Stay informed: Sign up for eNews Subscribe

Across Atlantic Ice The Origin of America's Clovis Culture

Read Chapter 2



The First American Settlers?

For decades the Clovis culture has been our anchor to understanding the peopling of the New World. It has been a touchstone for archaeologists, a rare "truth" and comfort regarding the unknowable archaeological past. They have recognized who was first in the Americas and whence they came. Even so, the interest in just how they came to be and where they came from has driven Clovis research, which has resulted in questions challenging even our very basic beliefs. What is the evidence that has started to shake the foundations of this long-accepted theory? Could it be that there were people in the Americas before Clovis?

The amount of evidence for Clovis has grown at a relatively fast rate since the first discovery of this archaeological culture. Clovis points have been found throughout most of continental North America, extending from the Atlantic to the Pacific, and from the plains of Alberta, Canada, into northern South America (figure 2.1). The wealth of data provided by the excavation and analysis of Clovis sites is giving us a new understanding of Clovis culture. Although Clovis hunters killed now extinct ice age animals, a conservative view of radiocarbon dates indicates that these people first appeared in the New World some 13,000 years ago, and in less than 200 years they had explored, exploited, and inhabited the two continents of the Americas. By comparison, the historic European residence in the Americas began barely 500 years ago. If the actual tenure of Clovis is only 200 years, they achieved the most rapid terrestrial expansion and extensive colonization in the history of pre-literate people.

The Clovis expansion across the Americas coincided with a period of dramatic climatic fluctuations that altered landscapes and their associated resources. Clovis people must have been highly adaptive generalists who could take advantage of a wide variety of resources. Their secular worldview must have centered on the edible and the inedible and how to attain the former. At the end of Clovis times we see users of descendant technologies adapted to every available ecological zone: tundra in the north; boreal forest with deciduous and coniferous woodlands in the northeast; deciduous forest and parklands; mixed hardwood and piney woods in the South; pluvial lakes and rivers in the relatively arid plains; desert in the southwest; and the tropics of Central America. Moreover, dredging along the Atlantic coast and sports divers on the Atlantic continental shelf have recovered Clovis artifacts from an ocean margin environment. Although Clovis sites and artifacts are rare on the West Coast and in the Great Basin and range country, scattered evidence around pluvial lakes and river courses suggests at least minimal Clovis occupation of these areas.

Fluted projectile points are the most readily identifiable Clovis artifacts, but there are many other items that make up the Clovis inventory. Among these are tools made from blades and flakes struck from specialized cores, plus bifacially flaked knives and adzes. Bi-beveled bone, antler, and ivory rods of unknown use, foreshafts (made to hold projectile points for use with a spear shaft), and projectile points as well as barbed harpoons are also characteristic of Clovis.

Settlement Patterns

The history of Clovis archaeology has had a profound influence on our perception of Clovis culture. The first recognized sites, Blackwater Draw and Dent, were found in semiarid environments; both included mammoth remains. Many other Clovis localities were subsequently discovered in the West. But even though archaeologists have become aware that the climate and landscape of North America were much different at the end of the Pleistocene than they are today, the popular view of Clovis continues to emphasize mammoth hunting in an arid to semiarid region. Yet Clovis artifacts and sites with Clovis-like fluted points have now been found from coast to coast and from Canada to as far south as Venezuela.

In North America, Clovis artifacts are relatively rare in the northeastern and north-central states, and some scholars believe this is because these areas were then recently deglaciated and not settled until late in Clovis times. Clovis sites are most abundant in the Mid-Atlantic and southeastern states and exhibit a wide diversity of activities, whereas diversity decreases to the north and west, where activities are focused on hunting. Western sites are either small campsites or kills that are sometimes associated with small temporary hunting camps. To the east and southeast, by contrast, they include not only kills but also stone processing areas, usually associated with large campsites; large overlook and upland surface sites; and at least one cave/rockshelter. Some of these differences in settlement patterns and site types may be the result of preservation or just plain luck in finding sites, but consistent patterns are emerging. The archaeological evidence suggests that the contrast between the east and the west can be explained by the late ice age prairie-forest border. The abundance of resources in the woodlands allowed at least semi-permanent settlements, while the prairies, plains, and deserts offered large game but little else at the time. Because plant and animal habitats were responding to abrupt and fluctuating climatic changes at the end of the ice age, this margin cannot be identified with great precision, but the general distribution of different Clovis site types gives a general indication of the location of this change.

Evidence of Clovis dwellings is sparse. Based on lack of evidence to the contrary, it is presumed that Clovis people occupying western sites had highly portable or ephemeral structures, but even hearth features are almost nonexistent. In the east, where larger campsites have been found, one would expect evidence of more substantial housing because of the length of time people could remain in this area. By and large, however, this is not the case, save for the appearance of a pattern of post molds at the Thunderbird Site in Virginia from which Errett Callahan has reconstructed the shadowy outline of a rectangular longhouse. If the evidence from Thunderbird has been interpreted correctly, the people who lived there may have constructed longhouses at sites where they intended to reside for extended periods or those to which they returned on a regular basis.

A discovery at the Gault Site in central Texas provides another example of a possible Clovis dwelling. This is a square cobble pavement with straight sides and distinct corners that measures almost 2 square meters. The sides of the pavement line up with the cardinal points (figure 2.2). As the Texas archaeologist Mike Collins has pointed out to us, the odds against this happening randomly are 90 to 1 (based on a 360 degree circle). The structure is in what would have been a wet valley bottom in fine clay deposits. Surrounding this pavement and resting at the same elevation is a scatter of blades, biface fragments, limestone clasts (possibly artifacts), and bone fragments, leaving no doubt about the pavement's association with an occupation. Collins's recent re-evaluation has questioned the Clovis association based on the lack of diagnostic Clovis artifacts as well as a series of dates indicating that this feature and its associated artifacts may be older than Clovis. Investigations into this issue continue.

There is, of course, a lot of speculation about how the feature was used. A likely possibility is that it was the floor of a small structure, constructed to elevate the occupants above the damp valley soil. Blade segments with evidence of grass and reed cutting were associated with the feature, and Collins imagines that the pavement was laid down and then covered with grass or mats, which in turn may have been covered with hides, making a comfortable platform on which to sleep. It is also possible that reed mats were used as walls to enclose the structure.

Another Clovis cobble pavement was discovered in 1948 in Texas at the Kincaid Rockshelter. This pavement covers a significant portion of the Clovis occupation level (there are more uses of the site in archaeological layers above Clovis), and artifacts and bones from food refuse were found in the cracks between the cobbles. A water seep presently keeps the interior of the shelter wet, and if the soil was damp during the Clovis occupation the cobbles may have been used to provide a dry surface. Why Clovis people selected this wet cave among the many dry caves throughout Texas remains a mystery. Nevertheless, these two sites establish pavement as a method known and used by Clovis people.

In order to understand the intensity and use of the Gault Site, intensive refitting analysis has been done there. Flaked stone artifacts are checked to see if knapping sequences can be reconstructed by fitting flakes into flake scars on flakes, cores, and tools. If pieces found near each other fit together, it is likely that they were made and used in that location, and if people used the site only temporarily, especially as a manufacturing locality, one would expect a lot of the remains to be clustered in refittable clusters. On the other hand, if there is a lack of fittable pieces it means there was substantial mixing of artifacts caused by long-term camping activities. The Gault study identified few refits. Since there is little indication in many of the deposits of natural processes moving the artifacts, the paucity of refits probably resulted from human activities that moved individual pieces away from where they were made. Moreover, tools of local chert that were used, worn, refurbished, worn out, and finally discarded are common in the collection and indicate that the Gault occupations were intense and long-term enough to result in this thorough mixing and reuse of artifacts.

