Abstract

In North America, there are enough sites with relatively large tool assemblages predating ~13,500 calibrated years before the present (cal yr B.P.) to allow assessment of the underlying characteristics of their shared lithic tradition. Their shared technological features involve the use of dual core-and-blade and biface technologies similar to those in the Northeast Asian Late Upper Paleolithic. These dual approaches were often merged to produce small projectile points, including stemmed point forms using an elliptical cross-sectional ogive design. Similar dual lithic technologies are found in assemblages in northern Japan dating to ~20,000 cal yr B.P. We suggest a group with a similar lithic technology became isolated somewhere in the vicinity of the Paleo-Sakhalin-Hokkaido-Kuril region, developing genetically into ancestral American populations. Between ~22,000 and ~18,000 cal yr B.P., a subset of this population migrated along the southern Beringian and Northwest coasts into the Americas. By ~16,000 to ~15,000 cal yr B.P., they had become widely dispersed across North America.

INTRODUCTION

In 2013, Collins et al. (1) defined seven cultural complexes or “patterns” in the Americas predating the Paleoindian period, which begins with the Clovis Paleoindian Tradition (CPT) currently estimated to have begun sometime between ~13,500 and ~13,000 calibrated years before the present (cal yr B.P.) (2, 3). These patterns helped to demonstrate that there was a wide array of lithic complexes found in the New World dating to the ~20,000 to ~13,500 cal yr B.P. period, but what was not clear is that most of these patterns share a broad technological similarity, linking them together into what we refer to as the American Upper Paleolithic (AUP), following Williams and Madsen (4) and Davis and Madsen (5). Despite heated late 18th century debates about the appropriateness of using “Upper Paleolithic” in discussing early New World lithic technologies (6), we think that its use is critical in linking the origins of these early technologies to the rest of the world’s archeological sequences.

In large measure, these complexes are based on the use of dual core-and-blade and biface technologies similar to those found in the Late Upper Paleolithic (LUP) of Northeast Asia, indeed, most of Eurasia. In the core and blade technological system, flat-faced, wedge-shaped, and occasionally Levallois-like and conical cores are used to produce elongate blades and/or blade flakes, which constitute one of the primary tool forms used by the initial occupants of North America. These elongate, mostly prismatic, blade flakes were often used as cutting tools without modification but were also retouched into a variety of other tools such as scrapers, drills, and projectile points. In the parallel biface technology, on the other hand, collateral flaking was used to produce proportionally flaked bifaces, including projectile points, with many of the removal flakes used expediently as modified flakes. These dual technological approaches were often merged to produce small projectile points, which vary in morphological form from site to site. Stemmed, lanceolate, and triangular forms were produced either by direct bifacial reduction or by bifacially reducing elongate blades produced from blade cores. Here, we review the lithic technologies used at a suite of sites dating to the AUP period to identify cross-cutting similarities and then compare the shared technological traditions to those found in Northeast Asia dating to and just before the AUP period.

BACKGROUND

The CPT represents clear evidence for a human presence in North America by ~13,500 cal yr B.P. (2, 3). However, the origin and form of the archeological record that predates Clovis remain subjects of ongoing debate (4, 5, 7–9). While it is reasonable to infer that populations must have been present before the emergence of Clovis technology, there is less consensus regarding how such earlier occupations should be identified archeologically. Nevertheless, a number of sites have been proposed as evidence for a pre-Clovis presence in the Americas, with reported ages ranging from ~20,000 to ~13,200 cal yr B.P. and possibly even as early as ~23,000 cal yr B.P., e.g., (10, 11). This latter age is derived from the dating of human footprints at White Sands, New Mexico, e.g., (12), but it has been suggested that the associated radiocarbon age estimates may be 7000 to 3000 years too old due to the uptake of older carbon in the dated plant material, e.g., (13–15). Regardless of their ultimate age, the White Sands footprints, dating to before ~20,000 cal yr B.P. or to “only” ~16,000 cal yr B.P., would still represent some of the oldest evidence of humans in the Americas and would be contemporaneous with the lithic assemblages we discuss here. If the older age is confirmed, then it would place the age of the initial occupation of lower North America well into the last glacial period. Not all of the sites claimed to hold pre-Clovis evidence are widely accepted, and to be conservative, we have identified only 10 where the stratigraphic and chronological relationships with demonstrable human artifacts are robust enough to provide strong evidence for pre-CPT occupations (Fig. 1) (4, 16). Even the validity of these 10 sites has been hotly debated between adherents of a “Clovis First” model for the initial occupations of the Americas and those who favor a “Pre-Clovis” model. We reference some of these pros and cons in the Supplementary Materials as a guide to these debates. However, many of the AUP lithic assemblages we include come from stratified sites, which also contain post-AUP components. These overlying assemblages contain diagnostic artifacts associated with age estimates corresponding to the accepted time ranges for those diagnostics, suggesting that the AUP chronologies are also largely correct. That is, the AUP assemblages stratigraphically underlie cultural deposits of Clovis age and younger and are therefore older.

Fig. 1. Map of locations and sites.

