A publication of the Archaeological Institute of America
The Monte Verde site in southern Chile was acclaimed in 1997 as the long-sought proof of a human presence in the Americas centuries before the rapid expansion of the Clovis culture around 11,500-11,000 rcbp (radiocarbon years before present) or 13,500-13,000 cal BP (before present in calendar years). It is thus a "paradigm-buster." By breaching the 11,500 rcbp (13,500 cal BP) Clovis temporal baseline, it implies "a fundamentally different history of human colonization of the New World" (Meltzer et al. 1997: 662). Acceptance of the antiquity of Monte Verde ensures that other sites, where much earlier pre-Clovis occupations have been claimed, will be less skeptically regarded.
Tom Dillehay excavated at Monte Verde from 1977 to 1985. In a series of articles that have appeared since 1982, he described two occupations. On the north bank of Chinchihuapi Creek was a complicated mass of logs, mastodon bones, plant remains, and round and angular stones, all sealed and miraculously well-preserved under a layer of peat. Radiocarbon dates put this MV-II occupation at about 12,700-12,000 rcbp (about 14,500-14,000 BP in calendar years). On the south bank, a more deeply stratified site, MV-I, comprising 26 stones (at least one, a basalt core, an unquestionable human-made artifact) and three possible hearths, yielded dates of more than 33,000 rcbp.
The first volume of the final site report, describing the environmental context, appeared in 1989. In 1997, the massive (1,071 pages) second volume of the final report, dealing with the archaeological material, was published by the Smithsonian Institution Press. Coincident with this publication, a delegation of Paleoindian specialists, under the auspices of the National Geographic Society and the Dallas Museum of Natural History, viewed the artifact collections in Kentucky and Chile, and visited what remained of the site (most of it had been destroyed by a bulldozer in 1988). These experts found the stone tools, "especially the projectile points and grooved spheres," from the MV-II site "compelling" and concluded "...we have no doubt that there are genuine lithic artifacts on the MV-II surface" (Meltzer et al. 1997: 661). They, like Dillehay himself, withheld final judgment on the older MV-I assemblage, but opined that "The chances seem good that these materials indicate a significantly early human occupation in the region" (Meltzer et al. 1997: 662). While affirming the cultural origin of the MV-II material and its Pleistocene age, the panel urged readers "to examine for themselves Dillehay's detailed volumes on the site and draw their own conclusions" (Meltzer et al. 1997: 660).
As the panel of experts suggested, I have closely examined Dillehay's 1989 and 1997 reports and previous publications. My review has raised troubling doubts about the provenience of virtually every "compelling," unambiguous artifact, as well as many of the more equivocally human-modified items and the radiocarbon-dated specimens. I found that the basic mapping and recording of the site were seriously flawed. The findspots of the handful of "compelling" MV-II artifacts--a quartzite bifacial preform (B-10-5), a nearly complete, willow-leaf-shaped point made of basalt (X-15-1 or D-10-1-1), point midsections A-1-26 and D-S-1, a ground slate perforator (A1300100), a rhyolite polyhedral core (A-1-4-1), and a quartz chopper (DW-12-4-10)--are inadequately documented. No photographs of any of these artifacts in situ have been published, nor are they depicted on any of the site plans. The final volume and earlier reports contain ambiguous and often contradictory provenience descriptions. Early reports contained allusions to three stone scrapers, still attached to their wooden hafts. A photograph of only one of these has ever appeared, and its precise findspot is not reported. Similar ambiguity pervades descriptions of a large grinding stone, four unretouched stone flakes said to have been found still lodged in the grooves of the wood slats that they had been used to gouge, three wooden mortars, a tent stake with knotted reed cordage attached, and eight of the radiocarbon samples. The most obvious artifact from MV-I, a basalt core (X-1E-1 or PZ-43-3), seems to have been ascribed at first to one of the MV-II excavation areas (Zone A) (Dillehay 1984), only later being located in one of the poorly mapped MV-I (Zone C) units, on the opposite side of the creek.
At first glance, my inquiry into artifact provenience at Monte Verde may appear to be a tedious and inconsequential exercise in hair-splitting. What difference does it make if a point fragment was found in Zone A or Zone D, in 1981 or in 1985? To state the obvious, accurate and precise description and illustration are minimal expectations for any site report. Archaeological excavation is the controlled destruction of a site. Rarely is this process closely monitored by anyone except the director and his or her assistants. A few guest experts may pay brief visits, but we can't all visit every site, and we can't bear witness to each and every significant discovery. After 80 or 90% of the site has been excavated, as at Monte Verde, the stratigraphy and the patterned distributions and associations of artifacts and features can only be reconstructed, in the abstract, by reference to a detailed, well-illustrated, and internally consistent report.
