Educ Stud Math (2007) 66:243-255 DOI 10.1007/s10649-007-9079-z Mathematical history, philosophy and education Michael Otte Published online: 27 March 2007 © Springer Seience + Business Media B.V, 2007 Abstract History of mathematics occupies itself describing processes of growth and development, whereas philosophy of mathematics is concerned with questions of justification. Both play an essential role within the educational context. But there is a problem because genuine historical studies necessitate ever greater particularity whereas mathematics and philosophy require generality and abstraction. The paper offers some methodological reflections about these matters together with two case studies from nineteenth century history of arithmetic and integration theory, respectively, which try to strike a balance between the directly opposed requirements Keywords Axiomatization - Arithmetic - Integration - Lebesgue - Grass mann - Mathematical generalization 1 Introduction One great problem of mathematics education is the seemingly static and infallible character of mathematical knowledge. Everything just is and thus means itself: P=P! This principle of identity ties at the heart of logic or exact science, and it is obviously directed against any historical or evolutionary concerns. P just means P! No comment or historical investigation, no psychological or philosophical consideration shall beable to add anything to the matter. Mathematics seems so immutable and absolute that the sociology and socio-cultural history of knowledge excluded it from their considerations since the very beginning (see e.g. Mannheim 1929). Within this context, it is frequently claimed, by mathematicians in particular, that mathematics has no history worth knowing. The newest state of the art of mathematics has taken up and reformulated in modern terms whatever appeared as worthwhile during its history. Any further concentration on the seesaw changes of history are of no use at all, mathematicians say, because all mathematical cognition conforms to the paradigm of contemporary mathematics, such that if there is a history at all, it is more of a pastime, that M, Otte (=) Institut fur Didaktik der Mathematik, University of Bielefeld, universitaetsstrasse, Bielefeld 33501, Germany e-mail: Michaelontra@aol.com © Springer244 M. Otte is, the history of mathematics is partly a dogma and partly gossip or small talk. The dogmatic attitude serves to exclude all alternative views of mathematics, whereas the small talk classifies them as off the record private ponderings Looking at mathematics in this way, however, leaves it as a set of completed works and finished theories that might sometimes reveal their secret beauty to the talented discoverer, but that could not be taught nor learned. Being a mere form of reality, or a reality sui generis, it has nothing to do with human activities or emotions. Such a view does not allow, for example, the consideration of unresolved problems. This is not good, because great problems and programs of their investigation amount to the largest part of the “real” history of mathematics. And it does not help to stimulate the spirit of creativity and truth in students. Yet the historical perspective on mathematics is essential to a spirit of truth and creativity. For instance, to perceive the changes in a thing certainly helps us to see it more clearly. The Continuity Principle, for example, in its manifold guises, has been the most important vehicle of mathematical generalization throughout the modem ages; it became dethroned only when a reductionistic spirit of “tigor” took hold. ‘One reason for this banishment or expulsion of the continuity principle and a motif for an illusionary search after an a-historical absolute rigor was that this principle had not always been well understood in terms of the relevant practices. Continuity is something, ideal; there is no perfect “uniformity of Nature,” such that conceiving of change always requires constructive efforts and depends on epistemological views. This brings me to a different kind of caveat, Historical considerations are often mere negative reactions to mathematical dogmatism and positivism, trying to use history as a means to debunk the assumption of a linear and logical development of mathematics as an illusion and sometimes meandering into abstract ideological criticisms. Mastery, Valéry says, presupposes that “one has the habit of thinking and combining directly from the means of activity, of imagining a work only within the limits of the means at hand, and never approaching a work from a topic or an imagined effect that is not linked to the means” (Valéry 1960, p. 40). The historical approach therefore cannot be justified by just pointing out “human” concems or interests. And even the broadest socio-cultural approach to the history of mathematics has to model its principal categories so as to make them applicable to the mathematical and cognitive questions at hand. We should hence aim at a historiography of mathematics where the facts are being illuminated by epistemological and cultural reflection Lakatos used to famously paraphrase Kant: “The history of mathematics, lacking the guidance of philosophy, is blind, while the philosophy of mathematics, turing its back on the most intriguing phenomena in the history of mathematics, is empty” (Lakatos 1970, p. 135). Differing from Lakatos, we do not believe, however, that a particular philosophy of mathematics could be “proven” right from history or that there is a completely “rational reconstruction” of the historical developments. For instance, there is nowadays much controversy conceming the use of the word “real” with reference to mathematical objects. No historical study can resolve these controversies and it would not even be the purpose of such studies. Their purpose is first of all to describe and analyze