Another example of a large Clovis site is the Carson-Conn-Short site on the ice age banks of the Tennessee River, which overflow waters from Kentucky Lake currently seasonally inundate. Although this site has been only preliminarily tested, it appears to be a major Clovis camping area associated with a chert source. A meter or more of midden (cultural refuse) is spread continuously along the terrace for up to a kilometer. The occupation debris probably accumulated from both long-duration encampments and repeated shorter visits during quarrying activities. Although the brief tests have recognized no evidence of dwellings, the presence of storage pits and possible heat-treating features indicate that this site has great potential for producing evidence of shelters if and when more detailed excavations are conducted.

Evidence of Settlement Direction

Almost fifty years ago Ron Mason observed that there were significantly more fluted points reported from the southeast than from the rest of the continent and that there was more variation in fluted point forms and technology in the east than in the west. Following the age-area hypothesis that the greatest number and variations of traits will occur near their center of origin and decrease as they defuse outward, he suggested the southeastern United States and not the West as the homeland of the Clovis complex.

Mason's estimates of the number of fluted points were based on impressions, not on empirical data. But as the years passed, surveys reporting eastern fluted point finds and sites increased at a rapid rate, while western finds remained relatively rare. Twenty years later Louis Brennan, the editor of Archaeology of Eastern North America (AENA), in an effort to revive Mason's hypothesis, organized and published a compilation of eastern fluted points. The AENA compilation covered the cis-Appalachian East from Quebec to Florida and recorded more than 5,800 points. Although the leading experts on Paleo-Indian archaeology were impressed by the data, they all agreed that the abundance of specimens alone was not proof of the antiquity or ultimate origin of Clovis. Even if the compilation demonstrated that there were more Clovis points recorded from the East, there were likely appropriate explanations-for example, the carrying capacity of the eastern habitats may have been more favorable to population growth, or there may simply have been more artifact collectors and archaeologists working in the East, where decades of agricultural activities likely plowed up more artifacts. Conversely, if there were fewer Clovis points recorded from the West, it was likely that greater erosion there had destroyed many Clovis sites and re-deposited fluted points. A study by the archaeologist Brad Lepper found a positive correlation between the numbers of fluted points reported from Ohio and the amount of acreage under cultivation and associated with contemporary population centers.

Can we evaluate whether or not a greater density of eastern Clovis points reflects better artifact visibility due to erosion and cultivation, plus more intense scrutiny and focused inventories? Certainly there are places where surface exposure is limited in the West, but this is not characteristic of most of the area. Virtually all of the grassland prairies have fallen to the plow, and surface erosion has been so extreme that a large percentage of the topsoil has been stripped off. We only have to look into the history of the U.S. Soil Conservation Service (now the Natural Resources Conservation Service) to see how widespread this has been and continues to be. Erosion has often reached below late Pleistocene levels, exposing massive tracts of Clovis and earlier deposits, leaving artifacts exposed along with gravels. This process was especially intense during the dust bowl days of the 1930s, and to a lesser degree during the 1950s. These severe droughts exposed an enormous area of what was once prime large-game habitat, especially for now-extinct Pleistocene species.

Some of the most fruitful artifact collecting episodes in North American archaeology took place on the Plains and in the Southwest during the dust bowl. Large private collections of Paleo-Indian (and later) points were assembled from the extensive exposures of late Pleistocene deposits in both regions. Although many Folsom (the fluted point culture that followed Clovis in the High Plains) and later Paleo-Indian sites were documented, Clovis artifacts were rarely found, even though numerous mammoth remains were reported.

Many western collections have been inventoried, and there are regional compilations similar to those in the east. For example, John Cotter and students from Denver University working under the direction of E.B. Renaud traveled throughout eastern Colorado, western Kansas, southwestern Nebraska, and northeastern New Mexico during the 1930s to record Paleo-Indian artifacts found by local collectors. Even though the study was conducted in an area where there was excellent exposure and visibility, as well as extremely active artifact collecting, Clovis points were poorly represented in the collections and accounted for less than 0.04 percent of the total Paleo-Indian point count. It seems reasonable to us to compare these Clovis artifact density results to those in the east, and this pattern has continued to the present, with a large difference in the number of Clovis points and sites found in the western and eastern states.

There is also a far greater diversity of Clovis descendant point forms and technologies in the east. These variations likely reflect local adaptations to varied ecological settings, but links among form, technology, use, and environment have not been established. In the east there are styles such as Debert (figure 2.3a), Cumberland (figure 2.3b), Redstone (figure 2.3c), Barnes (figure 2.3d), and Crowfield (figure 2.3e), to name just a few. Only a single variant, Folsom, is present throughout the western Plains (figure 1.1a), and there is a fluted variety in Alaska (figure 2.3f). Folsom likely developed as part of a bison-hunting specialization, and this point type expands along with the expansion of the grasslands that became the bison range. Consequently, we contend that the greater site density and diversity in point styles in the east support the conclusion that the Clovis tradition has a greater time depth there.

Although most fluted points have been found out of context and on modern ground surfaces in the east, there are major Clovis deposits buried in early terraces of eastern rivers. The densest accumulations of Clovis artifacts have been encountered at Shawnee Minisink on the Susquehanna River in Pennsylvania, Thunderbird on the Shenandoah River in Virginia, Carson-Conn-Short on the Tennessee River in Tennessee, Williamson in Virginia, Adams in Kentucky, Sinclair Site in Tennessee, and the Pine Tree and Topper Sites on the Savannah River in South Carolina. The Gault Site in central Texas, possibly the largest accumulation of Clovis artifacts yet identified, is situated on a small stream associated with springs. Many of these dense surface accumulations of Clovis artifacts are near or even at chert sources and have artifact assemblages that indicate an extensive range of domestic activities beyond simply obtaining tool stone. No comparable sites are known in the west.

The few western Clovis campsites, such as Murray Springs in Arizona and Sheaman in Wyoming, are much smaller than the eastern sites, and most are at or near kill sites with associated processing areas, such as Lehner and Murray Springs in Arizona and Lange-Ferguson in South Dakota. The Colby site, in Wyoming, has been interpreted as a meat cache. Even though there are excellent quarry locations with high-quality tool stone used by Clovis flintknappers, associated campsites have not been found. We suggest that the large eastern sites represent a complete cross section of Clovis population that camped on or near stone sources for extended times or for shorter terms but revisited the sites many times, whereas the paucity of evidence found at the western quarry locations, kill/processing sites, and campsites suggests that these localities were for short-term use or specific activities performed by a small subset of the Clovis population.

Large blades are rare in western Clovis assemblages but extremely common at many southeastern sites. The scarcity of blade technology in the west is not related to the availability of high-quality stone. Excellent sources of flaking material (e.g., Alibates dolomite from the panhandle of Texas, quartzite from the Spanish Diggings in Wyoming, and jasper from southeastern Wyoming) are available throughout the west and were used by Clovis people for projectile points but seldom for blades. The differential distribution of large blade technology may be related to chronology; that is, blades are also absent from late Clovis assemblages such as Thunderbird in the east. Or they may have been used in specific tasks that might have been gender related. Use-wear analyses of blade tools suggest that they were used for hide working and for the construction of matting, both frequently female tasks in historically documented foraging societies. The dearth of blades may signify that hide procurement and processing and the establishment of more durable living structures were not part of western activities. Along with the minimal occupation debris left at campsites, the lack of blades may point to small, mobile exploration parties rather than larger bands. However, it can be argued that blades occur in the same proportions to projectile points in the west as in the east, since Clovis artifacts are relatively rare in the west. Unfortunately, it is difficult to evaluate the circumstances of the occurrence of blades as few reports include relevant chronological or use-wear data.

