Você está na página 1de 15

On Reconstructing Tiwanaku Architecture Author(s): Jean-Pierre Protzen and Stella E.

Nair Reviewed work(s): Source: Journal of the Society of Architectural Historians, Vol. 59, No. 3 (Sep., 2000), pp. 358371 Published by: University of California Press on behalf of the Society of Architectural Historians Stable URL: http://www.jstor.org/stable/991648 . Accessed: 17/04/2012 15:43
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

University of California Press and Society of Architectural Historians are collaborating with JSTOR to digitize, preserve and extend access to Journal of the Society of Architectural Historians.

http://www.jstor.org

On

Reconstructing
Architecture

Tiwanaku

JEAN-PIERRE
AND STELLA E.

PROTZEN
NAIR

of University California, Berkeley

wanakuis about70 kilometersdue west of La Paz, dardized, but they leave one wondering what were the uses, and functionsof the spacesthis the seat of Bolivia'sgovernment,at an elevationof types, characteristics, T meters above sea level.' The site is thought architecture created, and how the people of Tiwanaku 3,840 of as the center of a civilization of the same name that movedin andout andthroughthese spaces.We suggestthat emergedhere around300 B.c.andlastedto aboutA.D.1100, the fragmentary remains are an insufficient basis from when it collapsed.At its apogee, Tiwanakuhad expanded which to gain an understanding the significance meanof or its sphere of influence far beyond its immediate environ- ing the architecture may have had for the people of ment into modern-daynorthernChile, southernPeru, and Tiwanaku. For that, we must first get a better picture of eastern Bolivia. The archaeologicalsite today consists of what that architecturewas (or might have been); we must two distinct areas.2The larger of the two, just east of the attempt to reconstructit. A reconstructionof Tiwanaku's modern town of Tiwanaku, centers around Akapanaand architectureinvolves uncovering,among other things, the the comprisesfive main structures: Kantatayita, Semi-sub- principlesthat governedthe layoutof its buildings,the relaterraneanTemple, Kalasasaya, tion of buildingsto buildings,the design of its appearance, Putuni, and Kerikala.The smallerareais to the south of the town and consists of the and the detailsof its construction. man-mademound of Pumapunku. Severalauthorshave tried their hand and offeredtheir The conquistador chroniclerPedro Cieza de Le6n vision of Tiwanakuarchitecture. and Reconstructions the site of describedTiwanaku,which he visited in 1549, as a place in an orthogonal grid of wide avenueswith generally depict ruins,ravaged time, with little more to be seen thanartifi- the variousstructures by strungout along them. An inspection cial hillocks,remainsof foundations, of walls,weathered of the actualsite plan does not reveal any obvious avenues bits pillars,a couple of statues, and scatteredbuilding stones.3 and shows that the structuresdo not neatly line up. One visitorsareconfronted with a verysimilarexperience. observes deviationsof as much as 5 to 6 degrees between Today's What they find areseveralerodedmounds,outlinesof court- the variousorientations. weathered the of yardstructures, Nevertheless, with the exception of the Semi-subteruprights, so-calledGateway the Sun(notmentionedby Cieza),somestatues(probably not raneanTemplewith its north-southaxis,most majorstructhose seen by Cieza), foundationstubbles,and jumbles of tures at Tiwanakuseem to be oriented roughly east-west. Alan Kolata sees in this orientationan embodimentof the buildingstones-but not a singlestandingoriginal building. The fragmentary remainssuggest an architecturethat solarpathas one of the organizingprinciplesfor Tiwanaku's was highly structured, rigidly ordered, and largely stan- urbanlayout.4The principle,he suggests,is carriedfurther Tw

?n ;ri

i.

01:
:' '.;.; .

\r of (afterManzanilla) Figure1 Reconstruction Akapana

in that "both Akapanaand the Puma Punku possess axial, constructedcentrallyinto their east andwest twin staircases facades."'Linda Manzanilla,who has excavateda stairway west side, suggested that it indeed may have on Akapana's At been a doublestaircase.6 Pumapunku, however,the westand ern staircasehas been entirelyexcavated, it is not a twin is it knownwhat the configstaircase.On neither structure the urationof the easternaccesswas. On Akapana, east side is still buried under the tailings of an earlier,massivelooting operation. At Pumapunkuthere is at this time no evidence of a stairway,simple or double, on its east side; that access has yet to be excavated. Other, more or less detailed, reconstructionsof individual structureshave also been proposed.Manzanilla,for as example,offersa view of Akapana a seven-tieredplatform mound (Figure 1). In her reconstruction, she shows each tier or terrace wall with a uniform height all around the mound. In doing so, she does not acknowledgethat the terrace walls on the northeast side, excavated by Gregorio Cordero Miranda from 1976 to 1979, and those on the northwestside, excavatedby herself in 1988 and 1989, difin fer significantly their configuration, execution,andheight

okcw

after of Figure3 Reconstructions Pumapunku Grasso,Mesa,and andafterEscalante Gisbert (lower) (upper)

n-

Figure 2 Verticaland horizontalconfigurationsof Akapana'seast, west, and south sides compared