Map showing the locations of major physiographic regions discussed in the text (A). Map showing the location of AUP sites in North America discussed in the text (B).

Only five of the 10 sites have reported lithic assemblages large enough to allow a reasonable assessment of the basic shared AUP lithic technology. These sites are Cactus Hill in Virginia, Meadowcroft Rockshelter in Pennsylvania, the Gault and Debra L. Friedkin sites in Texas, and Cooper’s Ferry/Nipéhe in Idaho (Fig. 1). The Friedkin site, located near the Gault site in central Texas, has a large lithic assemblage of well more than 100,000 artifacts dating to the AUP period, but analysis of the assemblage is not yet complete, and we provide only a limited description of its AUP component. As we noted, the White Sands footprint site is important chronologically, but no associated lithic materials have been identified as yet. Sites such as the Schaefer and Hebior mammoth sites in Wisconsin, Page-Ladson in Florida, and Paisley Caves in Oregon also appear to provide reliable evidence of pre–~13,500 cal yr B.P. occupations but are less useful in defining similarities and differences in basic lithic technology due to the paucity of lithic material they contain. However, their limited lithic collections include the combined blade or blade flake and biface production technologies characteristic of AUP assemblages and/or bifacially produced stemmed points and are pertinent as complementary or supporting sites. While this sample size of sites with notable lithic assemblages is smaller than the total number of sites bearing AUP cultural evidence, it is sufficiently consistent that it provides a preliminary picture of the AUP’s archeological record. We provide brief descriptive summaries of these sites in the Supplementary Materials.

We restrict our comparison of AUP lithic complexes to those from North American sites located south of the southern limits of the Cordilleran and Laurentide ice sheets. In South America, there are relatively few sites with large lithic assemblages predating ~13,500 cal yr B.P. Similarly, we do not consider the earliest lithic assemblages from eastern Beringia that date from ~14,500 to ~13,000 cal yr B.P. and include wedge-shaped microblade technology, microblade and flake tools, and teardrop-shaped projectile points made on thin flakes and preforms (17, 18) because it is possible, if not likely, that the lithic assemblages from these sites are associated with a later technological tradition that differs from the one from which the North American AUP assemblages seem to be derived.

Principal American Upper Paleolithic assemblages

Cactus Hill

Cactus Hill is a stratified dune site adjacent the Nottoway River on the coastal plain of southern Virginia (Fig. 1), e.g., (19) (see Supplementary Materials). Laminated sand layers at the site contain early Archaic, Clovis, and AUP deposits separated by 10 to 20 cm of culturally sterile sand (20). Chronology at the site suggests that the initial occupation of Cactus Hill was ~18,000 cal yr B.P. or earlier. While there are several dating issues, which contribute to uncertainty about this early age, the relative stratigraphic separation between the AUP and CPT horizons and the differences in technological characteristics between the lithic materials in the two horizons make it clear that the lowest cultural deposits date to some time before the CPT era.

Cactus Hill AUP lithic technology

McAvoy and McAvoy (19) identify Early and Late Pre-Clovis subperiods from excavated artifacts found at the Cactus Hill site (these assemblages are combined in Table 1). The Early Pre-Clovis subperiod includes polyhedral blade cores and core blades (Fig. 2). The Late Pre-Clovis subperiod is primarily represented by polyhedral blade cores, blades, and bifacial projectile points but also includes linear blade flakes, flake tools, unifacial end scrapers and side scrapers, tabular and irregular shaped grinding and abrading stones, and bifacial preforms [(19); figures 5.96 to 5.99]. The two small triangular points produced using bifacial collateral flaking are morphologically very similar to the “Miller” point from Meadowcroft Rockshelter but lack basal edge grinding. The authors also suggest that the recovery of a fragmentary burned bone “small projectile point tip” and abrading stones may signal the presence of an early bone tool industry at the Cactus Hill site (19).

Artifact	Category	Meadowcroft	Cactus Hill	Gault	Cooper’s Ferry	Debra L. Friedkin	Totals	Abrader	Bifaces	Blades/bladelets	Burins	Blade cores	Discoidal cores	Levallois-like cores	Graver/perforators	Modified flakes	Projectile points	Retouched blade/flake tools	Unifaces	Scrapers	Tool subtotal	Debitage
–	4	1		–	6
2	1	38	6	1	59
13	101	86	6	19	226
–	10	10	2	1	23
–	8	6	2	2?	18
2	–	4	–	1	7
–	–	–	2	–	2
2	–	33	–	1	36
2	5	250	21	17	295
1	2	10	34	12	59
4	5	5	–	5	19
–	?	14	2	–	18
–	5	7	–	13	25
26	141	464	76	83	790
298	~670	27,285	1,206	15,490	44,949

Table 1. Tool types in five AUP lithic assemblages.

Friedkin site tool quantities are based on data reported in (29) with the additional 12 projectile points reported in (30). Hence, these Friedkin site tool quantities reflect preliminary minimum values, as the final artifact analysis is incomplete.

Fig. 2. Core and projectile point from Cactus Hill.