Even when the various indications of greater Clovis population in the east are accepted, they are usually still interpreted as evidence of a west-to-east settlement pattern. For example, David Anderson and Jack Gillam produced geographic distribution maps based on a systematic database compiled from fluted point reports in official government files. When plotted at the county level across the continental United States, the density of fluted points is clearly greater in the east (figure 2.4). On the basis of this information Anderson and Gillam proposed several colonization models-all based on the assumption that Clovis arrived in the west and expanded to the east. Accordingly, they explain the high density of eastern points by supposing that the first migrants were few in number and that their population grew as their territories expanded southeastward along major rivers such as the Missouri and Mississippi to eventually end up on the eastern seaboard. The eastern environment was much more productive than the western, resulting in a large human population concentrated in several geographic centers.

But there are lines of evidence that seem to effectively challenge the eastward-expansion-of-Clovis hypothesis. One argument involves the nature of caching behavior. Caches are places where large numbers of artifacts or small groups of projectile points were buried. Clovis people left three varieties of caches: groups of blades, groups of projectile points, and mixes of projectile points, biface preforms, and tools. Currently, the best-known Clovis caches are west of the Mississippi River, but a growing number are being identified farther east. The caches may have been made for various purposes, ranging from logistical caching of tools for later retrieval to ritual offerings never intended to be reclaimed. In a thought-provoking paper, the Southern Methodist University archaeologist David Meltzer hypothesized that when Clovis people colonized a new, uninhabited landscape, they anticipated the possibility of difficulty in finding raw materials for tool replacement. To prepare for such an eventuality, they created insurance caches of artifacts across the landscape as they moved away from known resources. Thus, caches served as resupply depots; if explorers didn't find stone sources in the new territory, they could resupply from a stash strategically near the margins of known territory. This strategy would be essential for exploring and learning new lands, making it possible to venture farther and more widely across otherwise unknown terrain. Presumably, if new stone sources were found the explorers would not need to return to a cache, which may account for the number of these features in the archaeological record.

We agree that the caching behavior pattern has great practical benefits during exploration ventures, and we contend that the source areas for the raw materials found in a cache clearly indicate the general areas and directions from which explorers came. To test the directionality of Clovis exploration, we assembled information on the long-distance transport to the western caches of stone identifiable to specific sources. In six out of seven caches (the seventh contained only local material), all of the stone obtained from sources more than 150 kilometers away originated from sources south or east of the cache location. Only the Green River chert in the Crook County, Wyoming, cache traveled from west to east, but it also went north. The single Clovis point found in this cache probably originated in the Spanish Diggings area to the south. Although this pattern is not conclusive, it does not support the notion that Clovis folks were exploring from the west or northwest; instead, it supports just the opposite direction of travel.

If Clovis culture originated in the east and spread westward, the eastern radiocarbon dates should be earlier than the western, and there are some indications that this is the case. Conversely, if Clovis originated in the northwest and moved toward the southeast, the earliest dates should be in the northwest. So far, they are not (figure 2.5). The oldest accepted radiocarbon assays for Clovis are 11,540±100 and 11,590±90 RCYBP (radiocarbon years before the present; see the introduction for information about this dating method), both from the Sloth Hole Site, Florida, a long way southeast from any northwestern entry port imagined in a Beringian origin hypothesis. The youngest dates associated with "Clovis-like" fluted points are 10,400 RCYBP, from Charlie Lake Cave in British Columbia, Canada, and circa 10,300 RCYBP from northwest Alaska.

If correct, these radiocarbon dates indicate at least a 500-year period during which eastern Clovis people honed their fluted point technology before expanding westward to the prairie margins. This extended period would account for the abundance and diversity of artifacts found in the east, along with large occupation sites with deep refuse middens. The younger western radiocarbon dates, minimal usage of the western landscape and stone resources, and presence of insurance caches of raw materials originating from distant southeastern quarries suggest that Clovis bands were expanding their territories out of the east and southeast.

It is our opinion that the rapid expansion of Clovis peoples into the Southwest and the Plains would have taken place at the onset of the Younger Dryas, a period of rapid climatic change when the weather turned cold and dry. It is clear that water tables were dropping in the western Plains and Southwest, as evidenced by hand-excavated wells at Blackwater Draw and Murray Springs, and this situation may have been more widespread than we currently understand. Plummeting temperatures and decreased moisture would have had a major impact on the climate and ecology, causing a significant reduction in plant and animal resources. If eastern Clovis relied on coastal estuarian resources as a primary subsistence element, a drop in sea surface temperature would have caused food shortages. Consequently there may not have been sufficient resources to support a large southeastern human population. Moreover, during this cold, dry period many of the larger animals, such as the mammoth, mastodon, sloth, camel, and horse, became extinct in North America.

Westward expansions may have occurred in different ways, including exploration along major rivers and the Gulf Coast. The river orientation of Clovis people continues in the prairies, with major concentrations of artifacts along rivers and especially at the junctions of major rivers and streams. For instance, the majority of Clovis points from Iowa have been found in counties along the Mississippi, Missouri, Iowa, and Des Moines rivers, and they are especially concentrated in the region adjacent to the confluences of these rivers. Moreover, the majority of recorded Iowa projectile points are made of raw materials from locations to either the east or the southeast.

Expansion onto the prairies and plains and ultimately to the West Coast took place near the end of Clovis times, and it appears that the Clovis use of the west was by small bands of explorers rather than large groups of colonizing people. This exploration was wide ranging, covering virtually all geographic areas and ecological zones, but it did not include long-term settlement.

That Clovis people met other peoples during these explorations is strongly hinted by the contents of the Fenn Cache in Utah. In this cache, along with the usual horde of typical Clovis bifaces and projectile points, was an item previously unknown with Clovis artifacts, a crescent bifacial artifact. Since this artifact form is thought to have its origin among Great Basin non-Clovis people, its presence in the Fenn Cache suggests that descendants of people who colonized the New World by expanding along the north rim of the Pacific Ocean encountered and exchanged technological ideas with Clovis peoples.

Once the large herbivores such as the mammoth, camel, and horse became extinct, the bison was left without effective competition for expanding grasslands, and herds flourished. As a reaction to this resource windfall, people in the Plains adapted to the new environmental conditions and focused their economy on bison. They modified their hunting equipment into more specialized and effective weapons tipped by the fully fluted projectile points we now call Folsom. The advent of Folsom technology coincides with the appearance of the first western large habitation sites, such as Adair-Steadman in western Texas, Lindenmeier in northern Colorado, and Hanson in northern Wyoming. These sites are comparable in size and contain tool kits that represent diverse activities such as we see in the large eastern Clovis sites.


Animal bones and other perishable specimens used to reconstruct subsistence activities are rare in their sites, but a few remarkable exceptions provide us a glimpse of the types of food Clovis people consumed. The primary animal remains found in western sites come from mammoths, and since these spectacular sites have been in the literature longer than those where a much more mundane diet is evident, the concept that Clovis people were mammoth specialists has become ingrained. The evidence even in the west is, however, beginning to change this image. At the Murray Springs Site in Arizona a small bison kill complemented a mammoth kill. A campsite nearby yielded broken artifacts from both the mammoth and the bison kills, leaving little doubt that the inhabitants were the perpetrators of both events. Further, there is a bison kill in western Oklahoma associated with Clovis points.