attitudes, rather than justify them. Unlike philosophers or educators historians are unlikely to be focusing upon justification. “Reality” is a culturally laden concept and is certainly more than a set of material facts or objects. Positivists were therefore suspicious of mathematics because the objects of mathematical study were not material. And mathematical positivists opted for formal definitions and rigorous proofs to escape such suspicions. We believe that it is this question that is addressed when Thom affirms that “the real problem which confronts mathematics teaching is not that of rigor, but the problem of ®@ SpringerMathematical history, philosophy and education 245 the development of meaning, of the ‘existence’ of mathematical objects” (Thom 1973, p. 202). And to tackle this problem, students should experience processes of generalization and abstraction, rather than being confronted with the abstract as such. We shall present in Sections 3 and 4 two historical case studies that are meant to illustrate and substantiate the following general hypothesis. During the nineteenth century there appeared a number of problems and certain transitions occurred, based on new theoretical ideas about the nature of mathematics and its objects. These developments ~ that may be cast in terms of a transformation of Euclidean axiomatics into set theory and formal axiomatics — made mathematics more abstract and less “explanatory.” It seems a curious historical fact indeed that abstract pure mathematics arose not least from the necessities of large scale or distant communication, the irony being that its formal character simulta- neously facilitates and impedes learning and communication. Learning difficulties result not least from the fact that the concept of “explanation” is central to our educational practices and aims, whereas modern science and mathematics do not provide explanations of anything in the sense desired. They are either too hypothetical and abstract or too instrumental and technical. Mathematics could not, however, be fruitfully organized and pursued at school or university as a primarily professional topic. Mathematical education, like other subjects, also must contribute to a common search for clarity on fundamental issues. It might be concluded therefore that axiomatics in the sense of Euclid, i.e. “logical reduction, analysis and organization of intelligible concepts and meaningful sentences, seems to remain an irreducible, fundamental too! of our thinking” (Casari 1974, p. 61). This does not mean that everybody would want to go back to the style and methods of Euclidean mathematics nor stimulate a mathematical fundamentalism that aims to reduce everything to elementary mathematics. But it does show, however, that modern mathematics has been divided in its epistemology and methodology since the beginning of modemity. The previous ontological developments concerning the nature of mathematics and mathe- matical objects were underpinned by epistemological changes whose understanding is important if we want to employ the history of mathematics for educational purposes. There ‘was a shift from direct or constructive approaches to indirect and analytical ones and from instrumental reasoning to relational thinking. In the next section I deal with this episte- mological shift. 2 Analytical and operational thinking The profound transformation of mathematics during the nineteenth and twentieth centuries that occurred under the banner of modern axiomatics in the sense of Hilbert or Peano has been one ‘of the most intriguing phenomena in the history of mathematics. In the nineteenth century mathematics started to be considered as “the science which draws necessary conclusions” (Peirce, CP 3.558). According to this view mathematics is free from cxistential concerns about its constructs. It is hypothetico-deductive reasoning. In its previous sense, axiomatics referred to what was considered as self-evidently true. Since the nineteenth century, the foundations of the mathematical edifice turned into mere hypotheses to be justified by their possible consequences. As Peirce put it, mathematics is concerned with ideal states of things; it is one of the ways we systematize and analyze our thoughts. It is even a process that reflects on itself. Mathematics is in a sense meta-mathematics or meta-knowledge. Peirce himself, contrary to the majority of philosophers and mathematicians, however, did not endorse a completely analytical ideal of mathematics, but rather conceived of mathematical relations as establishing objective possibilities within a dynamical reality. D Springer246 M. Otte Even the analytic attitude brought by the modem axiomatic movement is an activity and as an activity it requires “objects” to act on. There are always unexpected facts and things without an explanation (even in pure mathematics as Chaitin (1998, 54) has emphasized, for example), contrary to Leibniz’ principle of sufficient reason and in contrast to his belief that essence precedes existence. We should therefore not exaggerate the analytical attitude, as if one could foresee all the consequences of one’s views or assumptions. Mathematics, I want to suggest, can be better conceptualized as a recursive interaction between relational and operative forms of reasoning. Operative thinking and a functionalist perspective introduced by this operativity were what made possible the transition from the mathematics of Antiquity to the algebraic outlook beginning with Descartes. A complementary aspect of this process, which was equally indispensable, may be called relational thinking. Leibniz thought that truth is constituted by proof; Descartes believed in evidence and thought proof irrelevant to truth. Leibniz believed in relations and theories, Descartes in facts and in the instrumental aspects. of problem solving. It is the relational