ ώ ν *(α π ε ρ ι) ο ρ γ ή ς à.
3· π ε ρ ί π α θ ώ ν ό ρ μ ή ς .
4· I am not o f course suggesting that the title is Aristotle’s own (though ό ρ μ ή , i i0 2 b2 i ,
h i 6b30, is a word he uses in this sort o f context) nor that he did not him self regard the E E as
a complete and unified whole. I see from Bodéus (1974, 459) that M asellis (19 56 , 348) also
identifies this work with A E C . N ot having had direct access to M asellis’s article I do not
know whether he supports his conjecture with a sim ilar emendation.
The Aristotelian Ethics in Antiquity 43
between that work and the eight books o f our Politics, listed at 75 in
the portion o f the list succeeding the disturbance o f the order.
All this o f course is conjecture. I make no claim for it other than
the exceedingly modest claim that it is as plausible as most o f the
conjectures which are current about the titles in the H erm ippus list.
Certainly it accords with all the ancient external evidence as well as
with the internal contents o f the treatises we possess. B u t some o f its
more unexpected features can be further confirmed, as we shall now
see.
TH E L IS T O F H E S Y C H IU S
Hesychius’ list contains 197 titles. T h e first 139 o f these form a
group which resembles so closely the list o f Diogenes that no
scholar doubts that in some manner it originates from the same
source. T h e remaining titles form an appendix, in four separate
parts. Both the main text and the appendix throw light on our
concerns.
T h e thirty-ninth title is an Ethics in ten books. M ost scholars
naturally enough identify this with the Nicomachean Ethics; but
they do not agree what value to place on the testimony. M oraux
regards the list as having been transcribed from the same original
catalogue as Diogenes’ list, but as having undergone a process o f
correction designed to eliminate apparent duplicates and to bring
the list into accord with the works o f Aristotle extant in the
Christian era. M oraux agrees with D üring that Diogenes and
Hesychius independently used the same original, and that in some
small and rather unimportant details Hesychius has preserved the
original better than D iogenes.1 H e thinks, accordingly, that the
figure ‘ 10 ’ is a comparatively late alteration to the original list. I f so,
it cannot be relied on as giving independent evidence o f a ten-book
Nicomachean Ethics prior to Aspasius.
It is possible, however, that the list in Hesychius is not so much a
defective copy o f the Hermippus list, as an independent inventory
o f the same library. A comparison o f the two lists yields a number o f
features which suggest this. For instance, though the titles listed
largely correspond, there is a great deal o f random variation in the
order in which they occur. However, no title travels very far from its
original position. It is hard to see what reason a copyist would have
I. M oraux, 19 5 1, 19 5 -2 0 9 ; D üring, 1957, 9 1.
44 The Aristotelian Ethics The Aristotelian Ethics in Antiquity 45
had to make so many tiny pointless changes in order : but it is easy 3 1-8 Logical and dialectical works (8)
enough to imagine books being replaced on shelves erratically, and 39 Ethics, ten books
catalogued at different times in different orders.1 40-4 Logical and dialectical books (5)
Once again, though, the ethical titles are scattered among the 45 On the idea (o f the good)
Logical and dialectical books. But this time, instead o f being 46-9 Logical and dialectical books (7)
distributed in pairs, with roughly a dozen books between each pair, 50 On what is better
we find them in approximately the same order,2 but singly, with 5 1-7 Logical and dialectical books (7)
roughly half a dozen books between each one. T h e list goes th u s:3 58 On the voluntary
i-16 Platonic and popular works 59-62 Logical and dialectical books (4)
17 Oeconomicus 63 On the noble
18 -2 3 Platonic and popular works (6) 64 Political works and theseis
24 On friendship, three books Probably it is idle to speculate on the cause o f this striking
25-9 Logical and dialectical works (5) relationship between the two lists. Certainly, it confirms the
30 On the passion o f anger hypothesis o f M oraux that a displacement o f the ethical titles has
taken place: and surely whatever explains the original misplace­
i . T h is impression is particularly strong in the case o f the group beginning at no. 6 in each
ment will also explain why the titles occur in one list in pairs
catalogue (nos. 1 - 5 coincide).
separated by a dozen titles, and in the other list in singletons
Diogenes Hesychius
6. Nerinthus, one book 6. Nerinthus, one book separated by half a dozen titles.1
7 · The Sophist, one book 7 · On wealth, one book M ost likely, the relationship between the two lists is to be
8. Menexenus, one book 8. The Sophist, one book
explained by a combination o f displacements o f books in the library
9 - Concerning love, one book 9 - On prayer, one book
10. Symposium, one book 10. Menexenus^ one book whose inventories are their basis, and emendations o f the texts of
I I . On wealth, one book I I . On noble birth, one book the lists between the time of Hermippus and Hesychius. M oraux’s
12. Protrepticus, one book 12. Concerning love, one book
explanation o f the ten-book Ethics on the latter basis has already
13· On the soul, one book 13- On the soul, one book
been mentioned. An explanation on the former basis is also
14· On prayer, one book 14. Protrepticus, one book
15· On noble birth, one book possible. Suppose that our earlier supposition, that the library of
It will be seen that in this section the titles contained are exactly the same, except that the Hermippus contained a five-book E E , plus two separately listed
Symposium has gone m issing in the second list. H owever, the order is varied in such a way
books making up the traditional Eudemian Ethics, is correct.
that the first four titles o f H erm ippus’ list appear as even-numbered titles in H esychius’ list,
and the last three titles o f H erm ippus’ list appear as odd-numbered titles in H esychius. It is i. It is not difficult to imagine circumstances which would explain these features o f the
hard to see what motive a copyist could have for such a sw itch; but if we imagine the books lists. Suppose, for instance, that a section o f the library was, for some reason, arranged as
arranged thus in the library, it is easy to see how the two lists could arise: follows, with the ethical works forming a separate, vertical group:
Nerinthus On Wealth H a lf a dozen Platonic works Oeconomicus
Sophist On prayer H a lf a dozen logical works On Friendship
Menexenus On noble birth H a lf a dozen logical works On the passions
Concerning love On the soul H a lf a dozen logical works Ethics in five books
Symposium Protrepticus H a lf a dozen logical works On the idea
I f one man catalogued in columns, and the other read the titles across, the first man would H a lf a dozen logical works On the better
produce a list very like H erm ippus’, and the second a list very like H esychius’ : the differences H a lf a dozen logical works On the voluntary
would be explained by the loss o f one book, and the slight reordering o f two others, between H a lf a dozen logical works On the noble
one cataloguing and the next. I f H esychius’ cataloguer went from left to right along each shelf in turn, from bin to bin, he
2. T h e order is identical except for the reversal o f the two titles O n the Idea’ and O n what would get the order we find in H esychius’ list; if a cataloguer went zig-zag, reading alternate
is better’ . shelves from left to right and then from right to left, he would get the order in H erm ippus’
3 - T h e numbers in brackets give the total o f intervening titles.
46 The Aristotelian Ethics
Suppose further that between the time o f Herm ippus and the
inventory which is the basis for Hesychius the remaining
Nicomachean books had found their way into the library. Unless we
suppose that at the same time the Eudem ian books had got lost, the
ten books o f Ethics now listed must be the original five Eudemian
ones plus five o f the Nicomachean books, presumably N E 1 - 4 and
10 (N E 5 and 6 being already among the five-book Ethics, and N E 7
being the book on the influence o f the passions1 ). That would leave
the two Nicomachean books on friendship. These would pre­
sumably be placed with the original Eudem ian book on friendship,
making now a total o f three books. And that is exactly what we do
find. In the place where the Herm ippus list has O n friendship : one
book’ the Hesychian list has O n friendship: three books’ .
In the Appendix Hesychiana we find, under no. 174, the entry O n
the Nicomachean E thics’ . M oraux thinks this is the N E itself. T o
do so, he has to postulate some very botched editing by a compiler
who (a) did not recognize the N E intended, on M oraux’s view, by
the correction in 39 ; and (b) quite misunderstood the work’s title,
inserting an ‘about’ . Surely this is more likely to be a work, not o f
Aristotle, but o f some scholar o f the period between Hermippus
and Hesychius, concerning the Ethics. Such a work would be more
likely to be included in a catalogue if it were itself a fragment of a
catalogue raisonné, discussing perhaps the troubled history o f the
transmission o f the N E to which the present study has borne
witness.
T H E M A N U S C R IP T T R A D IT IO N
T h e manuscript tradition o f the E E is very inferior to that o f the
N E . T h e oldest manuscript o f the N E is the tenth-century
Laurentianus 8 1, 1 1 (Bekker’s K b); the E E does not survive in any
manuscript earlier than the thirteenth century, from the end o f
which date the M S . Vat. gr. 1342 (Pb) and the Cambridge M S .
Cant. Ii 5.44 (C c). Altogether twenty-one manuscripts o f the E E
survive, the great majority from the fifteenth century. In a masterly
study Dieter Harlfinger has shown how they can be grouped into
two main classes: the Recensio Messanensis, o f Southern Italian
origin, whose earliest representatives are Pb and C c ( = Susem ihl’s
i. T h e title being, in our manuscripts o f H esychius’ list, now further corrupted, a well-
meaning scribe having corrected π ep l π α θ ώ ν ο ρ γ ή ς into π epi π ά θ ο υ ς ο ρ γ ή ς .
The Aristotelian Ethics in Antiquity 47
family Π 1 ), and the Recensio Constantinopolitana, o f Byzantine
origin, of which the archetype is the M S . Laurentianus 8 1, 15 o f the
1420s (L ) .1
Since the time o f Bekker it has been customary to say that the
manuscripts o f the E E do not include the text o f the common books
o f the A E , but merely refer the reader to the text o f the N E .2 In
fact, as Harlfinger has shown, eleven o f our twenty-one manu­
scripts, all deriving from the Recensio Constantinopolitana and
dependent on L , set out the text o f the A E in full within the context
o f the E E . It is only in manuscripts in which the text o f the N E
precedes that o f the E E that copyists, to save themselves from a
duplication o f labour, have contented themselves with a backward
reference to the fifth, sixth, and seventh Nicomachean books. No
argument, therefore, to an original Nicomachean home for the
disputed books can be drawn from the failure o f a number of
manuscripts to set them out in full in their Eudemian context.
One might hope, however, that a stemmatological study o f the
manuscript traditions o f the books o f the A E might throw light on
the problem. If, for instance, the Eudemian tradition o f these books
turned out to be independent of, and superior to, the Nicomachean
tradition, that might support the hypothesis that the books belonged
originally in a Eudem ian context. In a long and patient study
Harlfinger has shown that this hope is delusory. In fact, he inclines
to think that the E E tradition o f the disputed books, represented by
L and its descendants, takes its origin from a point somewhere in
the genealogy o f N E manuscripts close to the vetustissimus K b.3
T h is, in turn, does not show that the N E is the original home o f the
common books, as Harlfinger is at pains to point out. For it may be
1. Harlfinger, 1 9 7 1 , 1 - 4 9 .
2. Such was the case in the unrepresentative manuscripts which Bekker collated. T h u s
M arcianus 200 (Q) and 2 13 (M b)atth e end o f E E I I I carried this remark: σ η μ ε ίω σ α ι δ τ ι το
8 κ α ι € κ α ι s1' τ ω ν Ε ύ Β η μ ίω ν η θ ικ ώ ν ε ν τα ύ θ α π α ρ ε ίν τ α ι δ ια το το μ ε ν 8 τ ώ e τ ώ ν
Ν ικ ο μ α χ ε ίω ν , το 8 ε ε τ ώ ζ, το 8e s1' τ ώ η τ ώ ν Ν ικ ο μ α χ ε ίω ν εν π ά σ ι κ α ι κ α τ α λ ε £ ιν
ο μ ο ια ε ίν α ι. P b contained merely the incipits o f books I V - V I o f the E E .
3. On the basis o f collations o f ten Bekker columns in the A E H arlfinger concludes : ‘Das
in D etailfragen mit gewissem Vorbehalt, in der allgemeinen Tendez aber als ziemlich
gesichert zu betrachtende Ergebnis ist folgendes : D er T e x t underes L a u r. 8 1, 15 (L) ist sehr
eng mit dem des L a u r. 8 1, 1 1 ( K b), jenes berühmten vetustissimus der E N und der Magna
M oralia, verwandt. . . . Dam it ist natürlich erwiesen, dass der T ext der kontroversen
Bücher, wie er uns in L vorliegt, aus einem K od ex der E N stammt, der im Stem m a dieser
Ethik irgendwo diesseits des Archetypus in der N ähe, aber nicht in der Deszendenz des
höchstens eine von mindestens zwei voneinander unabhängigen Fam ilien repräsentierenden
Lau r. 8 1, i i (K b) angesiedelt werden muss’ ( 19 7 1, 43).
48 The Aristotelian Ethics The Aristotelian Ethics in Antiquity 49
that whereas the earliest E E manuscripts (those available; for view o f the dependent nature o f the Eudem ian tradition. There are
instance, to Aspasius) contained the disputed books, there came a six places in which there are discrepancies between the papyrus and
time when the only surviving manuscripts were ones in which the the manuscripts. In four o f these the papyrus agrees with K b against
E E followed the N E and the common books were therefore L . 1 In two it agrees with L b against L . 2 In no case does it agree with
represented in the E E context only by their mcipits or by a back- L against K b. T h e tradition which— in these passages— the
reference. Then, at a stage later than that represented by the K b readings of L approach most closely to is that represented by M b. 3
manuscript o f the N E , a scribe wishing to have a complete text o f
1. T hree o f these are simple errors in L (α ύ ξ η σ ις for α ν τ η in I i4 2 bi 3 ; où omitted in line
the Eudemian Ethics would need to obtain his text o f the common bi4 ; μ ο ν λ ε υ ό μ ε ν ο ς for β ο υ λ ε υ ό μ ε ν ο ς in bi4). T h e fourth is the reading h iw p ia r a i at
books from an existing Nicomachean version. Harlfinger concludes i i 4 2 bn in which L agrees with M b.
2. T h e readings εά ν τ ε instead o f έά ν τ ε κ α ι at H 4 2 bi5 , and the reading ε υ δ α ιμ ο ν ία ν at
that the manuscript evidence does not permit us to settle the the well-known crux in H 44b6.
question o f the original context o f the disputed books. 3 - In the two readings mentioned in the previous footnote, and the one non-trivial reading
Students o f the text o f the N E differ in their valuations o f its most mentioned in n. i, L is in agreement with M b; it agrees with K b only where the latter agrees
with M b. Since, according to Gauthier (1970, 3 1 1 ) xMb derives from a tradition independent
important m anuscripts; and their relative value appears to vary
o f K b and L b, this tiny sample o f readings does not accord with the results o f H arlfinger’s
from book to book. Bekker based his edition principally on four much more extensive survey.
m anuscripts: in addition to K b he used L b (Parisiensis 1854, 12th
cent.), M b (M arcianus 2 13 , 15th cent.), and Ob (Riccardianus 46,
14th cent.). Bonitz and Susemihl regarded M b, in spite o f its late
date, as being second in importance to K b. Jackson’s edition of book
A o f the A E was based on a wider collation than previous ones: for
that book he regarded L b as the most trustworthy, though he
conceded that K b was the richest in preserving good readings
elsewhere lost. Byw ater, whose edition is the one most familiar to
English readers o f the N E , accorded an enormous primacy to K b,
while at the same time attaching perhaps undue importance to the
readings o f the medieval Latin version, which he regarded as
‘almost as good as a Greek manuscript to us’ .1
T h e oldest witness to the text o f the Ethics is a single papyrus of
the second century.2 It contains only two small fragments, from the
common book B ( i i 4 2 bn - i 7 and ii4 4 z6 - i i ) ; in two places it
supports the readings o f L b against K b, to that extent corroborating
Jackson’s judgement against Byw ater’ s. B y comparing the text o f
this papyrus with that o f the best N E manuscripts on the one hand,
and that o f the Eudemian tradition represented by L on the other,
one might hope to confirm or discredit H arlfinger’s tentative
conclusion that the text o f the A E , as it stands in our Eudemian
manuscripts, is not independent but derives from the Nicomachean
stemma. T h e comparison goes some way to confirm H arlfinger’s
1. On the manuscripts o f the N E see Bywater, 18 9 2 ; Jackson, 1879, ix - x ii; Gauthier,
1 9 7 ° , 30 T-1 4.
2. Oxyrhynchus Papyrus no. 2402; Lobel, Roberts, T urner, and Barns, 1957, 12 4 -6 .

CH APTER 2
Cross-references in the
Ethics
H a v in g examined the external evidence for the relationship
between the Eudemian and Nicomachean Ethics I turn henceforth to
the internal evidence concerning the original context o f the
disputed middle books. For, as I observed earlier, this is the
question which should be raised and answered before any attempt is
made on internal grounds to settle anything about the systematic or
chronological relationship between the two Ethics. For if this is not
first settled, then that Ethics to which the disputed books rightfully
belong— whichever it may be— is being unfairly judged on the
basis o f an unnecessarily fragmentary text ; and in any case, as we
shall see, there is very much more internal evidence available to
indicate the true home of the disputed books than there is to
indicate which o f the two treatises, as a whole, is chronologically
prior. Failure to observe these obvious methodological points led in
the case o f scholars as gifted in different ways as Schleiermacher
and Jaeger to fallacious and indeed circular argumentation.1
In such an inquiry the first essential though unexciting task is to
examine the cross-references internal to each o f the ethical treatises
to see whether those in the undisputed Nicomachean books fit the
content o f the controverted books better or worse than those in the
undisputed Eudem ian books. In the present chapter I will examine
the forward-looking references in the first five Nicomachean books,
and the first four Eudemian books— excluding those which find
their fulfilment clearly within those books themselves— and
compare them with the disputed books.
i. Jaeger argued that the disputed books m ust belong to the N E because they contained a
view o f phronësis which suited the N E not the E E ; but A E B is itself the main evidence for
attributing such a concept to the N E (cf. above, p. 4). Schleiermacher argued that the N E
could not be by Aristotle because o f the inconsistency between the view o f virtue in the early
books and in book 6 ; and then went on to attribute the disputed books to the E E
(Schleierm acher, 1835).
Cross-references in the Ethics 51
T h e early Nicomachean books are not rich in forward-looking
references.1 When the three lives are introduced in 1.5 the
consideration o f the life o f contemplation is postponed until later
(1096*7) ; and a little further on we are told that the element o f self-
sufficiency in happiness needs to be looked at again (io g 7 bi4). Both
o f these promises seem to be kept within the undisputed
Nicomachean books, the latter in chapters 9 and 10 o f book 9, the
former in the second half o f book 10. B u t after the introduction o f
the Mean in book 2 we read, at the end o f the preliminary list o f
mean states: ‘About these there will be another occasion to speak;
but with regard to justice, since the term is used in more than one
sense, we will later distinguish between them and show how the two
kinds are severally mean states ; and similarly with regard to the
rational virtues’ ( n o 8 b6 ff.). T h e ‘other occasion’ to speak about
the listed virtues is presumably the detailed treatment o f them in
books 3 and 4. T h e reference to justice will fit well the treatment in
A E A : there he does indeed distinguish between two forms o f
(particular) justice, namely distributive and corrective, and argues
that each o f them is concerned with a mean ( i i 3 i ai7 ff. and
i I3 2 ai9 ff.). However, the final remark about ‘the rational virtues’
does not seem to fit the content o f the disputed books : at no point in
them does Aristotle attempt to show that the intellectual virtues are
mean states. 2
A more apt foreshadowing o f the content o f A E B comes earlier
in book 2 where virtue is provisionally described as acting in
accordance with right reasoning: ‘Th ere will be a discussion o f this
later, and o f what right reasoning is, and what is its relation to the
other virtues’ ( n o 3 b32-4). Right reasoning on the topics appro­
priate to the virtues, we are told at 1 1 44a24 in A E B , is wisdom ; and
there follows immediately a discussion o f the relation between
wisdom and the other virtues. (T he points made here about the
relation between wisdom and the virtues are also made in N E 10,
1 1 78at6, but too briefly to be the fulfilment of the promise o f N E
2·)3
1 . Examinations o f the cross-references can be found in Spengel (18 4 1, 489 if.), Fritzsche
( 1 8 5 1 , xxxiv ff.), G ran t (188 5, 47 ff.), Festugière (19 36, xx x-xliv), Fragstein (1974, 398 ff.).
2. F o r this reason, and because the expression λ ο γ ικ α ί aperçu does not occur elsewhere
in Aristotle, most commentators regard the clause as an interpolation. However, the reading
is attested by Aspasius, and at E E 1 2 2 1 a12 Aristotle lists wisdom as a mean between cu nning
and simplicity.
3. Though Aristotle’s words fit A E B well, Aspasius’ commentary at this point does not:
52 The Aristotelian Ethics
Finally, at the end o f book 4 we read: ‘N either is continence a
virtue, but something intermediate; it will be discussed later; now
let us talk about justice.’ G rant thought that these words had been
added by an editor, after an abrupt and interrupted ending to book
4, ‘to give the book a seeming union with the three Eudem ian books
which were now to be grafted on’ (Grant, 1885, 94). I f there are
independent grounds for regarding the middle books as Eudemian,
then there may be reason for regarding the present passage as an
interpolation; but at the present stage o f the argument it would be a
petitio principii to do so.1
Taking the text o f N E 1- 4 as it stands, therefore, we can say that
it leads us to expect a later treatment o f justice as a mean, in two
parts; o f right reasoning as a virtue; and o f continence. T h is fits
well enough the content o f the disputed books : if one could not
learn their detailed contents from these slender allusions, equally
one cannot claim that they contain anything that contradicts them,
except for the reference to intellectual virtues as means.
T h e references foward in the early books o f the Eudemian Ethics
are very much fuller.
(1) É E 1.5, 12 16 *37 :
S e ife t δ ε ό τ ι 6 ο ρ θ ό ς λ ό γ ο ς ε σ τ ιν ή φ ρ ό ν η σ ις, έ χ ε ι δ ε π ρ ο ς τ α ς α λ λ α ς ά ρ ε τ α ς ή
φ ρ ό ν η σ ις ώ ς η γ ε μ ο ν ικ ή τ ις . δ η λ ω θ ή σ ε τ ο .ι δ ε π ε ρ ί τ ο ύ τ ο ν , ό τ α ν κ α ί α υ τ ό ς λ ε γ η .
ε κ ε ΐν ο 8 ε ύ π ο μ ν η σ τ ε ο ν , ό τ ι τ ο κ α τ α το ν όρθον λ ό γ ο ν π ρ ά τ τ ε ιν ου μ ό ν ο ν τ ο ΐς ε χ ο ν σ ι
τ α ς τ ε λ ε ία ς α ρ ε τ α ς π ρ ο σ ν ε μ ε ι, άλλα κ α ι τ ο ΐς μ η ε χ ο υ σ ι τ α ς τ ε λ ε ία ς , π ρ ά τ τ ο ν σ ι γ α ρ
ο ν τ ο ι, o>s άν ό ο ρ θ ό ς λ ό γ ο ς ύ π α γ ο ρ ε ύ σ ε ιε ν , ου μ η ν ε χ ο ν τ ε ς τ ο ν λ ό γ ο ν α π ο δ ε ικ τ ικ ό ν
α λλ ά τ ο ν δ ο ξ α σ τ ικ ό ν , ό τ ι τ ο ν τ ο μ ε ν π ρ α κ τ ε ο ν , τ ο ΰ τ ο δ ε οι) π ρ α κ τ ε ο ν . T h ere are several
striking differences between A spasius’ remarks here and A E B. First, in Aristotle there is no
suggestion o f any hegemony o f wisdom over the other virtues : there is rather a partnership.
Secondly, though Aristotle does distinguish ( ii4 4 b27) between those who act κ α τ α τ ο ν
ο ρ θ ό ν λ ό γ ο ν and those who act μ ε τ ά τ ο υ ό ρ θ ο ν λ ό γ ο υ he does not connect this at all with a
distinction between perfect and non-perfect virtue (the terms are used in A E to draw a
different distinction: ii2Q b2 i ) ; nor would Aristotle agree with the distinction between the
apodeictic and doxastic logoi; for him, wisdom is the virtue o f the doxastic faculty ( ii4 4 b14).
T h is passage perhaps suggests that the N E on which Aspasius commented did have a
discussion o f wisdom rather than the complete lacuna conjectured earlier (p. 33), but it shows
also that it is not the one which we have as A E B . U nless we are to say that he took the forward
reference to be merely to later passages in N E 2 (e.g. n o 7 aI) and that the un-Aristotelian
remarks are his own unprompted additions.
i. A possible trace o f disorder in the N E text is found in the commentary o f the
anonymous scholiast on this passage: ύ σ τ ε ρ ο ν δ ε ε π ι π λ έ ο ν ε ρ ε ΐ π ε ρ ί ε γ κ ρ α τ ε ία ς κ α ι
κ α ρ τ ε ρ ία ς κ α ι μ α λ α κ ία ς , μ ε τ ά τ ό ε ξ ή ς β ιβ λ ίο ν ; see above, p. 37· T h e singular is odd,
suggesting that A E C succeeded A E A in his text. B u t this in turn is contradicted by the same
scholiast’s commentary on the text discussed above, n o 3 b33, which reads: ρ η θ ή σ ε τ α ι δ ’
ύ σ τ ε ρ ο ν π ε ρ ι α ν τ ο ν , ε ν τ ω ε κ τ ω ό π ο ν π ε ρ ι φ ρ ο ν ή σ ε ω ς λ ε γ ε ι.
Cross-references in the Ethics 53
It is clear that all relate happiness to three lives, the life of the politician, of
the philospher, and of the hedonist. Now of these the pleasure which is
concerned with the body and its enjoyments1 is no difficult matter to
decide what and what kind of thing it is and how it is produced ; what we
need to inquire about them is not what they are, but whether they
contribute to happiness or not, and in what way they contribute, and
whether some pleasures should be attached to living a noble life, and if so
whether these should; or whether these should be partaken of in some
other way while there are some other pleasures in accordance with which
the happy man may reasonably be said to live a life of pleasure and not
merely a life free from pain. But these matters must be investigated later.
It has been shown by Festugière that this passage admirably fits the
discussion o f pleasure at the end of the third disputed book.
Though the end o f that discussion is described as having dealt with
the question what pleasure is, the discussion is far less concerned
with the nature o f pleasure than the corresponding one in N E 10
and is almost wholly devoted to the question o f the relationship
between pleasure and the supreme good.
All the questions asked here are answered in A E C : bodily
pleasures contribute to happiness because the happy man needs
bodily goods so that his characteristic activities may not be impeded
( I I 53t>I7); there is a right way o f enjoying food and wine and sex,
and it is only excess in them that is blamable (i i5 4 ai6 ); the happy
life is one o f pleasure, not just o f freedom from pain (i i5 3 a28), but
the pleasure which constitutes happiness and indeed the greatest
good is not bodily pleasure but is the unimpeded exercise o f virtue
( n 53bi 3)·
(2) T he passage o f E E I continues thus : ‘First we must consider
virtue and wisdom, what the nature o f each is, and whether they, or
the actions which arise from them, are part o f the good life.’ T h is is
an accurate prediction o f the structure o f the E E including the
disputed books : the account o f pleasure promised comes after the
treatment o f virtue (E E II and I I I and IV) and o f wisdom (E E
V = A E B). T h e relation between the moral virtues and the parts o f
happiness is stated in A E A at i i29 bi8 , where it is said that legal
justice, which combines the virtues o f courage, temperance,
meekness, etc., is concerned with the things that produce and
i . Professor D . J . Allan has pointed out to me that if Stew art’s emendation (18 9 2,220 ) τα ς
σ ω μ α τ ικ ό ς α π ο λ α ύ σ ε ις is read at this point, there is a striking coincidence o f expression
with A E C , ii4 8 a5.
54 The Aristotelian Ethics
preserve happiness and its parts. T h e intellectual virtues o f
practical and philosophical wisdom are described in A E B as
producing happiness, the latter as ‘a part o f total virtue’ (i i44a5).
W hile the promise o f E E I is fulfilled to the letter in the disputed
books, it is impossible to see a fulfilment for it within the
undisputed Eudem ian books; those therefore, like Schächer, who
maintain that the E E never contained more than five books, have to
regard the whole passage as a false start (Schächer, 1940, 66).
(3) A t 1 2 1 8b 1 2 we are told that the good which is the final end o f
things achievable by man is the good which is the object o f the
discipline or disposition which governs all others. ‘T h is ’, Aristotle
goes on, ‘is politics and economics and wisdom. F o r these
dispositions differ from others in being o f this kind ; whether they
differ at all from each other will be explained later.’ 1
T h e later explanation is given, as promised, at 1 i 4 i b23 in A E B .
W isdom, politics, and economics are all the same disposition, but
they connote different exercises o f that disposition, concerned with
different subject matters. ‘W isdom’ in the broad sense is the general
name for the disposition: it may be exercised about an individual’s
own affairs (that is wisdom in the narrow sense) or it may be
concerned with a wider scope. I f it concerns his household, then it is
economics ; if it concerns the state, then it is politics. Any exercise o f
wisdom concerning the state is politics in the broad sense; but the
exercise of wisdom in this way may be expressed in legislative,
executive, or judicial decisions ; only in the two latter cases can it be
called ‘ political’ in the narrower sense.2
(4) T h e E E , like the N E , after introducing the notion o f right
reasoning into the concept o f virtue, promises a further con­
sideration o f its nature. ‘What right reasoning is, and what is the
mark to which we must look in order to specify the mean, must be
examined later’ we are told at i2 2 2 b8 .3 A s with the N E reference,
the promise to examine what right reasoning is is fulfilled in the
examination o f wisdom in A E B ; but the second question about the
I. ο ) θ τ ε τ ο ίιτ ' αν ε ιη α ν τ ο τ ο α γ α θ ό ν το τ έ λ ο ς τ ώ ν ά ν θ ρ ώ π ψ π ρ α κ τ ώ ν . τ ο ΰ τ ο δ ’ έ σ τ ί
τ ο νττο τ η ν κ υ ρ ία ν π α σ ώ ν , α υ τ ή δ ’ ε σ τ ι π ο λ ιτ ικ ή κ α ί ο ικ ο ν ο μ ικ ή κ α ι φ ρ ό ν η σ ις.
δ ια φ έ ρ ο υ σ ι γ ά ρ α ί τ α ι a i ε ξ ε ι ς π ρ ο ς τ α ς α λ λ α ς τ ώ τ ο ια ν τ α ι ε ΐν α ι■ π ρ ο ς δ 'α λ λ ή λ α ς ε ΐ
τ ι δ ια φ ε ρ ο ν σ ιν , ύ σ τ ε ρ ο ν λ ε κ τ έ ο ν .
2 · A schematic tree o f the different branches o f wisdom is given by von Fragstein (1974,
232 f.).^
3. τ ι ς δ ο ο ρ θ ο ς Aoyoç, κ α ι π ρ ο ς τ ι να δ ε ι ορ ον α π ο β λ έ π ο ν τ α ς λ έ γ ε ιν τ ο μ έ σ ο ν ,
ύ σ τ ε ρ ο ν έ π ισ κ ε π τ έ ο ν .
Cross-references in the Ethics 55
mark which specifies the mean is merely repeated, but not
answered, in bo okB (i i3 8 b22). For an answer we have to wait until
E E V III. (The question is not raised in the undisputed N E books.)
(7) In his discussion o f voluntariness Aristotle commends the
legal distinction o f events into voluntary, involuntary, and
premeditated ; even if the distinction is not exact, he says, they do
grasp the truth in a certain manner. ‘B u t we will speak about these
matters in our inquiry into justice’ (i2 2 7 a2). This promise is kept
when in chapter 8 o f A E A a detailed treatment o f voluntariness
replaces the simple threefold legal division with a many-branched
schema which preserves its main lines while allowing for more
complicated cases. T h e exactness o f this forward reference is
striking: it is by no means something to be taken for granted that a
treatise on justice will contain a discussion o f the various degrees o f
mens re a.
(8) In the last chapter o f E E II Aristotle asks whether what
virtue does is to safeguard one’s choice o f end or one’s reasoning.
Som e people, he says, think the latter ‘but what does this is
continence: it is this which keeps the reasoning uncorrupt. Virtue
and continence are different. We must speak later about them,
because this is the reason why those who do so regard virtue as that
which makes the reasoning right’ ( i2 2 7 bi5 ff) . It is not difficult to
find in the disputed books a treatment o f the difference between
continence and virtue— that is one o f the major themes o f A E C —
but it is not at first obvious how what is here said about the relation
between continence and reasoning fits what is said later. On the one
hand, we are told in A E B that it is temperance which preserves the
wisdom which is identical with right reasoning ( 1 1 40b1 2 ; 1 144b28) ;
yet temperance is the virtue which in A E C is distinguished from
continence ( i i5 2 a i 3). On the other hand, correctness o f reasoning
seems to extend beyond the bounds o f virtue not merely to the
boundary between continence and incontinence, but to the
boundary between incontinence and vice; for we are told in A E C
that the incontinent man has a good prohairesis ( i i 5 2 ai7) which
must include right reasoning ( 11 3 9 “24) ; and indeed the incontinent
man is often described precisely as the man who does not stick to his
reasoning ( i i 5 i a20, a32, b25).
T h e inconsistency between the E E and the A E is only an
apparent one : indeed, if it were a genuine inconsistency it would be
internal to the A E itself. It is true that there are incontinent men
56 The Aristotelian Ethics
whose reasoning is perfectly correct: but besides these weak
incontinents there are the headstrong incontinents whose passions
are so strong that they are lead astray by passion without
deliberating at all ( i i 5 0 big ff.). Consequently it is only o f the
continent that it can be said without exception that their reasoning
is safe. On the other hand the difference between the continent man
and the temperate man is not a matter o f reasoning but o f feeling :
the continent man has unruly desires, but unlike the incontinent he
is not led astray by them ; whereas the temperate man does not feel
unruly desires at all ( i i 5 2 ai). None the less continence resembles
temperance, and can be called temperance by analogy ( i i 5 i b32);
and it is surely that which Aristotle was doing when, in reporting a
popular etymology o f the Greek word for ‘temperance’ (sophrosyne )
he said that it was what preserves wisdom (sözei ten phronesin). T h e
imprecision o f this passage is corrected by the later distinction:
what the temperate man is here contrasted with is not the continent
man but the man ‘corrupted by pleasure and pain’— the
intemperate man whose first principles are perverted ( ii4 0 b20).
(9) At the conclusion o f the discussion o f intemperance in E E
I I I Aristotle promises to treat with greater exactitude ‘o f the class o f
pleasures’ when he comes to deal with continence and incontinence.
A E C o f course contains a detailed treatment o f pleasure in
conjunction with a treatment o f continence and incontinence: but
some scholars have complained that despite the full discussion the
class o f pleasures with which intemperance is concerned (those of
touch, o f a certain kind) is no more precisely defined than in E E I I I
(i2 3 0 b2 5 - i 2 3 i a25). But though the class o f pleasures was
precisely defined in E E I I I , A E C shows in chapter 4 how this class
can be viewed as a species o f various genera o f pleasures according
to particular classifications, in chapter 5 how the class itself is a
genus o f which natural and perverse pleasures are species, and in
chapter 7 how even in the case o f natural and necessary pleasures
one must distinguish between excessive and defective instances.
These passages come not in the section o f A E C devoted exclusively
to pleasure, but, as the preannouncement stated, in the section
devoted to continence and incontinence.1
i . T h e objection, first made by Spengel, was forcefully taken up again by K ap p ( 19 12 ,2 3 ) .
T h e best treatment o f the passage is by Festugière (1946, xx iv -ix , lxx), who shows that the
subdivisions o f objects o f pleasure in A E C 7 give three classes: (a) necessary, bodily
pleasures which provide the field o f temperance and intemperance ( i i 4 7 b24 ff.) = the
Cross-references in the Ethics 57
(10) At i2 3 4 a28 Aristotle says that ‘as will be said later, every
virtue is found in one way by nature and in another way with
wisdom’ . T h is fits admirably the account o f natural virtue in A E B
( ii4 4 bi - i 7 ) which concludes ‘in the part o f the soul concerned with
character there are two kinds o f virtue, natural virtue and full-
blooded virtue, and full-blooded virtue does not come about
without wisdom’ .
( 1 1) Like N E 4, E E IV ends with an announcement o f a
treatment o f justice (i2 3 4 bi4).
All these eleven forward-looking allusions, then, fit without
difficulty the disputed books ; but there are other passages which are
not so easy to locate as correspondences to previous announce­
ments. This was pointed out by Spengel (18 4 1, 489 ff.): Aristotle,
having said that beasts cannot be called happy at 1217*26 , goes on:
‘ N or can anything else whose name does not indicate that it shares
in its nature in something divine; it is by some other mode of
participation in goods that some o f them have better lives than
others. But that this is so must be considered later.’ Dirlm eier may
be right to say that Aristotle here allows for the possession of
happiness by divine and superhuman entities like the aether (1969,
19 1); but it is not this that he promises to consider later. It is more
naturally taken to be the applicability o f happiness, or sharing
goodness, to animals. That is how von Fragstein takes it (1974, 31)
and refers forward to A E C, 1 14 8 °i5 ff. : ‘ some things are pleasant
by nature, and o f these some . . . in accordance with the species o f
animals and men. ’ But a passage which is more obviously concerned
with the relation between animals and the supreme good is A E C,
H 53b25 ff·
And indeed the fact that all things, both brutes and men, pursue pleasure
is an indication of its being somehow the chief good. . . . But since no one
nature or state either is or is thought the best for all, neither do all pursue
the same pleasure; yet all pursue pleasure. And perhaps they actually
intermediate pleasures o f 1 148*25; (b) the pleasure-making things that are choiceworthy in
themselves (1 i4 7 b25) but admit o f excess = the naturally choiceworthy pleasures ( 1 148a24),
namely honour, victory, riches; and (c) the things naturally unchoiceworthy ( 1 148a24), not
naturally pleasant but only pleasant through innate or acquired corruption ( 1 148ai6 fF.). This
seems the best interpretation o f a difficult passage. C f. also D . J . Allan, 19 7 1. T hese divisions
are necessary to complete the account o f the relationship between temperance and bodily
pleasures : because the vice o f intemperance concerns only excesses o f natural pleasures, the
disposition corresponding to unnatural pleasures o f touch and taste is not vice but bestiality
(ii48b25- ii49ai).
58 The Aristotelian Ethics Cross-references in the Ethics 59
pursue not the pleasure they think they pursue nor that which they would the E E was ever concluded by Aristotle according to his plan, and
say they pursue, but the same pleasure; for all things have by nature did not contain the A E as we have them, then it must have
something divine in them. contained something very similar indeed.

T h is mysterious passage shows, I think, that Aristotle again has

in mind the topic raised in the passage from E E I ; but it seems to
indicate a slightly different point o f view. He agrees in both
passages that only things which have something divine in them can
enjoy the supreme good; but in the A E passage he seems to be
willing, as he was not in the E E passage, to countenance the idea
that animals lower than man have something divine in them. But o f
course not even in the A E passage does he suggest that the animals’
pursuit o f pleasure, which may be a pleasure in some sense identical
with the pleasure which is the supreme good, entitles them to be
described as happy.
In this case, then, we have an echo rather than an answer to the
question postponed for discussion in book one. Bu t from the other
eleven passages considered in the E E we could form quite a detailed
picture o f what is to follow the treatment o f the ethical virtues in E E
I —I I I . There is to be a treatment o f justice containing a detailed
discussion o f the relation between the voluntary, the involuntary,
and the premeditated, with special reference to legal terminology ;
there is to be a discussion o f the nature o f right reasoning, and a
treatment o f the relationship between wisdom and on the one hand
political and economic skill and on the other hand natural virtue;
the place o f wisdom in over-all happiness is to be settled. There is to
be a treatise on continence and incontinence, which will draw a
precise distinction between continence and virtue, and which will
contain a detailed classification o f pleasures. There is to be a treatise
on pleasure which will settle which are the pleasures which
constitute happiness, and what contribution to happiness is made
by bodily pleasures in particular.
T h e picture which is presented fits admirably the disputed books
as we have them ; and it is an incomparably fuller picture than that
which emerged from the brief allusions in the Nicomachean Ethics.
At the very least we can say this: from the forward-looking
references in the early books o f the E E we can be certain that
Aristotle planned to include in the E E a treatment o f the topics
which are dealt with in the A E ; that the A E as we have it fits
perfectly the description of that treatment as planned ; and that if

CH APTER 3
The Style o f the
Treatise on Justice
T h e first o f the disputed books is in some respects more self-
contained than either o f the others, and it has consequently
sometimes been treated as a separate unit. Fischer and von
Fritzsche, for instance, who presented an array o f arguments to
show that A E B and A E C belonged to the Eudemian Ethics,
thought that A E A was a Nicomachean book in the main, and
pointed to the apparently purposeless repetition in chapter 1 1 as an
indication that the end o f the book alone was a Eudemian fragment
(Fischer 1847, 12 ; von Fritzsche, 18 5 1, xlvi-xlvii). On the other
hand, even within the first ten chapters there are often difficulties in
following the argument which have led many commentators to
suspect doublets and dislocations in the text: so much so that
Jackson in his commentary o f 1879 printed a substantially
rearranged text. Recently, Row e (19 7 1, 100-7) has argued that the
unevenness o f the text is to be explained by postulating two strata,
an earlier E E version having been overlaid with new N E material.
M ost recently von Fragstein (19 7 4 ,16 2 -2 14 ) has urged that the text
is substantially in order, has offered frequently convincing accounts
o f the purpose o f the apparent repetitions, and has assigned the
whole book to the E E .
In the present chapter I shall go through the first chapter o f the
book as a sample o f the whole and consider what evidence is to be
gathered from vocabulary, style, and method o f argumentation as to
the original provenance o f the book. W e shall find that even a single
chapter is rich in clues to its origin.
We may notice first that the transition from E E I I I to the treatise
on justice is smoother than that from N E 4. Indeed there is some
indication that N E 4 is incomplete. T h e list o f mean states which
Aristotle in book 2 ( 1 i0 7 a2 8 - i io8bio) promises to deal with before
treating of justice concludes with shame and righteous indignation
The S tyle o f the Treatise on Justice 61
(nemesis). But book 4 ends with the treatment o f shame, followed by
the abrupt forward references to treatments o f continence and
justice already quoted. Aristotle’s failure to keep his promise to
discuss righteous indignation has been noticed since antiquity, and
variously explained.1 There is no similar gap left in the transition
from E E I I I to A E A.
T h e first chapter o f A concerns the distinction between legal or
general justice, or obedience to law as a whole, which is the same
state of character as virtue as a whole, but considered in relation to
others ; and particular justice which has its own field o f operation
just as courage and temperance have. T h e argumentation o f the
chapter presents no difficulties o f substance, and the doctrine is
consistent with both E E and N E . F o r indications o f its provenance
we have to look at details o f terminology and method : and o f these
there is no lack.
Consider, for instance, the first paragraph.
With regard to justice and injustice we must consider (1) what sort of
actions they are concerned with (2) what sort of a mean state justice is (3)
when something is just what are the extremes it is a mean between. Let our
investigation follow the method of the previous discussion. We see that all
men mean by justice that state of character which makes people capable of
doing what is just and makes them act justly and wish for justice; and
similarly by injustice that state which makes them act unjustly and wish
for injustice. Let us too then lay this down at the start as a general basis.
(AE i i29a3~i 1)
T h e practice o f asking, concerning a state o f character, ‘ what
things it is concerned with’ is common to both N E and E E (e.g. N E
i i i 5 a5, n i 7 b25, n i 9 b23, etc.; E E 1228*27, i 23° a34, I2 3 lS39,
etc.). In speaking o f justice as concerned with actions (praxeis)
Aristotle might seem here to be closer to the N E (which said that
virtue concerns passion and action, pathë kai praxeis, 1 i09b30) than
to the E E (in which the fullest definition of virtue describes it as
being concerned with pleasures and pains i2 2 7 b7). Bu t in the E E
too virtues are described as having particular types o f praxis
for their province (e.g. 12 3 3 ^ 2 ) — a formulation which is not
i. T h e anonymous scholiast comments on the lack o f a discussion o f nemesis (C A G xx, 204),
and there is a lengthy discussion o f Aristotle’s reason for omitting it in Albertus M agnus.
T h ere is no discussion o f the topic in the manuscripts o f Aspasius because they do not
preserve the whole o f his commentary on book 4, breaking o ff at i2 2 8 a33-
02 The Aristotelian Ethics The S tyle o f the Treatise on Justice 63
incompatible with the idea that virtue is concerned with pleasure in in all the Aristotelian ethics that the occurrence o f a medical analogy
a particular type o f action (ibid. i2 3 3 a38). in a passage cannot provide a decisive piece o f evidence for its
T h e reference to previous methodological considerations appurtenance ; but a comparison o f the N E and the E E shows that it
( i i 2 9 a5) fits the E E better than the N E . T h e methodological is in the latter that medicine is most systematically used as a model
considerations o f the N E concern the impossibility o f exactness in for ethical m ethod.1
moral science and the importance o f beginning with things better i. W. Jaeger collected together the passages in which medicine is used as a model for
ethical method in the N E and A E (Jaeger 1957, 58 ff.). T h ere are a number o f passages
known to us (N E 1.3 , i094bn ff.) and it is to this that, according
throughout the N E in which medicine appears as an example o f a techne on an equal footing
to the anonymous scholiast ( C A G xx.205), Aristotle is here with other arts such as housebuilding, rhetoric, steersmanship, etc. T h e important passages
referring. But what he goes on immediately to do is to make where it is singled out (along with the kindred discipline o f γ υ μ ν α σ τ ικ ή ) are: io g7 ai 3 :
medicine is concerned with the individual, not the universal ; medicine is practical 1 i0 5 bi 2 -
reference to a universally held opinion, and lay it down as a
1 8 ; the mean is introduced via medical examples n o 6 bi ff.; the importance o f a combination
hypothesis. It is the E E which, in its methodological chapter 1.6 , o f general knowledge o f the art and detailed provision for the individual in medicine and
I 2 i 6 b 2 6 ff., states that the opinions— preferably universal— o f men education ( n 8 o b7 ff.). Jaeger does not consider the use o f the analogy in the E E but says
are to be taken as evidences and exam ples.1 ‘ References to medical discipline and its methods are frequent in it, often occurring in
passages that correspond to those in the Nicomachean Ethics, but they also occur in places
Finally, the laying down o f hypotheses is an activity characteris­ where there is no mention o f it in the latter work. ’ T h e following list o f E E and A E references
tic o f the E E rather than o f the N E . T h e G reek word corresponding documents this:
to ‘lay down’ is used in this technical sense only once in the N E E E I: i 2 i 4 bi4 - 2 5 Constituents o f happiness to be distinguished from con­
ditions o f happiness, as constituents o f health from con­
( n o 4 b27); it is used nine times in the E E and once in the A E
ditions o f health.
( 1 1 39a6) in addition to the present passage.2 T h e word ‘ hypothesis’ I 2 i4 b34 Comparison between medical and ethical chastisement.
in its technical sense is not used in the N E ; in the E E ( i2 2 b28, i2 i 6 bi7 -2 5 Ethics like medicine a practical science.
I 2 i 8 bi 6 ff. M edicine defines end and proves goodness o f m eans; does
I2 35b3 °) it is5 ar,d the use of hypotheses in mathematics is often
not prove goodness o f end (health).
compared to the role o f the goal in practical reasoning (i2 2 7 a9, b29) EE II: i22o a2~3 Virtue like ε υ ε ξ ία is a whole consisting o f parts.
as it is in the A E ( 1 1 5 i a 17 ; cf. ii4 4 a24). 12 2 0 2 5-3 8 Comparison between genesis o f ε υ ε ξ ία and o f virtue.
i2 2 2 a30~40 Analogy between medicine and ethics with regard to the
A fter this introduction, two preliminary points are m ade: a state
mean.
o f character is not like a science, which can lead to opposite results; 12 2 6 *9 -15 Neither in medicine nor in ethics do we choose ends.
instead, it has a contrasting opposite, the study o f which may throw 1226*35 Doctors deliberate but not spellers; two types o f error in
medicine.
light on itself. So in this case, from a plurality o f senses o f ‘injustice’
122 7 27-30 Health and disease both objects o f medicine in different
we can conclude to a plurality o f senses o f ‘ justice’ . T o illustrate ways.
these points, medical analogies are used: health is not like a science i227b27-33 In medicine and other τ ε χ ν α ι the end is the principle and
either (12 29 “ 1 5 - 1 7 ) ; studying bad bodily condition will tell us hypothesis.
E E III: I228b35 ff.,
about good bodily condition ( 1 229s 19 -2 3). T h e close use o f medical 122<)b20 ff. Comparison between courage and health.
analogies for ethical topics in this way is particularly characteristic A E A: 12 2 Ç3 1 2 ff. Points about opposites illustrated by health and ε υ ε ξ ία .

o f the E E and A E . M edical analogies o f course occur so frequently ii37ai4-27 T h e doctor’s knowledge compared with knowledge of
justice.
1I3^a3° Comparison between the mean in justice and in medicine.
1 . Burnet, ad loc., recognizes that the reference m ust be to the use o f έ ν δ ο ξ α . Others take AE B: ii38b30-2 ‘What right reasoning prescribes’ no more helpful than
τ ή ν α ύ τ ή ν μ έ θ ο δ ο ν τ ο ΐς π ρ ο ε ιρ η μ έ ν ο ι? to mean n o t‘the method described in the previous ‘what medicine prescribes’ .
discussion’ but ‘the method used in the previous discussion’ . T h e difference is not o f great ii4ia22-33 Ethics, like medicine, is species-oriented.
importance : the introduction o f justice here parallels the introduction o f courage at 1228 a28 1 141 bi 4 ff. Value o f knowledge o f particular in medicine.
in the E E just as it fits the description o f method in the E E . W hen E E I says that we must use H43b2i~32 Apparent uselessness o f medicine and wisdom compared :
the evidence o f the φ α ιν ό μ ε ν α it means o f course the opinions o f others, not ‘the facts’ ; medicine neither sufficient nor necessary for being healthy.
see Owen, 19 6 1. i i4 4a4 Wisdom comparable not with medicine but with health.
2. In the E E ύ π ο κ ε ίσ θ ω and ν π ό κ ε ιτ α ι are used in the technical sense at I 2 i8 b3 7 ; U45a7 Relation o f wisdom to philosophy is like that o f medicine to
I 2 I 9 a IO, a2 9 , b2 8 ; I 2 2 0 a7 , a2 2 ; I 2 2 2 a6 ; I 2 2 3 b3 0 ; I 2 2 7 b2 5 · health.
64 The Aristotelian Ethics
Aristotle’s final methodological remark before proceeding to
the examination o f justice is that if one o f two opposites has a
plurality o f senses, so does the other. T h is is applied in the case o f
justice : there is more than one kind o f injustice, therefore there is
more than one kind o f justice. T h e practice o f distinguishing
between the senses o f words, and asking in how many different
senses words are used, is, along with the technical vocabulary which
goes with the practice, more characteristic o f the E E than o f the
N E ; the former, it has long been observed, is more interested in the
technicalities o f logic ( i i2 g a2 ^ -6 ) .1
H aving distinguished between ‘ just’ as ‘equal’ and ‘just’ as
‘according to the law ’, Aristotle defines the subject matter o f the
first sort o f justice and injustice as ‘ goods subject to good and bad
fortune, goods which taken without qualification (haplös) are always
good, but for a given individual are not always good’ (i i29 bi-6 ). H e
at once abbreviates this description to the expression ‘ta haplös
agatha', literally ‘ the without-qualification goods’ . T h is does not
mean that they are goods, no matter what may happen, or goods,
come what may ; just the opposite in fact : it means goods which are
good in the absence o f special circumstances to upset their
goodness. From the contexts in which Aristotle uses the expression
it is clear that ‘prima-facie goods’ or ‘things good in the abstract’
A E C: i i 47 a5-8 Exam ple o f medical syllogism.
i i S ° b32 - 5 Com parison between evil ethical states and types o f disease.
E E V II: I 2 35 b32_a7 D istinction between goods simpliciter and others illustrated
from medicine.
i2 3 0 ai8 -2 4 H ealth taken as prim e analogate for analogy.
I2 3 7 b22 W ish not enough for health nor for friendship.
E E V III: I248bi4 Virtue and health wholes consisting o f parts.
! 248b34 Wealth no more benefits vicious than full diet an invalid.
I2 38 b2 2 -5 T h e standard by which the doctor judges what is healthy
I. T h e following lexical facts give an indication o f this. T h e expression π λ ε ο ν α χ ώ ς
λ έ γ ε σ β α ι occurs only once in the N E ( i i 2 5 bi4). In the much briefer A E it occurs twice
(112 9 * 2 4 ; H 4 2 bi7 ) and in t h e £ £ five times (i2 3 0 bg ; I2 34 bi9 and 20; 1236 *7; i2 4 7 b28). Its
opposite μ ο ν α χ ώ ί λ ί γ ΐ σ θ α ι occurs only in the E E (i2 3 4 bi9).
A synonymous expression, ττολλα χώ ς λ έ γ ε α θ α ι occurs in the A E ( i i 3 6 b29) and in the
E E ( i 2 i 7 b27) but not in the N E .
A particular instance o f plurality o f senses, 8 ιχ ώ ς λ έ γ ε σ θ α ι, occurs twice in the A E
(1 i4 0 b3 1 ; H 5 2 b27) and twice in the E E (12 17 * 3 6 ; I228 bi8) but never in the N E . Iσ α χ ϋ ς
λ έ γ ε σ θ α ι occurs once in each o f the N E (1096*24) and the E E ( ΐ2 ΐ-;°ζ η ) . T h e question
π ο σ α χ ω ς λ έ γ ε τ α ι is raised only in the A E ( i i 2 9 a3 i).
Again, ό μ ο ν ΰ μ ω ς and its cognates, which occur three times ill . 11·. A alone, and from time
to time in the E E (e.g. 12 3 6 * 17 , b25) appear only once in the whole N E (i096b27).
σ υ ν ώ ν υ μ ο ς occurs only in the A E (113 0 *3 3 ). See below, p. 140.
The S tyle o f the Treatise on Justice 65
would be a more accurate translation than ‘absolute goods’
‘ unqualified goods’ ‘ intrinsic goods’ ‘things absolutely good’ .1
T h e prima-facie goods are mentioned often in the E E and in the
A E but never in the N E .2 For instance, we are told that ‘ Some
goods are good in the abstract and others good for individuals
without being good in the abstract; and the same things are good in
the abstract as are pleasant in the abstract’ (i2 3 5 b32 ff.), and a
comparison is developed between those things which are good and
pleasant to a healthy body and those things which are advantageous
and pleasant to an invalid. T h e final book o f the E E contains a long
discussion o f the prima-facie goods; it uses the expression ‘ natural
goods’ (physei agatha) as synonymous with ‘ goods in the abstract’
(haplös agatha). (T he synonymy is most obvious from the passage
I249a8 - i7 , where also ‘exterior goods’ is used as third synonymous
expression.) There we are told that if a man is good then the things
which are good in the abstract are good for him as an individual also
(i2 4 8 b27); and we are given a list o f these prima-facie goods:
honour, wealth, the good qualities o f body, good fortune, and
pow er.3 These, we are told, are good by nature but can be injurious
to those o f bad character.4 These goods are contrasted with virtuous
character and virtuous activity (i2 4 8 b37) which are the finest class
o f goods and cannot be misused. A man o f common goodness may
pursue the virtues for the sake o f the natural goods; the man of
supreme virtue (the kalos kagathos) chooses natural goods for the
sake o f virtue (i2 4 9 a6 - i5 ) .5
T h e prima-facie goods, then, play an important part in the
systematic development o f the theory o f virtue and goodness in the
1. Rackham (Loeb E E ) and Ross (Oxford N E ) translate ‘things absolutely good’ . G ran t’s
‘W hat are abstractedly goods’ is more correct, but seems a little clumsy.
2. T h e nearest to the concept is at i i 5 7 b27 S o k €Γ γ ά ρ φ ιλ η τ ό ν μ ε ν κ α ί a ip e r o v το
ά·7τλώ5 α γ α θ ό ν η ή δ υ . T h e N E does not talk o f a class o f α π λ ώ ? ά γ α θ ά nor does the notion
o f α π λ ώ ς ά γα θ'ον here mentioned play any part in the systematic development o f the N E
theory o f goodness.
3. T h e list which the Scholiast gives in commentary on the passage o f A E A — health,
strength, beauty, good bodily condition, wealth, reputation, and political power—
corresponds very accurately to the list in the E E , if we take the first four as instances of
‘good qualities o f body’ (C A G xx.208).
4. T h is is a different point from that made in the N E that goods can be dam aging: for one
o f the examples given there (i094bi9) that men have died because o f their courage is an
example o f a good which is not just prima facie good but κ α λ ό ν according to the E E
classification.
5. Other places where the E E uses the terminology are I228 bi8 ; i2 3 0 b34; i2 3 8 a3 ff., b5 ;
66 The Aristotelian Ethics The Style o f the Treatise on Justice 67