Typical quartzite blade core from the Cactus Hill AUP assemblage (above). One of two small triangular points from the AUP assemblage at Cactus Hill produced from collateral bifacial flaking (below). The point is 35.5 mm from base to tip. From (19); used with permission.

Meadowcroft Rockshelter

Meadowcroft is a well-stratified rockshelter along a tributary of the Ohio River in western Pennsylvania containing an archeological record spanning the entire precontact period (Fig. 1) (see Supplementary Materials). The chronology for 11 stratigraphic units is controlled by more than 50 radiocarbon dates. These are in the correct stratigraphic order and are associated with diagnostic artifacts of the appropriate periods. The earliest AUP occupations are found in lower Stratum IIa, which dates to “…sometime between 13,955 and 14,555 radiocarbon years ago” (21).

Meadowcroft AUP lithic technology

The AUP lithic assemblage at Meadowcroft is somewhat smaller than the other primary assemblages discussed here, consisting of 26 worked tools and 298 unused pieces of debitage. It is, however, sufficiently large that it provides an adequate assessment of the overall technology found at the site during the AUP period. These numbers differ somewhat from those presented in some of the Meadowcroft publications, e.g., (22, 23), primarily because we do not include the lithic assemblages from middle and upper Stratum IIa. The numbers represented here are a product of the most recent analysis of the Stratum IIa lithic assemblage. The assemblage is derived from the lower portion of Stratum IIa, beginning 40 cm below the surface of the stratum and consists of a mix of bifacial tool production and core-and-blade technologies.

The 13 blade or blade fragments in the Meadowcroft AUP assemblage are “…small, prismatic blades that were detached from small, prepared cores” (23). No blade cores were recovered during the excavation, but the nearby undated, but possibly contemporaneous, Krajacic site contains both similar blade forms and small, cylindrical polyhedral cores, which “…precisely parallel the core reduction strategy…” posited for the AUP blades at Meadowcroft (23). The blade collection includes a crested blade or lame à crête, which may possibly indicate that prismatic blade core production is also represented in the Meadowcroft AUP assemblages. Several of the blades have negative flake scars on their dorsal surfaces, indicating prior removals of other blades not represented in the recovered assemblage (Fig. 3).

Fig. 3. Prismatic blade specimens from the AUP component at Meadowcroft Rockshelter.

“A” and “B” are the dorsal and ventral surfaces of a crested blade, which appears to be a product of prismatic blade core preparation. Other prismatic blades are oriented to show their dorsal (C, F, G, and I) and ventral (D, F, H, J) surfaces. Reproduced from Adovasio et al. (23) (copyright 1999, SAGE Publications; reprinted by permission of SAGE Publications; RightsLink license no. 6090510866501).

The AUP bifacial technology at the site appears geared to the production of bifacial cutting tools and flakes. Two discoidal core fragments appear related to flake production. The limited collection of bifaces and bifacially worked tools includes a unique form that Adovasio et al. (24) call “Mungai Knives.” These consist of “…bilaterally retouched rhomboidal flakes…” worked only along the flake margins. One complete biface found on a “living surface” in lower Stratum IIa was identified as a resharpened lanceolate projectile point termed the “Miller Point” (24). This was produced using collateral flaking for shaping and retouch rather than the thinning method used to produce CPT lanceolates. The point has a basally thinned, edge ground, and ~18-mm-long haft, which tapers slightly from ~23 mm at its juncture with the point blade to ~21 mm at the square base. Although the blade has been resharpened and its original morphology is unknown, the point does not appear to have been shouldered.

Gault

Gault is an open site located along a small spring-fed stream on the ecotone between the central Texas hill country of the Edwards Plateau and the Blackland Prairie on the adjacent coastal plains (Fig. 1), e.g., (25) (See Supplementary Materials). More than 150,000 AUP artifacts, termed the “Gault Assemblage,” were recovered from the lowest two stratigraphic units (units 1 and 2) above bedrock in Area 15 of the site. These were separated from a similar number of CPT artifacts in unit 4 by a 10- to 15-cm-thick zone of markedly reduced lithic debris. The age range for units 1 and 2 is between ~21,700 and ~16,700 cal yr B.P. (25).

Gault AUP lithic technology

The Gault Assemblage contained within units 1 and 2 at the Gault site is by far the largest collection of AUP artifacts in North America. Of these artifacts, 464 diagnostic and/or classifiable whole or fragmentary lithic artifacts were identified, including bifaces, projectile points, prismatic blade cores and blades, unifaces, gravers, burins, and modified flakes. Tools were made almost exclusively on local Edwards Plateau chert, which is available within the site at exposures along Buttermilk Creek.