thought system that allowed man to develop mathematics and science and that became dominant at the beginning of the nineteenth century, when algebra was transformed from a “language” into a science of structures. The operative conceptual schemata themselves, in a way, had to become an object of thinking. Algebra is meta-algebra, it is “algebra on algebra” as Sylvester (18141897) once remarked with particular reference to the algebra of determinants. Number could have been turned into a language for all of mathematics already during the seventeenth century, but “arithmetization” became a program during the nineteenth century only after the algebraic perspective had sensitized mathematics to the structural view (Otte and Jahnke 1981). There have therefore been two different trends in the foundational debate of mathematics during the nineteenth century, each of which emphasized nearly exclusively one of two fundamental cognitive operations: the search for similarities and relations or the drawing of definite distinctions. The arithmetizing program, being of the latter king and based on rigor, searched to solve the foundational problems in a reductionistic manner, by defining all mathematical concepts in terms of some basic reified entities, ultimately the natural numbers. Complex numbers, for example, were for Cauchy nothing but pairs of real numbers. The axiomatic movement, in contrast, as anticipated in the work of Poncelet or Grassmann, tried to employ, 50 to say, a synthetic top-down strategy, solving the foundational problems of mathematics by extending and generalizing its relational structures and its rules of inference, thereby enlarging the applicability of its theories. These two views, the synthetical and the analytical, have always more or less made up the essence of mathematical activity (Otte and Panza 1997). Mathematical activity thus occurs by a sort of recursive interaction between reification and generalization. Mathematical objects are but hypostatic abstractions. “Hypostatic abstraction” means that a thought becomes the object of another thought. This is very fundamental for modem mathematical thinking (see for example Otte 2003, pp. 218-220). Thurston describes it under the label of mental compression Mathematics is amazingly compressible: you may struggle a long time, step by step to work through some process or idea from several approaches. But once you really understand it and have the mental perspective to see it as whole, there is often a tremendous mental compression (Thurston 1990, p. 847). So mathematics is creation or production and depends in its development on hypostatic abstraction, that is, on abstraction from action, rather then from objects. Abstraction must, © SpringerMathematical history, philosophy and education 247 however, be accompanied by the inverse process of interiorisation of the explicit; a process which tums the formerly explicit and detached into a means of the subject's feeling, thinking and acting. The axiomatic method of Hilbert and Noether is nothing but the “highest point” of relational thinking in mathematics and it began to enter first year university classrooms in Germany some years afler World War II. I remember the enthusiasm with which we received the first edition of Graeub's Lineare Algebra (Linear Algebra) published in 1958 with its coordinate free treatment, after having been accustomed to the tedious and clumsy calculations in terms of coordinates and matrices of the older books. But the weaker or more conservative students and those from physics did not readily follow Gracub’s axiomatic and structural presentation, It is not quite obvious what caused the principal difficulties. It seemed, however, that those students did not really believe in the objectiveness of conceptual arguments or proofs (remember Thom’s verdict). These students wanted direct calculations and elementary proofs, that is, proofs that were maximally “self-contained.” Such proofs should reveal a theorem to be true by the light of the very terms that contain it, analytically true. No conceptual constructions or additional intuitive hypotheses should be required. Skemp had called this type of thinking instru- mental understanding, and had contrasted it with what he called “relational understanding” (Skemp 1976). “Relational thinking” is the notion by which modern mathematics and science has been characterized, since Cassirer's famous book Substanzbegriff und Funktionsbegriff (Sub- stance and Function) of 1910. And relational thinking is the great obstacle of everyday knowledge, which tends to identify knowledge with reality. Science and mathematics, in contrast, teach us how to look at reality and how to analyze it. What the history of mathe- matics may teach us, then, is the spirit of truth, rather than absolute mathematical truth. In the remaining sections, we shall present some illustrative case studies of controversies and difficulties of mathematical developments, seen as the interaction of analysis and synthesis. 