E E . But, as Grant showed long ago, the notion is widespread T o show that the law covers the whole field o f conduct covered
throughout the disputed books also: at i i3 4 aio we hear o f ‘what is by the virtues already considered, Aristotle says :
useful in the abstract’ ; at i i4 3 b3 we are told that the just ruler does
The law commands the doing of the acts of a brave man (e.g. not leaving
not take more than his share o f prima-facie goods for himself, and at
one’s post, not running away, and not throwing away one’s arms) and
i r 37a20 we are told that justice exists among people who share in those of a temperate man (e.g. not committing adultery, not being wanton)
prima-facie goods. T h e distinction between goods in the abstract and those of a meek man (e.g. not committing assault or defamation) and
and goods for individuals is made the basis o f the rebuttal o f similarly with regard to other virtues. ( 122t)biy- 23)
Speusippus’ attack on pleasure in Book A E C (i i5 2 b2 5 ).1 We hear
F o r our present purposes the interest o f this list lies in the selection
also o f things which are bad in the abstract ( i i2 9 b8 ; i i5 3 b2).2 This
o f these three particular virtues as paradigms. It is in the E E that
terminology seems still to be one o f the strongest indications o f a
these three virtues have pride o f place (E E I I I , chapters i, 2, 3 ; also
link between the disputed books and the Eudemian Ethics.
i i 2 i ai5 - 2 3 in book II). In the N E both the preliminary list
Having used the notion o f prima-facie goods to distinguish the
(1 i0 7a2 8 -i io8bio) and in the full-fledged treatment o f the virtues
subject matter o f justice in the sense o f equality or fairness, the
in books 3 and 4 meekness appears well down the list; and so far
justice which contrasts with greediness (pleonexia), the taking o f
from being a paradigm it is twice described as a virtue that is strictly
overlarge shares, Aristotle turns to the wider sense o f justice in
nameless ( n o 8 a5, i i 2 5 b26).1
which it is obedience to law. In this sense, he says, ‘ we call “ just”
O f the justice that is law-abidingness Aristotle goes on to say that
those things that produce and preserve happiness and its com­
it is (1) perfect virtue, (2) the use o f perfect virtue, (3) using one’s
ponents for the political society’ (i i2 g bi 8). Once again a tiny point
virtue in regard to others, (4) not a part o f virtue, but total virtue.
o f terminology gives an indication o f the rightful place o f the
E very one o f these expressions is a Eudemian turn o f phrase, as I
treatise on justice. In speaking o f ‘happiness and its components'
shall now proceed to show.
Aristotle is here aligning him self with the Eudem ian view,
At E E 12 19 ^ 8 , when Aristotle has reached his definition o f
according to which happiness is activity in accordance with all the
happiness, we read :
virtues, moral and intellectual, so that each type o f virtuous activity
is a component o f total happiness, and against the Nicomachean Since happiness, we saw, is something perfect, and life can be either
view o f happiness as the activity o f one supreme virtue.3 perfect or imperfect, and virtue likewise (for there is total virtue and
partial virtue) and since the activity of imperfect things is imperfect, it
1. See Festugière, 1946, L X X .
follows that happiness is the activity of perfect life in accordance with
2. G rant notes ‘ It is added that “ men pray for these and seek after them, but they should
not; they should pray that the absolute goods may be goods to them individually, and that perfect virtue.
they should choose what is good for themselves.” T h is is in the same style with Eth. Eud.
Perfect virtue, then, is the totality composed o f the individual
V I I .x ii.1 7 : το ζ η τ ΐ ΐ ν κ α ι eϋ χ ε σ θ α ι π ο λ λ ο ύ ς φ ίλ ο υ ς. But to say what men “ ought to pray
fo r” is not after the manner o f A ristotle.’ Prayer is mentioned in the N E , once, in the not very virtues as parts: it is what is called later in the E E kalokagathia
serious context o f the man who prayed to have a crane’s gullet ( 1118 * 3 2 ) ; but the idea that the (i2 4 9 ai8). Legal justice, he here says, is identical with this perfect
desirable situation is to have the prim a-facie goods be good for oneself individually parallels
E E 12 4 9 *13 even more strikingly than the Eudem ian passage quoted by Grant.
virtue: or, to be more accurate, it is a particular aspect o f perfect
3. O f the twenty uses o f μ έ ρ ο ς , μ ό ρ ιο ν in the N E six are for parts o f physical objects, four o f virtue, its utilization in regard to others ( 1 i29 a26, 31). T h e ability to
parts o f the soul, four o f parts o f motion, two o f parts o f a community, three in expressions make use o f virtue in regard to others is itself added perfection; it
like κ α τ ά μ έ ρ ο ς . T h e N E nowhere speaks o f parts or components o f virtue, vice, or happiness.
T h e nearest it comes to doing so is at 114 0 * 14 (parts o f the common good) and u 8 0 b30
gives another sense in which justice is ‘ perfect’ virtue ( ii2 9 b3 i) .2
(science o f law-giving part o f science o f politics). T h e A E on the other hand constantly speaks
o f parts o f virtues and vices ( 1 13 0 a8, “ 14, *23, *34, bi 1 ; 1 1 3 1 * 1 ; 1144*5) as well as o f sciences 1. See above p. 23 and below p. 232.
( I I 4 I ’ I 3 i I I 43 a3 )· I " the E E we often hear that happiness is a whole o f parts ( i 2 i 4 b5 ; b28; 2. Assuming that the ό'τι in line 1 129b3 i is correct (despite the disagreement o f K b ) the
12 16 *4 0 ; I 2 i9 bi4 )an d parts o f virtue(s) are mentioned at I 2 i6 b6; 1219 *38 , “2 3 ; 12 2 0 * 3 ,“ 14 ; reason there given for the perfection o f justice is not the same as the one presupposed by lines
parts o f vices at 12 3 1* 2 2 ; 12 3 2 *13 . T h e E E and the A E are altogether happier than the N E 2 5 - 3 1 ; cf. the Scholiast, C A G xx.20 9 .31-210 .8 which makes much more sense than the heavy
about applying mereology to abstract objects. See below, p. 126. weather made o f this passage by most commentators, cf. G au th ier-Jo lif (19 59 , 342).
 68 The Aristotelian Ethics The Style oj the Treatise on Justice 69

T h e notion o f the use or utilization o f virtue is a Eudem ian one. chapter o f the A E in the same detail would be tedious and
T h e Greek word for ‘use’ is employed in the N E overwhelmingly unrewarding: each chapter, however, can be made to yield the same
for the use o f physical objects (eighteen uses out o f a total o f twenty- minute indications o f the original home o f the disputed books.
five). W e hear o f one science using another (io94b4) and there are Rather than proceed with a line-by-line examination of the style o f
references to our making use o f our senses ( n o 3 a3o ; n i 8 a28) of the A E , I shall present the evidence in a cruder but briefer manner
pleasures ( n o 5 ai2) o f irony ( i i 2 7 b3o). Only once in the whole in the form o f tables exhibiting the results o f vocabulary counting.
work do we hear o f the use of virtue ( i i 7 9 b3). In the Eudemian T h e presentation will be cruder in this way in that it will not permit
Ethics, on the other hand, the notion o f the use o f a hexis is account to be taken o f differences between senses o f the same lexical
ubiquitous ai6 , 18, 2 5 ; I 2 i9 b2, b4 ; i220 a3 3 ; i 225bi 2 ; item ; but it will have the advantage o f providing data sufficiently
1 22 7a23) and there are whole sections devoted to the topic o f the use copious and sufficiently well defined to enable statistical methods of
and abuse o f sciences and virtues (i2 7 ai ff. ; i246a26~b36).1 argumentation to be applied. T h e crudity o f the classification will
It has already been shown that the use o f the part-whole do no harm : it may prevent the detection o f stylistic differences
terminology in connection with virtue is another Eudemian between texts where a finer analysis might enable us to discri­
characteristic. T h u s the chapter ends, as it began, with Eudemian m inate; but it cannot have the effect o f producing imaginary
echoes.2 distinctions where there are none in reality. If, then, by vocabulary
T h e close examination, then, o f this first chapter o f the disputed counting and statistical analysis o f a simple kind there appear clear
books, has revealed that almost every sentence contains tell-tale differences between the A E and one of its competing environments,
indications that the original environment o f the books is the it is unlikely that later more refined analysis will overturn any
Eudem ian and not the Nicomachean ethics. T o go through every conclusions thus reached.
1. In the preliminary N E discussion o f happiness we read: ‘ It makes no little difference
whether we place the suprem e good in possession or in use, in state or in activity’ ( ioq8b32 f.).
I f these two pairs o f alternatives are meant to mark a single contrast we have here the
terminologies o f the E E and o f the remainder o f the N E brought together. An act o f a
particular virtue in the E E will be described as an activity o f that virtue ( τ ή ς α ρ ε τ ή ς
ε ν έ ρ γ ε ια ) or as a use o f that virtue (passages listed in text). T h e N E 's preferred expression is
activity in accordance with virtue’ (έ ν ε ρ γ ε ια κ α τ ’ α ρ ε τ ή ν ). T his expression is o f very
frequent occurrence in the N E (e.g. 1098*7, " 1 7 ; i0Q9b26; U 7 4 bi8 ; i i 7 0 b26; H 7 7 a9 ~ i 2 , 18,
24) and does not occur once in the E E . (We do however hear o f activity in accordance with
passion, I2 2 0 bi7 , I2 2 3 a39.) Whereas in t h e A ^ normally a person is said to act in accordance
with his hexis, in the E E his life, or his hexis acts ( 1 2 1 8 * 3 1-3 , 38, b2 ; 1220*8; cf. N E
But altogether the N E is much fonder o f the terminology o f activity ( ε ν ε ρ γ ε ια ε ν ε ρ γ ε ΐν )
than the E E : see below, p. 142.
H ere as usual the A E follows the E E usage. T h e activity words occur only seventeen times,
thirteen o f these in the section on pleasure in A E C. In the m ajority o f cases the subject o f the
ε ν ε ρ γ ε ια is a έ ξ ις , not a person (1144=6; 114 7 * 7 ; I I 52b3 5 ; 115 3 * 14 , bio ; ε ν έ ρ γ ε ια o f life,
1 r 54 a7 )· T h e expression ε ν έ ρ γ ε ια κ α τ ’ α ρ ε τ ή ν is not used. Only in one place is there an
echo o f the Nicomachean term inology: U 3 0 ai 7 : κ α τ ά μ έ ν γ α ρ τ α ς ά λ λ α ς μ ο χ θ η ρ ία ς ό
ε ν ε ρ γ ώ ν α δ ικ ε ί.
2. T h e sentence έ σ τ ι μ έ ν γ α ρ ή α υ τ ή , το δ ’ ε ίν α ι ο ν το α υ τ ό ( i i 3 0 ai2 ) does not give
much help for our investigation. T h e only other sentence o f this form in the ethical writings is
also in the disputed books r I 4 i b24. Though the precise way in which the sentence should be
read is not clear, its general sense is clear from the many other passages in Aristotelian
writings where it occurs (listed by G au th ier-Jolif, 19 5 9 ,34 3). ju stice and virtue are the same
hexis, considered in two different relationships ; justice and virtue are the same thing, though
what it is to be just is not what it is to be virtuous ; in a m ore modern terminology, ‘ justice’ and
‘virtue’ have the same reference but not the same sense.

CHAPTER 4
Particles and Connectives in the
Aristotelian Ethics
T h e problem presented by the books common to the Nicomachean
Ethics and the Eudemian Ethics is one which particularly invites
investigation by stylistic methods. It is natural to inquire whether
the three books which make a double appearance in the manuscript
tradition, as books 5, 6, and 7 o f the Nicomachean Ethics and as
books IV , V , and V I o f the Eudemian Ethics, were originally written
for one context rather than another; and one obvious way of
undertaking such an inquiry would be to ascertain whether in style
the common books show a greater similarity to one or other o f the
contexts in which they appear. Surprisingly, although scores o f
Aristotelian scholars have written about the problem o f the
common books and have pronounced confidently on their prov­
enance, no systematic study o f their style or comparison with their
Nicomachean and Eudem ian contexts has ever been m ade.1 T he
present chapter contains the beginning of such a study : it confines
itself to a single aspect o f style, vocabulary choice; and within
vocabulary choice it concentrates upon a small fragment of
vocabulary, namely the use o f particles and connectives. E ven in
such a small area, it will be found, it is possible to detect very
striking differences between the degrees to which the common
books resemble the competing environments offered by tradition.
Particles and connectives offer a number o f advantages as a
subject o f stylistic study. Students o f Greek style have a
considerable advantage over students o f English style in that Greek
is rich in brief words which express nuances o f thought-connection
and mood which in English are more likely to be expressed by the
tone o f voice or the tilt o f the eyebrow. T h e frequency o f such
particles is not affected by variations in subject matter as the
i . T h e history o f scholarly discussion o f the disputed books can be conveniently studied in
Schächer, 1940, and in R ow e, 19 7 1, chapter I I I .
Particles and Connectives in the Aristotelian Ethics 71
frequency o f most nouns, verbs, and adjectives is; moreover they
occur with sufficient frequency and regularity to enable standard
statistical techniques to be used to describe and draw inferences
from their distribution. Indeed, as we shall see, about one-quarter
o f any Aristotelian text consists o f uses o f these comparatively
insignificant and topic-neutral words. M oreover, such studies as
have been carried out suggests that the use o f particles and
connectives in Greek prose may vary in characteristic ways from
author to author.
How far authors are consistent in speech habits such as
vocabulary choice is a matter o f keen debate : fortunately the scope
o f the present inquiry does not necessitate any participation in the
debate. N either here nor elsewhere am I questioning the consensus
o f scholars that both the Eudemian and the Nicomachean Ethics are
genuine works o f Aristotle; nor am I putting forward any
hypothesis about the fixity or fluidity o f his stylistic habits. I am
merely attempting to discover and compare the regularities to be
observed within the undisputed Nicomachean books, within the
undisputed Eudem ian books, and within the books which are
disputed between the two. B y seeing how far the regularities in
the disputed books resemble the regularities in each o f the treatises,
we may hope to obtain an indication o f the context to which they
originally belonged without, for present purposes, making any
assumptions about the chronological order o f the N E and E E , or the
temporal distance which separates them, or the process by which
books belonging in one context became included also in another
context. Speculation on all these matters should follow, and not
precede, an unpartisan study o f the texts o f the disputed books and
their rival contexts.
Though it is surprising that scholars have so frequently
attempted to settle the provenance o f the A E without a serious
study o f their style,1 it is not surprising that the particular features
o f style studied in the present chapter should not have been
investigated. T h e counting o f particles and connectives by hand is a
tedious and inaccurate business: the availability o f machine-
readable texts o f Aristotle, o f computer programs for word-counts
i . An honourable exception here should be made for S ir Alexander G ran t’s edition o f the
Nicomachean Ethics (London, 1885), which contains an excellent and almost comprehensive
index verborum— the work o f John K eb le— and many judicious observations on particular
features o f the style o f the common books (which he believed to belong to the Eudemian
Ethics, then regarded as Eudem us’ work).
72 The Aristotelian Ethics
and concordances, and the development of statistical theory and
techniques during recent decades make stylometric studies incom­
parably easier to perform, and potentially more fruitful in outcome,
than they were in the 1860s when R u d o lf Eucken wrote his
dissertation De Aristotelis dicendi ratione (Göttingen, 1866) which,
to the best o f my knowledge, is the only previous study o f the
Aristotelian vocabulary which is the subject matter of the present
chapter.
Eucken studied Aristotle’s use o f particles in the Metaphysics,
Physics, de Anim a, N E , Politics, Rhetoric, and Poetics. H e wrote at a
time when the almost universal opinion o f scholars held the E E to
be the work o f Aristotle’s disciple Eudem us; but he studied the
usage o f that work, along with Theophrastus’ Characters and a
number o f other pseudo-Aristotelian works, in order to investigate
how far Aristotle’s style differed from that o f other members of his
school. His work was divided into eight chapters, each devoted to a
single particle or group o f particles ; his conclusion was that there
was an astonishing variety o f usages and constructions to be found
in the Aristotelian treatises, which he hoped might be someday o f
use in settling the order o f their composition. O f most interest in
our present context is an appendix: ‘Excursus de discrimine
dicendi rationis, quod in libris Ethicis invenitur.’
T h e individual books o f the Ethics, he maintained, differed so
much from each other in particle use that they could hardly have
been written in a single stint by Aristotle. T h ey fell into four
groups: books 8, 9, and 10 resemble each other closely; furthest
from them in style are books 5, 6, and 7, which again form a unified
group. T h e early books stand between, with the first two
resembling 5, 6, and 7 more than they resemble 9 and 10, and the
third and fourth most resembling the final books. T h e final books
are written in a fluent and elegant style, less argumentative than
Aristotle’s usual manner : thus epei, which is very common in all
Aristotle’s other works, is rarely found (once in 8, twice in 9, once in
10) ; in book 7 it occurs twenty-five times and nearly as commonly in
books 5 and 6. In the early books, it ranges between six occurrences
in book 2 and one in book 4. Sim ilarly koste : four times in book 8,
four in book 9, five in book 10, contrasts with twenty-two times in
book 7, twenty-three in book 5, and seventeen in book 6. Books 1
and 2, he said, to some extent resembled the middle books (though
hoste was much less frequent in them) and books 3 and 4 resembled
Particles and Connectives in the Aristotelian Ethics 73
the final books. In the books in which it rarely appears, Eucken said,
it could be said to be replaced by dl\ indeed, in books 8, 9, and 10 de
is used with a frequency which marks these o ff from every other
work o f Aristotle’s. K a i de and kai de are used as conjunctions in the
final books with an unparalleled frequency: in book 9 the former
expression occurs eleven times and the latter nine times ; whereas in
the three books of the A E they occur only five and two times
respectively. T h e use o f te also, particularly when used without any
other conjunction, is a characteristic o f the final books; a mark,
Eucken says, o f their quasi-poetical character. Te . . . te \s to be
found in the first and final books, not in the A E ; kaitoi, mentoi, and
alia men are more frequent in the A E than elsewhere; whereas ou
men, occurring once only in the A E , occurs in the third, fourth and
tenth more frequently than in any other work o f Aristotle.
T h e conclusions which Eucken drew from these striking
observations were modest. Unusual though the stylistic features o f
the later books o f the N E were, there was no reason to regard them
as alien to the philosopher’s style : it was simply a matter o f particles
frequent elsewhere being rare here, and usages elsewhere rare being
here frequent. T h e most that should be said was that the ten books
of the Ethics were written at different times and only later brought
together. No conclusion could be drawn about the disputed books,
because the style o f the Eudemian Ethics differed hardly at all from
that o f genuine works o f Aristotle ; there was certainly no reason to
deny the disputed books to Aristotle once one accepted that the N E
did not come from Aristotle in the order in which we now have it.1
It was a pity that Eucken, having noticed striking differences
between the A E and other books in the Ethics, did not compare the
A E features with the E E to see whether they were not actually
closer to the E E in those points in which they stood out. H e was
deterred from doing so by the prevailing scholarly dogma o f the
time, that to attribute the A E to the Eudemian Ethics would be to
deny that they were genuinely Aristotelian. T h e success o f W.
Jaeger in the present century in convincing scholars that the
Eudemian Ethics are genuine— though not based on any close study
o f the evidence comparable to Eucken’s— has enabled us to look at
i. ‘Quod si statuimus quam facile fieri potuerit, ut libri V , V I, V II et Ethicis Nicomacheis
et Eudemeis insererentur apparet, utri autem scripto magis apti sunt (de qua re homines docti
valde dissentiunt) ex dicendi ratione vix dijudicari potest, quia Ethicorum Eudemeorum
sermo fere non distare videtur ab Aristotelis ipsius scriptis’ (op. cit. 77).
 The Aristotelian Ethics Particles and Connectives in the Aristotelian Ethics 75
Occurrences o f thirty-six Particles and
T a b l e 4 .1. the question with a more open mind. W e shall see that even in the
Connectives in the NE, EE, and AE narrow field o f particle choice there is a great deal o f evidence which
confirms Eucken’s observation that the A E stands out from the rest
o f the Nicomachean Ethics. We shall find also that the Eudemian
NE AE EE
Ethics are not as internally inconsistent as Eucken observed the N E
αλλά (inc. <xAA’) 254 214 31 1 to be.
άν (inc. καν) 363 97 181
T h e first step in our inquiry is to discover the number o f
äpa (inc. dp’) 36 31 31
γάρ ΙΟΙΙ 368 occurrences o f the most frequent particles in the N E , E E , and A E .
577
ye (inc. y ’) 76 20 T able 4 .1 sets out the results o f such a count. T h e counts were made
38

OO
Se (inc. δ’) 723 1034 on the Oxford I C L i960 A computer by the C O C O A concordance
8η 223 46 75 and word-count program, from machine-readable texts o f the
διό 69 44 71 Ethics prepared by the Thesaurus Linguae Graecae o f Irvine,
i àv 31 17 15 California ; the editions used were B y water’s Oxford Classical T ext
>
€i 181 112 168 for the N E and A E , and Susem ihl’s Teubner edition for the E E .
eïirep 25 15 4 Each entry in T able 4 .1 records the number o f tokens o f a given
67T£l 20 42 60 word-type to be found in the text o f the N E (undisputed
€TL 57 48 45 Nicomachean books), A E (disputed books), and E E (undisputed
y 446 182 3X9 Eudemian books) respectively. (T he distinction between word-
καθαττζρ 70 H 10
tokens and word-types is familiar to philosophers: tokens are
κ α ί 2633 994 1633
counted as a typographer would count, and types as a lexicographer
μ έν 591 325 451
μ ,έ ν τ ο ι 2 7 17 would count, so that ‘Break! Break! B reak!’ contains three tokens
Μ 247 144 230 but only one type.) T h e final line o f the table gives the total number
μ ,ή ν 25 II 19 o f tokens occurring in the respective texts. A glance at the table
οθεν 34 I 7 confirms Eucken’s observations, as that de is very much more
ο ΐο ν 136 70 107 frequent in the N E than the A E , and that höste occurs more often in
δταν 34 47 49 the A E than in the N E ; and it shows that in these two respects, and
δτ€ (inc. oV) 34 5 28 others, the A E feature is reproduced in the E E . Som e o f the
o n 163 133 192
expressions which were discussed by Eucken, however, do not
ον (inc. ου/c, ουχ) 534 288 380
figure in the table: a case in point is the conjunction prin. These
ονδέ (inc. οι5δ’) 232 79 113
οΰν 92 absent expressions are ones which are too infrequent to enable a
235 97
OVT€ 82 statistical argument to be based on them. Even if we assume that an
73 74
πότβρον 22 II 21 author is totally regular in his speech habits, so that any difference
πώ ς 47 17 28 in the number o f occurrences o f a given word between two equal­
r e (inc. τ ’) 162 56 91 sized samples o f his prose is totally explicable by chance, we still
τ ο ίν υ ν I 4 17 need a sample large enough to contain at least ten occurrences o f the
ώ ς inc. ώ?) 203 67 141 word before we can use the sample to form a reliable estimate o f the
ώσπερ 98 80 123 frequency with which the author uses the w ord.1
αιστβ 38 61 96
Total words in text 39525 17041 26330 i. T h e reason is well explained by Ellegârd : ‘ I f we postulate that the relative frequency o f
a word in an author’s usage is o oooi, or one in ten thousand, an actual sample (using this
76 The Aristotelian Ethics
In the present instance, our over-all strategy will be to endeavour
to predict, on the basis o f the undisputed Nicomachean books, how
many times a given word would occur in the disputed books if its
frequency there was the same as that in the Nicom achean; and to
predict on the basis o f the Eudem ian books, how many times a word
would occur in the disputed books if its frequency there was the
same as the Eudemian one; and then to compare these expected
frequencies with the actual ones and to inquire whether the
differences between the expected and actual frequencies are such as
are explicable by chance alone. F o r this comparison not to be
misleading, we must restrict ourselves to words occurring with
sufficient frequency to enable us to say— either on the basis o f the
Nicomachean evidence or on the basis o f the Eudemian evidence—
that they would be expected to occur ten times in the A E . Since the
N E contains 39525 words and the A E contains 17 0 4 1 words, a word
occurring at the same rate in the N E as in the A E would occur
(39525/1 7041) X 10 times, or 23· 19 times in the N E if it occurred
ten times in the A E ; and since the E E contains 26330 words, a word
occurring with equal frequency in the A E and E E would occur
(26330/17041) X io tim e s,o r 15-45 times in t h e £ £ if it occurred ten
times in the A E . Since words can occur only a whole number o f
times, we must restrict our consideration to words occurring either
twenty-four times or more in the N E , or sixteen times or more in
the E E .
On this basis a word such as prin (occurring twice in the N E and
thrice in the E E ) must be left out o f consideration. But even with
this restriction, we can find thirty-six particles and connectives to
compare. From one point o f view, this may seem an insignificant
part o f Aristotle’s vocabulary: the E E , for instance, has a
vocabulary o f 4750 w ord-types,1 so that the particles might be said
to make up only three-quarters o f 1 per cent o f its vocabulary. But
word to refer to consecutive pieces o f text also) o f ioooo words by this author may well
contain no instance o f the word at all, or even as many as threeor four. (T h is can be calculated
by means o f the Poisson form ula.) B ut this also means that a sam ple o f ioooo words is far too
small to yield a reliable estimate o f the actual frequency o f such a word. If, on the other hand
we take samples o f iooooo words, the observed frequencies will range between at least 4 and
at most 16 in 95 per cent o f the samples . . .. G enerally speaking, to arrive at a realistic
estimate o f the relative frequency o f a word we need a text sample which is about ten times as
big as the inversion o f the word's relative frequency’ (1962, 14).
i. T h is figure is based on an unlemmatized word-count in which, e.g. δ ιώ κ ω and δ ιώ ξ ω
count as two types.
Particles and Connectives in the Aristotelian Ethics 77
in fact a very large proportion of the text is made up o f particles: of
the 26330 tokens in the E E 6861, or 26 05 per cent, consist of
occurrences o f one or other o f these thirty-six words.
Looking at the raw scores recorded in T able 4 .1 is not by itself
very informative. Apart from the striking discrepancies o f the kind
already noticed by Eucken, the over-all impression is one o f great
similarity o f usage: a word like kai which is frequent in one o f the
three treatises is frequent also in the other two, and words like ge
and ara which are comparatively rare in one are comparatively rare
also in the others. T h is is, o f course, what one would expect in three
works o f the same author— and indeed some o f the features, such
as the extreme frequency o f kai, would be expected to be com­
mon to all writers o f Greek. Is there any way in which one
can discover whether the over-all pattern o f the A E use o f par­
ticles revealed by this table resembles more closely the N E or
the E E ?
A very simple way to calculate a measure o f relative resemblance,
which can be carried out by anyone with a pencil and paper, and a
minimum o f mathematical skill, is to calculate a coefficient o f rank
correlation. T h is statistic, which is designated by the G reek letter p
and called ‘ Spearm an’ s rho’ after its inventor, is familiar to many
people involved in education since it is frequently used in order to
measure the degree o f correlation between gradings and assess­
ments o f the same candidates by different examiners. It is a measure
o f how closely two rank orderings resemble each other. T o apply it
in the present instance the first step is to rank the particles in order
o f preference for each treatise— the most frequent particle being
given the rank 1, the second the rank 2, and the last the rank 36. The
results o f doing this are shown in T able 4.2. (Where two particles
occur the same number of times— as poteron and men in the A E —
both are given the same rank— in this case 3 1 , and the next most
frequent is given the next-but-one rank, in this case 33.) In this
table it is already more obvious that certain particles are more
popular in one treatise than another: an, ge, de, kathaper, hothen,
oun, for instance, are much higher up in the N E 's preference
ranking than in that o f the E E , while epei, hotan, hoti, hösper, and
haste are higher up in the E E list; and in almost all these cases the
A E ranking mirrors the E E ranking rather than the N E ranking.
B u t to get a precise figure for the over-all resemblances, we need to
apply the formula for calculating the Spearman rho, which will
 The Aristotelian Ethics Particles and Connectives in the Aristotelian Ethics 79
T a b l e 4 .2 Ranking o f thirty-six Particles and enable us to compare the rankings pairwise. T h e formula is
Connectives in the NE, EE, and AE
P N (N 2 - i )
NE ΑΕ ΕΕ
where D is the difference between the rank numbers o f any given
άλλα 8 6 7
particle and N the total number o f individual particles in the
âv 7 11 10 ranking (in this case 36 ).1 I f we compare the N E ranking and the
apa 26 25 20 A E ranking and go through it line by line, calculating the difference
γάρ 3 3 3 in each line (2 in line 1, for alia, 4 in line 2 for an and so on) and
γε 19 26 25 squaring the difference each time, and then add up all the squares
δε 2 2 2 and multiply the total by 6 we get the numerator o f the fraction,
δη 12 22 20
6 X (D 2 ), which is 6 x 734 = 4404. T h e denominator is 36 x ( 1296 —
διό 22 23 21
1) =4 66 20. R ho is therefore 1 —(4404/46620) = 0 906. I f we
εάν 30 27 29
calculate rho for the A E and the E E in the same manner, we get the
y
€l Η 10 II
result 0 970. Since rho can vary between 1, for perfect correlation,
e ïn e p 3 1 29 36
22
and o, for complete absence o f correlation, the calculation confirms
€TT€L 34 24
u
€TL 20 24
the impression that the three treatises correlate well together (as
23
6 6 would be expected on the assumption they are all by Aristotle) but
V 7
καθαπερ 21 30 34 that the A E resembles the E E more than it resembles the N E . But
καί I I I the difference between the two statistics is not sufficiently large for
μ εν 4 4 4 us to be sure that the greater resemblance between the A E and the
μ εν το ι 35 33 32 E E is not the coincidental result o f chance. T o investigate this we
μή 9 8 8
need to look more closely at the behaviour o f the individual
μην 30 3 ΐ 3 1 particles.
όθεν 28 36 35
Spearm an’s rho takes into account only the differences in ranking
οΐον 17 ι6 15
between the two lists. A more accurate measure o f correlation is
όταν 28 21 23
given by Pearson’s product-moment coefficient (r) which, applied
ΟΤ€ 28 34 27
ότι to the present material, takes into account not only the difference in
15 9 9
ού 5 5 5
the order of preference o f particles, but also the size o f the
ούδε II Η Η preference: it is based not on the simplified data presented in T able
οΰν ΙΟ 12 ι6 4.2, but on the full data presented in T able 4 .1, which gives not just
ούτε 20 Γ5 19 the rank order o f Aristotle’s particles in each treatise, but the actual
ττότ epov 33 3 ΐ 29 number o f times they are used. I f we calculate Pearson’s r for each
πώ ς 24 27 27 pair o f ethical treatises, for the thirty-six particles, we find that the
Τ€ ι6 ι8
!9 correlation is very good in each case,2 though closer between the A E
τοίνυν 36 35 32
and the E E than between the A E and the N E . r for A E and N E is
ώς !3 17 12
1. T h e procedure for calculating Spearm an’s rho can be found in any statistics textbook,
ώ σπερ ι8 ι3 13 e.g. A ry and Jacobs, 1976, 227 ff. An alternative method o f dealing with ties is to give the
ώ σ τε 25 ι8 17 value « + 0 - 5 to two items tying for rank n. With this procedure, and a formula available for
tie-correction, the value o f p for N E and A E is 0-9047.
2. T h e formula for Pearson’s r is not given : it is complicated, and in practice nowadays the
statistic is seldom worked out with pencil and paper. See A ry and Jacobs, 1976, 16 2 - 8 1.
8ο The Aristotelian Ethics Particles and Connectives in the Aristotelian Ethics 8i
0 9848, r for A E and E E is 0-9954. T h is very high correlation is in T h e relationships between the treatises can be represented
part due to the effect o f the five particles which are commonest in graphically. A scattergram is a graph in which a single dot or cross is
each o f the three treatises (kai, de, gar, men, ou). I f we remove these used to represent the position o f an individual in two dimensions :
particles from our list, and study the remaining thirty-one particles, the pattern formed by the dots illustrates the correlation. Each cross
we find that that the figures for r are now: for A E and N E 0-8015 ; in our scattergrams will represent a single word-type (e.g. alia).
for A E and E E , 0 9753. Each axis o f the graph will be a scale to measure the frequency o f

T a b l e αλ Scattergram o f Occurrences o f Particles and Connectives in the T A BLE 4.4 Scattergram o f Occurrences o f Particles and Connectives in the
NE and AE EE and AE
214 000 214000
X

/
/
/
/
/
/
/
/
/
171-200 /
/ 171-200 /
/
/
/
/
/ /
/
Q.
/ /
X
/
/ /
/ /
X
/ X /
128-400 / 128-400
/ /
/ /
y O /
x / X /
/ Γ3
/
a /
/

/ X
X /

/
/
X
/ x
X /
85-600 85-600 /
/
x / X /
/
X

x x/
X x/
/ X
/
/
X

/
/ *
/ x

/ / *
f
X

X
X
X

X
42-800 42-800 / -x x
* / ?
/ /
/ X /
/ /
/ X /
/ / x
X X /X
/ x x X
' XX *
/ x
/ X
0000 0000 'r χΧ
« e°
N E particles Λ = 0-8015 E E particles Æ= 0*9753
Standard deviation o f Y d ata against regression line is 31-46 Standard deviation of Y data against regression line is 11-60
82 The Aristotelian Ethics Particles and Connectives in the Aristotelian Ethics 83
that word in a given ethical treatise : if we wish to correlate the N E T a b L e 4 . 5 Frequencies o f Particles and Connectives in the NE, EE, and AE
and the A E , for instance, we can plot the N E frequencies along one
axis, and the A E frequencies along another. I f Aristotle’s use of NE AE ΕΕ
particles in the two treatises was totally consistent, there would be
άλλά frequency (%) 064 1-26 ι·ι8
perfect positive correlation between the frequencies in each treatise,
standard error 0-04 009 0.07
and the crosses in the scattergram could be connected with a single αν f. 0-92 0-57 o-6g
straight line. In fact, o f course, the crosses scatter to a greater or less s.e. 0-05 006 0-05
extent, spreading in either direction from a regression line— i.e. a âpa f. 009 0 18 0* 12
line which can be calculated on the basis o f the data as giving the s.e. 0-02 003 0-02
best possible estimate o f a word’s frequency in one treatise given γάρ f. 2-56 2 16 2·Ι9
information about its frequency in the other. T able 4.3 gives a s.e. o-o8 01 I 0-09
scattergram for thirty-one particles in the N E and the A E : each ye f. 0 19 0 12 ο ·ΐ4
cross represents an individual particle (they are not identified s.e. 0-02 0-03 0-02
Se f. 468 4-24
individually so as not to clutter the graph): the position of each 3’93
s.e. 0 11 0 15 0’ 12
cross is determined by the particle’s frequency in the A E , as given
δη f. 0-56 0-27 0-28
by the vertical axis, and its frequency in the N E , as given by the s.e. 004 004 0-03
horizontal axis. T h e dashes give the regression line around which διο f. 0 17 026 0-27
the actual values cluster: the standard deviation against the s.e. 0-02 0-04 0-03
regression line is a measure of the degree of their spread. Table 4.4 eau f. o-o8 0 10 ο·ο6
gives the same graphical representation o f the data for the A E and s.e. 001 0-02 0·0Ι
>
the E E . It will be seen that just as the correlation between the A E ei f. 0-46 0-66 ο·04
and the E E is higher than that between the A E and the N E , so the s.e. 0-03 0-06 ο·ο5
crosses in the A E graph are much less scattered and more €17T€p f. 006 0-09 002
compactly clustered around the regression line, with a much s.e. 001 0-02 οοι
> /
€7ret f. 0-05 0·25 0·23
smaller standard deviation.1 T h is brings out in a visible manner the
s.e. O-OI 0-04 003
closer relationship between the A E and the E E which the study o f
en f. 0 14 0-28 01 7
particle usage suggests. I f one wished to predict the frequency o f a s.e. 0-02 0Ό4 003
word in the A E , information about its frequency in the E E would be f. 113
V 1-07 1*21
much more useful than information about its frequency in N E . s.e. 0-05 ο·ο8 007
M ore informative than the raw scores for each particle are the καθάττ€ρ f. o-i8 ο·ο8 004
scores expressed as a relative frequency, as in T able 4.5. H ere the s.e. 0-02 0-02 ΟΟΙ
frequency scores are the ratio o f the number o f occurrences o f καί f. 6-66 5·83 6-20
the word-type in question to the total number o f word-tokens in the s.e. 0-13 ο·ι8 ο ·ΐ5
text. T h us, since kai occurs 994 times in the A E , which contains μ έν f. 15 0 i ‘9 1 ι·7 ι
17 0 4 1 words, its relative frequency is 0-0583. T o avoid a superfluity s.e. 0-06 ο·ιο ο·ο8
o f zeros, the frequencies are expressed in the text as percentages, so μ ίν τ ο ι f. 001 0-04 ο·ο6
s.e. o-oo 0-02 002
that the frequency o f kai is given as 5 83 per cent, and the figure o f
μη f. 0-62 0-85 087
ï. T h e graphs, and other sim ilar graphs later in this book, were produced by the B IV A R s.e. 0-04 ο·οη ο·ο6
packagc for bivariate analysis, written by M r. F . Pettit o f the O xford com puter teaching staff μ ην f. o-o6 oo6 00 7
as a statistical demonstration program. s.e. o-oi 0-02 002
84 The Aristotelian Ethics Particles and Connectives in the Aristotelian Ethics 85
T able 4.5 (contd.) certain word-type in an author’s text we are looking on each word-
token as a binomial event, that is as an event which may have one o f
NE AE EE
two possible outcomes : either the token is an instance o f the type in
O0€V f. 0-09 . 0*01 003 question or it is not; just as each toss o f a coin must be either heads
s.e. O-OI 0*01 001 or tails. T h e frequency given in the table records the proportion o f
otov c1. 0*41 0*41
0-34 such binomial events which had the first o f the two possible
s.e. 003 0*05 004 outcomes. T h e standard error is calculated in accordance with the
όταν f. 009 0*28 0 19
statisticians’ formula for the standard error of a proportion, σ per
s.e. 001 0*04 0-03
on cent = sjpqjn where sigma is the standard error, p is the proportion,
ore f. 0*09 0-03
s.e. O-OI O-OI 0*02 q = i —p, and n is the number o f events (in this case, word-
OTL f. 0-41 078 tokens in the whole text).
o*73
s.e. 0-03 0*07 0-05 We are treating each o f our texts as providing data about the
ού f. I 35 16 9 1*44 vocabulary usage o f its author at the time o f writing : each text is
s.e. 0*06 O'IO 0*07 taken as a sample drawn from the indefinitely large universe or
ουδέ f. 0-59 0*46 o*43 population o f G reek prose which Aristotle wrote, or could have
s.e. 0-04 0-05 0*04 written, consistently with his stylistic habits at the time o f writing.
ουν f. ° ‘59 0-54 o*37 What these habits were is something that we can now only discover
s.e. 004 0*06 0-04 by a study o f the texts themselves ; and we have to call in aid the
ovre f. 0'i8 0-43 0-31
statistical theory o f sampling in order to discover what inferences to
s.e. 0Ό2 0-05 003
features o f the larger population we are entitled to draw from
π orepov f. 006 0*06 0-08
0*02 0*02 features o f the sample text, and what degree o f error our
s.e. 001
πώς f. 0-12 O-IO 0Ί Ι generalizations are subject to. When we compare the A E with the
s.e. 0*02 0Ό2 0*02 N E and the E E in turn, one thing which we wish to discover is
Τ€ f. O4I 0-33 0-35 whether the A E and the N E , or on the other hand the A E and the
s.e. 0-03 004 0*04 E E , can be regarded as two samples from a single population.
τοίννν f. 0-00 002 0-06 Clearly, as T able 4.5 shows, there are differences between the
s.e. 0*00 O-OI 0-02 frequencies in the A E and the frequencies in the N E , and also
œs f. 0-51 0*39 0-54 differences between the frequencies in the A E and the frequencies
s.e. 0*04 0*05 0*04 in the E E : but we wish to know whether these differences, in one or
ώ σ περ f. 0*25 o*47 o*47 other or both cases, can be regarded as due solely to chance or
s.e. 0*03 0*05 0*04
whether they mark a genuine stylistic difference. F o r if any two
αιστε f. 0*10 0-36 0-36
0-04 samples o f any data are drawn from a given population, there will be
s.e. 0-02 0-05
some differences between the statistics o f the samples which call for
o· 12 against does not mean that 0-12 o f the text consists o f ge, but no other explanation than chance variation in selection, as hands
that 0 Ί 2 o f a per cent o f the text consists o f ge. dealt at bridge from the same pack by fair dealers differ from deal
Associated with each frequency given in T a b le 4.5 will be found a to deal.
standard error. T h is is a measure o f how much reliance can be T h e use o f the standard-error formula enables us to separate out
placed on the frequency in question as an indicator o f the author’s chance differences from ones which call for some other explanation.
general usage, on the assumption o f a certain regularity in his It is possible to calculate the probability o f two samples from the
vocabulary habits. When we are counting the occurrences o f a same population having statistics which differ from each other by
86 The Aristotelian Ethics Particles and Connectives in the Aristotelian Ethics 87
multiples o f the standard error. As a rule o f thumb, where the the A E resembles the E E considerably more than it resembles the
statistics o f two works differ by more than twice the sum o f the N E . O f the thirty-six particles studied, the A E frequency is closer
relevant standard errors, it is improbable that the differences to the E E frequency than to the N E frequency in twenty-seven
between them can be accounted for as chance variations among two cases, and closer to the N E frequency than to the E E frequency only
samples from the same population. T h is rule o f thumb, with the in nine cases. I f we apply our rule o f thumb, and ask whether the
calculations on which it is based, rest on certain assumptions about differences between the respective features are more than twice the
the nature o f the population, which in the case o f literary texts may sum o f the standard errors o f the proportions, the situation becomes
be questioned, and will be questioned later in the present study. But even more striking, as can be seen in T able 4.6. T h is sets out in the
for the moment, let us return to T able 4.5 to see whether a case o f each particle in turn whether the A E frequency ‘agrees
provisional judgement can be made as to how far the A E , the N E , with’ — i.e. differs by less than twice the sum o f the standard errors
and the E E can be regarded as samples from a single population from — the N E frequency and the A E frequency. It will be seen
exhibiting only chance differences, and whether the differences from the table that the differences between the N E and the A E are
between the A E and the N E are more or less significant than those in a majority o f cases— twenty out o f thirty-six— too great to be
between the A E and the E E . attributable to chance in the form o f sampling error. Whereas the
I f we look first at the frequencies alone, we can already see that differences between the A E and the E E in every case but two (eiper
and hote) are such as to require no explanation other than chance.
T a b l e 4.6 Agreements and Disagreements between AE, NE, and EE in T h is result is very striking: can we achieve a more precise
respect o f Particle Use evaluation o f the significance o f the result and o f the probabilities
involved ?
Particle Does Α Ε Does A E Particle Does Α Ε Does A E T o calculate the significance between two sampling proportions
agree with agree with agree with agree with statisticians proceed as follows. First, the value o f a statistic z is
NE ? EE? NE? EE? calculated by dividing the observed difference between the
proportions by the appropriate standard error o f the difference.1
αλλά No Yes μη No Yes T h e theoretical probability o f the difference between two pro­
>/
av No Yes μην Yes Yes portions o f samples reaching a particular z value is known in
âpa Yes Yes δθε,ν No Yes advance and can be ascertained from tables to be found in any
γά ρ No Yes oiov Yes Yes
statistical textbook: the probability o f a z value greater than 19 6 ,
ye Yes Yes όταν No Yes
δε Yes Yes ore No for instance, is 0 0 5 , or one in twenty; the probability o f a z value
No
8η No Yes ΟΤΙ No Yes greater than 2 58 is 0 0 1 or one in a hundred. I f the probability that
διο Yes Yes ού No Yes the difference arises from sampling error falls below an acceptable
id V Yes Yes ou8e Yes Yes level, then the null hypothesis, i.e. the hypothesis that the two
î
€L No Yes οΰν Yes Yes samples have been drawn from the same population, is rejected, and
eïnep Yes No ούτε No Yes
€7Τ€ί No Yes TTorepov Yes Yes I. The value of z is calculated from the data by means of the formula
»/
€TL No Yes π ώ ς Yes Yes
V Yes Yes τβ Yes Yes z = ---------- where p =
καθάπερ No Yes τοίνυν Yes Yes , + */«2 JÏ »!+»2
κ α ί No Yes ώ ζ Yes Yes In testing the difference between the N E proportion and the A E proportion p l is the N E
μ ίν No Yes ώ σπ ερ No Yes proportion, p 2 the A E proportion, «, is the number o f words in the N E (39525), n2 the
μ ίν τ ο ι No Yes number in the A E (17 0 4 1). T his formula, and the explanation o f the method o f testing the
ώ σ τε No Yes
significance between sample proportions, is derived from Caulcott, 19 7 3, 85.
88 The Aristotelian Ethics Particles and Connectives in the Aristotelian Ethics 89
the differences between them are described as statistically signi­ probability o f two samples coming from the same population is only
ficant. It has to be decided in advance what level o f probability is 0 00005, or one ' n 20000 : the values for an, dë, and hosper are all
regarded as acceptable : the decision is to a certain extent arbitrary between four and five. For z = 5, the probability is only 0-0000006,
and to a certain extent guided by the purpose o f the test (e.g. we less than one in a million and a half ; the £ values for alia, epei, hotan,
must ask whether it would have more serious consequences to reject hoti, oute are all above this. On the other hand, all but three of the
the null hypothesis when it should not be rejected, or to accept it differences between the A E and the E E are statistically insignificant
when it should be rejected). T h e two levels most commonly used at our chosen level. Eiper and hate had already been revealed as
are the 0 0 5 and 0 0 1 level: for our purposes the appropriate mavericks by the application o f our rule o f thumb : oun, where the
procedure is to test for significance at the 0 01 level (it is more difference previously appeared to be just within the limits o f
important not to discover imaginary stylistic differences than it is to sampling error, is now shown to be just outside them.
be sure of discovering all genuine ones). When we say that it is enormously unlikely that the differences
T ab le 4.7 gives the z values o f the differences between the A E between the A E and the N E should have come about by chance,
and N E and o f the differences between the A E and E E for each that does not o f course mean that we can say without further ado
particle. It will be seen that twenty-one o f the differences between that it is enormously unlikely that the A E and the N E were
the A E and N E are statistically significant at the 1 -per-cent level. originally part o f the same work. F o r the differences, though not
Som e o f the values are extremely significant: at 2 = 4-00 the due to chance, may be due to a number o f other things : a conscious
T a b l e 4.7 Significance o f the Difference between Frequencies o f Particles in wish by Aristotle to vary his usage in one part o f his work ; the
the AE, NE, and EE influence o f subject m atter; varied degrees o f revision and
polishing, and the like. What our study has shown so far is that it is a
Particle N E and E E and Particle N E and E E and much more economical hypothesis to regard the A E as belonging
A E value A E value A E value A E value originally to the E E ; for almost all the differences between the A E
of z of z of 2 of z and the E E in respect o f the vocabulary so far studied call for no
other explanation than chance.
αλλά 7'39 0-69 μη 2-90 Q'31 We cannot yet be said, however, to have satisfactorily quantified
άν 4-24 !-5I μην 006 029 the likelihood o f chance accounting for the observed differences
âpa 2-88 1-73 θθ€Ρ 3·52 1 55
γάρ between the N E and A E in respect o f particle usage. F o r the
2-82 0-22 oîov I '2 I 0-07
ye 1-99 όταν statistical procedures which have been adopted are ones which are
0-75 5-48 1-94
Se 2-31 1-63 0Τ€ 2-36 2-84 strictly applicable only in situations where the outcome o f each o f
Sy 4-67 0-29 δτι 060 the binomial events to be counted is independent o f the outcome o f
5-57
διό 2-04 Ο-23 ον 3 09 2-04 each other event. In the case o f the occurrences o f a word in a text
èàv o-8o i*6o ονδβ 1-82 0-53 this assumption is clearly unrealistic : the choice o f a word at one
ei 3-03 0*24 ονν 079 2-65 point in a text affects the likelihood o f its being chosen at other
eiirep 1-02 3’54 OVT€ 5'35 2-09 points in the text. T h e matter has been vividly illustrated by A. Q.
èirei 6-46 0-39 π οτζρον 0-40 0-57 M orton (1966, 70):
£Tl 3-48 2-44 πώ ς 0-62 0-21
Kai occurs at a rate of about 005, so that two successive occurrences of kai
V 0-63 1-37 Τ€ 1-43 0*30
would be found, if the occurrence was a matter of pure chance, at a rate of
καθαπερ 2-69 1-91 TOLVVV 2-43 I9O
005 χ 005 = 00025, i.e. in four hundred words of text you could expect to
καί 3-69 i *57 ώς 19 1 2-10
p.év find two successive kais. In Greek prose two successive kais are found
3-56 1-49 ώσττζρ 4-32 0-03
μ έν το ι about once in a million words of text.
3-12 Γ02 ώστβ 6-83 on
But though the choice o f a word has an effect on the choice o f other
90 The Aristotelian Ethics Particles and Connectives in the Aristotelian Ethics 91

words in its immediate context, it seems reasonable to expect that study in individual books; and sim ilarly, only those with an
this effect can be discounted if we make our samples sufficiently occurrence in the E E o f fifty or more will repay study in the
large : there seems no reason to expect that the fact that Aristotle individual five books o f the E E . I f we exclude from our list o f
used kai on a particular line o f the first page o f book i o f the N E particles those which do not occur either seventy or more times in
should affect his decision to use it on a particular line o f the last page the N E or fifty or more times in the E E , we reduce our list from
o f book io. One might well hope that samples as large as the A E , thirty-six to twenty-four, with the omission o f ara, ean, eiper, eti,
N E , and E E would be big enough to allow the local in­ mentoi, men, hot hen, ho tan, hote, poteron, pos, toinun. It will be noted
terdependence o f word-choices to be discounted. But this hope that the new shorter list consists entirely o f particles whose
must be put to the test. What size a sample has to be for the frequencies in the A E and E E differ by only insignificant degrees.
occurrence o f a given word in a text to conform to the conditions of O f the new list, fifteen show significant disagreements between N E
simple sampling is an empirical matter which must be discovered and A E , while nine agree in N E and A E usage.
by counting and calculation. T o discover whether the A E , N E , and Table 4 .10 sets out the occurrences o f the twenty-four particles
E E display the degree o f internal homogeneity which would justify in each o f the seven books o f the N E , and Table 4 .11 in the five
making statistical inferences about their relationship, we must books o f the E E . T h e chi-squared value for each particle is
break up the text o f each o f them into smaller units, and count the calculated in a way which may be illustrated from the case o f ei in
occurrences o f words in each o f these new samples. A significance the E E , whose occurrences are given in T able 4.8. We first set out
test known as the chi-squared test enables one to assess whether the its occurrences in the form of a table known as a contingency table.
occurrence o f the word complies with the conditions o f simple
sampling or whether there are statistically significant differences
between the rates o f occurrence in different samples. T a b l e 4.8 Occurrences o f e i in the EE
T h e obvious smaller units into which to break the treatises are
E E book Occurrences of ei Occurrences of Total words
the individual books that compose them. Considered as samples
other words in book
these books have the disadvantage that they are o f different sizes,
which complicates some o f the calculations; but taking them as I 18 3366 3384
separate wholes has the advantage o f keeping open the possibility, II 27 4672
4645
suggested by many scholars, that they may have existed once as III 52 7043 7095
separate units which were perhaps only later combined into the V II 52 8566 8618
more lengthy treatises that we know. Obtaining statistics for them V III 19 2542 2561
separately will not only help us to assess the degree o f reliance Total 168 26162 26330
which can be placed on our conclusions concerning the treatises as a
whole : it may also enable us to detect any anomalies which may
make individual books stand out from the other books to which they
are traditionally united. We wish to know whether the differences in the figures from
Som e o f the particles we have so far studied occur too book to book can be explained simply as the result o f sampling
infrequently to make it profitable to study their occurrences in error. I f so, we would expect the proportion o f eis in each book to be
particular books. As before, we cannot make reliable estimates o f the same as the proportion in the whole of the E E . Accordingly, we
frequencies from samples unless the expression in question can be work out, for each book, the expected occurrences o f ei and the
expected to occur a minimum o f ten times within them. Since the expected occurences o f other words, on the assumption that the
N E consists o f seven books, this means very roughly that only proportion remains thus constant. T h e results o f the calculation are
particles with an over-all occurrence o f seventy or more will repay inserted in brackets in Table 4.9.
92 The Aristotelian Ethics Particles and Connectives in the Aristotelian Ethics 93
statistically non-significant and the occurrence o f ei in the E E
T a b l e 4.9 The Calculation o f Chi-squared
conforms to the conditions for simple sam pling.1
E E book Occurrences of el Occurrences of other Total A study o f Tables 4 .10 and 4 .1 1 reveals how far, within the N E
actual (expected) actual (expected) and within the E E , Aristotle was regular in his habits o f use of
particular connectives. T h e tables contain a number o f surprises.
I 18 (21-6) 3366 (3362-4) 3384 K a i, which because o f its extreme commonness and apparent topic-
II 52 (45'3) 7043 (7049-7) 7095
III 27 (29-8) 4645 (4642-2) 4672 T a b l e 4.10 Particles and Connectives in Individual Books o f the NE
V II 52 (55-0) 8566 (85630) 8618
V III 19 (16 3) 2542 (2544-7) 2561 I 2 8 9 10 X 2 f°r
3 4
Total 168 26162 26330 6 degrees
of freedom