(Figure2). It is by no meansobvioushow these wallswould have met at a corner.7 of are The reconstructions Pumapunku numerous.One Dick Ibarra Grasso, Jos6 Mesa, and Teresa Gisbert, by imposes a mound on Pumapunkuwhere there is a sunken court for all to see (Figure 3). Another,by JavierEscalante Moscoso, altogethermore plausiblethan the former,nevertheless does not acknowledgethat the areaof the platforms on the east side, where Escalanteputs a building,is almost 2 meters lower than the adjacentwings, or that the canals puncturethe terracewalls ratherthan being underground (see Figure 3). Edmund Kiss, a German building official and (Baurat), proposedwhat must be the most extravagant platformarea flamboyantreconstructionof Pumapunku's It is completewith plans,perspectives, sections, (Figure4).8 and cutawayviews. However fancifulKiss'sreconstruction is, at least it has the merit of incorporatingactualbuilding stones, even if, as we shall show, most of these stones are used in the wrong position. All these hypothetical reconstructions,faulty as they may be, are harmlesswhen comparedto the actualreconstruction of the Kalasasayaperpetratedby Carlos Ponce Sangindsin the late 1950s and early 1960s (Figure 5).9To take but one example,to rebuildthe easterngatewayto the all Kalasasaya, the evidencePonce had was a faintoutline (a sort of groundplan)carvedinto the top landingof the stairway, which was uncovered by a French expedition in late 1903 (Figure6). There was no clue to how high the gateway might have been or how the gatewaywas connected to the
ON RECONSTRUCTING TIWANAKU ARCHITECTURE

359

i-i?iiiiiiiiliiili :::i:-::::;

:::l:~?:"~~I~~l~ii'B~S~B~;~:~r18~9~~1"'~ :::
?

~PU"X~8~as~.~?~ll***?r~)?l~?;(i~rl~:?:~ :_.::_ ;
::::::: --:::

iii
rr". :

i- ; .r
- ? ??..:-; ?-~?~~~? ?? ..~;i,.: ii

:-::::?~- ~i;;i :i:s-?-:- I~i;-;--li.l-.?i --::::::?a'il;i~i:::~;-1~;5~-~iii:i':: :-i: ii? i :_I-i-iiii~__ ii:;i-l-i :-:': ::::? -? --'-:--':::::::::- ~ii;i~ei--iii$ni"i : i::::: : ..: :: ?-: ----:?::.::::: -_::t:i---:-:: i

:-:: .

::
;E --:?::? ???-~? [~

;...----???i

???? ?:?
: ?? ?-?:?x? ???

i : -::::::;::-i?'':" ::: .' ::?::-::---?: :.:..:. : :

: i-:-::-:::-.::::1--,:i:i::,~s~:':I`:-j --?:: -.-: :::: :::'. :

of von from 1937) Kiss,Das Sonnentor T77huanaku... Figure4 Reconstruction Pumapunku, Edmund (Leipzig,

li?: :.:~:?:

:-: ::'::
.:: ?::::

: :.. : . ..:. :::::::: .-::: : ii:: ::::::::w ::.:::::::--..i

as Figure5 Eastern gatewayto the Kalasasaya reconstructed by PonceSangines

JI.

j"

20

i,rtt

29 21 ALS ur

side walls (if there even were side walls) and to the two upright stones flanking the base of the stairway.What is in of perhapsmost objectionable Ponce'sreconstruction the is that it gives a totally distortedview of the tecgateway tonics of Tiwanaku architecture. Eduard Sekler distinguishes between the structure,the construction, and the tectonics of a building.10 structurehe means the strucBy tural principleon which the building is conceived, meaning, in the case of Ponce'sgateway,gravitywall and simple beam-the lintel. Construction refersto the materialization of the structural in this case,mortared stones.And principle: tectonics refers to how structureand construction finally, are perceivedor read:in this case, a massivecore clad in a veneer of thin stones. To our knowledge,the typical constructionat Tiwanakuis not one of cores and veneers, but of massivewalls laid up in stones without mortar." and The above-mentioned reconstructions attemptsof for example,by PaulGoldstein-to likenthe architecture Omo in the Moquegua Valley in southern Peru to raise the question of what actually Tiwanakuarchitecture, constitutesTiwanakuarchitecture. Goldstein argues:
For Tiwanaku,the monumental representation of state power crystallized in a characteristicceremonial architecture whose and its altiplanosatellites. In archetypes are found at Tiwanaku the broadest sense, the key elements can be categorizedas (a) artificial terraced mounds, (b) rectangularenclosures, including walled precincts and sunken courts, and (c) a complex of doorways and staircasesthatchanneledaccess to a ceremonialcore.12

22
6 4

&DIS ;..6057.....-..---...S. -

outline carved the top landing the stairway into of as Figure6 Faint recorded Posnansky by
360
JSAH / 59:3, SEPTEMBER 2000