3 Numbers: a-priori objects or human inventions? The axiomatic method, its changes and the debates around it, shows perhaps more clearly than any other single item the transformation mathematics has undergone during the last two centuries or so, Up to the middle of the nineteenth century, mathematicians were for the most part divided according to whether mathematics was supposed to deal with real meanings, like geometry or mechanics, or whether it was the result of our own mental constructions, like arithmetic. On April 9th of 1830 Gauss wrote to Bessel, for example: “We must modestly admit that whereas number is merely the product of our mind, space has a reality outside our mind as well, to which we cannot a priori prescribe its laws” (quoted from Otte and Jahnke 1981, p. 30). ‘These differences gradually disappeared when it was realized that the application of algebra to geometry can be seen as the algebraization of geometrical or mechanical constructions, rather than as a mathematical description of objects. The dominant focus of concer shifted away from the “interface problems" between knowledge and the external world and moved towards the problem of the intemal dynamics of knowledge and cognition. Mathematicians began to reflect more profoundly about their own constructions and activities, rather than about the givens of the external world. As Norbert Wiener writes: “To us, nowadays, the chief theme of the mathematicians of the Romantic period may sound most unromantic and repelling. The new mathematics ® Springer248 M. Otte devoted itself to rigor. ... What the new generation in mathematics had discovered was the mathematician; just as what the Romantics had discovered in poetry was the pot and what they discovered in music was the musician” (Wiener 1951, p. 96). We know that the axiomatization of numbers began about two thousand years after Euclid’s axiomatic presentation of geometry, with the publication of Hermann Grassmann’s small textbook Lehrbuch der Arithmetik of 1861. The traditional recursive definitions of addition and multiplication are due to Grassmann: x+0= x'0 sx + (yt 1) = ety) +15 O;x*(y + 1) = (ety) Fx. In this way addition and multiplication of natural numbers are derived from one single operation: x+1. By using these definitions Grassmann proved all the other principal properties of arithmetical operations (associativity, commutativity ete.). Grassmann’s axioms included also the induction principle. Grassmann’s exposition essentially corresponds to the characterization “which is customary in present day abstract algebra” (Wang 1970, p. 70). It ‘met, however, with strong criticism at its time. Hilder (1859-1937), for example, believed that arithmetic, being a human construction through and through, different from geometrical knowledge, is essentially based on some fundamental and elementary synthetical conceptions, which are characteristic of the human mind. Hélder objected to Grassmann’s axiomatic presentation by claiming “that the situation here is different from that in geometry where the fundamental notions and principles do not result from the processes of mathematical deduction” (Hélder 1892, p. 591). According to Hilder, statements referring to recursive operations are not axioms because arithmetic is a priori and not objective. He continues, If one wants to accept these formulas as arithmetical axioms, one would be led to introducing similar axioms for countless notions of number theory and analysis and the number of arithmetical axioms becomes infinite (Hélder 1892, p. 592). This is a rather ambiguous statement. It could mean that Grassmann’s treatment is incomplete as it did not treat the difficult part, showing that the formulas in question are in fact necessary and sufficient. But as things are, it is more appropriate to say that Hélder had not really appreciated the new understanding of mathematical axiomatics in the sense of Hilbert (or Grassmann), namely as being something hypothetically postulated and to be evaluated by its logical and mathematical consequences, rather than expressing some fundamental objective truths, given by intuition. What one does in fact is to compress one’s computational experience into arithmetical axioms. Definitions and statements of constructive arithmetic are statements about rules of construction. The axioms are, however, not complete descriptions that would quasi-mechanically generate all the truth about numbers. ‘They are more like partial views of this area of human experience and activity. Axiomatic schemata never characterize definite objects but characterize concepts or classes of objects. There is some vagueness and generality here. The axiomatic method is a kind of algebra, and in algebra, the opposition between the synthetic and analytic, or between determinism and indeterminism represented by the pair operation and variable, is driven to the extreme. Formulas like, a+b=b+a or similar ones express an analytic approach, whereas recurrent definitions represent synthesis. Hilder disagreed in principle with the axiomatic approach because it cannot reach or establish the completely definite and particular, but remains a calculus or an algebra of © SpringerMathematical history, philosophy and education 249 general relationships. In his “Antrittsvorlesung” (Hilder 1899) afier having been appointed professor of mathematics in Leipzig, he even criticized Hilbert’s axiomatization of geometry on such grounds. Numbers, Dedekind said, are created to better distinguish things. This kind of conceptualism has recently been described by Tharp. Tharp puts forward the idea that mathematical assertions should be regarded as expressing relations among concepts. Mathematics, from such a perspective, is to be based on largely arbitrary and very narrowly specified definitions. Tharp explains the claims of his conceptualist position with a comparison to fiction. As an illustration of the role of fiction in his argument Tharp presents the following very short story. “The only people in our story are Gertrude and Hamlet. Gertrude is a queen. Hamilet is a prince, and Gertrude is Hamlet's mother.” Tharp continues: “Given these two stipulations which constitute our story, various consequences follow from the meanings of the concepts ‘prince,’ ‘queen’ and ‘mother,’ and are evidently true-in-the-story: for example no princes are queens; Gertrude and Hamlet are distinct; Hamlet is not Gertrude’s mother. None of these conclusions follow logically from the given story, however” (Tharp 1989, pp. 168-169). All the truths gained from this story are analytical and one cannot inquire into questions or facts that were not presented. One cannot, for example ask, what color Hamlet’s eyes have. Mathematical objects such conceived