ili:*p**r l?

RnnxaMININc TTwaNAKU's Unsnx RpNEwALTHROUGH GNOUND-PPWNTRATING RNNNN

AND EXCEVATION:

THp Rpsurrs oF THREE FrBrn SpasoNS

Michele L. Koons

IxrnorucrroN

uring the 2002,2004,and 2005 field sea-

sons, over 100,000 m2 of ground-pene-trating radar (GPR) data were collected

to understand the spatial layout and arrangementof architectural features within the monumentalcore of Tiwanaku (fig. 12.1). The major objectiveof this survey was to better understand the archi-tectural transition and reorganization that oc-curred around AD 800, which roughly correlateswith the ceramic transition from Tiwanaku fV toTiwanaku V Since GPR data are collected inthree dimensions, the relative depths of structuresand features can be stratigraphically correlated todifferent construction sequences at the site. Based

on significant reflections in the GPR data, excava-

tions were placed in numerous locations aroundthe site. By incorporating excavations and GPRdata, we can project information from a series ofstrategically placed units across a larger area withgreat spatial precision. Using this method, nu-merous structures have been identified and par-tially excavated within the monumental core.

Many of these structures are related to the con-struction phases before and after the architecturaltransition that occurred around the ninth centu-

ry AD. This chapter will outline the methodolo-gy employed in the location of these excavations,

their archaeological yields, and interpretations.

BncrcnouxP

BetweenAD 500 and 1050,Tiwanaku emerged as

the primary ceremonial, political, and economic

center in the southern Andes (Browman 1978,

3 27 ; Coutare 2002,xvii; Janusek 2 004b). Situated

approximately 15 km southeast of Lake Titicaca

in highland Bolivia, at its apex G AD 800-1000)

Tiwanaku's influence stretched as far west as the

Chilean and Peruvian coasts and into the eastern

Amazon rain forest (Browman 1978,331; Cou-ture 2002,32). The central core of Tiwanaku was

the main trade hub for the region and consisted

of monumental architecture that was expanded,

rebuilt, and reintegrated over time while the dy-namics and organization of the city morphed and

changed.During the Late Formative period (250 BC-

AD 500),Tiwanaku was likely similar to other set-

tlements in the Titicaca Basin that were character-

izedby sunken court architecture (Hastorf 2005;

Janusek 200+b,109). By roughly AD 500 it had

become the largest center in the region, consisting

of the imposing Akapana, the Kalasasaya, and the

Semisubterranean Temple (fig. 12.1). Residential

compounds stood around the major monuments.The varying and distinct material assemblages

found within rhese compounds have led scholars tocontend that they were associated with specifrc ay-

llas, or kinship groups, from different communities

t47

148 Mrcnnrn L.I(ooNs

Akapana

FrcunE L2.I The 2002,2004, and 2005 GPR survey areas overlaid on an aerial photograph of the monumeni.core of Tiwanaku and highlighting the major monuments,

throughout the Tiwanalm sphere of influence(Couture 2002; Janusek 2003a, 2004b; Kolata2003c). The ayllu system is a dynamic alliance ofreal and fictive kin that worship a common ances-tor and sacred places in a landscape. It is well doc-umented for the Inca period and today, but isthought to have deeper temporal roots. Tiadi-tionally, people of one ayllu owned land in differ-ent geographical regions so that the members ofthat ayllu had access to multiple resources (Al-barracin-Jordin 2003, 96). Therefore, rhe com-pounds at Tiwanaku would have serwed ayllumembers not only residing at the site but alsofrom various hamlets and villages across the Tiwa-naku sphere of influence.

In addition to containing diverse materialculture, the Tiwanaku IV residential compounds,or barrios,were constantly renovated oveithe pe-

riod of at least three centuries leading up to :major architectural reorganization around AD800. This transition is also rnarked by a chanqe irvessel forms and iconography between TiwanakufV and Tiwanaku V ceramics found throughoutthe site (Janusek 2003b).

Around AD 800 the Tir.r'anaku fV compouncl-.were razed and ne\', more elaborate structureswith restrictive architectLrre were built, such as thePutuni complex and the final phases of the Aka-pana East complexes (fig'. 12.1) (Counrre 2002:

Janusek 2003a). It has been suggested that the archi-tectural transforrnation occurred because compe ti-tion between ayllu members residing within these

compounds increased as Tiwanaku influence ex-panded over the region. This shift in architecrurehas been interpreted as a move toward a more ex-clusive and elite-centric semlement (Couture 2002).

Cneprpn 12: RporuuINING TrweN'cKu's IJRSAN RtNBlvet r49

The research I present here, however, may

suggest a different prtt"ttt. Based on large-scale

GPR surveys and excavations, in some areas str-uc-

tures were razed and not rebuilt' Rather, theywere

cleared to make room for more public space dur-

ingTiwanakuV This increase in public space cor-

,"Ipond, with the height of Tiwanaku influence

thrtughout the region and could be a response to

an inc"rease in the number of pilgrims traveling to

the important ceremonial center' Below I present

evidence suggesting that the monumental core of

Tiwanaku *"s .h"ngi"g in a way that was able to

accommodat. -or" people' I contend that the

grander architecture was a synbol of the increas-

irrg i-portance of the demonstration of wealth

"n"d pt.ttige among the competi"g "/! groups'

b.rt .rot necessarily a qmrbol of elite exclusivity'

GPR AT TIWANAKU

GPR is a near-surface? remote-sensing technique

capable of subsurface three-dimensional ima gin g

(Cionyers 2004). This method is used to spatially

-"p ,obsorface features' such as middens, and an

"rrny of architectural features, such as walls and

floors. The basic GPR method involves one an-

tenna transmitting energy into the ground and

another antenna receiving electromagnetic radar

pulses reflected from buried objects or distur-

trn.", in the subsurface. For this project, radar

data were collected in equally spaced straight

ffansect lines within rectilinear grids' As radar en-

erg'y moves through the ground, the velocity of

th"""*urr", will change depending on the electric'

and to some extent, the magnetic, properties of

the materials through which they pass' A greater

contrast in the electric and magnetic properties

between two materials results in a stronger reflec-

don, and therefore greater amplitude, of the re-

tlected waves. A large number of subsurface re-

t'lections are collected along each transect line in

a grid. These reflections are collected in time

1nr'rror"conds [ns]), which can later be converted

into depth after determining the velociq' of the

energ'y traveling through the ground (Convers

and Lucius 1996).