Camel bones and possible tools made of camel bone have been found at several Clovis sites, but there is no direct evidence that the Clovis people killed these beasts. At a Hell Gap Paleo-Indian bison kill on the outskirts of Casper, Wyoming, archaeologists found the remains of a camel that met his end several millennia before the bison. A Clovis point found in the vicinity lead to speculation that the hunter who lost the projectile point killed the camel.

Mastodon and mammoth remains along with Clovis artifacts made from ivory and horse bones have been found in submerged Florida sites, fitting the western stereotype of the mighty big-game hunter. However, western-style mammoth kill sites are rare in the east, where most complete Clovis projectile points are isolated specimens from upland settings. This suggests a procurement strategy in which an individual hunter or small group stalked small herds or even single animals, such as deer, that would have been common in the eastern woodlands. Burned fragments of caribou bone have been found in hearth features in the northeast, suggesting that late Clovis people hunted a variety of tundra and steppe animals.

At the Shawnee Minisink Site in Pennsylvania, fish bones and hawthorn seeds were found in what appears to be a summer camp, pitched between a stone quarry and the Delaware River. At the Aubrey Site on the banks of the Trinity River near Dallas, Texas, evidence was found of mammoth, bison, ground sloth, and even small mammal and reptile procurement. Farther up the Trinity at the Lewisville Site, hearths with burned bone suggest a summer encampment where the main foods included small mammals, amphibians, reptiles and reptile eggs, and baked mud dauber larvae. And at the Kincaid Rockshelter in southern Texas, the Clovis menu included alligators, turtles, badgers, raccoons, and mice. Thus, it appears that Clovis peoples were capable of exploiting almost any environment they encountered and adapted readily to local fare. Still, it is noteworthy and unexplained why mammoth remains, with only a slight hint of smaller game resources, dominate the High Plains and western sites.

Flaked Stone Technology

Clovis subsistence was accomplished in large part through the use of flaked stone tools. Although there are minor variations in their overall tool forms, there are nonetheless enough similarities that one can easily distinguish Clovis and descendant fluted point assemblages from all other archaeological cultures and time periods. Clovis sites exhibit very specific material culture and behaviors (table 2.1). By and large, regional differences developed in Clovis and post-Clovis tools through time, but they can all be linked through a list of related production and artifact traits that we posit have common origins.

[Table 2.1 about here]

Much ink has been spilled over the significance of fluted point variation, but from our point of view the artifact forms recovered in most archaeological excavations usually represent the end point of a tool's use, making stylistic comparisons difficult. After manufacture, artifacts were used, broken, repaired, broken, repaired again, and eventually discarded, only to be found millennia later by archaeologists, who try to make sense of their discoveries. Occasionally we get a glimpse of what Clovis points originally looked like, in the form of caches and hunting losses. However, given the state of the majority of artifacts we find, what is the meaning of variation? Are there common denominators we can use as rules of classification? We know of many definitions of what a classic Clovis point looks like. If the "type Clovis point" from Blackwater Draw defines the type (figure 2.6n), how does it compare to the rest of the Clovis points from the same site? There is in fact a wide range of variation in Clovis point forms from Blackwater Draw (figure 2.6a-h and k-n). A typologist would have trouble identifying several of these as Clovis, yet their context indicates that they are all part of the same flaked stone assemblage. Differences like these are evident even in kill sites, but variation is even more apparent in large habitation sites such as Gault in central Texas.

Clovis people used a unique technology to produce a distinctive style of projectile point that can be distinguished even from other fluted points that are their regional successors. Even so, there is not a single form of Clovis point, and we see variations through time and space. Moreover, repairing broken points could change their length-to-width and base-to-blade proportions and affect the shape of the base or the area designed to accommodate hafting. In some cases, a point was modified and used as a knife or other type of tool. All Clovis points were probably not intended for identical uses, because Clovis hunters stalked and killed a variety of game ranging from mammoths to mice. Additional variation was introduced by differences in raw material and in the skill and experience of the knapper. Even the amount of time a flintknapper had to devote to the production of a replacement spear tip-for example, in emergency situations-could have resulted in variation from the norm.

Clovis projectile points were manufactured with a generalized bifacial flaking technology; that is, flakes were removed from both faces of the specimen during its manufacture (chapter 1 for more about bifaces). This is in direct contrast to unifacial technology, in which flakes are removed from one face only. While there are examples of Clovis preforms and points that were fluted on only one face (figure 2.6q) and some points that exhibit no fluting (figure 2.6r), there are quite a few that were flaked primarily on one face (figure 2.6a-c, i-j, and o-p) with the other only partly modified, occasionally by the removal of characteristic flute flakes from the base (figure 2.6i-j). Edge grinding normally extended from the base to the maximum width of Clovis projectile points, which is usually about a third of the length from the indented base. The indented bases were edge-ground as well, and some have slightly flared ears. The point edges are usually slightly convex and taper to a sharp but not elongated tip. For the most part, Clovis points are relatively thin and are thickest at about the location of the greatest width.

What continues to interest us is not how much variation there is, even with the aforementioned circumstances, but how distinctive Clovis points, tool types, and manufacturing technology are across much of the Western Hemisphere over a span of several hundred years.

The evidence from a variety of sites, including stone procurement areas, campsites, caches, and especially workshops, makes it clear that Clovis biface production was a highly developed technology. It is also evident that although Clovis shares many traits with other biface technologies, it has certain unique techniques and combinations of techniques. Clovis people used many of the by-products of their bifacial technology for tools, and indeed the technology was probably perfected in part to produce these other items. Particularly important were the large, flat biface thinning flakes, especially useful as butchering and skinning tools.

Another distinctive aspect of the Clovis flaked stone industry was the selection of exotic materials, especially in the west. Clovis knappers went far out of their way to obtain the highest-quality and most colorful stone and acquired flake stocks and artifacts from vast distances, through exchange, direct procurement, or both. Clovis knappers had the pick of the litter at stone outcrops because they were often the first people to explore many of the geological areas they encountered. One stone type they sought out was quartz crystal (see Bradley et al. 2010, plate 1). Although high-quality optical quartz flakes reasonably well, its brittleness makes it a challenge to work. Why they chose this stone when many other, more suitable types were readily available is not clear. Perhaps it was because of its special optical or perceived spiritual characteristics. Certainly crystals have held special significance for people worldwide and throughout time. Although many non-Clovis knappers used rare stones, the degree to which Clovis people went to obtain colorful raw material is unparalleled in the prehistory of North America.

When large pieces of stone were available, the preference was to make big bifacial flake cores, to create flakes as large and elongated as possible (figure 2.7a). These cores were part of the mobile Clovis technology and provided a portable stock of stone material. They were manufactured from the glassy stones obtained from local outcrops or gravel deposits. When a stone of suitable material was found, it was worked down to reduce its size and weight and to produce a desired shape. This process guaranteed the quality of the stone before it was carried away and made the core fairly compact and easy to transport. The core provided the raw material from which large, thin, flat, and straight flakes could be removed as needed.

