Variability in the use of Ob1 and Ob2 material

Earlier it was noted that there was variability in obsidian knapping quality both in original geological exposures and in artifact materials. Obsidian containing heterogeneities such as air bubbles, ash particles, fractures, and occasionally inclusions, were observed in different areas of the Chivay source. While the majority of tools were produced on homogeneous glass (referred to here as "Ob1"), a substantial fraction of flakes and cores, and some tools, were observed in the "Ob2" material containing heterogeneities. When the Ob2 heterogeneities are very small, the knapping quality of the material appears to have not been compromised albeit the material is visibly less pure. A large percentage of the flakes in Block 3 were Ob2 obsidian, while all of the projectile points made from Ob2 obsidian were found in Block 2 in the high puna.

Visual characteristics of Chivay obsidian vary across the source area as well. Colors recorded for artifacts during the phase II lithics analysis included black, grey, transparent (clear), and dark banding was visible in grey and transparent obsidian. Banding resulting from concentrations of tiny magnetite crystals is common in light grey and transparent material. High numbers of transparent flakes were encountered in Block 1 and Block 2, which is along the direct route from the Chivay source to the Lake Titicaca Basin where obsidian from the Colca is known as the "transparent type" (Giesso 2003). Obsidian color and translucence was difficult to quantify during lab analysis, and it is possible that the distinction between grey and clear obsidian may be connected to flake thickness where thinner flakes appear to be less grey due to greater quantity of light passing through.

Temporal control

Temporal evidence in this area came principally from three radiocarbon dates acquired from the workshop test unit [Q02-2u3] that placed activities in the workshop area between the Terminal Archaic and the Middle Formative. These dates are important because they are earlier than evidence of discrete production activity has been found elsewhere in the region, and furthermore it serves to explain the dearth of ceramics in the workshop area because the activities begin during the Preceramic period.

Unfortunately, quarrying at the Q02-2 quarry pit itself was not able to be directly dated because no datable organic materials were encountered in the Q02-2u2 test unit. Inference from three possible lines of evidence suggests that the excavation of this quarry pit occurred during the Early Formative Period (2000 - 1300 cal BCE). These lines of evidence include:

(1) Analysis of the workshop Q02-2u3 test unit, 600m downslope of the quarry pit in question, shows evidence of the arrival of larger cores and intensification of production in a distinct event during the Early Formative.

(2) The total lack of ceramics associated with the quarry pit itself, the aceramic ancient road "Camino Hornillo" [A03-268] leading away from the quarry, and the sites in the immediate vicinity of the quarry are consistent with the early dates of activity at the workshop as ceramics should not be expected.

(3) The only diagnostic artifact found along Camino Hornillo [A03-268] was an obsidian projectile point in the 4f style that is diagnostic to the Terminal Archaic.

It is possible that obsidian hydration dating may provide evidence linking obsidian flakes excavated from the quarry pit Q02-2u2 with obsidian from dated layers in the workshop Q02-2u3 unit through relative dating. Brooks (1998: 447-451) ran obsidian hydration on ten samples with Glascock from MURR on obsidian from Juscallacta, but the resulting dates were far older than expected, confirming a general skepticism among archaeologists regarding hydration dating as an absolute dating method. An application of hydration dating at the Maymeja quarry would have to overcome two principal obstacles. First, hydration dating has been shown to be unreliable in areas with large temperature fluctuations and particularly in places where there are diurnal temperature changes (Anovitz, et al. 1999) as is certainly the case in the Chivay source area. Second, the samples from the two contexts may not be comparable because the obsidian flakes from the workshop unit Q02-2u3 are often in saturated soils as they are adjacent to a bofedal, while in contrast the obsidian at the quarry pit Q02-2u2 is in very dry sandy ash, and therefore environmental moisture is much greater in one circumstance than in another. Obsidian hydration rates are sensitive to the amount of moisture in the vicinity of the obsidian sample.