The depth to which radar energ;v can pene-

trate is putii"lly controlled by the frequeno- of

the antenna. A lower-frequency antenna (in the

100 megahettz lMHzl range) produces longer

-avelen-gths thal can potentially p enetrate 1" "l

-

er into ih" gtonttd, sometimes up to 50 m in fa-

vorable conditions' A higher-frequency antenna

(in the 900 MHz range), however' wrll be able to

p.n"arr," less than f m in most conditions' but

*lat "

far higher resolution than a lower-frequen-

cy antenna'*A. trade-off therefore exists between

d.pth of penetration and resolution' More influ-

.rriirl ao ih" ,rr...r, of using GPR are the soil and

sediment mineralogy, ground moisture? and to-

pography. Electrically conductive or magnetic

*r?"titft will absorb radat energy very quickly

and prevent its transmission to depth, regardless

ri aft. frequency of antenna' In this project' sedi-

ments *"r" pri-"rily dry and the ground surface

was cleared of ,r"g"tation and other obstacles' We

were able to maintain consistent contact between

the antenna and the ground surface, producing

excellent energy propagation and resolution up to

about 1.5 m using a 400 MHz antenna'

The ability to measure distance in the ground

is one of the most powerful aspects of GPR' When

all transects collected within a grid are vertically

,lign"d, they can be processed to yield,amplitude

,h!" -upr, which are three-dimensional maps that

show the differences in reflected amplitudes across

a given surface at various depths' Archaeological

features are more readily apparent' and their spa-

tial arrangement is more easily identified'

2002 RT,SEARCH

In 2002, Kimberly llenderson of the University

of Denver collected a large radar grid measuring

220 m north-south by f60 m east-west' to the

west of the Akapana and directly south of the

Putuni complex (ng' rZ.tl' Prior to Henderson's-Cpn

,o*"y, ,cholars had hlpothesized that the

,t"" ,o the west of the Akapana was the focal

point of the urban landscape and the-center of so-

.iul, .nltor"l, and political activity during Tiwa-

naku fV and V (janusek 2004b; Kolata 1993'

2003c). As noted above, an architectural change

occurred around AD 800, where earlier buildings

were dismantled and destroyed to accommodate

the construction of new buildings and plazas'

This change in architecture reflects a shift in the

dlnamics -and

social organization of the city

150 Mrcnnro L.I(ooxs

ffiSlice 3. '-Tiwanaku

V

around the eighth cenrury (Couture 2002;Janusek 2004b; J. Moore iOVAtS. lfenderson(2004,81) positioned the GpR grid in 2002 totest whether or not this architecirral shift couldbe detected in distinct stratigraphic layers ofsub_surface remains in the GpR ieflection profilesand amplirude slice maps. Overall, her goals wereto understand what architecturrt nl"a socialchanges took place during the transition from Ti_wanaku fV to Tiwanaku V and to see if thesechanges could be detected v,ith GpR and small_scale excavations (Henderson 2004).

The results of Henderson's GpR survey

;h1wgd multiple sffucrures and areas of interest,including

-a recrangular 40 by 50 m strucrure, a

possible plaza area, and a sffucture in the south_ern sector ofthe survev area. Based on the archi_tecture apparenr in the amplitude slice maps andlinited excavailons (four 3 by 2 munits [L1_L4l),Henderson concluded that her amplitude slicemap 3 (6-9 ns, or 30-45 cm below the surface)

Putuni-likestructure

contained features and sffucrures related to Ti_wanaku ! and that amplirude slice map 5 (12_15ns, or 60-75 cm below the surface) representedsubsurface fearures from Tiwanaku fV ing. n.Z;(Henderson 2004).

2004 AND 2005 Excn.erroNs oF THEWEsr Sron or.T.rJE AK,A?ANA

Based on Henderson's processed and interpretecldata from 2002, in 2004 and 2005 I placed an a,l_ditional eighteen excavation unirs on the v.estside of the Akapana. These units ranged from 5 br5 m to 3 by 2 m and concenrrated & three mainsections of the sur-vey area. First, excavations \\,erelocated to investigate a possibl e plaza area locar_ed in the center of the grid wheie no significanrradar reflecdons were recorded. Secoid. unir-.were located in the norrhern portion of the sur_vey area in and around the 40 by 50 rn rectanzu_lar structure. Finally, excavations were placecl ir

Projeft P.A.P.A.2002GPR 5urvey Alapana5lice5{12-l5nsi

40 50 60 70 80 90 100110120Otherstructure

FrcuRp 72 '2 Hendettt"t :]t:" 3 ,(l'ft) and 5 (right) GPR maps annotated to highlight the architectural featuresof interest from Tiwanaku IV and i slice 5 ri,ro*, H".rd.ersont excavation units (L1_L4).

Cna.ptpn 12: Rppxu,rrMNc TrweNer<u's URBAN RlNpwer 151

:::e southern part of the sulvey area where a

.naller structure was identified in the amplitude

.-rce maps (frg.12.2).