As Goldstein says, these elements are archetypaland thus may not uniquely describe Tiwanaku architecture;many

archaeological sites from the Peruvian coast to the high Andes featuresome or all of these elements.13 What, then, are the design elements that give Tiwanakuarchitecture its culturalidentity and distinguishit from other architecture? What is the underlying order that governs the layout of buildingsand the relationshipsamong buildings?What are the buildings'spatialcharacteristics? What about the proand dimensions of the buildings and their eleportions And last but not ments, the composition of their faqades? constructionmaterialsand least, what are the characteristic techniques? These may seem dauntingquestions, given how little and thereis atTiwanaku, one authenticstandingarchitecture if can wonderswhereone shouldturnfor answers, answers be stufound. We took our cues from two nineteenth-century dents of Tiwanaku architecture,the pioneering Leonce Angrandand Alphons Stiibel.Angrand,a Frenchdiplomat, producedin 1844 the firstknownsite plan of Tiwanakuand measuredand recordednumerousbuildingblocks in great In detail and with great accuracy.14 1877 Stiibel,a German geologist, also measured scores of building stones with utmost precision and documentednumerousconstruction in of details.His workwaspublished 1892with the assistance Max Uhle.15Uhle added an historic dimension to Stiibel's work,and,aswe shallsee, the two of them cameup with severalvery interestingobservations aboutTiwanakuarchitechavemostly been forgottenor ignored. ture that Following in Angrand'sand Stiibel's footprints, we turned our attention to the many building blocks littering the site. If Tiwanaku architecturecan ever be visualized, then it has to be pieced together from a chaos like the one found at Pumapunku,a jumble of enormous stone slabs, fragments of elaborate gateways and windows, and hundreds of beautifully carvedyet mutilated building blocks strewn all over the landscape.

The niche itself has beveled jamb and head chambranle. revealsflaringopen towardthe back.A Type 2 niche is set within a stepped chambranle, it too has beveled reveals and out towardthe back. Icons of both types of niches flaring are also found. The niches of Type 2 further assume two distinct shapes:one, Type 2a, is nearly square;the other, Motifs Type 2b, is elongated and gate-like in appearance. and constructionfeaturesincludeconcentriccircles,crosses There arestep moldings,rabwithin crosses,and "arrows." bets, and steppedrabbets.Finally,there aresocketsfor construction clamps of a variety of shapes and sizes, and complicatedcutouts that we came to call "hoisting grips" for lack of a better term. These details assist in orienting the stones and in establishingrelationshipsamong stones. Clamp sockets, for example,are at the top of a stone in a horizontal plane; so are hoisting grips. Some motifs are incomplete on one stone, suggesting that there might be another stone with the missing part. And, as will become clearlater,steppedrabbetsarealongverticaledges with the step at the top, and step moldings generallyrun horizontally with the broaderstep above. Stone Types and Many of the stones scatteredaroundPumapunku other of the site showa striking to one another, both parts similarity in design and dimensions.Stiibel and Uhle speculatedthat the Tiwanakuhad a kit of standardbuilding blocks from In whichthey assembled structures Pumapunku.18 our the at own work,we remeasured of the samebuildingblocks, many not in an attemptto duplicate or Angrand's StiibelandUhle's details. work,butwith an eye on tool marksandconstruction In the processwe couldindeedconfirmthatmanystonesare perfectreplicasof one another.But there are also building identicalin design, that show significant blocks, apparently dimensionalvariations.Such blockscould obviouslynot be one as exchanged for another, is the casein a truekit of builddo ing blocks.Yetthe strong similarities indicatethat there were repetitivefeaturesin the architecture, that certain and designprinciplesprevailed. To date, we have identified some fifteen stone types, with abouttwentyvariations.The following exampleshave providedus with insights into Tiwanakudesign and composition. The so-calledEscritoriodel Incais a well-knownstone; it was drawn and recorded by several nineteenth-century explorersand travelers(Figure 7). It is called the "Desk of the Inca"perhapsbecauseit looks like miniaturizedarchitecture. This stone, which is only about 1.5 meters high, gives us some ideas of how design elementswere arranged:
ON RECONSTRUCTING TIWANAKU ARCHITECTURE 361

The Stones
Design Elements and Construction Details On inspection, the stones and stone fragments reveal a repertoire of motifs and construction features that are appliedin variouscombinations.Openingsare generallyset within a recessed frame, or chambranle: structuralfea"A ture, often ornamental, enclosing the sides and top of a doorway,window, fireplace, or similar opening. The top and piece, or lintel, is calledthe transverse the side pieces or the ascendants."16 jambs There are niches of two types:Type 1 and Type 2.17A Type 1 niche is set within an inverse trapezoidal,recessed

I-

": j--

.~
.

i~--ii~ii~i?i---:i-;:L

.....

Figure 7 Escritoriodel Inca

Figure 9 Two variationsof H-stones, with "crossed" baron back (left and right)and plainback (center)

step moldings were horizontalbands serving as separators between differentarrangements. Type 2a nicheswhen used alone form a horizontalrow.When combinedwith Type 1, Type 2 is set above Type 1, and the cross motif appearsin verticalrows.It was alwaysassumedthat the Escritoriowas an isolatedstone. Yetthere existtwo fragmentsthat suggest that there were more such stones, if not exactlythe same, then very similar:FragmentA and FragmentB. Fragment who workedat the site B was believedby ArthurPosnansky, But from about 1904 to 1945, to be an architrave. on close one findsthe beginningsof nichesbelow the step inspection molding that are very similar to those of the Escritorio. Similarly,there are the bases of niches on Fragment A. Dimensionally,the two fragmentsand the Escritorioall fit into the same compositionscheme (Figure8). we From the stones with an "arrow," learn that some stones were executedin left-handedand right-handedversions. This stronglysuggeststhatsymmetryin architectural compositions was important to the Tiwanaku builders.

shouldbe standingwith the half cross Stoneswith "crosses" and the stepped rabbetat the top. These stones, too, were executed in left- and right-handedversions. There are a great many niche stones around the site as well as in the town and in people'syards.Some have just one niche, others have two, and some have dentils set above and between the niches. The stones we have come to call H-stones were executed in two variations.In one variation,the reverseside of the stone is a flatsurfaceflankedby straightrabbetsandstep moldings on the sides; in the other, one finds a "crossed" barwith the horizontalbarendingin step moldings(Figure 9). Stones with a trapezoidaland beveled recesswe identified as a sort of capstone that neatly completes the transand verse of the chambranle the head of Type 2a niches on, Anothertype of stone, similarto the prean H-stone.19 say, vious one, but with a recessed panel, we did not immediately know how to interpret.

del Figure 8 Escritorio Inca B (middle) (left). Fragment and FragmentA (right)stones all fit into the same vertical composition.