The bifacial cores, reduced by the removal of flakes to make tools, eventually became too thin to produce suitable flakes. Depending on the size of the biface, it would then be made into a knife or a spear point. When the core was too large for a single point, it could be broken (figure 2.7b) and the resulting pieces used to make two or more points (figure 2.7c). Pieces made from these fragments but not finished often retain a bit of the break in the form of a square edge (figure 2.7a', b', and c'). We should also point out that bifaces in various phases of production could be used as expedient tools, such as knives, when necessary. A multiple-use approach to stone use is extremely efficient for mobile groups, who only carried tested and useful stone.

Clovis biface technology is distinguished by the use of overshot thinning throughout the reduction process (figure Intro.5). We think this technique had two main purposes: to massively thin and flatten the biface, and to produce large, flat flakes for use as tools. It was also a handy way to remove irregularities on an opposite edge, such as squared-off areas. Some projectile points exhibit overshot flake scars, including the results of removing the small portion of edge opposite the platform side.

[Sidebar 2.1 around here]

Projectile point production, whether from pieces of large biface cores, unmodified raw material, or large flakes, followed the same general sequence: once a basic bifacial outline was achieved, thinning and regularizing became the main objectives. Although overshot flaking remained the primary approach, two additional but related techniques, normal thinning and full-face thinning, were also employed. Because overshot is difficult to accomplish and the outcome hard to control, platforms were individually made and carefully prepared to enhance success. The biface edge from which the flake was to be struck was turned or beveled toward the face from which the flake was to be removed. A segment of the turned edge was then isolated from the adjacent edges by the removal of small flakes from the flaking face until the striking platform protruded not only down toward the flaking face but also out from the adjacent edges. These carefully isolated platforms were centered on a ridge between two flake scars. Isolated Clovis platforms were relatively wide and straight, as opposed to the convex isolated platforms produced by many other thinning techniques. When flaking preparation was finished, the striking surface was heavily abraded. This abrasion commonly curved over onto the adjacent flaking face. Finally, force was applied with direct percussion. The angle of the blow had to be nearly parallel to the biface surface and strong enough to produce a large flake. Even if everything went just right, it took incredible skill to control the flake termination so that it removed only a small portion of the opposite edge. In our experience, this highly distinctive platform preparation and flake spacing has resulted in flakes that are as culturally diagnostic as the finished points.

[Sidebar 2.2 around here]

The second biface flaking technique used by Clovis flintknappers, full face, produced flakes that traveled most of the way across the face of the artifact and terminated smoothly at or near the opposite edge (figure 2.8a-c). This method was common but was done in conjunction with overshot flaking. It is actually the same procedure, just with "premature" terminations. Platform preparation remained about the same as for overshot flaking.

The third thinning technique, diving, was practiced by Clovis knappers to a lesser extent than the other two thinning techniques and consisted of driving flakes to the middle of the biface, ending in hinge fractures. These scars were then met by flakes coming from the other edge (see chapter 1).

Basal treatment of Clovis preforms was the same as the treatment at the sides. Bases were thinned by striking flakes from a bevel or isolated and ground platform. During the early and middle stages of preform reduction, basal thinning was part of the overall thinning process (figure 2.8d-f). The only real difference between side and base thinning was that the central basal-thinning flakes followed a low convexity formed by flake scars originating on the sides of the preform. Also, because the bases are narrow relative to the width of the removed flakes, there is little choice of where the flakes could be removed, so both faces were thinned from about the same location.

Basal thinning brings us to the thorny issue of fluting. Along with biface thinning, probably the most sought-after skill in modern North American flintknappers is fluting. It can certainly be asked what the big deal is, if fluting is just thinning at the base. The challenge is that it incurs a much higher risk of breakage than lateral thinning. There are two reasons for this: the difficulty of striking the right spot, especially if the piece has an indented base, and the stress caused by hitting the end of a long, thin piece, which can result in a break across the middle (termed end shock).

Archaeologists and collectors have focused attention on a single characteristic of Clovis points, a distinctive flute scar on the finished pieces. It is unclear why the final basal thinning flakes (called channel flakes) were removed on most Clovis points in spite of the high risk of breakage. Large flakes were frequently removed to thin the bases of bifaces throughout the production process and not just during the final stages, indicating that those removed near the end of the sequence were not technologically different than the earlier ones. Examination of the platforms of the basal thinning flakes shows that they were prepared in the same manner as platforms for other biface thinning flakes. Technologically, flutes were just like any other major thinning flake and not an extraordinary invention.

Functionally, fluting thinned the base, facilitating the point's hafting onto a spear or foreshaft. But our experimentation with hafting and using fluted and unfluted points shows no significant difference in effectiveness. If fluting was a superior way of preparing a point for hafting, why wasn't it commonly used and reinvented many times? Clovis people themselves may have valued its symbolism as expressed by style rather than its function. Since fluting was a risky technique, perhaps the channel scars were retained to testify to the skill of the knapper. When finishing a fluted point it is easier to flake into the channel scars from the sides than it is to avoid intersecting them. An example of this is seen in the post-Clovis Dalton technology, whose pieces were fluted, sometimes to the tip, before the channel scars were almost completely removed by further flaking to finish the pieces. We envision that fluting might have become part of a pre-hunt ritual.

There is considerable variation in the way Clovis points were finished after fluting. Some have only minimal pressure retouch along the edges to remove major irregularities. Others show more extensive invasive pressure retouch. Still others have the entire circumference finished with abrupt bifacial retouch, leaving the large, flat percussion flake scars prominently visible on both faces. Finally, some Clovis points exhibit all-over pressure finishing that is well spaced and meets in a distinct midline. Examination of collections from throughout North America does not show these variations to be regional, and it is common to see all of them in collections from a single area. Thus, it is unclear whether the differences have any meaning beyond individual knapper preference. Additionally, most of the points we find have been reworked, and the finishing techniques may not represent original manufacture.

Consistency does occur to the extent that even a group of points with different finishing techniques can usually be confidently identified as Clovis. Clovis fluting differs in concept and expression from the other fluted forms that are generally considered to appear later. Clovis channel flakes originated from the biface plane of the preform, and when both faces were successfully thinned the result was a base with a knife-like edge and both channels originating from about the same place.

Finally, Clovis points' lower edges were dulled by polishing or grinding. It is thought that this was a means of keeping them from cutting the lashings that held them to a spear shaft. This is not necessary for projectile points, however, and many later point types do not have edge grinding. But edge grinding might have been useful if a hafted piece was designed to be both a point and a knife. This was probably the case with Clovis points-but not Folsom points, which it's difficult to argue were designed to be knives, even though they also have ground edges. For Folsom, edge grinding might have been a traditional technological trait, or perhaps served a different function.

Along with bifaces, Clovis knappers had a well-developed blade technology, especially in eastern North America. Although we've seen a wide range of possible flaking approaches for bifaces, most North American archaeologists are unfamiliar with the equally diverse range of approaches in blade technologies in other parts of the world. In our experience, the manufacture of blades was just as complex and varied as biface technologies.

Mike Collins from the University of Texas at Austin published the first comprehensive description of Clovis blade technology, but even he agrees that he has only scratched the surface. We have examined large numbers of Clovis blades, cores, and blade-making debitage (discarded pieces), and the following description is based on our impressions. More detailed analyses need to be done on the large collections from sites such as Gault in Texas and Carson-Conn-Short in Tennessee.