Intensified obsidian production

Episodes of intensified production were apparent in test excavation units at both the quarry pit and the workshop. At the quarry pit, the upper most levels (2 and 3) of the test unit showed evidence of concentrated activity, and because this test unit was in a debris pile, the strata are reversed so that levels 2 and 3 are the result of some of the deepest quarrying work at the quarry pit. In level 2a a number of discarded obsidian nodules were encountered measuring approximately 7 cm in length, and the thickness and angle of these strata indicated that quarrying was active and was resulting in a build-up in the discard zone. The low quantities of bifacial retouch or culturally-flaked stone of any kind in these upper levels at Q02-2u2 suggests that during the final episodes of excavation at the quarry it was purely whole nodules that were being extracted and transported away.

At the Q02-2u3 workshop, down the slope, intensified production was most apparent in levels 5 and 4 with evidence of the availability of larger cores taking the form of the discard of large cores and cortical flakes. These levels are interpreted as representing greater regularity in reduction strategies because they were thicker levels and production was more abundant, yet artifact morphology and variability was consistently low. In level 5, evidence was encountered that suggests that some proportion of cores were being transported away while others, even relatively large cores, were discarded at the workshop. In level 4, a similar strategy was in place, where it appears that medium and large cores with certain desirable characteristics were exported after some reduction, and some large flake blanks were also exported, but a fairly large percentage of cores and large flakes were discarded, suggesting that there was an abundance of material. Level 4 was thicker than other natural levels and it showed distinct evidence of abundance in large nodules, and relatively wasteful production, with large cores being sampled and discarded, along with the discard of large flakes. The presence of Kombewa flakes in levels 4 and 5, a flake-as-core technique, suggests that some variability in knapping strategies was practiced.

These characteristics changed dramatically in level 3 when discarded cores and flakes became much smaller and more advanced reduction seemed to have been occurring. The number of retouched flakes drops significantly, but concurrently the number of broken bifaces increased to thirteen, indicating that some advanced reduction was occurring. Activities in level 3 are distinct from all other levels and difficult to characterize because they seem contradictory. On the one hand, reduction strategies were more variable as some advanced reduction was occurring and heavy rotation and conservation of cores was taking place; while on the other hand, there were relatively large cores (cluster #3 type cores) being discarded as well as some cortical flakes in the level 3 assemblage. The peculiar pattern detected in level 3 may be due to a combination of a return to more local and variable production, combined with scavenging from the richness of the discarded material in the previous occupation level. By the ensuing level, level 2, the assemblage appears to have continued the pattern of variable reduction strategies but with lessened access to large nodules. In sum, the evidence from intensification at the quarry and the workshop suggests that workshop reduction began in the Terminal Archaic, quarry excavation and workshop production intensified and became more regular during the Early Formative, and ultimately during the final levels workshop activities returned to more haphazard localized production.

Links to the pastoral economy

The final pattern evident in the intensified obsidian production that was documented in the southern part of Maymeja is the correlation between pastoralism and obsidian production. There is the simple fact that it is physically easier for pastoralists to access and circulate obsidian. Burger and Asaro (1977: 41) note that tool-quality obsidian is found on geologically young volcanoes which, in the Andes, are from the Late Miocene and are typically found above 4000 masl where pastoralists reside. Furthermore, herders have the cargo animals to transport nodules. Based on the 2003 survey of the Chivay source, obsidian is available in a number of locations between 4950 and 5000 masl on the flanks of Cerro Hornillo. Obsidian lag gravels, often material with heterogeneities, blanket the Maymeja area and occur on a wide range of slopes on the eastern flanks of Hornillo. Obsidian is also eroding from a gully on the north side of Maymeja in an exposure labeled Q02-1. While there is reason to believe quality obsidian nodules lie underneath the tephra rich soils in a number of areas around Hornillo, it is surely no coincidence that the most high volume and persistent source of water in the region is adjacent to the Maymeja workshop, and the quarry pit is upslope to the east, as close to the workshop as possible along the 4970 masl contour.

Three principal patterns link obsidian production with pastoralism in this area:

(1) Maymeja production. The Maymeja workshop is adjacent to a water source and a rich grazing area, and a maintained, but relatively clean and unoccupied, corral [A03-127] sits adjacent to the Maymeja workshop. Construction in this area follows the margin of the rich bofedal, small circular structures and eroded terraces were evident here, and obsidian flakes litter the area. The second highest densities of flaked obsidian were found in the lower, northwestern portions of Maymeja close to the modern estancia [A03-570]. It is possible that these two zones were occupied simultaneously, as each residential sector adjoins a separate, large bofedal.