362

JSAH

/ 59:3,

SEPTEMBER

2000

i..

I
r.

Figure 10 Monolithicblind diminutivegateway Figure 11 Composite blind diminutivegateway

?OM

:
,

:.'

??:./

,~A.

Diminutive Gateways and Other Assemblages Blind Diminutive Gateways


Crucial to our understanding and interpretation of this and other stones at Pumapunku was our "discovery"of two fragments of what was once a monolithic blind diminutive gateway,48 centimeters wide and 95 centimeters high (Figure 10). One might argue that blind gateways are not reallygates but more closely representniches. Niches, however, are unlike gatewaysin their basic form. Cross sections in plan and elevation reveal the difference: in the case of niches, there is a smooth, beveled transition from the opening to the back of the niche, whereas in the blind gate there is a deliberate recess, a distinct hint of a shallow space into which the opening leads, which is absent in the niche. This particularblind diminutive gateway made clear the use of the lintel stone with the recessed panel:it is the ceiling of the small space. It also gave us clues to the positioning of the step molding found on so many stones and led us to conclude that among the possiblestone assemblies are blind diminutive gatewaysthat are not monolithic but are assembledfrom manystones (composite),andthat these composite gatewaysare either plain or assembledfrom Hstones. In each case, there are the appropriately placedrabbets and T-shaped clamp sockets to receive and attach a backing(Figure 11). Such lintel stones could also be combinedwith H-stones to form a row of blind diminutivegatecombinationwas anticipated Stiibel ways.This particular by and Uhle, although they did not record any actual lintel stones.20 A lintel stone with only half a "cross"could be

S0,
a.0Moa
0.50M 0.50M

ON RECONSTRUCTING TIWANAKU ARCHITECTURE

363

OMM

A and (middle, Pumapunku right) Fragment (left) Figure14 Little compared

olo
o5Mu 0.eM0.50m o.YMu

Figure 12 This combination of H-stones was anticipated by Stibel and Uhle

The Gateways
The famous Gateway of the Sun-Tiwanaku's emblemAt certainlywould suggest that buildings existed.21 Pumafinds-besides the standardizedstones-the punku one fragmentsof gatewaysthat are reminiscentof the Gateway of the Sun. Stiibel and Uhle identified fragmentsbelonging to at least three, possibly four, gateways.22 The gatewaysat Pumapunkushare a number of traits with the Gateway of the Sun. They are (or were) monolithic, that is, cut from a single slab of stone, includingthe threshold.The gatewaysare plain on one side and divided into several fields by a step molding on the other. The revealsof the jambsandthe doorheadarebeveled;they flare open to one side, such that the opening of the actualdoorway is larger on one side than on the other. By analogyto the niches,we inferthat the plainside with its smalleropening is indeed the front or outside of the gateways,whereas the modulatedside with the larger opening is the back or inside. On the plain,or front,side the doorwaysare flankedby recessesor pockets,one on eitherside of the two rectangular doorway:the functionof these pocketswill become evident later.The greatfrieze on the Gatewayof the Sun is specific to this gateway,althoughthe other gatewaysmay have provisions for an equivalentdecoration.The step molding that dividesthe backside at abouttwo-thirdsof the height of the gatewayswraps around, or roofs, the doorway head, thus definingdistinctfields:one eachon eitherside of the gateway below the step molding,one each aboveit, and a field in the area.The fields above the step middle, the "wraparound" each containone or two niches of Type 2a, andthe molding fieldsbelow containone niche each of Type 2b. It shouldbe noted that the Gatewayof the Sun, althoughmonolithic,is

' t ., ......... .. i;.',-


F.....

......

fits composition (right) inwith Figure13 Gateway vertical A's A Fragment

combined to form yet another kind of blind diminutive gateway(Figure 12). There exist two fragments of what was a monolithic open diminutive gate that we call Gateway A. What is remarkable about GatewayA is that it fits in neatlywith the Escritorio-typestones, for example,with FragmentA (Figcould ure 13). The gatewayknown as "LittlePumapunku" A. The outlines of step be attachedto the backof Fragment moldings on Fragment B match the spacing between the step moldings on the Little Pumapunku.The two stones matchin height, and the U-shaped crampsocketsat the top of each stone line up (Figure 14). The architecturewe have reviewedthus far is either at the monumentalscale of artificialmounds, platforms,and sunkencourts,or it is diminutive,at roughlyhalf scale.Was there any building-sizedarchitectureat Tiwanaku?
364 JSAH / 59:3, SEPTEMBER 2000