We see two different blade-making technologies in Clovis, but they share a common goal and may be variations of the same general method. One approach to initiating blade making was to select a piece of stone with one or more long natural ridges-a flat nodule or a tabular piece-to create a core around. The knapper would build a platform above one of these natural ridges, and the first blade removed would have a natural dorsal surface (figure 2.9a). The second approach was to produce a single bifacial edge vertical to the axis of the nodule to serve as the initial guiding ridge (figure 2.9b). The first piece removed in this process is known as a crested blade. When blade precores are bifacially flaked all the way around their circumference, the resulting cores have a back ridge that is often retained throughout the blade-making sequence. In Clovis there is little evidence of this, perhaps an important distinction when looking for an ancestral technology.

After the removal of the first blades, there is a wide range of possible choices for continued blade production. The most common approaches used throughout the world are continuing with a single platform, detaching most blades in a single direction; and with opposed platforms, striking blades from both ends of a core. Although we occasionally see blade scars originating from opposite ends of Clovis blade cores, it is clear that most blades were struck from the primary (proximal) platform, while those originating from the opposite (distal) end were removed while fixing mistakes.

Clovis had two blade technologies: employing a core with a single flaking face, and removing blades from the entire circumference of the core. The first technique produced cores that are termed wedge-shaped (figure 2.10a). The discarded core remnants tend to be flat faced and have flat backs, with either a natural cortical surface or flake scars originating from the sides and meeting in the middle. A distinct aspect of Clovis blade making, especially from these wedge-shaped cores, is the apparent intent to make highly curved blades (figure 2.10b). This is unlike any blade-making tradition we are familiar with anywhere else at any time. In other blade technologies, heavily curved blades were the by-products of core face rejuvenation. This was necessary when the face of the core became too flat to remove another straight blade. A blade or blades had to be removed from one or more corners to make the face more convex, allowing for the removal of another straight blade.

Another characteristic of Clovis blade making is that the knappers often kept removing blades even when the core became so greatly reduced in size that it could produce only small blades. Some people call these microblades, but we prefer to reserve that designation for the highly formalized technologies where small blades were the primary product rather than just the end of the line of a large-blade technology. Hence, we refer to the small Clovis blades as bladelets. This is a distinction made by Old World archaeologists as well. It is evident that Clovis people were intentionally making these bladelets, but their purpose has yet to be discerned.

The second blade-making approach is different in that the cores have blades removed from around their entire circumference. The resulting cores, described as conical (figure 2.10c), are roughly circular in cross section and taper toward the distal end. Their sequence of manufacture is unclear. They were probably made from large nodules or cobbles rather than shaped bifacial precores. It is also curious that many of the blade scars on these cores indicate that relatively straight blades were made from them, even though very few such blades have been found in archaeological sites. Perhaps straight blades were the intended product and most of them became tools that were subsequently used and modified to the point that the original blade form was obliterated. A few end scrapers made on straight blades have been recovered, and many of the expended examples may have also been made on straight blades. A few examples of other tool types made on straight blades are also known.

While conical cores are relatively uncommon, most are large and do not look as though they have yielded many blades. They could be blade precores that were made for export from the main quarry sites or campsites, but blade cores and debitage are rare at smaller camps and kill sites. It is also possible that some of the wedge-shaped cores were originally conical, though the lack of blade scars on their backs argues against this idea.

There is another curious characteristic of the conical cores: on most specimens the final flaking was the rejuvenation of the platform, a step usually subsequent to the removal of the last blade. Almost no negative bulbs of percussion remain on any of the blade scars on the cores. Since this core preparation happened after shaping, the objects we have found may have been precores rather than finished and discarded blade cores. It is also possible that the large conical cores were some type of tool. We would like to see use-wear analysis of conical as well as exhausted wedge-shaped cores, some of which could make excellent adze edges thanks to the angle between their platform and core face. A lot of work needs to be done before these blade technologies are really understood.

The remainder of the Clovis flaked stone technology is quite simple. Most blades or biface flake tools are unifacially retouched. Implements associated with kill/processing sites tend to be made on large biface flakes, with the retouching serving as both resharpening and tool backing. Although many of these pieces are called side scrapers, this simply indicates that the retouch is on the long side rather than the end (figure 2.11a). Microscopic examination of the edges indicates that most of these tools were used for cutting.

Certain types of borers (figure 2.11b), denticulated blades (figure 2.11c), gravers and multiple gravers (figure 2.11d-g), and end scrapers on blades (figure 2.11h-i) are fairly diagnostic Clovis tools. End scrapers made on flakes are often sub-triangular in outline (figure 2.11j), and some examples have a slight spur on one or both corners (figure 2.11k). Concave scrapers are present but not common (figure 2. 11l). Micro-end scrapers (figure 2.11m) and retouched blades (figure 2.11n) also appear but may be undercounted in some assemblages.

Borers are elongated drill-like tools usually made on flakes but occasionally made from broken bifaces and blades. This tool type is rare in the west and tends to be associated with large campsites. It was probably used to gouge out sockets and to drill holes in wood.

Denticulated blades are those whose edge or edges have been unifacially flaked to produce a series of notches that form a toothed pattern. Marilyn Shoberg has reported that microscopic examination of these tools indicates they were used for cutting meat.

Gravers have a small, sharp point on the edge of a flake. Occasionally there is more than one point on the same tool, and these are termed multiple gravers. There is speculation that gravers were used to engrave bone and other hard materials, but they are so delicate that they could only have been used on soft matter such as hide or leather. They do work well for incising designs on leather. Another suggestion is that they were tattoo needles. A microscopic study by the archaeologists John Tomenchuk and Peter Storck found that multiple gravers were used as compasses for drawing or etching circles.

Adzes are an infrequently recognized tool type in Clovis assemblages (figure 2.12). These are relatively heavy tools that would have been hafted so that their working edge was perpendicular to the handle, like a hoe. They were usually highly formalized and are well described in the post-Clovis Dalton Complex in the southeast, but they have been recognized in Clovis assemblages only recently. For the most part, Clovis examples seem to be less formal than Dalton adzes, but as more of the former are identified it is likely that this impression will change. Adzes are designed primarily as heavy woodworking tools, but they may be used for other tasks as well.

Burins are flaked stone tools produced by striking away a portion of a flake, blade, or biface in such a way as to form a nearly vertical edge and a chisel-like end. This implement is common in many Old World Upper Paleolithic assemblages and in some cases is the dominant tool form. So far they are rare in Clovis tool kits, but there is evidence of burin use on the edges of broken flakes and core ridges. It is possible that flakes were intentionally broken to produce edges that would serve as small planes and possibly engravers. These tools are called pseudo-burins.

Although eastern Clovis knappers knew about heat treatment to improve the flakability of raw material, they rarely used it, and the only evidence of this we are aware of in western Clovis assemblages is from the Anzick Site in Montana. The Carson-Conn-Short Site in Tennessee contained great quantities of burned chert cortex as well as color-altered tools and flakes, suggesting that heat-treatment was used at this site, but more research needs to be done to verify this interpretation.

Artifacts Made from Perishable Materials

With the exception of Florida, where underwater sites have provided excellent preservation conditions, Clovis sites have produced relatively few bone, antler, ivory, or other perishable material artifacts, so this significant aspect of Clovis material culture is relatively unknown. Nevertheless, bone and ivory artifacts that exhibit substantial intentional shaping and modification have been recovered. Expedient bone tools-those that show use but little formal shaping-are also present in many Clovis sites, especially mammoth kill locations.