ExcevATroNS To rHE WESTOF THE AKqPAN,{

The center of the 2002 GPR grid (fig. 12.2) showsi:i absence of significant radar reflections. Aerial::rotography of the site shows that this area is a

:.ctangular depression where water collects during--:-- $-et season, and both llenderson and I hlpoth--rzed that this was possibly aplaza (fig. 12.3). Rea-: rs for the lack of strong GPR reflections within-j-s area may be due to a high clay content and::oisrure level, which increase signal aftenuation.lhree 3 by 2 m excavation units were placed;.-rund the perimeter of the depression to see if a

'r:-rsurface material change that was not readily ap-:iient in the GPR maps could be detected to sug-i:st the presence of a plaza.

ln unit L5 we found a post-contact-period,uill lining the depression's eastern limits (fig.-1.3). IJnits L6 and L7 were placed to see if a

:- -,'rr surface could be identified, even though this::rture was not readily identified in the GPR:-aps. Excavations revealed a gravel surface, which, : identified as a floor. This floor had dense cul-:r:al deposits, including human bones, camelid-*rneS, and ceramics, and was about 40 cm below:-l: surface in L6, and 55 cm below the surface in

-5. The floor and associated ceramics and bones

from both of the units have been tentatively datedto Tiwanaku [V It is important ro note that thisgravel floor was not continuous across an entirelevel of L6; rather it was apparent in patches. UnitL7, although lacking the same gravel floor or thesame density of faunal and ceramic material, didcontain the remains of a looted burial and someutilitarian pottery. Since this burial was obviouslydisturbed and deeper than the gravel floor (about110 cm below the surface), it is possible that if a

floor did exist in the layers above the interment,it would have been destroyed when the burial was

looted.The gravel surface to the west of the Akapana

could be representative of a once-open area. Hen-derson (2004) described a gravel layer in unit L4,which was within the same general area as L6 andL7. However, the gravel stratum she identifiedwas roughly 20 cm below the floor found inL6,which may suggest that rather than one continu-ous plaza surface? there were multiple surfaces atdifferent levels. Excavations to the north of thehlpothesizedplaza area further support this claimand will be discussed below. Nthough no plazawas definitively identified, there were variousgravel layers found that suggest that this area westof the Akapana was reused and renewed multipletimes over the length of the occupation duringTiwanaku fV and V The artifact-laden surfacesare sharply contrasted with a very clean, artifact-free gravel floor found on the east side ofthe Aka-pana, as will be described below.

$l*- s*.'

wew$ Frcunp 12.3 Hypothesizedplaza area on the west side

of the Akapana, and theapproximate locations ofexcavations L5 thr ough L7 .

This photograph was takenduring the rainy season,

and water has accumulatedinside the rectangulardepression-

152 Mrcnnln L. Kooxs

ExcnvrrroNs rN THE NoRTHERNSecron Wnsr oF THE Arrpexe

A total of eight excavation units were placed justto the north of the excavations in the plaza area(fig. 12.a). Three of the excavations focused onthe east wall of the rectangular 40 by 50 m struc-ture visible in the GPR image (L9,L11, and L13).One unit was located inside the structure (L10),and four were placed to the east of the strucfurein other areas with significant GPR reflections(L8, LI +, L2 3, and L2 +). I hypothesized that thisrectangular structure, like the Kalasasaya and thePutuni, also had an entrance facing east. In theGPR reflection profiles (fig. 12.5),I noted in a pla-nar reflection what could be in situ architecture.Units L9 and Ll1 were placed to investigate thesource of this reflection and to see if the entranceto the structure could be identified.In L9 and L11we found a flagstone pavement and foundationwall (fig. 12.6) that related to the planar reflectionwithin the GPR reflection profiles. Only a 2 mstretch of the foundation stones were in situ. Thefoundation stones contained basal moldings that

indicate that they once supported a masonn- :--perstructure (Skidmore and Yates 2005).

To the east of the foundation was a red, si-.

sand construction fill with several set cut Stofl::llenderson found this same qpe of fill in unit L:which investigated the western wall of the sa:: .

structure (fig. 12.2). The same filI has also b.=-explained by both Vranich (1999), in the case

the Pumapunku, and by Couture and Samp...(2003), for the Putuni complex. The presence

this construction fill suggests that this was a p1'--

form structure with an interior court similar -

the Putuni. Given this, I refer to the struclure .'the Putuni 2, or P2.

To the west of the foundation wall within L -

was a gravel floor composed of small and mediu:,-size greenish gray pebbles. This gravel floor r'.-capped with red clay similar to the aggregate th,supported the walls of the Putuni (Couture 20t. i249). Abutting the gravel aggregate surface. i:-rectly to the south of the southernmost in sr:,foundation stone, was a flagstone pavement (fi:12.6). This pavement was made up of cut ir:-desite stones fitted together to form a tiled su- -

Ll3 . ".

* ,...

mrplitude

T::! ''o

!['*

m,;[J -''''I;:,'

ro#olitude

Highplitr"

fr@,*ri ....l r20(

ampl

Ftcunp 72.4 Location of excavations in the northernsector ofthe 2002 survey atea,

ll..e8&.4

25. ,

2a fir1E f.

L10 6

to* rfl'p**g.f--{

,

5F 1"* {'q

0r4g& ,A -: I .

0 5 10 1

LI1.#;.

'' gth::i

tr-:,.'t0 20 25

Grid 6

W

40 qls so ss ob 6s 70 7s 80

'w.w

"e,.

;*s*.-,:

w"'#**-"b

a* Cr

Grid 3

L't4. r

I&_''

;"

Crrqprpn 12: Rr'.Bxq.rr,rrMNG TrweNeru's URSAN RrNnwer 153

GPR Reflection Profile 45I

^1Cl;lFIFr

22222-5

FIcunp 12.5 Unirs L9 and L11.GPR reflection profile 457,showing reflection producedfrom the flagstone pavement.

Unit L9

Foundation

Unit Ll'l Profile45'l

Ftcunr 12'6 Units L9 and L11- Plan view of the eastern wall of the rectangular strucrure (p2), showin gtheflag.stone pavement and foundation stones.

face. The flagstone pavement was missing sec-tions, and several of the stones were uneven anddisrurbed. Farther to the south of the pavementwas a scatter of cut andesite blocks and a largestone block.

The andesite block in L11 could have been anorthostat, or a large stone slab, set berweencoursed masonry a grpical Tiwanaku masonrymethod (Protzen and Nair lggT).It also bearssimilarity to a lintel noted by Courure (2002) in

the structure she excavated to the east of thePutuni. The combination of paving stones and apossible lintel suggests that the area was near theentrance to this platform.