not a completecomposition;the Type 2a niche on the upper left haspartof its chambranle missing,whichwouldhaveto be completedon anotherstone. The three gatewaysof Pumapunku, althoughshattered into pieces, can be reassembledinto recognizableform as follows. -~~ Gateway III It is the broken gateway that can be reconstructed most accurately;all its pieces with the exception of the threshold are still there. The width of the doorway can be calculated precisely to 77.4 centimeters thanksto the incised meanderfrieze above the doorway's front. The constituent parts of the frieze are repeated with such regularitythat it is easy to reassemble it with precision in spite of obliterated or shattered sections. The height of the doorway opening can be measured directly on the remaining jamb pieces, from the doorhead to the traces of the broken-off threshold (Figure 15). The gateway'sback side shows that the opening was roofed by a triple-steppedmolding (Figure 16). This gateway,like the Gatewayof the Sun, was not a whole composition:the left ascendant the chambranle the Type2b niche of of on the rightjambwouldhavehadto be continuedon another to buildingstone attached the gate.This samejambfragment revealsthe purposeof the pocketswe haveseen on the Gateway of the Sun:here the pocket houses clamp sockets.The two T-shaped sockets-one parallel,the other perpendicular to the main plane of the gateway--suggest that more building blocks were to be added to the gatewayin either plane.AdditionalrecessedandplainT-shapedclampsockets on the narrow sides of the stone slab further support the notion that the gatewaywas partof a more extendedwall.23
__

FMU

Xi

E]

_ /

_"

front Figure 15 Gateway III,

LL[

f
m 0.50 0-

'

Figure 16 Gateway III,back

?..
?s .. ~. :".

r
"i ?:~.

':7
O.IM

Figure 17 Gateway II,back (left) and front (right)


ON RECONSTRUCTING TIWANAKU ARCHITECTURE 365

i-iii:iiirL-

:??:?: ::?

Gateway I Of this gateway,only the right jamb and a fragmentof the left jamb remain (Figure 19). The step molding roofs the doorwayin two steps,not three as on the previoustwo gateways. We also noticed a slight variationin the dimensions of its Type 2b niches as comparedwith the previous two gateways,a detailto which we will returnlater.Little of the gateway lintel has survived,but what is left is enough to affirmthat, like Gate II, it had no incised frieze. Instead, where the friezewould be, thereis a largerecessedareainto which, again, some equivalentornamentationcould have been set.

-i-li--

of II Figure18 Nicheiconson the jambreveal Gateway

Gateway II GatewayII is very similar Although in general appearance to GatewayIII, it differsfrom the latterin two aspects(Figure 17). First, it has niche icons on the jambreveals,a feature that is unique to this gateway(Figure 18). Second, it does not have a meanderfrieze. Instead,where the frieze is incisedon GatewayIII, GatewayII has three narrowledges stepping back. On these ledges one finds tiny T-shaped crampsocketsregularlyspaced,suggestingthat some ornamentation was attached there in lieu of an incised frieze. GatewayII has T-shaped crampsocketsof all kindsin similarplacesas GatewayIII, andthus it mayhavebeen set into a similarcontext.

Akapana Gate In the summer of 1995 we recorded the remains of yet another gateway on the summit of Akapana(Figure 20). The pieces of this gateway have to our knowledge never been documented. Stiibel and Uhle did mention the remains of a gateway on the plateau of the mound.24 Whether these remains are identical with the pieces we recordedwill probablynever be known.This gatewaydiffers from the aforementionedgatewaysin that it is not a monolith but a trilithon,that is, a gatewayassembledfrom an enormouslintel andtwo jambstones. Only fragmentsof the lintel and a candidatefor the left jambstoneremain.A of cut stone depictedin Squierhas all the characteristics our jambstone,and could be identicalwith it.25The rightjambstone, no longer existing,mayhavebeen the one illustrated by Stiibel and Uhle.26It is of interest that Bernab6Cobo, who visited Tiwanakuin 1610, and probablyagain around

4'

o,1u

oI
I, Figure19 Gateway back(left)andfront(right)
366 JSAH / 59:3, SEPTEMBER 2000

1620, described just such a gate: ". .. fifty feet to the east of

it (Acapana)remains standing a large gate of only three well-wrought stones, to each side its own, and another on top of both."27 The Akapanagatewayhas triple-steppedroofing and lacksthe niche in the field below the step molding. It shows a very unusual construction technique. The underside of the lintel is cut at an angle, the exact angle of the bevel of the doorhead flaring open to the inside (Figure 21). The top of the jamb is cut at the same angle to receive the lintel. With this configuration, the bottom of the Type 2a niche above the step molding had to be carvedout of the beveled surface of the jamb. These details bespeak a remarkable in sophistication stereotomy,or the artof stonean understanding,if not knowledge,of cutting, presuming descriptivegeometry.

OM:

"

:...

?C :.