(2) Local Scale. Throughout the highland portion of the 2003 survey, especially in Block 2, concentrations of obsidian flakes were almost always associated with pastoralist site features. A road leads away from the quarry pit that suggests that loaded caravans departed to the south from the Chivay source along a route referred to as Camino Hornillo. This route is in an unsurprising location, as the road climbs out of Maymeja along the least-steep possible route, and following this road caravans could soon join the regional trail network near Lagunas Lecceta, just a few kilometers to the south.

(3) Regional Scale. The consumption of Chivay obsidian is largely associated with pastoralist sites. It expands on a regional scale in the Terminal Archaic with pastoralism, and on the western foothills of the Andes both obsidian and herding corrals are rare below an altitude of 1200 masl.

The simple association of obsidian with pastoral activities of shearing and butchering appears plausible on one level and such applications of obsidian probably facilitated its widespread distribution (Section 3.6). Despite this functionalist association between obsidian and pastoralism, several particulars of obsidian consumption patterns run contrary to this view. First, high-quality cherts with excellent fracture properties are available throughout the Titicaca Basin and given the quantity and size of Titicaca Basin herds, particularly in the Late Formative and onwards, one would expect much greater quantities of obsidian consumption as herd sizes grew. In fact, obsidian consumption appears to have been relatively reduced as populations and herds grew during Late Prehispanic times. Second, in the Chivay obsidian consumption zone sufficiently large flakes for shearing and butchering (for example, 2 cm or larger) are relatively rare in both surface and excavated contexts. Curation of obsidian in tool form does not appear to have been a priority except for in the production of projectile points.

The links between pastoralism and obsidian circulation appear to have more indirect than the functional connection would suppose. The development of regional caravan exchange, including long-distance circuits and the relative ease of transporting weight, served to move stone into regions where such materials were rare. The second probable connection is the social and symbolic link that obsidian provides between consumers, non-local trade networks, and representation of alliance and affiliation. Obsidian possession was one of a number of ways of differentiating ones self. Evidence of access to non-local raw materials, mediated through camelid caravans and pastoral economy, was one possible way to demonstrate social connections and access to goods and knowledge in other forms. Additional aspects of the social and symbolic links with obsidian exchange will be discussed below.

Pastoral bases and caravan stops

The most extensive pastoral features in the Upper Colca survey area lie in the dense Block 2 occupation along the eastern toe of the Huarancante lava flow. The area has been occupied fairly consistently since the Early Archaic and a distinctive land use pattern was documented in 2003 related to pastoral occupations, including corrals and associated residential structures, along the margins of the lava flow. As is characteristic of pastoral occupations, the structure of these sites is blurred by multiple non-contemporaneous occupations, refuse disposal is shallow, and stone corral walls and rock shelters serve as the few permanent feature types that persist through time from previous pastoral occupations (Nielsen 2000: 480-483). The Block 2 survey revealed that many of the larger estancias contain multicomponent sites, indeed most of these choice locations are occupied to this day in some capacity. However, the small and mid-sized pastoral bases in the area show more variability, as ceramic distributions suggest that some sectors were settled while others were abandoned at times in prehistory. The number of active corrals and the herd size potential in a given time period is difficult to estimate due to this pattern of shifting locations for small and mid-sized pastoral bases.

Based on systematic survey work in 2003 it appears that a number of the animal control features in Block 2 exceed the local capacity for seasonal grazing and were perhaps linked to short-term stays by passing caravans. These large animal control structures primarily take the form of vestiges of corrals that appear as bases of stone enclosures on top of low mounds rising from surrounding pampa. While it is difficult to estimate animal numbers due to shifting pastoral occupation, seasonal changes in land use, and the brevity of the 2003 observational period, the conditions in Block 2 appear to be the "Caravan rest areas" that were first discussed in Section 3.2.5.