Excavations in a site near East Wenatchee in central Washington that has been interpreted as an artifact cache have yielded an assemblage of stone artifacts and a group of bone implements. Of interest here are twelve bone tools that were recovered in direct association with large fluted bifaces and finished fluted projectile points. A thirteenth bone artifact remains in place, and there is evidence that a fourteenth was destroyed by carnivore activity. Although preservation is only fair to poor, some are well enough preserved for detailed description. Most, if not all, are slightly flattened cylindrical bone rods beveled on both ends (figure 2.13a-b). The bevels are always on the same side, but usually at slightly different angles at the opposite ends. These rods were made from mammoth limb bones or the antlers of large cervids. Some retain evidence of the cancellous tissue (the spongy interior bone), but grinding and polishing have obliterated the initial stages of manufacture. Most of the flattened, beveled surfaces exhibit shallow incisions perpendicular to the long axis of the rods. This hachuring (incised checkerboard-like cuts) is considered to be intentional roughening to aid in hafting, rather than decoration.

The bone rods vary substantially in size, with lengths ranging between 125 and 276 millimeters, with a mean of 225 millimeters. Widths range between 14 and 30 millimeters, but eleven of the twelve range between 24 and 30 millimeters. It is interesting to note that groups of rods have the same width measurements: three are 24 millimeters wide, four are 26 millimeters wide, and two are 29 millimeters wide. Altogether they have a mean width of 25.5 millimeters. If the smallest is not included, the mean width is 26.5 millimeters. Width was clearly more standardized than length.

Three of the double-beveled bone rods from the East Wenatchee Site exhibit simple decorative embellishment. Two have a zipper-like incised design along the midlines of their interior surfaces (figure 2.13a). Unfortunately, preservation is poor and details of the individual incisions are not clear. The third specimen exhibits small curved incisions on the back surface of the beveled end. These shallow incisions cover the surface and resemble fingernail impressions.

Mike Gramly, one of the archaeologists who investigated the East Wenatchee cache site, has proposed that the bone rods found there were composite sled shoes, tied together and lashed to wooden sled runners. The bevels on the ends of the rods would have allowed enough overlap to produce a continuous composite runner. He notes that similar bone sled shoes were used by Inuits but readily admits that much more study needs to be done before this interpretation can be supported or rejected. We have difficulty accepting it. Skids for sled runners seem highly unlikely because of the complexity of the lashing that would have been required and the inevitable interference with traction that lashing would have caused. One might also expect distinct wear patterns to develop quickly, but there are none. Furthermore, the decorative embellishments would not have been visible.

Another Clovis cache site, Anzick in southwestern Montana, contained more than 132 artifacts and bone tools buried in a collapsed rock shelter.This site was similar to the East Wenatchee Site, as it contained a large number of bifacially flaked stone artifacts, including Clovis points, a few unifacially retouched tools, and bone rods. A notable difference from East Wenatchee was the presence of the partial remains of two subadult human skeletons. The remains of a toddler consisted primarily of skull fragments, and the artifacts were all stained with red ocher. The other individual has been shown to be unassociated with the Clovis materials.

Eleven bone rods were among the artifacts collected from the disturbed deposits of the site: two complete rods, four fragments with beveled ends, and five midsections. One complete rod is much like those from East Wenatchee and is beveled at both ends (figure 2.13b). The other is beveled on one end only, with the opposite end indented and tapered mostly flat but slightly rounded (figure 2.13c). Six of the seven bevels are incised, primarily in a crosshatch pattern, and there was organic residue thought to be resin on the beveled ends.

The complete rods are about the same length as those from East Wenatchee, but they are proportionally narrower. They measure 220 and 281 millimeters in length, but the width at their widest point ranges between 15 and 20 millimeters, with a mean of 17.9 millimeters. As with the East Wenatchee specimens, there seems to be a greater standardization of width than length.

The archaeologists Larry Lahren and Rob Bonnichsen have hypothesized that the double-beveled bone rods from Anzick were foreshafts for spears. Clovis points would have been hafted onto one of the bevels, and the other would have been lashed to a wooden main shaft. But hafting Clovis points onto the bevels of the bone rods by inserting a wedge would have produced a bulky and weak joint connection. Since this construction would have made penetration beyond the hafting area difficult, we think there are more plausible functional interpretations for these artifact forms.

Five closely related Clovis sites have been excavated in the San Pedro Valley, Arizona, not far from the border with Mexico. While three of these sites are animal kill and butchering locations, two, Naco and Escapule, may represent mammoths that died after being speared and eluding their Clovis hunters. The only formalized bone tool recovered from any of these Arizona sites was excavated at Murray Springs (figure 2.13m). It is 259 millimeters long, 21 millimeters thick, and shaped like an eyebolt. The head, 58 millimeters wide, has a hole 25-30 millimeters in diameter bored through the center. The handle is 34 millimeters wide where it joins the head and gradually tapers to an approximate width of 21 millimeters at the opposite end. The hole was purposely beveled, the most pronounced beveling being at the top and bottom on opposite sides of the tool. C.Vance HaynesJr. and E.Thomas Hemmings have convincingly argued that this unique artifact is a spear or dart shaft straightener.

Bone and ivory objects recovered during the excavations of the various Blackwater Draw, New Mexico, Clovis localities include roughly fashioned expedient tools made of mammoth bone and carefully shaped formal tools, also mostly made of bone. All of these were directly associated with flaked stone tools, including Clovis points, and with mammoth remains. Among the formalized artifacts are four points, an awl, an ivory billet, a possible flaker or ornament, and a bead preform.

Only one of the points is complete, and it has been identified as antler rather than bone. It has an oval cross section and is beveled at one end and tapers to a point at the other. Oblique scratches are visible on the bevel. The pointed tip of this specimen shows impact fracturing, and because it was found associated with a mammoth carcass it is probably a projectile point (figure 2.13d). Its length and width are 237 and 17.2 millimeters, respectively, falling within the range of the bone rods from the Anzick Site. The other three specimens are fragmentary, but all closely resemble portions of the complete artifact.

An ivory tool found in the Blackwater Draw Clovis sediments was made from a short quarter section of mammoth tusk that was ground into a cylindrical shape (figure 2.13l). This tool is 46 millimeters wide and 34.8 millimeters thick and would have measured more than 73.5 millimeters long when complete. It has been described as a burnisher or a hammer used for flintknapping. One end is slightly convex and polished; the other end is damaged by impact and longitudinal flake scars that are characteristic of hafted flintknapping hammers.

Among the other bone tools is a small bone fragment that was ground to a narrow, sharp point to serve as a perforating tool for hide working or a basket-making implement. A small, rounded rectangle with perforations incompletely drilled from both ends is a possible bead preform. A fragment of a possible canine tooth exhibits polishing and has been interpreted as an ornament or flaker, but it is too incomplete to adequately assess its function.

A modified section of a mammoth tusk tip was also found at Blackwater Draw. Although this is not a formal tool, it is an important specimen because it illustrates some of the techniques used by Clovis people to work ivory. The slightly curved section was removed by a sequence of scoring and chopping around the diameter of the tusk. After its cross section was reduced in mass, it was flexed until it snapped away from the rest of the tusk. It measures 730 millimeters long and 97 millimeters in diameter at the base. What sort of tool or tools were intended to be made from this piece of ivory is unknown, but it could have been further sectioned into beveled rods like those from East Wenatchee and Anzick or projectile points similar to those found in other localities at Blackwater Draw.

