Research ArticleGEOLOGY

Evidence of an early projectile point technology in North America at the Gault Site, Texas, USA

See allHide authors and affiliations

Science Advances  11 Jul 2018:
Vol. 4, no. 7, eaar5954
DOI: 10.1126/sciadv.aar5954


American archeology has long been polarized over the issue of a human presence in the Western Hemisphere earlier than Clovis. As evidence of early sites across North and South America continues to emerge, stone tool assemblages appear more geographically and temporally diverse than traditionally assumed. Within this new framework, the prevailing models of Clovis origins and the peopling of the Americas are being reevaluated. This paper presents age estimates from a series of alluvial sedimentary samples from the earliest cultural assemblage at the Gault Site, Central Texas. The optically stimulated luminescence age estimates (~16 to 20 thousand years ago) indicate an early human occupation in North America before at least ~16 thousand years ago. Significantly, this assemblage exhibits a previously unknown, early projectile point technology unrelated to Clovis. Within a wider context, this evidence suggests that Clovis technology spread across an already regionalized, indigenous population.


Current research on the early human occupation of the Americas no longer recognizes Clovis as the expression of a founding population (1, 2). Increasing diversity, range, and time depths within the expanding database of sites predating Clovis attest to greater complexity in the early record (3) than previously thought. Archeological opinion on the nature, timing, arrival, and peopling scenarios remains divided (46). Despite this, there is increasing evidence to support a number of contemporaneous (7) and older (2, 8) cultural manifestations at least 2 thousand years (ka) before the appearance of Clovis (9). This includes the Western Stemmed Tradition (10), Beringian assemblages (11), and Eastern Seaboard sites (1214) in North America alongside the El Jobo/Monte Verde and fishtail bifacial technologies and edge-trimmed traditions in South America (1517). These technological patterns require careful and systematic evaluation to address the nature and timing of both the early occupation of the Americas and, subsequently, the origins of Clovis (Supplementary Materials).


Initially identified in 2002, excavation at Area 15 of the Gault Site (Fig. 1) was undertaken to explore evidence of early cultures in Central Texas. Research focused on the manufacturing technologies, their relationship to Clovis, and the associated age of this assemblage. This report focuses on the optically stimulated luminescence (OSL) ages obtained from the lowest deposits in Area 15 that contain a material that predates Clovis, referred to as the Gault Assemblage.

Fig. 1 Location of the Gault Site, Texas, USA.

The site occupies an upstream first- to second-order floodplain of the Buttermilk Creek valley, which has reliable springs and high-quality chert outcrops. The setting is an ecotone with mesic, riparian flora in the valley floor and xeric-adapted plants on the adjacent uplands. Regionally, the setting is in the Balcones Canyonlands, itself an ecotone, where resources of the limestone uplands (the Edwards Plateau) interface with the Blackland Prairie on the adjacent coastal plains.

The Area 15 excavation block, located within the valley floor, is a 56-m2 grid, which was stepped down in meter increments. Excavation was conducted in 1 m × 1 m squares within this grid in arbitrary levels of 5 cm deep from 93.00 m and below (based on an arbitrary site datum of 100.00 m). The upper ~1.8-m deposit is a midden, common to Central Texas (18). Below this is a ~1.2-m-thick silty clay deposit atop a ~0.20- to 0.50-m-thick fluvial gravel. The gravel rests on Cretaceous-age limestone bedrock of the Comanche Peak Formation (Supplementary Materials).