The biface collection includes 38 whole or fragmentary bifaces and 10 projectile points, which are characterized by a focus on creating proportional cutting tools (Fig. 4A). Most of these were produced using collateral flaking for shaping and retouch. There are rare overshot and end thinning flakes, but their presence appears to be based on chance rather than design and the paucity of such flakes distinguish the Gault Assemblage from the overlying CPT assemblage (25). The bifacially worked projectile points include small stemmed and lanceolate forms (25). The only whole specimen is unusual in that it is produced from smoky quartz and is rather crude as a result. Of the remaining nine, five are basal fragments and four are undiagnostic tip fragments. Of the basal projectile point fragments, three are stemmed points with shoulders [Fig. 4A, (b) to (d) and (f)] and two other point fragments lack shoulders [Fig. 4A, (g) and (h)]. All but one of the stemmed points have slightly to moderately concave bases, and all the stems expand slightly from the neck to the base. All these stem fragments are small, with neck widths of 8 to 12 mm, basal widths of 12 to 16 mm, and haft lengths of 12 to 15 mm. Three of the points bear edge-ground haft margins [Fig. 4A, (b), (c), and (e)]. The single stemmed point, which retains a portion of the blade [Fig. 4A (f)], has beveling along one edge. The two fragments that lack shoulders have hafts that range from 21 to 22 mm at their upper limits to 19 to 22 mm at their bases and both have slight basal concavities. Only one of these unshouldered points is edge ground along its stem margins. The unshouldered point fragments are too generalized to be distinguished from later Paleoindian forms, but the other stemmed points are quite distinctive and are unlike any later Paleoindian or Archaic point forms found in the Texas region (25).

Fig. 4. AUP bifacial tools and projectile points from the Gault and Debra L. Friedkin sites.

(A) Examples of the biface technology in the AUP Gault Assemblage including large bifaces [(a) and (j)], projectile point fragments [(b) to (f)], and a lanceolate biface exhibiting collateral parallel oblique flaking (i). Modified from Williams et al. (25) [RightsLink license no. 609052136540]. (B) AUP projectile points from the Debra L. Friedkin site; (k) triangular lanceolate point, (l) lanceolate stemmed point, (m) lanceolate stemmed point midsection with base and blade sections, (n) lanceolate stemmed point midsection with base and blade sections, (o) lanceolate stemmed point midsection with base and blade sections, (p) point tip, (q) lanceolate stemmed point base, (r) point midsection, (s) beveled point tip, (t) beveled point tip, (u) beveled point tip midsection, and (v) point tip. Modified from Waters et al. (30) (RightsLink license no. 6090530177149).

A distinctive core-and-blade industry is represented by 86 blades and blade fragments and six blade cores or core fragments. Three cores are complete enough to be characterized as either formal flat-backed blade cores (two) or semiconical cores (one). All three exhibit a similar manufacturing method used to produce a regular series of blades, including the establishment of a core platform, a unidirectional blade face, a flattened back, and modified sides (26). The production sequence maintained blade face curvature and allowed for the continued manufacturing of blades as is evident on longitudinally curved blade faces on the three cores. Most of the blades can be characterized as true blades [sensu (27)], but both crested and cortical blades are also present. The length of the complete blades ranges from 70 to 147 mm. Four of the blades have been retouched along one or both margins.

The Gault Assemblage shares broad similarities with the later CPT assemblage at Gault in terms of biface technology, blade production, and flake reduction (25, 26). However, there are specific differences in all these areas, suggesting that as the Gault Assemblage technology was replaced by the patterns of the CPT, some aspects of technological toolkit were retained and modified, while others were substantially changed (26). While biface production constitutes a major component of the overall lithic technology in each assemblage, both the manufacturing strategies and the biface forms that were produced differ markedly. Biface manufacturing in the Gault Assemblage was used as a method adopted to produce nonprojectile point tools, whereas in Clovis, biface manufacturing appears to be specifically related to the production of Clovis points, with occasional use as expedient tools. Likewise, while blade manufacturing was also a major goal of lithic production in both assemblages, the types of cores and resulting blades in the two assemblages differ, suggesting that the Gault Assemblage blade manufacturing technology was “ancestral” to the more standardized CPT blade technology (26). The early lithic assemblage at the Friedkin locality, which dates to between the Gault and Clovis assemblages, is characterized by a more elaborated bifacial technologies and may be transitional between the two.

Debra L. Friedkin

The Debra L. Friedkin site, located ~500 meters downstream from Gault (Fig. 1), also preserves AUP components stratified below Clovis. Excavations at Friedkin uncovered a pre-Clovis archeological component known as the Buttermilk Creek Complex (28, 29). During the initial fieldwork phase (2006 to 2009), 15,573 artifacts were recovered from a 20-cm-thick deposit beneath the Clovis horizon, including 83 tools and 15,490 pieces of debitage. This assemblage includes evidence of biface manufacture, blade and bladelet production, and core reduction strategies. A second excavation phase conducted in 2015 to 2016 expanded the assemblage to approximately 100,000 artifacts, including 328 tools and 12 complete and fragmentary lanceolate and stemmed projectile points (Fig. 4B) found 15 to 20 cm below the Clovis/Folsom zone (30). Among the tools confirmed from the ~15,500- to 13,500-year-old sediments are 40 late-stage biface fragments, as well as blade segments, bladelets, scrapers, a discoidal core, and snap fracture tools (radial and bend-break types), most commonly made on flakes but also on bifaces and retouched flakes. Artifact analysis is ongoing, and a full description of the 328 tools in the Buttermilk Creek Complex is forthcoming.