άλλα 37 3D 60 38 37 22 30 16-27
>/
αν 60 18 53 50 40 55 87 33-50
F o r each cell we now square the difference between the actual 128 76 170 29-41
γάρ 151 195 135 156
and expected values and divide the result by the expected value: ye 18 20-41
3 17 15 3 5 15
thus for the first cell (eis in book I) we square (2 1-6 -18 ) and divide 8é 267 216 302 295 298 234 239 159 4
the square by 21-6, obtaining the result 0 6 0 . When we have 8η 28 10 42 28 31 38 46 16 37
performed this operation for each cell in the table, we add all the 8 ιό 5 10 Η II 10 10 9 4-51
>
results together, and that gives the value o f chi-squared for the table et 44 l3 37 15 9 21 42 4 12 8
as a whole. In this case chi-squared = 2 46. €7T€L 4 6 4 2 I 2 I N.A.
T h e significance o f a chi-squared has to be discovered by y 63 35 93 67 58 71 59 14-96
consulting a table; but first one must ascertain the number o f καθάπβρ 19 3 3 3 IO 21 II 36-75
και 419 298 398 440 362 320 396 17-97
degrees o f freedom associated with it. T h e number o f degrees o f
μ έν 75 108 104 85 84 67 68 42-85
freedom is the number o f ways in which the data represented in the
μή 28 9 61 59 30 34 26 4 3 11
table are free to vary independently o f each other within the oîov 16 21 20 18 8-50
!3 33 15
constraints set by the characteristics o f the total. In each book, δ τι 21 22 16 10 21 2665
39 34
given the number o f occurrences o f ei, the number o f occurrences o f ον 41 48 91 115 80 65 94 35-24
words other than ei is not free to vary independently but is fixed: ον Se 18 22 43 43 31 27 48 14-50
therefore in each line o f the table we have only one degree o f r 766
ουν 25 32 35 32 30 39 42
freedom. G iven the number o f eis in each o f the first four lines o f the οντ e 16 16 12 15 5 4 5 23-18
table, the number in the fifth is not free to vary but can be deduced τε !7 14 II 22 24 26 48 30-85
by subtraction from the total : therefore the degrees o f freedom (i)S 19 19 42 43 25 23 32 13.89
from line to line are four. T h e number o f degrees o f freedom o f the ώ σπερ 12 14 15 18 6 13 20 8-48
ώ σ τε 11 4 5 4 5 4 5 N.A.
table as a whole is four by one, i.e. four. I f we enter a table o f chi-
squared in a statistical textbook at the entry for four degrees o f Totals 5644 4242 6338 5953 5827 5291 6230
freedom we find that the probability o f a chi-squared o f 2 46 is
F o r 6 degrees o f freedom, a value o f 16 8 1 or more is significant at the i-per-cent level.
above 0 50: a distribution o f ei parallel to that in the E E would be
found by chance in one case out o f two. Clearly, the differences I. T h e procedure for calculating chi-square can be found in any statistics textbook; e.g.
between the rates o f occurrence in one book and the next are Caulcott, 19 73, 1 1 4 ff.
94 The Aristotelian Ethics Particles and Connectives in the Aristotelian Ethics 95
T a b l e 4 .11 Particles and Connectives in Individual Books o f the EE other particles all occur in at least one o f the treatises in accordance
with the conditions o f simple sampling, and therefore enable
I II III VII V III X 2 for inferences to be drawn from the frequency o f their use in either the
4 degrees N E or in the E E to an expected frequency in the A E .
of freedom Before proceeding to the comparison with the analogous data for
αλλά the A E , we may notice that our tables reveal that the E E , despite the
32 7§ 43 112 46 !4-°7
ay 26 greater variation in the length o f the individual books that compose
34 34 56 31 I 5-38
γάρ 68 104 225 1249 it, displays a considerably greater degree o f uniformity in respect of
Ϊ31 49
ye 5 !3 3 10 7 626 particle usage. Disregarding the four particles irregular in both, we
Se 103 287 198 3Ö2 84 13-01 find that fourteen out o f the other twenty in the E E occur at rates
δή 5 25 8 22 !5 139 3 showing only insignificant differences between books. In the N E
8 to 0 17 ΐ4 35 5 13 9 3 only ten are similarly regular; twelve display significant fluctu­
et 18 52 27 52 19 246 ations in rate o f occurrences, and in two cases the chi-squared test
€ΤΓ€1 I ι8 II 22 8 7-16 is not applicable.1
>/
38 ι:5 6ο 67 39 2600 I f the A E was originally part o f the E E , we would expect that
η
καθάττερ 4 3 2 I I N.A. those regularities which are displayed in the particle use in the E E
καί 165 40Q 312 526 ι6 ι 14-26
μ έν would continue to be regular, and regular at approximately the
58 12 1 84 148 40 0-55
μη 26 same rate, in the A E ; similarly, if the A E was originally all o f one
74 40 8ο ίο io-oo
οίον 21 10 piece with the N E one would expect it to reflect the regularities to
15 32 29 1-73
δτι 24 56 20 67 be observed there. W e can therefore use the fourteen observed
25 8 -6 6
ού 45 100 46 !33 56 17-83 regularities from the E E , and the ten observed regularities from the
ovSe 27 28 ι1 29 12 ! 3'25 N E , to make predictions as to the regularities in particle usage to be
ονν 15 19 Π 29 17 8-8o expected in the A E . T able 4 .12 , which sets out the figures for the
ούτε !3 20 25 22 2 13-67 individual books o f the A E , enables us to begin to compare these
Τ€ 17 25 ι8 24 7 4 15 expectations with the observed data to see whether the
ώς Η 25 ι8 67 17 176 5 Nicomachean hypothesis or the Eudemian hypothesis best fits the
ώσττ€ρ 9 23 23 52 ι6 10-94
facts.
ώ στΐ II 35 12 29 9 5-07 T h e most striking feature o f T able 4 .12 is that it shows the A E as
Totals 3384 7095 4072 8 6 ι8 2561 much more uniform than either the E E or the N E . In twenty-two
out o f twenty-four cases the differences between the individual
F o r 4 degrees o f freedom, a value o f 13 28 or more is significant at the i-per-cen t level.
books are statistically insignificant. Only in the cases o f ë and hösper
are the between-books differences significant at the 1 -per-cent
neutrality might have been expected to occur with constant level. T h is fact is not negligible in view o f the fact that some
frequency, is irregular in appearance in both the N E and the E E : scholars have wished to solve the problem o f the disputed books by
there are differences between the rate o f occurrences in individual
books which, in each treatise, are just significant at the i-per-cent i . F o r a chi-squared test to be applicable, at least 80 per cent o f the cells in the frequency
level. Sim ilarly the negation signs ou and ou te occur at significantly table must have an expected value o f no less than five. Because e-nei and w a r e each occur less
different rates in different books within each treatise, and so does than forty times in the N E this is not so. T h e inapplicability o f the chi-squared test in this
case is o f course not because the difference between the N E and E E in this respect is too
the particle an. It is not, then, the commonest or most apparently
small, but because it is too enormous. T h e test is inapplicable for analogous reasons to
topic-neutral particles that occur with the greatest regularity. T he κ α θ ά π ε ρ in the E E .
Φ The Aristotelian Ethics Particles and Connectives in the Aristotelian Ethics 97
T a b le 4.12 Particles and Connectives in Individual Books o f the AE dë, dio, hoion, oude, oun, hos) where we can compare the regularity
found in the N E with that found in the A E ; and thirteen cases (gar,
χ 2 for 2 degrees ge, de, ei, epei, men, më, hoion, hoti, oude, oun, te, hôste) where we can
A B C of freedom compare regularities in the A E and the E E . We already know, from
T able 4.6, that in six out of the eight N E cases, and in every one of
αλλά 88 43 83 2‘55 the thirteen E E cases, the differences between the frequencies in the
âv 34 32 31 3·78
γάρ 133 102 133 993 T able 4.13 Comparison o f two contexts for the AE in respect
ye 5 5 IO 1-68 o f Particle Use
δε 280 161 282 2-29
δη 12 20 14 8-94 Particle A E and N E A E and E E
διό 15 15 14 2-15 χ 2 for 9 degrees χ 2 for 7 degrees
€1 53 20 39 5·°9 of freedom of freedom
è n ei 10 7 25 8·6ι
V 90 3 0
6 2 1 2 - 2 8 αλλά jr o 8 1 6-86
καθάπερ I 3 10 7'79 âv s ^-44 22-12
καί 362 240 392 1-50 γάρ 40-48 I4'5I
μ ίν 114 71 140 4-27 ye 28-22 8-79
μη 54 3 1 59 i -oo §é 23-84 17-67
οΐον 34 10 26 5-25 δη 46-27 22-95
ότι 45 39 49 1-76 διό 10-90 16 4 1
ού 114 62 I 12 1-54 ei S2 'H 7-68
ούδί 20 28 3 1 7-06 €71-et 66-07 1 6 12
οΰν 21 0-89 >/
3 2 39 27-41 4°'49
η
otire 20 2 4 3 0 4-26 καθάττερ 55'37 N.A.
re 27 IO 19 2-89 και 34 '10 18-50
ώς 3 0 1 4 2 3
1-46 μ έν 57Ί 1 7-30
äianep 2 3 ! 3 4 4 1 0 - 4 5 μη 48-63 11*21
ώ σ τζ 2 3 1 7
21 0-43 ο ΐο ν 15-41 7*01
δη 54-63 10-62
Totals 6430 4198 6413 ον 44-14 22-56
Fo r 2 degrees o f freedom, a value o f 9-21 or more is significant at the irper-cent level.
ov 8e 24-74 20-52
ονν 933 1 5-56
OVT € 52-20 21-33
splitting them up, assigning one or more o f them to the N E and the Τ€ 38-31 7-12
remainder o f the E E . 1 ώς 20-49 25-29
T h e irregular occurrence o f hösper contrasts with the regularities ώσπβρ 40· 8q 21'40
to be found in both N E and E E ; the irregularity o f ë parallels its ώστε 5 r 39 5*55
irregularity in the E E , and conflicts with the prediction o f regularity
from the N E . Otherwise, all predictions o f regularity from either F o r 9 degrees o f freedom, a value o f 21-66 or more is significant at the 1-
per-cent level.
E E or N E are fulfilled by the A E . We now have eight cases (alla, de, F o r 7 degrees o f freedom, a value o f 18-48 or more is significant at the i~
i. T h u s Rowe, op. cit. 9 0 - 1 1 4 sees A E B as being wholly Nicomachean, whereas A and per-cent level.
C, he thinks, have a Eudem ian base. Values o f χ 2 significant at the i-per-cent level are italicized.
98 The Aristotelian Ethics Particles and Connectives in the Aristotelian Ethics 99

treatises as a whole are not significant ; but we are now in a position T a b l e 4.14b ώ στε tn the traditional EE. Proportions in per cent
to make a comparison which takes account not only o f the frequency 0-6
of the particles in each treatise considered as a whole, but also of the
distribution o f the particles in the particular books o f the treatises. 0-5
We can place the three books of the A E in their rival contexts in
turn, and see which experiment produces the more homogeneous 0-4
resulting whole. We do so by putting the seven books o f the N E
together with the three books of the A E , and applying a chi-squared
0-3
test to the ten-book whole thus produced (the traditional Nico­
machean Ethics); and similarly, lumping together the E E and the
0-2 ί-
A E and applying a chi-squared test to the resulting eight-book Eude­
mian Ethics. T h e results o f this experiment are shown in T able 4 .13.
0- t
It leaps to the eye that in respect of particle usage the traditional
Eudemian Ethics is a much more homogeneous work than the
traditional Nicomachean Ethics. In the former, there are only nine 0-0
3 4 5 8
cases where the differences between books are significant ; in the Books

latter the differences are statistically significant in twenty out o f

T a b l e 4.14c ε π ε ί in the traditional NE. Proportions in per cent
twenty-four cases. I f we restrict ourselves to a consideration o f the
cases recently isolated, we find that only in four of the eight cases 0-4

where an N E regularity corresponded to an A E regularity can the

two homogeneous treatises be combined to form a third homo­ 0-3 -
geneous whole; whereas in the thirteen cases where E E and A E c<u
agreed in displaying regularities, the regularities are preserved if we ” 02
combine the two treatises together. E very irregularity to be found aα> ------
in the Eudemian Ethics containing the disputed books is already to 01

T a b l e 4.14 Bar Charts o f the Use o f ε π ε ί and ώ στε in the traditional .... ,j _ l_ . . Ί
00
NE and EE 1 4 5 6 10
Books
T a b l e 4.14a ώ σ τε in the traditional NE. Proportions in per cent
0-6 r
T a b l e 4.140 ε π ε ί in the traditional EE. Proportions in per cent
0-4 r
0-4

C- 0·3
" 0-3 C
α> 1)
u
&
0-2 ê 02

0 1
0-1

00
00 2 3 4 5 6
4 5 6 10 Books
Books
100 The Aristotelian Ethics Particles and Connectives in the Aristotelian Ethics 10 1

be found in the E E alone; and even those particles (such as alia) T a b l e 4.15 (contd.)
which display irregularity in the narrower context o f the E E no
longer do so if placed in the wider context o f the traditional
Eudemian Ethics. I 2 3 4 8 9 10

T h e way in which the disputed books look much more at home in

8η f. 0-50 Ο 24 0-66 047 0 -5 3 0-72 0 -7 4
a Eudem ian than in a Nicomachean context can be illustrated O -IO 009 010 o’ io o-io 0-09
s .e . O -II
graphically in the cases o f the particles höste and epei. T able 4 .14 8 ιό f. 009 0-24 0-22 o-i8 0- 1 7 019 014
gives bar charts o f the occurrences o f these particles, first in the s .e . 005 o*o6 0-05 0-05 0-05 006 0-05
5
Nicomachean Ethics taken as including the disputed books as its Cl f. 078 0-31 0-58 0-25 015 040 0-67
fifth, sixth, and seventh books; and then in the Eudemian Ethics s .e . 009 0*1 0 0-08 009 009 009 0-09
taken as including the disputed books as its fourth, fifth, and sixth επ ^ ι f. 0-07 014 006 0-03 0Ό2 004 0-02
books. T h e sore thumb which is conspicuous in the first chart o f s .e . 0-03 0-03 003 0-03 0-03 003 OO 3
each pair is noticeably lacking in the second one of each pair. 1 f. Γ 12 0-83 147 1-13 ΓΟΟ 1 -3 4 0 -9 5

We must now take a further look at the irregularities discovered s .e . 0*14 o-i6 013 014 0- 1 4 015 0-13
καθάπερ f. o*34 0-07 0-05 005 0- 1 7 0 40 o-i8
between books within the undisputed N E and the undisputed E E .
s .e . 0-06 006 0-05 005 o-o6 006 0-05
Is it possible to discover whether these irregularities in the use of 6-21
καί f. 742 7'02 6-28 7 ’39 605 636
particular particles are concentrated in particular books o f these 0-38 031 032 0-32
s .e . o* 33 0 -3 3 0 -3 4
treatises which render anomalous an otherwise homogeneous whole ? μ εν f. 164 1 -4 4 1 -27 i 09
1-3 3 2 -5 5 1 -4 3
T h e raw scores given in Tables 4 .10 , 1 1 , and 12 do not render this s .e . o * i6 o-i8 015 016 016 017 0-15
perspicuous : the matter is more easily studied in Tables 4 .15 and 16 μή f. 0-50 0- 21 096 099 0-51 064 0-42
which give the proportions and standard errors o f the particles in s .e . 0-10 011 o-io o-io o-io O II 010
individual books. οΐον f. 0-28 031 0-52 0 -3 5 0 -3 4 0 -3 4 0-24
s .e . o-o8 008 0-07 0-08 0-08 008 0-07
δη f. 0 -3 7 052 0-62 0-27 017 064 0 -3 4
T a b l e 4.15 Frequencies of Particles and Connectives in Individual Books s .e . 0*09 009 0-08 0-08 0-08 009 0-08
o f the NE ον f. o* 73 ΙΊ3 144 1 -9 3 1 -3 7 1-23 151
s .e . 0-15 018 014 0- 1 5 0- 1 5 016 015
I 2 3 4 8 9 10 ονδε f. 0-32 0-52 0-68 072 0 -5 3 051 0-77
s .e . 0-10 O II o-io O 'I O o-io 011 010
αλλά frequency ονν f. o*44 0 -7 5 0-55 054 0-51 0-74 067
(%) 066 071 0-95 064 063 042 0-48 s.e. 0*10 011 o-io 0 10 o-io o-i i 0 10
standard OVT € f. 0-28 038 0 19 0-25 0-09 0-08 0-08
error 0 11 0 12 0 10 o-io 0· 10 o-i i 0 10 s.e. 0*06 006 0-05 005 0-06 006 0-05
αν f. i 06 042 0-84 0-84 069 1-04 I 40 Τ€ f. 0-30 0 -3 3 0 17 0 -3 7 0-41 049 0 -7 7
s.e. 0 13 0 15 0-12 0-12 0 13 0-13 0 12 s.e. 0-09 009 0-08 0-08 0-08 009 0-08
γάρ f. 2-27 17 9 2-38 286 335 2'55 2-50 ώς f. 0*34 0 -4 5 0-66 072 0 -4 3 0 -4 3 0-51
s.e. 0-21 0-24 0*20 0-20 0‘2I 0-22 0-20 s.e. 0*09 010 009 009 009 0- 1 0 0-09
ye f. 0-32 0-07 0-27 0-25 0-05 OO9 026 ώ σπερ f. 0*21 0'33 0-24 0-30 o-io 0-25 0-32
s.e. 006 007 0*05 006 006 0-06 006 s.e. 0-07 007 006 006 0-07 007 0-06
Se' f. 473 509 4*76 496 5 11 4-42 3 84 ώστ€ f. 0*19 009 0-08 0-07 0-09 008 0-08
s.e. 0-28 0 32 0-27 0-27 0-28 0-29 0-27 s.e. ΟΌ4 004 0-04 0-04 0-04 004 004
102 The Aristotelian Ethics Particles and Connectives in the Aristotelian Ethics 103
T a b l e 4.16 Frequencies o f Particles and Connectives in Individual Books T a b l e 4.16 (contd.)
o f the EE
I II III V II V III
I II III V II V III
οΰν f. 0-44 0-27 0-36 0-34 0-66
αλλά frequency s.e. O -IO 0-07 0-09 0-07 0-12
(%) 0 -9 5 IIO 0-92 1 -3 ° i 80 ούτε f. 0-38 0-28 0-54 0-34 0-08
standard s.e. o-io 0-07 0-08 0-07 01 i
error 019 013 016 0-12 0-21 τε f. 0-50 0-35 0-39 0-28 0-27
αν f. 077 0-48 0 -7 3 0*65 1*21 s.e. o-io 0-07 0-09 0-08 0-12
s.e. 014 010 0-12 0*09 Ο 16 (ï)S f. 0-50 0-35 0-39 078 0-66
γάρ f. 201 i -85 2- 23 261 191 s.e. 0-13 0-09 o-i 1 0-08 0-14
s.e. 0-25 0-17 0-21 016 0-29 ώ σπερ f. 0-27 0-32 0-49 o-6o 0-62
γε f. 015 018 006 012 0-27 s.e. 0-12 0-08 o-io 0-07 0-14
s.e. 006 0-04 OO5 004 ΟΌ7 ώ στε f. ο·ι8 049 0-26 0-34 0-35
δε f. 304 4 -0 5 4 -2 4 4-20 3-28 s.e. o-io 0-07 0-09 0-06 ο ι 2

s.e. 0 '3 3 0-23 C 28 0-21 038

δη f. 0- 1 5 0 -3 5 O I 7 Ο 26 0 -5 9 T h e formula for calculating the standard error o f the proportions
s.e. 009 0-06 OO 8 006 0· 10
given differs from that used for the ethical treatises as wholes and
διό f. 000 0-24 O 3O 0 '4 I 0-20
s.e. 009 006 OO 8 0' 06 0 · 10
explained on pp. 84-5 above. Earlier, we were treating the ethical
€1 f. 0-58 0-60 074 treatises as being each a sample from the indefinitely large
0 -5 3 0 -7 3
s.e. 014 009 012 OO9 0· 16 population o f Aristotelian prose : now, we are regarding each book
επ εί f. 0-03 0-25 0-24 0-20 Q '3 1
as a sample from a population o f known size and known
s.e. 0-08 006 OO7 0-05 009 characteristics, namely the whole treatise from which the book is
V f. I ' Î2 1-62 I-2 8 0-78 1-52 taken. We are asking, of each book, whether its characteristics are
s.e. 9
O I 0-13 0· 16 012 022 those which would be expected o f a sample drawn randomly from
καθαπερ f. 012 0-04 004 001 OO4 the population constituted by the whole treatise. In such a case, in
s.e. OO3 0-02 003 0-02 OO4
calculating the standard error o f the sample proportion we make use
και f. 4-88 661 6-68 6 · 10 629
o f the known proportion, for each word, o f the population as a
s.e. 041 029 035 026 Ο 48
whole. T h e formula used to work out the standard error o f the
μ εν f. 171 i '7 i ι·8ο 1-72 156
s.e. 0-22 015 019 014 Ο 26
sample proportion is
μη f. 0-77 104 086 093 0 -3 9 (s e p ) = y / { 6 (i —d)jn}
s.e. 016 011 014 010 0 · 18
οΐον f. 044
where p is the observed proportion in the sample, Θ is the
0-45 045 0 -3 4 039
s.e. 0· II 008 009 0-07 013 population proportion and n is the sample size.1 We can use the
ότι f. 071 079 078 098 standard error thus calculated as a standard unit to express the
0 -4 3
s.e. 015 010 013 009 017 observed difference between the sample and population pro­
ού f. r 33 141 141 1 -5 4 2- 1 9 portions as a 2;-score thus:
s.e. 0-20 014 017 013 0-22 ρ —θ
ουδέ f. 080 039 036 0 -3 4 0 -4 7
" — se p
s.e. O II 008 010 007 013
I. Caulcott, 1973, 83.
104 The Aristotelian Ethics Particles and Connectives in the Aristotelian Ethics 105
W here a 2 score is 2 58 or more (or —2 58 or less) there is less than difference between the sample frequency and the population
one chance in a hundred o f the word-frequency in question having frequency was calculated in each case; these values are not given in
occurred simply as the result o f random sampling, and we say that the table, but wherever a 2-score had an absolute value o f more than
the difference between the frequency in the particular book and the 2-58 the proportion figure in the table is printed in bold type: the
frequency in the treatise as a whole is statistically significant at the bold-type values, therefore, are those which differ from the over-all
i-per-cent level. In T able 4 .15 the proportions and standard errors frequency in the N E by a statistically significant amount. Table
for each particle in each book o f the N E is given, the standard errors 4 .16 sets out analogous data for the E E , and Table 4 .17 for t h e ^ £ :
being calculated on the basis o f the known frequencies for the N E as in each case the standard errors are calculated on the basis o f the
a whole, as already given in Table 4.5. T h e z-value for the frequencies for the whole treatises given in T able 4.5, and in each
case values differing from the treatise value by an amount
T able 4.17 Frequencies of Particles and Connectives in Individual Books significant at the i-per-cent level are in bold type.
nf the AE T h e first fact which meets the eye on studying these tables is the
great general regularity o f particle use to be found in the ethical
A Β C treatises. O f the 360 values recorded o f frequencies in these tables,
only thirty-three, roughly 9 per cent, are anomalous in the sense of
Frequency Standard Frequency Standard Frequency Standard
error error error being different by a significant amount from the over-all frequency
(%) (%) (%)
o f the treatise to which they belong. In all other cases chance is an
αλλά i *37 0 14 10 2 0-17 1-29 0-14 adequate account o f the differences observed. It should be stressed
âv ο·53 0-09 0-76 0’ 1 2 0-48 o-oç again that in the anomalous cases no inference, about authorship,
γάρ 2-οη 0·ι8 2'43 0-22 2-07 0·ι8 chronology, interpolation, context, etc. can be drawn from the
γ€ 0' 11 0-04 ΟΊ 2 ΟΟ5 0 16 0-04 anomaly alone: all the anomaly means is that some explanation
Se 435 0-25 3-88 031 4-40 ο·25 other than chance is needed. But such explanations are often very
δη 019 ο·ο6 0-48 ο·ο8 0-22 ο·ο6
easy to find. F o r instance, it will be noticed in T able 4 .16 that the
διό 0-23 ο·ο6 ο·36 ο·ο8 0-22 ο·ο6
ΐ frequency o f kai in the first book o f the Eudemian Ethics is
6t 0-82 ο·ιο 0-48 0 12 0-69 ο·ιο
6776t ο·ι6 ο·ο6 ο ·ΐ7 ο·ο8 ο·ο6 significantly less than the frequency o f kai in the other books o f the
0-39
Γ40 0Ί 3 0'7 ΐ ο·ι6 ο·97 E E (it is indeed much the lowest to be found in any o f the books in
ÿ 0-13
καθαπζρ 0-02 0-04 ο·οη 0-04 0 16 0-04 the three tables). Scholars have long noticed in book I o f the E E
καί 5*63 0-29 5'72 ο·36 6* 11 0-2Q evidence o f careful stylistic revision— hiatus is avoided, for
μ εν ι ·77 ο ·ι7 ι ·09 0-21 2 18 0-17 instance— and it may well be that when polishing up his style
μη ο·84 0·Ι I ο·74 0-14 0-92 ο -ιι Aristotle, like many other writers, removed paratactic constructions
οΐον 0-53 ο·ο8 0-24 0·Ι0 041 ο·ο8 in favour o f more elegant periods; this would account for the
δτι ο·ηο ο ·ιι ο·93 0-14 076 ο ·ιι anomalous value. There may be similar particular explanations for
ού 1-77 ο·ι6 1-48 0-20 1-75 ο·ι6 the other particular anomalies: what is important in the tables are
ού 0-31 ο·ο8 ο·07 ο·ιο 048 ο·ο8 the regularities revealed by the absence o f anomalies in the great
οΰν ο·5θ ο·ο9 ο-so Ο ΙΙ ο·6ι 0-09 majority o f the cases.
ΟΌΤ€ 0-3 1 ο·ο8 0-57 0 10 ο·47 ο·ο8
Τ6 I f we look at T able 4 .15 , we notice that book 2 stands out to some
0’42 ο·οη 0-24 009 ο·3θ 0-07
ώς ο·ο8 0 10 036 ο·ο8 extent from the other books: seven out o f the twenty-five
ο·47 ο·33
ώσττ€ρ ο·36 0-09 Ο ΙΙ ο·09 o-oç anomalous values to be found in the N E books occur in this book
° '3Ι
ώστβ 036 ο·οη ο·40 009 ο·33 o-oj alone. What is most striking is that in each of these cases the book 2
value is nearer to the E E mean proportion than it is to the N E mean
ιο6 The Aristotelian Ethics Particles and Connectives in the Aristotelian Ethics 107
proportion. This affinity between N E 2 and the E E may deserve only crumb o f comfort for those who defend a Nicomachean origin
further investigation. for some o f the common books is that B ( N E 6) is as comfortable in
T able 4.16 shows that the E E is more homogeneous in respect o f the ten-book Nicomachean context as are any o f the undisputed
particle use than the N E : there are eight anomalies in i20 cases(6 .7 books. But that does not alter the fact that it is more at home in the
per cent) as against twenty-five anomalies in 168 cases (14 9 per Eudemian context (where it displays only one anomaly, in the use of
cent). Books I and V I I I , with two anomalies each, stand slightly ê) than in the Nicomachean context (where it displays three).
apart from books II, I I I , and V II, which are highly regular with the We have noticed that the common books, taken by themselves,
unusually high gar score for book V I I as the only anomaly. M ost form one o f the most homogeneous blocks in the Aristotelian ethical
regular o f all the ethical books are the common books o f the A E : no corpus. Can we proceed further and see whether this uniformity is
anomaly at all is to be found in any o f the three books. T h is is surely
a striking result in view o f the fact that many scholars have claimed T a b l e 4.18 Particles: Books as Samples from the Traditional NE as a
that the common books form a patchwork o f writings from different Single Population
periods, some from Eudem ian original material, some from
Nicomachean. I f this is patchwork it is highly remarkable I 2 3 4 5 6 7 8 9 ΙΟ
patchwork, with the pieces so well matched as to make a more
regular pattern than the originals from which the pieces have been αλλά 0 0 0 0 X 0 I 0 I I
cut. aw 0 I 0 0 0 0 I 0 0 I
We may use the same data to test the fit o f the A E into its two γάρ 0 I 0 0 0 0 0 I 0 0
ye I 0 0 0 0 0 0 0 0 0
rival contexts in the following manner. I f we compute the mean
δε' 0 0 0 0 0 0 0 0 0 I
proportion for the ten books made up o f the N E plus the A E , we can
δή 0 0 0 0 I 0 I 0 0 I
again compute standard errors and z-scores for each o f the διό 0 0 0 0 0 0 0 0 0 0
frequencies of each o f the ten books, and record anomalies as we did >
€l 0 0 0 I I 0 0 I 0 0
for the seven-book N E , and similarly we can compute mean €7761' 0 0 0 0 0 ο I 0 0 0
proportions for the eight books o f the traditional Eudemian Ethics y 0 0 I 0 0 0 0 0 0 0
( E E plus A E ) and with their aid reckon standard errors and z- καθάπερ I 0 0 0 I 0 0 0 I 0
scores for each particle in each book. T h e scores so computed και I 0 0 I 0 0 0 ο 0 0
cannot be set out without an inordinate degree o f reduplication of μ έν 0 I 0 0 0 ο I 0 0 I
data: but Tables 4 .18 and 19 show the anomalies which are μη 0 0 I I 0 0 0 0 0 I
produced by this procedure. It will be seen that the effect is to make οΐον 0 0 0 0 0 0 0 0 0 0
ΟΤΙ 0 0 0 I 0 I I I 0 0
the N E considerably less homogeneous than it was (there are now
ού I 0 0 I 0 0 0 0 0 0
fifty-one anomalies out o f 240 cases, or 2 1 3 per cent) while the E E
ού&€ 0 0 0 0 I 0 0 0 0 0
is made very slightly more homogeneous (with twelve anomalies in οΰν 0 0 0 0 0 ο 0 0 0 0
192 cases, 6 2 5 per cent); the E E remains very much more OVT€ 0 I 0 0 0 I I I I I
homogeneous than the N E . I f we look in particular at the disputed Τ€ 0 0 I 0 0 0 0 0 0 I
books, we find that in their Nicomachean context they display ώς 0 0 0 I 0 0 0 0 0 0
eighteen anomalies (25 per cent), in their Eudem ian context only ώ σπερ 0 0 0 0 0 ο I I 0 0
two (2 8 per cent). N E book 2, which stood out in a purely ώ στε 0 0 0 0 I I I 0 0 0
Nicomachean context, is now more at home with the disputed
books to keep it com pany; in the new arrangement the most o denotes an absolute £-score less than 2 58; 1 denotes a z -score greater than 2-58 and
anomalous o f the undisputed Nicomachean books is book 10. T h e therefore significant at the i-per-cent level.
ιο8 The Aristotelian Ethics Particles and Connectives in the Aristotelian Ethics 109
displayed within each o f the books in question? T h is is o f interest the contents o f the samples which result may be indicated as
given the prevalence o f theories which divide the disputed books follows:
into pieces o f varying dates. Altogether the common books total
1 . i i29 ai~ 30 b8 ; bk. A chs. 1 - 2 ; general and particular justice.
1 7041 words. It is therefore possible to divide them into seventeen
2. i i3 0 b8 -3 2 ai 5 ; bk. A chs. 2 - 4 ; distributive justice.
samples, each o f approximately 1000 words, and repeat the tests
3. 1 13 2 111 5 - 3 3 ^ 2 ; bk. A chs. 4 - 5 ; corrective justice and
which we have carried out on the books as a w hole.1 T h e limits and
reciprocity.
4. i i3 3 a2 2 -3 4 a20; bk. A chs. 5 -8 ; justice as m ean; political
T a b l e 4.19 Particles: Books as Samples from the Traditional EE as a justice, nature, and law.
Single Population 5. i i 3 4 a26 -36 b3 2 ; bk. A chs. 8 -9 ; justice, voluntariness, and
involuntariness.
I II III IV V VI V II V III 6. i i 3 6 b32~38bi ; bk. A chs. 9 - 1 1 ; equity, aporiai about
justice.
αλλά 0 0 0 0 0 0 0 I
7. i i 3 8 bi- 4 0 ai ; end of bk. A , bk. B chs. 1 - 3 ; intellectual
αν 0 0 0 0 0 0 0 I
virtues, man as agent.
γάρ 0 0 0 0 0 0 0 0
8. 1 1 40“ i —4 1 b2 ; bk.B chs. 4 - 7 ; art, wisdom, and learning.
ye 0 0 0 0 0 0 0 0
9. i i 4 i b2 -4 3 a5; bk.B chs. 7 - 1 0 ; wisdom and its parts and
8e I 0 0 0 0 0 0 0
δη 0 0 0 0 0 0 0 I
satellites.
S to I 0 0 0 0 0 0 0 10. i i 4 3 a5~44b5; bk. B chs. 1 0 - 1 3 ; wisdom in relation to
el 0 0 0 0 0 0 0 0 learning and intellect.
enet 0 0 0 0 0 0 0 0 1 1 . ii4 4 b5“ 46a6; bk. B ch. 13 , bk. C , chs. 1 - 2 ; wisdom and
y 0 I 0 0 I 0 I 0 virtue; opinions on akrasia.
K a d a ire p 0 0 0 0 0 I 0 0 12. ii4 6 a6~47b8; bk.C chs. 2 - 3 ; akrasia and knowledge.
και I 0 0 0 0 0 0 0 13. i i4 7 b8 -49ai2 ; bk. C chs. 3 - 5 ; the sphere o f akrasia.
μ έν 0 0 0 0 0 0 0 0
14. i I49ai2 - 5 0 bi2 ; bk. C chs. 5 - 7 ; different kinds o f akrasia.
μη 0 0 0 0 0 0 0 0
15. H 50bi2~ 5 2ai 3 ; bk. C chs. 7 - 1 0 ; continence, incontinence,
ο ΐο ν 0 0 0 0 0 0 0 0
intemperance.
O Tt 0 0 0 0 0 0 0 0
I 0 0 0 0 0
16. i i5 2 ai 3 —53bi 3 ; bk. C chs. 1 0 - 1 3 ; is pleasure a good?
ού 0 0
ού& έ I 0 0 0 0 0 0 0 17. H 5 3 bi3 “ 54b35 ; bk. C chs. 1 3 - 1 4 ; conclusion o f discussion
ουν 0 0 0 0 0 0 0 0
o f pleasure.
OVT€ 0 0 0 0 0 0 0 0 T able 4.20 gives the raw scores for each particle in each sample.
re 0 0 0 0 0 0 0 0
Proportions, standard errors, and z-scores were computed as
WS 0 0 0 0 0 0 0 0
before; these statistics are not reproduced, but where the z-score
waTTep 0 0 0 0 0 0 0 0
0
revealed an anomaly the raw score has been printed in bold type.
w o re 0 0 0 0 0 0 '0
T h e final column for each particle gives the value o f chi-squared for
ο denotes a 2-score less than 2 58; I denotes a 2-score greater than 2· 58 and therefore 16 degrees o f freedom, calculated in the manner explained above on
significant at the i-per-cent level. p. 92.
Inspection o f T able 4,20 suggests that the general regularity we
i. The samples are not exactly equal in word-length. For technical reasons it was easier to
deal with samples measured by number of lines than with samples measured by number of have observed in Aristotle’s use o f particles continues to be
words. observable even at a comparatively microscopic level. Tw enty-one
 110 The Aristotelian Ethics Particles and Connectives in the Aristotelian Ethics 111
anomalies occur out o f 408 cases, a mere 5 per cent. N one o f the
sample sections displays a remarkably large number o f anomalies ;
S3 E
sample i i has the largest number, namely four; this section,
-ιo-i *-
'M b Ό Γ»O
I 00 I- «^V interestingly enough, is one where Aristotle is recording, perhaps to
ωo 1)
o>
I O nC h- some extent in their own words, the opinions o f other philosophers
ΐΛ Ο n 1 OO ! ! I
χ '0 £ r·)
« o on the nature o f akrasia. Once again, there is no real support for the
view that the common books o f the A E are patchwork o f disparate
material.
rO ’t - X -φ 1 A word o f warning is necessary, however, about the significance
4.20 Particles and Connectives in Seventeen Samples of the AE

o f the statistical results at this point. T h e tests described on pp. 84,

87 and 1 03, are all based on the assumpton that the sampling
N «i fO UON distribution o f the values to be studied is a normal one. In fact,
where we are dealing with the proportion o f a number o f discrete
events falling into one o f two classes (as we are with the occurrences
o f words) we know that the appropriate theoretical sampling
distribution is not the normal distribution but the binomial
sO sO t-* ro o o N distribution. M oreover, when the proportion o f events falling into
one of the two classes is very small (as it is in the case o f rarely

χ 2 of more than 32 significant at the i-per-cent level for 16 degrees of freedom.

τί- O' Π rn O N occurring words) the appropriate discrete distribution is not the

Scores in bold type indicate 2-values (absolute) of more than 2-58 (/><ο·οι).
00 Ό η μ rh ro
binomial, but some other such as the Poisson distribution.1 None
the less, it is regular statistical practice to use the binomial
distribution instead o f the Poisson, and the normal distribution in
turn instead o f the binomial, provided that we are dealing with a
00 N 00 <-> h rsO ■-« m <-< ο -'ί* ·-> Ο O r*-> N N ·-· 'f Ό reasonably large number o f trials ; for in such cases the distributions
O 1· O ' O ir > \C
approximate very closely to each other, and the use o f the normal
distribution greatly facilitates computation. In the case o f the
00 so η o 00 i statistics for the N E , A E , E E as wholes, and for the individual
books as wholes, the length of the books (and therefore the number
ci tj- im *-> t+i o f trials in question, each word-choice counting as a trial) was
i r n nO•Ή *-< Ό
co^foo Lnoc ιη-Φτί-Ν μ μ γον
sufficiently great for this approximation to be reliable. Bu t now that
able

N HH we have reached the consideration o f samples o f no more than 1000

Γ- Γ-· <N ro iTi
N rn words, the effect of treating the distribution as a normal one may
T

become misleading in the case o f the less frequently occurring

particles. T h e approximation o f the Poisson distribution to the
r^i r* t"· N η ro W 00 N m Ο Η M fO O O -φ I-H ro if) N t-. u*> >-i frj
normal one is sufficiently close provided that the number o f trials
t-t Ο M NN NH
multiplied by the proportion of successes is at least five. In the case
o f the 1 000-word samples, therefore, the approximation is reliable
vy in the case o f particles occurring with a frequency o f at least 0-50 per
a b *w
5- Vy
VÖ VÖ V
ε -< ^ m'm 'p-'2
60 w -vut; *?=- 's 3 vè 'p-i.2 h '3 «-! *3 w ϊ δ δ i . See Caulcott, 1973, 2 8 -3 3 .
- e *Ö N ^ »O 00 * * 3. 3. O sO O o O O t- 3 ‘3 e3
112 The Aristotelian Ethics
cent. We have already encountered in connection with the chi-
squared test the fact that where individual cells o f the table o f values
have an expected value o f less than five in more than 80 per cent o f
cases, the test is unreliable. F o r analogous reasons, in Table 4.20
where the chi-squared test is inapplicable, the z-values on which
the specification o f values is based cannot be regarded as any precise
measure o f statistical probability.
In a later chapter we shall see a method o f circumventing some of
the difficulty o f accurately measuring the statistical significance o f
stylistic features even in small samples. In the present one, enough
has been done to show that in respect o f the use o f particles, the
common books o f the A E , whether considered as a single unit, or as
individual books, or in iooo-word samples, resemble the Eudemian
Ethics far more than they resemble the Nicomachean Ethics.
CH APTER 5
Prepositions, Adverbs, and Pronouns
in the Ethical Treatises
PREPOSITIONS
like particles, furnish abundant material for the
P r e p o sit io n s,
statistical study o f style. T h ey occur with considerable frequency
and are comparatively, though not totally, free from fluctuations
due to change o f subject matter. T able 5.1 sets out the occurrences
and frequencies o f the nineteen prepositions commonest in the N E ,
A E , and E E . It will be seen that tokens o f these prepositions
constitute roughly 6 per cent o f the total text to be considered.
Prepositions occur slightly less frequently in the A E than in the E E ,
and considerably less frequently in the A E than in the N E ; but
neither of the differences is statistically significant.
In the majority o f cases the frequency in the A E is not
significantly different either from that in the N E or from that in the
E E . But once again the differences which do exist show that the A E
resembles the E E more than it does the A E . I f we plot the
occurrences o f each o f the prepositions in the A E against those in
the N E we find that the correlation between frequency in the A E
and frequency in the N E is 0 87 for the nineteen prepositions : the
corresponding correlation between the frequency in the A E and the
frequency in the E E is higher, 0-97. T h e favourite preposition in the
N E is en; in the A E and E E it is kata with the accusative.
When we apply the test described above in chapter 4 to see
whether the treatises can be regarded as samples from a single
population, we find that only in three cases are the differences
between preposition usage in the E E and the A E statistically
significant at the 1-per-cent level. T h is is in the case o f the
prepositions dia with the genitive, eis with the accusative, and epi
with the dative. In every one o f these cases there is also a significant,
and in fact larger, difference between the usage in the N E and the
114 The Aristotelian Ethics Prepositions, Adverbs, and Pronouns in the Ethical Treatises 115
T able 5 .1 Nineteen Common Prepositions in the N E , A E , and E E are five cases, on the other hand, where the N E differs significantly
from the A E without a correspondingly significant difference
Preposition NE AE EE between A E and E E . T h is is in the case o f aneu (twice as frequent
in A E and E E as in N E ) , en (much more popular in the N E than in
Occur­ Propor­ Occur­ Propor­ Occur­ Propor-
tion rences tion rences tion the others), epi with the accusative (another Nicomachean fa­
rences
°/
/0
°/
/0
°//0 vourite), para with the accusative (very much less frequent in the
N E ) , and peri with the genitive (similarly infrequent in the N E ) . It
av€v 18 0-05 23 0 13 31 0*12 is again possible that some of these differences may be due to the
άττο 64 0 16 22 0 13 58 0-22
effect o f subject matter: but it is unlikely, since the subject matter o f
δ ια -f acc. 245 0-62 138 ο·8ι 199 Ο76
8 id 4- gen. 62 0· 16 8 004 OI3 the N E and that o f the E E so closely resemble each other. Once
34
eis 83 0-21 13 008 48 0 18 again, the most economical hypothesis to explain the data is that the
iK (εξ) 130 0-33 50 029 63 0-24 common books, as they stand, belong to the E E rather than to the
€V 482 1-22 117 069 223 0-85 N E.
ε ν ε κ a ( —tv) 49 0· 12 22 0 13 56 0-21
è n l+ a c c. 85 0-22 19 O II 38 ΟΊ4
€7n'-f gen. 75 ΟΊ9 37 0-22 48 o-i8 AD VERBS
επ ί + dat. 82 0-21 17 o-10 55 0-21
κατά + acc. 363 0-92 152 089 249 0-95 After particles, connectives, and prepositions, adverbs present a
μ ετά 42 O II 24 ΟΊ4 31 0-12 promising field for the statistician. T h ey occur frequently and in a
παρά ■+ acc. 25 o-o6 36 0-21 40 O-I5 variety o f contexts, and being indeclinable they are easy to collect
77apa + g./d. 23 0-06 3 o-18 6 0-22
116
from an unlemmatized print-out o f a word-count. M any adverbs in
■περί+acc. 233 0-59 90 °'53 0-44
π ε ρ ί + gen. III 0-28 88 0-52 105 0-40 G reek as in English are formed by terminal modification o f a
7rpos + acc. 255 0-65 107 063 171 0-24 descriptive adjective : in general such adverbs will not be the subject
VITO 61 0-15 20 0· 12 34 0*13 o f a special study here, but will be treated as forms o f the adjective.
Total of 19 2488 6-29 986 5’79 1605 6 10 T h ere are many other adverbs and adverbial expressions whose
study reveals striking differences between the Nicomachean and
T h e figures for each preposition were obtained from a word-count made by the C O C O A Eudemian Ethics which can then be used for comparison with the
program. As the program did not discriminate between cases where the preposition occurred
with different cases, the figures for such expressions as epi are the result o f an apportionment AE.
o f the com puter total between the different cases on the basis o f an independent hand-count One such group of expressions is a set o f adverbial modifiers o f
o f the complete text. degree: hekista (least), hltton (less), hikanös (enough), lian (too
much), mällon (more), malista (most), panu (altogether). T h e
usage in the A E . T h e differences probably show simply that Nicomachean Ethics is uncommonly fond o f expressions o f this
prepositions are not quite as topic-neutral as might at first appear: group: as will be seen from T able 5.2 they make up over 1 per cent
the higher frequency o f epi with the dative in both N E and E E by o f its entire text (as against 0 32 per cent for the A E and 0-55 per
comparison with A E is due largely to the occurrences o f eph' hemin cent for the E E ). This difference, as the table shows, can be
in the discussions o f voluntariness in the second book o f the N E and observed throughout the various books o f the treatises : the lowest
the second book o f the E E . proportion in any book o f the N E ( 0 7 1 per cent in book 1) is
T h us, in sixteen out o f nineteen cases the differences between the approximately the same as the highest proportion in any book o f the
E E and the A E are only such as might be expected by chance; and E E (0 7 3 per cent in book III), and no book in the A E contains these
there is no non-chance difference between the E E and the A E not words in so high a proportion (highest is 0 45 per cent for book C).
matched by a greater difference between the N E and the A E . There W hen the A E is compared with either the N E or the E E the
ii6 The Aristotelian Ethics Prepositions, Adverbs, and Pronouns in the Ethical Treatises 1x7
T a b l e 5.2 A d v e r b ia l M od ifiers o f D egree in In d iv id u a l Books o f the N E , T a ble 5 .3 ά π λ ώ ς a n d μ ό ν ο ν in In d iv id u a l Bo oks o f the N E , A E , a n d E E
A E , and E E
P ro p o rtio n
Propor­ B ook άπλώς μόνον T o ta l 0/
/0
tion
Book ήκιστα TjTTOV ίκανώς λίαν μάλλον μάλιστα πάνυ Total 0/
/0 NE
I 5 4 9 0 - 15
NE 2 6 0-21
3 9
I I 3 6 2 H 13 I 40 071 I 6 0-09
3 5
2 0 7 I 0 17 8 I 34 o-8o I 2 0-05
4 3
3 I 14 0 I 33 14 5 68 i -07 8 10 0-24
4 Η
4 3 6 0 4 51 13 2 79 132 0 2 2 0-03
9
8 3 9 0 0 20 20 8 66 113 10 0 0-08
5 5
9 I 2 I I 26 32 2 65 I -22
10 I 10 2 2 66 ι·ο 6 Total 27 21 48 0 - 12
3 33 15
Total 10 51 II 10 19 4 121 21 4 18 i 06 AE
A 23 II 34 0-53
AE B 16 0-38
7 9
A 0 4 2 0 8 I 0 15 0-23 C 28 0-67
15 43
B 0 2 I 0 4 4 0 II 026
C 0 6 I I 16 0 29 Total 58 35 93 0-55
5 0-45
Total 0 12 4 I 28 10 0 55 0-32 EE
I I 9 10 0-30
EE II II 0-21
4 15
I 0 0 0 I 13 8 0 22 065 III 2 0-19
7 9
II 0 2 0 I 17 6 0 26 0'37 V II 40 0-66
17 57
III 0 5 I 0 18 8 2 34 0'73 V III 0-39
4 5 9
V II I 3 I 2 27 24 0 58 0-67
V III I 0 0 0 I 0 6 0-23 Total 63 37 100 038
4
Total 2 10 2 4 79 47 2 14 6 0-55
the occurrences o f these words in the individual books o f the
differences, for the group of words as a whole, appear greater treatises. T h e lowest scoring books o f the A E and E E (E E I I , with
than those to be expected as a result o f sampling error : when the 0-21 per cent and E E I I I with 0 1 9 per cent) score just below the
A E is compared with the E E there is an absolute z-score o f 3^47, highest scoring books o f the N E (2, with 0 21 per cent and 8, with
when it is compared with the N E the £-score is 8 8 1. But the 0-24 per cent). Ju st as, in T able 5.2, we saw that the A E has a more
differences between the A E and the N E are very much greater than than Eudemian distaste for the Nicomachean favourite adverbs, so
those between the A E and the E E , and both diverge in the same here we see that the A E has a more than Eudem ian liking for the
direction: the occurrence o f these Nicomachean favourite ex­ adverbs most popular in the E E . Bu t in this case the differences
pressions is much less in the A E than in either the E E or the N E . between the A E and E E proportions are not significant, while those
T w o adverbs which, by contrast, are Eudem ian favourites are between the A E and N E proportions are.
monon (only) and haplos (without qualification). T able 5.3 sets out T h e adverbs isös (perhaps) is a distinct Nicomachean favourite,
 118 The Aristotelian Ethics Prepositions, Adverbs, and Pronouns in the Ethical Treatises 119
occurring in the N E with a frequency o f ο ·ι8 per cent by
N O Ο N iri fN N 1Λ o H O 00 comparison with 0 05 per cent in the A E and 0 04 per cent in the
ro 0 —1 N- NH
> 0 N ^ ro M ro Γ-» r o rf- ur> -*f
Ö ό Ö 0^ ό ό 0 0 Ο Ο Ο Ο
E E . It turns out that this is just one o f a number o f tentative
N—^ 'w '
£ 00 "ΦΟΟ r^ o m ΙΛ Th Μ Λ h O' bl fO ’t 10 (VJ ^ expressions which are markedly more popular in the N E than in the
& Ο N NH — Η 1Λ H fO N A E or the E E . T h e five commonest o f these— isös (perhaps), dokei
E E

à
(seems), doxeie (would seem), eoike (seems likely), phainetai
vo (appears)— are studied in Table 5.4 which gives the occurrences,
and

On W Ο N O Ο "Φ u-j vo t θ' N O ^ O'

« Ci book by book, of each. It will be seen that these expressions,
AE,

Ö
-Θ - between them, add up to over 1 per cent o f the whole Nicomachean
text. B y this crude measure, the E E is only half as tentative as the
NE,

O
OO 00 N m t}· 00 w
>no r-ο o
<P~ ·"< HH VT) N Μ ΙΛ N E (these expressions amount to 0-43 per cent o f the text) and the
00
in In d ivid u a l Books of the

P*
A E is even less tentative (0 36 per cent). Other, less frequent,
* expressions o f tentativeness, not shown in the table, exhibit the
0 0 *-« 0 0 0 0 M Ο N u"> ro N N O OO ro im
<N Tf same pattern, such as tacha (possibly) and adelon (unclear), which
-ö
occur between them twenty two times in the N E , four times in the
C E E , and twice in the A E .
+1 •2 ·-< O' r-** n o 00 n t »Λ h O « o n « ro
4-t ro « Ο Γ>» N N >■* o ro r o r o ro N fO i/ l ΙΛ r o ■φ It will be noted that the tentative expressions occur very much
* Ο ~ Ο ~ ~ ~ « 0 Ö Ö 0 ö ό ö ö ö less frequently in N E 2 than in any other Nicomachean book: the
u ^ 00 00 ro Ο 00 ΓΟ rj- N 1Λ t « r-Ό o m oo
CL, *
>»
o ^ r»so r- o N « N Ο N N Tf
n
hH proportion is only 0-19 per cent. Here, as in particle usage, N E 2
■
<t·
resembles the E E pattern in some respects more than the pattern o f
Certainty

its fellow Nicomachean books. I f we leave this book aside, we find

that the lowest-scoring N E book (N E 4, with 0 72 per cent) scores
O' ^
N
ir> 0 0
W
Ο
IX M
N O' l>
OO
N ro i— O N ro Ο 00 O Ov
considerably higher than the highest-scoring E E or A E book (E E
Expressions of Doubt and

I I I , with 0 56 per cent). Once again the A E is more un-

Nicomachean than the E E is; but the difference between A E and
0 0 0H4 «Μ Μ
w>oo ■-« H
H ** « T f (S o
E E is again statistically non-significant, for the group considered as
l-H *H M
a whole.
Table 5.4 shows, in addition to these five expressions o f
tentativeness, three contrasting expressions o f certainty or clarity—
Ο Ο ΙΛ r o ro Ο N N n n V)
anagke (necessarily), delon (clearly), phaneron (obviously).
Ο ιλ t Ο Th
MM Th
M ji Consistently with the hypothesis that the E E is less tentative than
vO
00 the N E , these expressions are about three times as popular in the
5.4

E E , where they make up 0 48 o f the text, as they are in the N E ,

N O* 'Φ N OO OO ^
hHN- M <S - N
O OO ^
hH ^■OO N N vi N >Λθ O where they account for no more than 0· 16 o f the text. T h e A E , with
T a b le

>-. i-η ro N-t M N o

00
° ‘35 Per centi is less prone to use these words than the E E , but not
l/i
3 Ο N r o ·-< 00 10 un o n n rt· 00 by a significant amount; its difference from the N E , however, is
o ►“< ·-· (-1 t-H Γ*>.
significantly large.
T h e data presented in Table 5.4, taken together, amount to quite
O
a striking demonstration o f the greater closeness o f the A E to the
£ £ E E than to the N E . T h e occurrence o f each o f the expressions in the
!20 The Aristotelian Ethics Prepositions, Adverbs, and Pronouns in the Ethical Treatises 121
T a b le 5. 5 Scattergram o f Occurrences o f Expressions o f Doubt and T able 5.6 Scattergram o f Occurrences o f Expressions o f Doubt and
Certainty in AE and EE Certainty in N E and EE
34-000 34 000
δοκεί δοκεί

δήλον
δήλονχ χ
/
/
/
/
27-200 . ... y / . 27-200
/
/
/
/
/
/
/
/
20-400 r , 20-400

1
/
/ C
r

\
\
(Λ
T / 0)

\
o

\
\
C/5
/ c

\
8 / U
u
c u
u φχVEpÔVyf 3O φχ νερό V

\
g o

\
\
/ O

\
\
/

\
O 13-600 13-600
/ X X χ ν ίγ κ η
/ 'ανάγκη
/
/
/
/
/
ίσως / Ίσω ς
X εοικε
’έοικεχ x /
6-800 / 6-800
/ φαίνεται χ
f X
/ φαίνεται δόξειε χ
/ δόξειε
/
/
/
/
/
0 000 / 0000
910 14 19 29 41 56 1 5 42 58 70 87 91
R = 0-9322 O ccurrences in E E R = 0-2366 N E occurrences

A E is not significantly different, when tested by the usual test,1 level for six out o f the eight expressions (all except dokei and dëlon).
from the occurrence in the E E ; by the same test, when we compare When we take the tentativeness group as a whole, there is no
the E E and the N E , the null hypothesis is rejected at the i-per-cent significant difference between the figure for the E E and the A E
I . See above, p. 84. (2 = 1*09), but there is between the N E figure and the A E figure
122 The Aristotelian Ethics Prepositions, Adverbs, and Pronouns in the Ethical Treatises 123
( 2 = 8 0 3); similarly with the figures for the certainty group as a P R O N O U N S A N D D E M O N S T R A T IV E S

whole (z for N E and A E is 4 32, for E E and A E it is 2 07). When we turn from indeclinable adverbs to pronouns and
We can present the data graphically in a pair of scattergrams in demonstratives, and later when we come to consider adjectives,
which the values for individual expressions in the A E are plotted nouns, and verbs, we encounter a new difficulty in handling the
against those in the E E and those in the N E respectively. T h e result data. T h e C O C O A program used for word-counting does not
o f this for the A E and E E is shown in T able 5.5, and for the N E and lemmatize : that is to say, each separate form o f a declining noun or
E E in Table 5.6. It will be seen that A E and E E correlate well conjugating verb appears listed in alphabetical order, not grouped
(r = 0-9322), whereas the correlation between A E and N E is so poor under the form under which the noun or verb would appear in an
that the regression line is valueless. ordinary lexicon. Almost all the data considered so far have
There seems to be less information for our purpose to be gained involved only a single entry in the print-out o f the computer word-
from the study o f other adverbs in the ethical treatises. Tw o count (the exceptions being contractions such as αΙΓ which get
temporal adverbs are Eudem ian favourites, hama (at once) and aei counted as forms o f the uncontracted word, and forms containing a
(always) ; in the case o f hama the A E frequency is closer to the N E terminal n, such as eoiken, which are counted with the unterminated
one, and in the case o f aei closer to the E E one, but in neither case form). When counting Aristotle’s usage o f a pronoun, a noun, or
are the differences between the A E frequency and either o f the verb, we have to group together by eye, and total by hand, the
others significant. T h e values are set out below, along with those o f occurrences o f each o f the separate unlemmatized forms which
three other frequent adverbs: together constitute the score for the single dictionary entry. There
is ample scope for human error in omitting a form or incorrectly
NE AE EE adding a total. T h is should be borne in mind henceforth.
T w o o f the commonest pronouns are houtos (this) and autos (self,
Occur­ Propor­ Occur­ Propor­ Occur­ Propor­ he, same). Between them they make up about 3 per cent o f the text
rence tion rence tion rence tion o f each o f the three ethical treatises. Though there are slight
°/
/0 °/
/0 °/
/0 differences between the frequencies with which they occur in the
1 / different treatises, these differences are not significant.
aei 24 ο·ο6 20 0 12 24 0-09 (Approximate frequencies for houtos: N E 1 5 1 per cent, A E 14 6
αμα 18 0-05 9 005 35 0 13 percent, E E 1 7 5 per cent; for autos: N E 1 44 per cent, A E 1-34 per
δλως 35 ο·ο9 15 009 24 0-09 cent, E E 1 5 0 per cent.) Sim ilarly the figures for the two Greek
ο μ ο ίω ς 94 0-24 29 0 17 50 0-19
words for ‘nobody’, oudeis and mêdeis, show no significant
ό ντω ς 51 ο ·ΐ3 20 0 12 34 0 13
differences between the treatises (provided that oudeis and outheis
are taken together, and likewise mêdeis and mêtheis). T h e figures and
proportions are:
In the case o f holös, houtös, and homoios the frequencies in each o f
the three treatises are too close together to be used for any NE AE EE
discriminatory purpose. T h e evidence provided by the use o f these
words cannot be used unless they can be suitably grouped with Occur- Propor- Occur- Propor- Occur- Propor-
other words— e.g. their parent adjectives— to yield statistically rence tion rence tion rence tion
0/ 0/ 0/
significant results. W e shall see later how this can be done, as it can /0 /0 /0

with some other adverbs here omitted because they occur, in

ouSei's 160 0-40 78 046 91 0-35
isolation, less than twenty-four times in the N E and less than μ η δ είς 43 0 ·ιι 14 008 21 0-08
sixteen times in the E E .
 124 The Aristotelian Ethics Prepositions, Adverbs , and Pronouns in the Ethical Treatises 125
But there are other pronouns and kindred adjectives which are
O' U
i G \0 VO O ' m oo to ■φ r> to o o w more popular in the Nicomachean treatise than in the Eudemian,
O' N 7*· O y^oo m so ro ro
2 « 2 e ^ Ο Ο Ο ^ O *-1 o
6 NH Μ »Lh and vice versa. Three Nicomachean favourites are hekastos (each),
' '— ' >W ' '— ' i ^ ~ ** ~
0 r o NH O ' ro n n cn in ιλ r- toioutos (such), and tosoutos (so big). T h eir distribution, book by
H + 00 OO SO sO l-H ΐΛ Ό i n t J- r o t
ro book, is shown in T able 5.7. It will be seen that, taken together,
5.7 Pronouns and Pronominal Adjectives in Individual Books of the NE, AE, and EE

these three expressions account for over 1 per cent o f the N E , but
Οί
o only 0· 57 per cent o f the A E and 0 66 per cent o f the E E . In the case
Q* η
«
N N
μη
r o o
N n
Ο M
00 OO m sO N co O PO o OO 'r> i-<
l··
ro w ci r* H Ht M Ç* o f none o f these words is there significant difference o f frequency
6^
between the E E and the A E , but in the case of the group as a whole,
ro N O ' co ro »-1 to -Tf ΙΛ rt* η o m so co t o ro and of toioutos considered singly, the differences between N E and
& N co (S
A E show up on the test as significant at the i-per-cent level. Table
t/i 5.7 also contains the figures for five words which are relatively more
O
(S fo t m m t> « N ro Γ*· O ' Γ'·» 0 0 O O frequent in the E E than in the N E and which between them
W h-l W Tf
N li*>
* constitute 1 32 per cent o f the E E text but only 0 96 per cent o f the
Co N E text. T h e frequency o f this group— allèlôn (each other), alios
O Tt- <- ►-( 0 0 'Φ r^· ro r ·- r o 0 0 00
Μ N «Ί 11 w N so n r o >-n r>
ro m
N N ri ^
r o r -· i n
N
ro
Tf (other), ekeinos (that), enioi (some), and heteros (the other)— in the
*0
λ A E is much closer to its frequency in the E E than to its frequency in
3 the N E . T h e differences between the A E frequency and the E E
O OO Ο O O' O' Ό (S to <n ro o *-· O ' in o o O ro
ro v© ro in frequency are non-significant in the case o f every word except
-0 allëlôn : it is clear from a glance at the table that the discussion o f the
topic o f friendship (in books 8 and 9 o f the N E and book V II o f the
^ uI _
’ t Ι Λ O '. T t O o c N ro η X Ο Ό Ο O0 T f O
E E ) causes a massive increase in the frequency o f this pronoun.
O ' τί* r o 0 0 t-ι Ο N o Ί- h 'Λ to 00 «η O ' t Ό
ÖÖ Ô Μ Μ Μ M ö O O O O O
When we compare the individual expressions between the N E and
5 Q . <3
ο ?\2 fo ΟΌΟ o ^ ïn v o r>· the A E v /e find highly significant differences not only in the case of
H + c in H 00 U-, o I'' o O i- ι Ό O0 Γ"
O' c o t J- tJ- r o <-*
allëlôn but also in the case of enioi and heteros. It will be seen from
the table that in the N E the figures for the Nicomachean group and
o the Eudemian group commonly approximate to each other ; in the
< !s
o ^ •0 0 O ' ro « (N Tj- O E E and the A E the Eudemian group in every book scores more
ro
b
o heavily than the Nicomachean group, and in the treatises as a whole
h
occurs at least twice as frequently as it.
tn
O Another pronoun which occurs significantly more frequently in
<» t oTt·«(Ti -τ|- vO
M M in Ό hh so the N E than in the other two treatises is the reflexive heautou
o μ cm ro «Ν o
o (oneself) : it is to be found eighty-three times in the N E , fourteen in
h
the A E , and twenty-four in the E E . It has not been included in
T able