The sediments in Area 15 are well stratified with diagnostic projectile points and associated artifact complexes in chronological order (Fig. 2) and, in many cases, are separated from one another by a decrease in debitage counts. The midden deposit contains Archaic projectile points in good stratigraphic order. Within the midden, there is a vertically constrained distribution of small (<1 cm) diagnostic Andice notching flakes recovered between 94.90 and 94.00 m (fig. S5). Field observations of the orientation of artifacts suggest that shrink-swell movement of the soil has reoriented some materials and favored the downward migration of some of the smallest artifacts, but the absence of diagnostic artifacts like Bell-Andice notching flakes (Supplementary Materials) below the appropriate age deposit suggests that this movement has been limited. The silty clay beneath the Bell-Andice occupation contains an ~50-cm-thick sequence of Late Paleoindian components in a stratified order overlying an ~25-cm-thick Clovis component. Below this is an ~65- to 80-cm-thick deposit overlying undulating bedrock that consists of silty clay and fluvial gravel deposits, containing the Gault Assemblage. A further indicator of the stratigraphic integrity of the site is the separation between cultural components. Furthermore, bioturbation, such as animal or root disturbances, was identified and excavated separately in the field. Debitage counts indicate that decreased cultural material frequencies occur between the Gault, Clovis, and Paleoindian components. There is an ~10-cm-thick zone of decreased cultural material between the Clovis and Gault components. This suggests a reduction in site activity or possible occupational breaks between the three cultural depositions.

Fig. 2 Stratigraphic profile of the Area 15 excavation block showing the diagnostic cultural materials and components alongside the stratigraphic sequence.

Diagnostic projectile points listed on the left were all found within the associated deposits (SMH, St. Mary’s Hall). Stratigraphic unit numbers are shown on the right, and the cultural horizons are highlighted in gray.

Stone tool assemblage

The stone tool assemblage recovered from the lowest, earliest deposits exhibits a small projectile point technology as well as both a biface and a blade-and-core tradition. The projectile points from the Gault Assemblage exhibit two stem morphologies: stemmed and lanceolate (Supplementary Materials). One stemmed projectile point (Fig. 3I) exhibits a slightly concave base, with concave lateral margins and short shoulders with beveled edges. In profile, this point is slightly curved, suggesting that it was manufactured on a flake. Two bifurcate stemmed points were also recovered (Fig. 3, H and J); both have a deep concave base, an expanding stem, and exhibit beveling. A small proximal tip with beveled edges was also recovered (Fig. 3K). These points were likely produced on flakes and predominantly manufactured using pressure flaking to shape and finish the points. These stemmed points are technologically and morphologically distinct from any later regional cultural manifestations. Superficially, they resemble point types within the regional Early Archaic yet differ in base treatment and blade bevel (Supplementary Materials). The two lanceolate projectile points (Fig. 3, X and Y) are similar in size, exhibit a concave base, and share similarities in the basal flaking and finishing. Only one point (Fig. 3X) is ground along the edges. Both points are snapped at the stem, but existing flaking pattern suggests comedial (midline) flaking. The lanceolate points superficially resemble Late Paleoindian types but do not fit any single point type from this period. A sixth point exhibits weak shoulders and a contracting stem (Fig. 3G). This point is made from smoky quartz and exhibits a central ridge produced from comedial flaking. Unlike the three points discussed above, the morphology of this point resembles Western Stemmed points, but its age places it outside of the known chronology for this type (7). All projectile points were recovered from undisturbed sediments within Area 15 with no evidence for the downward movement within their excavation units. This projectile point assemblage is unlike anything in the early archeological record of the Americas and indicates complex behavioral activities associated with a group or groups who colonized the New World.

Fig. 3 Gault Assemblage artifacts

(A to D, F, and L) Bifaces. (E) Blade core. (G) Quartz projectile point. (H and I) Projectile points. (K) Projectile point tip. (M, V, and W) Blade. (N) Unifacial tool. (O and T) Gravers. (P) Discoidal biface. (Q) End scraper. (R to U) Modified flake tools. (X and Y) Lanceolate projectile points. Descriptions are given in the Supplementary Materials.

Alongside these points, approximately 150,000 artifacts, consisting mainly of debitage, have been recovered from these lowest two units (see strata 1 and 2 in the Supplementary Materials). To date, 184 flaked stone artifacts have been analyzed (Supplementary Materials). These include the distinct stemmed projectile points, blades and blade cores, bifaces, and flake tools. The Gault Assemblage shares this generalized biface and blade-and-core lithic tradition with the overlying Clovis materials but differs significantly in the following ways.