Unit L13 was located 15 m north of L9 in anarea where the GPR amplitude slice maps showedreflections of what we thought could be thenortheast corner of strucrure P2 (figs. 12.4,12.7).Within the first 30 cm, rocks were unearthed inthe northern profile of the unit, but none of them

154Mrcsnre L. KooNs

were shaped or appeared in siru' These jumbled pavemenr to rhe easr was a flexed buria.l of a .. -rocks were probably the remains ofa conduit, be- -rl" -r,rr craniar deformation oriented ro . .cause' although no srones were found in situ, the no;rth. Nthough the sfull fron this burial v.aspattetn of the surrounding rocks and

the- rench

";;;dy deliberately elongated, the bones r, .. .shape of very dark brown !T tt qpi*r of n siru n-Jr""iulyud ro nore the sqr)e (flat, round, or .,: -conduits noted elsewhere. riis conduit.was with- garrar) (BIon 200i).in the upper levels ofthis unit and mayhave been " irr. inrenr of L10 (fig. 12.+)was to inr-es..-associated with the platform structure. The con- grr. , p..p"ndicularintersection of two linear:--duit was also not very deep below the surface, i".tion, apparent in Flenderson,s GpR anrl -suggesting that it may have been used to remove rude s.hce map 5 between 12 ard 15 ns (60--,water ftom the top of the platform strucfure and cm). This intersection occurred in a deeper sJ: - .eventually emptied i1,o t larger conduit. than srrucru re p2,so I hlpothesized that this ic ,_The excavation of L13 continued to the south ture would b" .rriier (T#anaku l1,1.In the up;-..;oftheconduit'whileu'edidnotfindthecontin- levels of the excavation and about 35-rl0 c.-uation of the foundadon from L9 and Ll l, we did below the surface, we {ound a gravel surface tt ;. -

find the same consrructionfill.Below this fill u'e dense faunal ancl cerarnic rnaterials (fig. 1l ifound the remains of a flagstone pavement possi- The gravel floor surface was similar to thar e-r- ,

bly related to an earlier construction (fig. 12.7). vated in L6,butthe densiqy of artifacts was mucl,The pavement was not of the same quality as that greater. It is possible that th" floor. found i1 L 1rin L9 and Ll1 and was covered in places-w_tth a i,urpn.rof th"interiorcourq/ard of p2.Thepar-thin layer of green ash- This tr-" gr!.t-ash layer tern of thick debris depositiin, however, is ,-er.is associated with dedicatory burials from a Ti* different from the fine red clay that capped th.r'lanaku fV compound beneath the Putuni and in relatively debris-free Puruni interior (Courur.the Akapana East residential structures (Couture and Sampeck 2003). This suggests thar the func-2002;Janusek2004b). Situated slightlyabove this tion of P2 could have been different from th.

FIcrnp 12.7 Llntt L13 and GpRreflection profile 458. This unit con-tained a human burial (with a skullshowing tanial deformation) and apavement that was 80 cm below thesurface, Unit L 13

Trace Number

Crrtprpn 12: RpExc,\TIMNG TtweNer<u's IJRBAIv RBNBwar- 155

::nction of the Puruni. Associated with the floor',1, as a stone foundation found in the southeast:,ortion of the unit (fig. 12.8). This foundation'' all is apparent in GPR reflection profile 437trq. 12.8). When amplinrde slice map 3 is exam-

-red, smaller walls are apparent within the center,i P2, suggesting that it may at one time have:,rntained rooms (fig'. 12.9). This is also apparent--,, a different pattern than the architecrural layoutrf the Putuni, further suggesting that the func-

tions of these two platform structures were dis-tinct from each other.

Below the floor with dense cultural debris weencountered the source of reflection producingthe perpendicular reflection in the GPR maps(see fig. 12.2, shce 5). This was a cobblestone fea-ture and can be seen as a high-amplitude reflec-tion in the GPR reflection profile a30 (fig. 12.8).Based on what is known about the architecture ofTiwanaku, this cobblestone surface could be part

02468

10v11:t4u16ElA- lll

2224262830

GPR Reflection Profile 430

Unit L 10

GPR Reflection Profile 437

FtcunE 12.8 Unit L10, with a dense layer of culturalmaterial,cobblestone wall or feature,and a corner to a foun-dation- GPR reflection profile 430 shows the cobblestone waII/fearure, and GPR reflection profile 437 highlightsthe corner foundation.

FIcURE 12.9 Northwesternportion of the GPR gridfrom the 2002 survey area.

Walls inside structure P2

suggest that it may have

rooms.

unit t-to tn

unit L10

r56 Mrcnpr-p L. KooNs

/of a foundation of an earlier structure. In unit L2Henderson also excavated a similar cobblestonefeature 60-64 cm below the surface while investi-gating the approximately 50 m long north-southoriented reflection seen in frg.12.2, slice 5. Cob-blestone foundations similar to the feature in L10andL2 were found below the Putuni and belongedto a Tiwanaku IV elite residential str-ucture (Cou-ture and Sampeck 2003). Residential compoundssuch as La K'arafi4 Akapana East 1 Mound, andAkapana East 1 also contained cobblestone foun-dations similar to the feature in L10 (EscalanteMoscoso 2003; Janusek 2004b). I suggest, there-fore, that these cobblestone features were part ofcompound walls from Tiwanaku fV residentialstructures that were razed during the architecturaltransition. It should be noted that Henderson(200+,105-9) suggests that the north-south cob-blestone feature could be an aqueduct similar toone found atTiwanaku's "sister city" of Lukurma-ta, located in the adjacent Katari valley (Kolata andOrdoff 1996, 137). Either interpretation of thecobblestone feature would conform ro the overallpatterns identified at the site, indicating rhat moreresearch is needed. Regardless of whether or notthis feature was an aqueduct or compound wall,the once residential nature of the area west of theAkapana is further suggested by the archaeologi-cal findings to the east of structure P2.