(η
O Table 5.7, however, since its between-books occurrence in the N E
h
O
ΜnO Π
t Η
ΜΌΗΌ w t*· O
ro'
b
ö H ff) fN| m m ο n r>
is even more irregular than that o f allëlôn'. forty-eight o f its
»Vy
Nicomachean occurrences occur in the single book 9.
Words concerned with multitude and magnitude, wholes, parts,
o N r o T j-0 0 O' O and shares form a group which repays study in the ethical treatises.
o « S<cuu
ff) §
o 3
o S3
H Though the commonest word for ‘all’, which comes in two forms,
H > > H
120 The Aristotelian Ethics Prepositions, Adverbs , and Pronouns in the Ethical Treatises 127
pas and hapâs, does not tell us anything o f interest— there are no
T H E D E F IN IT E A R T IC L E
significant differences between the frequency o f either o f these
forms, or in the relationship between them, from treatise to One might expect in advance that if any word at all displayed
treatise— many other words in the group are favourittes o f one or regularity within the work o f an author, it would be the definite
other treatise. T h e Eudemian Ethics is more interested in wholes article, so colourless and so topic-neutral. When we compare the
and parts,1 as is shown by the figures for occurrences o f the word for use o f the different cases o f the article in the three ethical treatises,
‘ whole’ (holos) and o f two words for ‘part’ (meros, morion) : this expectation is not fulfilled. Between the E E and the A E we find
statistically significant differences between the frequencies for two
NE AE EE o f the cases o f the article, the masculine dative singular to( i) and the
masculine nominative plural hoi. Between the N E and the A E we
όλος 40 S» 35 find statistically significant differences in seven cases (the frequen­
μ (ρος 12 28 14 cies for ho, hë, to, hoi, tous, tois, and tais). But the most striking
μ οριον 8 10 17 feature which differentiates the use o f the article in the N E from its
Total 60 ( = 0-15%) 76 ( = 0-45%) 66 ( = 0 25%) use in the E E is the much greater preference which the latter has for
the singular forms. T h e figures below give the proportion (per cent)
T h is interest, it will be seen, is shared in an exaggerated form by the o f each book constituted by (a) singular forms o f the article, (b)
A E , which in respect o f these words as a group differs significantly plural forms o f the article, (c) the sum-total o f both singular and
from both the other treatises. T h e Nicomachean Ethics has a liking plural forms.
for two words to do with parts and sharing (koinos, ‘common’ , and
loipos, ‘rest’, for which the figures are respectively: N E 48, A E 12 ,
Book NE i NE 2 NE 3 NE 4 NE 8 NE 9 N E 10
E E 2 1 ; N E 39, A E 3, E E 2) ; but has a more general preference for 8-32 7-86 7-96 8-14
Sing. 7-26 9-3I 7-5I
words to do with multitude (such as oligos, ‘ few ’, and polys, ‘m any’) Plur. 5-00 5-82
5-7 I 4'47 5-46 5-05 5-97
and magnitude (such as pleion and pleistos, the comparative and
Total 12-97 14-31 I 3-33 12-79 13-32 I 3-01 14-11
superlative o f polys ; and mikros, ‘sm all’, and its comparative elattön
and superlative elachistos). The figure for occurrences o f these
Book EE I EE II E E III E E VII E E V III
words are as follows. Megas (‘great’ ) with its comparative and
Sing. 9-25 954 8-73 9-98 7'97
Plur. 541 362 4-92 3-78 3-20
NE AE EE 13-76 11-17
Total 14· 66 I 3 J6 13-65
oAiyoç 27 3 8
770Aii? Γ45 26 76
ττλίωνρπλίΐστος 75 33 39 It will be seen that four out o f five o f the Eudemian books have a
μ ικ ρ ό ς 56 3 21 proportion o f singular forms o f the article o f more than 8 5 per cent.
έλά ττ ω νρςλά χισπ ος 27 31 17 Only one o f the Nicomachean books, N E 2, whose similarity to the
Total 330 ( =ο·83% ) 9 6 ( = ο ·56%) ι6 ι ( Eudem ian books has several times been remarked, has a score in
excess o f this. I f we take the plural forms, only the first o f the
superlative is also a Nicomachean favourite, but it will be more Eudem ian books has a score o f more than 5 per cent here; all except
convenient to consider it later in a different context. book 4 o f the N icomachean treatise score over 5 per cent. Here, as so
often, the disputed books display the Eudem ian features in an
I. See above, p. 66. accentuated form. T h e figures for these books are as follows :
128 The Aristotelian Ethics
Book AE A AE B AE C
Sing.
10-90 9-3 I 9’37
Plur. 2-72 4 10 4-04
Total 13-61 I 3'4 I 13-41
Altogether, the proportion o f all occurrences o f the article
constituted by singular forms is 59-7 per cent in the N E , 69· 5 per
cent in the E E , and 75-7 per cent in the A E .
G R O U P IN G IN D IC A T O R W O R D S
T h « words which we have so far studied amount to some 53 per
cent o f the entire treatises. In each treatise, roughly 25 per cent
consists o f particles and connectives, 6 per cent of prepositions, 9
per cent o f adverbs, pronouns, and similar words, and 13 per cent of
the definite article. In respect o f every feature so far in which there
was a significant difference between Nicomachean and Eudemian
usage the common books resembled the Eudem ian pattern more
than the Nicomachean. Undeniably the results so far presented
amount to a substantial argument in favour o f assigning these
disputed books to an original Eudemian context.
However, there are two ways in which the case so far presented
has been incomplete. In the first place, many o f the vocabulary
habits studied in the present chapter have been too irregular to form
the basis o f any rigorous statistical argument. From time to time a
test has been used to determine which frequency differences were
significant and which were not. T h is procedure in the present
context is useful as a rough discriminator: but the result o f the test
can only be regarded as giving an accurate measure o f the
probability o f a difference in frequency being the result o f chance if
the data in question conform to the conditions o f simple sampling.
In chapter 4 a chi-squared test was undertaken to ensure that this
condition was met, and it was found that in the case o f the majority
o f particles studied it was indeed satisfied. In the present chapter no
such test has been undertaken: but it is obvious to the naked eye
when we study the book-by-book occurrences o f several o f the
expressions that more than chance fluctuations occur between one
book and another. It would be desirable to find a procedure which
would enable us to tighten up the statistical argument from one
treatise to another.
Prepositions, Adverbs , and Pronouns in the Ethical Treatises 129
In the second place, even the most powerful argument to show
that, considered as a single unit, or even book by book, the
Aristotelian Ethics belong with the Eudemian would be insufficient
to settle the age-long scholarly dispute about the origin o f the
common books. F o r, as has been remarked, many scholars favour a
solution which sees the common books as made up o f different
strata from different periods, form ing a patchwork. At the end o f
chapter 4 we considered the occurrence of particles in each o f
seventeen iooo-word samples o f the A E : the results were
interesting in that they showed no trace o f patchwork composition;
but in the case o f many particles the frequencies were too small for
complete confidence to be placed in the result. It would be desirable
to employ a method o f studying the small samples which would
generalize the inquiry to include words occurring even less
frequently than particles, while enabling firmer statistical con­
clusions to be drawn.
We can achieve both these results if we take words together in
groups and count not the occurrence o f single words, but the
occurrence o f a group, the occurrence o f any word in the group
counting as an occurrence o f the group. T o serve our purposes, a
group must satisfy two conditions. First, the group must be one
which is characteristic either of the Nicomachean or the Eudemian
Ethics: it must consist in general o f words which are either
Nicomachean favourites or Eudem ian favourites; and the occur­
rence o f the group as a whole must be substantially more frequent in
one of the treatises than in the other. I f this is not so, the occurrence
o f the group within the A E will be useless as a discriminator and
will yield no information about its likely original context. Secondly,
the group must be one which occurs frequently enough in the
treatise o f which it is characteristic to enable us to make a reliable
prediction o f its occurrence in any randomly selected iooo-word
sample o f material homogeneous with that treatise. I f this is not so,
it will be impossible to tell, when the actual occurrence in the
sample differs from the expectation derived from one o f the rival
contexts, whether this is due to a genuine difference o f style or
merely to a chance sampling fluctuation.
We can ensure that our groups satisfy the former condition by
calculating, for each word and group a distinctiveness ratio.1 In the
I. T h e notion o f distinctiveness ratio is adapted from Ellegârd, 1962, 20 ff.
130 The Aristotelian Ethics
present context, the D .R . of an expression is the ratio o f its
frequency in one treatise to its frequency in the other. L e t us
express the D .R . o f an expression occurring in the ethical treatises
as its Nicomachean frequency divided by its Eudemian frequency.
T h u s, a word which occurs more frequently in the Nicomachean
Ethics will have a D .R . greater than unity, a Eudem ian favourite
will have one between one and zero, and a word which is used with
the same relative frequency in each will have a D .R . o f one.
We may expect the D .R . o f many words, in two treatises by the
same author, to be very close to one: words with such a D .R . will
obviously be useless as discriminators. For our purposes we must
look for words, or groups o f words, whose D .R . is some distance
from unity— say more than 1-4 or less than 0-7; words, in other
words, which occur about three times or more in one treatise for
every twice they occur in the other. W hen we group words, it is not
necessary that every word in the group should have a marked D .R .,
but it is important that the group as a whole should have: that is,
that the occurrence o f a-word-from-the-group should be a
considerably more frequent event in one treatise than in the other.
T o ensure that our group enables predictions to be made o f its
occurrence in samples as small as 1000 words it is necessary to insist
on a minimum absolute frequency as well as a characteristic relative
frequency. In practice, we need groups which will enable us to
predict an absolute occurrence o f not less than ten in each sample of
1000 words. Since the Nicomachean Ethics is 39525 words long, this
means that we can make predictions based on Nicomachean
frequencies o f 400 and above: for if a word or group occurs 400
times in a text o f about 40000 words it can be expected to occur ten
times in a iooo-word sample homogeneous with it. T h e Eudemian
Ethics is 26330 words long, and so for predictions based on
Eudem ian frequencies we need words or groups occurring at least
260 times in the E E .
W e shall look, therefore, for groups o f words with the following
characteristics. A group should be either a Nicomachean favourite
group with at least 400 occurrences in the N E and a distinctiveness
ratio o f at least 1 -4 ; or it should be a Eudem ian favourite group with
at least 260 occurrences in the E E and a D .R . o f less than 0-7. On the
basis o f the Nicom achean favourite groups we can calculate a
Nicomachean expectation for each o f our seventeen samples o f the
A E , and on the basis o f the Eudem ian favourite groups we can
Prepositions, Adverbs, and Pronouns in the Ethical Treatises 131
calculate a Eudem ian expectation. B y comparing the expected
number o f occurrences o f each group in each case with the actual
occurrences, we can determine for each sample in turn whether it
resembles the Nicomachean Ethics more than the Eudemian. Any
substantial patching, say, o f an original Eudemian text with
Nicomachean material should display itself in a fluctuation between
one sample and another in respect o f comparative resemblance.
It turns out that it is possible, from the data that we have already
recorded, to assemble six Nicomachean and six Eudem ian groups
answering to these specifications. T h e first Nicomachean group is
formed by taking together five particles popular in the N E : de,
eiper, kathaper, hothen, and oun. Taken as a whole the group has a
D .R . of 2 00 and occurs 587 times in the N E , so that words from the
group could be expected to occur 14 85, i.e. fifteen times, in a iooo-
word sample homogeneous with the N E . T h e second group
consists o f the single preposition en, with a D .R . o f 1-44 and a total
o f 482 occurrences in the N E . T h e third group has already been
studied as a group : the qualifying adverbs hëkista, hêtton, hikarns,
[fan, malIon, malista, pany occur altogether 4 18 times in the N E , and
the group as a whole has a D .R . o f 1 9 3 . E ven more distinctive is the
group o f expressions o f tentativeness already studied : isos, dokei,
doxeie, eoike, phainetai. T h is group occurs 404 times and has a D .R .
o f 2 3 7 . T h ree Nicomachean favourite pronouns and
demonstratives— hekastos, toioutos, and tosoutos— form another
group with a D .R . o f 1 -56 and 407 occurrences in the N E . T h e final
Nicomachean group is made by putting together the two
Nicomachean favourites koinos and loipos with the ‘multitude and
magnitude’ group of oligos with mikros and polys and their
comparatives and superlatives. T h is group has a total occurrence in
the N E o f 4 17 and a D .R . o f 1-5 1.
T h e six Eudem ian groups are made up as follows. T h e
conjunction alia is popular enough to constitute a group on its own,
with 3 1 1 occurrences in the E E and a D .R . of 0 54. T h e two
connectives dio and holt form a group occurring 263 times with a
D .R . of 0 58. A third group can be made up from three more
Eudemian connectives, hôsper, hôste, and epei : this highly character­
istic gr6up has a D .R . of 0 38 and occurs 279 times in the E E . We
have already seen that anagkê, delon, and phaneron form a
characteristically Eudemian set o f expressions, but as they make up
only 0 48 per cent o f the Eudemian text they are not by themselves
 T ab le 5.8 Nicomachean Favourites in Seventeen Samples o f the AE: Part One

Group N E D.R. Expecta- Actual occurrences in samples Mean

total tion in —--------------------------------------------------------------------------- actual
1000 i 2 3 4 5 6 7 8 9 10 i i 12 13 14 15 16 17 occur­
rences

N I {de, eiper
kathaper, hothen
oun) 587 2-00 1485 88 10 7 7719 9 9 10 12 10 10 12
N II (en) 482 14 4 12-19 10 20 5 6 5 9 ii 6 4 5 7 6 3 6 6 2 5 6-8
N III (hêkista,
he!ton, hikanos,
Itan, mâllon,
malista, pany) 418 1-93 10 58 3 2 2 2 1 5 2 1 5 3 4 3 3 6 3 6 2 3-2
N IV (isös, dokei,
doxeie, eoike,
phainetai) 404 2-37 10-22 9 2 2 2 1 4 2 3 5 4 2 2 1 2 4 3 8 3-2
The Aristotelian Ethics

N V (hekastos,
toioutos, tos-
outos) 407 1-56 10-30 4 3 6 8 2 5 2 7 9 8 7 7 1 1 2 8 5 1 5-5
N VI (oligos,
polys, pleön,
mikros, etc.,
koinos, lotpos) 417 1-51 10-55 20 12 12 5 6 5 2 5 2 2 o 2 8 7 7 3 62
Total N I-V I 27 I 5 6869 42 55 37 21 36 41 28 37 32 34 27 30 36 36 33 31 34-8

T able 5.9 Eudemian Favourites in Seventeen Samples o f the A E: Part One

Group E E D.R. Expecta- Actual occurrences in samples Mean

total tion in ------ -------- --------------------------------------- ' actual
1000 i 2 3 4 5 6 7 8 9 10 i i 12 13 14 15 16 17 occur­
rences

E I (alia) 0-54 1 1 *81 !3 II 7 15 17 18 10 !5 8 14 16 7 II 13 l 3 13-8

3 11
E II (hoti, dio) 263 0-58 9-99 15 3 II 13 6 i i 8 17 8 15 7 8 13 4 15 10 98
E III (hosper,
höste, epei) 279 0-38 io-6o 6 8 II 8 12 9 5 8 8 10 8 13 16 20 15 12 I I -4
E IV (anagkê,
dëlon, phaneron,
monon, haplos,
mentoi, toinun) 256 030 9-72 14 12 5 13 10 13 6 7 12 4 6 9 l 3 16 7 11 5 96
E V (aneu, heneka,
para + acc.
pen + gen.
hama) 267 048 10-14 9 6 2 7 16 10 13 14 13 16 17 12 16 6 9 5 6 10-8
E VI (ekeinos,
heteros, alios
ii
Prepositions, Adverbs, and Pronouns in the Ethical Treatises

enioi, allëlôn) 380 0-73 14-43 17 12 16 17 10 12 23 10 9 13 13 12 8 8 13 6 12-4

Total E I-V I i 756 66 69 74 62 52 73 71 72 54 79 75 51 75 65 67 66 59 72 52 65-8
133
Σ34 The Aristotelian Ethics
frequent enough to make up one o f our test groups. T h e group
remains too small i f we add to it the 0 38 per cent o f the text made up
by the characteristically Eudemian adverbs haplôs and monon. We
may add further two particles, mentoi and toinun, which we saw in
chapter 4 to be Eudemian favourites. T h is brings our group to a total
o f 256 occurrences, near enough to 260 occurrences to predict an
expectation in 1000 words approximating to ten. T h e D .R . o f the
group so formed is 0-30: it is thus an extremely characteristic
Eudemian group. For the next group we take four prepositions
popular in the E E — aneu, heneka, para with the accusative, and peri
with the genitive— and add to them the popular adverb hama. This
gives us a group with D .R . 0 48 which occurs 267 times in the E E .
T h e pronouns and pronominal adjectives which we noticed earlier as
characteristic o f the E E — allëlôn, alios, ekeinos, enioi, heteros— will
make up a sixth and final group, as they occur in total no less than 380
times. But the group is not as distinctive as the others, with a D .R . as
high as 0 73.
T able 5.8 compares the expected frequency per 1000 words of
the Nicomachean groups with the actual frequencies in the
seventeen approximately iooo-word samples o f the A E , and T able
5.9 compares the expected frequency o f the Eudemian favourites
with the actual occurrences in the same samples. T h e final line o f
each table gives the mean occurrence per sample. As in the case of
the particles studied in the seventeen samples in chapter 4, the
results are strikingly uniform in their tendency. In none o f the
samples does the actual occurrence o f the Nicomachean indicator
words reach higher than 80 per cent o f the Nicomachean
expectation; in most o f them the actual occurrence is about h alf the
expectation, and the mean is exactly 0 5 1 o f the expectation. I f we
turn to the Eudem ian favourites we find that in nine o f the sixteen
samples the actual occurrence surpasses the Eudemian expectation ;
and the mean o f the actual occurrences is 0 99 o f the expectation. In
every single sample the Eudemian expectation is approximated
better than the Nicomachean one: once again there is no sign o f
patchwork. Indeed, even in the case o f the Nicomachean favourites
the actual occurrences in the seventeen samples are closer to the
expectation which would be derived from the Eudemian use o f the
same expressions than to the Nicomachean expectations. T h is can
be seen in the following list:
Prepositions, Adverbs, and Pronouns in the Ethical Treatises 135
Group I II III IV V VI
A'-Exp. 14-85 12*19 1058 10-22 10 30 10-55
£-Exp. 7-42 847 5-48 43° 6-60 6-99
Actual mean 99 6-8 3-2 32 5'5 6-2
Bu t of course the Eudem ian frequency o f these words (which ex
hypothesi are much rarer in the E E than in the N E ) is not as accurate
an indicator o f the frequency to be expected in the A E i f that is
homogeneous with the E E as the Nicomachean frequency is o f the
frequency to be expected in the A E if that is homogeneous with the
N E . In the case o f these groups, therefore, the failure to meet the
N E expectation is o f greater moment than the degree o f
approximation to the E E expectation.
T h e grouping procedure we have adopted goes some way to
remedy the incompleteness o f the statistical argument hitherto.
Grouping words helps to smooth out the fluctuations o f individual
words from book to book o f a treatise and to approximate the
conditions o f simple sam pling; the greater frequency o f group
occurrences in comparison with word occurrences also means that
the underlying binomial distribution approximates more closely to
the normal distribution presupposed by the most easily employable
significance tests. T h e grouping also means that we are able to apply
the large amount o f information about vocabulary frequency
available from the comparatively lengthy texts o f the N E and E E to
small portions o f the A E no larger than 1000 words in length. We
shall see in chapter 6 how this procedure can be carried even
further.
T h e particular groupings adopted have been to a considerable
extent arbitrary. It is important that the reader should appreciate
that this does not affect the validity o f the statistical procedure
adopted. T h e application o f statistical techniques to vocabulary
choice is based on the fact that the choice o f a word-token in a text
can be considered as a binomial event which can have one o f two
outcomes : either the word-token in question is an instance o f a
word-category or it is not. F o r the statistical techniques to be
applicable what is necessary is that it should be unambiguously
decidable whether the token falls under the category or not. But it is
no more difficult to decide whether a given token, say the
I 36 The Aristotelian Ethics
seventeenth word in the second o f the disputed books, is a token o f
either o f the types hoti or dio than to decide whether it is a token o f
the type hoti ; and indeed a decision whether a word-token counts as
an occurrence o f one o f our groups is exactly the same kind o f
decision as the decision whether it is an instance o f the lexicon-
word-type logos (i.e. is it logos, or logon, or logon, etc.). The
argument could have proceeded in the same way if we had made our
groups o f 400 Nicomachean occurrences, or 260 Eudemian occur­
rences, out o f words sharing a common distinctiveness ratio, or a
common initial letter. T h e grouping has been chosen to reflect
incidentally certain features o f the N E and E E other than their
statistical relationships— attention has been drawn, for instance, to
the fondness o f the N E for adverbs o f degree and expressions o f
tentativeness. T h e judgement that the N E is more tentative than
the E E is, o f course, one that goes beyond the merely statistical
evidence: it involves judgements— as the statistical data do not—
on the meaning o f the words being counted. T h e reader will be able
to evaluate for him self such judgements in the light o f the data and
o f his knowledge o f G reek; but the judgements could all be
mistaken, and the groupings totally incongruous, while leaving
unaffected the statistical argument to the effect that the A E
resembles the E E much more than it does the N E . 1
i. T h e grouping procedure is an adaptation o f that to be found in Ellegârd, 1962. M y
method differs from his principally in that I have adapted the notion o f distinctiveness ratio
so that it fits a problem o f determining between two rival contexts for a text, rather than a
problem o f assigning a text to one o f an indeterminate number o f authors, and in that I have
grouped my words on the basis o f syntactical and semantical criteria rather than on the basis
o f D .R .
CH APTER 6
The Use of Technical Terms
in the Ethical Treatises
M o st o f the words studied so far have been ones which are
comparatively little affected by changes in subject matter : function
words which are likely to occur with comparative frequency
whatever the topic o f discussion. Few words are completely
unaffected by changes o f topic: at the end o f chapter 5, for instance,
it could be seen from T able 5.8 that words such as ‘m ore’ and ‘ less’
occur with more than their usual frequency when justice is the
subject being considered. And in philosophy above all, no word is
completely topic-neutral ; for even the most colourless words such
as ‘i f ’ and ‘ therefore’ may themselves provide the topic o f
discussion. But in comparison with most nouns, verbs, and
adjectives the words we have studied suffer little fluctuation due to
subject matter.
In the present chapter we shall turn to the remainder o f the
vocabulary o f the ethical treatises to see what information it can
yield about the original context o f the disputed books. We can no
longer hope to be able to trace regularities which continue from
book to book with only sampling variations : the best we can aim to
do is to detect the influence o f stylistic preferences in fluctuations
which are the resultant not only o f style but principally o f topic. We
cannot aim to locate the origin o f the A E by asking whether there is
one o f its rival contexts which it differs from only in statistically
insignificant ways : we must expect it to differ significantly from
each o f the N E and E E and it will be only from the comparative
magnitude o f the differences that we may hope to draw any
conclusions. G iven that the topics treated in the A E differ from
those treated in the N E and E E more than these do from each other,
we may well expect that the N E and E E will resemble each other
more than either of them resembles the A E . But, as before, we shall
be on the look-out for words in which the N E and E E do not closely
γ38 The Aristotelian Ethics The Use o f Technical Terms in the Ethical Treatises 139
resemble each other, words with a high distinctiveness ratio. For other verbs o f saying (phëmi makes up only 12 per cent o f the verbs
any such comparison to be meaningful we shall need words which, o f saying in the N E as against 20 per cent in the E E and 24 per cent
though not topic-neutral, are not connected exclusively with topics in the A E ) . L ike phëmi, lego is both absolutely and relatively more
which are treated only in the A E , or only in the N E and E E . W e shall popular in the E E than in the N E (56 per cent in the E E vs. 5 1 per
therefore not consider names o f particular virtues and vices, as cent in the N E ) ; while both e r ö and eipon are both absolutely and
such, nor words for the cluster o f concepts surrounding friendship relatively more popular in the N E than in the E E (making up
(discussed only in N E and E E ), nor the names o f the intellectual between them 38 per cent as against 24 per cent). In the case o f erô
characteristics listed in A E B , nor the like. W e shall discuss words the A E approximates more closely to the E E frequency, and in the
connected with the topics which to a greater or lesser extent run case o f eipon to the N E frequency ; but in neither o f these cases are
through all three o f the treatises: verbs o f saying and thinking; the differences, considered in isolation, significant. I f we turn to
logical, metaphysical, and psychological technical terms of particular forms o f the verbs, we find that legetai is an E E favourite
Aristotle’s philosophy; general evaluative expressions and words (frequency 0 1 3 vs. 0-08 per cent in the N E ) whereas eirëtai is an
linked with pervasive ethical concepts rather than particular ethical N E favourite ( 0 1 5 vs. 0 09 per cent). T h e A E goes with the N E in
topics. T h e borderline between usable and unusable material will respect o f legetai (0 08 per cent), with the E E in respect o f eirëtai
not be a sharp one, and arbitrary decisions o f exclusion or inclusion (0 08 per cent). In neither case are the differences significant.
will have to be taken from time to time. Fortunately, as will be seen,
such arbitrary decisions are unlikely to have affected the over-all
L O G IC A L T E R M IN O L O G Y
outcome.
V E R B S OF S A Y IN G Scholars have often remarked that the E E is ‘ more logical’ than
the N E . T o verify this one may construct a list o f logical terms,
T h e commonest verbs o f saying in the ethical treatises are lego, Aristotelian technicalities, and investigate whether their occurrence
phëmi, and the available tenses o f ero and eipon. The figures for these is more frequent in the E E than in the N E . T able 6.1 sets out the
verbs are: results o f such an inquiry. It will be seen that the twenty-two logical
terms and their cognates which are listed in the table occur more
NE AE EE than twice as frequently in the E E as in the N E , amounting in the
E E to about an eightieth o f the entire text. T h e frequency o f these
Occur­ Propor­ Occur­ Propor­ Occur­ Propor­
rences tion % logical technical terms is even higher in the A E , where about one
rences tion % rences tion %
word in every sixty-five is a term o f logic from the list. Part o f the
Aeyto 189 0-48 97 149 high frequency in the A E is no doubt due to the fact that the second
0-57 0-57
φ ημί 44 01 I 46 0-27 54 0-21 o f the three books o f the A E deals with the intellectual virtues; but
epio 103 026 32 0-19 43 Ο 16 the results shown in the table cannot be explained simply as the
€LTTOV 36 0-09 17 o-io 22 o-o8 result o f variations in subject matter. J f that were so, the
Total 372 094 192 1- 13 268 ΙΌ2 frequencies in the N E and the E E should resemble each other in
contrast to the E E . In fact, in spite o f the difference in subject
It will be seen that verbs o f saying, as a whole, are more popular in matter between the A E and the E E , we have once again a situation
the E E than in the N E , and more popular again in the E E : but these in which the difference in over-all frequency between the A E and
differences do not appear to be significant. W hat is significant is the E E is not significant by our test, whereas the differences between
greater frequency o f phëmi in the A E and E E than in the N E : this either o f them and the N E are highly significant.
comes out both absolutely (where the difference between the A E Five other words which are too homespun to be called logical
and the N E is highly significant by our test) and relatively to the technicalities, but which occur frequently in logical contexts,
14° The Aristotelian Ethics The Use o f Technical Terms in the Ethical Treatises 141
T a b l e 6.1 Logical Terms in the NE, AE, and EE
NE AE EE
NE AE EE
Occur­ Propor­ Occur­ Propor­ Occur­ Propor­
α κ ρ ιβ ή ς, α κρ ίβ εια 10 3 5 rence tion % rence tion % rence tion %
ά μ φ ισ β ητεω , ά μ φ ισ β ή τη σ ις II 5 II
αναλογία, dmAoyoç 9 27 16 αληθής 41 0·Ι0 31 ο·ι8 33 0-13
αναλυτικά 0 2 3 λόγος 106 0-29 84 ο·49 107 0-41
ά ντίκ ειμ α ι, άντιτιθτημι, ονομ,α 7 0-02 18 on 8 0-03
ά ντίθεα ις 17 8 7 ορθός 33 ο·ο8 49 0-29 30 o-ii
άποδείκνυμι, ά πόδειξις, φευδής 7 0-02 10 006 5 0-02
δ είκ νυ μ ι 6 16 15 Total ο·49 192 113 0-70
194 183
άπορεω , απορία, διαπορεω 24 13 23
άποφάναι, άττόφασις,
κατάφ ασις, φάσις 6 II 4
M E T A P H Y S IC A L T E R M S
άφορίζω, διορίζω , ορίζω ,
π ροσ διορίζω 41 20 37 We may turn from logical technicalities to the technical
δ ια ιρ εω , δ ια ίρ εσ ις 9 8 20 expressions o f Aristotelian metaphysics. T able 6.2 sets out the
ελ εγχ ω , ελεγχος I I 2 figures, book by book, for words expressive o f eight fundamental
επ α γ ω γ ή I 3 3 concepts which contribute to the construction o f the metaphysical
(ναντίος, εναντιόω,
ύπεναντίος, ύπεναντιόω 29 framework o f the ethical treatises. It will be seen that four o f the
37 74
καθόλου 12 22 expressions, or groups o f expressions, are Nicomachean favourites,
9
λ ο γίζεσ θ α ι, λ ογισ μ ός 6 10 25 and four are Eudem ian favourites. T h e notion o f becoming,
λόω, λυσις 0 6 3 expressed in the noun genesis and the verb ginomai, is twice as
μ ετα φ ερω , μεταφορά 2 4 5 prevalent in the N E , where these expressions form 0 70 per cent of
ομώ νυμος, συνώνυμος I 3 2 the text, as in the E E where they form only 0-28 per cent. (In the A E
π ολλα χώ ς etc. λέγεσ θαι 3 5 7 these expressions form 0 40 per cent.) T h e Aristotelian technical
συλλογισμός, πρότασις, terms for operation or actuality, the noun energeia and the cognate
σ υμ π ερ α ίνειν, όρος 2 40 37 verb energein are likewise favourites o f the N E , with a frequency
τίθτημι, ύ π ο τίθ η μ ι, ύπόθεσ ις 28 10 25 there o f 0-33 per cent against o-io per cent in the A E and 0-13 per
ύπολαμβάνειν, ύπόληφ ις 10 iS 5 cent in the E E . T w o other notions which occur much more
Totals 236 261 338 frequently in the N E are those o f perfection (expressed by the
°'6o% ) ( = i ' 53%) ( = 1-28%) adjective teleios and the verb teleiod) and that o f something’s being
naturally proper (oikeios) to something else. It will be seen that there
display the same pattern. T h ey are alëthës (with alëtheia), orthos is a considerable fluctuation from one book to another o f the N E ,
(with orthotês), logos, and pseudës (with pseudos and diapseudesthai), and two books o f the N E (3 and 4) score less highly in respect o f this
and onoma. Here again we have a substantially greater frequency in group than the highest-scoring book o f the E E (II). T h e fluctuation
the E E than in the N E for the group as a whole, and a frequency in cannot be due entirely to subject matter, since the highest-scoring
the A E greater than either; but here for the group as a whole, and for E E book has very much the same subject matter as the two lowest-
orthos in particular, the A E frequencies are significantly different scoring N E books. I f we take the figures for the treatises as a whole,
from either o f the other treatises. we find that in every case a test shows the differences between the
 142 The Aristotelian Ethics The Use o f Technical Terms in the Ethical Treatises 143
E E and the A E to be non-significant, and those between the N E and
w—I -f- U \0 O OO N OO ΡΊ O OO Ό ro O t o
« Γ-* t-* ro the A E significant. Bu t any conclusions to be drawn from this must
Ό (S Ό O r^» r>· t O' wi N to o
o 'S Q, ό ο ö 6 ô ö ö be drawn in the light o f the marked between-books fluctuation.
~ r° 9 ^0 C
h T
O Oh
2 o
ts© ^ 10 N h o m N t r-* ro moo N « Γ-* T h e first Eudem ian favourite is the notion o f principle, expressed
ro >-i ro t ro ro O \ÛO h o n t-- N o m Ό
N by the noun archë and its cognate verb arc ho. These expressions
</>No
Metaphysical Technical Terms in Individual Books of the NE, AE, and EE

H * M w m 00 Μ n tm t>· O ' Ο Ό Ο t T able 6.3 Scattergram o f Occurrences ofMetaphysical Technical Terms in

» Έ δ I—
I I—I sO Μ H-. Tj- W (S Μ N o NE and AE
73000
3 * φ ν σ ις
.ä
00 co. ro
μ μ ro t
t t
a. γίν ο μ α ι )
a
t>
3
o ιλ μ Ο Ο in (s\û t O ' r o »Λ Γ-» Ο 00 ro O 58-400

<
<N ro

\
Nw

\
60 X α ρ χή

\
*

\
\
y
^ 3 y
5 έ< X N (N ο N tO C m ro 'O
vo m O' t o ofS -< y
■ - i N N CS y
9*·*0
"Ö y
r
y
y '
-j- W<\0 1 * ro n O ' «Λ ι λ χ <N 00 o o 'o' 'o' m ir> m ro 43-800
u->

<
t η v© r o N >Λ h

\
ro t 1 0 Γ» m m t- ι t
£ w o

\
ό 0 Ö ^6 Ö

\
10 ώ ^ P S sO
'—■ e«

\
B.‘C 0 t^ o o m u
u y
O' m t <s 3 ενόεχομαι y
U
υ X X y
O. 3 0 σνμβζti VCtJ y S
S '0 Ul *
y
h j! ϊ ii O ro O OO m o t O' o m o «
y
ro 10
Ί 'δ y
*· t- 29-200

\
\
C/i

\
V )W t η m NH ο Ό Ο
** N
o
S
Vi»
3 ενέργειχ
6.2

^ ^
N Q. OO Ό Γ'·* * * ^ sO O Ο Γ-» O t N X
vVy Q-
Ί» n m r-*
14-600
T a b le

-Ô -è

S
t> 5a. ^τέλειος
Ό N η* f- 00 N O O r o OO O l> N h · Γ" sO N
« η t
" vu
S/ v2
§ N t fO n i r j u*> 10 M 1-H N vO o
c<
X οικείος

ΓΊ ro 'i-oo O' O ^*4 <5 M U 0 000

§ a
« - s ^
H £ Ä = 0-3692 NE occurrences
144 The Aristotelian Ethics The Use o f Technical Terms in the Ethical Treatises 145
TA B LE 6.4 Scattergram o f Occurrences o f Metaphysical Technical Termsin expression endechesthai alios echein, to be capable o f being
EE and AE otherwise) and sumbainein (which occurs frequently in the form
φύσιΐ kata sumbebëkos, meaning per accidens). Finally the E E is much
7 3 000 7
/
more interested than in the N E in the notion o f nature : the words
/ physis and physikon occur in the E E with a frequency o f 0-39 per
v iv o u tx i /
X
/ cent, in the N E with only 0 1 6 per cent. In the A E they occur at a
/
/
rate o f 0-43 per cent, a frequency higher, but not significantly so,
i than in the E E . Once again there is a considerable variation from
/
58-400 / book to book, one book o f the E E scoring less than 0 50 per cent for
/ this group while four books o f the N E score above that figure.1
αρχή
/ * Table 6.3 shows the result o f plotting the A E scores for each of
/
/
these words against the N E scores : the correlation, it will be seen, is
/ very low, 0 3692. T able 6.4 shows the result o f plotting the A E
/
j scores against the E E scores: the correlation this time is high, r
/
43-800 /
being 0-9576.
/ Other metaphysical technical terms, such as ‘potentiality’,
c /
υ
U dynamis, plus dynamai, and dynaton (N E 7 6 = 0 -19 per cent, A E
ui Ιν δ έ χ ο μ χ ι σ υ μ β α ίν ω
O
o
3
X 41 = 0 -24 percent, E E 59 = 0-22 percent) and ‘ function’, ergon(N E
* /
o J 63 = o x 5 per cent, A E 16 = 0 09 Per cent> E E 4 5 = 0 · 17 per cent)
b) /
t are not sufficiently distinctive of one or other treatise to serve as
29-200 / discriminators.
/
/
/
/
/
P S Y C H O L O G IC A L T E R M S
/
/ In the realm o f psychology as in metaphysics we can identify
/
r , . x Nicomachean and Eudemian favourite expressions. T h ere are two
ε ν ε ρ γ ε ί <x
14-600 /
verbs for ‘think’ — axioö and oiomai— that are more frequent in the
/
/ Eudemian Ethics: the figures are N E 20, A E o, E E 29 and N E 18,
τ έ λ ε ιο ί/
V
A E 24, E E 32 respectively. But it is with affective, rather than
/ cognitive terms, that we see a distinct pattern o f preferences
/
emerging. T ab le 6.5 sets out the occurrences in each book o f three
/ . *
/ O tK C tO S
/
Nicomachean and five Eudemian sets o f expressions relating to
0-000 affective states and attitudes. Haireomai, ‘choose’ and its verbal
8 12 33 36 44 74 83
R = 0-9576 EE occurrences adjective hairetos are more popular in the N E ; whereas the more
technical word for ‘ choice’ , prohaireomai, with the adjective
occur with a frequency between them o f 0 32 per cent in the E E ,
and approximately the same frequency (0 33 per cent) in the A E ; I. T h e one constant featu reof T ab le 6.3 is that for every book o f the NE , the frequency for
they are just over half as frequent ( 0 1 8 per cent) in the N E . T h e E E the N E favourites is higher than that for the E E favourites ; while for every book o f the E E
and for every book o f the A E the frequency for the E E favourites is higher than that o f the N E
is fond of two words to express contingency, endechesthai (which favourites. T h is constancy is all the more striking given the violent fluctuations which the
often occurs in the context of, or as an abbreviation for, the table displays otherwise.
 146 The Aristotelian Ethics The Use o f Technical Terms in the Ethical Treatises 147
prohairetos and the noun prohairesis, is preferred in the E E . T h e
_rt + ' iL 00 ^Μ « N^H θ' ηN θ' ΐΛ ΙΛ ΐΛ N t^· verb bouleuö with its cognates, closely connected in sense with
O |3
0
H H N T )- N θ' μ l"-
Ö Ö« Ö O (S O O O prohaireomai, is also commoner in the E E than in the N E . F o r
N Oo g-fi •à 3 ^ 3 >S >2- Ä -S-
h 5 l -8 Ο O l O OO « N <N OO ‘ want’, as a general verb o f appetition, ephiesthai is common in the
N E , oregomai in the E E . (The commonest verb for ‘want’, boulomai,
1 g
O t>
w 'Vif Su is slightly more frequent in the E E — the proportions are N E 0 24,
w O
δ ö O Th Γ>» *Λ r«-5 ΙΛ N r< N fO N A E 0 08, E E 0 27 per cent for the word and its cognates— but the
o o N E and E E frequencies are too close to make it worth including in
w the table.) T h e words thymos and epithymia, with the verb
< epithymein, words for the two lower parts o f the Platonic tripartite
4. -
ui O' y VW CS O n f^ N O Ο ΙΛ i* OO N HH θ ' O rt- -φ OO
soul and their functions o f anger and sensual desire, are
Z νζ? O.
Q-*°
*0 considerably commoner in the E E than in the N E . T w o verbs for
Kn
vO O U“> N Ο O f- taking pleasure, hëdesthai and chairein, are popular in the N E . Once
«!* OO 4 >—I >-H OO ^ O O O *-<

•Î again there is considerable fluctuation due to subject m atter: in

■S3'Wv0~î both the N E and the E E the majority o f the occurrences o f the
=*- §· O f^ O O sO Eudem ian favourites occur in the books in which prohairesis is
being explicitly discussed; but again differences due to subject
-E -S
matter are combined with differences due to vocabulary pre­
e 3 -
'= § ference : both within the prohairesis book itself, and in the other
Ο Γ-* Ο O U“> N
books taken separately, the Eudem ian frequency o f the Eudem ian
°1
°^c§. favourites is much higher. T aking the groups as a whole and the
works as a whole, it will be seen that as usual the A E frequencies are
o ' '0' '*3- 'rT
— -)- J. Co 'Z '
sp <N
m 3 3 "
u*> Γ » O '
Tf-
m
OO
O ? ■"t· N N
^ O O ?
1-1 N O N close to the E E ones and distant from the N E ones. F o r the
« w 0 0 s
e ^ r ° -I- o C
6 Ö Ö Ö Ö Ö Ô Ö 6^ Ö Ö
'— '
Ö Ö
'■—*'
Ö
' '— '
O
'— ' Nicomachean group the E E frequency is 0 22 per cent and the A E
^ i g .2 OO 0 0 N O ΙΛ U-, \θ ir·, OO OO O w N 00
fO t J- s©
N
N m 1-4 m
frequency is 0-21 per cent: in each case less than half the N E
$ frequency o f 0 54 per cent. For the Eudemian group the E E
3
i~3· N vn 00 N Ό 10 O' Γ"* O O - IT) r>- O ' 0 r»
N
frequency is 0-82 per cent and the A E frequency is 0 7 7 per cent : in
Ü ΙΛ
X each case much higher than the N E frequency o f 0 45 per cent.
Ö One psychological term which is much more frequent in the E E
Ο Ο ~ Ο O -
^COo than in the N E is the word ‘soul’ itself : psyche. T h e figures here are
î ftP“ N E thirty-one (0-08 per cent), E E fifty-five (0 2 1 per cent). In
o d respect o f this word the A E resembles the N E rather than the E E ,
UJ N ^VU with fourteen occurrences (0 08 per cent).
J
QQ
<
T E C H N IC A L T E R M S O F E T H IC S
H
ΙΛ N Several o f the technical terms o f Aristotle’s ethics, such as the
names o f particular virtues and vices, are useless for our inquiry
since their occurrences are virtually confined to single locations in
O tq -h (S η 'ψΟΟ 0- 0-2< eo u the treatises. T h ere are, however, a number of terms which occur in
O tti a § 1-1 « Μ I—H Λ
CO ^ o a variety o f contexts in whose use we can detect Eudem ian and
t- ^ e2 > > h
148 The Aristotelian Ethics The Use o f Technical Terms in the Ethical Treatises 149
Nicomachean preferences. Seven such Eudem ian favourites are H ere again we see a combination o f the effects o f subject matter and
listed as follows: o f vocabulary preference: but even so, by our test the differences
between A E and E E are insignificant, those between A E and
NE AE EE
N E significant.
Occur- Propor- Occur- Propor- Occur- Propor- Another group o f Nicomachean-favoured ethical terms are terms
rences tion % rences tion % rences tion % for excess and defect, prominent in discussions o f the theory o f the
mean:
α ίτιο ς
(inc.
NE AE EE
αιτία ) 35 ΟΟ9 8 0-05 63 0-24
βία,
Occur­ Propor- Occur-■ Propor­ Occur­ Propor­
β ία ιον 12 0-03 3 0Ό2 47 ο·ι8
rences tion % rences tion % rences tion %
έξις 59 0-15 57 ο·33 64 0-24
έκώ ν 11 003 30 ο·ι8 25 0*10
ελ λ είπ ω ,—φις 74 o-iç 12 0-07 28 ou
εκούσιον 42 O il 24 ο ·ΐ4 45 ο ·ΐ7
ύπερβαλλω ,—
άκώ ν II 003 8 ο·ο5 II 004 βολή 97 0-25 34 0-20 38 0· 14
ακούσιον 21 0-05 II ο·ο6 23 0-09
υττερεχω,—οχη 29 0-07 8 0-05 21 008
Total Ι9Ι 0-48 141 ο·83 278 ι·ο6
Total 200 0-51 54 03I 87 0-33
It will be seen that taken together the words o f this group occur Though terms for excess in general are more frequent in the N E ,
twice in the E E for every once in the N E . T h e rate in the A E , for the the E E is fonder than the N E is o f expressing excess by hyperochê
words as a group, is between the two and closer to the E E ; but for rather than hyperbole. T h e differences between the N E and the A E
certain words taken individually, such as aitios and hia, there are for for elleipö and elleipsis and for the group as a whole are significant ;
once striking differences between the A E and the E E . none o f the other differences is. T w o verbs o f obligation, deô and
Besides these Eudem ian favourites, there are some ethical opheilö, make a final group o f Nicomachean ethical favourites: deö
concepts whose expressions are more frequent in the Nicomachean (N E 274, A E 75, E E 128) has a D .R . o f 1-43 and opheilô(N E 2 1 , A E
Ethics. T h e notion o f a life, or way o f life, expressed by the noun bios o, E E 4) has a D .R . o f 3-5.
and the verb bioö, is much more Nicom achean than Eudem ian; and
the notion o f happiness, though obviously pervasive in both ethics,
E V A L U A T IV E E X P R E S S IO N S
finds expression in the adjectives eudaimôn and makarios with its
cognates, and in the noun eudaimonia, more than twice as often in Besides words which have a more or less technical role in ethics,
the N E than in the E E . F o r the three treatises we have: the ethical treatises are naturally rich also in common evaluative
expressions, words which with varying nuances are more or less
NE AE EE
equivalent to the word ‘good’ with its adverb, comparatives, and
Occur- Propor- Occur- Propor- Occur- Propor- superlatives. It is interesting to study the use o f the common word
rences tion % rences tion % rences tion % for ‘good’, agathos, and its adverb ‘ well’, eu, and its various
βίο ς etc. 10 comparative and superlative forms, kreittön and kratistos, ameinön
75 019 5 0-03 0-04
ευδα ίμ ω ν etc. 100 ο·25 ο·ο8 38 ο ·ΐ4 and aristos, beltiön and beltistos', with it one can study the words
Η
μα κά ριος etc. 37 0Ό9 I 0*01 4 0-02 iepieikes\ ‘halos', ‘ spoudaios', which have technical meanings in
particular contexts but can also be used as general pro-words. T h e
Total 212 ο·54 20 0 12 52 0-20
results o f studying in each book the use o f these words— along with
 I

The Aristotelian Ethics The Use o f Technical Terms in the Ethical Treatises 151

N O O^toc w oo lO or-i O u"> t three words for instrumental goodbess, or utility, sympheron,
tJ- io r-*
^ i 3, ό ό ö ïïphelimon, chrësimon— are shown in T able 6.6. Once again clear
l>HMir5N Nicomachean and Eudemian preferences em erge; but for once the
O 'C S N NMo i-ifO
iO mO o IH t
Xn O' t w oo
If) N t M Lr-*
T)
A E does not completely share the preferences o f the E E . It
continues to resemble the E E more than the N E , the coefficient o f
correlation between the E E and A E usage being 0-8185 as against
Evaluative Expressions in Individual Books of the NE, AE, and EE

TA B L E 6.7 Scattergram o f Occurrences o f Evaluative Expressions in the EE

n t o >- t n o α> ο ο ΙΛ η ο οο and AE
68000
αγαθός

32 ΓΟ CO U"> HH t m n n ro o

/
/
O'' »Λ 0Ο fO N fO O O' /
O -Ι M l- H Q O t < 0 \© /
N 54-400 / ............
/
/
-3 w+ i.S Co o ' θ ' o ' Co' ro t 1> ^ 'το οο O
iO /
O ' θ ' γ-~> O ' ΟΟ GO Μ ο Ο ΟΟ ΙΑ)
ό /
ό Μ Μ μ
Hgb fii 'w ' 3 , 3 - 3 , 3 , ■
— ■'— ' 3 ,3 , /
ο ΟΟ 00 |> Μ t " Μ οο to φ ΟΟ LT1 1-1 ro O ' ι ο /
ι/Ί co -<*■ LD ΙΛ θ ' Γ" t CO Ν σ> ro Ν Ν ro r o /
t
/
/
/
ό
5* 3* Μ CO μ to o ο Μ t Ό u-i Ο IM Μ Μ Γ­ Μ4 40-800
/
t
rt /
/
/
/
00 * Λ T h -Η Ν ΟΟ Μ ι/ι ιλ N N /
hi /
ε η ιεικ ή ς /
/
27-200 /
/
t O N Ό /
σπουδαίο'; /
X /
/
6.6

j /β ελ τΐω ν
r*><*< *■< t U") θ ' Ο t Ο CO n r- m o
Table

Ö ro ro Π η f#; f ) |s
’ καλός
X
/
13-600 /
N N O t σύμφερόν εν

* /
/
/
<N LO N O' *-i O ro 10 O O' ' *.
κρεά τω ν
χρ ή σ ιμ ο ί
X
ta - « ro tco OO 15 m < ffl U ·« k )"S S S 5 S 0-000
S; o ^ ë ^ « > « S 11 23 38 48 56 73
H ES > H R - 0-8185 E E occurrences

I
152 The Aristotelian Ethics The Use o f Technical Terms in the Ethical Treatises 153
T able 6. 8 Scattergram o f Occurrences o f Evaluative Expressions in the groups o f expressions) while the A E and the E E are very close
NE and AE together (0 58 per cent each)— and this despite the fluctuation in
68 000
χγχθόί the frequency o f epieikès due to the fact that while normally it is
something o f a synonym for agathos it is used in the first common
book, and there alone, in a special technical sense. B u t with the
Eudemian favourites the pattern is unusual : the words are much
more frequent in both the N E and E E than in the A E , and the A E is
54-400 closer to the N E than to the E E .
/
/
/
/ GLO RY AND THE FA M ILY
/
/
*
/ I f we search through word lists o f the N E and E E , looking for
/ words with a high distinctiveness ratio, two rather surprising and
/
40-800 / interesting groups o f Nicomachean favourites emerge. We discover
/ that the Nicomachean Ethics has a consistently higher score first in
/
/ words which concern glory, honour, praise, and their converses
/
f humiliation, shame, and blame, and secondly in words for family
/
/ relationships and for youth and age. Aischros, ‘shameful’, with the
/
/ compounds formed from it, is a Nicomachean favourite, as is
χ επ ιεικ ή ς
/
27-200 / geloion, ‘ridiculous’ , and the verb psegö, ‘blame’, with its adjective
/ psektos. So, on the other hand is the verb epainein, ‘to praise’ , with
/
its adjective epainetos, and the noun time, ‘honour’, with its cognate
> σ π ο υ δ χ ΐο ς /
/ verb timân and adjective timios. T h e N E is also fond o f the adjective
/
β ε λ τ ίω ν ^ / megas, ‘ great’ , with its comparatives meizôn and megistos and of
/
/ χ κχλότ compounds formed from the root mega l such as megalopsychos and
/ εν
13-600 megaloprepës. T h e figures for the treatises are:
συμ φ έρο ν^/
/ à μεινω ν
/ NE AE EE
/
/ X ,
/ κρ α χτώ ν Occur­ Propor­ Occur­ Propor­ Occur­ Propor­
rences tion % rences tion % rences tion %
0000 004 003
αισχρός etc. 53 0 13 7 9
Λ = 0-8066 S E occurrences γελοίος 17 004 0 ο·οο 5 0Ό2
ènaivéai etc. 64 0 16 7 0-04 31 0-12
ο·8ο66 between the N E and A E , but it does not correlate very
μ ίγ α ς ,
highly with either. T h e result o f plotting the A E uses against each μ ΐγ α λ - 039 6 0-04 80 0·30
156
o f the other treatises is shown in T ables 6.7 and 6.8. τ ιμ ή etc. 92 023 II ο·ο6 19 ΟΌ7
In this group o f expressions it will be seen that the Nicomachean φ ίγ ω etc. 24 006 8 005 12 0-05
favourites follow the pattern with which we are now fam iliar: the
Total 406 10 3 39 0-23 165 0-63
N E stands on its own (with a proportion o f 1 05 per cent for the six
I 54 The Aristotelian Ethics The Use o f Technical Terms in the Ethical Treatises 155
H ere the E E frequency for the group is just under two-thirds that

o ccu r­
ren ces
ro Ό u-> NO N rs
for the N E ; the A E is significantly less than either the N E or the in NO O
în rt- U-) N NÔ O' ro
EE. m

O O u*> « tr> O
I f we take a group o f words referring to youth vs. age and family n «ri
relationships— adelphos, ‘brother’ , goneis ‘ parents’, neos ‘youth’, vO O O ΓΟ N
CH
LTi
N O "*·
pais and its cognates, ‘ child’ and ‘ childish’ , and patêr and patrikos, U~) O N iri es
ci
‘father’ and ‘paternal’, uios, ‘ son’, teknon, ‘child’, and mëtêr, 0
3
' - HH O nO O ►H
‘mother’, we again find that the group as a whole is substantially co

more frequent in the N E than in either o f the other two treatises. m N OO OO O' N

T h e group occurs 129 times in the N E (0-33 per cent), 28 times in

s
Ή. N 0 O N N 00 O
^1*
00
HH rs
the A E (ο ·ι6 per cent), and 53 times in the E E (0-20 per cent). E NO N rr» to 10 Th
« N
O m ro O ^}-
rn
TESTIN G GROUPS OF IN D ICATO R WORDS AGAINST SM ALL O' O ro ro NH
■'t’
SAM PLES
OO co n 00 N Ό «N
N
In respect o f the words studied in the present chapter, as in the r-« 00 ■Ί" N N sD nO 00
N
earlier ones, the A E as a whole resembles the E E much more than
s© Tf N N
the N E ; though the considerable book-to-book fluctuation within N t—1
LO m 0 IT)
the E E and the N E makes the results more difficult to interpret than O O N
es
in chapters 4 and 5 above. We may now try to apply the findings o f Tf- to On τί- co 00 NH
Tt"
the present chapter, as we applied the findings o f chapters 4 and 5, ro O' O' N co O CO Ό
N
to small samples o f the A E to see whether the material we have been
N in O ht 0 O
studying allows us to detect any sign o f piecemeal construction or fS

stratification within the A E . 00 N 00 IT)

ÎS
T h e first step, as before, is to attempt to construct groups o f Λ fi m N 00
υ nO O'

OOOI
words with a sufficiently high frequency and a sufficiently marked n N Tt" 9
S e
On Ο fn 0 0 Ö in
distinctiveness ratio. I f we consider first Nicomachean groups, we X 'fi HH
ω
find that the Nicomachean favourite metaphysical terms provide a
group o f 523 words with a D .R . o f 2 74, and the Nicomachean û μ
N
°9 NO ΙΟ
Γ*"·
sO

favourite ethical terms (happiness-words plus excess-and-defect ά n N 1-4 hH HH

m N 00

%
NO
words) a group o f 4 12 words with a D .R . o f 2 1 3 . T h e Nicomachean n l-H O
0 0
10 Tj- î vO 00
evaluative words occur 4 18 times in the N E and have a
IC
distinctiveness ratio o f 1 -82 : they will form a third suitable group. A o is H 1? ~ *s? -S -Si •S s o C
a <υ .^ « ^ c a "'s "δ 5
fourth group can be made up o f the ‘glory’ words (406 occurrences, 3 SJ 6 ~ V. . -«ï
a -2 1 1 S, <u ^
with a D .R . o f 1 6 4 ) ; to make a fifth group we can take the ‘fam ily’ « ^ ü ■
S'a .8 „ ® '«■^
«i, » 5δ
c- ■a s u ü 4çj-> -S 1^ 8 β ? -«
fc *I .—
s r
words (145) and the ‘obligation’ words deo and opheilo (295 <U Ü "<Ss *- •<* jr e
u V" ur «,.$<
Si -S
2 <
£ R rf -S ^O ea Sa“ e . -5
. ~ C S
-s
occurrences) to make a group o f 440 with a D .R . o f 1 5 3 . T h e Ο & H '—■-ft <4. o — r Ê i - s <3 «
ÎS u u ^ a !§ w 3 »»Ci a1) 3. «S *
Nicomachean psychological words (2 15 occurrences) are not HN S « G ω «
C s? ίτ" ,s ^ -u 1-11 u V Ss χ __ - I*- ? ^
sufficiently numerous to make a group : we can, however, add them
*». a» g ss
•S J- l* ^ "S -«S o
to some other assorted Nicomachean favourites to make a final, s lu 1 1
H
 The Use o f Technical Terms in the Ethical Treatises 157
rag-bag group. Into this bag we can throw the two Nicomachean
verbs o f saying, ero and eipon (139 occurrences), and four other
words which the N E specially favours : (a) diapherô and diaphora,
N E 92, A E 20, E E 30 ; (b) homoiös, N E 94, A E 29, E E 50; (c)
chronos, N E 40, A E 5, E E 12 ; (d) radios, N E 3 1 , A E 5, E E 8. T h is
will give us a total group with 6 1 1 occurrences and a D .R . o f 16 7 .
S ix Eudem ian groups can be constructed to match these six
Nicomachean groups. T h e first is provided by the Eudem ian verbs
o f saying, lego and phemi, plus the Eudemian verbs o f thinking,
oiomai and axioö: these give a group with 27 1 occurrences and a
D .R . of 0-68. F rom the logical and quasi-logical words we can select
the occurrences o f logos, logismos, and its verb, enantion, aporeo and
its cognates, diaireô, and orthos : this will make up a group with a
total o f 279 occurrences in the E E , and a D .R . o f 0 5 1. T h e
Eudem ian metaphysical terms give a group with an occurrence of
267 and a D .R . o f 0 51 ; if we add psyche to the volitional terms we
get a group o f psychological terms occurring 270 times with a D .R .
o f 0 64. T h e Eudem ian ethical terms occur 278 times, and make a
group with a D .R . o f 0-45; the evaluative ones 357 times with a D .R .
o f 0-75.
Arm ed with these twelve groups, we can set out to record their
occurrences in each o f the seventeen samples o f the A E with which
we are familiar from chapters 4 and 5. T h e outcome o f this exercise
can be seen in Tables 6.9 and 6.10. It will be seen that once again the
overwhelming impression is o f a greater similarity with the E E than
with the N E . In none o f the samples does the actual occurrence of
the Nicomachean indicator words reach more than 72 per cent o f
the Nicomachean expectation; in most o f them the actual
occurrence is less than half the expectation, and the mean is only
0 47 o f the expectation. In all but two o f the samples, on the other
hand, the actual occurrence o f the Eudem ian favourites is above 58
per cent o f the Eudemian expectation and in four cases the
Eudemian expectation is exceeded. T h e mean o f the actual
occurrences is 0 8 3 o f the expectation. However, in the case o f these
groups V I I - X I I , unlike the groups studied earlier, the influence o f
subject matter occasionally triumphs over stylistic preference.
W hile in fifteen out o f the seventeen cases the Eudem ian
expectation is approximated by the sample better than the
Nicomachean expectation, in two cases the Nicomachean expec­
tation is better approximated. In each case, the reason is not far to