The Sheaman Site, in eastern Wyoming near the boundary between South Dakota and Nebraska, also produced a bone or antler rod in association with bifacial reduction flakes, a Clovis-like projectile point, several flaked stone tools, and bison bone. All of these artifacts were concentrated slightly below a burned and buried paleosurface. The rod is a flattened cylinder that tapers slightly to a tip broken in a distinctive manner that suggests it impacted a resistant material with considerable force. The other end terminates in a flat bevel that is lightly incised with a crosshatch pattern. The artifact's length (not including the missing portion) is 203 millimeters, and it is 13 millimeters wide. Since it has impact damage reminiscent of the damage to the bone projectile point from Blackwater Draw and it is proportionally longer and narrower than the bone rods from Anzick and East Wenatchee, we conclude that it was used as a projectile point.

Numerous ivory and bone projectile points / foreshafts, locally termed pins, have been recovered from rivers and springs in northern Florida, but because of the acidic upland soils typical of the southeastern United States, no carved ivory artifacts have been found in terrestrial Paleo-Indian sites in Florida. All of the Florida ivory specimens were found in submerged river channel locations that were inundated by inland water table rise during the Holocene. The largest concentration, which accounts for more than half of the sample, comes from the lower karst (limestone with cracks and caverns) section of the Aucilla River. Other river basins that have produced ivory objects include the Santa Fe, Ichetucknee, and Oklawaha.

Most of the known Florida ivory specimens have been recovered from stratigraphically mixed late Pleistocene deposits, where they are associated with extinct fauna, including mammoth and mastodon. Flaked stone projectile points found at these sites include Clovis fluted and an unfluted southeastern point type named Suwannee, some varieties of which are pre-Clovis. Many of the ivory and bone implements are also points, and there are good indications, because of their extraordinary preservation, that they were made from fresh rather than fossil ivory and mastodon bone, strongly suggesting a direct association between humans and mastodons. The points are similar in shape and relative proportion to those from Clovis sites in the western United States. Generally speaking, they are roughly cylindrical and taper to a sharp point at one end. The opposite end terminates in a flat bevel (figure 2.13e). Recently an incomplete example with a single barb was found (figure 2.13f).

The length of these points is highly variable because after they were broken during use they were reworked into progressively shorter weapon tips, just like stone points. The best-preserved specimens show that their surfaces were burnished to slickness except on the beveled end, which was roughened and grooved for hafting. In addition, two of the ivory points exhibit incised decoration extending from the interior end of the bevel down the shaft toward the tapered end. One of these designs is a zigzag pattern (figure 2.13e), and the other is a series of parallel lines that run transverse to the long axis (figure 2.13g).

Jim Dunbar, an archaeologist who has researched these Florida sites, argues that the ivory were foreshafts with fluted stone points hafted to the scored bevel and the pointed end inserted in a socket in the main shaft. His interpretation does not include a separate wedge on the opposite side of the stone point from the bevel, as suggested for the Anzick pieces. He bases this reconstruction on the breakage patterns observed on the artifacts. Although this hafting method would be less bulky than the wedge method, the connection between the stone point and the ivory foreshaft would be weak. Our experience with replicated Clovis weaponry indicates that penetration through the hide of an animal would have been severely impeded by either of these configurations. Thus, as with the Anzick hypothesis, we think that alternative interpretations are more plausible. For instance, taking the well-documented bevel-hafting method used by historic Inuit hunters as a model, we conducted experiments using ivory rods as projectile points and found that they are extremely lethal and penetrate thick modern elephant hide deeply with little difficulty. We think it likely that these were points, probably used on thrusting spears. Damage to many of them supports either interpretation.

There are other possible Clovis bone points, but because of poor preservation they are not common. A cylindrical bone point of unknown association was found in southeastern Saskatchewan in the early 1900s. There are no dated geological or artifact associations for this specimen, but the probability that it was made from the long bone of a mammoth or mastodon and its resemblance to bone points of known Clovis age suggest that it was of early manufacture. Another long bone point has been described from Lower Klamath Lake in southern Oregon. The presence of blue silt stains on the point and on associated mammoth bones suggests that they are contemporaneous. The similarity of this specimen to other artifacts of known antiquity also favors a Clovis-age origin. Finally, two bone points were recovered from Sheridan Cave in Ohio (figure 2.13h). Although they were not in direct association with Clovis artifacts, Ken Tankersley, the excavator of the site, has made a convincing argument that they are at least Paleo-Indian. All of its characteristics fit well with the other Clovis bone points.

C.Andrew Hemmings has identified ivory objects from Sloth Hole, Florida, as atlatl hooks, the projections on the end of spear throwers where the spear was seated (figure 2.13i-k). The discovery of these hooks supports the idea that Clovis hunters used these effective weapons for hunting big game. On the other hand, an expected artifact form in Clovis assemblages that has been elusive is an eyed needle. But two fragmentary specimens made of ivory were recovered from a river context in Florida. It is not possible to say that they were definitely associated with Clovis artifacts, but since they were made from fresh ivory and mammoths went extinct during Clovis times, this seems likely.

Not all bone tools need be made in formal styles. There has been a lot of controversy over the use of percussion flaking for the manufacture of expedient tools from the bones of large mammals, especially mammoths. Studies such as the Ginsberg experiment, the experimental butchery of an elephant, demonstrated conclusively that fresh, heavy cortical bone could be processed using techniques akin to stone knapping into cores that produce sharp-edged bone flakes. Furthermore, a growing number of flaked mammoth bones are being found at Clovis sites, such as Lange-Ferguson in western South Dakota, where bone flaking has been identified and described.

Several formalized bone and antler tools and manufacturing technologies have been documented for the Folsom period, 10,900-10,300 BP, which immediately followed Clovis on the High Plains of North America. Folsom artifact types include fluted projectile points; various kinds of flaking tools, including an elk antler tool that we think was used for fluting points; eyed needles; cut and incised bone disks and fragments (possibly ornaments or gaming pieces); and a hide flesher, a tool probably used to remove unwanted tissue from animal hides, made from a bison tibia. Virtually all of these materials have come from sites interpreted as base camps. Manufacturing techniques include grooving, incising, abrading, chopping, polishing, and drilling. Since Folsom probably derives from Clovis, it is likely that these artifact types and manufacturing techniques extend well back into Clovis times. We expect them to be found if and when Clovis campsites with good organic preservation are excavated.

Artistic Expression

Evidence of Clovis ornamentation is minimal and rare, and the representation of animals and humans in cave art is totally absent, with the possible exception of a couple of mammoth figures in the west. Although bone artifacts probably served a utilitarian function, some were decorated with incised hatching marks or zipper or zigzag designs, as noted above. It is possible that the double-beveled bone rods were conjoined into burial or other ceremonial objects. Probable bone bead preforms and a tiny pebble with a natural hole in the center found in the Clovis level at the Shawnee Minisink Site in southeast Pennsylvania, as well as beads from Blackwater Draw, indicate that Clovis peoples probably wore decorative or ritual objects.

Small incised stones have been recovered from the Clovis level at the Gault Site in central Texas. These stones are mostly thin limestone slabs that are natural in the area. The incising is mostly geometric designs, especially hatching and crosshatching (figure 2.14a and c), but at least two stones may be etched with animal representations. Some people think one of these resembles a running fox or a type of canine (figure 2.14b), and the other seems to represent several fletched spears stuck in an animal (figure 2.14d). The latter stone has the same basic design on both sides. Similar incised stones appear at other Clovis sites but are rare because of poor preservation in acidic soils or differences in site function or simply because they have not been recognized. However, a figure of a mastodon was recently found incised in a fragment of a long mastodon bone eroding out of the sediment at Viro Beach, Florida (figure 7.7h). While it has no direct association with Clovis, it was likely produced during that time or possibly even earlier.