Excavations were placed outside of the eastwall of structure P2 to investigate additional high-amplitude reflections that were hypothesized tobe other walls and floors not associated with P2

€tg. D.q- Units L8 and L14 were excavated dur-ing the 2004 field season; units L2 3 andL24, dtr-ing the 2005 field season. These four unitsdemonstrate evidence for multiple layers of occu-pation, far more than the layers just below thePutuni itself, as described by Courure and Sam-peck (2003). This evidence includes hearths, util-itarian ceramics, stone workshops, and multiplelevels of renewed floors, and suggests that duringEarly Tiwanaku f[ the monumental core was a

residential area similar ro the other residentialcompounds excavated around the core. At somepoint during Late Tiwanaku fV the residentialarea was renovated multiples times, as is evi-denced by numerous superimposed debris-ladenfloors. These floors could have been plaza areas

for the increasing number of festivals and c. ,

monies held within the core. A similar pattei:floors continues through Tiwanaku V It app.

that during the span of occupation the area t,- -' ,

east of structure P2 was consistently in flur ,-may have been utilized by a variety of difle ;. -

people for different purposes. as is suggest..the diverse and dense artifact assemblage t:.-more and Yates 2005).

The dense debris within the context o- -

gravel stratum from L5 and L6 and aroun.l : -

suggests that trash was either discarded in this .." .

or that the cultural debris was intentionalh- inc -

porated into the matrix of the gravel floor. B-

and Janusek (2004, 126) note that dedicatonferings are found in association with the c '

str-uction fill of many of the razed and r€fli',' :high-status structures at Tiwanaku. These o-. -'ings include dismembered human and cailr..remains as well as smashed ceramics and o:-:"materials, and are found in the fill levels oi -Akapana and below the Putuni.The densin- ol ,-debris, and the presence of human and can=-bones, may suggest that these surfaces also ac:.as dedications associated with renewal. Wher- .-the gravel surface on the west side of the Alap.:was itself an offering or just a floor with a sr.,-deal of debris, it demonstrates that dlis x1g2 -,', ,:

used for activities that were probably assocr.t:.

with great numbers of people participatins -large-scale events.

ExcevrrroNs rN THE SourHER\Secron WEsr oF THE AKcPANI

Excavations in the southern sector of the GP--

survey area focused on an area where the GPR r.-flections suggested a large, near-surface stmcru::(fig. 12.10). This structure, which we are referrin.to as Wl , was located roughly 1 0-3 0 cm belou' tr.surface. The full extent of W1 at this time has n, .

been excavated, but based on the images product -from the radar data,we can suggest that the struc-ture possibly extends to the south and east of th.excavated area. The radar image slice maps als

indicate another earlier structure below \\-1.which has been confirmed with test excavationsThe earlier structure is also quite shallow, and n,

more than 45 cm below the surface. Due to ti.

Crrcprpn 12: RBExl,rvrNrNG TrweNeru,s IJRSAN RBNpw,qr. r57

slightly uneven topographical surface and theshallow narure and varying depths below the sur-face of W1, at this time we cannot determine if allt.lle unexcavated linear GPR anomalies belong toWl or the earlier srrucrure. Hopefully this can beresolved in the future with more directed excava-tions or by employing a higher-frequency anren-na that can detect subtle changes with higher res-olution of the near field.

From 2004 through 2006, Wl was excavatedby C. Wesley Mattox. He found what we believeto be a small-scale ceremonial structure with aninterior patio, rooms or chambers surroundingthe patio, and a compound wall. The patio wailined with cut ashlars, which were likely removedfrom an earlier structure, since only the interiorperimeter was flush. "Borrowed" ashlars were alsoused to construct the sunken courward of theTiwanaku V Akapana East 1 north compound(|anusek 2003a). Nthough the interior of thepatio of Wl was not fully excavared, the radardata reveal that it is relatively free of internalstructures. llowever, a 2 by 2 m excavation of asingle area of high-amplirude reflections in thecenter of the patio uncovered a large andesiteblock, but its purpose and relation to the struc-ture at this tirne is unknown.

Ftcunn, 12.10 Southernsector of the 2002 GPRsurvey area with plan viewline-drawn map of theexcavation units and archi-tectur e encountered- Theline drawing is slightlycransparent so rhe GPRreflections can be seen.

Line drawing courtesy ofW. Mattox.

To the west of the patio is what we believe tobe a water-retaining feature, possibly akin to a re-flection pond, with laminated sediments. It wassurrounded by a rectangular foundation ofashlarsand divided into a northern and southern area byan internal ashlar pavement. The ashlars are alsorecycled stones from another structure. A verysmall and finely constructed drainage conduitwith a stone stopper was located in the northeastinterior wall of the fearure. This suggests that thefeature was periodically filled with water anddrained, possibly in association with certain cere-monies performed in the strucrure. The signifi-cance of this and other water features on the sitewill be discussed below.

To the south of the patio, rwo 4 by 2 m roomshave been identified that may have been used forthe staging of small-scale ceremonies that tookplace within the compound. To the north of thepatio, we believe we have identified the northerncompound wall. Excavations to the north of thiswall revealed a great deal oftrash and debris, sug-gesting that it was outside the perimeter of theformal structure (Wes Mattox, personal commu-nication). Within the GPR image slice maps, thecompound wall appears to extend to the east be-yond the excavation and abuts with a north-south

158 Mrcrrm,n L. KooNs

/

I

running wall. The radar image also suggests thepresence of numerous other walls and features tothe south and east of the excavated area.

Preliminary ceramic analysis suggests thatWl dates to the final phases of Tiwanaku. How-ever, its association with the deeper structure,and, therefore, the changes that occurred be-tween the uses of these two strucfures, remainsunknown.

200+ AND 2005 GPR oxTHE EAST SInp op THE AI.APA\-I

In 2004 and 2005I collected an additional6,i^m2 of GPR data on the east side of the Akap" .

(fig. 12.1). A modern fence surrounding the ir:..'monuments made this survey difficult, which is r, -

there is a diagonal strip of data apparently mis.,- -

flom the GPR maps, seen here in black (fig. 11. - -

Stones from theKantatallita

Ftcunp 12.11 Annotated GPR survey area from 2004 and 2005 showingvarious strucrures, fea-tures, and conduits.

lp ';L*.Walls -

Structures orterrace

0

Cntprpn 1 2 : Rprxu,rrNrNc TrweNeru's URBAN RrNpwer r59

The amplitude slice maps produced fromprocessing the GPR data show a circular featurero the south of the survey area, large blocks likelyassociated with the Kantatallita, numerous con-iuits, and four other possible structures (fig.11.11). Several of these fearures were investigat-ed through excavation during the 2005 field sea-son and will be discussed here.