158 The Aristotelian Ethics
seek: in the case o f sample 3 (which reaches 56 per cent o f the N -
expectation, but only 34 per cent o f the Eudemian) the high
Nicomachean score is due to the twenty occurrences o f the verb deô
in the discussion o f corrective justice and reciprocity. In the case o f
sample 6, where the actual occurrences reach 62 per cent o f the N -
expectation but only 58 per cent o f the ^-expectation, the
explanation is to be found in the technical use o f epieikês in the
discussion o f equity, which boosts the score for TV-group-IX to
twenty-two. T h is latter anomaly in fact is the result o f the greater
similarity o f the A E to the E E : in the A E and the E E the use o f
epieikês as a synonym for ‘ good’ is rare, because o f the reservation o f
the term for the technical use in A E A. T h at is why epieikês appears
as an TV-favourite and yet is common in sample 6, where the
technical sense is discussed. T h us, the two anomalies reveal the
limits o f the comparatively crude vocabulary tests applied rather
than any genuine Nicomachean element in the samples o f the
disputed books. Allow ing for discrepancies due to particular
subject matters (the high score o f A £ -g ro u p -X II in samples 16 and
17 , for instance, is due to pleasure being there the topic o f
discussion) the over-all picture remains as before, o f general
similarity between the A E and the E E . There is no support for the
patchwork theory: the two sections which approximate the N E
expectation better than the E E expectation are not places which
scholars on other grounds have regarded as particularly
Nicomachean, and in each case if the scores o f the groups I - V I (see
above, pp. 13 2 -3 ) are added to the scores o f groups V I I - X I I the
sections emerge as over all closer to the E E expectation.
We may conclude our stylometric examination, therefore, with
the verdict that the overwhelming weight o f the evidence suggests
that the common books are more at home in the context o f the
Eudemian Ethics than in the context o f the Nicomachean Ethics. Our
twelve groups o f Nicomachean favourites and our twelve groups o f
Eudem ian favourites have provided twenty-four independent tests
to apply to the disputed books. Tw enty-three out o f these twenty-
four independent tests give an unambiguous answer that the
common books, considered as a whole, resemble the E E more than
the N E } W hen we take the twenty-four tests together and apply
I . O nly the twelfth o f the Eudemian groups suggests otherwise. H ere the actual
occurrence in the A E reaches only 43 per cent o f the Eudem ian expectation. T h is failure can
only partly be explained through considerations o f subject matter.
I
The Use o f Technical Terms in the Ethical Treatises 159
them as a group to each o f the seventeen iooo-word samples o f the
A E , they indicate in the case o f each individual sample that it is
closer to the E E than to the N E . N one o f our tests gives any support
to the theory that the A E is a patchwork o f mixed Eudem ian and
Nicomachean material. T h e conclusions o f these test-groups are
confirmed by those o f other vocabulary tests (such as the study o f
the singular and plural forms o f the definite article) which are
independent o f the vocabulary elements which contributed to the
groups constituting the twenty-four tests.
Since our tests have covered some 60 per cent o f the total
vocabulary o f the ethical treatises, it is inconceivable that their
results should be overturned by further study o f the vocabulary.
Since they have been restricted to vocabulary, they leave open the
possibility that further information about the relationship o f the
A E to the E E and the N E may be obtained by the study o f syntax:
but it is in the highest degree improbable that the over-all picture,
o f a close resemblance between the A E and the E E , and a much
more distant relationship between the A E and the N E , should be
seriously disturbed by further stylistic investigation, however
sophisticated.1 T h e most plausible and economical explanation o f
I. One stylistic test was carried out which does not depend on vocabulary counting: a test
o f the syntactic category o f last words o f sentences. M ichaelson and M orton have suggested
(1970 , 192 ff.) that a study o f the last words o f sentences (defined as portions o f text
terminating with a period, colon, or question-mark) enables samples from different authors
to be discriminated. (For a criticism o f this view, see Johnson, 1973, 92 ff.) For this test,
sentences in Aristotle were divided into three classes : those ending in a noun, those ending in
a verb, and those ending in some other part o f speech. Each book o f each o f the ethical
treatises was then run o ff against each other to see whether the differences between them, as a
whole and pairwise, were statistically significant, the chi-squared test being used to
determine statistical significance. It turns out that Aristotle differs from other G reek authors
who have been studied in this manner in that an unusually high proportion o f his sentences
end in parts o f speech other than nouns or verbs. T h e application o f the test to the ethical
writings give the following results : D ifferences between individual books o f the E E are non­
significant (12 57 for 8 degrees o f freedom). D ifferences between the books o f the A E are
non-significant at the i-per-cent level (chi-squared 11-82 for 4 degrees). T h e differences
between the A E taken as a whole and the E E taken as a whole are non-significant (0 73 for 2
degrees). T h e differences between the A E and the N E however are statistically highly
significant: chi-squared is 30-25 for 2 degrees o f freedom. It is difficult to know how to
interpret these results, however, for it turns out that in respect o f this feature the N E itself is
highly inconsistent: taking the seven books o f it together chi-squared for 12 degrees o f
freedom is 72-28. T h is inconsistency is caused by books 4 and 9 : if these two books are
removed the rem ainder o f the N E is internally consistent and differs only in a statistically
non-significant way from the A E and the E E . I know o f no other reason for regarding books 4
and g as standing apart from the rest o f the N E . Both books differ in this respect markedly
from a test-block o f 6687 sentences drawn from a number o f Aristotelian works in various
fields. F o r an account o f the application o f the last-word test to the entire Aristotelian corpus,
see K en ny, 1977.
 The Aristotelian Ethics
the evidence assembled in the last three chapters is that the
common books o f the A E , just as they now stand, belonged
originally to the Eudemian Ethics. H aving established this high
probability on the grounds o f statistical stylistics as well as of CH APTER 7
external historical evidence, we must now turn to the evidence
provided on the topic by the philosophical content o f the treatises. Wisdom in the
Aristotelian Ethics

In the half-century since the publication o f W erner Jaeger’s

Aristotle it has been almost universally accepted by scholars that the
Eudemian Ethics represents an early stage and the Nicomachean
Ethics a late stage in the development o f Aristotle’s ethical theory.
Jaeger’s theory o f development was based in large part on a change
which he claimed to detect between the two ethics in Aristotle’s use
o f the term phronësis. In the E E phronësis was used, he claimed, in
the Platonic sense o f a speculative, philosophical wisdom, uniting
ethics and metaphysics; in the N E the term refers to practical
wisdom, and is sharply and explicitly contrasted with sophia or
philosophical wisdom. T h e difference between the two is explained
by the development in Aristotle’s thought from a Platonizing
idealist ethic to a more empirical and pragmatic moral theory.
T h e contrast which Jaeger drew between phronësis in the E E and
in the N E was in reality a contrast between phronësis in the E E and
in the A E : his account of the nature o f phronësis in the N E rested
almost entirely on the description o f that virtue in the disputed
book B. T h e theory o f a development, so far as concerns the
phronësis argument, rests heavily on the assumption that the
disputed books are Nicomachean: an assumption which Jaeger
him self did almost nothing to defend, mentioning it only in an
inadequate afterthought footnote. G iven that we have shown that
there are strong stylistic reasons for regarding the disputed books,
as they stand, as belonging with the Eudemian Ethics, it would be
possible to argue with as much justification as Jaeger that this
Ethics, containing as it does the developed theory o f phronësis,
must be the work o f the later, mature Aristotle. Instead, we shall
proceed more cautiously, and study separately the phronësis-theory
o f each o f the three treatises : we shall compare that o f the N E and
that o f the E E in turn with that o f the A E , and ask what contrasts
162 The Aristotelian Ethics
can be drawn between the N E theory and the E E theory, and which
o f the two fits better with that o f the E E . We shall not, in fact,
attempt to draw from this comparison any positive conclusion
about the relative chronology o f the treatises: for rightly under­
stood the data do not permit the drawing o f any such conclusions.
Jaeger’s contrast between the sense o f phronësis in the E E and N E
has been severely criticized by scholars a number o f times— notably
by M argueritte, Gadam er, and Leonard. These criticisms have
gained widespread acceptance, especially since being canonized in
the commentary o f Gauthier and Jo lif.1 None the less, it is worth
while to look afresh at the concept o f phronësis in the ethics, since
even those who have discarded Ja eger’s view on this point
frequently continue to accept his general theory o f the development
o f ethical theory in Aristotle, and his assumption that the disputed
book B is Nicomachean. T h us Row e, in his monograph The
Eudemian and Nicomachean Ethics, while admitting ‘ the kernel o f
Jaeger’s discussion o f the ethics, his account o f the history o f the
term phronësis. . . has been shown conclusively to rest on a
misinterpretation o f the texts’ , goes on to argue at length that
there is no sharp distinction between ethics and the theoretical sciences in
E E . . . the practical and theoretical tend to merge into one another ; and
this is reflected in the close relation envisaged between speculative and
practical thinking. But in N E, the distinction between the two spheres is
complete. Ethics and the theoretical sciences no longer have anything in
common, since their subject matters are now established as being totally
different in kind. Correspondingly, the rational faculty is now divided into
two; the phronësis of E E becomes the areie of one half, and sophia is
appropriated for that of the other.
He goes on to conclude ‘ in principle, Jaeger’s view o f Aristotle’s
development was right’ (op. cit. 1 1 , 70-3).
L ik e Jaeger, Row e regards the disputed book B as Nicom achean;
he regards indeed all the disputed books as being Nicomachean as
they stand: books A and C , he thinks, are based on and supersede
lost Eudem ian parallels, but B , he thinks, was cut from whole cloth
for the N E . As with Jaeger, so with Row e, there seems a certain
i. ‘ Une meilleure intelligence de la “phronësis” de l’Ethique à Nicomaque . . . jointe à une
meilleure compréhension de l’évolution doctrinale d’Aristote et à une connaissance plus exacte
de la chronologie de son oeuvre . . . achève ainsi de ruiner l’hypothèse de Jaeger sur
l’évolution du concept de phronësis entre PEthique a Eudème et l’Éthique à N icomaque : ici et
la, c’est la même sagesse, pratique mais normative, que désigne le mot de phronësis’ ( 1959, ii,
469).
Wisdom in the Aristotelian Ethics 163
circularity in the argument for the Nicomachean context. F irst, a
contrast between the N E and the E E is built up on the assumption
that B belongs with the N E ; then we are told that the original home
o f B must have been the N E because it fits the doctrine o f the N E
better than that o f the E E .
The whole of V I is built up on the distinction between phronësis, the aretë
of the deliberative faculty, and sophia, the aretë of the contemplative; and
since E E actually merges deliberative and contemplative thought, VI
cannot have been intended for it, unless we suppose that Aristotle took up
two opposing positions in the same work (p. 109).
Supposing we grant, for the moment, that the E E does not
distinguish between deliberative and contemplative thought with
the clarity that B does : even, so, before concluding that B belongs
with the N E , ought not we to inquire whether the N E , outside the
disputed books, makes the distinction made in B ? Otherwise, may
not the situation be that neither the N E nor the E E contains the
fully worked out theory o f phronësis ; and for the simple reason that
in whichever o f them B originally belonged, B was the place chosen
for the full working out o f the concept? T h e only sound procedure
is to study the theory o f A E in its own right and compare with N E
and E E in turn. T h is we shall now do.
T h e doctrine o f the A E , and in particular o f book B , concerning
phronësis can be summed up under four heads, (i) Phronësis is an
intellectual virtue concerned with the truth about mutable matters
and the whole good o f man. (2) Phronësis is the virtue o f a particular
part of the rational soul, and is distinguished from other intellectual
virtues by being deliberative rather than intuitive and practical
rather than theoretical. (3) Phronësis is indissolubly wedded to
moral virtue, providing the right reasoning necessary for the
exercise o f virtue, and dependent on virtue for the correctness o f its
own starting-points. (4) T h e union o f phronësis and moral virtue is
dependent on the pre-existence o f certain natural qualities,
intellectual and affective. L et us develop each o f these points in
turn, and compare the A E doctrine with that o f its competing
contexts.
(1) Phronësis as an intellectual virtue. ‘ In classifying the
excellences o f the m ind,’ we are told at the beginning o f book B , ‘ we
said that some were o f the moral character, and others o f the
intellect. Moral excellences we have discussed at length ; let us now
 The Aristotelian Ethics
discuss the others’ ( i i 3 9 ai - 2 , trans. Greenwood). T h e nature o f
the aretê o f anything depends on the ergon o f that thing: the ergon of
every intellectual part is the production o f true and false
judgements (i 139*29). That, at least, is its ergon in the sense o f its
characteristic activity, its ergon whether it is working well or ill ; its
activity when it is working well, its good activity, and therefore its
ergon in another sense, is truth alone ( i i 3 9 bi2). T h e intellectual
virtues are then excellences which make an intellectual part o f the
soul come out with truth. There are five states o f mind which have
this effect— technë, epistëmë, phronësis, sophia, nous, which we may
translate as skill, knowledge, wisdom, learning, and
understanding— and so five candidates for being intellectual
virtues. Being truth-productive is a necessary but not a sufficient
condition for being an intellectual arete : skill fails to pass some o f
the other tests for being an aretê ( ii 4 0 b2 2 -4 ); the other four
qualities remain as virtues, though not all independent o f each other
( i i 4 i a2-8 , b3~ 5 ; 114 2 * 2 6 -3 1). Phronësis is defined as a ratiocina­
tive, truth-attaining quality concerning what is good for human
beings ( ii4 0 b5, b2 i) . T h e fact that it is ratiocinative, and has the
particular subject matter it has, distinguishes it, as we shall see,
from the other three intellectual virtues. T h us far the A E .
In the N E , too, we find that phronësis is an intellectual virtue : it is
listed as a paradigm o f such at 1 i0 3 a6 when the distinction between
moral and intellectual virtues is introduced. T h e relation between
having an aretê and doing one’s ergon well is also familiar in the N E
( i098a8 ff. ; 1 10 6 *17 ff.) ; but we are not told the particular ergon o f
the intellectual part o f the soul. We are, however, told implicitly
that phronësis is ratiocinative— is concerned with a logos— since
moral virtue is defined as consisting in the observation o f a mean
‘ which is defined by a logos such as a wise man would use to
determine it’ (110 7*2). Since we learn from here, and from the final
book (117 8 * 10 ), that it is wisdom which sets the standard for the
moral virtues, we might well conclude that the subject matter o f
wisdom is the whole good o f man ; but this is not spelt out in the N E
as it is in the A E , though we are o f course told, right from the
beginning, that there is a discipline called politikë which has human
good as its subject matter (io g4a27). Bu t the relationship between
this and phronësis is left obscure.
L e t us now turn from the N E to the E E . Here we find that
phronësis is not given as an example o f the intellectual virtues when
Wisdom in the Aristotelian Ethics
these are first introduced (at 1220 *4 -14 ) and indeed it appears at
12 2 1* 12 in what looks like a list o f moral virtues. But in every
other respect the doctrine o f the A E is to be found more clearly in
the E E than in the N E . From the last book o f the E E it is clear that
phronësis is an intellectual virtue: it is a virtue (i24 0 b35) which is in
the logistikon or ratiocinative part o f the soul ( 1 246b23) ; it is in the
ruling part o f the soul (i240bi2) and it gives commands (epitattei,
I249bi5). All this accords with the description o f intellectual
virtues as belonging to the rational part which qua rational is in
command o f the soul (tou logon echontos ho epitaktikon esti tës psychés
hë logon echei, 1220*9). As in the A E and the N E , an aretê makes the
ergon o f its possessor good : as in the A E , the ergon in question is the
production o f truth. We are told at i 2 2 i b29 that there are aretai
corresponding to different parts o f the soul: the aretai o f the
rational part are the intellectual virtues ‘ whose ergon is truth, either
about how things are, or about bringing into existence’ . 1 T he
distinction between these two kinds o f truth fits perfectly the
distinction between the subject matter o f phronësis and sophia in
book B : sophia, we are there told, unlike phronësis, is not concerned
with bringing into existence (oudemias esti geneseös, i i 4 3 b2o).
Phronësis is ratiocinative in the E E : ‘ It is clear that [lucky people] do
not succeed by means o f wisdom, because wisdom is not
unreasoning but can give reason why it acts as it does’ (1247*14).
Finally we are told in the first book that the good which is the end o f
human action is the subject matter o f the supreme discipline o f
politikëkai oikonomikëk a iphronësis. These three hexeis, we are told,
are alike in this respect: how they differ will be explained later
( i 2 i8 bi5). T h e forward reference well fits a passage in B,
i i 4 i b24 ff. There we are told that politikë and phronësis are the same
hexis with a different einai : different applications, we might say, o f
the same mental quality. Wisdom applied to one’s own affairs is
phronësis in the narrow sense ; wisdom applied to the affairs o f the
polis or to the affairs o f one’s household is phronësis in a broad sense.
Politikë, too, has broader and narrower senses : in the broadest sense
it includes the wisdom o f the legislator, o f the executive, and o f the
judiciary ; more especially the word applies to the two latter, and par
excellence politikë is wisdom in executive decision ( i i 4 i b3o).
(2) Phronësis and other intellectual virtues. In the A E one o f the
I. ώ ν € ρ γ ο ν ά λ ή θ ίία , η π ε ρ ί τ ο υ π ώ ς « χ€ΐ, 17 xrept y e v e a c ω ς.
 The Aristotelian Ethics
main functions o f phronësis, one o f the characteristic activities o f the
phronimos, is deliberation, bouleusis: this is said not only in the
endoxon which introduces the principal treatment o f phronësis
( ii4 0 a2 5 ff.) but also in the passage contrasting phronësis with
sophia ( i i 4 i b9 ff.) and in the discussion o f euboulia ( ii4 2 b32).
Wisdom is concerned with human affairs and with matters that can be
objects of deliberation. We regard good deliberation as the pre-eminent
characteristic of the wise man, and no one deliberates about things that
cannot be other than they are, or that are not directed to an end, a good
attainable by action (i i4 ib9—13).
It is because wisdom is concerned with things that can be other than
they are that it is the virtue o f a different part o f the soul from sophia
or learning, which is concerned with unchanging and eternal
matters. Book B began with a distinction between two parts o f the
rational soul, based on the principle that like is known by like: the
logon echon which was previously distinguished from the alogon is
now divided into the logistikon (named from the sense o f logizesthai
in which it is equivalent to bouleuesthai, deliberate) and the
epistëmonikon which is concerned with the eternal truths.
Repeatedly we are told that each o f these parts has its proper aretë :
sophia, apparently, for the epistëmonikon, and phronësis for the
logistikon now also called doxastikon ( 1 i3Qai 6 ; ii4 0 b20; i i 4 3 bi 5 -
17 ; i i4 4 a2~ 3; i i4 5 a4). T h e other intellectual virtues turn out to be
parts o f either sophia or phronësis, and to be thus located in one or
other o f the soul-parts mentioned : there is no need to go beyond a
bipartition o f the rational soul. Th us, sophia consists o f nous (of one
kind) plus epistêmë ( i i 4 i b3 -4 ): these two, then, will be, appro­
priately enough, in the epistëmonikon. Similarly, gnome', synesis, and
nous (of another kind) share the same subject matter as phronësis
(1 i4 3 a25~b6) and therefore inhere in the same part o f the soul as it
and seem not to be separable from it as a distinct virtue (compare
i i 4 3 a25 with H 4 3bi5 ff.).
T h e psychology o f the A E , and especially o f B , is easy to
reconcile with that o f either the N E or the E E . Both o f these work
regularly with a tripartite soul : a vegetable element, an appetitive
element, and a rational element. T h e vegetable element, whose
ergon is nourishment and growth, is called in the last chapter o f N E
i and in the first chapter of E E II the threptikon ( n o 2 bi 1 ; i 2 i g b22
and 38) ; the name phytikon is also used (N E 1 i0 2a32 ; E E i 2 i 9 b38,
Wisdom in the Aristotelian Ethics 167
accepting V ictorius’ emendation o f the manuscripts). Both treatises
call this an irrational part o f the soul (N E i i0 2 a3 2 : tou alogou to
men ; E E i 2 i g b32 : heteros alogon morion), and both o f them dismiss
it as irrelevant to ethics. T h e irrational part o f the soul that is o f
importance is one which, unlike the threptikon, is under the control
o f reason: in the N E it is described as participating in reason but
distinct from reason, fighting reason in the incontinent man,
obeying reason in the continent man, and harmonizing with reason
in the virtuous man ( n o 2 bi 4 - n o 3 a2); in the E E too it is said to
participate in reason, having by nature the power to obey and listen
and to follow the rational part ( i 2 i9 b2 8 -3 o ; i220 ag - i 1). In the
N E this is introduced as part o f the irrational soul, but we are later
told that you can if you like call it rational, in which case there will
be two rational elements ( 1 i0 3 a2) ; in the E E , on the other hand, it is
introduced as a part o f the rational soul ( i2 i9 b28), but is then later
described as an irrational part (12 2 0 “ 10 ); but it is clear that there is
no real difference here between the doctrines o f the two treatises. In
the N E it is also called the epithymetikon kai ho lös orektikon : the part
o f the soul for desire and appetite ( x i0 2 b30) ; and similarly in the E E
the part which is ruled by reasoning is the part concerned with
orexeis kai pathëmata, appetites and passions (i2 2 0 ai) and it is
called aisthetikon kai orektikon ( i2 i9 b22)— an anticipation o f the
name the scholastics gave it when codifying Aristotelian psy­
chology, the ‘ sensitive appetite’ . It is ‘the part o f the soul that is
capable o f following in accordance with prescriptive reasoning’
(i2 2 0 b5). In each o f the treatises the ethical virtues are virtues o f
this part o f the soul : we gather this in the N E from being told that
the distinction between ethical and intellectual virtue is based on
this distinction o f parts o f the soul ; in the E E we are told explicitly
that an ëthos is a quality o f this part o f the soul (i2 2 0 b20) and the
connection o f this with the notion o f ëthikë aretë via the concept o f
pathos or passion is spelt out at great length (i2 2 0 b8 - i 2 2 i b38). T h e
Aristotelian alogon metechon logou corresponds to the two lower
parts o f the tripartite soul o f Plato’s republic: the epithymetikon and
the thymoeides. A trace o f this Platonic anatomy o f the soul can be
found in the N E 3, at n i 7 b24, where it is said that bravery and
temperance ‘seem to be the virtues o f the irrational parts’ ; this
probably also gives us a clue why this part o f the soul is called the
epithymetikon kai holos orektikon in the N E — the other orexis which
Aristotle has in mind is thymos. In the E E , on the other hand, orexis
ι68 The Aristotelian Ethics
is divided into three species: epithymia, thymos, and boulësis
( i2 2 3 a27). T h e first two are no doubt the hormai, or drives, o f the
irrational part spoken o f at i24 7b9; boulësis, on the other hand,
which involves a judgement that something is good, is no doubt the
same as the hormai apo logismou o f i2 4 7 bi9. H ere too is the germ o f a
later scholastic faculty: the intellectual appetite or will.
N either the N E nor the E E makes a distinction between two
parts o f the rational faculty with the clarity o f the A E , and indeed
both o f them, when distinguishing between the rational and
irrational part o f the soul, describe the rational part in terms
appropriate to the logistikon o f book B rather than to the whole
bipartite logon echon o f that book : the E E says that it participates in
reason by prescribing (tö(i) epitattein, i2 i9 b3o), the N E introduces
it as what ‘ exhorts to the best’ (i i0 2bi6). However, it is not difficult
to insert the divided rational soul o f B into the slot provided by the
rational soul in the framework o f the early books o f the N E and the
E E . In these early books, whichever set o f them is rightly taken as
leading on to B , Aristotle would merely be refraining, wisely, from
introducing a greater number o f technical divisions than he needs at
a given stage o f exposition. Whether the s i s is taken as belonging
with the N E or the E E we have the same didactic procedure as in
Plato’s Republic : a division o f the soul into parts in a provisional
manner, followed by a further division o f the topmost part into sub­
elements geared to different subject matters. (Compare Republic
439 d and 478 a.)
In the early books o f the N E sophia appears from time to time
alongside phronësis ( i098b24 ; 110 3 * 5 ): whether it is used as a quasi­
synonym, or to mark a contrast, and if so on what basis, is not clear.
In the early books o f the E E there are two hints o f the distinction
between logistikon and epistëmonikon made explicitly in B . T h e first
is the passage at i 2 2 i b29 already cited where we are told that the
output o f the intellectual virtues is truth ‘either about how things
are or about bringing into existence’ : in terms o f the psychology o f
B , truth about how things necessarily are would be the output o f the
epistëmonikon, and truth about bringing into existence would be
that o f the logistikon. T h e logistikon also could very well be referred
to in E E i220 b25 by the expression to bouleutikon tës psuchës, which
we are told is the one which studies the final cause or wherefore (hou
eneka) o f such things as walking; for when the logistikon is
introduced in B we are told that logizesthai is equivalent to
Wisdom in the Aristotelian Ethics
bouleuesthai; the logistikon is called also the dianoëtikon kai
praktikon ( 1 1 39a29) and this term in its turn appears in B to be
synonymous with dianoia heneka tou kai praktikë ( i i 3 9 a30). T he
logistikon is in B also called the doxastikon ( ii4 0 b26); in an obiter
dictum in the E E treatment o f friendship also we learn that there is a
special part o f the soul for doxa (i2 3 5 b29).
It is, however, the final books o f the N E and E E that it is most
instructive to compare with book B in respect o f psychological
terminology and theory. In N E 10 happiness is identified with
activity in accordance with sophia (1 i79 a32), and sophia is described
as something which is related to philosophy as knowing is related to
seeking (1 i7 7 a24-6). T h is fits well the sophia o f book B , which is
illustrated by famous philosophers and which has as its subject
matter divine, honourable, and useless things. As in B there is a
contrast between learning and wisdom as between theory and
practice, so in N E 10 we are told that the sophos can act in isolation,
while phronësis appears along with moral virtue as a component o f a
second-rate type o f happiness which is essentially social ( 1 1 77a3 1 ;
ii7 8 ai6 -22). But there is in addition to this resemblance between
A E B and N E 10 a great difference. Whereas in B sophia is the virtue
o f a particular part o f the soul, and there is no suggestion that the
epistëmonikon is transcendent or immortal, in 10 sophia resides not
in the soul but in the nous, which is divine, or the divinest thing in
us; the life o f nous is superhuman and is contrasted with that o f the
syntheton or body-soul compound; the moral virtues and phronësis
are virtues o f this compound, but the aretë o f nous is capable o f
separate existence ( 1 1 7 7 s 14, b2 6 -g ; 78s 14-20). M oreover, it is in
this nous that we must see the essential human being ( 1 i78 a2 ff.) : a
contrast with B , where a human being is defined as essentially a
chooser ( i i 3 9 b5)·
With an effort, it is no doubt possible to regard the nous o f N E 10
as being none other than the epistëmonikon o f A E B ; and certainly
the tension between the conception o f human nature dominant in
10 and that o f B can be found within xo itself (compare i i 7 7 b27
with i i78 a7). T h e fact remains that though B uses nous in a variety
o f senses, it never uses it in a way resembling that o f 10.
Nous in B is often used, like and sometimes with dianoia, as a
general word for the whole human intellectual apparatus, for the
cognitive as opposed to the affective half o f the mind. T h u s in the
second chapter o f B we are told that the three things controlling
 The Aristotelian Ethics
truth and conduct are perception, nous, and orexis, and man is
defined as appetitive nous (113 9 * 17, b5). But commonly nous is a
name for an intellectual hexis : and close inspection o f the text o f B
reveals that there are four different intellectual conditions which
bear the name.
First, and most explicitly, nous is the grasp o f the first principles
o f theoretical science; the understanding o f unproven necessary
truths which is the basis o f epistëmë (1 i40b3 i - i i 4 i a9). Together
with epistëmë., we are told, this nous constitutes sophia ( i i 4 i ai9).
B u t there is also another theoretical nous, which is concerned with
individuals and is a kind o f perception ( 1 i4 3 b6) : this is the kind o f
understanding which we must have o f a particular instance if we are
to grasp it as an instance o f something general; the kind o f vision
whereby we see that a triangle is the simplest figure into which more
complicated figures can be analysed ( 1 1 4 2 “29). What makes it
appropriate to use the same name for the grasp o f very general and
abstract principles and for the appreciation o f features o f an
instance or diagram before one’s eyes? It is appropriate, Aristotle
would say, because in each case we are dealing with something
immediate and intuitive: in each case there is a contrast with proof
and reasoning.
Corresponding to these two types o f theoretical nous there are
two types o f practical nous. Such, at least, I take to be the message o f
one o f the most difficult passages in the whole o f the Ethics,
1 1 43a35_t>4 :
Nous is concerned with extremes in both directions: it is nous and not
reasoning whose objects are primary terms and ultimate particulars. There
is one nous of the unchangeable and primary terms in the realm of
demonstration, and another of the ultimate and contingent element, and of
the other premiss, in practical reasoning; for these are the principles of the
wherefore; for universale come from particulars, and for these you have to
have a perception which is nous.
It is common ground among commentators that in this passage
Aristotle is teaching that there is a practical form o f nous which
corresponds to the particular nous on the speculative side. But in
claim ing that there are here two forms o f practical nous— one
corresponding to the general abstract nous o f the speculative
faculty, as well as one corresponding to the particular nous— I am
departing from the majority o f commentators. When Aristotle says
Wisdom in the Aristotelian Ethics 171
that there is a nous dealing with ‘ the other premiss in practical
reasoning’ (en tais praktika is. . . tes heteras protaseös) I take it that
he means a nous dealing with the major premiss o f practical
reasoning, the premiss which contains reference to the end : this will
correspond to the intuitive understanding o f speculative axioms,
since axioms in mathematical reasoning, play the same role as the
wherefore, or hou heneka, in practical reasoning (A E C , i i 5 i ai7).
T h e phrase I render as ‘the other premiss’ is commonly rendered
‘the second premiss’ : most commentators take it that there is just
one practical nous, concerned with the changeable particulars which
are the topic o f the minor premiss o f a practical syllogism. T h is
makes irrelevant the introductory sentence ‘Nous is concerned with
extremes in both directions' and renders unintelligible the feminine
plural hautai in the sentence ‘ these are principles o f the wherefore’ .
I f we take it that Aristotle is saying that nous is concerned with both
extremes in both realms, practical and theoretical, then all becomes
clear: we have universal speculative nous (the nous ‘kata tas
apodeixeis ton akinëtôn horön kai proton1 o f 1 i4 3 b2 and 1 i40b3 i ff.),
particular speculative nous (the nous o f particulars which is aisthësis,
o f i i4 3 b6 and 1 i4 2a29), particular practical nous (the nous which is
‘en tais praktikais tou eschatou kai endechomenou’ o f i i 4 3 b2), and
universal practical nous (the nous o f the hetera protasis o f i i 4 3 b3).
A s in speculative matters, so in practical matters, nous provides the
starting-point and the stopping-point for reasoning ( 1 i4 3 bi 1). T h e
practical reasoning which is the existence o f phroriësis begins with a
conception o f an end, a hypolëpsis about the purpose o f conduct
( ii4 0 bi7 ); this right conception o f an end can itself be called
phronësis ( ii4 0 bi 3 ; i i4 2 b34). Practical reasoning ends with a
judgement about what is to be done, a self-addressed command
(114 3*9) or a piece o f advice to another ( 1 1 4 3 “ ! 5). T h is too can be
called nous (114 3*27) whether or not it is backed up in a particular
case by a statement o f reasons ( i i 4 3 bi4).
There is thus a parallel between the structure o f phronësis in the
logistikon and sophia in the epistëmonikon which can be illustrated as
follows :
epistëmonikon logistikon
{
nous (of axioms) nous (of end) Λ
logos = epistëmë logos—euboulia yphronësis
nous (of particulars) nous (of particulars)J
172 The Aristotelian Ethics
T h e pattern nous-logos-nous is common to both sides o f the
division. Ju st as sophia can be described as epistëmë plus nous
(‘ knowledge with a head on it’ i i 4 i ai9), so phronësis could have
been described as euboulia plus nous : instead, euboulia is described
as phronësis minus nous, as what you are left with if you discount that
element o f phronësis which is the right conception o f the end
(1 i4 2 b32~4). T h e particular nous is, in each case, the perception o f
an instance (of a term or a principle) as an instance : this may be
either with a view to acquiring the general term or instance (e.g.
n 43b5) or to applying it (e.g. U 4 3 bi5). Aristotle’s fullest account
o f particular nous comes when he is contrasting phronësis with
epistëmë :
That wisdom is not knowledge is clear : for as has been said it is concerned
with particulars, since action is particular. It is something which perceives
by nous ; for nous is concerned with terms of which there is no deduction,
and wisdom with particulars of which there is no science but only sense
perception—not the perception of the proper sensibles, but that whereby
we see in a mathematical case that a triangle is an ultimate element,
another stopping-point. This wisdom is perhaps better called perception :
it is a different kind from the wisdom we were talking about earlier.
( 1 1 4 1 a^4 3 1)1
Aristotle nowhere gives an example in the A E o f what is perceived
by this phronësis-nous: he relies on the analogy with the
mathematical case o f the analysis o f a figure and with cases o f the
exercise o f a skill such as medicine ( 1 1 4 i bi7 ff.). Presumably
instances o f the kind o f thing he has in mind would be the
perception that A needs cheering up, that B is being offensive, and
that one is hurting C ’s feelings : an awareness, we might say, o f the
morally relevant features o f one’s action and the situation in which
one is acting. Som eone who is blind to this kind o f thing— who does
not, as Aristotle puts it, have an eye for this kind o f thing ( 1 i4 3 bi4,
1 I 44a3 ° ) — will never be able either to acquire or to apply general
principles about the good life or admirable action.
In all these passages nous is an acquired or natural condition o f
the soul or one o f its parts : it is not a substance or faculty which can
i. Reading α ν τ ιλ η π τ ικ ό ν in 1142*25 with Kb, and understanding it in an active sense, as
suggested by the correction in the manuscript Ο3 α ν τ ιλ η π τ ικ ή ; reading also at 1142*29 30
α ΰ τ η μ ά λ λ ο ν α,ΐσθησις ή φ ρ ό ν η σ ις with the majority of the best manuscripts and taking
α ΰ τ η with φ ρ ό νη σ ις. So read, the text agrees with, rather than contradicts, the passage at
1 1 4 3 * 3 5 ff·
Wisdom in the Aristotelian Ethics 173
be contrasted with and judged superior to the whole soul-body
compound as it is in N E 10 at i i7 7 b28. B u t if we compare the A E
use o f nous with that o f the E E we find no such difficult contrast as
we find when we compare the A E and the N E . Nous is uncommon
in the E E until the final book : in the earlier books it appears only in
the popular phrase noun echön (having common sense) and in
conjunction with theos in the refutation o f the Platonic Idea o f the
Good ( i 2 i4 b3 i ; i2 3 7 b38 ; i 2 i 7 b3 i). In the final book, however, it
appears a number o f times, in senses which match those we have
seen in the disputed book B.
T h us at 1 246“ 1 3 - 14 the incontinent man is described as an
intemperate man with nous (pös akolastos . . . echön noun) : this is no
doubt a popular saying o f some kind, but it suits very well the
theory o f B and C , according to which the incontinent man differs
from the intemperate man by having a correct appreciation o f the
end— the nous tës heteras protaseos o f 1 14 3 b3 ■ At i24 7a3o nous
appears along with nature and tutelage as a possible cause o f
successful action: here it must mean generic intellectual ability,
including both phronësis and technë. A t i24 6bio epistëmë and nous
are linked together as they are in the account o f sophia in B ; at
i24 8 ai 7 f f . nous and bouleusis— the constituents o f phronësis
according to our account o f B — appear together as the conditions o f
right desire. A t 1 24 8 ^ 9 we are told that (moral) virtue is the tool o f
nous, just as at i240bi 1 we were told that phronësis was the employer
rather than the employee o f virtue. In all these passages in book
V I I I o f the E E nous appears in senses familiar from B : either as the
generic cognitive apparatus, or as the intuitive element in learning,
or as the intuitive element in wisdom.
It is the final chapter o f the E E that is the closest parallel to N E
10, and here the word nous is not used. Aristotle is looking for the
criterion which the good man should use in his choice o f the good
things o f nature and fortune. H e says :
Here as elsewhere one should conduct one’s life with reference to one’s
superior, and more specifically to the quality of one’s superior’s activity. A
slave, for instance, should look to his master’s, and everyone to the
superior to whom he is subject. Now a human being is by nature a
compound of superior and inferior, and everyone accordingly should
conduct their lives with reference to the superior part of themselves.
However, ‘superior’ is ambiguous: there is the sense in which medical
science is superior, and the sense in which health is superior; the latter is
174 The Aristotelian Ethics
the raison d’être o f the former. It is thus, that matters stand in the case of
our intellectual faculty. For God is not a superior who issues commands,
but is the raison d'être of the commands that wisdom issues. But ‘raison
d'être' is ambiguous, as has been explained elsewhere—this needs saying,
since of course God is not in need of anything. To conclude: whatever
choice or possession of natural goods—health and strength, wealth,
friends, and the like—will most conduce to the contemplation o f God is
best: this is the finest criterion. But any standard of living which either
through excess or defect hinders the service and contemplation of God is
bad. (i249b6 -2 i)‘
T h e word ‘ nous', as I have said, does not appear in Aristotle’s text,
but von Arnim proposed to insert it in place o f each occurrence o f
the word ‘ G o d ’ , whose presence he attributed to a Christian
interpolator. T h e proposal has no support from any manuscript or
version, and is implausible in itself— what Christian would assent
to the proposition ‘G od is not a superior who issues commands’ ?—
i. M y translation of this passage, like every other translation that has been offered, is
controversial at a number of points. I follow Bekker’s text exactly, even with respect to
punctuation; with the exception of the misprint ά ρ χ ο ν τ ε ς for ά ρ χο ν το ς in line 8 . None of
the emendations that have been proposed seems to me to be necessary to make the text
intelligible, and few of them seem even to improve it. ‘Superior’ is not always the most
natural translation of α ρ χ ώ ν but seems the best English word to capture the two senses
present in the Greek of ‘higher in a scale of value’ and ‘higher in a chain of command’ (as in
‘superior officer’). There is no reason to emend π ρ ο ς τ η ν ε ξ ιν κατά τ η ν εν ε ρ γ ε ία ν inline7 ;
what is meant is that the slave should help his master to do whatever his master is doing as
well as it can be done: he should strive to see that his master’s ε ν έ ρ γ ε ια has the ε ξ is of
goodness. The most difficult clause to make sense of as it stands is έκ α σ το ν δή δεο ι π ρος
τ η ν έ α υ τω ν α ρ χ ή ν ζην. In this clause I think 8η should not be emended: Aristotle is
drawing a conclusion, not adding a second premiss. The genitive plural εα υ τώ ν is puzzling:
Rackham’s emendation i v α ν τω is attractive, and certainly expresses clearly what I take to be
the sense. The plural is perhaps to be explained, as my translation suggests, as looking back to
the implicit plural in έ κ α σ το ν ; or it may be due to the self being looked on in this passage as a
plurality of parts. The genitive is as it were partitive: ‘that part of themselves which is an
α ρ χ ή ’. I am unable to settle the parallel ambiguities of ο ϋ τω and θ ε ω ρ η τ ικ ό ν in the
sentence ο ν τω δ ’ί χ ε ι κ α τ ά το θ εω ρ η τικ ό ν . Does ο ϋ τω mean ‘like the foregoing (viz.
“ superior” ) in being twofold’ or ‘like the foregoing (viz. health) in being a raison d ’etre' ? Does
θ εω ρ η τικ ό ν mean the intellectual faculty in general, or the speculative faculty in particular ?
Elsewhere in the E E the use of the verb θ εώ ρέω and its derivatives would support either
reading: the wide sense is to be found at 12 i7b26 ff. ; i226b26; 1245*23, and the narrow sense
at I 2 i 4 “ i 3 ; I 2 i 5 b i 3 ; I 2 i 6 b i 3 ; I 2 i 9 a i 7 , b2 6 ; I 2 2 i b 5 ; 1227 ' % b2 Q ; other passages are
themselves ambiguous or use the verb in the literal sense o f‘see’. 1 incline to think that the
faculty meant here is the theoretical intellect, the ε π ισ τ η μ ο ν ικ ό ν of B : this fits the
commoner use of the verb in the EE. But provided that the first sense o f‘thus’ is taken with
the broader sense of ‘intellectual faculty’ , or the second sense with the narrower sense, there
is no difference in the over-all resulting significance of the passage. I owe to Professor D. J.
Allan the neat translations of ου εν εκ α as raison d’etre and of σ ώ μ α τ ο ς α γ α θ ά as “ health and
strength” .
Wisdom in the Aristotelian Ethics 175
so that it is not surprising that it has not been taken up by other
commentators. M ?n y distinguished scholars, however, while
rejecting the proposal as an emendation have accepted it as an
interpretation : the word ‘ G od’ here, they claim, does not refer to a
transcendent G od o f the universe, but to the nous immanent in each
o f us, which is described in N E 10, at i i 7 7 a2 8 - 3 i, as something
divine. I f this interpretation is correct, then the expression I have
translated ‘the contemplation o f G o d ’ must be translated as ‘ G o d ’s
contemplation’— the genitive must be taken to mark the subject,
not the object o f the contemplation; and the phrase rendered ‘ the
service and contemplation o f G o d ’ needs to be taken accordingly as
‘ the service o f G od, and contemplation’ . T h us there will be found
in this passage a twofold division o f the intellectual faculty : one part
in which inheres prescriptive wisdom, the other which is called
‘ god’ and issues in contemplation. T h e twofold division will
correspond to the division between the logistikon and the epi­
stëmonikon in B , while the divine status accorded to the con­
templative nous gives the present passage a close resemblance to N E
io .1
Ingenious though it is, this interpretation is not in the end
credible. T h e previous chapter o f the E E contains several
references to G od , which are clear references to the cosmic G od o f
the universe, who is not to be identified with, but is explicitly said to
be more powerful than, nous and science. T h is God is contrasted
with ‘the divine in us’ , which in that context appears not to be any
speculative faculty, but human reason in general.2 It is hard to
accept a quite unprepared switch to a use o f ‘ G od ’ to refer to the
faculty o f contemplation : this would be so even if there were any
parallel for such a use, which there is not.3 Bonitz offers no parallel
either for the subjective genitive with theoria ; and it can hardly be
said o f the speculative faculty that it ‘ is in need o f nothing’,
1. See D. Wagner, 1 9 7 0 , passim.
The text is in part corrupt and is not easy to restore completely with
2 . I 2 4 8 * i 5 - b4 .
confidence. Fortunately I do not need to discuss its emendation—other than to say that I
accept Spengel’s reading of “28, τι' οι!ν αν κ ρ ε ΐτ τ ο ν κ α ι ε π ισ τ ή μ η ς εϊη κ α ι νοΰ π λ η ν
θ εό ς, which is supported by the old Latin version of the De Bona Fortuna— for the points
made above remain true on any of the proposed reconstructions including von Arnim’s.
3 . D. Wagner has devotedly combed classical literature for descriptions of nous, and other
abstractions, as divine (op. cit. 1 2 8 - 3 9 ) . But describing something as divine, or as a god, is
quite different from using the word ‘god’ baldly to refer to it. Politicians and leader writers
frequently say that inflation is a cancer ; but if I read that someone has died of cancer, I do not
take this to mean that his death was due to the depreciating value of money.
176 The Aristotelian Ethics
particularly in a context in which— on this view— wisdom is said to
be issuing commands for its benefit. Finally, it is possible to make
perfect sense o f the passage without giving the word ‘ G o d ’ any such
strained sense.
When Aristotle says that a human being is by nature com­
pounded of superior and inferior, it is not the composition o f soul
and body that he has in mind, but the division between the
irrational and the rational part o f the soul. I f we apply to this case the
general rule that one should look to one’s superior, we infer that a
human being should govern his life in accordance with what is
required by his rational soul. But there are two different kinds o f
arche : in matters concerning our body we need to take account o f
the requirements (i.e. prescriptions) o f medical science; we also
need to take account o f the requirements (i.e. needs) o f our health.
Both medical science and health are archai in different ways : and
the latter is the raison d'être o f the former. W e must apply this to the
mind : there too there is an arche like medicine, and an arche like
health. T h e archë corresponding to medicine is wisdom, which like
medicine issues commands. So applying the rule about looking to
one’s superior, we must obey the commands o f wisdom, and take
account o f the needs o f that which is related to wisdom as health is
related to medical science. So far— up to 1 2 ^ 1 4 .— all is reasonably
clear. But now we naturally want to ask: well, what is it that stands
in the same relation to wisdom as health does to medical science?
Instead o f giving us an immediate answer, Aristotle goes on ‘For
G od is not a superior who issues commands. . . .’ T h is sentence,
and especially the ‘for’ is at first baffling in the context, and provides
the principal reason which has made commentators want to change
either the reading or the sense o f ‘G o d ’ so as to make Aristotle refer
here to the speculative intellect.
T h e relevance o f the sentence becomes clear if we connect it with
the previous chapter. T h ere we learnt that the cosmic G od was the
supreme arche in the soul, superior to logos, epistëmë., and nous. If,
without further ado, we applied the rule that everyone should live
according to the requirements o f his superior as a slave lives
according to the requirements o f his master, we would most
naturally conclude that in matters o f the soul we must live in
obedience to the commands o f G od. Bu t this would be, in
Aristotle’s view, to make an erroneous move. We must first
distinguish between senses o f archë: fo r G od is not a ruler who issues
Wisdom in the Aristotelian Ethics 177
commands. We thus see the appropriateness o f the ‘for’ : it explains
why the distinction just made had to be made. It is indeed true, even
according to the final chapter o f the E E , that G od is the supreme
archë o f the soul : but not in the sense in which the analogy with
slave and master would suggest.
Shall we say, then, that it is G od who stands to wisdom in the
relation that health stands to medical science? T h e rule about
looking to one’s superior would then be applied in this w ay: we
must obey the commands o f wisdom, and take account o f the needs
o f G od. But that in turn would be absurd : since God o f course is not
in need o f anything. T o prevent this absurd conclusion being drawn,
Aristotle has to make a further distinction. Health was related to
medical science, we said, as its raison d'etre or hou heneka. But the
hou heneka o f an institution or activity may be either that which it
seeks to attain, or that which it seeks to benefit. Aristotle often
invokes the distinction, without offering a clear instance to illustrate
it (Phys. I I .2, 19 4 *35-6 ; De An. II.4 , 4 i5 b2 -3 , b2 0 - i ; M et. Λ ,
i0 72b2). Everyday examples o f what he has in mind are easy to find :
if I go into a toyshop to buy a toy railway for my young son, the
railway is the hou heneka o f my shopping trip in the one sense, the
boy is the hou heneka in the other sense. M et. Λ , i0 7 2 bi applies the
distinction to the F irst M over: as Ross paraphrases the passage, ‘ A
final cause in the sense o f that whose good is aimed at cannot be
found among unchangeable things, but a final cause in the sense o f
the good aimed at can; it moves by being loved’ (Ross, 1924, 373).
T h is is a close parallel to the present passage: Aristotle is here
saying that G od is not a raison d'être whose good is being aimed at,
but a good whose attainment is the raison d'être o f wisdom’s
commands.
Well, then, what is the hou heneka whose good is being aimed at,
and what is it to attain G od? Aristotle gives no explicit reply, but
the answer is clear enough from the concluding passage : that which
benefits from the commands o f wisdom is that which serves and
contemplates G od ; and this service and contemplation is itself the
only kind o f attainment of G od which is possible. And what is it that
serves and contemplates G od? Either the soul as a whole, or its
speculative part : which o f the two is intended must depend on what
is meant by ‘service’ here. And this, notoriously, is a very difficult
problem.
T h e only other place in which Aristotle speaks o f serving G od, or
178 The Aristotelian Ethics
the gods, is in Pol. H , 1329*30 ff., where the expression clearly
means ‘ worship’ in a literal, liturgical sense. Commentators have
found it astonishing that in the last chapter o f E E Aristotle should
offer as a criterion for the goodness or badness o f choices their effect
on liturgical worship. Accordingly, they have preferred to take
‘serving G o d ’ in the sense o f ‘cultivating the intellect’ , citing as
parallels N E 10, 1 i7 9 a23, where we are told that the therapeia o f the
nous endears us to the gods, and Plato, Timaeus 90 c 4 where the
expression therapeuonta to theion will bear this sense. B u t if the bare
theos in this context cannot mean nous, as we have argued, this
interpretation is excluded.
I f we are to look for Platonic parallels a much more appropriate
one is provided by the Euthyphro. T h ere a long section (12 *2 4 -
i4 ai) is devoted to a discussion o f he tön theön therapeia in which
Socrates argues against Euthyphro that the notion is unintelligible.
T h ere are a number o f themes paralleling the present text:
therapeia cannot be for the benefit o f the gods, whose condition we
cannot hope to alter for the better; therapeia is what slaves give to
masters ; but what slaves do for their masters depends on what their
m asters’ ergon is: what then is the splendid achievement, the
pankalon ergon, which is to be achieved by the gods with our
service? ‘Polla kai kala’ replies Euthyphro to Socrates, who
promptly turns to some inconclusive teasing.
T h e service o f the gods which Euthryphro has in mind includes
prayer and sacrifice: but it includes also acts o f justice such as
Euthyphro’s attempt to punish a m urderer— the endeavour which
gives the whole dialogue its framework. I f Aristotle does have the
Euthyphro in mind here, then the service o f G od could well include
acts o f moral virtue. Th ese are the kalai praxeis o f the kalos kagathos
which are the subject o f the first part o f the same chapter; they
could well be regarded as the polla kai kala which we, under the
arche o f G od ( 1248*25 ff.) find our fulfilment in performing and by
which we make our contribution to the splendour o f the universe.
T h is interpretation o f ‘ therapeia' is highly speculative: but given
the excessively laconic style o f the conclusion o f the E E , no
interpretation can be anything else. It is certainly not alien to
Aristotle’s manner in the E E to defend the moral opinions o f the
plain man against the paradoxes o f Socrates.
It is with relief one turns from decoding the cryptogram o f E E
V I I I to looking for a parallel in the A E . T h e search is not, o f course,
Wisdom in the A ristotelian Ethics 179
a difficult one. It has long been recognized that the final chapter o f B
is very close to the conclusion o f the E E . H aving established that
wisdom and learning are the excellences o f the two parts o f the
rational soul, at 1 14 3 bi8 ff. Aristotle inquires what use each o f these
virtues is ? Learning does not inquire how to make men happy, and
the acquisition o f wisdom is surely no more necessary in order to
become happy than the acquisition o f medical science is necessary
in order to become healthy. ‘M oreover, it would seem absurd if
wisdom, being inferior to learning, should be put in authority over
it: as seems to be implied by the fact that whatever produces
something is superior to it and issues commands about it’ (1 i4 3 b3 3 -
5). These aporiai are solved as follows. Learning does not produce
happiness as medical science produces health: it is, rather, a
constituent o f happiness. Wisdom is not separable from virtuous
activity as the possession o f the science o f medicine is separable
from healthy activity. Finally
wisdom is not in authority over learning or the better part, any more than
the science of medicine is in authority over health ; it does not make use of
it, but provides for its coming into being; the orders it issues are not issued
to it but issued for its sake. To say it was in authority over learning would
be like saying that political science was superior to the gods, because it
issues commands about all the affairs of the state. (114 5*6 -11)
Here, as in E E , wisdom is something which issues commands
(epitattei: a word popular in the E E , but occurring only once in the
N E ) . Here, as in the E E , the prescriptive role o f wisdom is likened
to that of medicine. H ere, as in the E E , it is a function o f wisdom to
promote the contemplative activity o f the soul: in the E E its
function is to promote the contemplation o f God, in the A E it is to
promote the activity o f the better part o f the soul, whose objects are
unchanging, eternal things, the most honourable things by nature,
the divine constituents o f the cosmos, and in general extraordinary,
marvellous, mysterious, and superhuman things ( i i 3 9 b2 3 ;
i i 4 i a20, bi-~7). In the E E there is stressed the relationship between
wisdom and the hou heneka in the sense o f hou : namely, G od. In the
A E there is stressed the relationship between wisdom and the hou
heneka in the sense o f h o(i) : namely, ‘the better part o f the soul’ .
B u t in substance the doctrine is the same.
T h e nous passage in N E 10 is the closest Nicomachean parallel to
these two passages o f the A E and E E , but it is much further from
 The Aristotelian Ethics
each o f them than either is from the other. Phronësis in N E does not
appear as the handmaid o f learning, but only as the yokemate o f the
moral virtues ; the activity o f contemplation is attributed not to an
element of the soul, but to a separable nous contrasted with the
human soul-body compound. In A E C , as in E E V I I I , 2, there is a
divine element in us: but as in E E V I I I 2, it is not the theoretical
nous : it is appetitive rather than speculative, the origin o f the right
sort o f desire for pleasure (compare i i S 3 b32 with 1248 *17-29 ).
(3) Phronësis and M oral Virtue. In N E 10 we are told: ‘Wisdom
is wedded to moral virtue, and moral virtue to wisdom, since the
principles o f wisdom are determined by the moral virtues, and
rightness in morals is determined by wisdom’ ( ii7 8 ai6 - i8 ) . T h is is
a neat summary o f the teaching o f all three treatises. In both the N E
and the E E a reference to wisdom is incorporated into the definition
o f moral virtue. In the preliminary general treatment o f the subject
in the N E virtue is defined as a state o f character expressed in choice
lying in a mean determined by reasoning in the way that the wise
man would determine it ( n o 6 b36)— a definition developed by
Aristotle from the commonly accepted account o f virtue as action
‘ in accordance with right reasoning’ (1 i0 3 b32). In the E E virtue is
defined as a state o f character which makes people capable o f
performing the best actions and puts them in the best disposition
with regard to the best and greatest good ‘the best and greatest good
being that which is in accordance with right reasoning, that is the
m ean’ (i2 2 2 a9 -io ). A fter the treatment o f magnanimity, Aristotle
fills this out: ‘ the virtue concerned with each subject matter judges
correctly what is the greater good and what is the less, just as the
wise man and his virtue would command, so that all the virtues go
with that virtue, or that virtue goes with all virtues’ ( i2 3 2 a25-38).
In each Ethics there is a promise to expand, at a later point, the
notion o f ‘right reasoning’ (N E 2, n o 3 b3 3 ; E E i2 2 2 b8). T h e
promise is not fulfilled in the N E , but both A E and E E contain pas­
sages which refer back to it. T o say that the mean is in accordance
with right reasoning, we are told in B (1 i3 8 b5 if.), is true but
uninformative : it is as i f one were to prescribe the medicines ‘that
medical science, and the man who possesses it, would command’
(1 i3 8 b3 i ) . 1 We must inquire, Aristotle says, what right reasoning
is, and what is the standard or criterion which determines it (horos).
i . The similarity of language between E E III and A E B is striking, a n e p ό φ ρ ό ν ιμ ο ς âv
κ*λ€νσ€ΐ€ κ α ί ή ά ρ € τη ( ΐ 232β3);οσα ή ια τρ ικ ή κ ζ λ ζ ύ ε ι κα ι ώ ς ό τ α ύ τ η ν ς χ ω ν ( ι ι $ * $ 2 ).
Wisdom in the A ristotelian Ethics 181
T h e first question is answered at some length in book B itself: right
reasoning in these matters, we are told at 1 i4 4 b23, is phronësis ; and
phronësis is the major topic o f the book. But what of the question:
what is the horos o f right reasoning? T h ere is no clear answer to this
question in book B : some commentators take this as evidence that
the book is in a fragmentary state, others that Aristotle though the
question was misconceived, others that we are to look for the answer
to book 10 o f the N E . 1 T he third answer seems to me more nearly
correct than the other two, as becomes clear if we consider the
analogy with medicine. What the Aristotelian doctor possesses and
the rest of us lack is not so much a set of rules o f thumb for
prescribing, but a knowledge o f the nature o f health itself : he knows
what the mechanism o f the body is, how its humours are balanced
when it is in good working order, what are the functions and
interrelations o f its parts. So in ethics, if we are to know the right
action to do we must know the nature o f a healthy— i.e. virtuous
and happy— mind : we must know what the parts o f the soul are,
what their functions are, and how they are related to each other.
T h is is why the horos can also be called the skopos (1 i3 8 b23) : it is a
right understanding of the end o f conduct, namely a good life, that
is necessary for right reasoning in morals, just as it is a right
understanding o f health, which is the end o f medical practice, that
is necessary for right reasoning in medicine. In the N E it is in book
10 that we are given the fullest account o f the nature o f the good
life— the contemplative activity o f nous— and to this extent those
who say that it is to N E 10 that we must look for the horos are
correct. But o f course in B itself in the concluding chapters the
relationship o f phronësis to sophia is discussed and the relationship
o f both to happiness is sketched (114 4 * 3 -6 ; 114 5 * 6 - 11) . T h e
relevance of this to the question of the horos is not however brought
out either in N E 10 or in A E B. F o r that we have to turn to E E V I I I .
In E E V I I I the question of the horos is taken up in language
reminiscent o f the beginning o f B.
I. T h e first is the view o f Allan (1970, 136) and the second o f Rowe, who writes ‘W ithin the
limits o f A ristotle’s system, no detailed criterion is in fact possible . . . either the question
“ what is the standard” means what it says, and the answer is that there is no such thing; or
else it is sim ply a misleading way o f putting the more general question, how we apply the
general rule to the particular case1, in which case the answer is the intuition o f the φ ρ ό ν ιμ ο ς
( 19 7 1, I I 2 - I 3 ) . Both Allan and Rowe are mistaken, I believe, in thinking that Aristotle is
here concerned with the formulation and application o f rules. T h e third view is that o f
D irlm eier (1969, 442) and other commentators.