'

O.~LI

I.
I
. .

.
F'

0:
0.50 m

0- M'a:A

0.50M

0.60M

m 0.50

Figure 21 Akapanagateway, construction

Figure 20 Akapanagateway, back

Figure 22 Comparisonof Scheme 1 (right)with Scheme 2 (left)


:' A

ON RECONSTRUCTING

TIWANAKU

ARCHITECTURE

367

Two Schemes

which the gatewayshave been carved,he thought that he The full-sized gatewaysjust considered,while very similar could inscribethem into a rectanglewhose diagonal,width, of in appearance,fall into two differentvertical composition andheightwerein the relationship 5:4:3.30Unfortunately, numbersdo not quite work out, but even if they schemes(Figure22). In otherwords,the verticallayeringof Escalante's the variouselements is not the same for all gateways.The did,we challengehis idea.As noted above,neitherthe slabof Gateway of the Sun and Gateway I follow one pattern, the Gatewayof the Sun nor that of GatewayIII accommoScheme 1; GatewaysII and III and the AkapanaGateway dates the full composition;furthermore,stones had to be follow another,Scheme 2. We surmise that the difference addedto completesome of the niches.It seemsutterlyillogstems from the number of steps in the step molding wrap- ical, therefore,to assumethat the slabs-the dimensionsof ping aroundthe doorway:a three-steppedwrappingpushes which are accidental,dictatedby what was availablerather the horizontalstep molding down fartherthan does a two- than a deliberatedesign of the whole composition-should system. stepped wrapping.Consequently,the fields below the step providethe basisfor a proportional Earlierin the projectwe were reasonablysure that we molding are proportioned differently, and so are the 2b niches in them. had evidence for a proportional system governing the dimensionsof niches.3" had lumpedtogether all niches We of a type in one set, obtainingforType 2a an averagewidthDimensions and Proportions to-height ratio of 1:1.0394, with a standarddeviation of The repetitivenessof certain measurements,the apparent 0.013, and for Type 2b a ratio of 1:1.4495,with a standard regularityof like elements,and the similaritiesamongcom- deviation of 0.043. The low standarddeviations gave us positions are intriguing. Like many researchersbefore us confidence that we were on the right track. However, on (whose work will be discussedbelow), we suspect that the finding that the gatewaysfall into two schemes, in which Tiwanakuused a system of preferredmeasurements, possi- the proportions of the Type 2b niches differ from one a system of proportions. scheme to the other,we had to reviseour earlierresults.We bly now divide the Type 2b into two sets, Type 2b, and 2b2, with averageproportionsof 1:1.5474and 1:1.341.Each set Module now shows an incrediblytight fit, the niches in either set There aretwo fundamental waysof buildinga measurement deviating from each other by no more than a millimeter. First,one maychoosea basicmodulefromwhichthe We conclude, therefore, that while the Tiwanakuhad set system. dimensions all architectural of elements,fromthe dimensions specificproportionsfor individualelements or motifs, they of the buildingto its smallestdetail,are expressedin either appliedthese proportions stubbornly, dependenton not but of multiplesor fractions the module.Second,one maychoose an overallcompositionor specific context. that of rulesof proportions relatethe dimensions one element to thoseof its immediate Rulesof proportions need neighbors. not be tied to a particular moduleand thus maybe indepen- Scaling Factor dentof scale.The two systemsarenot exclusive eachother; Gateway A appears to be a perfect miniature replica in of be combinedinto a singlesystem.What kindof sys- almostall its detailsof some of the full-sizedgateways,such theymay tem did the Tiwanaku if any? as Gateway II at Pumapunku.The front of Gateway A use, Posnanskyand Escalantesuggestedthat the Tiwanaku shows the gateway in its double-stepped, recessed chamused a unit of measurementcalled a luk'a,that is, an arm's branle,as well as the two "pockets"with T-shaped cramp length, or roughly 60 centimeters.28Through all our sockets so typical of the largergateways.On its back side, measurements, we could not only not corroborate this the gatewayis roofed with a triple-steppedmolding similar hypothesis, but we were unable-until now-to detect a to that of GatewayII (Figure23); it has the corresponding convincingunit of measurement,or module, fromwhich to Type 2b2niches below the step molding but is lackingthe derive other measurementseither as multiplesor fractions Type 2a nichesaboveit. Instead,thereis a sizablerecessthat of this module. could have receiveda stone with a niche or some other element. The apparentclose affinitybetween the two gateways Proportions had stronglysuggeststhat the Tiwanaku some scalingfactor Escalante used Pythagorean tri- that allowed them to maintainthe relativeproportionsof proposedthatthe Tiwanaku anglesto proportiontheirbuildings.29 Takingthe slabsfrom similar elements independent of scale. Protzen has been
368
JSAH / 59:3, SEPTEMBER 2000

Figure 23 Gateway A and

GatewayIIcompared
t

LL

bW

QC

searchingfor a linearor nonlinearmonotonic functionthat would providethe desiredtransformation. Aftersome fumhe did come up with the function bling,
x' = tg30?x.

also more complex than in the modular system, and ratios between widelyseparated be veryhard calculate. to partsmay Becauseof this structure, such a system gives morescope for and experimentation variation ...34

This function predictscorrespondingvalues within a margin of errorof ? 1% or less. Note that the value of tg300 is 0.5774, makingthe smallerarchitectureslightlylargerthan half scale (Figure 24).32 Through our investigationof dimensionsand proportions we had hoped to find rules that apply generally; instead, we get a picture of infinite sensibilitiesto specific conditions. This situationis not unlike that describedbyJ. J. Coulton of the early Greek architects, who probably "used a system similar to the one set out by Vitruviusfor the Ionic order."33 Coulton explainsthis system:
Inthis the rules do not relate each element to a single common