ExcevetoNs oN THEEesr Smn op rHE AKAPANA

Ln 2005, excavation units were placed in variousl,rcations on the east side of the Akapana to inves-dqate the source of significant radar reflections

from the 2004 data set (fig. 12.12).t In excavationunits 51 and 52 we encountered in situ architec-nrre in the form of north-south linear walls offoundation stones (fig. 12.13). This shallow ashlarfoundation was potentially associated with theKantatallita, and the stones were likely reusedfrom an earlier structure, much like the ashlars inW1 and the Akapana East 1 north compound. Aclean gravel floor was encountered roughly 30 cmbelow the surface and just below the base of thefoundation stones, suggesting that the foundationwas a later addition. Situated within the gravelfloor in the northwest corner of unit S1 was an el-evated capstone of a conduit. This capstone was

concave and had a hole in the center that was

15

10

Units 59 and S1O

Flcunr 12 . 1 2 Northern portion of rhe 2004 and 2005 GPR survey area, wirh the locations of the 2005 excava-;ion units highlighted.

MrcHBr-n L. KooNs

Flcunn 12.13 Units Sl tothe north and S8 to thesouth show the foundationstones that created thehigh-amplitude refl ectionshown in black in figure12.72.

FIcunp 12.14 Conduit in thenorthwest corner of unit S1. Theconcave stone with a hole is at thesame level as the clean gravel

floor. The round stone in the cen-

ter ofthe concave stone is a plugfor a hole that would have

drained into the conduit below.

plugged with a perfectly fitted carved stone (fig.12.14).This feature was obviously a drain, as fur-ther excavation revealed the conduit to continuedirectly below the level of the gravel floor. The

gravel floor sloped toward the center of the ele-

vated capstone so that water would have been

funneled through the hole and into the conduitsystem below.

I

Crrqprpn 12: Rrinxlr,rrNrNc TrwaNaru's UnsAN RrNBwer r6t

What I hlpothesize to be rhe same floor wasencountered in units 53 through 57.In 55 and 57this floor was roughly 60 cm below the surface; in53 and 54, about 45-50 cm below the surface; in56, about 50 cm below the surface. Excavation bythe Direcci6n Nacional de Arqueologia de Bolivia(DINAR) on the immediate east side of the Aka-pana in 2004*5 also revealed what may be thesame floor. The er.idence that this one continuousfloor extends over such a large portion of the eastside of the Akapana suggests that this was onelarge open area. The surface of the gravel floorrvas relatively clean, but ceramics from just belowthe level of this floor were mainly monochromeand utilitarian in narure. Due to the cleanliness ofthe floor, it could nor be dated, but the dearth ofstrucnrres and features above the floor suggeststhat it was in use in the final phases of Tiwanakuoccupation. This pattern of clean floors is also ap-parent in Akapana East and the K'arupata plat-form at Lukurmata and suggests a potentially im-portant pattern, which is especially apparenr incontrast to the densely debris-laden floors on t1lewest side of the Akapana $anusek 2004b).

Servicing the gravel floor surface in 56 and 54were conduits, which can be easily seen in theGPR amplitude slice maps (fig. 12.11) and thephotograph of the excavation (fig. 12.15).Associ-ated with the conduit in 54 was a burial of a

bound female, face down and oriented east. Theunique orientation along the route of the winrersolstice axis and face-down position with hands

possibly bound behind the back indicate that thisperson may have been a sacrificial offering. Theburial was within the gravel floor and within thesame level as the capstones of the conduit, sug-gesting that the conduit and human offering wereassociated.

Directly below the gravel floor in 55 and 57were the remains of walls and associated domes-tic undecorated ceramics (fig. 12 . 1 6). This followsthe pattern established throughout the site, whereearlier structures were razed to accommodatenew construction. In the case of 55 and 57, theearlier structure was razed and replaced with a

gravel plaza.The GPR amplitude slice maps show anoth-

er structure immediately to the east of unit 57.This str-ucture is on the same level or a levelabove the gravel floor (roughly 45-50 cm belowthe surface) and has a double exterior wall andfairly large rooms (figs. 12.1 l,12.I2).However, itis also possible that these reflections are associat-ed with a series of rerraces descending to thenorth (see Ernenwein and Koons 2007). Thesffatigraphy of 53 may confirm the presence ofthe terraces. Within 53 we noted the gravel floor,and below this floor was a large amount of debrismixed with clay and silt and containing large lens-es of ash. Quite possibly. this was fill placed tolevel out and terrace this northwardly slopingportion of the site. Flowever, this debris is similarto the debris found outside of the residentialcompounds farther east (lanusek 2003a), which

FtcuRp 12. 1 5 Plan view of units 54 and 56, showing the conduits present in both. Just above and encompass-ing the Iayer of the conduits is a clean gravel floor. A sacrificial victim is in association with the conduit in 54.

162 NlrcHpr-s L. KooNs

Figure 12.16 Units 55 and 57,facing west. The origin of theGPR reflection in profile 34was the Tiwanaku IV or earli-er wall. The subtler reflectionabove the earlier wall was

from the gravel floor. Unit 5

could indicate that the reflections correspond tocompound walls. At this time, the source of therefl ections needs more investigation.

Trw,qxeru Wemn Mexecnupxr

Based on patterns within the GPR maps, togeth-er with the excavations of 51, 54, and 56 andother known excavations at the site, it can be de-termined that the numerous, linear, noncardinal-ly oriented reflections correspond to conduits.On the east side, there are many shallow conduits,mostly between 45 and 60 cm below the surface.Based on the excayations, we know that they arelocated directly below, or within the same level as,

a patio or plaza surface. By tracing the patternsand depths of the highly visible GPR reflectionscreated by the unexcavated conduits, we can de-duce the location of plaza areas. These patternsindicate that a huge open area characterized theeast side of the Akapana during the later phases ofsite occupation.