i82 The Aristotelian Ethics
T h e doctor has a standard by reference to which he distinguishes a healthy
from an unhealthy body and the degree to which each activity is advisable
and wholesome, and beyond which in either direction it ceases to be so.
Sim ilarly in regard to actions and choices o f things which by nature are
good but not praiseworthy, the good man should have a standard o f
possession, choice, and avoidance concerning abundance and scarcity o f
wealth and other gifts o f fortune. E arlier we said “ in accordance with
reasoning” — but this, like saying in m atters o f diet “ in accordance with
m edicine and m edical reasoning” , is true but uninform ative. ( i24Qa2 i - 6 ) 1
Aristotle then goes on to give, as an answer to the question about
the horos, the rule that everything should conduct its life with
reference to its superior, which we considered in the previous
section. T h e resemblances to B in the reference to the doctor, and in
the use o f the phrase ‘true but uninformative’ strike all com­
mentators. T h e resemblance may seem too striking: if B belonged
originally in the same context as E E V I I I , would not this passage be
a useless reduplication o f the earlier one? Bu t setting aside the fact
that the undoubted E E and the A E itself contain a number o f
apparent doublets, a closer look at the text shows that it is not mere
repetition o f the B passage. Aristotle does not say here, as he does in
B, that we must inquire what is the orthos logos: the answer he goes
on to give is an answer to the question what is the horos— the horos
of kalokagathia or perfect virtue, as is said in the concluding
summary (i2 4 9 b24). T h u s E E V III can perfectly well follow A E B
as part o f the same treatise: a twofold question is raised at the
beginning o f B , and the first part o f the question is there answered;
the second half o f the question is raised in V I I I , in terms
reminiscent o f the twofold question o f B , and the second h alf o f the
question is then answered.
Som e have disputed that the final passage o f E E V I I I is an
answer to the question raised in B. Row e writes: ‘E E V I I I , 3
severely restricts the application o f its suggested horos: it is
introduced as a standard for “ actions in connection with and
choices o f things naturally good but not praiseworthy” . . . there is
no justification in the text for supposing that E E is giving a standard
for actions in general’ (op. cit. n o ). L ik e some other scholars, Rowe
must be treating as an editorial addition the final statement that it
has just been said what is the standard o f perfect virtue. There is no
i. Reading, with Ross, ύ γ ιζ ιν ό ν for ev ύ γ ιά ίν ο ν in line 24, and φ υ γ ή ς n ç p l for n e p l
φ υ γ ή ς in line b3, after Allan and the L atin version.
Wisdom in the Aristotelian Ethics 183
reason for regarding this statement as spurious ; but even if one does
so, the discussion in E E V I I I does give the general standard for the
exercise o f virtue. T h e ‘things that are good by nature but not
praiseworthy’ are health, strength, honour, birth, wealth, and
pow er(i248 b24, b2 8 -3 o ; i249ai o - n ) . T h ese are the subject matter
o f the virtues of magnanimity, magnificence, and liberality (E E III,
I 2 3 i b28 ff.). T h e subject matter o f the virtues o f courage,
temperance, and meekness, on the other hand, are the passions of
the irrational soul. F o r these too a standard is sketched in the last
chapter o f the E E : ‘ T h us it is, too, with the soul, and this is the best
standard for the soul, to be as little as possible conscious o f the
irrational part o f the soul qua irrational’ (i249 b2 2 - 5 ) .1 T h us the last
chapter of the E E does give a standard for the exercise o f the six
virtues which are discussed in the official treatment o f the virtues in
E E I I I and which together constitute the perfect virtue or
kalokagathia o f the first part o f the final chapter. Other praise­
worthy states, like dignity and candour and wit, are no doubt
difficult to bring within the scope o f the criterion offerred by E E
V I I I : but then according to the E E these states are not virtues
(i2 3 4 a25) and since they do not involve prohairesis (i2 3 4 a26) they
do not call for a horos to guide choice.
As has been said, it is ground common to all three treatises that
wisdom and moral virtue are inseparably linked: but how and why
this is so is spelt out at much greater length in the E E and the A E
than it is in the brief quoted passage which is the N E ’s only
contribution to the topic. T h e A E has a long argument to the
conclusion that ‘ it is impossible to be good in the strict sense
without wisdom, or wise without moral virtue’ ( ii4 4 b30 -2). It is
indeed possible to do good actions— the actions that a good man
would do— without possessing either wisdom or moral virtue; but
actions only express virtue if they are done out o f prohairesis and for
their own sake (i i4 4 ai3~2o). Even good actions which are not the
expression o f virtue do o f course call for some skill in planning: but
this does not mean that there can be wisdom without virtue, for the
planning is the expression not o f wisdom, but merely of
intelligence— intelligence which could be used for evil ends as well
as good. Intelligence is a necessary, but not a sufficient, condition of
wisdom: intelligence only becomes wisdom if accompanied by a
i. Reading ο ν τ ω for το ύ το , το for τ α , and α λ ό γ ο υ lor άλλου (Fritzsche).
184 The Aristotelian Ethics
correct understanding o f the supreme good for man (that ‘ right
conception o f the end’ which is itself called ‘wisdom’ at i i 4 2 b34).
But only a morally good man has a correct understanding o f the
nature o f the supreme good : wickedness perverts us and deludes us
about the ultimate grounds o f action (cf. 1 i40bi 1-20 ). So wisdom is
impossible without moral virtue (1 i4 4 a2 i - bi).
M oral virtue, conversely, is impossible without wisdom.
Without wisdom, and even in childhood, one can feel drawn to
justice, temperance, courage; but such good tendencies, without
wisdom, can be positively harmful, like the strength o f a blind man.
Only wisdom will turn these natural virtuous inclinations into
genuine moral virtue. Socrates went too far when he said that each
virtue was itself a branch o f wisdom; but he was right in thinking
that virtue was impossible without wisdom. Those on the other
hand who say that virtue is determined by right reasoning (kata
ton orthon logon) are understating the case: it is not enough to
act in accordance with a man’ s instructions, as one might act in
accordance with a doctor’s instructions without oneself under­
standing medicine. T h e choices o f the virtuous man must be
determined by right reasoning in his own person ‘Virtue is not just a
state determined by right reasoning, but a state accompanied by
right reasoning: and right reasoning about these matters is
wisdom .’ T h us virtue without wisdom is as impossible as wisdom
without virtue ( ii4 4 bi~3o).
It is noteworthy that in this passage Aristotle corrects as
inadequate the formula which he him self gives in N E 10 to explain
the way in which wisdom and the virtues are interlinked. Nowhere
in the N E itself is it suggested that this formulation is a provisional
or abbreviated one: in the treatment of virtue in the early books, no
reason is given why thephrommos who defines the mean must be the
same person as the virtuous man who is acting according to the
mean.
T h e question whether wisdom and virtue are separable is
discussed at length at the beginning o f E E V I I I in the form o f an
inquiry whether wisdom can be abused. Here, as in A E B , Aristotle
is arguing against the Socratic identification o f virtue and
knowledge.1 Som e things, he says, can be used for their natural
I. The passage (1246*26- 36) is a difficult one, and like the final chapter of E E V III it has
suffered from the well-meant attentions of emendators who were unwilling to make a
sufficiently patient effort to understand the text handed down. Undoubtedly it is corrupt in
Wisdom in the Aristotelian Ethics
purpose and also for other purposes ; and they can be used for other
purposes in two different ways, either qua the kind o f thing they are,
or quite per accidens. Thus an eye can be used for its natural
purpose, to see straight; you can also (if you push on one eye in the
way described in Problems 958*5 ff.) use your eyes to see double:
even this perverse use o f them is a use o f them qua eyes. Selling an
eye, on the other hand, or eating one, counts as a per accidens u se.1
Knowledge, too, is capable o f multiple use : you can use it correctly,
or use it to make a deliberate mistake, e.g. in orthography— using it,
as it were, like ignorance.2 T h is is a sort o f perverse use o f one’s
hand in writing: there is the third kind o f use o f one’s hand when it
is used as a foot by dancers who dance on their hands and use their
feet as hands.
Now if every virtue was a branch o f knowledge, virtue could be
misused like knowledge. You could use justice as if it were
injustice: a man doing unjust things would then be acting unjustly
out o f justice, just as one can behave ignorantly out o f knowledge.
places, but there is no need for such extensive reconstruction as is proposed, e.g. by M oraux
in M oraux and H arlfinger ( 19 7 1, 2 8 1-3 ).
In my paraphrase I follow Bekker’s text with the following exceptions.
In line 26 I omit φ ίλ φ with P.
In line 27 I read η f/ αν for ή δ ν following Jackson and Dirlmeier.
In line 30 I accept von Fragstein’ s neat emendation ό τ ι ή σ τ η ν â ’ ό φ θ α λ μ ω for ό τ ι 8 '
όφ θαλμω .
In line 35 I accept Spengel’s emendations el δη for ή dr/ and àpera i for άρισται.
In line 36 I read ei'r/ âv for e v a v and in line 37 α δ ικ ή σ ε ι for cl δ ίκης i l with Spengel.
In line b4 I propose to read eti el instead o f è vel, which removes the inconsistency between
!246 b4 and ” 35.
In line 8 I accept Victorius1 reading.
In line ίο I omit e n with P.
In line 1 5 I am attracted by von Fragstein’s proposal to fill the blank in the manuscripts (η . . . .
σ φ ι . . . .) with ή ο νο υς φ ρ ο ν€ f.
In line 20 I read i v τ ω ά λ ό γ ψ for μ,ζν τ ω λ ο γ ω following Susem ihl.
In line 24 I accept Victorius’ emendation o f κ ό λ α σ ιν a v into α κ ο λ α σ ία ν .
In line 25 I accept Spengel’s proposal o f i o n 8k for in t re.
In line 28 I read οΰ for ούν ό.
In line 29 I propose éa v evfj ή for i à v fj.
In line 3 1 I follow Jackson in inserting a after y à p .
In line 36 I read γ ν ώ σ ς ω ς for γ ν ω ς after Sylburg.
I also depart from time to time from Bekker’s punctuation, as will be clear from the
paraphrase.
1. Commentators with a philological background find this bizarre, and hasten to emend
the text. Philosophers greet the examples as instances o f their familiar stock-in-trade.
2. Several scholars (especially Allan) have drawn attention to the protracted parallel
discussion o f deliberate mistakes in the Hippias M inor (373 c ff.). T here as here there is a
discussion o f a deliberate distortion o f one’s vision (374 d).
ι86 The Aristotelian Ethics
But i f this is impossible, as it is, it is obvious that the virtues are not
branches o f knowledge. E ven if you say that one is not strictly
ignorant out o f knowiege, but merely behaves incorrectly in the way
that an ignorant man would, the argument still holds. F o r you
cannot do, out o f justice, the things you would do out o f injustice.
B u t even if most virtues cannot be misused, perhaps phronësis
can. F o r if phronësis is a branch o f knowledge, a piece o f truth, then
it will perform in the same way as knowledge does. T h at is, it will be
possible to act stupidly from wisdom, and go wrong in the same way
as a stupid person. D oing stupid things out o f wisdom might be
called either acting wisely, or acting stupidly. I f the former is
the correct description (so that any action done out o f wisdom is
wisely done, so that the ‘ use’ o f wisdom is single and not multiple)
then the first o f the two descriptions just suggested is incorrect:
doing stupid actions from wisdom will be acting wisely, not acting
stupidly.
L et us take the two possible descriptions o f this situation in turn.
Suppose that doing things out o f wisdom is acting wisely. This
would parallel the case where making deliberate mistakes in
grammar or medicine expresses a knowledge o f grammar or
medicine. But the two cases are not parallel : the misuse o f grammar
and medicine is possible only because there is a superior branch of
knowledge perversely em ploying the skill. But no branch o f skill
can thus boss wisdom, which is itself the boss o f all; nor can science
or understanding (presumably because they would not lend
themselves to perversely employing anything.) N or can moral
virtue boss wisdom : wisdom, being the virtue o f the superior part o f
the soul, makes use o f the virtue o f the inferior part : it is the em­
ployer, not the employee. What other candidate is there for the role ?
Suppose on the other hand that doing stupid things out of
wisdom is acting stupidly. L et us start from two popular dicta:
incontinence is a vice o f the irrational part o f the soul; the
incontinent man is almost an intemperate man— he is an in­
temperate man who still has sense. Incontinence, so conceived,
m ight be thought to be a case o f the misuse o f wisdom. Bu t this
thought is mistaken. It is indeed true that strong desire can pervert
the reason and cause it to make judgements opposite to those o f nous.
B u t in that case, o f course, we have neither wisdom nor
incontinence. What we have is a case o f a bad state o f the irrational
part o f the soul causing a change for the worse from a good state o f
Wisdom in the Aristotelian Ethics 187
the rational part. It is clear that we shall have a change the other way
round if we have a good state o f the irrational part, and stupidity in
the rational p art.1 T h is will amount to our second possibility:
justice will be capable o f being used either justly (when the reason is
in a good state) or unjustly (when there is stupidity in the rational
part; and wisdom will be capable o f being used stupidly (when
there is evil desire in the irrational part). But this will not be all :
there will also be the possibility o f using injustice justly, and
stupidity wisely, i.e. acting wisely out o f stupidity. For (since surely
virtue is stronger than vice) it would be absurd if vice in the
irrational part could pervert the virtue in the rational part and turn
it into ignorance, while virtue in the irrational part cannot turn
stupidity in the rational part into wisdom, i.e. right judgement about
what is to be done. A fter all— on this view— wisdom in the rational
part makes intemperance in the irrational part act temperately : isn’ t
that what continence is ?
But all this is absurd : especially the idea o f acting wisely out of
ignorance. T h e very analogy on which our opponents rely tells
against them here. Intemperance may pervert medical skill or
grammar, but the opposite o f intemperance does not convert
ignorance into knowledge; because the opposite o f intemperance
does not contain the extra that would be needed to do so. It is virtue
as a whole— moral and intellectual— that has this sort o f superiority
over vice : the just man can do all that the unjust man can, just as in
general power ranges wider than impotence.2 (Virtue, o f course, is
not itself a power : but it includes as necessary parts the powers o f
intelligence and natural goodness, as we shall see in the next
section.)
So two conclusions emerge. First, wisdom and the hexeis o f the
irrational part o f the soul3 go hand in hand— to this extent Socrates
was right that there is nothing stronger than wisdom, because there
is no such thing as the misuse o f wisdom by vice. Secondly, wisdom
is not knowledge, because it is incapable o f being misused : hence
this second part o f Socrates’ thesis is wrong. Wisdom is not
knowledge, but another type o f cognition.
1. Commentators discuss whether the text here (bi6) and below (b2 i) should read ά γ ν ο ια
or ά νο ια . T h e difference is immaterial to the sense: ά γ ν ο ια about the subject matter o f
wisdom is a v o ia .
2. C f. the parallel text in A E A, i J 3 7 ai8.
3. Susem ihl’s emendation o f a i άλλου €%€ΐς into a l τ ο ν ά λ ο γ ο υ though
unnecessary, catches the sense exactly.
ι88 The Aristotelian Ethics Wisdom in the Aristotelian Ethics 189
T h us, in E E V I I I . i, the thesis that wisdom and moral virtue are i i 7 9 b2 i ff. But in the N E there is no close parallel to the A E
inseparably linked is confirmed by a many-pronged reductio ad teaching on natural virtue and inborn intelligence.
absurdum o f the suggestion that one might occur without the other. In the E E , however, the doctrine is clearly stated in the same
Each theoretically possible such combination and permutation is terminology. A t I227b29 ff. we are told that prohairesis is not
examined in turn and shown to lead to absurd consequences. T h e concerned with ends but means; ‘ hitting on the things that must be
doctrine is in perfect accord with, but in no way a mere duplication done for the sake o f the end belongs to another power’ — the
of, the closing chapters o f A E B. formula is that used to introduce deinotës in B. At i2 3 4 a28 we are
(4) Phronësis and Nature. One o f the more interesting theories of introduced to the natural virtues. ‘F o r every virtue, as will be
the A E is the account of the relationship between inborn and explained later, exists both naturally and otherwise, that is, in
acquired virtue. According to H 44bi ff. there are both cognitive conjunction with wisdom.’ T h e natural virtues then listed are
and affective natural qualities : in the cognitive faculty, correspond­ natural justice and natural temperance, just as in B at 1 i4 4 b5- T h e
ing to the acquired virtue o f wisdom, there is the natural power of fit is perfect.
intelligence or deinotës·, there are also natural traits o f character (epi Aristotle’s teaching on wisdom, then, so far from supporting
tou ëthïkou): ‘W e are just, capable o f self-control, brave and so on Jaeger’s theory o f the evolution o f his ethical theory, provides a
from the moment o f our birth’ ( ii4 4 b5-6). These natural virtues, number o f arguments to confirm the conclusion that the disputed
unlike the virtues strictly so called, can be possessed piecemeal books belong with the Eudemian rather than the Nicomachean
( ii4 4 a35): virtue strictly so called is only possible in conjunction Ethics. A study o f Aristotle on phronësis destroys a large part o f the
with wisdom ( i i 4 4 bi6). Intelligence differs from wisdom in two basis on which Ja eger’s theory was constructed.
ways : first, it is tactical rather than strategic, concerned not with the
formation o f a prohairesis but with its efficient execution ( 1 1 44a2 1 ) ;
secondly, it is capable o f misuse, and can help execute bad choices
as well as good ones, whereas wisdom is impossible apart from
virtue ( ii4 4 a3o).
Readers o f Aristotle are sometimes puzzled by this. I f we have to
acquire virtue in order to become wise, and we cannot become wise
without virtue, how can we ever acquire either virtue or wisdom ?
Are we not trapped in a practical vicious circle? T h e difficulty is a
spurious one. It is as if someone were to allege a difficulty about
getting married. H ow can anyone ever become a husband? F o r to
be a husband you need to have a w ife; but a woman cannot be a wife
unless she has a husband! As a single union simultaneously turns a
man into a husband and a woman into a wife, so the wedding o f
wisdom and virtue makes what was deinotës into phronësis and what
was physikë aretë into kyrws aretë.
T h e distinction between wisdom and intelligence is drawn
elsewhere in thc A E (C, i i 5 2 ai o - n ) but does not figure in the N E .
Nor are the physikai aretai mentioned as such, though just possibly
Aristotle has them in mind when at 1 i0 5 b23 he contrasts with hexeis
the dynameis or capacities to be affected by various emotions. T h e
need for a good natural grounding for virtue is stressed, as at