Cooper’s Ferry/Nipéhe

The Cooper’s Ferry/Nipéhe site is located along the lower Salmon River in west-central Idaho (Fig. 1), e.g., (31) (see Supplementary Materials). Geoarcheological study of two excavation areas at the site defined a stratified sequence of deposits, including a lower loess deposit lithostratigraphic unit 3/lithostratigraphic unit B3 (LU3/LUB3) that contains an AUP occupation with age controls provided by 18 14C age estimates. Bayesian modeling of radiocarbon data places the start of occupation to between ~16,045 and ~15,725 cal yr B.P. Lithic debitage and stemmed projectile points recovered in situ outside and stratigraphically below dated pit features indicate an older as yet undated human occupation.

Cooper’s Ferry/Nipéhe AUP lithic technology

Excavations of items deposited within the sediments of LU3 in Area A produced 161 pieces of debitage and 29 stone tools, including two Levallois-like cores, two fragmentary stemmed projectile points, a projectile point blade, four biface fragments, three prismatic macroblades, a burin spall, and 19 modified flakes (Fig. 5) (31). A pit feature, PFA2, was excavated downward from the surface of LU3 at ~13,210 cal yr B.P. and contained 724 pieces of debitage and 13 lithic tools, including four stemmed projectile points, two unifaces, three blades, two Levallois-like linear macroflake products, one unidirectional core, two modified flake tools, and a hammerstone (32–34). Three pit features in Area B, F78, F108, and F151, were constructed and buried in LUB3 sediments ~15,785 cal yr B.P. and contained a total of 12 complete and fragmentary stemmed projectile points, a biface fragment, a burin spall, a small fragment of a unidirectional core, and 321 pieces of debitage. A fourth pit feature (PFP1) was found within LUB3 sediments, but its upper limits were truncated by previous excavations conducted in the 1960s, thus obscuring its upper stratigraphic characteristics. This pit feature contained 14 stemmed points but no datable organic samples (35). Pedogenic calcium carbonate coatings were observed on the lower faces of each point, identical to those seen on other early points from F78 and F108 (36), which indicates that the PFA2 points were buried in LUB3 before the Rock Creek soil developed at the site ~14,000 cal yr B.P. (31, 36). Two other stemmed points were found in LUB3 loess above and below the pit features.

Fig. 5. Stemmed AUP projectile points from the Cooper’s Ferry/Nipéhe site.

Labeled brackets at top of each set of projectile points signify the features and lithostratigraphic units associated with these artifacts. Modified from Davis et al. (31) (RightsLink license no. 6090520467321) and Davis (36) [published under the Creative Commons Attribution license (CC BY 4.0, https://creativecommons.org/licenses/by/4.0/)].

Complementary/supporting American Upper Paleolithic assemblages

There are several other North American sites with lithic assemblages of the appropriate age, which have technological characteristics similar to those summarized here. Several of these have rather large stone tool assemblages, but issues with chronology and stratigraphic placement, make these assemblages difficult to assess. We consider only four complementary sites where we feel that chronological and stratigraphic controls are sufficient to allow comparison to the larger assemblages described above.

Page-Ladson

The Page-Ladson site is contained in a sinkhole along the Aucilla River in the Florida panhandle (Fig. 1), e.g., (37) (see Supplementary Materials). Comprehensive dating of the depositional sequence produced 71 14C age estimates on wood preserved in the submerged deposits, with 24 of these coming from a stratigraphic column bracketing lithic materials. Seven, with an average age of ~14,550 cal yr B.P., are directly related to a biface and other flake tools (37). The site’s AUP lithic assemblage recovered during the two excavation phases is limited to a broken biface, two flakes showing evidence of use, and an additional 11 pieces of debitage (37, 38). The flakes are described as bifacial thinning flakes, and the pointed biface is interpreted as functioning as a knife. One of the bifacial thinning flakes has a ground platform and is more than twice as long as it is wide. This blade flake may be derived from the initial forming of a blade core but by itself is insufficient to demonstrate that a core-and-blade technology was present at the site. Together, the pieces of debitage suggest that they were produced from the resharpening of at least three different tools (38).

Schaefer and Hebior mammoths

Schaefer and the nearby Hebior site are two mammoth (Mammuthus primigenius) butchering sites in southeastern Wisconsin located on the margin of what was the fluctuating Laurentide Ice Sheet at the time the bones were deposited (Fig. 1), e.g., (39, 40) (see Supplementary Materials). The age of the Schaefer mammoth, which appears to have been butchered (41), is controlled by 13 14C dates directly on mammoth bone and another 17 on wood macrofossils recovered from the peat in and around the bones (40), which average about 14,750 cal yr B.P.

While the lithic assemblage at Schaefer is limited to the two recovered blade flakes, they are diagnostic in that they represent the production of blade flakes for use as a principal tool type by AUP foragers. Both flakes are made from local chert. One of them is a prismatic blade, termed a “crest blade” by Joyce (40). This lame à crête would probably indicate use of a blade core technique to produce these blades. The Hebior mammoth was associated with two chert bifaces, a chert flake, and a flaked chopper made on dolomite (39, 42, 43).