The Gault biface assemblage exhibits the prevalent use of comedial (midline) flaking indicative of proportionally thinning a biface, which closely parallels the reductive technology used to create the Gault Assemblage projectiles. Clovis, however, demonstrates the use of full-face and overshot flaking to produce thinned bifaces. In addition, the flake striking platforms produced during manufacture are larger and less prepared than the Clovis flake platforms (19).

In contrast, the blade-and-core assemblage shares more commonalities with Clovis technology. Both technologies exhibit flat-backed blade cores that use a single blade face with an acute platform and unidirectional blade removals as well as conical cores that have blade removals around the circumference of the core. Evidence from the blade platforms indicates that the Gault Assemblage generally exhibits less preparation than Clovis platforms.

The similarities and differences suggest that there is no single linear trajectory toward Clovis technology within the Gault Assemblage. Instead, parts of the technological repertoire, like the blade-and-core tradition, appear to have continued in the Clovis levels at the Gault site, while the projectile points and the biface traditions underwent significant changes. In a broader context, the technologies present in the Gault Assemblage appear to represent a unique pattern within the early human occupations of the Americas that indicates a regional adaptation after the initial colonization of the New World.


Four OSL samples were collected from the lowest cultural bearing deposits (see strata 1 and 2 in the Supplementary Materials) in Area 15 of the Gault Site (Fig. 4 and table S1). Samples were collected by horizontally hammering 4.5-cm-diameter steel tubes into clean open sections. After collection, the sample holes were slightly enlarged to obtain in situ gamma spectrometric measurements at all locations.

Fig. 4 OSL ages in association with the stratigraphic units and the projectile points recovered from the Gault Assemblage (not to scale).

Stratigraphic unit numbers are shown on the left, and the cultural horizons are highlighted in gray. Clovis ages have been reported elsewhere (see text).

Samples were prepared at the Department of Geosciences at Murray State University, Murray, KY under low red fluorescent lighting fitted with Lee 106 filters. Pure quartz grains were obtained using standard OSL preparation methods, including treatment with 10% HCl and 30% H2O2, to remove carbonates and organic material, respectively. The samples were wet-sieved to obtain a 32- to 63-μm grain size fraction, and a heavy liquid separation using lithium polytungstate was performed to isolate quartz. No HF etching was applied.

Luminescence measurements were conducted at McMaster University, Hamilton, ON using a Risø OSL/TL-DA-15 reader fitted with a 7-mm-thick Hoya U-340 filter. Measurements were carried out using blue diodes (470 nm) operating at 90% power (~30 mW/cm2). Laboratory irradiations were performed using a calibrated 90Sr/90Y radioactive source attached to the Risø luminescence reader. Blue LED’s (light-emitting diode) multigrain aliquots were prepared on 9.8-mm-diameter stainless steel discs using a silicon spray and a 0.5-mm mask (~72 grains).

An initial equivalent dose (De) estimate was made by comparing the natural OSL signal of four aliquots to their OSL signal after a given dose. A second identical regenerative dose was applied to the same four aliquots, and the Infrared Stimulated Luminescence (IRSL) signal was measured as a check for feldspar contamination. All aliquots had an IRSL-to-OSL ratio of <1%, suggesting no significant feldspar contamination in the prepared samples. A dose recovery test was used to determine the preheat temperature (160°, 200°, 240°, and 260°C) at which a given dose could be best recovered. A preheat temperature of 200°C and a cut-heat temperature of 160°C produced a dose closest to the given dose and was used in all subsequent De measurements. Thermal transfer tests that were carried out to assess the possibility of charge transfer from light-insensitive shallow traps to light-sensitive OSL traps showed no significant thermal transfer (20). During this experiment, the OSL signal of several aliquots was first bleached by a 400-s exposure to blue LED’s. Apparent De’s were then calculated using the SAR protocol (21), but applying different preheat temperatures to different aliquots. Significant thermal transfer [>1 gray (Gy)] was only observed with the application of preheat temperatures more than 260°C.