CH APTER 8
Happiness in the
Aristotelian Ethics
T h e references to happiness in the A E are b rief and sometimes
cryptic: but they are o f great importance in connection with the
question o f the original home o f the disputed books. For, as we shall
see, one o f the most important differences between the N E and the
E E lies in the different treatment o f happiness offered by each of
them. T h e difference is not simply one of presentation, but of
substance also. T h e N E sees happiness as constituted essentially by
the contemplative activity o f nous: this is the only happiness really
worthy o f the name, and the life o f wisdom and the moral virtues is a
second-rate kind o f happiness. In the E E on the other hand
happiness consists in the ideal functioning o f every part o f the soul :
the activity o f contemplation is only one, admittedly the highest
one, among a family o f activities which constitute the happy life.
T h is contrast, while it has been ignored or questioned by some
commentators, seems to me capable o f decisive demonstration.
Less easy to settle is which o f these two conceptions o f happiness is
the one to be found in the common books. I shall argue that in
general the A E in this respect as in others resembles the E E more
than the N E : but some o f the evidence, it must be admitted, is
ambiguous, and from one passage it is clear that while writing the
common books their author was at least aware o f the possibility of
the Nicomachean conception. Bu t before studying the texts o f the
A E it is necessary to examine at some length the relevant parts o f
the E E and the N E .
T o institute a comparison between the E E and the N E is in many
parts o f the work very easy. T h e same topics are often treated in the
same order, and one can tick o ff the arguments as they occur in each
treatise, noticing occasional omissions and variations on one side or
the other. T h e first books o f the treaties are not so simple to
compare. T h ey deal, broadly speaking, with the same topic:
Happiness in the Aristotelian Ethics 191
happiness. T h ey follow, broadly speaking, the same method : they
review important current opinions on the topic (with particular
reference to Plato) and then use the notion o f ergon to reach
Aristotle’s own definition o f happiness. And the definitions reached
in the end— in the case of the E E not until the first chapter o f the
second book— seem remarkably similar. F o r the N E ‘activity o f
soul in accordance with virtue, and if there are more than one
virtue, in accordance with the best and most complete’ is given as
the definition o f human good (1.6 , i098ai6 ); in the E E the
concluding definition of happiness is ‘activity o f complete life in
accordance with complete virtue’ (2 .1, i2 i9 a39). In each work,
having reached his own definition, Aristotle goes on to show how
well it fits with received opinion (N E 1.8 ff., E E II. 1), and takes as
his next topic the distinction between parts o f the soul as the basis of
the division o f virtues into moral and intellectual (N E 1 .1 3 , E E
I I . 1). But behind this over-all similarity there is a great difference
o f plan, o f style, o f proportion and even, once one begins to
examine it closely, in doctrine. There is nothing like the close
paragraph by paragraph parallelism we can find in later books.
T h e first book o f the E E is tightly structured: though the
structure escaped some earlier commentators, it has been well
brought out in a recent analysis by Row e (19 7 1, 15 ff.). W e begin
with a solemn literary introduction— unusual in Aristotle— which
leaves us in no doubt that the subject o f the work is happiness.
Inquiries may be practical or theoretical: ours, it is implied, is
practical, but theoretical considerations will be appropriate from
time to time. T h e first point o f inquiry is what is a good life and how
is it to be acquired ( I .i, I 2 i4 ai5). Immediately we are presented
with five candidate answers to the second question (by nature, by
learning, by training, by grace and by luck) and three candidate
answers to the first (wisdom, virtue, and pleasure).
So far all seems clear, but we may be brought up short by the
beginning o f the next chapter. It would be a mark o f great folly, we
are told, if one can choose one’s own way of life, not to organize
one’s life in pursuit o f some goal or end (skopos or telos), whether
honour or reputation or riches or culture : so the first thing which
must be done is to decide in one’s own case (for oneself?) what are
the constituents and the necessary conditions for a good life.
T h e relationship o f this passage to what precedes is puzzling. Are
honour, reputation, riches, and culture meant as alternative
IÇ2 The Aristotelian Ethics
answers to the question o f chapter one : what constitutes a good life ?
Or are they alternative descriptions o f the three canonical lives : the
life o f culture being the same as the life o f wisdom, the life o f riches
being perhaps the same as the life o f pleasure, and ‘honour’ and
‘reputation’ being both meant as synonyms for the life o f virtue?
N either answer seems very plausible : but the question arises from a
m isunderstanding o f a passage near the beginning o f the first
chapter where we are told that theoretical questions are to be
discussed as occasion arises. T h is is commonly understood as a
reference to a fairly distant future discussion— either in undefined
future works outside the E E , or later in the E E itself (for the latter
view see Rowe, 19 7 1, 15). But surely it is a reference to what
immediately follows: though the treatise is a practical one, there
will be occasion for discussing theoretical matters, and right at the
beginning we have to discuss the question ‘What is the good life and
how is it caused ?’ T h e practical question ‘What goal in life should I
pursue?’ is not introduced until the beginning o f chapter 2.
Obviously the two questions are intimately connected : as Aristotle
says, in order to answer the practical question one needs to reflect
on the theoretical one, and in particular to have an eye on the
theoretical distinction between essential constituents and necessary
conditions. But the two questions are not identical. F o r suppose
that the answer to the second part o f the theoretical question was :
happiness is caused by luck or by unmerited divine favour. Or
suppose that the answer to the first part was: happiness is an
absence o f all striving. In either case, it would be absurd to answer
the question ‘H ow should I organize m y life?’ by saying ‘ Pursue
happiness’.
T h e working out o f the relationship between the theoretical and
the practical question is the key to many o f the puzzling features of
the prologue (1. 1-6 ) o f the E E and shows how a number o f passages
which have been thought to be repetitious are not so and need not
be explained as survivals from earlier layers o f composition. One
purpose o f the prologue is to show that in the subsequent discussion
o f happiness only the opinions listed in the first chapter need to be
taken seriously. In the case o f the opinions about the origin o f
happiness this is stated immediately : the five listed ones (provided
that one takes ‘learning’ in an extended sense, so that any action
originating in human thought can be said to be the result o f
learning) exhaust the types o f causes there are ( 1 2 1 4a3o) ; in the case
Happiness in the Aristotelian Ethics 193
o f the opinions about the nature o f happiness it takes longer to
show, since everyone has something to say ( i 2 i 5 bi6). We do not
have to consider the opinions o f the very young, the sick, and the
mad ; nor do we have to consider the opinions o f the multitude
( i 2 i4 b35). T h is may seem surprising, as Aristotle shortly after­
wards devotes quite a long section (1.5) to a consideration o f the
multitude, and since in general, in the words o f G .E .L . Owen,
‘when Aristotle discusses the views o f “ the many and the wise” it is
the second party that gets the shorter shrift’ (1968, 154). B u t what
Aristotle is rejecting are the theoretical opinions o f the many about
the nature and origin o f happiness ; what he treats with respect, as
philosophical data are the everday choices and decisions (actual and
hypothetical) o f the ordinary man.
What we are to consider are the opinions which occur in the
traditional aporiai about happiness, namely those listed in the first
chapter. T h e practical importance o f the question ‘W hat is
happiness?’ is obvious; but Aristotle takes time to show the
importance also o f the other question, ‘What is its origin?’ I f
happiness originates by chance (which here includes grace) or
nature, then it will be beyond most people’s reach and they can do
nothing about it ( i 2 i 5 ai5). Aristotle adds a third theoretical
question— whether happiness consists in a quality o f a person or a
quality o f his action— which he says will be very useful in clearing
up the philosophical problems, and therefore indirectly in answer­
ing the practical question. It will indeed : the distinction between
hexis and energeia, I shall argue, is central to his own solution to the
relationship between the traditional three lives and the nature of
happiness. M oral virtue and wisdom, though different hexeis, are
exercised inseparably in a single energeia, so that they are not
competing but collaborating elements in happiness, and pleasure is
identical with the unimpeded energeia o f the appropriate hexeis : so
that happiness, understood as energeia kat' areten, has the
characteristics o f the three philosophical opinions about the nature
o f happiness.
Chapters 4 and 5 offer two different proofs that only these three
views need taking seriously. Chapter 4 starts from the different lives
or professions, and shows that the only ones worth taking seriously
are those chosen by people o f means, which are the political,
philosophical, and apolaustic life, corresponding to the three
opinions; people’s views on happiness depend on their walks o f life,
IQ4 The Aristotelian Ethics
and Anaxagoras is quoted as an intimation o f Aristotle’ s own
eventual composite view. Chapter 5 starts from the objects o f
voluntary actions. I f we ask what a worthwhile life is, we may go
through the various things which occur in life, and ask which o f
them make life worth living (or, as Aristotle says, worth choosing).
T h ere are some occurrences, e.g. sickness and pain, which make
people want to give up life: clearly these do not make life worth
living. There are the events o f childhood : these cannot be the most
choiceworthy things in life since no one in his right mind would
choose to go back to childhood. In adult life there are events which
bring neither pain nor pleasure in themselves : all the things which
we do choose in the sense that we do them voluntary, but only as
means to another end and not for their own sakes ; clearly these, in
themselves, cannot be what makes life worth living, and no one
would choose to live for the sake o f them however long they were to
continue.
I f life is to be worth living it must surely be for something which
is an end in itself. One such end is pleasure. Consider first the
pleasures o f food and drink and sex. Only a slavish person would
choose to live a life consisting o f these alone without any aesthetic or
intellectual pleasures, for it would be a life no different from a
beast’s. N or does the pleasure o f sleep make life worth living, for
continued sleep would be the life o f a plant. We are left therefore
with three possibilities for what makes life worth living: a
combination o f the pleasures o f food and drink and sex with the
other bodily pleasure o f the senses, which is the apolaustic life o f
Sardanapalus or Sm indyrides ; the life o f virtuous action (the life o f
the real politician, not like the false politicians who are only after
money or power); and the life o f scientific contemplation
(represented, once again, by Anaxagoras).
T h e conclusion o f both chapters 4 and 5 — both the consideration
o f the kinds o f life which people choose to live, and the
consideration o f the types o f answer which they would give to the
question ‘ what makes life worth livin g?’ show that ‘ all connect
happiness with one or other o f three lives, the political, the
philosophical, and the voluptuary’ (1.5, I 2 i6 a27). T h e purpose o f
the prologue up to this point, then, has been to select the opinions
on the nature o f happiness which are to be taken as the starting-
point o f the inquiry. W e have reached the same point by three
different routes. F irst, we have noticed that philosophers regard
Happiness in the Aristotelian Ethics 195
happiness as consisting in one or other or some combination o f the
three goods o f wisdom, virtue, and pleasure; and we have
eliminated the views o f non-philosophers in favour o f these three
which belong to the traditional aporiai on the topic ( i 2 i4 b2Q -
I 2 i 5 a9). Secondly, we have started from an exhaustive division of
the types o f life men lead, and have decided that only three, those
chosen by men o f means, are worth considering, and have noted
that these correspond to the three goods first listed ( i 2 i5 a34).
Th irdly, we have started from the answers to the question ‘What
makes life worth living?’ and reached the conclusion that the
choices made by men and the answers they would give to this
question once again lead to the point that only three candidate
answers need to be taken seriously.
These three answers provide the ‘true but unclear statements’
which are the starting-point for the inquiry. T h e inquiry will be
devoted to sifting out the truth from the confusion in the views : a
start is already made in the final statement o f them in the prologue
( i2 i6 ai 1 —26) when we are invited to distinguish the genuine
politician, who values virtuous actions for their own sake, from the
pseudo-politician who uses them as a means to an end ; when we are
told that it is pleasure itself rather than particular sensual pleasures
that the voluptuary sees as happiness; when the philosophic life is
specified as one o f scientific knowledge rather than o f wisdom. T h e
importance o f these nuances will become clear only after Aristotle
has presented his own account o f happiness. F o r happiness,
according to him is the exercise, as an end in itself, of all the virtues,
moral and intellectual ; and according to him pleasure is the greatest
good, but only because pleasure is identified with the exercise o f all
the virtues.
T h e traditional opinions about the three lives and the cor­
responding goods, besides providing the starting point o f the
inquiry, dictate the plan o f the work, as Aristotle spells out
immediately ( i2 i6 a28 ff.). T o sort out the truth from the falsehood
in the voluptuary’s opinion there is no need to inquire into the
nature o f bodily pleasures and their causes ; all that is obvious
enough. What needs to be investigated is whether these are
constituents o f happiness, or only connected with it in some other
way (perhaps as necessary conditions) ; and whether there are other
pleasures which make the happy life a pleasant life. These inquiries
are postponed : an answer to the questions is in fact to be found in C
196 The Aristotelian Ethics
o f the disputed books : bodily pleasures are necessary to prevent the
obstruction o f virtuous activity ( i i 5 3 bi8) and the pursuit of the
right amount o f them is itself an exercise o f virtue ( 1 1 54a1 5 —18 ; cf.
i 2 3 i ai7 in I I I .2).
T h e inquiry immediately to be undertaken is that into virtue and
wisdom. (I speak o f ‘the inquiry’ ; they cannot be regarded as two
separate inquiries entirely, because the division between virtue and
wisdom presupposed in the traditional trichotomy is something
which Aristotle is going to correct, substituting for it the distinction
between ethical and intellectual virtues.) T h e word ‘ wisdom’, used
in the traditional distinction to denote the goal of a philosophical
career, will be reserved by Aristotle for the practical intellectual
virtue which is the foundation o f the ‘ political life’ but which
also subserves the theoretical contemplation characteristic of the
philosopher. T h e way in which Aristotle concludes his inquiry into
virtue and wisdom in the E E is something which cannot be
described without settling the question whether the E E contains
the disputed books o f the A E . I f the disputed books do belong here,
then we can say that from 1.7 until the end o f disputed book B is
devoted to answering the question raised in I2 i6 a3 8 - b2 ; but we
cannot say without qualification that books I I to IV o f the inclusive
Eudemian Ethics deal with virtue, and book V deals with wisdom,
since that is to state Aristotle’s definitive treatment in terms o f a
provisional one which is merely his traditional starting point.
T h at Aristotle was at this point well aware o f the possibility o f
structuring the relationship between virtue and wisdom in various
ways is immediately apparent : for he goes on to talk o f Socrates,
who like him self disagreed with the dichotomy between virtue and
wisdom, but who unlike him self thought that the division contained
not too few but too many members, since, on the Socratic view,
virtue and wisdom were identical.
Socrates is introduced principally, however, to usher in the
methodological considerations which end the prologue o f the E E .
T h e present inquiry, unlike theoretical sciences, has an essentially
practical purpose : our aim is not merely to know what justice is, but
to be just. On the other hand, it is a genuinely philosophical inquiry,
concerned to discover not just what is to be done but why it is to be
done, operating by means o f argument, but with due respect for
received opinions which contain a deal o f truth however obscurely
and confusedly expressed ( i2 i6 b26 -2 7 ai8).
Happiness in the Aristotelian Ethics 197
After this prologue ( i 2 i 7 ai8) Aristotle takes his start from the
generally accepted, but as yet unclarified, statement that happiness
is the greatest and best o f human goods ( i 2 i 7 a2 i) . ‘H um an’ is first
clarified : we are concerned with goods what are human both in the
sense o f being good fo r men (vs. goods for gods or animals) and
attainable by men. It is of good which are human in both these
senses that happiness is the best ( i 2 i7 a22, a4o).
For the rest o f E E I Aristotle devotes him self to elucidating the
word ‘best’ in the popular account o f happiness. People say that the
best o f all things is auto to agathon, the absolute good, whose
goodness is not dependent on that o f anything else (so that it can be
called ‘the first good’) and which is the cause o f goodness in every
other good thing. Aristotle is prepared to agree that the best o f all
things is the absolute good, but there are three rival opinions as to
what the absolute good itself is. First, there is the Platonic opinion
that it is the Idea o f Good ( i 2 i7 b6 -i6 ), against which Aristotle
argues at length ( i 2 i 7 bi 7 - i 2 i 8 a38). Secondly, there is the view
that the absolute good is a common, univocal universal goodness :
this is briskly disposed of with the remark that no practical
discipline— not to mention the highest practical discipline,
ethics— is concerned to bring about the presence o f so thin and
ubiquitous an attribute in its subject matter ( i2 i8 bi-6 ). Aristotle
concludes by presenting the third possible opinion as his own view.
It is clear, then, that neither the Idea of Good nor the universal, goodness, is
the absolute good that we are looking for: for the one is unchangeable and
unattainable; the other, while it is changeable, is not an object of
attainment. But the wherefore, in the sense of end (to hou heneka has telos)
is the best : it is the cause of what leads up to it and the first of all. It is this,
then, that is the absolute good : the end of everything achievable by man (to
telos tön anthropö(i) praktôn) (i2 i8 b7~i3)
‘B est’, then, in the popular definition o f happiness, is clarified with
reference to the notion of purpose. When we ask for the why and
wherefore o f any human action, we can be told that it is to be done
for the sake o f something else; we can ask in turn for the why and
wherefore o f that something else; when we reach a point where
there is no further answer to the question, we have reached the
‘ wherefore in the sense o f end’ . It is the worthwhileness o f the end
o f an action which makes the actions leading to it themselves worth
while : in this sense an end is the cause of the goodness o f the means
 The Aristotelian Ethics
to it. T h e best o f all human goods, then, would be a good which
appeared at the top end o f every chain o f practical reasoning: that
would be an absolute good, an independent good on which the
goodness o f every other human good depended, as the goodness o f
health-producing drugs or regimes depends on the goodness of
health itself ( i 2 i8 bi7 ~ 2 i) . T h is supreme good is the subject matter
o f the supreme science of ethics and the virtue o f wisdom.
A t this point the first book o f the E E ends. Aristotle has reached
this stage by reflection on the dictum that happiness is the best o f
human goods. N ext, at the beginning o f book II, he throws into the
discussion two further premisses: first, what is common to the
supporters o f the three lives, that happiness is either wisdom or
virtue or pleasure; and secondly an inductively established
definition o f arete— ‘virtue’ in its broadest sense o f ‘excellence’ — as
‘the best state or condition or power o f anything that has a use or a
function (ergon) ( i 2 i 8 b3 5 ; I 2 i 9 a i ) . W ith the help o f these further
assumptions ( i 2 i8 b39) he then proceeds to a close-knit argument
designed to elucidate further the nature o f happiness as the best o f
human goods ( i 2 i8 b3 2 - i9 a38).
On all three o f the traditional views, the best goods are in the
soul: so that best o f all things that we are looking for will be
something in the soul. N ow the things in the soul fall into two
classes: some are states (or powers), others are activities (or
processes). So the best thing in the soul will be either the best state
or the best activity. Consider first the best state. I f a thing has an
ergon (as the soul does) then its best state is aretë. So far, then, we
can conclude that the best thing in the soul is either aretë or the best
activity. So now consider the best activity (b3 2 - a6, “2 9 -3 1). We
have to approach this through a consideration o f the value o f an
ergon. T h e best ergon must be the ergon o f the best state: for in
general an ergon takes the rank o f the state whose output it is. But
which o f these two, the ergon or the state producing it, is better?
Clearly the ergon: for an ergon is an end: and we saw in the final
stage o f E E I that an end is the cause o f goodness in the things
whose end it is. (It is a general Aristotelian principle that if X causes
Y to be P , X must be P -e r than Y : e.g. only if X is hotter than Y
will X heat Y .) 1 This principle about the relationship o f ergon and
hexis applies to all kinds o f ergon, and in particular to the case where
i. L ater incapsulated in the scholastic dictum p r o p t e r q u o d u n u m q u o d q u e t a le , et illu d
m a g is .
Happiness in the Aristotelian Ethics 199
the ergon is an energeia or activity (in the way that the ergon o f sight
is seeing) rather than a product o f activity (such as the health
produced by healing or the houses that are the output o f
architectural skill).
Applying these considerations about ergon to the activities o f the
soul, we can draw the interim conclusion that if the activity o f the
best state o f the soul (i.e. aretë) is an activity which is also an ergon,
rather than the production o f an ergon, then it will be the best thing
in the soul that we are looking for. F o r it will be better than any
other activity (by the theorem that an activity takes its rank from the
state producing it); and it will be better than the best state which
produces it (by the theorem that an ergon is better than the state
whose output it is); and that state was ex hypothesi the best state;
and everything in the soul is either a state or an activity ( i 2 i 9 a9 - i 8 ) .
All that remains, then, to inquire is : what is the activity o f aretë,
and is it its ergon or not? T o discover this, let us ask what is the ergon
o f the soul’s aretë. T h e ergon o f a good X , or o f the goodness o f an A",
is a good ergon-o f-a n -X : thus the ergon o f a shoemaker is a shoe, and
the ergon o f excellence in shoemaking, or o f a good shoemaker, is a
good shoe. Applying this to the case in point, the ergon o f soul is
life,1 so the ergon o f the aretë or excellence o f a soul is a good life
( i 2 i 9 a 2 6 - 8 ) . B u t the life that is meant here is active, waking life,
and this is identical with the activity o f the soul. So the best thing in
the soul which we are looking for as the essence o f happiness turns
out to be the activity o f aretë or the activity o f a good soul
( i 2 i 8 a 3 5 ) . 2 Bu t aretë can be either complete or incomplete,
according to whether we mean aretë as a whole, or a particular
partial aretë; and life, whose active stage is the energeia o f the soul,
likewise can be complete or incomplete (in the sense o f running its
1 . € σ τ ω ψ υ χ ή ς e p y o v τ ο ζ η ν r r o i e î v . It makes nonsense o f the argument to translate this,
as Solomon, Rackham, and D irlm eier do, ‘the work o f the soul is to produce living’. Ισ τ ω
7Γo € Î v is to be construed on the model o f e a n v i S c i V , with t t o l c i v to be taken in the third o f
the senses given in Bonitz s.v., namely ‘ to posit1 : ‘L e t us take leave, then, to posit that the
e r g o n o f soul is life.* T h ere is no need, once this has been understood, to emend the following
manuscript reading τ ο υ B k χ ρ ή σ ι ς κ α ι έ γ ρ η γ ο ρ σ ι ς : it means ‘ . . . is life, and o f life the
active and waking part’ .
2. T h e structure o f the argument is simpler and clearer in the recapitulation (a29~34) : the
best goods are in the soul ; everything in the soul is έ ξ ι ς or c v e p y c i a ; so the best is either the
best £ ξ ι ς or the best è v é p y e t a ; the best I f ι ς as ά ρ € τ ή \ the best i v é p y c t a is the ev ep y eta o f
the best e fts , a n d is b e t t e r t h a n th e b e st e f t? : so happiness, the best good, is the i v é p y c ta o f
the α ρ ε τ ή o f the soul, i.e. the i v i p y e i a o f a good soul. In the full-fledged deployment the
argument is complicated by the introduction o f the notion o f e p y o v .
200 The Aristotelian Ethics
proper course or not). So happiness— which is something perfect
and complete— must finally be defined as ‘activity o f complete life
in accordance with complete virtue’ ( i 2 i9 a39). His definition thus
completed, Aristotle goes on to confirm it by showing that it is in
accord with universally received opinions ( i 2 ig a3 9 b26).
Though the argument is always compressed, and sometimes
sketchy, and though the style lapses at one point into mere jotting
( i 2 i8 a36), the discussion o f happiness in the E E is clearly very
carefully thought out and tightly structured. It is also in perfect
accord with the philosophical method sketched in the prologue.1
When we turn from the first part o f the E E to book i o f the N E
we find a different and far less organized structure. T h e N E starts
not from the notion o f eudaimonia but from that of goodness. The
subject matter o f ethics (politikë) is the good for man, the end of
action for the sake o f which all else is desired (an ultimate end whose
existence is presented in a passage o f notoriously suspect validity,
i094ai8 -2 2 ).2 M ost people would agree that this supreme good is
happiness (however much they must disagree about what precisely
happiness consists in) (i095a 18 -2 6 ); but some people, the
Platonists, argue that the supreme good is an absolute good which is
the cause of all other thing’s goodness. Aristotle then goes on to
expand the general view that the supreme good is happiness by
outlining the three traditional lives (lo g s'314-96*10), and to
criticize the Platonic view by providing a series o f arguments to
show the incoherence o f the Idea o f the Good (i096ai2 - 9 7 a4).
Aristotle begins his own account by saying that the good we are
looking for must be teleion by comparison with other ends— that is,
it must be something sought always for its own sake and never for
the sake o f anything else; and it must be autarkes or self-
sufficient— that is, it must be something which taken on its own
1. In recent years Allan (19 6 1, 3 0 8 - 11) and Row e ( 19 7 1, 15 -2 6 ) have done much to
elucidate the structure o f the E E discussion o f happiness. Though I disagree with Allan on
one or two points and with R ow e on many, my interpretation o f the passage owes much to
theirs.
2 . In my paper ‘H appiness’ (1966) I attempted to defend the N E from the charge o f
committing a quantifier-shift fallacy at this point. Subsequent discussions with a number o f
people (in particular M r. J . O. U rm son and M rs. L . Brown) have made me less confident that
the N E can be defended along the lines I suggested. B u t whatever the case with the N E , it is
clear that the E E is guiltless on this point. F or i f Aristotle is guilty o f the fallacy he is charged
with, he must think it is a logical truth that, if there is choice at all, there is a single supreme
end. B u t E E I, i 2 i 4 b6 ff. admits it as a possibility— though a mark o f folly— not to organize
one’s life in view o f an end.
Happiness in the Aristotelian Ethics 201
makes life worth while and lacking in nothing. Eudaimonia has both
these properties (i0 9 7ai 5 h2 i). What, then is eudaimonia? T o
elucidate this, we must consider the ergon o f man. M an must have
an ergon, because particular types o f men (e.g. sculptors) do, and
parts and organs o f human beings do. What is it? N ot life, not at
least the life o f growth and nourishment, for this is shared by plants,
nor the life o f the senses, for this is shared by animals. It must be a
life o f reason concerned with action: the activity o f soul in
accordance with reason. So the good o f man will be his good ergon :
the activity o f soul in accordance with virtue, and if there are several
virtues, in accordance with the best and most perfect (teleiotatën)
virtue (i09 7b22-98 ai8).
Though on a superficial reading the accounts o f happiness in the
first books o f the N E and E E appear rather similar, closer
investigation reveals substantial differences in structure, in argu­
ment, and in doctrine. E ven when the same technical terms
appear— such as ergon, teleion, auto to agathon— they are put to very
different use. And the concluding definitions o f happiness
enunciate radically different theories o f its nature.
T h e structure o f the N E treatment o f happiness is much less neat
than that o f the E E . T h e flow o f the argument is frequently broken
by scattered and repetitive methodological remarks (i094bn -
95ai4 ; i095a3 0 -bi4 ; i098a2 6 -b9). T h e connection between the
discussion o f the three lives and the refutation o f the Idea o f the
Good is not brought out. T h e opinions o f others on the topic,
instead o f being worked into the argument as they are in the E E , are
presented in an enormous album o f frequently disjointed remarks
in chapters 8 - 12 . T h ere are a number of inconclusive passages
where the argument comes back to its starting-point and a fresh
beginning has to be made : sometimes this is even remarked upon in
the text (‘T h e argument has come around to the point where we
were before’, 1097*23).
Rowe, who has detailed the incongruities o f the N E text,1
considers that they provide evidence that the E E represents an
earlier stage o f Aristotle’s development. H aving written the E E , he
suggests, Aristotle changed his mind on a major point; he wrote the
N E as a revision o f the E E , but the revision was only partially
i . H e even somewhat exaggerates them. T here is no need, for instance, to attribute to the
N E the suggestion that ‘the Idea o f the G ood’ might be an answer to the question ‘ what is
happiness?’ (Rowe, 19 7 1, 27). In this section Row e acknowledges his debt to K ap p (19 12 ).
202 The Aristotelian Ethics
complete, leaving ugly traces of the original pattern. It seems a little
curious to treat a more careful and polished treatment o f a topic as
prima-facie evidence o f an earlier stage o f composition: it would be
at least as plausible to suggest that the E E represents a definitive
ordering of material which had been only partially moulded into
shape at the time o f writing o f the N E . T h e final scrapbook o f
received opinions, R ow e thinks, provides evidence against those
who regard the E E as post-Aristotelian. ‘ I f E E is spurious, and was
constructed on the model o f E N , then its Peripatetic author shows a
remarkable ingenuity; he has evidently abstracted certain parts o f
the list presented by E N , worked them into the structure of I, i -7,
and still managed to construct a highly coordinated series o f
argum ents.’ T h e rewriting would certainly call for ingenuity. But a
suggestion that the E E I is later than N E need not be a suggestion
that E E I is non-Aristotelian ; and it would surely be rash to assume
that the ingenuity involved in turning N E 1 into E E I would be
beyond the powers o f an Aristotle.
T h e principal difference in structure between the E E and the N E
is that whereas the E E begins with the concept of happiness, and
goes on to discuss the nature o f goodness in the course o f
commenting on the terms o f a traditional definition o f happiness,
the N E begins with the supreme good as the subject matter o f
Ethics, and goes on to consider happiness as the most popular
answer to the question ‘what is the supreme good?’ B u t the
differences between the accounts, as has been said, go deeper than
mere presentation. In the E E , unlike the N E , Aristotle is prepared
to accept the notion o f an absolute good (auto to agathon)} In both
ethics in the construction o f the definition o f happiness use is made
o f the concept o f ergon·, but the ergon in question in the E E is the
ergon o f the soul, in the N E the ergon o f man. T he E E lays much
more stress than the N E on the causal role o f the supreme good : the
final cause that is the first cause in the order o f goodness-causation,
the cause o f the goodness o f other goods.
i . D . J . A llan comments well on the significance o f this. ‘[In the N E\ he treats as
equivalent the assertion o f a Platonic Idea and the use o f the expression “ absolute” . T he
search for the absolute good is consequently a wild goose chase. T h e author o f the E E joins
with enthusiasm in the search, and is in full agreement about the description o f the bird that
is being hunted. It would seem to me that the key to this difference in the style o f criticism is
not that the Eudem ian version shows closer sym pathy with Platonism . . . but that the
passage is integral to the discourse on method which immediately precedes it. T h e Platonist
is to be disabused o f his errors by reasoning from the premisses which he admits, and which
are true, but not in the way in which he supposes’ (19 6 1, 309).
Happiness in the Aristotelian Ethics 203
The question is often raised whether the E E is more Platonic
than the N E . M ost o f those prepared to give an answer give one in
the affirmative : but scholars in recent years have come more and
more to realize that the question has too many facets to be given a
simple yes-no rep ly.1 In the present context of the discussions o f
happiness we can say that the N E resembles Platonic ethical theory
more than the E E does in making the, good, rather than happiness,
the primary ethical concept; on the other hand the ergon argument
in the E E is closer than the N E one is to the parallel section in the
Republic (352 e-354 a) in appealing to the notion o f the ergon o f the
soul rather than to the philosophically more suspect notion o f the
ergon o f man. T h e criticism o f the Idea o f the Good is more
thoroughgoing, and conducted from a less sympathetic standpoint,
in \ h e E E than in the N E : in the words o fD . J . Allan ‘ its manner o f
introducing the discussion, placed side by side with the
Nicomachean equivalent, can only seem cool and abrupt’ .2 T h e
discussion in the E E , as G . Verbeke has pointed out, is more closely
linked than that in the N E to general Aristotelian theories o f
causation, potentiality, and actuality.3
T h e most important difference between the accounts o f
happiness in the N E 1 and in the E E I - I I is that the former
identifies happiness with a single dominant end, the activity o f the
highest virtue; whereas the latter views happiness as an inclusive
end, the activity o f all the virtues o f the rational soul in the broadest
sense. T h e N E says that the good for man is ‘activity o f soul in
accordance with virtue, and if there is more than one virtue, in
accordance with the best and most perfect (teleia/ ( iog8ai6). In
the last chapter o f the book we are told that there is indeed more
than one virtue— there are, for instance, moral and intellectual
virtues— and hence we conclude, unconditionally, that happiness is
activity in accordance with the best and most perfect o f these many
virtues. (Compare ι ι ο 2 35~7 n o 3 a2 - io .) In the E E we are told
that happiness is activity ‘o f complete life in accordance with
complete (teleia) virtue’ ; and the word ‘complete’ has just been
unambiguously glossed when Aristotle has said that ‘life is either
complete or incomplete, and so also virtue— one being whole
virtue, another a part ( i2 ig a35-9). So that when in the E E Aristotle
1. T his has been so especially since G . E . L . O wen’s classic lecture ‘T h e Platonism o f
Aristotle’ (1965).
2. Op. cit. 309.
3. Verbeke, 19 5 1, 83.
204 The Aristotelian Ethics
goes on to distinguish the parts o f the soul and the virtues that
correspond to them, and to list examples o f moral and intellectual
virtues, we know that their activities are all supposed to be part o f
happiness. The virtue o f the soul that figures in the definition o f the
end o f m an— we learn at 1220*4— is a virtue which is constituted by
the several virtues o f the different parts o f the soul.
T h e contrast I have just drawn has long been familiar to
scholars.1 It has recently been questioned by Professor J . L . Ackrill,
who argues that it is a mistake to regard Aristotle’s idea of
eudaimonia in N E 1 as the idea of a ‘dominant’ end. T h e concluding
clause o f the N E definition ‘if there are more than one virtue, in
accordance with the best and most complete’ should not be
understood as referring to a particular supreme virtue, one o f many,
but rather to total and comprehensive virtue, the sum o f all virtues
(Ackrill, 1974, 15 - 18 ).
Ingenious though it is, this interpretation does not seem to me
tenable. In the previous chapter of the N E Aristotle has already
made clear that he does not regard happiness as an inclusive end,
but as a dominant one, in the passage in which he says that when we
are looking for happiness we are looking for something which, when
not added to anything else, is most choiceworthy— clearly, if it is so
added it is more choiceworthy with even the least additional good.
I f happiness were meant as an inclusive end, as the sum total of
goods sought for their own sake, it would be absurd to speak o f
goods additional to happiness.2 When Aristotle speaks o f ‘ the best
1. e.g. Verbeke, 19 5 1, 84: ‘L e point de vue adopté dans Y Ethique Eudémienne est
indéniablement plus large que la position prise dans Y Éthique à Nicomaque: dans ce dernier
ouvrage Aristote cherche l’activité propre de l’ homme, celle qui le distingue de tous les autres
vivants, et pour y arriver, il procède par élimination progressive pour ne retenir que l’activité
intellectuelle ou la contemplation; dans YEthique Eudémienne le Stagirite affirme simplement
la supériorité des biens psychiques sur les biens extérieurs et il en conclut qua la perfection de
l ’homme se trouve dans l’activité de l’âme : mais il ne dit pas que cette activité serait purement
intellectuelle ou contemplative: on ne voit pas qu’ il exclut d ’autres activités psychiques.’
2. T h is interpretation o f i097b was adopted in my paper ‘ Happiness’ (1966). It is
criticized by Ackrill, p. 12, who writes ‘ It is indeed only if one is willing, with K en ny, to treat
“ happiness” as a fair translation o f the word eudaimonia that one can feel the slightest
temptation to take the self-sufficiency passage as he does.’ I agree that ‘happiness’ is an
im perfect translation o f eudaimonia, which means something more like ‘ worthwhile way o f
life’ ; I use it in preference to such a periphrasis on grounds o f tradition and brevity. B u t my
interpretation o f this passage is a traditional one, which was adopted by Eustratius, Albertus
M agnus, and Thom as Aquinas, who were comparatively free from the temptations presented
by the distracting overtones o f the English word. T hus Eustratius (C A G xx.65) γ ίν ε τ α ι γ α ρ
to TTρ ο σ τ ιθ ε μ ε ν ο ν υ π έ ρ ο χ η τ ο ύ α γ α θ ό ν κ α ί α ύ ζ η σ ις , ώ ς μ ε ίζ ο ν α ε κ τ ή ς π ρ ο σ θ ή κ η ς
τ η ν ε υ δ α ιμ ο ν ία ν γ ιν ο μ ε ν η ν α ίρ ε τ α )τ ε ρ α ν ε ΐν α ι α υ τ ή ν μ e r ’ α υτού, η α ιρ ε τ ή ε σ τ ιν κ α ι
Happiness in the Aristotelian Ethics 205
and most teleia virtue’ the word ‘teleia' admittedly has several
senses, one o f which is ‘complete’ ; 1 but it is hard to take the word
for ‘best’ to mean ‘comprehensive o f m any’ rather than ‘better than
the rest’ . ‘A r e të ', like the English word ‘ virtue’ , can be used as a
mass-noun (as in ‘a man o f great virtue’ ) or as a count-noun (as in
‘a man of many virtues’ ); but on A ckrill’s view Aristotle is made to
switch from the mass-noun to the count-noun use and back again to
the mass-noun use within a space o f ten words.
T h e traditional view sees the clause ‘ if there are more than one
virtue, in accordance with the best and most complete’, as keeping
open a place for the eventual doctrine o f N E 10 that happiness is the
activity o f the supreme virtue o f sophia. Even Ackrill does not try to
deny that in N E 10 a ‘dominant’ view o f happiness is adopted. No
doubt we are wise not to take it for granted that N E i and N E 10
were written in a single stint : but there is evidence that when N E i
was written Aristotle was thinking o f the topic o f N E 10 (1096*5)
and that when he wrote N E 10 he had in mind N E 1 ( 117 7 * 11 ) . ‘ I f
happiness is activity in accordance with virtue, it is reasonable that
it should be activity in accordance with the most excellent virtue :
and this will be the one o f the best part o f us.’ T h is is either nous or
something like it, so ‘ the activity o f this in accordance with its
proper virtue will be perfect (teleia) happiness’ . I f teleia here means
‘final’ rather than complete, that suggests that it meant the same in
the N E i passage to which reference is m ade; if it means ‘ complete’
then again it implies that there is nothing else in happiness other
than the contemplative activity o f nous. Aristotle then goes on to
show that theoretic contemplation possesses all the qualities which,
according to Book 1, were, in popular opinion and in truth,
properties o f happiness. T h u s: it is the highest activity, most
continuous and durable (117 7 * 2 1 ; cf. n o o bi 1 - 1 7 ) ; the pleasantest
(*23 ff.; cf. 1099*7 ff·); most self-sufficient (*27; cf. io97b6 ff.); it is
loved for its own sake ( ii 7 7 bi ff.) and therefore teleion in the sense
explained at 1097*30 if. I f Aristotle underwent a spectacular change
in his view o f happiness between Book 1 and Book 10, as Ackrill’s
view seems to imply, he wrote book 10 in such a way as to cover up
the change entirely. But even if Ackrill is right about N E 1, the
Λ7ιό ν τ ο ς α υτο ύ. T h e traditional interpretation criticized by Ackrill is ably vindicated by
Clark (1975, 15 3-4 ).
I . In its immediately succeeding use, the word means ‘ complete’ (i09 8 ai8 ); the use
previously explained was ‘always chosen for itself and never for the sake o f something else’
(ί097“3°).
206 The Aristotelian Ethics
contrast between the E E and the N E remains: for at no point does
the E E adopt the view o f happiness as identical with theöria which
Ackrill agrees is presented in N E 10.
T h e complete virtue, whose exercise in a complete life
constitutes eudaimonia according to E E I, is the sum total o f the
virtues discussed in the central books o f the E E , and is treated again
as a unified whole in the first part o f the final chapter o f E E V I I I ,
under the description kalokagathia (i2 4 8 b8 -i6 ). Clearly, anyone
who deserves the description kalos kagathos must have all the
individual virtues, just as a body can only be healthy if all, or at least
the main, parts o f it are healthy. Aristotle then develops the theme
that natural goods, like health, and wealth are beneficial only to a
good man : to a bad man they may be positively harmful. But among
good men we must distinguish between those o f a utilitarian cast,
who pursue virtue for the sake of non-moral goods, and the kaloi
kagathoi who pursue the natural goods only for the sake o f the
virtuous actions for which they are useful. T h e two marks o f the
kalos kagathos— that he has all the virtues, and that he pursues the
virtues for their own sake and other things for the sake o f them— are
not really distinct : already in the case o f the particular virtues we
were told that genuine virtue involves the choice o f to kalon
(i2 2 9 a2 if. ; i2 3 0 a29~32). T h e new character introduced here is not
the kalos kagathos, but the mere agathos, the ‘Laconian’ character
who pursues virtue as a means to an end : he— like the continent and
incontinent persons discussed in A E C and from time to time in the
E E — occupies a position intermediate between the virtuous and
vicious characters who are presented in an exaggeratedly well-
defined trichotomy by the doctrine o f the mean. T h is last
distinction between the agathos and the kalos kagathos sets in final
clarity the ideal o f the happy life as one o f complete virtue
(i2 4 9 ai8).
Though we are told that kalokagathia is complete virtue, we are
not at this point explicitly reminded that happiness is the exercise o f
complete virtue. Bu t Aristotle goes on immediately :
Pleasure has already been discussed: what kind of thing it is, and in what
sense it is a good; and how things which are pleasant simpliciter are fine
simpliciter, and things which are good simpliciter are also pleasant. But
there cannot be pleasure except in action: and so the truly happy man will
also have the most plesant life. This human quest is not an idle one.
(i249ai8 -2 i)
T h e first o f the backward references in this passage is intelligible
Happiness in the A ristotelian Ethics 207
only if A E C , or something like it, preceded in the original
Eudemian Ethics. T h e next points are both made in the book on
Friendship: Things which are pleasant simpliciter are fine simpli­
citer because that most rightly deserves t0 be called pleasant which
the wise man calls pleasant; and to him ;t is good and fine things
that are pleasant (i2 3 s b 3 6 -3 6 a7). T h in gs which are good simpliciter
are also pleasant, because it is naturai goods which are good
simpliciter and these are naturally pleasant: this natural pleasure is
nature’s road to virtue (i2 3 7 a5-9).
Thus for the ideally virtuous man the concepts ‘ agathon’, ‘hëdu\
‘ kalon’ coincide in their application. I f what is pieasant for a man
differs from what is good for him, then he is not yet perfectly good
but incontinent ( i 2 37a8 -9 ); if what js good for him does not
coincide with what is fine for him, then he is not yet kalos kagathos
but only agathos: for the kalos kagathos the natural goods o f health
and wealth and power are not only beneficial but fine, since they
subserve his virtuous activity (12 4 9 ^ ). So for him ^ goodness,
fineness, and pleasantness coincide. 'J'he bringing about o f this
coincidence is the task o f ethics (1247*3). But whereas something
can be kalon or agathon whether it is a hexis or an energeia ( 1 248b35~
7, b23-4 ) it is only an energeia or praxis that can be pleasant. So it is
in the fine activities o f the good man that the highest pleasure is to
be found and that pleasure, goodness, and fineness meet. But the
fine activities o f the good man are the activities o f perfect virtue
with which happiness was identified in book one. So the treatise has
reached completion: Aristotle has carried out the promise o f his
first paragraph to show that happiness combined the three
superlatives— finest, best, and pleasantest— o f the Deliac in­
scription ( i 2 i4 a7-8).
T h e quotation just given is not, ho\vever, the last paragraph o f
the E E . It is followed by the passage already discussed concerning
the horos o f kalokagathia. F o r at 1249*2 j Aristotle has yet to finish
showing the truth contained in the true but unclarified premiss
that happiness belongs to one of the three traditional lines.1 T h e
discussion o f virtue and o f pleasure and o f kalokagathia has shown
that what the seekers o f pleasure and the pursuers o f virtue seek is to
be found in Aristotelian ideal. But what 0f the Anaxagorean life o f
I. I think Jaege r was well inspired to regard A ris to tle ideal as a synthesis o f the three lives
(235-8). Joh n Cooper says ‘ T h e study ot the three lives no mor£ provides the structure o f the
Eudemian Ethics than it does that o f the Nicomachean ; Jaeger was at straws here with
which to prop up his theory o f Aristotle’s d e v e lo p m ^ - (Cooperj , 45) T h is seems tQ
me imperceptive.
2o 8 The Aristotelian Ethics
philosophic contemplation? Are the values o f that life too
incorporated in the inclusive programme o f eudaimonia ?
L et us continue to leave aside, for the moment, the A E , and try to
answer the question from the E E context alone. I f kalokagathia is a
synthesis o f the virtues o f parts o f the soul in the way that health is a
synthesis o f the health o f various parts o f the body, then it must
include the virtues o f the intellectual parts o f the soul as well as o f
the passional part. Wisdom is alluded to as a necessary part o f
kalokagathia at i2 4 8 b3 i : but is the healthy functioning o f the
speculative part o f the soul part o f the complete virtue which is
happiness? T h e final section about the criterion o f kalokagathia is
there to tell us not only that is it part o f happiness itself but that it
sets the standard to which the activities o f the other virtues must
conform if they are to remain within the realm o f virtue and
happiness. Not, o f course, that reference to the contemplation and
service o f G od settles how much o f the other virtues one should
have ; that would be absurd, since virtue is something o f which
there cannot be too much. But virtuous action consists in executing
choices about the right amount o f other things— the passions and
external goods which are the field o f operation o f the moral
virtues— and it is reference to the horos which is necessary to decide
what is the right amount o f such things. T h is does not conflict with
the doctrine that virtue must be chosen for its own sake, because it is
kalon; the horos provides the measure, not the motive, o f virtuous
action. Brave, temperate, and generous living is an end in itself,
chosen for its own sake, and as part o f happiness in its own right.
B u t what particular behaviour in concrete circumstances counts as
virtuous living cannot be settled without consideration o f the
contemplation and service o f God. Aristotle’s favourite analogy
between spiritual and physical health may help us here. T h e
healthy man has a good appetite for the right sort and the right
amount o f food. T h e right sort and the right amount o f food means :
the sort and amount o f food that conduces to health. But the healthy
man, unlike the hypochondriac, does not eat his food because it
conduces to health; he eats it because he has a good appetite and he
enjoys his food.
T h e contrast, then, between the inclusive, organic view o f
happiness in the E E , and the dominant, intellectualist one o f the
N E is clear and profound, Instead o f a single life offering us all the
values sought by the promoters o f the three traditional lives, the
concluding section o f the N E offers us a first-class, perfect
Happiness in the Aristotelian Ethics 209
happiness consisting o f the exercise o f sophia, and an alternative,
second-class career consisting in the exercise o f phronësis and the
moral virtu es.1 N ow which o f these conceptions o f happiness
matches better the remarks about happiness in the disputed books ?
In A E A we are told that ‘we call those acts just that tend to
produce and preserve happiness and its parts for the political
community’ ( i i2 9 bi9). T h is seems to presuppose the E E view of
happiness as a whole with parts, rather than the N E view o f
happiness as a single dominant activity— and one which at its best
can be pursued outside a community ( i i 7 7 a33). In B we are told
that wisdom studies ‘the things which will make a man happy’ : the
plural perhaps fits the E E conception better than the N E one, but it
is hard to be certain whether the G reek expression (ex hôn estai
euiaimôn anthröpos) refers to means towards happiness or actual
constituents o f happiness (cf. E E I 2 i4 bi 2 f f . and i 2 i 6 b2 i). T h e
crucial passage in B for our concerns is the following answer to the
objection that learning and wisdom are superfluous because
unproductive: ‘T h ey are indeed productive: not like medical skill
in relation to health, but like health itself ; it is thus that learning is
productive o f happiness : for being a part o f virtue entire by being
possessed and being operative it is productive o f happiness’
(114 3 * 3 -6 ).2 T h e passage is difficult enough to deserve five pages of
discussion in Gauthier and J o lif s commentary. T h e difficulties are
these. (1) T h e passage begins with a plural verb (‘ they are
productive’ ) and then we are offered a single subject (‘ learning’) :
how much o f the passage is about both learning and wisdom, and
1. I agree with C ooper ( 1 9 7 5 ,1 57 - ^°) against G au th ier-Jo lif (19 59 ,8 9 3-6 ) that this is the
only possible interpretation o f i i 7 6 b2 6 -a22, and Aristotle’s doctrine cannot be taken to be
that there is a single happy life consisting primarily o f the exercise o f learning and secondarily
o f the exercise o f the other virtues. Cooper shows convincingly that this is not a possible
reading o f the crucial sentences : κ α ι τ ω ά ν θ ρ ώ π ω δ η ό κ α τ ά τ ο ν νο υ ν β ίο ς , ε ϊπ ε ρ τ ο ύ τ ο
μ ά λ ισ τ α ά ν θ ρ ω π ο ς , ο δ τ ο ς αρα κ α ι ε ύ δ α ιμ ο ν ε σ τ α τ ο ς . δ ε υ τ ε ρ ω ς δ ’ ό κ α τ α τ η ν ά λ λ η ν
α ρ ε τ ή ν . But Cooper’s own view, that the second-rate life o f N E 10 is the same as the mixed
life o f £ £ , seems as untenable as Gauthier’s. T h e only positive argument he offers for it as
that it would be strange i f the mixed life championed in the E E were not so much as
mentioned in the N E . T h at is so, only if the N E is later than the E E . And for this we need an
independent argum ent— which, o f course, C ooper believes he has : the psychology o f N E 10,
he says, is closer to the de Anim a, which he assumes is late (pp. 175 ff.). D evereux (19 77, 8),
who shares Cooper’s faith in the lateness o f N E 10 , has shown how little support the N E text
gives to Cooper’s interpretation of the two types o f happiness.
2. ε π ε ιτ α κ α ι π ο ιο ΰ σ ι μ ε ν , ο ύ χ ώ ς ή ια τ ρ ικ ή δ έ ύ γ ίε ια ν , άλλ’ ώ ? η ύ γ ίε ια , ο ύ τ ω ς ή
σ ο φ ία ε υ δ α ιμ ο ν ία ν , μ έ ρ ο ς γ ά ρ ο ΰ σ α τ ή ς δ λ η ς α ρ ε τ ή ς τ ω ε χ ε σ θ α ι π ο ι ε ί κ α ι τ ω
ε ν ε ρ γ ε ΐν ε υ δ α ιμ ο ν ία ν . L ik e von Fragstein ( ΐ 974 > 2 47 ) 1 prefer to read ε υ δ α ιμ ο ν ία ν with
Pap. O xyrhync. 2402, L b and Ob and Eustratius; but i f one reads ε ύ δ α ίμ ο ν α with M b
(‘ makes a man happy’ ) the sense is much the same.
210 The Aristotelian Ethics
how much about learning alone? (2) W hat is it that health is
supposed to be productive of? (3) W hat is meant by saying that
learning, as part o f virtue entire, is productive o f happiness ?
I answer the difficulties as follows. (1) It is true o f both wisdom
and learning that they are productive o f happiness in the way that
health is productive : but it is true only o f learning, and not also of
wisdom, that it is not productive in the way that medical skill is. For
both learning and wisdom are constituents of happiness; but
wisdom contributes to happiness not only in being a constituent of
it, but in prescribing for the sake o f another constituent, namely
learn in g(i i4 5 a9). (2) What health produces is neither happiness (as
some commentators have thought in contradiction to E E i 2 i9 b20-
2) nor health (as others have thought, in spite o f the relation
between health and health being m anifestly one o f tautologous
equivalence rather than efficient causality) but the activities o f a
healthy body. Health is a hexis which produces, causally, the
energeiai characteristic o f it. Ross caught the sense exactly in his
note on his translation: ‘ I.e., as health, as an inner state, produces
the activities which we know as constituting health.’ It is in this way
that wisdom and learning produce happiness : they are the hexeis
whose activities constitute happiness, as is spelt out in the final part
o f the passage. (3) L earnin g is only a part o f virtue entire; it would
be pointless o f Aristotle to remind us o f this here if he intended to
say, as many commentators make him say, that it is the whole of
happiness. It is because it is part o f virtue entire, and because the
activity o f virtue entire (the ‘complete virtue’ o f E E I) is happiness,
that the possession plus exercise o f learning is productive of
happiness: learning is one o f the hexeis which causally produce
energeiai which in their totality are happiness.
T h u s understood— and this way o f understanding it, I would
contend, is the natural way o f understanding it even on grounds o f
its internal structure alone1 — the passage is in perfect accord with
the doctrine o f E E I and E E V I I I , but in flagrant contradiction with
that o f N E 10 and more muffled contradiction with that o f N E 1. It
i . N o one has taken more pains to square this passage with N E 10 than Gauthier and Jo lif,
and even they adm it that the most obvious way o f taking the passage ‘bouleverse tout l’édifice
de la morale aristotélicienne en faisant entrer la sagesse pratique,— et avec elle la vertu morale
qui en est inséparable,— dans la constitution même de la fin suprêm e, qui partout ailleurs[ ! ]
consiste dans la seule contemplation’ (1959, 543)· T h eir own interpretation, which on the
basis o f comparison with fr. 6 W o f the Protrepticus makes Aristotle say that learning is an
improper part (i.e. the whole) o f happiness, is rightly found unwarranted and fanciful by
Cooper (19 75, 1 1 2 - 1 3 ) .
Happiness in the A ristotelian Ethics 211
clearly envisages an inclusive, not a dominant conception of
happiness.
There remains one further passage on happiness in the disputed
books. It is a long passage in the discussion o f pleasure in A E C,
which I shall first quote and divide into numbered sections to
facilitate further discussion.
( i ) I f certain pleasures are bad, that does not prevent the chief good from
being some pleasure, just as the chief good may be some form of
knowledge though certain kinds o f knowledge are bad. Perhaps it is even
necessary, if each disposition has unimpeded activities, that, whether the
activity (if unimpeded) of all of our dispositions or that of some one of
them is happiness, this should be the thing most worthy of our choice; and
this activity is pleasure. Thus the chief good would be some pleasure,
though most pleasures might perhaps be bad without qualification. (2)
And for this reason all men think that the happy life is pleasant and weave
pleasure into their ideal of happiness—and reasonably too; for no activity
is perfect when it is impeded, and happiness is a perfect thing; this is why
the happy man needs the goods of the body and the external goods, i.e.
those of fortune, viz. in order that he may not be impeded in these ways.
(3) Those who say that the victim on the rack or the man who falls into
great misfortunes is happy if he is good are, whether they mean to or not,
talking nonsense. (4) Now because we need fortune as well as other things,
some people think good fortune the same thing as happiness ; but it is not
that, for even good fortune itself when in excess is an impediment, and
perhaps should then be no longer called good fortune; for its limit is fixed
by reference to happiness. (ii5 3 b7~24, trs. Ross)
(1) In the first section o f this passage, it is very striking that
Aristotle appears to leave open the question whether virtue is the
activity o f all the virtues (for the ‘ dispositions’ he has in mind are
clearly virtues) or o f some one o f them. T h is is the only place in the
E E or the A E where this is done, and it cannot but remind the
reader o f the definition o f the human good in N E i as ‘ activity o f
soul in accordance with virtue, and if there are more than one
virtue, in accordance with the best and most complete’ . I f any
passage o f the A E has the ring o f a Nicomachean interpolation, it is
surely this one. Y et the style, in other respects, is not more
Nicomachean than Eudem ean; and it would be ironic to regard this
passage as evidence o f a Nicomachean origin for the disputed books
when it was the very same passage which— because o f its
identification o f the supreme good with a pleasure— seemed to
Aspasius (C A G χ ίχ .15 1) to provide evidence that the context was
212 The Aristotelian Ethics
not by Aristotle but by Eudemus. In fact a closer reading o f the
passage shows that there is a crucial difference from the
Nicomachean parallel. In N E i the disjunctive definition of
happiness appeared in a conclusion o f Aristotle’s own argument,
and had as its purpose to keep open an unobvious alternative he was
later to argue to be the correct one. Here, it appears as an ad
hominem argument against those who refuse to identify pleasure
with the supreme good— whoever they may be, whether Plato,
Speusippus, or other Platonists perhaps even including an earlier
Aristotle. T h e disjunction can very well be understood as intended
to allow, dato non concesso, a premiss which Aristotle rejects:
‘W hether happiness is (as argued in the present work) the
unimpeded activity o f all human virtues or (as our opponents
allege) the unimpeded activity o f one particular human virtue,
either way happiness is unimpeded activity and therefore pleasure.’
Certainly this passage allows assumptions to which Aristotle, on his
own account, would never have assented at the time when he wrote
either the N E or the E E : as that the majority o f pleasure are bad
haplös.
(2) T h at the happy life is rightly thought to be pleasant is
common ground in the N E ( ιιη η &22) and the E E (i249 a2o). Both
treatises also allow that happiness m ay be impeded by lack of
external goods (N E 1 i7 8 b33 f. ; E E i2 4 g b2o) ; but o f course the way
in which lack o f money ‘ impedes’ liberality is more intimate than
that in which it impedes contemplation. T h e present passage would
fit either type o f prevention.
(3) Th at a virtuous man is happy in the midst o f suffering and
misfortune is a view which is criticized both in the N E (1096*2) and
by implication in the E E ( i 2 i4 bi4) : it was, as Burnet and Gauthier
showed, a view dear to Plato and the author of the Protrepticus
(G auth ier-Jo lif, 1959, 32).
(4) In both ethics the mistaken identification o f happiness with
good fortune is mentioned ( N E 1, 2099b8; E E 1 2 1 5 * 1 2 ff.); in both
we are warned that excessive good fortune may make virtue and
happiness difficult (N E i i 2 4 a30 ff. ; E E i24 9b2o). T h e language of
the warning in the E E , like the one here, makes use o f the technical
term horos.
W e have now considered all the references to happiness in the
A E . W e can see that, despite the difficulties in interpretation o f two
o f the most crucial passages, they add up, on balance, to a theory of
Happiness in the Aristotelian Ethics 213
happiness and its parts and its relation to virtue and fortune which
is closer in both substance and terminology to that o f the E E than to
that o f the N E . T h is impression is confirmed when we consider that
the Eudemian Ethics itself makes a m uch more coherent whole when
the A E is included in it than if it is omitted. T h e intellectual virtues
are then dealt with at a length proportionate to that devoted to the
moral virtues; the relation of the Aristotelian ideal o f happiness to
the Anaxagorean contemplative life is spelt out by the discussion o f
sophia and its relation to eudaimonia; the promise in E E I to discuss
the value o f pleasure and its relation to human goodness is kept in
A E C. I f the A E belongs with the E E then there is consistently
presented a view which combines an inclusive conception o f
happiness as the exercise of all the virtues with an insistence on the
primacy and supreme value o f philosophical contemplation. On this
view, wisdom and the moral virtues find expression in actions
which are both parts o f happiness in themselves and also promote
the contemplation which is the supreme element in happiness. Acts
o f the moral virtues contribute to happiness in the way that good
breathing contributes to good singing and not in the way in which
(say) being early to bed and early to rise may do so. T h ey are things
which a happy man does qua happy, as the good singer’s good
breathing is an inseparable and essential part o f his good singing.
Finally, the summary in E E V I I I .3 o f the whole course o f the
discussion fits admirably a Eudemian Ethics which contains the A E
and displays to the careful reader the elegant structure o f the work
as a whole.1
T h e -contrast that we have drawn between the N E view of
happiness and the E E view o f happiness, and the comparison we
have made between each and the observations on happiness in the
A E , once more confirms the hypothesis that the disputed books
originally belong with the E E . In conclusion, we may note that our
discussion also enables us to throw light on a crux in the E E . In the
discussion (1 i2 0 b3 5 - 2 i ai0) o f the doctrine o f the mean in E E 2
Aristotle presents a list o f triads o f which the final member is
‘cunning-stupidity-w isdom ’ . As many o f the triads in the list are
i. T h is is well brought out by D irlmeier in his excellent note on 12 4 9 *17 : ‘V I I I 3 ist
überhaupt ein K ap . de Rückbeziehungen: am Anfang der Verweis au f E E I I I ; sodann ist die
Existenz von E E Iv ( = N E 5, verteilende Gerechtigkeit) vorausgesetzt, ferner die von E E V
( = N E 6, Phronësis) une eben auch E E V I ( = Ε Ν η , L u st).’ T his note is the more impressive
because Dirlm eier believes that the original E E m iddle books are lost.
214 The Aristotelian Ethics
names o f moral virtues and vices, this passage has seemed to many
to be in such flagrant contradition with A E B that the paradox could
be resolved only by excising the offending triad, or by denying the
E E to Aristotle, or at the very least by denying B to the E E . N ow in
fact the triad contains many entries which are not moral virtues and
vices (the most obvious being ‘gain -loss-ju st’). Aristotle’s aim is
not to show that wisdom is a moral virtue, but to show that here too
there can be an application of the mean to the case o f wisdom.
Aristotle, at 12 2 1* 3 6 shows how it applies in the case o f the choice of
means. But the contrast between the inclusive and the dominant
notion o f happiness shows us how the doctrine o f the mean applies
also to the choice o f ends. T h e parameter to which the mean is to be
applied is given by the question: how many values does a man
pursue? T h e stupid man pursues too many, he follows one little
project after another: he is guilty o f the folly we were warned
against in E E I, i 2 i 4 b9. T h e wise man pursues the right number:
he follows the Aristotelian ideal which combines the values o f the
contemplative, political, and hedonic lives. T h e cunning man
pursues a single dominant goal and is ruthless about other values.
An intemperate man who pursued pleasure, come what may,
would, provided he was intelligent, provide an obvious example o f a
cunning man (cf. A E i i 4 2 bi8 ; ii4 4 a28). B u t so, if I am right,
would the man who gave him self to the single hearted and
unrelenting pursuit o f philosphy without regard for the moral
virtues. A person who organized his life entirely with a view to the
promotion o f philosphical speculation would be not wise but
cunning, not phronimos but panourgos. T h e type o f person whom
many regard as the hero o f the Nicomachean Ethics turns out, by the
standards o f the Eudemian Ethics, to be a vicious and ignoble
character.1
i. Cooper (19 75 , 164) says ‘Aristotle in the N E conspicuously avoids saying that his
theorizer will be a virtuous person. . . . H e will not possess the social virtues or any other
virtues, because he will lack the kind o f commitment to this kind o f activity that is an essential
characteristic o f the virtuous person.’ D evereux points out on C ooper’s interpretation, there
is nothing to prevent the contemplative from being quite ruthless in pursuing his goal. F or
example, he may by betraying a friend gain a large sum o f money and thereby assure him self
years o f leisure for philosophizing. W hat would hold him back?’ (1978). I am not sure
that C ooper’s interpretation o f N E 10 is correct, on this point; but i f it is, the contemplative
surely deserves the condemnation which the E E , by m y conjecture, would wish to pass on
him.
CH APTER 9
The Dating o f the
Aristotelian Ethical Treatises
I t is the almost universal opinion o f Aristotelian scholars that the
E E was composed earlier in Aristotle’s career than the N E . A s to the
Magna M oralia, scholars do not agree whether to view it as a
genuine work o f Aristotle: but those who treat it as authentic are
unanimous in regarding it as the earliest o f the ethical treatises.
Scholars are far less united about the absolute chronology o f
Aristotle’s ethical works than they are about the relative chro­
nology. T h is is because the opinions concerning the order o f the
treatises are based rather upon a priori theories o f Aristotle’s
philosphical development than upon the historical data to be found
in the treatises themselves. T h is is particularly true in the case o f
the Magna M oralia, which is unusually rich in topical allusions
permitting a fairly precise dating: the dating which the allusions
suggest is quite other than the one which is accepted by the scholars
who regard it as genuine. But it is also true in the case o f the E E and
the N E , where the rather sparser historical data lend no support to
the view that the E E is earlier in date. T h is is particularly the case if
the A E is located in its rightful context in the E E , so that the
historical allusions in the A E can be drawn on in order to give a
terminus post quem for the date o f the E E . T h e present chapter will
examine the evidence for dating the ethical treatises which can be
drawn from their texts. It will not attempt to present an overall
theory o f Aristotle’s ethical development, much less o f the progress
o f his philosophy as a whole.
Before examining the textual evidence in detail, let us look briefly
at the chronologies for the ethical treatises which scholars have
proposed within the framework o f the known dates o f Aristotle’s
life. We know that he was born in the year 384 ; he came to Athens in
367 and remained there in the Academy until approximately the
time o f Plato’s death in 347. For the next three years or so he was at
2i6 The Aristotelian Ethics
Assos under the patronage o f his friend and future in-law,
Herm eias the tyrant o f Atarneus. In 343/2 he was summoned to
M ieza by Philip o f M acedon to act as tutor to his son Alexander.
After Philip’s death and Alexander’s accession Aristotle returned to
Athens in 335/4, and taught there some thirteen years until just
before his own death in 3 2 2 .1
W. Jaeger, who regarded the M M as inauthentic, assigned the
E E to the middle forties, during Aristotle’s stay in Asia M inor ; the
N E belonged, he believed, to the final Athens sojourn. Gauthier
and Jo lif, in both editions o f their massive and authoritative
commentary on the N E , accept this chronology, dating the E E
shortly after 348/7 and the N E between 335 and 330.2 Som e more
recent scholars, while accepting Jaeger’s order for the E E and the
N E , have argued for the authenticity o f the M M as an early work of
Aristotle’ s : thus Dirlm eier proposed that they date from his period
in the Academ y.3 T h e chronology put forward by D üring in his
article in R E Supplem ent-Band x i.3 3 1 ff. dates the Magna M oralia
before 360, the Eudemian Ethics between 355 and Plato’s death, the
N E and the existing A E to the period between 334 and 332.
T o date the Magna M oralia as it stands to a period before Plato’s
death is quite impossible, for the following reasons.
(1) A t i Kj7b2 1, to illustrate the distinction between demotes and
phronësis, the author says ‘E ven a bad man may be called intelligent :
for instance M entor was generally thought intelligent, but he was
not a wise m an.’ Commentators are unanimous in seeing here a
reference to M entor the Rhodian, who in 341/0, by ingenious and
treacherous means, entrapped Hermeias and sent him to the court
o f the Persian K in g where he was crucified. T h e imperfect tense is
often, though not always, taken to indicate that the sentence was
written after M entor’s death (c. 337/6).4 T h e allusion would be
clear to Aristotle’s own circle ; Aristotle him self was deeply moved
by H erm eias’ death and wrote a paean in his praise which later led
to his own prosecution for impiety. T h is tells, as supporters o f the
authenticity o f the M M have insisted, against the theory that the
work was written long after Aristotle’s death, when the memory o f
1. See D üring, 1957, 249 ff. T h e crucial dates are given by D ionysius o f Halicarnassus in
his letter to Am m aeus: see Chroust, 1973, 16-24.
2. G authier, 1970, 6 1 - 2 .
3. D irlm eier, 1958, passim.
4. So von Arnim , 1924, 10.
The Dating o f the Aristotelian Ethical Treatises 217
M entor was less vivid : 1 but it tells even more strongly against those
who wish to place the M M in the 360s.
(2) A t i2 0 3 a23 the writer gives a list o f evildoers on the grand
scale. Along with Phalaris, he names Dionysius and Clearchus. It is
not clear whether the first or the second Dionysius was meant:
either o f them, by their ill treatment o f Plato as well as their other
misdeeds, could deserve to figure in such a list. Clearchus is most
likely the tyrant o f Heraclea in Pontus, once a pupil o f Plato, whose
sanguinary reign lasted from 364/3 to 353/2 .2 T h e reference would
perhaps be more natural after his death ; at the very earliest it cannot
be much before 360.
(3) A t i20 5ai9 -2 3 the writer illustrates a point about the
individuation of dispositions by a reference to the knowledge o f
grammar ‘ in Lam pros and in N eleus’ .3 No likely Lam pros is
known; but Neleus is now universally taken to be Neleus o f
Skepsis, the son o f Coriscus and nephew o f Theophrastus, who
inherited the latter’s library at his death c. 285 (Strabo, xiii.608).
N eleus, Strabo tells us, was a pupil o f Aristotle. Unless he inherited
the library at an improbably advanced age, he can hardly have been
old enough to know grammar during the 360s.4
(4) A t I 2 i2 a4 the writer is discussing the difference between
being well-disposed to someone and being their friend. ‘ I f there
were people as perhaps there were, well disposed to Darius far away
in Persia, it does not follow that they were friends o f his.’ 5 T h e
reference is almost certainly to Darius Codomannus, who certainly
1. I agree with D üring (1966) ‘D er Sinn von Nam en wie Alkibiades oder Lysander
war zu jeder Zeit verständlich. M entor kann nicht zu dieser Kategorie von Persönlichkeiten
gerechnet werden. N ur wenn dieser “ heimtückische M ann” noch in aller Gedächtnis war,
hatte es einen Sinn, ihn als Beispiel hinzustellen.’
2. R E , s.v. ‘K learchos’ . One manuscript adds to the list the name Ε ύ μ ά ν θ η ς . N o such
name is known. Ε υ μ α θ ή ς was one o f the T hirty T yrants, but there seems no reason to single
him out (Xenophon, Hell. 1 1 . 3 2 ; R E vi.1075). Perhaps the text should read Έ ρ ιά ν θ η ς .
Erianthes, the commander who wished on his own authority to destroy Athens after
Aegospotami, might well have lived in Athenian memory as an arch-evildoer. N o other
credible name ending in α ν θ η ς is given in Hansen.
3. T h e reading Ν η λ ( ΐ o f the manuscript M , though not accepted by Bekker and
Susem ihl, has been universally read since the article by Wilamowitz (1927).
4. Von Arnim , who was anxious to place the date o f N eleus’ birth as early as he plausibly
could, concluded that he was probably born in the fifties, and would be about eighteen in 334,
the date which at that time von Arnim favoured fo rth e /ΜΛί. H e argues convincingly that the
knowledge o f grammar in question is not that o f the professional philologist, but o f a
schoolboy (von Arnim , 1928).
5. où γ ά ρ et τ ι ς η ν Δ α ρ ε ίω ζ ϋ ν ο υ ς èv π έ ρ σ α ις ό ν τ ι, ώ σ π ε ρ ίσ ω ς η ν, ε υ θ έ ω ς κ α ι
φ ιλ ία η ν α ύ τ ψ π ρ ο ς Δ α ρ ε ίο ν .
2 i8 The Aristotelian Ethics
had well-wishers among the anti-Macedonian party in Athens.
Whether or not the imperfect tense indicates composition after
D arius’ death in 330, the reference cannot be earlier than his
accession in 336 and is most likely after his great battles against
Alexander at the Granicus in 334 and Issus in 333.
(5) T h e M M , when discussing deliberation, says that no one has
to deliberate how to spell the name ‘Archicles’ . T h e reference may
o f course be to some member of the immediate circle o f the author
o f the M M who has, like Lam pros, been forgotten. Certainly it is
not easy to find an Archicles whose name would be on everyone’s
lips : but the best known o f those who have come down to us was a
trierarch in a battle fought in 334/3 (C IA 11.804“ 18).
Taken together these references make impossible a date for the
M M in the sixties, and make most likely a terminus post quem for the
work of about 335. T h is fact is, o f course, well known to the
partisans of the view' that the M M is an authentic work o f Aristotle’ s
and the earliest o f his ethical treatises. Von Arnim , who was the first
and most candid o f the twentieth-century champions o f the
authenticity o f the M M , accepted that the references proved that
the M M could not be earlier than 334, and concluded that it, and
consequently the two later treatises the E E and the N E , belonged to
Aristotle’s final years o f teaching in Athens (1924, 1 0 - 1 2 ; 1928,
i ff.).1
L ater partisans o f an Aristotelian M M have been reluctant to
follow von Arnim here. In the opinion o f all those who have
compared the works in the last century and a half,2 the M M is
philosophically naïve and crude by comparison with the E E and the
N E . I f we are to attribute it to Aristotle himself, then, it seems that
we must assign it to his apprentice years in the Academy rather than
to his magisterial period in the Lyceum . We shall have to postulate,
therefore, to explain the historical allusions, a revision of the work,
or updating o f the course, in the late thirties. M oreover, the stylistic
differences between the M M and the Aristotelian ethical treatises
1. A t the time o f writing the later monograph, von Arnim believed he had found
confirmatory evidence for this date for the Μ ΑΊ. Rhetoric B must be later than 338, since it
refers to the spccch o f Dcm ades after the battle o f Chaeronea ( i4 0 ib32). But the treatment
o f several passions in chapter 9 o f that work shows that Aristotle had not yet worked out the
doctrine o f the mean, which is clearly to be found in the M M . T h is therefore must be some
years after 338 (loc. cit).
2. T h e qualification is necessary because Schleierm acher, writing in 18 17 , regarded the
M M as the only authentic Aristotelian ethical treatise (18 3 5 , 123).
The Dating of the Aristotelian Ethical Treatises 219
have impressed other scholars more deeply than they impressed
von Arnim. Consequently, it has been postulated that this updated
juvenile Aristotelian treatise has reached us through another
hand— whether, as D üring supposes, the updating itself was
entrusted to someone else, or whether, as Cooper suggests, the
updated course given by Aristotle has been preserved only in the
lecture notes o f one o f the members o f its audience.1
Now1 o f course every scholar is at liberty to postulate as many
juvenile treatises by Aristotle as he wishes, and as many revisions as
he thinks necessary to explain the present state o f the manuscripts
o f the corpus. M oreover, if he is prepared to regard as detachable
from the postulated treatise all chronologically locatable references
and all quantifiable features of style, his hypothesis will be
conveniently secure against any possibility o f objective falsification.
B u t even safeguarded with this protective covering, a hypothesis
such as that o f Cooper can hardly escape the charge o f superfluity.
I f we are to postulate, in order to explain the un-Aristotelian style of
the M M , that w hat we have is the published notes of a student w ho
attended his course, we do not need to postulate further a juvenile
version of the course in order to explain the crudity o f the
philosphical discussion. For the crudity to be explained is not a lack
o f overall philosophical vision or comprehensive organizing design :
it is a constant botching o f the details o f a philosophical argument.
And this kind of incompetence can perfectly adequately be
explained by the mediocrity o f the note-taker who has already been
postulated to account for the style. T h e present state o f the M M is
overdetermined if we explain it both by the immaturity o f Aristotle
and by the imperfect accuracy o f his recording disciple.
In itself, the hypothesis that the M M is a student’s presentation
o f his lecture notes of a course given by Aristotle in his final Athenian
period is an attractive one. It explains, first o f all, the appearance of
the work in the Aristotelian corpus better than the still most
popular hypothesis o f totally post-Aristotelian origin. It explains
why the historical introduction to the work ends with Plato:
something difficult to explain if the work was written by a later
peripatetic ( n 8 2 a3o). It explains how the author can refer to
Aristotle’s doctrine o f the syllogism as his own ( i 2 0 i b25). T h e
crabbed syllogistic manner of much o f the work, so well brought out
I. Cooper, I<)73, 327-49·
220 The Aristotelian Ethics
in B rin k’s m onograph,1 sorts better with the diligence o f a student
polishing up his lecture notes than with Aristotle’s ubiquitous and
lordly contempt for the technical terms and methodological devices
o f his own invention. I f not associated with the hypothesis o f an
effectively vanished juvenile treatise, the theory does justice in an
unforced way to the topical allusions which forbid the assignment
o f the treatise to a period earlier than the middle thirties. Indeed, it
almost reconciles the parties to the traditional debate. For there is
not a very great difference between a scholar who says that the M M
is not authentic Aristotle, but is the product o f a Theophrastean
peripatetic, and a scholar who says that the M M is an authentic
work, but has reached us through the hand o f Theophrastus or
another o f Aristotle’s disciples. T h e difference between them,
rather, is not so much a difference about the historical prob­
abilities, as a difference in the rigour o f their standards o f authen­
ticity.2
I f the M M is a student’s version o f a course given by Aristotle,
there is no reason why the course it records should not be a course
for which we have Aristotle’s own lecture notes : i.e. either the E E or
the N E . In fact, as we have seen, the M M follows the E E slavishly,
in the order of topics and the manner o f their treatment, for almost
the whole o f its length. B y far the simplest hypothesis to explain the
present state o f the M M is that it is a student’s published notes o f
the course which we have in its authentic form as the E E . With the
exception of a few b rief passages, which will receive special
consideration shortly, there are no differences between the content
o f the E E and that of the M M which cannot be explained by a
combination o f a pardonable degree o f incompetence in under­
standing, and a modest talent for editorial revision, on the part o f a
disciple o f Aristotle present at the lectures.
Scholars have failed to see this obvious and attractive explana­
tion o f the evidence largely because o f their conviction that the
E E is an early work. I f the M M is based on notes taken at a course
given during Aristotle’s final Athenian sojourn, then it cannot be a
record of the E E if that belongs to the period in the Academ y under
Plato or in Assos with Hermeias. But we must ask ourselves what
1. Brink, 1933.
2. T o take a modern parallel : students o f W ittgenstein differ in the amount o f authority
they ascribe to G . E . M oore’s notes o f his lectures in the early thirties, or Barrett’s
reconstruction o f his lectures on aesthetics.
The Dating o f the Aristotelian Ethical Treatises 221
really solid evidence there is for assigning an early date o f this kind
to the E E . Von Arnim , as has been remarked, thought that the E E
belonged to the final Athenian sojourn. A n examination o f the
evidence will show that there is nothing to prevent, and m uch to
commend, such a chronology.
T h e E E , like the M M , contains a number o f references to
contemporary persons and events. T h e mention o f Coriscus at
1220 *19 and i24 0 b25 has been thought to confirm a dating for the
E E in the early forties : for Coriscus was probably among the group
around Aristotle in Assos. But Coriscus, whether or not he listened
to lectures in Assos, occurs as a Jo h n D oe figure in so many o f
Aristotle’s writings, assigned by scholars to so many different
periods o f Aristotle’s life, that it is impossible to base any
chronological hypothesis on the use o f his name. Other references
are prima facie more useful for establishing a terminus post quem for
the E E .
(1) ‘ Gourmands pray’, we are told at 1 2 3 1 “ 16, ‘ not to have a long
tongue but to have a crane’s gullet: like Philoxenus son o f E ry x is.’
Several gourmands seem to have borne the name : Philoxenus son o f
Eryxis is possibly identical with the dithyrambic poet Philoxenus o f
Cythera (c. 435-38 0) who seduced Dionysius I ’s mistress Galatea.
A t all events he must have been old enough to be active during the
first part o f the fourth century, for Athenaeus tells us (5.220) that
Aeschines mocked Anaxagoras for having so disgusting a pupil as
Philoxenus son o f E ryxis. His being mentioned, therefore, is no
help in assigning a date to the E E .
(2) ‘A man who is in love endures many dangers,’ says Aristotle
at i229 a23, ‘like the man at Metapontum who murdered the tyrant,
and the person in Crete in the story.’ ‘D ie beiden Beispiele sind
nicht zu verifizieren’ says Dirlm eier, commenting on the passage
(1969, 315). T h e second reference is indeed too vague; but it is not
difficult to find a story to fit the first. Plutarch (M oralia 760 c) refers
to Antileon o f Metapontum, along with Harmodius and
Aristogeiton, as instances o f tyrannicides motivated by love. T h e
story o f Antileon is told in detail by Parthenius in a passage drawn
from the peripatetic historian Phanias o f Eresus (F H G II.298). In
love with the young Hipparinus, he carried out various dangerous
exploits to gratify his beloved’s whim. He met his death after
assassinating the tyrant Archelaus o f Heraclea in order to rescue the
young man from his attentions. A monument was erected to him
222 The Aristotelian Ethics The Dating o f the Aristotelian Ethical Treatises 223
when, shortly after, the city o f Heraclea regained its former the education o f his so n .1 But if so, the Pammenes mentioned can
democratic constitution. hardly be the well-known Theban general, though the name is not a
T h is allusion would assist in dating the E E if only we knew the common one. Prodicus o f Selym bria appears in Plin y’s apostolic
date o f the death o f Archelaus. Heraclea was not founded until 432, succession o f Hippocratic doctors between Hippocrates and
so that the story cannot belong to the sixth century, as some Chrysippus, the teacher o f Aristotle’s medical grandson
commentators, misled by Plutarch’s coupling o f Antileon with Erasistratus.2 T h is would place him in roughly the same generation
Harm odius and Aristogeiton, have suggested. T h e time at which as Aristotle, but does not enable us to fix his date with any precision.
the cities o f M agna Grecia, and Sicily, one after the other, were As for the story o f the harpist and the king, M ichael o f Ephesus tells
reverting from tyranny to democracy was during Tim oleon’s march us that some scholars thought it was Alexander while others
down the Italian coast and his expedition to dethrone Dionysius II thought it did not suit that monarch’s character.3 M odern scholars
o f Syracuse in 344 (see Hackforth, 1927, 288 ff.). But I have been on the basis o f a similar story in Plutarch have thought the allusion
unable to find confirmation for the conjecture that Archelaus’ is to Dionysius o f Syracuse.4 B u t would Aristotle have referred to
assassination is to be assigned to this period. either Dionysius as basileus? Despite its rich allusiveness the
(3) In the tenth chapter o f book V I I Aristotle is discussing passage leaves us in total uncertainty about its date.
friendships in which what is given and what is taken differ. Th ese (4) At i 2 3 i an Aristotle quotes an elegant remark o f the
often lead to trouble, he says. musician Stratonicus. No one knows for certain the dates o f this
wit, whose principal stock-in-trade appears to have been the kind o f
This is how it happens in love affairs, since in them one party pursues the
joke that N ew Yorkers like to make about Philadelphia. A survey of
other as a pleasant person to live with, but sometimes the other the one as
useful, and when the lover ceases to love, he having changed the other all the witticisms attributed to him — a depressingly unhilarious
changes, and then they calculate the quid pro quo, and quarrel as Python experience— shows that they cannot all have been made by the
and Pammenes used, and as teacher and pupil do in general (for knowledge same person, unless he lived from about the time o f Alcibiades’
and money have no common measure) and as Prodicus the doctor did with return to Athens in 407 until Ptolemy Soter declared him self king in
the patient who offerred to pay his fee with a discount and as the harpist 304·5
and the king fell out. (i243b17-25, trs. Rackham)1 Altogether, the undisputed books o f the E E offer us tantalizingly
T h e passage is rich in historical allusions. From the Greek it is clear little help to anchor their date. T h e various allusions can be made to
that Python and Pammenes were in the relationship o f teacher to fit with, but in no way enforce, a dating o f the E E during Aristotle’s
second Athenian period.
pupil.* T h ere is no difficulty in finding a famous teacher o f the
name o f Python: the Byzantine rhetorician, who came to Athens in T h e matter is altered if we restore the disputed books to their
343 as an ambassador from Philip ofM acedon, and who may indeed rightful place as part o f the E E . F o r in the disputed book C there is a
have conveyed the monarch’s invitation to Aristotle to undertake reference which is universally taken by commentators to be a
reminiscence o f the court o f Alexander. A t 1 i5 0 bi3 we are told o f a
Xenophantus who was unable to restrain his laughter on some
1. ο ϊο ν σ υ μ β α ίν ε ι έττί τ ώ ν ε ρ ω τ ικ ώ ν , ο μ ε ν γ α ρ δ ιώ κ ε ι ο ις [τον] η δ ν ν e m το α υ ζ ή ν ,
ο δ’ ε κ ε ίν ο ν ε ν ίο τ ε ώ ς χ ρ ή σ ι μ ο ν ό τ α ν δ έ π α ύ α η τ α ι τ ο ΰ έ ρ ά ν , ά λ λ ο ν γ ιν ο μ έ ν ο υ ά λ λ ο ς
γ ί ν ε τ α ι , κ α ι τ ό τ ε λ ο γ ίζ ο ν τ α ι τ ί άντ'ι τ ι ι ’α ς , κ α ι ώ ς Π ν θ ω ν κ α ι Π α μ μ έ ν η ς δ ιε φ έ ρ ο ν τ ο 1 . I f Aristotle was in Athens at the time o f receiving the summons, as seems to be implied
κ α ι ώ ς δ ιδ ά σ κ α λ ο ς κ α ί μ α θ η τ ή ς ( ε π ισ τ ή μ η γ ά ρ κ α ι χ ρ ή μ α τ α ο ύ χ έ ν ι μ ε τ ρ ε ίτ α ι) , κ α ι by his best-known joke (fr. 619, Rose) : έ γ ώ έ κ μ ε ν ’ Α θ η ν ώ ν ε ις Σ τ ά γ ε ιρ α ή λ θ ο ν δ ια τ ο ν
ώ ς ‘ Π ρ ό δ ικ ο ς ό ια τ ρ ό ς π ρ ο ς τ ο ν ά π ο δ ιδ ό ν τ α μ ικ ρ ό ν τ ο ν μ ισ θ ό ν , κ α ι ώ ς 6 κ ιθ α ρ ω δ ό ς β α σ ιλ έ α τ ο ν μ ε γ α ν , έ κ δ ε Σ τ α γ ε ίρ ω ν εις ’ Α θ ή ν α ς δ ιά το ν χ ε ιμ ώ ν α τ ο ν μ έ γ α ν .
κ α ί 6. L ik e Rackham I accept Jack son ’s emendation τ ί ά ντ'ι τ ίν ο ς ; but I reject Ή ρ ό δ ικ ο ς . 2 . Pliny, N .H . 29.2. Editors emend both Pliny and Aristotle to read ‘H erodicus’. B u t this
2. Jackson is surely correct in drawing this conclusion from the oAo k in the manuscripts involves gratuitously attributing to Pliny not only a confusion between master and pupil but
(19 0 0 ,4 2). W hether D irlm eier is correct in concluding that Python and Pammenes were also also a muddle about birthplaces.
lovers seems to me more doubtful : though a stormy affair with a male pupil is o f course not 3 . C A G X X .4 6 4 .
ruled out by the rollicking heterosexual domesticity attributed to Python in Athenaeus, 4. Plutarch, M oralia 33 f.
5-499- 5. R E iv A . 1 . 3 2 6 gives the dates o f his activity as 4 1 0 - 3 6 0 . See G ow , 1968, 7 0 .
224 The Aristotelian Ethics
solemn occasion and broke into a guffaw. We know from Seneca
(De Ira ii.2) that Xenophantus was a musician in the service o f
Alexander. M oreover, his period at Alexander’s court must have
been the very beginning o f his career, since he was still active
enough to play for the funeral o f Dem etrius Poliorcetes in 28 3.1 As
long as book C was thought to belong to the N E , or to a post-
Aristotelian Eudemian Ethics, this passage excited no remark : but if
it is indeed an allusion to the M acedonian court musician, and the
A E belongs with the-Ei;, then we must date the E E well after Plato’s
death, well after the period in Assos, and most probably therefore
during the final Athenian period : the same period, in fact, to which
we were led by the historical allusions to date the Magna M oralia.2
I f we turn to the N E , we find a very solid terminus post quem for
the work in the allusion, in the discussion o f courage, to a battle ‘At
the temple o f H erm es’, i i i 6 bi9- Thanks to the well-informed
second-century scholiast (C A G xx.16 5) this can be identified with
an event during the Sacred W ar in 353 at Coronea in Boeotia.3 We
know therefore that at least this part o f the N E cannot have been
written earlier than the period just before Plato’s death. W e cannot
assume from the casual nature o f the allusion that the battle was still
news at the time o f writing : a little later, at 1117 * 2 7 , there is an
equally off-hand reference to a battle at the L o n g Walls o f Corinth
in 392. T h e reference to the splendid character o f Eudoxus in N E
10 .2 ( i i 7 2 bi5) has seemed to many to betray the tones o f the
obituarist. According to the most usual chronology, Eudoxus died
about 35 5.4 From the tensing o f the reference to Speusippus in
1. See Berve, 1926, s.v., citing Plutarch, Demetrius 53.
2. In the same passage there is a reference to a play o f Theodectes, whom Plutarch
(Alexander 17) tells was a pupil o f Aristotle along with Alexander. T h is confirms the
identification o f Xenophantus: but the exact dating o f Theodectes’ dramatic career is
uncertain. Susem ihl (R h . M us. liv.6 31 ff.) dates his first tragedy to 353, and Pickard-
Cam bridge placed his death (which we know occurred when he was forty-one, Plutarch, X
Orat. 837 d), in 334. T h is dating fits the ancient literary evidence best; but on the basis o f the
Didaskatiai other scholars (W ebster, 1954, Diehl, R E v A .17 2 2 ) place his career six or more
years earlier. O f Carcinus, the other dramatist mentioned here by Aristotle, all we know is
that he attended the court o f D ionysius I I . It would assist us to date the A E if we could be
certain o f the identity o f Σ ά τ υ ρ ό ς ό φ ίλ ο π ά τ ω ρ in 114 ^ '34, the son o f Sostrates according
to C A G X X .4 2 6 . M ore than one Pontic king bore his name in the fourth century (G auth ier-
Jo lif, 1959, 624).
3. See R E , s.v. Onomarchos.
4 . So Heath, 1 9 2 1 , 3 2 2 - 3 4 . M erlan, i 9 6 0 (like von Fritz, 1 9 3 0 ) , has argued for a redating
o f Eudoxus’ lifespan to 3 9 5 - 3 4 2 , making him survive Plato.
The Dating o f the Aristotelian Ethical Treatises 225
i ig 6 b7 it is impossible to tell whether he was alive or dead at the
time o f writing, which may therefore be before or after 339. Plato is
twice quoted in the present tense ( n o 4 bi 2 ; i i 7 2 b28): no
conclusion can be drawn from this, for so is Hom er ( 1 1 i6 a22 etc.).
Nothing in the N E forces us to a date much later than 3 5 3 ; but
equally, nothing disproves the traditional assignment o f it to
Aristotle’s last years. Certainly he cannot have been a young man
when he introduced his course with the remark that a youth was no
fit student o f ethics (1095*2).1
T h e historical data, then, are perfectly consistent with our
hypothesis that the M M , while incorporating a small amount of
material from an already existing N E , is essentially based on a
course given by Aristotle during his last years at Athens which
corresponds in substance with our text o f the E E . But we must
examine more closely the cases in which the text o f the M M
diverges from the structure and content o f the E E .
N o great difficulty is presented by the fact that the M M begins
with a summary o f the thought o f previous ethical thinkers which
does not appear in the E E . T h e content o f the M M introduction
(118 2 * 10 -3 0 ) would not be beyond the ability o f a moderately
talented student to construct for himself. On the other hand,
Aristotle no doubt gave the course which we possess as the E E on
more than one occasion : and most o f us rewrite the beginning o f our
lecture courses more frequently than we rewrite the rest o f them.
T h ere is evidence in the text o f the E E that the first chapters have
been more carefully revised than the rest o f the work : they and they
alone are at pains to avoid hiatus.
T h e incorporation o f the Nicomachean material presents more o f
a problem. T h e two most substantial Nicomachean portions o f the
M M are the amalgam o f quotations identified by Allan, introduced
with the words ‘this can be seen from the Ethics' (1 i8 5 bi5 ) ,2 and the
section on self-love in chapters 13 and 14 o f book 2 ( i2 i2 * 3 8 - i3 b2).
1. T o illustrate the point that we do not deliberate about what is beyond our control E E
1226*29 says ‘W e do not deliberate about affairs in In dia’ ; N E i i i 2 a28 says ‘N o Spartan
deliberates about what form o f government would be best for the Scythians.’ I f we are to
attach any importance to this difference, we m ight say that the Scythians became a poor
example o f things beyond our ken with Philip’s intervention in Scythia in 339, while the
Indians remained a good example until Alexander’s invasion o f India in 327. B y the end o f
Aristotle’s life both examples must have looked a little passé.
2. See above, p. 10.
220 The Aristotelian Ethics
N ow it is a startling coincidence that at each o f the points where the
M M draws extensively on the N E , there occurs in the text o f the E E
an allusion by Aristotle to an earlier work o f his— allusions which
scholars, blinkered by the dogma o f the priority o f the E E , have
universally found baffling.
A t 1 220b 1 1 in the E E , during the discussion o f the way in which
moral virtue concerns pleasure and pain, we read ‘A fter this comes
the classification o f passions, powers, and dispositions made en tois
apFllagmenois\x T h e three Greek words have often been found
puzzling and various emendations have been proposed. D . J . Allan
has suggested two possible translations o f the manuscript reading:
‘ in the cancelled version’ or ‘ in the separate section’ .2 In either case,
Aristotle is referring here to an appendix. Rackham ’s comment on
the passage is interesting: it is, he says, ‘perhaps a reference to N E
i i0 5 b20, inserted in the belief that the E E is the later work’ (19 35 ,
247). Certainly, except for very slight variations, the E E text and
the N E text are here very similar.3 T h e corresponding passage in
the M M (1 i86a9 ff.) comes right in the middle o f the Nicomachean
cento introduced with a reference to the Ethics ,4 If, as the words en
tois apëllagmenois suggest, the E E contained (or the E E lectures
appealed to) an appendix o f superseded material at this point, and if
the superseded material was in fact the N E or part o f it, then we
would have a full explanation o f why the M M in this area
apparently deserts its usual model, the E E , for verbatim copying o f
the N E .
i2 2 0 bi i is not the only place in the E E where Aristotle refers to
his own earlier work. In the discussion o f self-sufficiency in the
1. μ ε τ ά τ α ν τ α ή δ ια ίρ ε σ ις ε ν τ ο ΐς ά π η λ λ α γ μ έ ν ο ις τ ώ ν π α θ η μ ά τ ω ν κ α ι τ ω ν
δ υ νά μ εω ν κ α ι τω ν εζεω ν.
2. Allan, 19 6 1, 3 1 2 ? J 9^6, 148. In support o f the first sense o f α π α λ λ ά τ τ ο μ α ι Allan
appeals to Plato, R ep. viii.559 b and ix.571 b.
3· T h e E E , as is its wont, is a little conciser and more precise, specifying passion by its
accompaniment with sensory pleasure and pain, and explaining that ε ξ ε ις are defined by
being κ α τ ά λ ό γ ο ν η ε ν α ν τ ίω ς .
4- M M 1 1 85bi4 reads: o n δ ε ή ε ν δ ε ια κ α ί ή ύ π ε ρ β ο λ ή φ θ ε ίρ ε ι τ ο υ τ ίδ ε ίν ε σ τ ιν ε κ
τ ω ν η θ ικ ώ ν · δ^Γδ* ύττερ τ ώ ν α φ α ν ώ ν τ ο ΐς φ α ν ε ρ ο ΐς μ α ρ τ υ ρ ιο ις . Allan comments : ‘T h e
attempts o f scholars to emend ε κ τ ώ ν η θ ικ ώ ν (see Susem ihl) show that they failed to
recognize the words from δ ε ι to χ ρ ή σ θ α ι as a verbatim quotation o f E N 1 10 4 * 12 .. . . With
this commences a passage o f about five T eubner pages in which phrases from the N E are
cited with some embroidery and expansion; that we find nothing sim ilar outside these pages
o f M M ; and that in view o f this it is only reasonable to translate ίδ ε ίν ε σ τ ιν ε κ τ ώ ν η θ ικ ώ ν
‘ it can be seen from the Ethics' (Allan, 1966, 14 2 ; see allan, 19 57, 7).
The Dating o f the Aristotelian Ethical Treatises 227
treatise on friendship, Aristotle makes an obscure critical reference
to a book in which the same topic has been treated.
If therefore one abstracted and treated knowledge as absolute and
independent and did n o t. .. (but this is obscure in the way it is written in
the book, though in fact it need not be obscure) in that case there would be
no difference between knowing oneself and knowing another instead of
oneself, and likewise no difference between another living instead of
oneself: whereas patently to be conscious and aware of oneself is more
desirable. Two of the things in the book have to be put together : life is
desirable, but so is goodness. (i244b29-45a)1
H ere a book— presumably one by Aristotle, else the author would
have to be mentioned2— is being criticized for being obscure, and
for excessively separating knowledge from goodness and regarding
life, identified with this knowledge, as the only desirable thing. Can
we locate an earlier work o f Aristotle which is criticisable on these
grounds ?
M ost authors treat the reference as being to a lost work, and are
therefore at liberty to invent for it a text suitable to be a peg for these
criticism .3 But if we turn to the parallel passage in N E 9, we can find
an extant argument o f Aristotle’s which passably fits the E E ’s
strictures. First, the passage (1 i7 0 a2 5 - b9) is intolerably obscure: it
contains a sprawling sentence o f 1 1 2 words whose construction is
impermeable even with the aid o f the five braces o f brackets
thoughtfully provided in Byw ater’s edition.4 T h e argument goes
approximately thus: life consists essentially in perception and
thought, both o f which are accompanied by self-consciousness : it is
1 . €i ο ΰν τ ι ς ά π ο τ ε μ ο ι κ α ι π ο ιη σ ε ιε το γ ιν ώ σ κ ε ιν α ύ το καθ'1 α ύ το κ α ι μ η . . . (άλλα
τ ο ύ τ ο μ ε ν λ α ν θ ά ν ε ι, ώ σ π ε ρ ε ν τ ώ λ ό γ ω γ ε γ ρ α π τ α ι , τ ώ μ ε ν τ ο ι π ρ ά γ μ α τ ι ε σ τ ι μ η
λ α ν θ ά ν ε ιν ), ούθ'εν αν δ ια φ ε ρ ο ι η το γ ιν ώ σ κ ε ιν ά λλ ον ά ν θ ’ α ύ τ ο ΰ ■ το δ ’ ό μ ο ιο ν τ ο ν ζη ν
ά ν θ ’ α υτο ύ ά λλον, ε ύ λ ό γ ω ς δ ε το ε α υ τ ο υ α ια θ ά ν ε σ θ α ι κ α ι γ ν ω ρ ίζ ε ιν a i ρ ε τ ώ τ ε ρ ο ν .
δ ε ι γ ά ρ ά μ α σ ν ν θ ε ΐν α ι δύο εν τ ώ λ ο γ ω , ό τι τ ε τ ο ζ η ν [κ α ι] α ιρ ετό ν , κ α ι ό τ ι το α γα θόν.
T h e text is corrupt in places, and I do not know how to emend it; but the general sense is
clear.
2. So D irlm eier (1969, 4 6 1): ‘ wir haben wieder jenes in Corpus Ar. einzigartige, a u f E E
und Poetik i454bi8 beschränkte, Sich-berufen a u f ein geschriebenes W erk, ein B u ch .’
3. T h e most promising attempt hitherto to locate a source in the lost works o f Aristotle for
the quotation is that o f G aiser (1967) who thinks that the logos referred to is the Protrepticus
(see also W idmann, 1967). T h e two passages which G aiser cites as parallels from modern
reconstructions o f the Protrepticus—frs. B 7 1 - 7 , 97-/O J Düring— are far less close to the
allusions in the E E than the texts in N R 9 and N E 10 discussed below.
4. According to the figures given by Wake, 19 57, and M orton and W inspear, 19 7 1, this
must be one o f the longest sentences in Aristotle. Wake, indeed, does not record any sentence
in the N E , E E , or A E longer than 1 1 5 words.
228 The Aristotelian Ethics The Dating o f the Aristotelian Ethical Treatises 229
indeed the consciousness o f one’s own existence that makes life So we must study together and feast together and share the best
pleasant. B u t a good man looks on his friend just as he looks on things in life so far as we can attain them. T h e contrary view 1 which
him self: therefore consciousness o f his friend’s existence will be made the solitary life seem as valuable as life in the society o f friends
just as desirable to him as consciousness o f his own. Hence, a good was based on a false comparison between man and god. I f men were
man must have friends, or he will lack something desirable and will like gods they would not need friends; but if they were like gods
not be truly happy. they would not have any thoughts either, except o f themselves. ‘For
T h ere seems something very strange about this derivation o f the us, our well-being has reference to something other than ourselves ;
desirability o f friends from the value o f self-consciousness. T h e in his case he is him self his own well-being’ . (i2 4 5 bi8 -i9 )
argument in the E E shares some o f the strangeness, but it does Both in the N E and in the E E Aristotle, then, Aristotle argues to
criticize one thing which seems a feature o f the N E argument: the the value of friendship from the value o f self-consciousness and the
exaggerated value placed on the abstract awareness o f one’s own or role o f the friend as an alter ego. But the E E is anxious to criticize a
other’s existence. N o doubt it is nice for the good man to know he is presentation of this argument which divorces knowledge too much
good (i I70b5); but he has to be good, to exercise his virtues, and not from other values, which treats human beings as if they were pure
just to contemplate his goodness with self-satisfaction. It is in the godlike intellects, and which does not sufficiently stress the need for
exercise of the virtues— in the performance of good and worthwhile friends to share in all the elements o f the good life. It is striking that
actions— and not just in the contemplation o f them that we should all these points o f criticism are the very points in which, in the
surely look for the value o f friendship. It seems to be along these previous chapter, we identified the differences between the E E and
lines that the E E develops its own argument. Looking at life simply the N E with regard to the nature and constituents o f happiness.2
in terms o f the possession and communication o f knowledge leads T h ere is no need to look elsewhere than to the N E for the logos
to the conclusion that the society o f others is superfluous (i2 4 5 ai - which is here being subjected to critical commentary.3
19 ); but it is obvious that we all want to share things with our Once again, the conjecture that the E E here refers to the N E is
friends, whether bodily pleasure, or artistic culture, or philosophi­ confirmed by a comparison with the M M . T h e second large
cal inquiry. Hence there must be something wrong with an Nicomachean borrowing in the M M occurs at I 2 i2 a2 8 -b24, in the
argum ent1 leading to an opposite conclusion. (i2 4 5 a20-8) T h e section on self-love which in the N E immediately precedes the
knowledge of one’s friends who resemble one in various ways is passage corresponding to the E E criticism o f the logos, and which
indeed a way o f knowing oneself; and because a friend is an alter ego first introduces the identification of a human being with his nous
one can look on his good as on one’s own. But the value o f so doing (1 i68b29-69a3) which is developed in the N E 10 account o f
will depend fundamentally on the value o f the good shared. contemplative happiness. T h e M M then goes on, like the E E , to
criticize the excessive assimilation o f man with God which is to be
To share even common pleasures and ordinary life with a friend is
found en tois logois.4 Once again, the presence of the unusual
obviously pleasant (for it is always accompanied by consciousness of him) ;
but it is even more pleasant to share the more divine pleasures. This is ό τ ι a c t ή& ιον Ε αυτόν deœ petv èv τ ω β έ λ τ ίο ν ι ά γ α θ ω . το ύ το δ’ € σ τίν o re μ € ν π ά θ ο ς , ore
because it is always pleasanter to see oneself enjoying a superior good : and δ ί π ρ ά ξις> o re £è erep ov τ ι. et δ ’ α υτόν i ζ η ν κ α ί ο ν τ ω κ α ι τ ο ν φ ίλ ο ν , èv ôè τ ώ σ ν ζ η ν
this may be now an emotion, now an action, and now something different σ υ ν ε ρ γ β ΐν , ή κ ο ιν ω ν ία τ ω ν èv reAet μ ά λ ισ τ α y e.
from either. I f it is pleasant to live well for oneself and a friend, and if 1. λ ό γ ο ς is again ambiguous at I245bi2.
2. N E 1 0 , 1 17 8 ^ 0 - 2 4 , provides an example o f the excessive assimilation o f man with G od
living together involves working together, their sharing will be above all in which exactly fits the complaint o f E E i244b23 and I245bi 3 - i g .
the things which make up the end of life. (i245a38-b4)2 3. Several particular passages in the course o f the argument suggest that E E is
commenting on N E : thus E E 1245*2-^ expands and modifies— admittedly not in a very
1. It is not clear whether logos in line 27 means ‘argum ent’ or is another reference to the perspicuous manner— the brief allusion to what is ώ ρ ισ μ € ν ο ν at i i 7 0 a2 0 ; and E E 12 4 5 * 15 -
book criticized in lines 3 1 and 35 above. 16 corrects H 70bi2.
2. α>στ€ κ α ί τα φ ο ρ τ ι κ ά μ-kv σ υ ν ή δ ς σ θ α ι κ α ί σ ν ζ η ν τ ω φ ί λ ω η δ υ eùAoycoç 4. T h e plural may mean that M M has other texts too in view, perhaps including ones in
( σ υ μ β α ί ν € ΐ γ ά ρ ϊ κ ζ ί ν ο υ ά μ α α ί σ θ η σ ι ς act')· μ ά λ λ ο ν r a s fle io re p a s ή δ ο ν ά ς . α ί τ ι ο ν δ ’ which the comparison with God was made more explicit. M M differs from E E here in that
230 The Aristotelian Ethics
Nicomachean material in the M M would be simply explained by
the presence in the E E o f a Nicomachean appendix: in this case,
chapters 8 and 9 o f N E 9 .1
M ost scholars will reject these conjectures out o f hand as
presupposing the existence o f at least parts o f the N E at the time
when the E E was written or delivered as a course o f lectures.
Certainly it is difficult to find a scholar who will seriously consider
the possibility that the N E antedates the E E : but the time has come
to look at the evidence for the priority o f the E E .
Probably the commonest ground for assigning the E E to an
earlier date has been the belief that it is more Platonic than the N E .
In so far as this claim rests on the actual doctrines o f the ethical
treatises the claim seems unfounded : as we have shown, the E E is
no m ore Platonic than the N E in its treatment o f wisdom, and
considerably less Platonic than the N E in its refusal to identify
happiness with the activity o f the separable nous. I f the claim rests
on the constant allusions to Platonic themes and terminology, then
it must be admitted that the E E is highly Platonic : it is shot through
with reminiscences o f Plato’s ethical dialogues from the Euthyphro
to the Philebus. But so too is the N E . I do not know that anyone has
undertaken a count to see which o f the two treatises contains the
more explicit and implicit allusions to Platonic texts. T h e exercise
would no doubt be futile : but it might well be that there are more
clearly identifiable Platonic citations in the E E . T h is would merely
be an instance o f a general feature o f the style o f the E E remarked on
long ago by G rant, a more copious and explicit manner o f
quotation. T h is stylistic feature may merely indicate that the E E is
instead o f using the theory o f divine self-contemplation o f Metaphysics as part o f a reductio ad
absurdum o f the comparison between man and god, it introduces (whether or not it endorses)
a comparison between m an and god as part o f a criticism o f the Metaphysics Λ view : i f a man
spends all this time gazing at his own navel we regard him as a dunce; how can we accept a
self-contem plating G od ? ( i 2 i 2 a3 6 - i 3 a8)
i. In one other place £ £ refers to a logos, in the criticism o f P lato’s Ideas. € τ ι κ α ι t o i v τ ω
λ ό γ ω γ ε γ ρ α μ μ ί,ν ο ν . η γ α ρ ο ν & ξμ ιά χ ρ ή σ ιμ ο ν a ν τ ο τ ο τ ο ν ά γ α θ ο ν ξ ϊδ ο ς tj π ά σ α ις
ο μ ο ίω ς ■ ζ τ ι ού π ρ α κ τ ό ν . T h is too would fit the N E well: after the arguments which are
common to both treatises the N E then goes on to an argument which depends on the
disjunction: either the definition o f good applies only to the idea, in which case μ ά τ α ιο ν
€ σ τ α ι το € Γδο? ; or else we have to suppose the same definition in goods o f diverse kinds which
are related only analogically (i0 9 0 b8 - 3 i) ; he then goes on to object that the idea ο ν κ α ν ς ϊη
'π ρ α κ τ ό ν ουδβ κ τ η τ ό ν ά ν θ ρ ώ π ω (b3 I -~5)· T h e note again is easily explained as a reference
to an appendix, or hand-out, o f Nicomachean material.
The Dating o f the Aristotelian Ethical Treatises 231
further on the way to publishable form than the N E ever w as.1 B u t
scholars are coming more and more to agree that A ristotle’s works
cannot be ordered on a scale o f deviation from Plato : his philosophy
was a lifelong development o f Platonic themes and a lifelong
confrontation with Platonic problem s.2
T o be sure, there is evidence that at different times in his career
Aristotle’s attitude to the Platonic theory o f Ideas varied between
different degrees o f sympathy and harshness. But it is unwise to
conclude that the sympathetic criticisms are early and the harsh
criticisms are late ;3 it is equally imprudent to draw the opposite
conclusion, imagining a pupil growing gradually more tolerant o f
his teacher’s doctrines under the mellowing influence o f age. I f we
compare the criticism o f the theory o f Ideas in N E 1.6 with that in
E E 1.8 I think it is undeniable that the criticism o f the E E is made
from a greater distance. In the N E Aristotle introduces his attack
with an apology and an admission that the inventors o f the theory
are dear and close to him : ‘ amicus Plato sed magis amica veritas’—
an apology which tactfully echoes Plato’s own remarks on Hom er
(i096an - i 7 ) . In the E E he begins: ‘the existence o f a Form not
only o f good but o f anything else is an idle abstraction : this has been
considered in multifarious ways both in popular and philosophical
writings’ (i2 4 7 b20 f.). T h e arguments in the N E share more
Platonic premisses than those in the E E , and are less sharply and
crisply developed.4 I f the differences between the two treatments
were to be settled by assigning a temporal order between them, it is
undoubtedly the E E that an open-minded reader would regard as
the later. But the reason for the differences may well be quite other
than a lapse o f time.
When nineteenth-century scholars sought to show that the E E
was post-Aristotelian they pointed to the fact that there are some
virtues for which Aristotle tells us in the N E he has to coin a name,
1. G rant, 1885, 30: ‘T h e Eudem ian writer . . . shows indeed a proclivity to indulge in
abundance o f literary quotations, and he quotes more fully and explicitly than A ristotle.’
2. F o r D irlm eier, Aristotle’s development is ‘Reifen, in Verbindung mit Plato, bis zuletzt’
(1969,568).
3. T h e best treatment o f the E E and N E criticism o f the Ideas is Flashar, in Festschrift
Schadewalt, 2 23-37 . Flashar notes that the E E contains two arguments not to be found in the
N E , and thinks that while all the N E arguments are directed against Plato the E E arguments
are against Xenocrates— though not necessarily after Plato’s death.
4 - Flashar’s own explanation— that the N E is more exoteric than the E E and therefore
more apology to Plato seemed called for hardly seems plausible in view o f the generally
more finished and polished form o f the EE.
232 The Aristotelian Ethics
but which are alluded to by that name without hesitation in the E E
and the M M . T h u s, at n o 8 a4 -9 we read:
In regard to anger also there exists an excess, a deficiency and a mean.
Although there are really no names for them we might call the mean
praotes (gentleness) since we call a man who occupies the middle position
gentle. O f the extremes, let the man who exceeds be called orgilos (short-
tempered) and his vice orgtlotês, and the deficient man aorgetos (apathetic)
and his vice aorgësia.
Again in book 4 Aristotle says that the middle here is nameless, and
the extremes almost so (1 i2 5 b26 ff.). B u t in the E E praotes is used
without apology at i2 2 0 b38, as is orgilotes (the corresponding defect
is here called analgesia) ; praotes is again used without hesitation in
I I I . 3, i 2 3 i b5 f f ., where various names are offered for the
corresponding defects, including orgilos. Anaisthësia is listed at
i 2 2 i a2 as the contrary vice to intemperance : at I2 3 0 bi4 we are told
that some people call those uninfluenced by pleasures anaisthetoi
and others give them other names. In N E at 1 i0 7 b8 Aristotle says
that such rare birds are nameless, and offers anaisthetos as a coinage.
Again, at i i 2 6 bn a virtue o f sociability is introduced which is
described as having no name (cf. n o 8 ai7 ) : Aristotle settles for
calling it philia. T h e name is used without embarrassment at 12 2 i a7
and i2 3 3 b30 in the E E . Such examples do not show that the E E is
post-Aristotelian: but the nineteenth-century scholars may well
have been right to see them as an indication that the E E postdates
the N E .1
It is not only differences in nomenclature which distinguish the
treatments o f the virtues in the N E (3-4 ) and in the E E (III). T h e
ampler and more vivid treatment o f the particular traits o f character
is one o f the features which contributes to making the N E longer
than the E E . In particular the virtues o f magnificence and
magnanimity are given longer and more loving treatment in the N E
(six and a half Bekker columns) than in the E E (less than three
columns).2 Gentleness, on the other hand, while treated like all the
virtues at greater length in the N E , is upgraded in the E E to third
place among the virtues. T h e lists in the two Ethics agree in their
content, though not in their ordering, with one exception. T h e N E
has a place for a nameless virtue o f proper ambition, a satellite
1. See Spengel, 18 4 1, 457.
2. S e e D . Rees, 19 7 1, for a detailed comparison between the treatments.
The Dating o f the Aristotelian Ethical Treatises 233
virtue to magnanimity ( i i 2 5 bi- 2 5 ) — it may be called philotimia
without the pejorative overtones o f that word. T h e E E knows no
such virtue. Instead, in its list of good passional means (i2 3 4 a24) it
has the quality o f semnotës (dignity), a mean between obsequious­
ness and self-will. N o doubt it is rash to draw any conclusions from
these variations in emphasis. But it is perhaps worth pondering the
following question. Who is the younger man, and who is the older
m an: the author o f the N E who prizes highly proper ambition,
magnificence, and magnanimity, or the author o f the E E who is
m ore attracted by dignity and gentleness?1
W e come finally to an argument which probably most scholars
today would regard as the strongest for dating the E E early, or at
least earlier than the N E : the treatments o f pleasure in A E C and
N E 10. M ost scholars are convinced that the treatment in A E C is
Eudemian, even if they are prepared to regard most o f the A E as
Nicomachean: for as part o f the N E it makes such a puzzling
doublet with N E 10 which makes no reference to it. M ost scholars
are also convinced that the treatment o f N E 10 is philosophically
more sophisticated than that in A E C : consequently, they say, since
A E C belongs with the E E and N E 10 with the N E , and N E must be
later than the E E .2
Recently, G . E . L . Owen has argued that the two treatments are
not doublets because they are not about the same topic. T h e A E
treatment offers to tell us what our real pleasures are; the N E
discussion is designed to tell us the nature o f enjoying by reviewing
the logical characteristics o f pleasure-verbs.3 It is a mistake, he
thinks, to draw a contrast, as many scholars do, between a treatment
which identifies pleasure with unimpeded activity and a treatment
which regards pleasure as an epiphenomenon supervenient on
activity.4
1. M y colleague, M r. J . G riffin o f Balliol College, has suggested that if the E E is later than
the N E the difference between the two treatments could be explained by disillusionment with
the career o f Alexander.
2. See Festugière, 1946; Lieberg, 1958; G au th eir-Jolif, 19 5 8 ,77 8 ff. Our stylistic criteria
support the common view that A E C ’s treatment o f pleasure is Eudem ian and N E io ’s is
Nicomachean : T ab le 9.1 shows the result o f comparing the two treatments with expectations
drawn from the E E and from the remainder o f the N E . T h e comparison would hardly have
been worth making, for the point is so generally accepted, were it not for theattem pt o f W ebb
(Phronësis 1977, 236) to show that it is the C treatment that is the Nicomachean one.
3. Owen, 1972.
4. T hus Anscom besays that pleasure ‘astonishingly, reduced Aristotle to babble, since for
good reasons he both wanted pleasure to be identical with and to be different from the activity
that it is pleasure in’ (19 57, 76).
234 The Aristotelian Ethics The Dating of the Aristotelian Ethical Treatises 235