Caravan rest areas

Larger sites that are the result of repeated occupation as rest areas are well documented in ethnoarchaeological accounts of camelid caravans (Lecoq 1988: 185-186;Nielsen 2000: 461-462, 500-504;West 1981: 70). According to these accounts, in the course of a multi-week caravan journey the team will rest periodically, usually for two or three nights, in order to allow for recovery by the animals and for conducting routine maintenance tasks by the caravan drivers such as repairing cargo panniers and stitching up the shoes made for the llamas' feet. Typically, these rest areas are situated adjacent to a rich grazing area with plentiful water and little competition for pasture access. The rest stops commonly are located immediately before or after a strenuous section of trail.

The Block 2 area is something of a natural bottleneck at the transition zone between the Colca valley and the puna, and it also contains the largest bofedal in the study area. The Block 2 puna is situated just outside of the rugged volcanic terrain that rings the Colca valley. It is the first extensive area of prime grazing encountered after caravans have crested the ascent out of the Colca, and the ascending animals would have presumably been loaded with agricultural cargo. It is therefore unsurprising that large corral facilities were encountered in this location.

The lithic and ceramic artifacts associated with these abandoned corral features are also telling. While local LIP and LH sherds are common in Block 2, the unslipped Chiquero-style pottery of Formative production, are also widespread in Block 2. Obsidian flakes are extremely common in the vicinity of these mounds. As demonstrated in the testing work at A03-39, small hearths and obsidian knapping at medium and advanced reduction stages occurred on the margins of these corral structures.

Obsidian knapping at rest areas

One plausible scenario for the quantity of obsidian in the Block 2 area would have caravan drivers resting for several days after the arduous climb from the Colca valley and knapping obsidian in the Block 2 area while their herds recover from the climb. Perhaps caravans would periodically venture into the Maymeja area of the Chivay source and recover larger nodules. Local herders also use the Block 2 puna and perhaps they were involved in obsidian procurement. Local herders, arguably more familiar with the obsidian source itself, could have brought obsidian to this area and passing caravans resting in the area may have traded for nodules and this would have contributed more obsidian to the Block 2 area.

The evidence from Block 2, however, is overwhelmingly of advanced stage obsidian reduction, and therefore if whole nodules were being exported from the Chivay source in whole form then they were not being knapped at Block 2. This evidence suggests that nodules were being reduced at the Maymeja workshop over in Block 1, or being exported from the region in whole form. Primary reduction and the use of large obsidian artifacts was not occurring in Block 2. As mentioned above, 10% of the flakes in Block 2 were over 4cm in length and neither cortical, nor non-cortical, flakes approach this size potential in Block 2. The absence of large flakes in this area, a zone rich in small obsidian flakes, is curious and it suggests that non-local caravans were responsible for the quarrying activity in Block 1. The larger (15-20 cm long), Ob1 nodules would have had to have been quarried, and some fraction of them knapped, in the Maymeja zone and then transported from the region without further processing in areas such as Block 2.

Was there multiethnic access at the Chivay source?

A principal question concerning the Chivay obsidian source area, when it was finally located in the 1990s, was whether cultural materials showing Titicaca Basin stylistic affiliation would be found at the source area. The lack of known Pukara or Tiwanaku materials in the vicinity of the source, as confirmed by the Upper Colca research project, suggests that access to the source was not physically restricted, and that perhaps neighboring groups were granted access as needed. Other plausible explanations are that, even in the Middle Horizon, material moved through down-the-line trade and no diagnostic, non-perishable materials were reciprocated back to the source area. Alternately, a freelance trading entity was responsible for transporting the material towards Titicaca, but this group traded exclusively with Titicaca Basin polities and curiously none of the material found its way into the Wari sphere despite the proximity of Wari-influenced sites in the lower valley. Given Tiwanaku's demonstrated strategy of placing colonies, sometimes integrated multiethnic, directly adjacent to resources of interest (Goldstein 1989), it seemed possible that a site with a Tiwanaku component would be found adjacent to the Chivay source.

The obsidian data do not appear to conform to the expected pattern of postulated multiethnic "artisan islands" within the vertical archipelago model proposed by John Murra (1972, pp. 442-443) in which rare or restricted resources such as metal ores, or here different obsidian types, are shared by different groups; in this case, each group (Huari and Tiahuanaco) would have had colonies to acquire the resource for the distant homeland (Burger, et al. 2000: 342).