are purely semantical entities, as analytical philosophy since Bolzano believed they are in fact. Thus the question of the nature of number, or of mathematical objects in general, is a question about the nature and ontological status of mathematical theories. It is beyond the scope of this paper to further pursue this latter question, but one should remember that it is commonly accepted that the axiomatic view might be incomplete in various respects 4 The theory of integration from Cauchy to Lebesgue The transformation of the theory of integration from Cauchy to Lebesgue at the beginning of the nineteenth century occurred in a positivistic spirit more concemed with the idea of arithmetical approximation and the language of inequalities (Grabiner 1981). Central to the new Analysis was the concept of continuous function. No matter how much Lagrange may assert and insist that a function is for him an abstract object, in his thought patterns it somehow is residually a mechanical orbit or perhaps a physical function of state, whereas in Cauchy orbits and forces and pressures are always functions, as they are for us today (Bochner 1974, p. 837). Cauchy, by exhibiting the importance of the notion of numerical function, in a sense completed Descartes reduction of geometry to arithmetic. He reduced all geometry to that of the straight line (Lebesgue 1966, p. 179). This kind of reductionism was very much criticized by Grassmann and others, who believed that mathematical theory should model the relations essential to a certain subject matter area (Otte 1989, pp. 5-13). Contemporaries believed that by this introduction of the notion of (continuous) mathematical function, mathematics should really become applicable and should be enabled to produce direct or literal truths of all kinds (Schellbach 1883, 16f). Mathematical knowledge was conceived to have a direct bearing on empirical reality. This positivistic attitude differed sharply from that of the axiomatic movement (Otte 1989, pp. 5~7) The basic property of an abstract mathematical function is its continuity. As Bochner observes, “the conceptions of function and of continuity have evolved simultaneously” D Springer250 M, Otte (Bochner 1974, p. 845). In his Introductio in analysin infinitorum of 1748 Euler defined a continuous function of a variable quantity as an analytical expression composed in any way of that variable and constants (Euler 1748, I, p. 4), such that “un simple changement de notation suffira souvent pour transformer une fonction continue en fonction discontinue, et reciproquement,” as Cauchy (1889, Oeuvres, series 1, 8, p. 145) observed. Cauchy, after having demonstrated the inconsistency of Euler’s efforts, revised the whole approach, transforming mathematics into extensional theory. It became clear that a continuous mathematical function had to be conceived of as an equivalence class of concrete representations of it, rather than to be identified with some of its possible representations — the axiom of extensionality furnishing the constitutive equivalence relation. The property of being continuous can be attributed to such a class only, rather than being a property of some representation of a function. A function in the sense of Cauchy or Dirichlet must thus be seen as an equivalence class of analytic expressions or formulae, where the equivalence relation is based on the axiom of extensionality. And Lebesgue generalized the function concept still farther by assuming that the equivalence relation should hold except for sets of measure zero. The development thus goes from continuous to measurable functions, which upon pragmatic inspection may be said to cover all functions considered in Analysis. Thus the development of the function concept reflects the development of Analysis during the nineteenth and carly twentieth centuries. In this manner, it is possible, since Cauchy, to single out sets of functions by certain of their properties and reason about them in general without representing them explicitly. For instance, instead of giving a linear function directly by f(x)=ax, Cauchy proves that a continuous function having the property f(x +») = f(x) +f(y) can be represented that way (Cauchy 1821, pp. 99-100). What Cauchy achieved really was a new understanding of mathematical concepts, according to which a concept is to be defined, as Schlick later said with respect to Hilbert’s axiomatic definitions, by the fact that certain conclusions can be drawn about it (Schlick 1925, p. 45). This operative understanding, making a concept essentially a scheme of action, is totally alien to traditional substantial thinking. “Indeed Cauchy (1789-1857) and Gauss (1777-1855) may rightly be called the first truly modern mathematicians” (Belhoste 1991, p. vii). And Klein (1896) says in his much-quoted and much-translated lecture on the “Arithmetization of Mathematics” that the early nineteenth century was “the time of Gauss and Abel, of Cauchy and Dirichlet” (p. 241). Cauchy was an incredible calculator, and many of his papers are full of ingenious calculations, such that he himself sometimes complained about the “innumerable calculations. ... Signs. Formulas...” (quoted from Belhoste 1991, p. 214). Cauchy, however, took these formulas not just in algebraic or formal terms but tried to attribute to them “a definite extent” (Cauchy 1821, Introduction). All relationships between quantities have always to be verified “by substituting specific numbers for the quantities themselves” (Cauchy 1821, Introduction). Starting in 1814 Cauchy also approached the theory of integration directly, that is, avoiding relationships with differential calculus, but assuming continuity (Cauchy 1823, p. 122), Cauchy thus dethroned the so-called Fundamental Theorem of the Calculus. After Cauchy there followed Riemann. With respect to the integral, Riemann started from necessary