A Speculative Reconstruction
Our colleague StephenMiller in Classicswas firstto notice the affinitybetween the H-stones and the centralsection of the Escritorio stone. Taking this clue, together with the compositionalfit of the Escritoriostone with the miniature GatewayA seen earlier,and the scalarrelationshipof Gateway A with the full-scale gatewaysas points of departure, Protzen speculated that the H-stones and "cross"stones might fit into a similar composition with some full-sized gateway.The height of the H-stones with "crossed"bars corresponds-give or take a millimeter-to the height of the 2b1 niches on gatewaysof Scheme I. Extrapolating the dimensionsfrom Scheme I, Protzen createdthe configuration shown in Figure 25. Searchingthrough our archiveof stones, we identifieda few more stones that seemed to fit in well with the proposed composition:a fragmentof a lintel (a step-molding stone) and, of course, niche stones. All the actualstones have the appropriate dimensionsand haveTsockets in the right locations, supportingthe shapedclamp proposedreconstruction. This reconstruction is entirely speculative since we have no evidence that such a configurationexisted. Yet it suggests other possibilities that we have just begun to explore.For example,we suggestedthat gatewayswere not freestandingbut were part of a longer wall, and that walls probably flanked the opening on the front side. In the reconstruction'sfront side, too, there are projectingwalls (Figure 26). Can the gatewaysnow be incorporatedwith the proposed configuration? This is a question for further research.
ON RECONSTRUCTING TIWANAKU ARCHITECTURE

module, but form a sort of chain, so that each element is derived the one, successivelyfromthe preceding usually immeone. Theratiosbetween successive partsare diately preceding

IL

itJ

__

lu
b

ol

C DE

the of Figure24 Graph relating dimensions Gateway to those of II A Gateway

369

Conclusion
We think that we have succeededin unlockingsome of the design principlesof at least some of Tiwanaku'sarchitecture and in understandingits structure,construction,and tectonics. The structuralprincipleis simple. It consists of load-bearingwalls and occasionallintels (beams);it is constructedin massivestone blockswithout mortarbut sometimes held togetherwith construction clamps.The tectonics consists of completely erasing the identity of individual stones in favorof a continuous surfacemodulatedby geometricmotifs.This is in sharpcontrastto, for example,Inca and which sharesthe samestructural architecture, principles similarconstructiontechniques,but where the individuality of the stones is often drasticallyaccentuated by deep sunkenjoints and unique shapes. We also know that, in spite of our findingsto date, we are not even close to comprehendingTiwanakuarchitecture. For example, in what relationshipdid the half-scale architecturalpieces of the Escritorio-type stones and the and relatedGatewayA stand dimensionally compositionally with the full-scale architecture,and what was their use or function? Stiibel and Uhle suggested that the half-scale architecturewas ideally suited for altars.We are intrigued by the idea since we find in our own history such examples as Gothic altars-with their arches, spires, rose windows, etc., carvedfrom wood or stone-that are miniaturereflecwithout.But as long aswe do tions of the largerarchitecture not know how the Tiwanaku half-scale architecturewas connected to the full-scale architecture we cannot pass judgment on Stiibel and Uhle's suggestion, nor do we yet have an alternativehypothesis. Stiibel and Uhle also wondered about the abundance of gateways relative to the scarcity of buildings found at What was their significance?The gateways Tiwanaku.3" certainlydo bespeaknotions of passageand movement, as our colleague William Conklin once proposed.36Yet to understandthe nature of this passageand the paths of this movement, and possibly appreciatethe symbolic meaning of the gateways, we first must know where the gateways actuallystood-something that is farfrom established-and was we must knowwhat theirrelationship to each other and to the buildingsof which they may have been a part.Buildings are different from almost every other artifactin the archaeologicalrecord:they are rooted to a site, they cannot be moved; buildings are big, at least big enough for human occupation;and buildings are eminently public in the sense that they impinge on all of us, they cannot be avoided. To apprehendarchitecture,it is not sufficient to contemplate a building as an object from a distance;one must experienceit, walkaroundand throughit, andbe in it.

i~Q\

P~2iiF~ia~ a

~8,

?: I ?.??;-.: ??;?: ? ~:.~ :i'::~

laow

.i.
Figure 25 Hypotheticalconfiguration,back

,?.

< ..?1:.

~P

.:.:??

Figure 26 Hypotheticalconfiguration,front

370

JSAH

/ 59:3,

SEPTEMBER

2000

To quote John Whiteman, ". .. we become unwillingly subject to architecture. It envelops us and controls our experience extensively.... Architecture ... works on us even when we are not looking."37 Obviously, we can no longer apprehend Tiwanaku architecture in this way: we cannot move through it or be in it. Yet an apprehension of this kind is a prerequisite for any plausible interpretation of the possible functioning and symbolic significance of the architecture. Whether we will be able, someday, to experience Tiwanaku architecture vicariously in some kind of virtual reconstruction is contingent on much-and more painstaking-work.