A central goal of the 2003 research at the Chivay source was to examine the source area for evidence of such "artisan island" sites, but no such sites were encountered.

Circuits, caravans and exotic goods

The circuit mobility model of Nuñez and Dillehay (Dillehay and Nuñez 1988;1995 [1979]) depicts the regional integration between "axis settlements" as articulated by circuits of caravan traffic (Section 3.2.6). Some of these circuits grew to become the dominant exchange routes in a region. Despite the historical and adaptational focus of the circuit mobility model, the model's emphasis on the integrating role of camelid caravan networks linking dispersed communities in the altiplano highlights the importance of regional context in the emergence of political power in the transegalitarian milieu of the Titicaca Basin Formative.

The dynamic and decentralized Nunez and Dillehay model is compelling for understanding regional obsidian distributions in two ways. First, this model emphasizes the regular interaction that linked dispersed communities, often second-tier communities, across broad spaces. Obsidian appears to have had some social and symbolic significance, but it was only moderately rare and therefore it appears to have circulated relatively widely and continually between regional centers and also within second-tier and smaller communities. Thus the significance of obsidian is not principally as a "wealth item" like precious metals that served to differentiate elites in a type of network strategy (Blanton, et al. 1996). Rather, obsidian circulation, and the parties responsible for procuring and circulating it, depicts the subsistence level economic and cultural links upon which early aggrandizers likely constructed their political strategy.

During earlier time periods, such as during the Terminal Archaic when obsidian was first being intensified, it could be argued that obsidian was a rare "prestige technology" (Hayden 1998) that would confer advantages on the owner and would otherwise serve to differentiate people. Helms (1992: 159) describes how those conveying exotic materials were "long-distance travelers or contact agents as politico-religious specialists" in contact with the mysterious and distant (see discussion in Section 2.2.5). Such associations would arguably have been more likely during the earlier, emergent stages of regular caravan networks (e.g., the Terminal Archaic) under the assumption that sustained contact and diminished scarcity of a non-local material like obsidian would have probably reduced the social or symbolic power of such goods. Nevertheless, obsidian appears to have retained symbolically exotic associations that persisted in some form given the Late Prehispanic contexts: concentrations of obsidian in ritual mounds at Tiwanaku, and unmodified Chivay nodules at the gates of Machu Picchu (Section 3.5.3). However, given its abundance in the Early and Middle Formative sites, and the occurrence of obsidian in both commoner and elite contexts in the Late Formative, it would be difficult to argue that obsidian was status-conferring due simply to its non-local origin in that time period.

Model A: Direct acquisition

The Direct acquisition model describes personal visits to source areas by members of consumption households and it specifically excludes exchange of these goods between households. As per this model one should expect high variability in procurement methods, reduction strategies, and architectural and perhaps ceramics styles. Obsidian is found in both surface contexts and through quarrying, and presumably quarrying began to be practiced as the obsidian available from surface collection became more depleted. Thus, there is evidence in variability in procurement methods and in reduction strategies, however it is difficult to ascertain whether this variability results from procurement differences from non-local visitation, or from diachronic changes in available material.

It was previously suggested (Section 3.7.1) that during the Formative Period non-local visitors from the Lake Titicaca Basin, a known consumption zone for Chivay obsidian, may have left evidence of their presence during personal procurement. In the southern Titicaca Basin a distinct fiber-tempered early pottery tradition was used, but in both the north Titicaca Basin and the Colca area the contemporary pottery was grid tempered. One form of evidence of non-local personal procurement from the southern Titicaca Basin might take the form of sherds of fiber-tempered pottery from utilitarian vessels, although it may be difficult to differentiate from the evidence of fiber-tempered pottery the acquisition of obsidian through personal procurement from trade-item reciprocation. However, very little pottery, local or non-local, was identified in the Maymeja zone of the obsidian source and it is therefore difficult to use pottery to evaluate any of the procurement models. In the adjacent Block 2 survey area, obsidian scatters were often spatially coterminous with "Chiquero" pottery described by Wernke (2003) as a Formative style in the main Colca valley. As the Formative ceramics chronology in the Colca is still being defined, and there may have been differences in ceramic styles by altitudinal and ecological zone as well, it is difficult to definitively state whether all the pottery that appeared to belong to the Formative in the Chivay source area was indeed of local production.