conditions, rather than from some sufficient properties of the functions to be integrated, like continuity, as Cauchy did. And in this way he introduced relational reasoning, based on proof analysis. Riemann defined a bounded function to be integrable on a closed interval, if and only if the partial © SpringerMathematical history, philosophy and education 281 sums or Riemann-Cauchy sums § = )°/(ti) (x; — xu-1)) approach a unique limiting value as the norm of the partition of the interval approaches zero. Riemann writes: “For S to converge it is necessary, besides that f(x) be bounded, that the sum of all of the intervals where the oscillation of f(x) is greater than s, for whatever value of s, could be made arbitrarily small, by means of an appropriate choice of the maximal length of the intervals (x;—xy-))” (Riemann 1892, p. 241; my translation). Riemann’s integrability condition was obviously the weakest under which the traditional procedure, based on Cauchy sums, retains a meaning. It is also important that Riemann thus distinguished between continuity and integrability and created the concept of an integrable function. Integrable functions in the sense of Riemann were regarded as the largest subclass of the class of ‘completely arbitrary’ functions that has played a role in the definition of a continuous function since the first half of the nineteenth century. It seemed thus “that Riemann had extended the concept of an integrable function to its outermost limits” (Hawkins 1970, p. 34). A further generalization therefore seemed unthinkable as long as Cauchy-Riemann sums were regarded as the only possible approach to the definition of the integral. Riemann’s vision remained, however, fixed to the traditional procedure of determining or calculating the value of the integral. In the case of the continuous functions the variation of the argument x controls the oscillation of the function and this led to evaluating the integral on the basis of the customary Cauchy~ Riemann sums, Lebesgue questioned precisely these traditional procedures. Lebesgue writes: “One could say that, according to Riemann’s procedure, one tried to add indivisibles by taking them in the order in which they were furnished by the variation of x, like an unsystematic merchant who counts coins and bills at random in the order in which they ‘came to his hand, while we operate like a methodological merchant who says: T have m(E;) pennies which are worth I.m(Ey) [have m(E) nickels worth 5.m(E>) I have m(E3) dimes worth 10.m(E3) etc. Altogether then I have S = 1.m(E;) +$.m(Eq) + 10.m(Es) +... (Lebesgue 1966, p. 181) Thus Lebesgue’s idea was to partition the range of the function instead of its domain. Any generalization of the concept of measure and of measurability would now obviously afford a generalization of the concept of the integral. Lebesgue’s new approach, following Borel, not only put the notions of measure and measurable set in a central place, but also further strengthened the methodology of axiomatic definition and proof analysis. We see this in the following statement of Borel, whose axiomatic approach to the concept of measurability and the results achieved in this manner were very important to Lebesgue: The procedure that we have employed actually amounts to this: we have recognized that 2 definition of measure could only be usefal if it had certain fundamental properties; we have posited these properties a priori and we have used them to define the class of sets which we regard as measurable. ...Define the new elements which are introduced with the aid of their essential properties, that is to say those which are strictly indispensable for the reasoning that is to follow (Borel, quoted by Hawkins 1970, p. 104) But it was Lebesgue, as Thomas Hawkins has observed, who “may be said to have created the first general theory of integration. Various definitions, theorems and examples existed prior to his work, but they lacked the coherence and completeness of a true theory” (Hawkins 1970, @ Springer252 M, Otte p-IX). The set of intergrable functions should form a reasonable space. The integral operator should, for instance, be subject to a continuity principle. That is, the limit of a series of integrable functions fir should be an integrable function F itself, for which infinite additivity ‘of the measure is obviously required, because F may have infinitely many points of discontinuity even though each of the fa has only finitely many such points. Now, as was said already, the “modem” theory of the Calculus since Cauchy began with a study of continuity. One of the first observations in such an investigation is that the only requisite for the definitions of and theorems on continuity is the availability of a notion of distance as a measure of proximity. This leads to the abstract notion of a metric space. A slightly deeper analysis of the relationship between a given metric on a set and the collection of functions continuous with respect to that metric shows that it is not the metric which is significant, but only those subsets which are open, where this notion of openness is defined as a generalization of the notion of an open interval of the real line. Two metrics ‘on the same set X determine the same classes of continuous functions if and only if'a subset of X is open with respect to one of the metrics if and only if it is open with respect to the other, Thus the task of characterizing continuous functions is equivalent to choosing a topology, that is, choosing a class of open sets. The topology of a space is a property of its global structure and is therefore appropriately to be characterized in set theoretic and axiomatic terms. Without Cantor’s set theory general topology and abstract measure theory would not have been possible. And the theories of measure and integration formed in tum for some time the most substantial single