Final De measurements were made on 48 aliquots for each measured sample. All measurements followed the SAR protocol (21, 22) on 0.5-mm multigrain aliquots of 32- to 63-μm quartz with a total stimulation time of 100 s. This fraction was targeted to isolate quartz-rich silt identified through petrography (23) that was presumed to be incorporated in the floodplain sediments through aeolian processes. The OSL signal was integrated from the first 0.4 s of the decay curve, and the subtracted background was integrated from the last 4 s. Aliquots were required to pass the following criteria for further analysis: <10% test dose error, <10% recycling ratio error, <10% recuperation, <10% palaeodose error, and a signal greater than 3σ above background. All De values incorporated an instrumental error of 1.5%.

Final De values used for age calculation were statistically modeled using the central age model [CAM; (24)] owing to low overdispersion (that is, <10%) and the normal distribution of De’s in each sample. A σb value of 0.045, calculated from dose recovery results, was added in quadrature to all De estimates to account for variability arising from the intrinsic luminescent properties (25).

External α and β dose rates were determined from the U, Th, and K concentrations of a small amount of sediment (~2 g) collected from each OSL sample and measured with neutron activation analysis and delayed neutron counting (conducted at the McMaster University Nuclear Reactor). Conversion of radioisotope concentrations was done using the data of Guérin et al. (26). Dose rates were calculated, assuming secular equilibrium in the U and Th decay chains. Dose rates were corrected for water content using laboratory-based measurements and for attenuation using factors from Brennan et al. (27) and Guérin et al. (28). External gamma dose rates were obtained in situ at all sample locations using a NaI(Tl) Harwell four-channel gamma spectrometer. Cosmic dose rates were calculated on the basis of the methods by Prescott and Hutton (29, 30) and calculated using a 2.0 g/cm3 of overburden density, assuming a linear sediment accumulation.

Ages were calculated by dividing the De modeled with the CAM by the corresponding total dose rate. In addition to the errors on the modeled De’s, the following systematic errors were incorporated into each age calculation: ±25% for moisture content and ±10% for cosmic dose rate. OSL age results are reported with 1σ errors in table S1. An average of the laboratory-measured moisture content from each sample of 20% was used for age calculation.

Disequilibrium measurements were not conducted on the samples at Gault (31). To investigate its potential impact on age estimations, Rn loss calculations were carried out using data tables provided by Guérin et al. (26). Ratios of 238U pre-Rn loss to 238U total U energy release were used to calculate multiplication factors to modify U concentration in the age calculations. Factors were calculated separately for alpha, beta, and gamma energy releases and then applied to find the effective U concentration for 100, 50, and 25% Rn loss. A U-only dose rate was obtained for each calculation and added back in to the total dose rate from Th, K, and cosmic rays (table S2).


The OSL samples displayed favorable luminescence characteristics including low overdispersion (<10%) and normally distributed De distributions. Moreover, the fast component contributed more than 90% of the signal measured in the first 0.4 s of stimulation, suggesting that the OSL signals from these samples are fast component–dominant. A representative decay curve and growth curve are shown in fig. S1. All measured De’s were significantly lower than their corresponding D0 values (~65 Gy). Equivalent dose distributions are displayed as histograms and radial plots in fig. S2.

The OSL ages presented here establish the presence of a cultural component, stratified below Clovis, and associated with ages older than ~16 ka (Fig. 4 and table S1). These OSL ages range from 21.7 ± 1.4 ka to 16.7 ± 1.1 ka and, within error, are in the expected stratigraphic order (Fig. 4 and table S1). On the basis of the results of OSL dating presented here, we find a mean age for the Gault Assemblage (n = 4) of 18.5 ± 1.5 ka.