Architecture at the Chivay source was also difficult to connect with the question of local versus non-local acquisition. Several circular structures approximately 2.5m in diameter were encountered in the area of the Maymeja workshop, but as the construction was eroded and appeared to have been expediently built, the association between these buildings and specific architectural traditions in the Colca valley or elsewhere is indeterminate. Wernke (2003) found that circular buildings were prevalent in the higher altitude portions of his survey. Further excavation in the workshop area may expose house floors or ceramics that indicate local or non-local cultural affiliation.

Direct acquisition certainly changed through prehistory from the earliest mobile foragers to the locals who procured material for household needs throughout the prehispanic past. Colca area residents probably procured obsidian for their needs from the Chivay source directly during visits to the high country that were embedded in hunting trips for viscacha, deer, and vicuña. Indeed, it was a local hunter from Chivay, Pedro Huaracha, that guided Sarah Brooks to the obsidian source in 1994 (Brooks 1998: 433-438). When non-local residents, such as residents of the Titicaca Basin, visited the Chivay source as they followed the Escalera-Lecceta route, it was perhaps for personal procurement but more likely it was in the context of caravan-mode circulation, either independent or administered, as described below.

Model B: Multiple reciprocal exchanges (down the line)

According to this exchange model, local upper Colca communities procured a sufficient excess of obsidian from the Chivay source to supply material into the exchange system for regional circulation. Two principal material expectations were described for this mode: first, the Chivay source area would contain local artifact styles and relatively low variability in procurement; second, that assemblages from the local Upper Colca communities would have some fraction of large obsidian flakes and perhaps cores consistent with the large nodules found in the Maymeja area.

As mentioned, ceramics and architectural evidence are in short supply at the Chivay source itself. There is spatial consistency in the procurement of obsidian, as the quarry pit and the workshop appear to have been repeatedly occupied over the years, but evidence from the both stratified test units show that quarrying and production was episodic. The earliest and the latest (upper-most) levels of the workshop test unit showed intermediate reduction events and some advanced reduction that may have been the result of irregular local visits to the source area.

One could argue that the intensified and regular production observed in the middle levels of the quarry pit test unit and the workshop test unit were the result of locals mining their adjacent obsidian source to meet the demands of the non-local reciprocal exchange network. However, two pieces of evidence contradict this proposal. First, as was described in Section 7.6, if locals were quarrying and producing quantities of obsidian at Maymeja one would expect some fraction of the large nodules to be represented in assemblages from local communities either in Block 2 or Block 3. In fact, very few flakes, cortical or non-cortical, over 4 cm in length were found in Blocks 2 and 3, while over 10% of flakes from the Block 1 Maymeja workshop were in that category. Second, the spatial evidence of intensification that was observed in production in the quarry and workshop was not reflected in evidence of intensified or concentrated obsidian deposits in the Block 2 and 3 areas. In Block 2, concentrations of obsidian flakes were found associated with corrals, but these were consistently small flakes. The question is if locals were responsible for intensified production at the Maymeja quarry, then which locals were conducting this work and why did they not use any of the large nodules in their local economy? The evidence suggests that locals were not involved in intensified obsidian quarrying and production in the Maymeja, but this does not preclude local residents guiding foreign caravans to the source area, or low intensity local procurement activity for local consumption and down-the-line exchange. Indeed, the concentrated obsidian deposits at Taukamayo in Block 3, although consisting of Ob2 material and relatively small flakes, may have been the result of local communities articulating with non-local caravans near the site of modern-day Callalli.