Paisley Caves

The Paisley Caves are a set of small closely spaced caves overlooking Summer Lake in southeastern Oregon (Fig. 1), e.g., (44). Chronology is controlled by more than 400 14C age estimates with more than 100 dating to earlier than ~12,800 cal yr B.P. Human fecal remains and stemmed points at the site date to before ~14,000 cal yr B.P., with the five earliest human coprolites dated to between ~12,500 and ~12,000 B.P. (~14,500 to ~14,000 cal yr B.P.) (45). Biface production was apparently the focus of the AUP knappers at Paisley Caves, although the number of formal tools is limited. Cutting tools consist only of a single biface and an edge-modified flake. The two older projectile points are broken but are both stem forms. The obsidian points were produced using bifacial collateral flaking, resulting in biconvex cross sections.

The stems on the two points are moderately long and tapered, ranging from ~25 to 40 mm from the base to slight shoulders where the blades begin. The distal ends of the stems are ~17 to 22 mm wide, with the stems tapering to slightly rounded bases of ~5 to 10 mm. The stem margins on the oldest point dating to between ~13,170 and ~13,314 cal yr B.P. are finely retouched and edge ground. The stem of the other possible AUP point is unground and has not been finely shaped by retouch. It may either have been broken during initial manufacture or was reworked and repurposed into a knife after it was broken. The shape and length of the blades are unknown.

There is no evidence of blade production in the Paisley Caves AUP lithic assemblage, but the number of tools in the collection is quite small, so the absence of blades may not be definitive. On the other hand, the amount of debitage is relatively large, and if blade production was common, then blade fragments might be expected to show up in the assemblage. Moreover, as Jenkins et al. (44) note, blades are extremely rare in later Western Stemmed Tradition (WST) assemblages, a Late Pleistocene to early Holocene cultural pattern defined by nonfluted stemmed and lanceolate projectile points found at many sites in western North America, e.g., (46–49). This WST pattern suggests that blades and formalized blade production may also have been absent at Paisley Caves.

Characterization of AUP technological attributes and patterns

A basic characteristic of AUP lithic technology is the production of linear flakes and blades from unidirectional, discoidal, polyhedral, and Levallois-like cores (Table 1). These flakes, blades, and spalls are used as informal tools and are further reduced to create a range of formal tools including bifaces, projectile points, gravers, unifaces, and burins. In some, but not all, AUP sites, projectile point production appears to be based, in part, on the relatively minimal reduction of flakes and blades, which often result in forms that tend toward plano-convex cross sections. Larger AUP biface technology is seen as well and appears to be based on more extensive reduction of larger flake spalls into bifacial preforms and finished bifaces with biconvex cross sections. AUP points are unfluted and appear in stemmed and lanceolate forms. AUP points are generally small, and examples from the Gault, Friedkin, and Cooper’s Ferry/Nipéhe sites are particularly diminutive—the smallest of which measured ~2 to 4 cm in length. AUP points appear in a range of two-dimensional hafting forms, including contracting rounded, pointed, and squared bases (Cooper’s Ferry/Nipéhe), concave and slightly contracting margins with squared bases (Meadowcroft), and expanding and slightly convex margins with square and concave bases (Gault). The stemmed AUP points generally have basal edge grinding, even including the smallest point form seen at the Cooper’s Ferry/Nipéhe site, which suggests that it was deployed in some kind of haft. Bifacial reduction is evident in the tools and debitage found at Cactus Hill, Meadowcroft, Gault, Friedkin, Cooper’s Ferry/Nipéhe, Hebior, Page-Ladson, and Paisley Caves. While bifacial collateral flaking to a midline is most common, overshot flakes were found in pre-Clovis contexts at Gault, Friedkin, and Cooper’s Ferry/Nipéhe. Further research is needed to determine whether this technological feature is unique to the CPT or has its origins in the AUP. The presence of smaller projectile points made on flakes or blades and larger bifaces made through more intensive reduction of bifacial preforms illustrates an important duality in AUP lithic technology that is reflected in later CPT and WST point forms where large and small point forms were used together. While the points were made through both bifacial reduction and modification of blade flakes, they can take the same general morphological shape. These shapes are varied in general, but small triangular forms are more common in eastern North America while stemmed forms are more common in the West. The AUP components at Cooper’s Ferry/Nipéhe also contain fragmentary prismatic blades made on cryptocrystalline silica and a unidirectional bladelet core made on fine grained volcanic material from PFA2 (34) and two fragments of prismatic blades made on chert.