application of Cantorian set theory. Characterizing topology in axiomatic terms may be helpful in understanding Lebesgue’s step of choosing a wider class of sets, measurable sets and measurable functions. It also helps to see what happened in proof-based pure mathematics. The original experiences of force, motion and acceleration gave rise to the mathematical concept of a differentiable function, Now differentiability is defined in terms of continuity and continuity in tum is conceived of in terms of a topology. By means of an axiomatic definition of the notion of topology, one liberates that dialectic of generalization and formal reasoning which charac- terizes the mathematical exploration and reconstruction of reality in mathematical terms. What counts are, on the one side, the exigencies of exact mathematical reasoning, rather then experienced reality as such. An object is replaced by a concept. On the other hand, conceptualization introduces an idea of indeterminacy and freedom into the reasoning process, because concepts are general meanings that do not determine their reference to the objective world. That is the perspective of proof analysis and generalization, Hence we see the emphasis given to the moder axiomatic approach by Borel and Lebesgue. With respect to his elementary construction of the notion of “area” Lebesgue observes, for example: “Even when we consider mathematics as an experimental science, it is important to show that the areas that we have just considered are completely determined by the usual axioms” (Lebesgue 1966, p. 50). And in the same in- service courses for teachers he stated that we do not need to know what things are (numbers, for example) as long as we know what statements about them one is allowed to make (Lebesgue 1966, 16). This advances the view that mathematics is mainly concerned with conditional if-then statements, as in modem axiomatics, rather than with categorical affirmations about intui- tively given objects, as in pre-modern axiomatics. This revolution of mathematical abstraction caused some uneasiness and sometimes “anxiety” in its time. Even Lebesgue’s attitude was full of contradictions. His qualms with the axiomatic method as well as his objections showed up very clearly after Zermelo had published his proof that every set can be well ordered in Mathematische Annalen in 1905. When Zermelo in the course of his argument explicitly formulated the Axiom of Choice, ®@ SpringerMathematical history, philosophy and education 283 legitimizing the use of infinitely many arbitrary choices, most mathematicians reacted critically to such a procedure, even though many of them had used such choices before, with a greater or lesser degree of awareness, in their own research in set theory, analysis and algebraic number theory. ‘Among the debaters was a group of French construetivists and quasi-empiricists of whom Borel and Lebesgue seemed the most critical, In December 1904 Borel had “finished a brief article, requested by David Hilbert as an editor of Mathematische Annalen, on the question of Zermelo’s proof” (Moore 1982, p. 93). Borel’s article stimulated an exchange of letters between himself, Hadamard, Baire and Lebesgue, which was finally published in the Bulletin de la Soc. Math. de France (see Hadamard 1905). This sequence of letters “remains a classic statement on the grounds for accepting or rejecting the Axiom” (of Choice) (Moore 1982, p. 98). Let us in particular look more closely at Lebesgue’s contribution. Lebesgue, considering applied mathematics the basis of pure mathematics, was in general critical of the axiomatic method, because it lacks constructive existence proofs and is thus incomplete. Lebesgue’s work was directly affected by the matter, since a weak, that is, denumerable, version of the Axiom is sufficient to construct Lebesgue non-measurable sets or functions (something Lebesgue believed does not really exist). On the other hand, the Axiom appeared indirectly in various places of Lebesgue’s own work. Hadamard had in his contribution to the exchange of letters, reminded his colleagues of a distinction, made by Tannery (1897), between functions that are “defined” and others that are “described.” A definition, according to this distinction, amounts to a mere existence claim, without providing the properties of the object defined, whereas a description presents the properties of some entity. The issue figured prominently in Cauchy's dealing with the problem of continuous functions, as we have seen, and it reappears at the tum of the twentieth century again in the discussion about Zermelo’s choice functions. Now, Lebesgue makes explicit description or definability the touchstone of his philosophy of pure mathematics, negating that one can prove the existence of a mathematical object, like a function, without describing it (Lebesgue in: Hadamard 1905, 2661). Existence as such is not relevant to mathematics. Thinking about an entity thus means thinking about some property or concept. By the time of writing his letter, Lebesgue would accept an infinity of choices only if an explicit rule could be provided for making them and this means that neither non-denumerable sets nor non-computable real numbers are admitted. Denumerable sets have, however, measure zero, such that any two measurable functions are equivalent when restricted to such a set, From the point of view of measure theory, the concentration on the particular and individual makes no sense at all, In fact, the whole idea of axiomatic presentation would lose its purpose. 