On the regional scale, down-the-line exchange is probably implicated in larger distributions of obsidian, particularly during the Archaic Period prior to the advent of caravan based mobility. A combination of direct, personal procurement and down-the-line exchange may have been the source of obsidian at sites like Asana and Qillqatani during the preceramic. There is evidence to suggest that during the Late Archaic these down-the-line networks may have become more segmentary and isolated. It is during the Late Archaic that Asana no longer contained Chivay obsidian, and projectile point styles during the Late Archaic became increasingly localized. Subsequently, in a dramatic change, during the Terminal Archaic and Early Formative regional exchange became far-reaching, the series 5 projectile point style appears with relatively little variation throughout the central and south-central Andes, and the knowledge of food production and other technologies became widespread. The evidence presented here suggests that exchange patterns, including obsidian procurement through down-the-line and direct acquisition, were relatively reduced during the Late Archaic. During the ensuing Terminal Archaic, with the initiation of regional caravans, major changes include an expansion of regular exchange and temporal consistency in obsidian supply at places like Qillqatani maintained through long-distance relationships as is described by Model C.

Model C: Independent caravans

The Independent Caravans model entails household level organization of long-distance caravans for the acquisition of specific goods from another region, or for moving goods between a number of communities in a loop and bartering for goods at various locations along the route. The material expectations of this type of obsidian circulation in the Chivay source area include episodic but intense procurement while a caravan was stopped at the obsidian source, and prompt transport of obsidian southward from the Maymeja source area to a principal thoroughfare that connects the Colca with regional travel routes.

The 2003 Upper Colca project results support an interpretation of obsidian production at the Maymeja quarry and the workshop through procurement and circulation by independent llama caravanners, with this mode occurring from the Terminal Archaic through at least the Early Formative. The 2003 results support this interpretation at a number of scales.

(1) Dates.The¹⁴C samples collected from the workshop test unit show that intensified production occurred between 2880 and 1260 cal BCE. Based on the evidence of intensity of production, levels 2 through 4 at the quarry pit test unit [Q02-2u2] are tentatively linked with levels 4 and 5 at the workshop [Q02-2u3].

(2) Camelid pastoralist emphasis.Why was obsidian from the quarry pit not reduced at the quarry pit? Instead it appears to have been transported down slope 600m to a site adjacent to water and to a lush grazing area. There are a number of links between pastoralism and obsidian production at the source area that have been explored in detail above.

(3) Transport.It does not appear that locals participating in down-the-line exchange networks were responsible for quarrying and then processing obsidian at the Chivay source workshop because obsidian flakes and cores in the local consumption sites were much smaller than one would expect if local pastoralists were quarrying for large nodules.

(4) Regional consumption.The temporality of consumption in distant sites like Qillqatani suggests that regular, caravan-based exchange was the mode of transport between the source and the consumption zone. It seems more like that caravan traffic, rather than Direct Procurement and Down-the-line exchange, would have resulted in Chivay obsidian consistently representing 10% to 20% of the lithics assemblage at Qillqatani for a period of several thousand years.

That is not to imply that procurement and production for other modes of transport, including personal acquisition and down-the-line exchange, did not occur in Maymeja; only that the intensified production associated with quarrying work and concentrated reduction activities appear to have been linked to caravan-based export. While the pastoral link and the intensified production seem irrefutable, the challenge remains to differentiate independent from administered caravans.

Model D: Elite-sponsored caravans

Caravans organized and administered by elites are described in colonial documents from the Lake Titicaca Basin. According to these passages, Titicaca Basin herders could pay part of their annual labor obligations by contributing their labor and their caravan llamas to elite-organized long-distance caravan ventures that procured non-local goods that were, in turn, redistributed to the Titicaca Basin community in the form of feasts. Material expectations at the Chivay source for obsidian procurement and circulation under elite-sponsored caravans were discussed in Section 3.7.4. It could be relatively difficult to differentiate independent from elite-sponsored caravans from archaeological evidence because the actual caravanners in either case are commoners from altiplano herding communities.

Evidence of elite-organized obsidian procurement was not found, nor were walls or other defensive structures encountered that would have been protecting the obsidian source for some kind of access monopoly. Social mechanisms for monitoring and control of the source area are a possibility in prehistory but evidence of the use of distinctive artifact styles as a signal of "ownership" of the source area were not encountered in 2003. This is consistent with other obsidian sources in the Andes, such as Alca and Quispisisa (Burger, et al. 1998;Burger and Glascock 2002;Jennings and Glascock 2002), where quarrying remains and workshops rarely contain discrete evidence of hierarchy at the source, or even cultural affiliation from specific groups known to have used those sources, such as the Wari. In North America, in an examination of Mississippian chiefdom hoe production and consumption, Cobb (2000: 195) found no evidence of hierarchy at the Mill Creek quarries over the course of four centuries of quarrying. Cobb compares this with ethnographically documented quarrying in New Guinea for ceremonial axe blades, and similarly at this quarry no evidence was found of hierarchy in production in a tribal society, despite the fact that the blades were often finished at the quarry site (Burton 1984).