von 18. Stiibeland Uhle, Die Ruinenstiitte Tiahuanaco, II, 38. part 19. See Protzen andNair, "WhoTaughtthe Inca Stonemasons," 29. fig. von 20. Stiibel andUhle, Die Ruinenstiitte Tiahuanaco, II, 38. part 21. See Protzen and Nair, "Who Taught the Inca Stonemasons," figs. 16 and 17. von 22. Stiibel and Uhle, Die Ruinenstatte Tiahuanaco, II, pl. 29, figs. part 2d. 2c, 23. See Protzen and Nair, "Who Taughtthe Inca Stonemasons," 32. fig. von 24. Stiibeland Uhle, Die Ruinenstitte Tiahuanaco, II, 27. part and in 25. George E. Squier,Peru:Incidents Travel Exploration theLandof of theIncas (London, 1877), 280, centerright. It should be noted that the picseveral ture is not of an actualview but a compositedrawingincorporating observations made by Squier.Although Squierwasone of the firstto actuthe ally photograph ancientremainsof Peru and Bolivia,his bookwasillusmade afterthe photographs. tratedwith engravings von 26. StiibelandUhle, Die Ruinenstatte Tiahuanaco, 38, fig. 22. pl. del 27. Bernab6Cobo, Historia NuevoMundo(1653). Preliminary study by Padre FranciscoMateos, 2 vols., Bibliotecade AutoresEspafiolesdesde la Notes Formaci6ndel LenguajehastaNuestros Dias (Madrid,1964), vol. 2: 196: This article is based on a paper read by the authors at the 39th Annual ha "... cinquentapies al orientedel (Acapana) quedadoen pie una portada of the Instituteof AndeanStudiesat Berkeleyin January1999. Meeting de solas trespiedrasbien labradas, cadalado la suya,y otraencima a grande 1. For a map, see Jean-PierreProtzen,with StellaNair, "Who Taughtthe de ambas." A of and Their Skills? Comparison Tiahuanaco IncaCutIncaStonemasons Tihuanacu-The Cradle 28. Arthur Man,2 vols. (New ofAmerican Posnansky, 2 (1997): 146-167, fig. 1. Stone Masonry," JSAH 56, no. York, 1945), vol. 1: 89. Escalante Moscoso, Arquitectura Prehispanica, 2. For site plan, see ibid., fig. 2. 392-397. del 3. Pedro Cieza de Le6n, Crdnica Peru,PrimeraParte(1553). Introduc339-400. 29. EscalanteMoscoso,Arquitectura Prehispanica, ci6n de FranklinPease, Nota de Miguel Maticorena(Lima, 1984), part 1, 30. Ibid., fig. 312. chap. cv, 282-285. 155-156. 31. Protzen and Nair, "Who Taughtthe Inca Stonemasons," 4. Alan Kolata,TheTiwanaku Mass., 1993),96-98. (Cambridge, 32. Anglesof 300 areeasy to construct,even withoutthe help of a compass. 5. Ibid., 97. With three sticksof equallength one can constructan equilateral triangle, Una en del 6. LindaManzanilla, (Mexico Akapana: pirdmide el centro mundo all anglesof which are 600. Markingthe midpointon one of the sides and City, 1992),41. joining that point to the apex of the angle opposite that side, one bisects en 7. JavierE EscalanteMoscoso, Arquitectura Prehispanica losAndesBolithis angle into two anglesof 300. 2nd ed. (La Paz, 1994), 134. vianos, Problems Structure and Greek Architects Work: at of von und Welteislehre 33. J. J. Coulton,Ancient 8. EdmundKiss, Das Sonnentor Tihuanaku Horbingers 3rd ed. (Ithaca,1977), 66. Design, (Leipzig, 1937). 34. Ibid. 200 de 9. CarlosPonce Sanginds,Tiwanaku: ahios investigaciones arqueoldgicas von 35. Stiibel and Uhle, Die Ruinenstitte Tiahuanaco, II, 26-27. part 227. (La Paz, 1995), 36. William J. Conklin, "Tiahuanaco Huari:ArchitecturalComparand 10. EduardSekler,"Structure, in Construction, Tectonics," GyorgyKepes, in isons and Interpretations," William H. Isbell and Gordon E McEwan, in ed., Structure Art andin Science (New York,1965), 89-95. Structure: and Prehistoric MonumentalArchitecture eds., HuariAdministrative 11. Escalante's argument to the contrary notwithstanding. Escalante State Government (Washington,D.C., 1991),281-291. 218. Moscoso, Arquitectura Prehispanica, 37. John Whiteman, "Criticism,Representationand Experiencein Con12. Paul Goldstein, "TiwankuTemples and State Expansion:A Tiwanku Architecture: Architecture Drawingin anAge of Criticism," and temporary in Moquegua,Peru,"LatinAmerican Sunken-Court 4, no. Temple Antiquity 6 Harvard Architecture Review (1987): 140. 1 (1993): 22-47, quotationp. 24. 13. Ibid. 14. Heiko Priimers,Die Ruinenvon Tiahuanaco Jahre 1848: Zeichungen Illustration Credits im undNotizenvonLeonce del vol. 13 (Mainzam Rhein, 1993), 385-478. Una en Manzanilla, Angrand, Akapana: pirdmide el centro Figure1. AfterLinda 15. Alphons Stiibel and Max Uhle, Die Ruinenstdtte Tiahuanaco von im mundo (Mexico City,1992) Hochlande altenPeru:Einekulturgeschichtliche aufGrundselbstindi- Figure 3. AfterJavierE EscalanteMoscoso,Arquitectura des Studie en Prehispanica los (Leipzig, 1892). gerAufnahmen 16. CyrilM. Harris,Illustrated Architecture (New York, Dictionary ofHistoric 1977), 104. 17. See ProtzenandNair,"WhoTaughtthe IncaStonemasons," 13, p. 5. fig.

Andes 2nd Bolivianos, ed.(LaPaz,1994)


und von WelFigure4. EdmundKiss, Das Sonnentor Tihuanaku Horbingers teislehre (Leipzig, 1937)

Allothers Protzen byJean-Pierre

ON RECONSTRUCTING

TIWANAKU

ARCHITECTURE

371

Você também pode gostar