Comparison of American and Northeast Asian Upper Paleolithic technologies

As Adovasio et al. (23) concluded decades ago: “Collectively, these data suggest that the first inhabitants of eastern North America employed a technologically standardized and sophisticated, small, polyhedral core– and blade-based industry of decidedly Eurasiatic, Upper Paleolithic ‘flavor’.” With the excavation and reporting of several additional well-dated AUP sites, this Upper Paleolithic “flavor” has become even more evident, leading to the obvious question of where this technology may have originated. The antecedents of the AUP may have been either in Europe or Asia, as Adovasio et al. noted, and a European origin for this American lithic technology has been suggested by some, e.g., (50). However, paleogenetic studies indicate that an Asian origin for Ancient Native American (ANA) populations (51) and the progenitors of AUP technology were most probably located somewhere in northeastern Asia. Unfortunately, these paleogenetic data are not yet sufficient to tell us where that location might have been.

One possibility is Beringia, the land bridge that connected Asia to the Americas during the Late Pleistocene. That bridge has long been held to be the most logical land route used by migrating ancient North Asians, e.g., (52), and until several North and South American sites were shown to predate the opening of an ice-free corridor between Beringia and the Americas south of the ice sheets, that route was long considered to be the only viable one. The hypothesis was so embedded in peopling of the America models that most of the same paleogenetic studies, which pinpointed a Northeast Asian source for those early populations, also identified an isolated standstill during that migration as a “Beringian standstill” without a shred of evidence that it took place in Beringia, e.g., (53, 54). There are, however, good reasons to question that widely held assumption.

First, there are no known stone tool complexes in Beringia that are old enough and/or similar enough to be possible progenitors of AUP technology. All lithic tool assemblages in eastern Beringia postdate ~14,500 cal yr B.P., lack large core-and-blade production, and are characterized by a microblade technology and a small “teardrop-shaped” projectile point pattern that is not found in the AUP (17, 18, 55). Later lanceolate and stemmed point forms appear to have been introduced to eastern Beringia from lower-latitude North America (56). Second, what is known of Beringian paleogenetics suggests that Late Pleistocene populations there dating to ~11,500 cal yr B.P. did not contribute to the populations that first occupied North America south of the ice sheets (57). Similarly, paleogenetics suggest that pre–Last Glacial Maximum (LGM) populations in far northeastern Siberia did not contribute to ANA populations and that those early populations were replaced later by different groups from East Asia (58), in keeping with archeological evidence that northern Siberia was largely abandoned by humans during the LGM (17, 59). In other words, present speculation that ANA populations and AUP technology originated in Beringia, e.g., (7, 8), is just that—speculation—and so we need to look elsewhere for a likely cultural progenitor of the AUP. The archeological record of the northwestern Pacific Rim’s Paleo-Sakhalin-Hokkaido-Kuril (PSHK) peninsula and its vicinities in continental East Asia is older than the AUP, is geographically most proximal to North America, and is therefore the more likely region where an AUP cultural progenitor may be found. That too is largely speculation, given that there is much we still do not know about the Paleolithic technologies around the PSHK region, with part of that problem likely due to limited English language publications. Intensive study of collections will be needed to support a meaningful comparison between LUP lithic assemblages in both northeastern Siberia and the PSHK regions and AUP lithic technologies.

Davis et al. (31, 36), Davis and Madsen (5), and Buvit et al. (60) suggest that the pre-Clovis–aged large and small stemmed projectile point, core-and-blade, and biface technology found at AUP sites such as in the LU3/LUB3 deposits at Cooper’s Ferry/Nipéhe share design and manufacturing links with Late Pleistocene–aged lithic technological traditions observed in the PSHK region (Figs. 1, 6, and 7). Paleogenetic studies indicate that the progenitors of the First Americans share genetic ancestry with LUP peoples from both southern Siberia and far eastern Asia, becoming geographically isolated sometime after ~25,000 cal yr B.P. (57, 61, 62) before expanding into the Americas after ~19,500 cal yr B.P. (62, 63). While paleogenetics has not yet pinpointed the exact location of their northeastern Asian residence, archeological evidence—particularly lithic technology—provides critical clues. The closest comparable projectile point forms in Northeast Asia to those predating ~16,000 cal yr B.P. found at the AUP sites discussed here are associated with LUP bifacial point-bearing sites in Hokkaido, (Fig. 6), e.g., (31, 36, 60).

Fig. 6. Comparison of stemmed points from Japan and North America.

AUP points from Cooper’s Ferry/Nipéhe LU3 (A, C, D, and F to H) and bifacial stemmed points from Hokkaido (B, E, and I to K). Modified from Davis et al. (31) (RightsLink license no. 6090520467321).

Fig. 7. LUP artifacts from the Okushirataki 1 site, Hokkaido, Japan.

Illustrated examples of bifacial stemmed points (A to D), blade core (E), blade tool (F), and biface (G).

Okushirataki 1

The Okushirataki 1 site was excavated during rescue campaigns conducted by the Hokkaido Archaeological Operations Center in 1997 and 2000. This site is important, as it provides a basis for establishing the age of LUP bifacial stemmed points in the PSHK region. A large excavation block covering 7752 m2 revealed 53 lithic scatters and 19 dense charcoal concentrations, primarily within an eolian sediment deposit designated as lithological unit IIa. In total, 830,243 lithic artifacts were unearthed, with 99,249 items found and mapped in situ. The site contains evidence of repeated occupation, beginning with an Early Upper Paleolithic (EUP) component characterized by a small flake-based tool assemblage. This was later followed by microblade technologies and bifacial points associated with subsequent LUP occupations (64–66).