The geography of the Chivay source area presents natural defensive features, such as restricted entrances to the Maymeja area that are only available along a half-dozen routes that are not cliff faces. None of the access routes showed evidence of defensive fortification or restriction. In fact, these routes were generally improved with a roadway on one exit [A03-268] and a short portion of stairway on another egress. The consumption patterns of obsidian from the Late Formative until the Inka period do not suggest that evidence of elite-sponsored quarrying would be encountered (Section 3.5.3) Assemblages at Tiwanaku suggest to Giesso (2003) that the state controlled the acquisition of obsidian from Chivay (Cotallalli) and basalt from Querimita, however there are a number of articulation points between the Chivay source and the Tiwanaku core 300 km to the south-east where the state could exercise control of the obsidian entering the core region. Recent perspectives on the probable ethnic diversity subsumed by the corporate strategies of the Tiwanaku state (Goldstein 2005;Janusek 2004;Stanish 2002) highlight the variability that might be expected in the nature of relationships between Tiwanaku elites and communities in its periphery. If the internal ethnic constituency of the Tiwanaku state, and the polities that preceded Tiwanaku in the Titicaca Basin, were more diverse than previously thought, then it is possible that many of the corporate themes were a reflection of the need to centralize the political economy because the population was relatively heterogeneous.

In the Chivay source area, the only distinctive evidence of non-local, state-affiliated materials were sherds and architecture in styles belonging to the imperial Inka. These sherds and structures are almost certainly mortuary facilities, although it is also possible that they are shrines associated with water emerging from the spring in Maymeja. Given the reduced importance of obsidian in the Inka sphere, it seems unlikely that the Inka presence in the Maymeja area was related to administered obsidian production.

Discussion

The above obsidian procurement and circulation models are not mutually exclusive, and the implications can be considered synthetically in light of previously discussed exchange models in the Andes. The evidence just presented suggests that while obsidian procurement and initial production at the Chivay source was predominantly direct procurement, and transport via direct and down-the-line models of circulation, there was an important period of intensified procurement of obsidian that occurred during the Terminal Archaic through the Early and Middle Formative Periods. Obsidian is available on the surface surrounding Cerro Hornillo and the many small scatters on the ridges and in shelters in this area suggest that informal production occurred at various times in prehistory. These scattered reduction sites were complemented by evidence of intensive production that occurred in the southern portion of Maymeja where it appears that the principal goal was acquiring larger nodules of obsidian.

Models that attribute the origins of long-distance trade caravans to administered trade based on elite strategies (Stanish 2003: 69) have similarities to the models that limit the inducement to long-distance exchange to elite competition (Berdan 1989: 99;Brumfiel and Earle 1987). Following this perspective, in a context of down-the-line exchange it was elite demands and finance for procuring non-local goods that prompted the development of regular, long-distance caravan exchange, and then trade expanded to accommodate the needs of the wider population. As discussed by Smith (1999: 113), these perspectives place multifaceted processes like exchange into dichotomous terms. Differentiating elite from commoner demands, and defining prestige and utilitarian goods across the wide prehistoric variability in time and space is especially difficult. In the Andes, if "virtually all trade in the Andes conforms to what Polyani referred to as 'administered trade'" (Stanish 2003: 69) then how does one explain the consistently high percentages of non-local lithics due to long distance trade, as documented in primarily Formative levels at Qillqatani? It is argued here that these percentages are in excess, in both quantity and regularity, of the sporadic incidence of obsidian one might expect from down-the-line exchange over 200 km of largely homogeneous puna. If this regular consumption of non-local goods is indeed the product of independent caravans that began in the Terminal Archaic then it demonstrates an upswing in long-distance interaction that was founded on camelid domestication but it predates evidence of social ranking.