5 Conclusion Mathematical ideas that appear extremely abstract and difficult at first sight become understandable from a historical perspective only. The transformation of processes into structures with which we have dealt here is quite instructive in this respect. History of mathematics occupies itself describing processes of growth and development, whereas philosophy of mathematics is concemed with questions of justification. Both play an essential role within the educational context. The historical perspective has provided insight into an analogous opposition inside mathematics itself. On the one side, mathematics played an important role as a force of D Springer28 M. Otte liberating the human mind from its traditional ontological bounds, It has been affirmed ihen ag ins that mathematical practices asserted since the early nineteenth century an increasing degree of autonomy vis-a-vis the reductionist search for meaning; not the least because of its uncompromising zeal for generalization, Cantor's fame affirmation, that complete freedom is the essence of pure mathematics, is but an expression of these facts. although a biased one. Stated differently, the development of arithmetic and also that of integration theory since Cauchy shows quite clearly that axiomatic set theory appears to be the best general Framework available for mathematical theorizing as well as for theorizing about about things that cannot be exhibited In concluding, one might say, as strange as it may at frst appear, “that whereas the new axiomatics, the axiomatics that arose fiom nineteenth century geometry and algebra, can really be seen only as a tool, although a very powerful one, of set-theoretical thinking, hich grew out of nineteenth century analysis, axiomatis in the sense of Euclid, i.e, logical reduetion, analysis and organization of intelligible concepts and meaningful sentences, aera tin an inreductible, fundamental tool of our thinking” (Casati 1974, p. 61). Perhaps it is exactly such a contradiction which makes the study of the history of mathematics worthwhile and revealing, References Belheste, B. (1991). Augustin-Louis Cauchy, Bern Heidelberg Now York Springer. Gochner, S. (1974), Mathematical reflections. American Mathematical Monthly, 81, 827-852. Gixati E1974), Axiomatical and set-theoretical thinking. Synthese 27, 49-62, Cauchy, A. L, (1821).Cowrs d‘anabyse algébrigue. Paris: Impr. Royale. Cauchy, A. L. (1823). Résumé des lecons données éI'Ecole Royale, Polytechnique sur le cateul infinitesimal, Paris: DeBure. Cauchy, A. L. (1889), Oewres completes IV, 5-261. Paris, France Gauthier-Villars. Chaitin, G. J, (1998). The limits of mathematics. Berlin Heidelberg New York: Springer. Euler, L. (1748). ntroductio in analvsin infinitorum. Berlin Heidelberg New York: Springer (Reprint, 1980), Grabiner, J. (1981). The origins of Cauchy's rigorous calculus, Cambridge, MA: MIT. Gracub, W. (1958). Lineare algebra, Berlin Heidelberg New Yorke Springer Hadamard, J. (1905). Cing lentes sur la théorie des ensembles, Paris: Rutan de la Société Mathématique de France, 33, 261-273. Hawkins, Th. (1970). Lebesgue’ theory of integration: Its origins and development, Madison, Wl: The University of Wisconsin Press, Holder, O. (1892). R. Grassmann, Die Zahlenlehre oder Arithmeti Grtingische gelehrte Anzeigen, 15, 585-595. Holder, O. (1899). Anschauung und Denten in der Geometri, Antrittsvorlesung Leipzig 1898 (Reprint, Darmstadt, 1963), Klein, F (1896). The arthmetizing of mathematics. An address delivered atthe public meeting of the Roy: leaclemy of Sciences of Gottingen, November 2, 1895 (I. Maddison, Trews ). Bulletin of the American Mathematical Society, Series II(2), 241-249, “Lakatos, 1. (1970). Falsifcation and the methodology of scien, research Programmes, In Lakatos/Musgrave (Cds), Criticism and the growth of knowledge (pp. 91-196), Cambrilge UP O springerMathematical history, philosophy and education 255 Lebesgue, H. (1966). Measure and integral. San Francisco: Holden Day. Mannheim, K. (1929). Jdeologie und Uopie. Frankfurt/Main: Schulte-Bulmkez (6th printing, 1978; English translation: 1936, ideology and Utopia. London: Routledge), Moore, G. H. (1982). Zermelos axiom of choice. Berlin Heidelberg New York: Springer. Otte, M, (1989). The ideas of H. Grassmann in the context of the mathematical and philosophical tradition since Leibniz. Historia Mathematica, 16, 1-35. ute, M. (2003), Complementarity, sets and numbers, Educational Studies in Mathematics, 53, 203-228. Otte, M., & Jahnke, N. (1981). Origins of the program of arithmetization of mathematics. In H. Bos et al. (Bds.), Social history of nineteenth century mathematics (pp. 21-49). Boston: Birkhéuser. Otte, M., & Panza, M. (Eds,) (1997). Analysis and synthesis in mathematics. Boston Studies in the Philosophy of Sciences, 196. Peirce, Ch, S.: CP = Collected Papers of Charles Sanders Peirce, Volumes I-VI, C. Hartshome and P, Weil (Bas.), Cambridge, MA (Harvard UP) 1931-1935, Volumes VII-VIIl, A. W. Burks (Eds.), Cambridge, MA. (Harvard UP) 1958 (quoted by no. of volume and paragraph). Riemann, B. (1892). Gesammelte mathematische Werke und wissenschafilicher NachlaB New York: Dover (Unabridged repr. of the 2nd ed., 1953). Schellbach, C. H. (1883). Ober den Inhalt des mathem. u phys. Unterrichts auf unseren Gymnasien. Berlin: ‘Teubner. Schlick, M. (1925). Allgemeine Erkenntnislehre. Berlin Heidelberg New York: Springer. Skemp, R. R. (1976), Relational understanding and instrumental understanding. Mathematics Teaching, 77, 20-27. ‘Tannery, J. (1897). De PInfini math, Revue générale des sciences pures et appliquées, 8, 129-140. ‘Tharp, L, (1989). Myth and mathematics. Synthese, 8/, 167-201 Thom, R. (1973). Modern mathematics: Does it exist? In A, G, Howson (Ed.), Developments in ‘mathematical education (pp. 194-212). Cambridge UP. Thurston, W, (1990). Mathematical education. Notices of the American Mathematical Society, 37, 844-850. Valery, P. (1960). Leonardo. Frankfurt: Insel Verlag (first printed in 1895). Wang, H. (1970). Logic computers and sets. Chelsea, NY Wiener, N. (1951), Pure and applied mathematics. In P, Henle a.0. (Eds.), Structure, method and meaning. NY: The Liberal Arts Press © Springer