Lithic scatter no. 53 at Okushirataki 1 yielded key evidence regarding bifacial stemmed points. A total of 9915 stone artifacts was recovered. Associated charcoal from feature no. 19 provided three radiocarbon dates with mean calibrated ages of ~21,400, ~21,360, and ~19,830 cal yr B.P. The youngest of these, ~19,830 cal yr B.P., is interpreted as the approximate age of the lithic assemblage. The elimination of older dates, combined with the occurrence of similar dates at other locations within the site, may suggest a past wildfire event. The lithic assemblage from this scatter includes bifacial projectile points, four of which being stemmed forms, as well as bifacial blanks or preforms, burins, end scrapers, side scrapers, and boat-shaped tools (Fig. 7). Other artifacts found at the site include retouched blades and flakes, prismatic blades, elongated flakes, blade cores, other cores, spalls, and a large quantity of debitage. Additional items such as anvil stones and a single obsidian nodule were also recovered. Obsidian dominates the lithic raw material assemblage, with minor occurrences of other cryptocrystalline rocks such as siliceous shale and agate. Chronological analysis of flake scar patterns, along with an extensive refitting analysis, indicates that the lithic artifacts in scatter no. 53 were primarily produced through blade-based, flake-based, and bifacial reduction techniques. The bifacial stemmed points, as shown in Fig. 7, were manufactured from the reduction of larger elongated flake blanks.

Hattoridai 2

The Hattoridai 2 site was excavated during 3 years of excavations conducted by the Hokkaido Center for Buried Cultural Property between 1998 and 2000 (64). The excavation covered a vast area of 6691 m2 on a very flat terrace along the Yubetsu River. A total of 65 lithic scatters and nine dense charcoal concentrations were found, primarily within eolian sediments of lithological unit IIa. In total, 798,648 lithic artifacts were unearthed, with 67,754 mapped in situ. The site contains an archeological sequence that begins with an EUP component, followed by an LUP component, which is characterized by various types of microblade technologies, and bifacial projectile points.

Lithic scatter no. 39 at Hattoridai 2 contains 3721 stone artifacts. Among these, 22 are bifacial projectile points, including 6 stemmed examples, along with 44 bifacial blanks or preforms, 3 end scrapers, 5 side scrapers, 6 retouched blades and flakes, 27 blades, 7 elongated flakes, 13 cores, 3 spalls, and 3594 flakes. Obsidian is the dominant raw material, with smaller amounts of siliceous shale, glassy andesite, and jasper. The artifacts were primarily produced through blade-based, flake-based, and bifacial reduction techniques. Charcoal from feature no. 5, which was a combustion feature directly associated with this lithic concentration, yielded three radiocarbon dates, with mean calibrated ages of ~16,750, ~16,620, and ~16,170 cal yr B.P. (64).

For the purpose of comparing AUP and eastern Beringian lithic technologies, the artifact assemblages from lithic scatter 53 at Okushirataki 1 and lithic scatter 39 at Hattoridai 2 lack microblade and microcore technologies, distinguishing them from other LUP toolkits in the region. However, the relative chronology and uses of microblades versus bifacial stemmed points in Hokkaido are complex and have been major research topics among local archeologists for a long time. Several assemblages contain both microblade reduction technology and bifacial stemmed points, although not always. For example, Hirosato- and Oshorokko-type microblade cores are found alongside Tachikawa-type bifacial stemmed points at sites such as Yoshizawa, Motomachi-2, and Nakamoto, while other bifacial point assemblages, such as those at Ochiai and Satsunai N sites, lack microblade reduction (67, 68). The reasons for this variation remain unclear. Some researchers attribute it to temporal changes, while others suggest functional and contemporaneous differences between sites, but both interpretations lack solid evidence.

Microblade technology appeared in PSHK around the onset of the LGM and began to spread in southern Siberia ~23,000 cal yr B.P. (59, 60, 69). It continued to spread across Siberia, particularly in regions where these novel inset weapons served, in part, as a solution to the challenges of provisioning during long, harsh winters—supporting tactics adapted to the risks of cold-weather large-game hunting (70). Nonetheless, microblade technology is also found in later periods, for example, as far south as southern Kyushu, Japan, in the LUP context (71). It has even been controversially argued to be in association with pottery by 15,000 cal yr B.P. (72). Therefore, the appearance, design, and functional purposes of microblade technology in relation to environmental conditions may require further assessments. In eastern Beringia, microblade technology first appears after 14,500 cal yr B.P., followed by the emergence of smaller, thinner, teardrop-shaped bifacial projectile points shortly after 14,000 cal yr B.P. (17). Microblade technology, despite its utility in Siberia and Beringia, was not transmitted into AUP sites south of eastern Beringia during the Late Pleistocene for reasons yet to be fully explained.

The appearance of fully bifacially flaked, stemmed projectile points in the PSHK region after ~20,000 cal yr B.P. mark