Conduits are also present on the west side ofthe Akapana. FIowever, these occur at va4ingdepths, suggesting that they were either con-structed at different times or served to drain dis-tinct surfaces not associated with a contiguousplaza like that present on the easr. The gravelfloors at variable depths on rhe west side do sug-gest that that there were multiple levels of open

GPR Reflection Profile 34

space drat may have been renovated at diif.:.times leading up to the transitiot .4t tto6sd. ri

I suggest to be a Tiwanaku fV compound r a1-

cated in the center of the west side GPR grid .

razed and never superimposed with other ar.tecture, just gravel surfaces. Therefore, it app. ,

that the west side of the Akapana, like the . '

side, was also open space during the final Ti'naku V occupation, although it may har,e b..confi gured differently.

Aside from the numerous subterranean c(

duits, a series of circular features were identii:.to the immediate east of the Akapana. Fist-, ',

12.11 shows the largest of these circular featu:".'in the center of the eastern GPR grid (labe i-"pool"). This particular feature has a diameterroughly 40 m. Evidence suggests that these circ --lar features were stone-lined and once filled r r:

standing water. The lack of GPR reflecdons t:--side the circle suggests that there was more cl.here than in other parts of the suruey area. Hi!.-clay content causes GPR signal attenuation antl ,

]ack of depth penetration. Tv,o excavation unt,.(S9 and S10; see flg.12.12) were placed inside thr.feature and confirmed that sediments were excefr-

tionally hard and clay-rich, with little intern,:-stratigraphy. Sediment consolidation like this oc-curs from standing water. It is also significant th::two andesite-lined circular pools were noted i;rphotographs from the 1890s by the Gennan ar-

Distance (m)

CnelrBn 12: RpBxu,rrNrNG Tru'eNeru's UneAN RpNpuar t63

chaeologist Max Uhle. One of these was to thesouth of the main core and near the Pumapunku.The location of the other pool has not been re-identified. It is possible that one of the poolsidentified through our geophysical method couldhave been the pool photographed by Uhle.

These pools are strikingly similar to qochas, orrain-fed pools, seen throughout the Andean high-lands. Qochas are agricultural features that dateto as early as the Initial period and continue to beused today (Erickson 2000). They are mainly cir-cular, but oval and small rectangular ones do exist.Most range from 0.1 to 4 ha in diameter (Erick-son 2000, 338). They have man1, functional uses,including water storage during the dry season,growing crops, and as a source of drinking water(Erickson 2000, 340-41). Throughout the Titi-caca Basin, series of qochas are connected to eachother by complex canal systems.

It is possible that the pools found to the easrof the Akapana were similar to qochas yet with a

more ritualistic or ceremonial function. Thesepools are in direct association with the elaborateconduit system, and their proximity to the monu-mental core suggests that water ceremonies playedan important role at the site. The water featurefound in structure W1 also suggests that r rater wasintegral to Tiwanaku ceremonial life. Further-more, Kolata (1993, 104; 2003b, 184-87) notesthat the Akapana contained an elaborate drainagesystem, which he claims synnbolically associated itwith a sacred mountain, such as Quimsachata tothe south or Illimani to the east. Kolata (1993)also notes that a moat surrounded the majormonuments, which can be directly related to theintense relationship the people of Tiwanaku hadwith water. These combined data sets demon-strate that water management and manipulationwere very important features of the site and de-serve further investigation.

Su m,rARy oF RESULTs

Based on the archaeological and geophysical re-search conducted between 2002 and 2005, wenow have a much clearer picture of the dynamicnature of the monumental core. The architectur-al transition that occurred sometime around theninth century was likely a slow and varied process

and cannot be viewed as a singular event. This issuggested by the numerous gravel floors encoun-tered on the west side of the Akapana that tenta-tively date to Late Tiwanaku IV The last phase ofthe site (Tiwanaku \D otr both sides of the Aka-pana is marked by fine ashlar foundations in manyof the constructions, including str-uctures W1 andP2 on the west side and the Akapana East 1 northcompound on the east side, and large open spaces

(Janusek 2003a,2004b). The fact that many ofthese buildings used "borrowed" ashlars does re-main puzzling, however. Nonetheless, it fits with\&anich's (1999) model that considers Tiwanakuto be like a religious amusement park, with theelaborate architecture visible to the public andthe "rougher" architecture hidden behind thescenes.

It is also apparent from this study that al-though there are similarities in the character ofthe east and west sides of the Akapana, markeddifferences do exist. This is most notable in thepresence ofa clean, artifact-free floor that extendsacross much of the Ilatarea on the east side of theAkapana and was in use during the final phase ofthe site. This clean floor covered earlier domesticstructures that were razed, suggesting that thepeople who were once living in close proximity tothe Semisubterranean Temple were displacedfrom the center of the site to make room for morepublic ritual space. Where these people were re-located is uncertain, but the numerous residentialcompounds immediately outside the main corewere densely occupied during Late Tiwanaku IVand Early Tiwanaku V (|anusek 2003a). There-fore, it is possible that they did not move very far.The clean floor on the east side is contrasted withthe dense and multiple levels of artifact-ladenfloors on the west side of the Akapana. This di-chotomy may be of some significance. The arti-fact-rich surfaces west of the Akapana may be

dedications associated with episodes of construc-tion renewal, as discussed. In contrast, clean sur-faces like those found in the open space east of theAkapana are noted inside residential compoundsand the Putuni. Whatever the meaning behindthese noted patterns, they indicate that very dif-ferent activities were performed on the differentsides of the Akapana during the later use of thesite.

164 Mrcrmrs L. KooNs

/ Numerous conduits are present on both sidesof the Atapana. On the east side, tracking theseconduits allows us to trace the level of the contin-uous gravel surface and determine its depth ofroughly 45-60 cm below the surface.Alarge pool,and possibly a series of other pools-interpietedfrom the integration ofmultiple geophysical meth-ods (Ernenwein and Koons 2007) and excavation

-exist on the east side of the Akapana, demon-smating that our understanding of the importanceand significance of water managemetrt

"rrd manip-

ulation at Tiwanaku is still in its infancy.Overall, the changes that occurred around AD

800 incorporated finer architecture and more openritual space. Not coincidentally, it was around thissame time that Tiwanaku influence reached itsapex across the region and can be seen in the re-mains from the Moquegua valley, peru, and Co-chabamba, Bolivia. Therefore, the introduction ofpublic space, along with finely faced architecrure,created a public arena for the increasing numberof people traveling into Tiwanaku with the ex-panding sphere of influence.