Have I subjected the Monte Verde final report and earlier publications to unusually intensive scrutiny? Indeed. For this is no ordinary site report, and it deserves more than a cursory inspection. The hefty 1997 volume has been heralded by pre-Clovis enthusiasts as the stone that will finally sink the "Clovis-first" model of the peopling of the Americas. As such, it has profound implications for archaeology, genetics, physical anthropology, and linguistics. Beyond the Americas, it has potential global significance for archaeological methodology and evolutionary theory. As the lithic analyst, Michael Collins, has observed (1997: 468), archaeology lacks "an adequate paradigm for understanding lithic artifacts and manuports modified little or none from their natural state." If, as he has concluded, 95% of the lithic assemblage of a site occupied by a band of Homo sapiens about 12,500 rcbp consists of such imperceptibly modified stones, does that mean that archaeologists have failed to recognize most of the extant record of more ancient human activity? At Monte Verde, unshaped pick-up tools--naturally rounded or cracked rocks from the nearby stream bed-- are argued to have served just as well as the very few carefully-chipped bifaces. Was this expedient South American industry a viable alternative to the lithic technology of the Old World, where hominids, even as early as 2.5 million years ago, found it necessary to shape stone tools by deliberate and skillful flaking?
I paid particular attention to the association of artifacts and wood and bone objects, because the excavator asserted that the context of the Monte Verde assemblage was as good as that of the Folsom points found embedded in Bison antiquus skeletons in 1927, and more conclusive proof of human agency than the merely geological co-occurrence of Clovis points with mammoth bones (Dillehay 1997: 18). At Monte Verde, there was functionally articulated association: "articulated objects, conjoined or attached, in a functionally integrated fit, like a hafted lithic tool, cordage wrapped around two posts, an in situ lithic tool in worked wood, or the presence of exotic organic residues on the edges of demonstrable stone artifacts." However, in examining the report closely, one finds that two of the three hafted lithic tools mentioned in early reports are neither described nor illustrated; the cordage was wrapped around stakes that have been attributed to clearly erroneous proveniences; at least one of the "in situ" tools in "worked wood" was positioned there by the excavator for a photograph (Dillehay 1997: 149); and the plant residues on lithic objects, ascribed to human use, cannot be definitively distinguished from those that represent only natural accretion in a plant-rich matrix (1997: 507). Because poor mapping and inadequate provenience data leave us uncertain of the actual relative positions of any materials recovered from the site, and Dillehay refers in some cases to objects that were found lying more than 30 meters apart as somehow "associated," the behavioral and functional inferences drawn from these associations are also questionable.
Although its 1,071 pages ostensibly represent an exhaustive analysis of the Monte Verde site, the final report fails to provide even the most basic provenience data about key artifacts. Those few clearly human-made objects (the point fragments, the bolas, the preform, the cores, and the perforator) are literally "compelling" because they force us to consider the likely human production or transport of the many other, more ambiguous lithic, wood, bone, and vegetal items with which they appear to be intermingled. However, alternative explanations of the mastodon bones, the piles of wood, and the peculiar suite of seeds and leaves can be advanced, which do not entail human agency, and to their credit, Dillehay and his colleagues have considered and even tested some of these possibilities, such as animal activity, or transport by glaciers, streams, or wind (e.g., Dillehay and Rossen 1997: 355). Their assessment was grounded on a uniformitarian assumption that the potential depositional and erosional processes in the region at 12,500 rcbp were the same ones that they observed during their research along Chinchihuapi Creek. Contemporary natural processes cannot account for the piles of mixed bone, wood, plants, and stones at Monte Verde. However, at the end of the Pleistocene, nearby glaciers were melting, sea level was changing, and huge proboscideans with poorly known dietary and behavioral patterns roamed this area. How could present-day studies of a sluggish stream in a stable landscape accommodate these now unobservable factors?
According to Adovasio and Pedler (1997: 576), the inspection team that visited Monte Verde evaluated the evidence with respect to three basic criteria:
(1) artifacts of indisputable human manufacture recovered in primary depositional contexts;
(2) clearly defined, unambiguous stratigraphy accompanied by a precise knowledge of emplacement mechanisms, site context and associations of recovered artifacts and ecofacts;
(3) multiple radiometric determinations showing indisputable internal consistency.
Discounting the earliest (13,565±250 rcbp [TX-3208]) and latest (11,790±200 rcbp [TX-5374] and 11,920±120 rcbp [TX-5376]) radiocarbon dates leaves a range of ca. 12,700-12,200 rcbp for the wood pieces at the site. These dates are reasonably consistent, given the normal error ranges of late Pleistocene dates. So the site appears to meet criterion 3. There are unquestionably human-made artifacts at the site, and they are assumed to be in primary context, so criterion 1 is met. However, so numerous are the inconsistencies and ambiguities in the reporting of crucial artifacts that "precise knowledge of site context and associations of recovered artifacts and ecofacts" has not as yet been demonstrated, so the site fails on criterion 2. Perhaps Dillehay and his colleagues can allay these concerns by re-examining their field notes to clarify the discrepancies in the final report. Until they do, Monte Verde should not be construed as conclusive proof of a pre-Clovis human occupation in South America.
See Fiedel (1999) for a longer critique of Monte Verde.
Back to Introduction