T a b le 9.1 Nicomachean and Eudemian Favourite Expressions in the Two them appear both in the A E and in the N E . At the bottom o f the
Treatments o f Pleasure scale there are the pleasures o f the sick (sick either in body or soul)
which are not really pleasures at all : these appear in A E at i i5 3 b33
Group Total in Expectation Actual Expectation Actual and i i 5 4 a32, and in B at i i7 3 b22 and i i 7 6 aig. N ext up the scale
N E less in N E in N E in A E C in A E C come the per accidens pleasures o f food and drink and sex as enjoyed
10 'I -5 10· 1-5 io -1-5 1 1-14 11-14
by the gourmand and the lecher : these appear in A E at 1 15 2 b35 ff.,
H 5 4 ai7 , a35~b3 where they are distinguished from the lower class
A-favourites
o f false pleasures.1 Th ese pleasures appear in N E at i i 7 3 b8 - i5 .
N I 542 42-38 45 24-48 18
N ext up the hierarchy come the aesthetic pleasures o f the senses.
N II 453 35·42 29 20-11 7 T h ese are divided into two classes— the pleasures o f the inferior
N III 38s 30 1 1 33 I 7-44 6
N IV 2 9 17 16-86 12 sense o f touch and taste, and the pleasures o f the superior senses o f
373 31
NY 29-48 30 16-98 6 sight, hearing, and smell. These are mentioned in A E at i i 5 3 a26
377
N VI 403 31-46 H I 7-39 7 and in N E especially at i i7 4 bi4 -7 5 aio and i i7 5 b36~76a2. Finally
there are the pleasures o f mind mentioned in A E at 1 15 3 s! and 20,
2533 19802 182 113-27 56
and in N E at 1 173h 17, i i 7 4 b2 i, and b34 and elsewhere.
Secondly, the nature o f pleasure and its relation to activity. T h e
passage in the N E which is usually thought to distinguish pleasure
from activity in a way which is incompatible with the A E is
Group Total in Expectation Actual Expectation Actual
EE in N E in N E in A E C in A E C H 74b23~32 :
10-1-5 10 1-5 1 1 —14 1 1 —14 Each sense has a corresponding pleasure as also have thought and
speculation, and its activity is pleasantest when it is most perfect, and most
Æ-favourites perfect when the organ is in good condition and when it is directed to the
E I 31 1 33-90 H 19-47 25 most excellent of its objects; and the pleasure perfects the activity. The
E II 263 28-67 16 16-46 24 pleasure does not however perfect the activity in the same way as the object
E III 279 3°"4 I II 17-46 25 perceived and the sensory faculty, if good, perfect it ; just as health and the
E IV 256 27-90 12 16-02 25 physician are not in the same way the cause of being healthy. . . . The
EM 267 29-10 12 16-71 13 pleasure perfects the activity, not as the fixed disposition does, by being
E VI9 380 4 14 2 39 23-79 18 already present in the agent, but as a supervening perfection, like the
190-88 104 109-98 130 bloom of health in the young and vigorous, (trs Rackham)
1756

H ow can this statement be reconciled, scholars have wondered,

I agree with Owen that the traditional contrast is mistaken, but I
find his own contrast unconvincing.1 T h e doctrine of the A E and
the N E seem to me to be identical with respect to the classification
o f pleasures, the nature o f pleasure and its relation to activity, and
the moral value o f pleasure.
First, the classification o f pleasure. There are five different types
o f pleasure, or pseudo-pleasure, considered by Aristotle, and all of
i. Owen has been criticized by G oslin g, 1973.
with the definition o f pleasure in A E ( 1 1 53s 14) as the unimpeded
activity o f a disposition in accordance with nature? T h e recon­
ciliation is not difficult, and the way to undertake it is briskly
indicated in Owen’s article:
The activities of the natural states which served as [AE |’s paradigms of
pleasure need not be enjoyable at all. Smith is exercising his wits on an
argument; but his wits are blunt, he is tired, the argument is tangled. So
I. T h ere is a difficulty at 1 15 4 bi 7 where the two types o f pleasure distinguished above
seem to be confused.
236 The Aristotelian Ethics The Dating o f the Aristotelian Ethical Treatises 237
[/Y£] is spelling out the further conditions that are requisite for pleasure— Finally, the relation between pleasure and goodness. Both A E
sharp -wits, impeccable object. (1972, 146) and N E regard the question ‘ is pleasure good or bad ?’ as too simple.
Both regard the question as only answerable after pleasures have
Such conditions are precisely what is covered by Aristotle’s
been distinguished and classified in the manner just discussed.
requirement in A E that an activity, to be pleasant, must proceed
Pleasure in a good activity is good, and pleasure in a bad activity is
unimpeded. You are at a wine-tasting; you are free from colds or
bad: this doctrine o f the N E ( ι ι η ^ τ η ) leaves it open— as only the
distracting flute-playing; then if you do not enjoy the wine either it
A E says explicitly— for the pleasure o f the best activity to be the
is a bad wine or you have a bad palate; there is no third alternative.
best o f all human goods.
T h e language o f N E is an expansion o f what is brusquely said at
I f we are to seek, then, to settle the temporal relationship
1 1 54b20 : those things are naturally pleasant which bring a healthy
nature into action (physei d’hëdea, ha poiei praxin tes toiasde between the E E and the N E by a comparison between the two
treatments o f pleasure, it must be to features o f composition rather
physeös). Pleasure ‘ perfects’ activity: that is to say, it causes the
than to matters o f philosophical substance that we shall have to
activity to be a good one o f its kind. T h e faculty and the object,
appeal in awarding the palm to one text rather than the other.
between them, are the efficient cause o f the activity ; if they are both
Festugière is the only author to have really argued a case for the
good, they will be the efficient cause o f a good activity, and
superiority o f the N E treatment : and recently Webb has shown that
therefore they too will ‘perfect’ activity, i.e. make it be a good
in the great majority o f cases F estugière has simply argued from the
specimen o f such activity. B u t pleasure causes activity not as
greater length o f the N E treatment to its being the more mature one
efficient cause, but as final cause : like health, not like the doctor.1
(W ebb, 1977). Apart from this, Festugière’s principal argument is
T h e physis o f A E is the good hexis o f N E , the praxis o f A E is the
that the A E treatment is largely a refutation o f Speusippus, while
energeia o f N E , and the poiei o f A E and the teleioi o f N E both refer
the N E treatment puts forward a positive view o f Aristotle’s own.
to the agency o f a final cause.
B u t Festugière is mistaken both about the doctrine o f Speusippus
It is true that Aristotle prefers to speak o f the activity as being a
(which he identifies with the theory in N E 10 that pleasure is
pleasure in A E and o f the pleasure arising from the activity in N E :
absolutely bad, 1172*28 , despite Aristotle’s denial that Speusippus
but this does not m ark a difference o f doctrine. It simply means that
held this view at 1 i5 3 b8) and about the method o f Aristotle (in A E
when he talks o f ‘activity’ in N E he is often talking o f impeded
C 7 - 1 0 Aristotle is practising the policy recommended at the
activities as well as unimpeded ones. T h e pleasure supervenes on
beginning o f E E , and carried out throughout that work and the A E ,
the activity to the extent that if you say an activity is pleasant you
o f squeezing the truth out o f a discussion o f received opinions).
are saying more than that it is an activity o f a specific kind : but you
There is no denying that N E 10 contains much that is o f the highest
are not saying anything more than that it is a good, unimpeded,
philosophical interest concerning the nature o f states, processes,
activity o f that specific kind. It is no more something that could
and activities;1 but it cannot be said that either the study o f these
exist in isolation from the activity than the höra could be peeled o ff
metaphysical questions, or the discussion o f philosophical opinions
those in their prim e.2
on the morality o f pleasure, has been worked into Aristotle’s own
ethical argument with anything resembling the density and
1. So, convincingly, G au th ier-Jo lif, 1959, 839. economy o f the treatment in A E C . Worst o f all, the treatise in N E
2. G oslin g’s comment on this passage is excellent: ‘Perfect activity is not a condition for
the occurrence o f pleasure. Perfect activity is what pleasure is. B u t activity is not what on.’ T h e passage most difficult to reconcilc with A E is not i i 7 4 b24 ff., but i i 7 5 b33 ο ν μ η ν
pleasure is, since activity needs the addition o f perfection to be pleasure, which perfects €OiK€ y e ή η δ ο ν η δ ιά ν ο ια e fv a i ουδ’ α ΐσ θ η σ ις (ά τ ο π ο ν y à p ), αλλά δ ια το μ η
activity not as a condition already present, but in the way in which any perfection added to χ ω ρ ίζ ε σ θ α ι φ α ίν ε τ α ι τ ισ ι τ α ύ τ ό ν . But this is consistent with the A E view i f δ ιά ν ο ια and
something makes it perfect (1 i7 4 b3 1 —33). T h e point is repeated in the remark so α ΐσ θ η σ ις are taken, as throughout the previous discussion, to refer to a faculty and not to a
unfortunately translated in terms o f the bloom on the cheek o f youth. It is, indeed, difficult to particular exercise o f a faculty.
get a non-redundant translation, but the sense is that pleasure perfects in the way their prime i. I have studied some o f the philosophical lessons to be learnt from N E 10, 4 5, in
perfects those o f full age. Sim ilarly, validity makes an argument perfect, humour a joke and so K en ny, 1963, ch. 6.
238 The Aristotelian Ethics The Dating o f the Aristotelian Ethical Treatises 239
10 leaves the reader unclear about Aristotle’s own views on the arguments for a Eudemian origin for the common books. In
crucial points o f the relationships between pleasure, activity, and particular, I have no strong convictions about the original date of
goodness. A charitable student can tease out o f the text a doctrine the N E : indeed, given the lack o f evidence for the existence o f our
consistent in itself and consistent with the theory o f A E C : but Nicomachean Ethics until the second century a .d . it may well be
patient and learned readers have been left in doubt whether misleading to speak o f the date o f the N E at any point in Aristotle’s
Aristotle in N E 10 did or did not identify pleasure with the relevant career. Loosely connected as its books are internally, and
activity, did or did not reject the identification o f the supreme uncertainly evidenced as it is by external testimony, it could well be
pleasure with the supreme good. N o one has been left in similar a collection o f material from various periods o f Aristotle’s life which
doubt by the brief and lucid arguments o f A E C. did not exist in a collected form until the time o f Aspasius. Such a
In m y view, it is as impossible to settle the chronological conjecture, however, could be supported, if at all, only by further
relationship between the E E and the N E by comparing the close work on the style o f individual Nicomachean books as part of
treatments o f pleasure as it is to do so by com paring the criticisms o f an over-all investigation of the stylistic features o f the Aristotelian
the Ideal theory. All I have tried to do, in each case, is to show that corpus. Only such a study could produce a general scholarly
the arguments which claim to prove by these means that the E E is consensus about the chronology o f Aristotle’s writings resembling
early and immature are built upon sand. N o doubt, when finally that which, since the work o f Cam pbell and Lutoslawski, it has been
pressed, most scholars would say that their belief in the priority o f the good fortune of Platonic scholars to possess.
the E E to the N E rests not upon any particular argument but upon
their over-all impression o f the respective philosophical merits o f
the works. Such judgements are, o f course, partly a function o f
variations in the fashionableness of criteria for judging philosophical
m erit : they are also very much a function o f how closely a text has
been read, analysed, and meditated upon. It will only be when the
E E has been for some time as carefully and widely studied as the N E
has been for centuries that we shall be able to make an unclouded
judgement about their comparative worth.
T h e contribution o f the present work has been, I hope, to show
that such a study can only be fairly undertaken if the disputed books
are replaced within the Eudem ian framework. I have presented a
succession o f arguments, historical, philological, and philosophical,
to prove that they fit within that framework very much better than
they fit within the Nicomachean framework in which they have
been so long read. In addition, I have tried to show that the
historical and philosophical arguments for the comparative lateness
o f the N E and the comparative earliness o f the E E are inadequate
and should never have secured the almost unanimous assent o f the
learned world. I have put forward a hypothesis concerning the
dating and origin o f the E E and the M M which, I flatter myself,
explains the phenomena no worse than those current in the
commentaries and in the periodicals. B u t I do not attach to these
conjectures the weight which I believe should be attached to my

BIBLIOGRAPHY
A c k r il l , J. L ., Aristotle on Eudaimonia, London, 1974.
A l l a n , D. J., The Philosophy o f Aristotle, Oxford, 1952.
------, ‘Magna Moralia and Nicomachean Ethics’ , Journal o f Hellenic
Studies 77 (1957).
------, ‘Quasi-mathematical Method in the Eudemian Ethics’, in Mansion,
1961.
------, ‘Aristotle’s Criticism of Platonic Doctrine Concerning Goodness
and the Good’, Proceedings o f the Aristotelian Society 64 (1963-4).
------, ‘Review of Dirlmeier 1962’, Gnomon (1966).
------, ‘The fine and the good in the Eudemian Ethics’, in Harlfinger and
Moraux, 1971.
A n sc o m be , G. E. M., Intention, Oxford 1957.
A r b e r r y , A. J . , ‘The Nicomachean Ethics in Arabic’, Bulletin of the School
o f African and Oriental Studies 17 (1955).
A r k in , H., and C o lto n , R. R., Statistical Methods, New York, 1970.
A r n im , H. von, ‘Die drei Aristotelischen Ethiken’, S B Wien 202 (1924).
------, ‘Arius Didymus’ Abriss der peripatetischen Ethik’, S B Wien 204
(1926).
------, ‘Die Echtheit der Grossen Ethik des Aristoteles’, Rheinisches
Museum 76 (1927).
------, ‘Eudemische Ethik und Metaphysik’, S B Wien 207 (1928).
------, ‘Neleus von Skepsis’, Hermes 63 (1928).
------, ‘Nochmals die Aristotelischen Ethiken’, S B Wien 209 (1929).
------, ‘Der neueste Versuch, die Magna Moralia als unecht zu erweisen’,
S B Wien 209 (1929).
A r y . D. and J acobs , L. C., Introduction to Statistics, New York, 1976.
B en d ix en , J . , ‘Bemerkungen zum siebenten Buch der Nik. Ethik.’,
Philologus 10 (1885).
------, ‘Übersicht über die neueste des Aristoteles Ethik und Politik
betreffende Literatur’, Philologus 11 (1856).
B er v e , H ., Das Alexanderreich aufprosopographischer Grundlage, M u n ic h ,
1926.
B ig n o n e , E., L ’Aristotele perduto e la formazione ßlosofica di Epicuro,
Florence, 1936.
B l o c k , I., ‘The order of Aristotle’s Psychological Writing’, American
Journal o f Philosophy, 1961.
B tblw xraph y 241
B odeus , R., ‘Histoire des oeuvres morales d’Aristote’, Revue philosophique
de Louvain, 1973.
B r in k , K . O., Stil und Form der pseudaristoteltschen Magna Moralia,
Ohlau, 1933.
B y w a ter , I., Contributions to the Textual Criticism o f Aristotle’s
Nicomachean Ethics, Oxford 1892.
C au lco tt , E ., Significance Tests, London, 1973.
C hroust , A-H., Aristotle, London, 1973.
C l a r k , S. R. L ., Aristotle's Man, Oxford, 1975.
C ook W ilso n , J., On the Structure o f the Seventh Book o f the Nicomachean
Ethics, Chs. 1-10 , Oxford, 1879.
C o oper , J. M., ‘The Magna Moralia and Aristotle’s Moral Philosophy’,
American Journal o f Philology 94 ( i973)·
------, Reason and Human Good in Aristotle, London, 1975.
D evereux , D ., Review of Cooper, 1975, forthcoming, 1978.
D ie l s , H., Doxographi Greeci, Berlin, 1879.
------, ‘Zu Aristoteles’ Protreptikos und Cicero’s Hortensius’, Archiv für
Geschichte der Philosophie 1 (1888).
D ir l m e ie r , F., Aristoteles, Magna Moralia, übersetzt und erlaütert, Berlin,
1958.
------, Aristoteles, Nikomakische Ethik, übersetzt und kommentiert, Berlin,
Ï959-
----- , Merkwürdige Zitate in der EE, Heidelberg, 1962.
------, Aristoteles, Eudemische Ethik, übersetzt und kommentiert, Berlin,
: 969-
D rossart - L u lo fs ., J., Nicolaus Damascenus: On the Philosophy of
Aristotle, Leiden, 1965.
D ü r in g , I., ‘Ariston or Hermippus’, Classica et Medievalia 17(1956), 1 1 -
21.
------, Aristotle in the Ancient Biographical Tradition, Goteborg, 1957.
------, Aristoteles, Heidelberg, 1966.
------, Aristoteles (Sonderausgaben der Paulyschen Realencyclopädie),
Stuttgart, 1968.
E lleg ä rd , A., A Statistical Method for Determining Authorship,
Gothenburg, 1962.
E uCk e n , R., De Aristotelis Dicendi Ratione. Pars Prima, Göttingen, 1866.
------, Über den Sprachgebrauch des Aristoteles, Berlin, 1868.
F estugi Èr e , A. J., Aristote, Le Plaisir, Paris, 1936.
F ischer , A. M., De Ethicis Nicomacheis et Eudemiis, Bonn, 1847.
F lashar , H., ‘ Ideenkritik in der Ethik des Aristoteles’, in Synusia,
Festschrift Schadewalt, Pfulling, 1965.
F o rtenbaugh , W . W ., Aristotle on Emotion, London, 1975.
F r a g st ein , A. von, Studien zur Ethik des Aristoteles, Amsterdam, 1974.
242 The Aristotelian Ethics Ihbltography 243

F r it z sc h e , A. T . H., Eudemi Rhodi Ethica (Aristotelis Ethica Eudemia), J o hn so n , H., ‘ The use of Statistics in the Analysis of the characteristics of
Regensburg, 1851. Pauline Writing’, New Testament Studies, 1973.
F K . von, ‘E u d o x o s vo n K n id o s ’ , Philologus lx x x v (19 3 0 ), 4 7 8 - 8 1 .
r it z ,
K a p p , E., Das Verhältnis der Eud. zur Nik. Ethik, Berlin, 1912.
F u r le y , D., T wo Studies in the Greek Atomists, Princeton, 1968. K ea r n e y , J. J., ‘Two notes on the Tradition of Aristotle’s Writings’,
G adam er , H .-G ., ‘Der aristotelische Protreptikos und die entwick­ American Journal o f Philology 84 (1963), 52-63.
lungsgeschichtliche Betrachtung der aristotelischen Ethik’, K e n n y , A., Action Emotion and Will, London, 1963.
Hermes 63 (1928). ------, ‘Happiness’, Proceedings of the Aristotelian Society 66 (1965-6).
G a iser , K . ‘Zwei Protreptikos-Zitate in der E. E. des Aristoteles’, ------, ‘A stylometric study of the Aristotelian Corpus’ CIRPHO, autumn
Rheinisches Museum N .F. 110, 4 (1967). 1976.
G a u th ier , R. A., Introduction to Second Edition of l'Ethique à L eonard , J., Le Bonheur chez Aristote, Brussels, 1948.
Nicomaque, Louvain, 1970. L ie b e r g , G., Die Lehre von der Lust, Munich, 1958.
G a u th ier , R. A. and J o l if , J . Y ., L'Éthique à Nicomaque, Louvain, 1958 M an sio n , A., ‘La genèse de l’oeuvre d’Aristote d’apres les travaux
and 1959. récents’, Revue néoscolastique de la philosophie 29 (1927).
G o h lk e , P., ‘Die Entstehung der aristotelischen Ethik, Politik, Rhetorik’, M a n sio n , S., Aristote et les Problèmes de Méthode, Louvain, 1961.
S B Wien 223 (1944). M a r g u er itt e , H., Review of Jaeger, ‘Über Ursprung und Kreislauf des
G o sl in g , J. C. B., ‘More Aristotelian Pleasures’, Proceedings o f the philosophischen Lebensideals’, Revue d'histoire de la philosophie 4
Aristotelian Society (1973) 15 ff. ( i 93°)·
Gow, A. S. F., Machon, Cambridge, 1965. M a s e l l is , V., ‘T raduzione e cataloghi delle opere aristoteliche’, Rivista di
G r a n t , A., The Ethics o f Aristotle, London, 1857. filologia e d'istruzione classica 34 (1956).
G reenw ood , L . H. G ., Aristotle, Nicomachean Ethics Book VI, M auru s , Silvester, Collected Works, ed. F. Ehrle, ii, Regensburg, 1885-6.
Cambridge, 1909. M ein e c k e , A., ‘Zu Stobaeus’, Zeitschrift f.d. Gymnasialwesen 13 (1859).
G r if f in , M . T ., Seneca, a philosopher in politics, Oxford, 1976. M er c k en , H. P. F., The Greek Commentaries of the Nicomachean Ethics o f
H a ckfo rth , R., ‘Sicily, 367 to 330 b .c .’ in The Cambridge Ancient History, Aristotle in the Latin Translation of Robert Grosseteste, i, Leiden, 1973.
VI, Cambridge, 1927. M e r la n , P., Studies in Epicurus and Aristotle, W iesbaden, i960.
H a l l , R., ‘The Special Vocabulary of the Eudemian Ethics’, Classical M ich aelso n , M . and M orton , A. Q., ‘The Authorship and Integrity of
Quarterly n .s . 9 (1959). the Athenaion Politeia’, Proceedings of the Royal Society of Edinburgh,
H a n sen , B., Rückläufiger Wörterbuch der griechischen Eigennamen, Berlin, 1973·
M in g a y , J., ‘Coniunctio inter Homines Hominum’ in Stern, Hourani,
I957·
H a r d y , W. F. R., Aristotle's Ethical Theory, Oxford, 1968. and Brown, 1972.
H a r l f in g e r , D., ‘Die handschriftliche Überlieferung der Eudemischen M on an , J. D., Moral Knowledge and its Methodology in Aristotle, O xford,
Ethik’ in Harlfinger and Moraux, 1971. 1968.
------and M oraux , P., Untersuchungen zur Eudemischen Ethik, Berlin, M oraux , P., ‘L ’exposé de la philosophie d’Aristote chez Diogène Laerce’,
Revue philosophique de Louvain (1949).
1971-
H eath , T ., A History o f Greek Mathematics, Oxford, 1921. ------, ‘Das Fragment Θ, 1 ’ in Harlfinger and Moraux, 1971.
H i c k s , R . D ., Diogenes Laertius (L o e b edn.), H a rv a rd , 19 2 5 . ------, Les Listes anciennes des ouvrages d’Aristote, Louvain, 1951.
J ackso n , H ., The Fifth book o f the Nicomachean Ethics of Aristotle, ------, Der Anstotelismus bei den Griechen, i, Berlin, 1973.
Cambridge, 1879. M orton , A. Q. and W inspea r , A. D., It's Greek to the Computer,
------, ‘ On some passages in the seventh book of the Eudemian Ethics, Montreal, 1971.
Cambridge, 1900. M orton , A. Q. and M c L em an , J. J., Paul, the Man and the Myth,
J aeger , W., Aristotle: Fundamentals of the History o f his Development (trs. London, 1966.
Robinson), Oxford, 1948. M u h l l , P. von der, De Aristotelis Ethica Eudemea, Göttingen, 1909.
------, ‘Aristotle’s use of medicine as a model of method in the Ethics', M u l l e r , A., Die griechischen Philosophen in der arabischen Überlieferung,
Journal o f Hellenic Studies 1957, 58 ff. Halle, 1873.
244 The Aristotelian Ethics
M unro , H. A. J., ‘Nicomachean Ethics, Book V : Eudemian Ethics, Book
IV ’, Journal o f Classical and Sacred Philology 2 (1855).
M u r r a y , O., ‘Philodemus and the good King ’, Journal of Roman Studies
55 (1965)·
N ew m an , W. L., Aristotle’s Politics, Oxford, 1887-1902.
N u y en s , F ., L ’Evolution de la psychologie d ’Aristote, Louvain, 1948.
O w en , G. E. L ., ‘Logic and Metaphysics in Some Earlier Works of
Aristotle’ in Aristotle and Plato in the Mid-Fourth Century, Goteborg,
i960.
------, Tithenai ta phainomena in Mansion, 1961.
------, The Platonism of Aristotle, London, 1965.
------, ‘Aristotelian Pleasures’, Proceedings o f the Aristotelian Society 72
(19 7 1-2 ).
P aret , R ., ‘Notes bibliographiques sur quelques travaux récents’ ,
Byzantion 29/30 (1959/60).
Aristotle and the Arabs, N e w Y o r k , 1968.
P e t e r s , F . E .,
R a bin o w it z , W. G., Aristotle’s Protrepticus and the Sources o f its
Reconstruction, Berkeley, 1957.
R ackh am , H., The Eudemian Ethics (Loeb edn.), London, 1935.
R e e s , D ., ‘ M ag n an im ity in the E E and N E ’ , in H arlfin g er and M o ra u x ,
1971.
R ist , J. N., Epicurus, an Introduction, Cambridge, 1972.
R ose , V., De Aristotelis librorum ordine et auctoritate, Berlin, 1854.
------, ‘Uber die griechische Kommentar zur Ethik des Aristoteles’ ,
Hermes 5 ( 18 7 1) .
Ross, W. D., Aristotle, London, 1923.
------, The works o f Aristotle (Oxford translation), vol. ix, Oxford, 1925.
------, Aristotle’s Metaphysics, Oxford, 1924.
R o w e , C. J., The Eudemian and Nicomachean Ethics: A study in the
development of Aristotle’s Thought, Cambridge, 1971.
S chacher , E. J., Studien zu den Ethiken des Corpus Aristotelicum,
Paderborn, 1940.
S ch leierm ach er , F ., Über die ethischen Werks des Aristoteles,
= Samtliche Werke, iii.3, Berlin, 1835.
S olom on , J ., ‘Translation of the Eudemian Ethics’, The Oxford Aristotle,
ix, Oxford, 1925.
S p e n g e l , L., ‘Über die unter den Namen des Aristoteles erhaltenen
ethischen Schriften’, Abhandl. der Bayer. Akademie 3 (1841 and 1843).
------, ‘Aristotelische Studien’, Abhandl. d. Bayer. Akademie, Munich, 10
(1864) and (1866).
S t e r n , S . M., H o u ran i , A. and B row n , V., Islamic Philosophy and the
Classical Tradition, London, 1972.
Bibliography 245
S tew a rt , J. A., Notes on the Nicomachean Ethics o f Aristotle, Oxford,
1892.
S to ck , St. G., Introduction in Ross, 19 25.
S u se m ih l , F., Eudemii Rhodii Ethica ( Aristotelis Ethica Eudemia),
Leipzig, 1884 (Amsterdam, 1967).
------5 ‘Die Lebenszeit des Theodektes,’ Rhenisches Museum, L IV .
T h e il e r , W ., ‘Die Grosse Ethik und die Ethiken des Aristoteles’, Hermes,
69 (1934)·
T it z e , F. N., De Aristotelis Operum serie et distinctione, Leipzig, 1826.
V er b e k e , G., ‘L ’idéal de la perfection humaine chez Aristote et l’évolution
de sa noétique’ , Fontes Ambrosiani, 25 (19 5 1).
W a g n er , D ., Das Problem einer theonomen Ethik bei Aristoteles,
Heidelberg, 1970.
W ak e , E., ‘Sentence-length Distributions of Greek Authors’, Journal of
the Royal Statistical Society, i960.
W e b b , P., ‘The relative Dating of the Accounts of Pleasure in the
Nicomachean Ethics’, Phronësis, 1977.
W ebst er , C., ‘Fourth Century Tragedy and the Poetics’, Hermes 82
(1954)·
W alzer , R., Greek into Arabic, London, 1984.
------, Magna Moralia und Aristotelische Ethik, Berlin, 1929.
W e h r l i , F., Die Schule des Aristoteles, Basel, 1959 ~.
W id m a n n , G. Autarkie und Philia in den Aristotelischen Ethiken,
Tübingen, 1967.
W ilam o w itz - M o ellen d o r ff , U. von, ‘Neleus von Skepsis’, Hermes 62
(1927).
 INDEX

Absalom , 37 bloom on cheek o f youth, 236

abuse, 185—6 BodéuH, K M 16, 18, 42, 240
A ckrill, J . L ., 104, 240 Boulesis, 168
action and passion, 61 bouleusts, 1 66
activity (energeia), 68, 1 4 1, 235 Brink, K . 0 -, 1, 9, 10, 2 2 0 - 1, 240
Adrastus o f Aphrodisias, 37 Brown, L ., 200
adverbs, 1 1 5 —17 B yw ater, I., 48, 75, 241
ais thesis, 17 0 -2
akriheia, 37 Cam pbell, L ., 235
Albertus M agnus, 6 1, 204 Carcinus, 224
Alexander, 2 16 , 2 18 , 2 2 3 -5 Caulcott, E ., 87, 241
Alexander o f Aphrodisias, 37 certainty, expressions of, 119
Allan, D ., 2, 9, 10, 53, 57, 1 8 1, 18 5 , 200-3, Chareonea, 2 18
2 2 5 -6 , 240 chi-square test, 9 0 -2 , 95
anaisthësia, 2 3 1 choice, 14 6 -7
analogy, 64 Chroust, A -H ., 17, 2 16 , 241
Anaxagoras, 194, 221 Cicero, 1 5 - 1 7 , 32
Andronicus, 17, 18, 27 Clark, S., 204, 241
anger, 42 Clearchus, 2 17
animals, 16, 57-8 dementia, 25
Anscom be, G .E .M ., 2 33, 240 Clement o f Alexandria, 38
Antileon o f Metapontum, 221 C O C O A , 75, 123
apëllagmena, 232 contemplation, 5 1, 163, 17 4 -9 , 208
A quinas, Thom as, 204 continence, 52, 55-5
A rberry, J ., 22, 240 contingency, 14 4 -5
Archelaus o f Heraclea, 221 contraries, 19
Archiles, 281 Cooper, J ., 1, 1 0 ,2 0 1 ,2 0 9 - 10 , 2 14 , 2 19 ,
A rius Didym us, 12 , 14 , 17 , 19 -2 2 , 28 Coriscus, 221
aretë, 16 4 -5 , 19 8 -9 , 205; see virtue Coronea, 224
Ariston, 39 correlation, measures of, 7 9 -8 1
Arnim , H . von, 1, 9, 13 , 14, 19 -2 0 , 2 16 - 18 , courage, 24
240 cunning, 2 14
article, definite, 12 7 -8
Aspasius, 5, 15 , 2 9-36, 5 1 - 2 , 61 D em ades, 2 18
Assos, 216 , 220 D arius, 2 17
Athenaeus, 37, 2 2 1, 222 degrees o f freedom, 92
A tticus, 36 demotes, 183, 188
autarkeia, 200 deliberation, 166
Dem etrius Poliorcetes, 224
De Bona Fortuna, 125 D evereux, D ., 209, 214 , 241
Bekker, I., 47-8 dianoia, 169
Bendixen, J ., 3, 240 D iels, H ., 17 , 2 1, 14 1
Bignone, E ., 15, 240 dignity, 233
binomial distribution, 1 1 1 Diogenes Laertius, 26-9, 32, 40
binomial events, 85, 135 D ionysius o f Halicarnassus, 2 16
B IV A R , 82 D ionysius o f Syracuse, 2 17 , 2 2 1, 223
 Index *49
248 Index
K eb le, J ., 7 1 non), i Mi Ho, 205
D irlm eier, F ., i , 10, 17 , 57, 1 8 1 , 185, 199, Gauthier and Jo lif, 25, 38 ,4 8 -9 , 6 7 ,2 0 9 -10 ,
K en n y , A ., 159, 237, 242 null hypolhcniN, 87
2 13 , 2 16 , 222, 227, 2 3 1, 241 233, 242 knowledge, 164, 170 Nuycnn, 1''., 3, J43
distinctiveness ratio, 12 9 - 3 1 glory, 15 7
distributions, n i , 135 God, 25, 174, ϊ 75, 17 6 - 7
dominant vs. inclusive ends, 204 Gohlke, P ., I , 242 ( ),
learning sophia 16 2 - 3 , 166, 168, 1 7*), 210 oreklikim, 167
Owen, U . K. I.., 3, 193, 203, 2 3 3 -6 , 243
doxastikon, 52, 166 Good, Idea of, 197, 202-3 lemmatization, 76, 12 3
O xyrynchus papyrus, 48
D rossart-Lulofs, J ., 23, 241 goods, classes of, 2 1, 25, 6 4 ,18 3 ,19 7 - 8 , 202 Leonard, J . , 162
D iiring, I., 1 , 10, 17 , 18, 33, 39, 40, 4 2 - 3 , Gosling, J . , 234 -6 , 242 L ieb erg, G ., 233, 242
I’ ammcnes, 222
2 16 , 2 19 , 241 gourmand, 8-9 life, ways of, 148, 19 2 -5 , 200
Paret, R ., 22, 243
Grant, A ., 33, 5 1 - 2 , 66, 7 1, 2 3 0 - 1, 242 lists o f Aristotle’s works, 39-46
particles, 7 0 - 1 1 1
Griffin, J . , 233 logical terminology, 139 -4 0
economics, 54 parts (of justice, virtue, happiness), 66, 1 2 5 -
Griffin, M ., 25 logistikon, 16 5 -6 , 168, 1 7 1 , 175
Elias, 17 6
grouping indicator words, 1 2 8 - 3 1 , 154 logos, orthos, 164; see right reasoning
Ellegârd, A ., 75, 129, 136, 241 Pearson, Κ ., 79
Lucian , 37
energeia, 68, 193, 19 9 ; see activity perfection, 67, 14 1, 200
happiness, 3, 7, 25, 5 1 , 53, 57, 66, 148, 19 0 - Lyo n s, J ., 22
Epicurus, 15 Peters, R ., 23, 243
2 13
epeikeiu, 25, 158 Pettit, F ., 82
Harlfinger, D ., I , 34, 46, 47-9, 202 M agna M oralia, 1, 9 - 12 , 14, 1 6 - 1 7 , 2 0 - 1,
epistëmë, 17 0 ; see knowledge Phanias o f Eresus, 221
health, 62-4, 17 9 ; see medicine 23, 29, 2 16 -2 0 , 2 2 5 -9
epistëmonikon, 166, 1 7 1, 17 4 -5 Philip o f M acedon, 222
Heracles, 222 magnanimity, 2 3 2 -3
epithymëtikon, 167 Philo, 2 4-5
H ermeias, 216 M ansion, A ., 14, 242
equality, 6 Philoponus, 39
H erm ippus, 27, 39, 46, 47-9, 202 manuscripts o f Aristotle, 1 , 2, 46-9
ergon, 16 4 -5 , 1 9 1, 198-9, 202 Philoxenus, 8, 24, 2 21
H esychius, 39, 43 M argueritte, H ., 40, 42, 162, 243
Erianthes, 2 17 Phocylides, 15
hexis vs. energeia, 19 3, 199, 207, 210 M asellis, V ., 42, 243
ethikë arete, see virtue, moral phronësis, 2, 3, 1 6 1 ; see wisdom
H eylbut, G ., 3 1 M aurus, S., 1, 243
euboulia, 1 7 1 - 2 phytikott, 166
hiatus, 105, 225 mean, 6, 7, 13 , 19, 24, 5 1 , 149, 2 1 3 - 1 4
Eucken, R ., 7 2 -3 , 241 Plato, 1 - 3 , 25, 16 7 -8 , 178, 197, 200, 203,
Hicks, R . D ., 29, 242 medicine, 6 2 ,1 7 2 ,1 7 6 , 17 9 -8 2 , 18 6 - 7 ,2 0 8 -
eudaimonia, 204; see happiness 209, 2 19 , 225, 230
Hipparinus* 2 21
Eudem us, 1 , 1 8 , 29, 30, 35, 39 9 pleasure, 4, 7, 26, 3 0 - 1, 53, 56-8, 147, 194,
homonymy, 64 M enander, 16
Eudoxus, 2 1, 29, 224 196, 20 6 -7, 2 1 1 - 1 2 , 233-8
horos, 25, 18 0 -3 , 207, 2 12 M ercken, H ., 32, 243
Eusebius, 36 Plotinus, 22
hou heneka, 17 4 -7 , *97 M erland, P ., 15, 243 Plutarch, 26, 2 2 1, 223, 224
Eustratius, 204
hylomorphism, 3 M entor, 2 16
Euthyphro, 178 Poisson distribution, 76, h i
hypothesis metaphysical term inology, 14 1
evaluative expressions, 149 Politics, 3, 6, 43
eyes, 185 Metaphysics, 3, 6, 230 politikë, 54, 16 4 -5 , 200
Ibn A bi U saibia, 17, 39 M etapontum, 221
exoteric works, 1 6 - 1 7 population, 85, 103
Ibn A l-N adim , 38 methodology, 62 Porphyry, 38
Ibn al-Qifti, 17, 39 M ichael o f Ephesus, 223 potentiality, 145
fam ily, 15 3 Ideas, 231 M ingay, J ., 20, 243 practical reasoning, 4, 109, 1 7 0 - 1
Favorinus o f A rles, 26-7 incontinence, 26, 42, 186, 17 3 M onan, J . P., 2, 3, 243 praotes, 232
F ez M S , 22 India, 225 M oraux, P ., 14, 17 -2 0 , 23, 28, 39, 5 1, 185, prayer, 66
Fischer, A ., 60, 241 insensibility, 14, 232
Flashar, H ., 2 3 1 , 241 243 prepositions, 1 1 3 - 1 5
intelligence, 18 3, 188 M orton, A ., 85, 159, 227, 243 prim a-facie goods, 64
focal meaning, 3 intemperance, 8 M iiller, A ., 38, 243 probability, 87
fortune, 16 intuition, 17 0 -2 M unro, H ., 243 Problems, 8, 9, 26
Fragstein, A . von, 1, 5 1 ,5 4 , 57 ,6 0 , 185, 209,
M urray, O ., 25 Prodicus o f Selym bria, 223
241 Jackson, H ., 48, 60, 185, 222
product-moment coefficient, 79
frequency, 82 Jaeger, W ., i, 3, 4, 9, 10, 1 7 - 1 8 , 50, 63, 73,
nature, 145 ; see virtue prohair esis, 188-9
friendship, 28, 33, 227-9 18 5 , 207, 2 16 , 242
Neleus o f Skepsis, 2 17 pronouns, 12 3 - 6
Fritz, K . von, 241 Johnson, H ., 15 9 , 242
Fritzsche, A. von, 1, 9, 5 1, 60, 241 nemesis, 24, 61 Protrepttcus, 2, 16, 17, 2 12
justice, 6, 5 1 , 55, 60-69
N ewm an, W ., 8, 243 psychological terms, 145, 166
Fu rley, D ., 15 , 241
Nicolaus o f Dam ascus, 22 P tolem y-el-G arib, 17, 39
Kalokagathia, 2 0 - 1, 65, 67, 182, 18 3, 206
Nicolaus o f Laodicea, 23 Ptolemy Soter, 223
G adam er, H -G ., 162, 241 K ap p , E ., i, 56, 2 0 1, 242
Nicom achus, 16 , 29, 32, 33, 39 Python, 222
G aiser, Κ ., 227, 242 Kearney, J ., 39, 242
250 Index
Rabinowitz, G ., 3, 17 , 244 Susem ihl, F ., 48, 75, 185
Rackham, H ., 65, 197, 222 syntax, 159
R ees, D ., 232, 244 syntheton, 169
Regenbogen, E ., 15
regression lines, 182 telos, 100, 1 9 1, 19 7 ; see happiness
Rhetoris, 25 temperance, 5 5 -6
riches, 2 1 , 35* 65 tentativeness, r ig
right reasoning, 54, 1 8 0 - 1 , 184 Theiler, W., 10, 244
R ist, J . , 15 , 244 Them istius, 38
Rose, H ., 34, 42, 244 Theognis, 15
R oss, W . D ., 40, 65, 17 7, 2 1 1 , 244 Theophrastus, 1 2 - 1 4 , 22
Rowe, C ., i, 3, 9, 27, 60, 70, 96, 162, 1 8 1, theoria, 174, 206; see contemplation
19 1, 20 0 -2, 244 therapeia, 19 8 -9
Thesaurus Linguae Graecae, 75
samples, 85, 90 threptikon, 166
Sardanapallus, 17, 194 throat, pleasures of, 9
Satyros, 224 tkymoeideSy 167
saying, verbs of, 138 Tim oleon, 222
Schächer, E ., 54, 70, 244 Titze, F ., 27, 40, 244
Schleierm acher, F ., i, 9, 50, 2 18 , 244 topic-neutral ity, 13 7
Scholiast, 36 -8 Topics, 42
Scythia, 225 triads, 22
self-consciousness, 27 truth, 164, 195
semnotes, 233 type vs token, 75
Seneca, 25, 224
sense o f words, 64 Urm son, J ., 200
sensitive appetite, 167 use o f hexisy 7, 24, 68-9
service o f G od , 17 7 -8
shame, 24, 6 0 -1 Verbeke, G ., 3, 203 4, 244
Sim plicius, 19 virtue entire, complete, 54, 66 -7, 19 1, 203,
singing, 2 1 3 205
sociability, 232 virtue, innate, 25, 57, 184, 189
Socrates, 24, 178, 184, 187, 196 virtue, intellectual, 52, 16 3 - 18 0
sophia (learning), 1 6 2 ,1 6 6 ,1 6 8 - 9 , *79» 209- virtue, m oral, 5 1 - 3 , 65, 167, 18 0 -6 , 106
10 virtues, lists of, 13 , 23-4 , 232
soul and its parts, 14 7, 16 3-6 , 168-9, voluntariness, 55, 148
187
Spearm an’s rho, 7 7 -9 W agner, D ., 174, 244
Spengel, L ., 1, 9, 12, 5 1 , 57, 185, 2 32, 244 Wake, E ., 227, 244
Speusippus, 66, 224, 237 W alzer, R ., i , 10, 22, 38, 244
speculation, 174 W ebb, P ., 233, 237, 244
Stagira, 223 W idm ann, G ., 227, 244
standard error, 84-6, 103 W ilamowitz, U ., 2 17 , 245
statistical significance, 88 will, 168
Stew art, J . , 53, 244 wisdom, 19 , 5 2 -4 , 1 6 1 - 8 1 , 2 0 9 -10 , 2 1 3 - 1 4
Stobaeus, 12 , 17, 19, 2 1 W ittgenstein, L ., 270
Stock, S t., 2, 244 W olfson, H ., 25
Stratonicus, 223
Suda, 39 Xenarchos, 19
superior, 17 3 - 7 Xenophantus, 223