Nthough today the word Tinkain manywaysis synonymous with ritual battle, the acruaimeaning in Quechua is "coming together" toachieve balance, and where this occurs is consid-ered sacred. In this respect, the term refers to an\-rwo parts that come together to make a whoie.This can be two branches of an irrigation canaimeeting up to form one large branch, the conflu-ence of two rivers, or the seam holding twopieces of cloth togerher (Harrison 1989; OssioAcufra 1992). The coming together of differentf'actions or moieties of an ayllu or multiple ayllusis also a form of Tinku. These unions occur ar a

central place known as a marka (see Janusek2004b,4t- 45).

Highland Bolivia in the past, as well as todal,,consisted of small hamlets or villages dispersedacross a region. Kin groups from these dispersedsettlements converged at a marka at imporranttimes of the year to participate in ceremonies andritual acts to achieve societal balance. The cre-ation of more public space at Tiwanaku at theheight of its regional influence suggesrs that itwas a marka. Competition among ayllu factions atthis center would have been fundamental to thefunctioning of the sociery as a whole.

As I have discussed, the different residentialcompounds around Tiwanaku have been associat-ed with distinct ayllu groups. ll'Jris interpretationis based on the differenr paterns of material cul-rure found in rhe different compounds (see Cou-ture 2003; Courure and Sampeck 2003; EscalanteMoscoso 2003;Janusek 2003a; Rivera Casanovas2003). As the importance of Tiwanaku increased,more and more people were traveling to the cen*ter, and the competition between ayllus also in-creased. The more elegant and elaborate archi-tecture has been interpreted as a reflection ofthisincreased competition and exclusivity (Couture2002). lfowever, rarher than this competitionleading to a more exclusive and elite-centric set-tlement after AD 800, the evidence I present heresuggests that it fostered a more inclusive environ-ment to accommodate pilgrims from all aroundthe south central Andes. In all the previously in-vestigated areas of the site, the earlier compoundswere razed and renewed with the more elaboratelater compounds. Until the research presentedhere, there was no evidence that some of the com-

Drscussrox

The ideas I present here about the jnclusive or-ganization of the monumenral core of Tiwanakuaround the time of the architectural transition atAD 800 are in line with whar is known aboutAn-dean social organization in general. The ayllukinship sysrem, which was largely based on dual-ism and the "balancing" of binary oppositions,was a major factor in the way Andean people or-ganized their settlements and other aspects oftheir lives for thousands of years (Gelles 1995;Goldstein 2005, 30; Zuidema t9B3,l9g0). Todayas well as in the past, this balancing was achievedthrough different means, including ritual battlesknown as Tinku. In Inca times, the state spon-sored and oversaw these competitions, where thedifferent ayllus took rurns leading the ,,battle."

The alternating of the "winners" and ,,losers,' ofthese battles was viewed as a necessity to achieveequilibrium within the society (Gelles I9g5).Overall, the competitiveness between ayllus func-tioned to catalyze and energize labor relation-ships and communal projects so that everyonebenefited from their ourcornes.

Crrprpn 1 2 : RBBxaITNING TrweNer<u's IJRBAN RBNBwar t65

pounds were razed and never rebuilt. Further-more, previous research did not indicate that peo-ple once residing immediately next to the Aka-pana were forced to move away from there toaccommodate more plazas and open space duringthe latter years of site occupation, as I have shownhere. \A,4th the more open space, the elaborateTiwanalm V architecture would have been visibleto many people who could now congregate in themonumental core. The elegant architecture canstill be interpreted as evidence for increased com-petition among the elite. F{owever, it also couldhave been used to instill a sense of awe in the pil-grims and, therefore, was a syabol of the overallpower and influence Tiwanaku had in the region.

Overall, I contend that the evidence present-ed here demonstrates that Tiwanaku was wherepeople from various regions came together.Moreover, it has been suggested that the wordTi-wannku is a derivative of Taypicala, which in Ay-mara means "stone in the center" (Escalante

Moscoso 1997, l0l). The competitive yet inclu-sive nature of the practices performed in the coreis reflected in the architectural changes that oc-curred over time. This interpretation is in linewith what we know about Andean organizationalprinciples and shows that these practices are root-ed in antiquity.

Coxcrusrox

The methodologv used here is a unique and effi-cient way to understand the potential significanceoflarge spaces containing buried architectural el-ements of the past, elements that were utilized formultiple purposes over long periods of time. TheGPR method of three-dimensional mapping inte-grated with selective excavations on such a largescale is a distinct way of conducting archaeologi-cal research. Placing strategically located excava-

tions based on significant radar reflections allows

us to extrapolate information from limited excava-

tions across a much larger area. Tladitionally, sin-gle compounds or complexes are completely orpartially excavated, then placed into the context ofthe entire site. Individual or isolated excavations

were not seen as a valuable research tool because

of the incredibly complex stratigraphy and multi-ple layers of occupation present at the site (Cou-ture 2002). Flowever, mywork shows that contexts

ofindividual excavations can be projected onto agreater area when GPR reflection profiles and

maps are consulted. Therefore, individual excava-

tions can then be used to tell a great deal about thenature of subsurface cultural remains that wouldotherwise be invisible. By employing the GPRmethod and selective excavation, I was able to il-luminate the changes that occurred throughoutTiwanaku's monumental core around AD 800.

This information can be integrated with data fromeach individual building to tell a more complexand holistic account of the changing nature ofTiwanaku's sociopolitical dynamics.

Norp

1. Unlike the units on the west side of the Akapana, the

units on the east side were named in accordance witha grid established by the Bolivian government. Be-

cause the numbering system is cumbersome and con-fusing, I renamed the units for the purposes of thisproject:

Original Unit Number

UnitN2059,E1015UnitN2062,E1018Unit N 2061,81026Unit N 2046,E 1023

Unit N 2050,81022Unit N 2049,8102+Unit N 2051,E 1022Unit N 2059,81014Unit N 205+,E 998Unit N 205+,F.999

New Unit Number

S1

S2

S3

S4S5

S6

S7

S8

S9

s10