Ages associated with the temporal diagnostic artifacts above the Gault component are in excellent stratigraphic agreement (32). This includes four OSL ages of 11.9 ± 0.8 ka, 12.9 ± 0.6 ka, 13.2 ± 0.6 ka, and 13.6 ± 0.6 ka from the Clovis component (Fig. 4) (32). These dates agree with the known Clovis range of ~13.5 to 12.9 ka (3335) and agree with the ages from other Clovis sites in Texas (3638). These data emphasize the stratigraphic integrity of Area 15 and the agreement between the temporal diagnostic artifacts and OSL ages (Supplementary Materials).

The OSL ages for these early levels at the Gault Site are in good stratigraphic agreement with the known, younger, temporal diagnostic artifacts and age estimates indicating the reliability of this dating sequence (32). The significant reduction in artifact frequencies between the Clovis and Gault Assemblage confirms the presence of an older, isolated, assemblage below Clovis. Given the SDs for these OSL ages (Fig. 4 and table S1), the Gault Assemblage is dated to at least 16 ka, which is the youngest possible age for this occupation; however, the time span suggests that the inhabitation of the Gault site began ~1 to 2 ka before.

The Gault Site differs from other OSL-dated sites relevant to the early occupation of North America for two reasons. First, there is a well-dated (32) long stratigraphic sequence above the Gault Assemblage with distinct and well-separated occupational horizons. In addition, the OSL De distributions are normal and exhibit low (<10%) overdispersion, which provide more confidence for the modeled final De values used in age calculation (Supplementary Materials).

Disequilibrium in the U-series is commonly present in carbonate-rich environments and can potentially change the dose rate over time, leading to inaccurate age estimations (39). Although disequilibrium measurements were not carried out for the Gault samples, several inferences can be made about the potential impact that this would have on the ages we report. The samples collected at Gault have U contents that make up a relatively small (~20%) contribution to the total dose rate, so any influence of disequilibria on the resulting ages may not be very significant. For samples studied with a similar U content, and assuming a >50% disequilibrium in the U-series chain, OSL ages have been in 8% error from the true age (39). This generally falls close to or within the 1σ age errors that we report. Moreover, our Rn loss calculations suggest that for the Rn loss to have a statistically significant effect on age calculations relative to 0% Rn loss, for nearly all samples, >50% Rn loss would have had to have occurred (table S2).

In general, U concentrations in carbonate rocks are quite uniform at approximately 2 parts per million (ppm) (40). Phreatic cements have been less well studied; however, a U concentration of 1.80 ± 0.75 ppm was found for a series of freshwater phreatic cements by Chung and Swart (41). The total U concentration in our Gault dating samples ranges from 2.13 ± 0.1 ppm to 2.48 ± 0.1 ppm, indicating the possibility that much of the dose rate comes from carbonate elements. This is expected when examining the lithology of the units. The Gault Assemblage layers are the major host of phreatic carbonates at Gault.

Phreatic carbonates routinely form in situ as postburial phases and occur as surface coatings on grains and larger elements. If these had formed during the burial history and had reduced pore volume by their presence, then they may have added U to the source of dose rate to the quartz grains, changing the bulk dose rate over time. On the basis of earlier work of U concentrations in carbonate rocks and phreatic carbonates cited above, it would probably have added U to the pore space volume at a similar concentration as the surrounding host material. As this occurred over time, incremental increases in bulk dose rate would have occurred, and thus, the bulk dose rate over the burial history would have been lower than observed at the time of sampling. Thus, any effect from the crystallization of U-containing phreatic carbonate would make the ages older, in the same direction as any effects of Rn loss >50%.

The evidence from Area 15 at the Gault Site demonstrates the presence of a previously unknown projectile point technology in North America before ~16 ka. The physical and cultural stratigraphic evidence recovered from Area 15, as well as the associated OSL ages reported here and elsewhere (32), are consistent in showing a coherent sequence of the Gault Assemblage, Clovis, Late Paleoindian, Early Archaic, and Middle/Late Archaic occupations over an apparent span of more than 16,000 calendar years (Fig. 4). This sequence corresponds well with previous studies in Central Texas (42). The distinct technological differences between Clovis and Gault Assemblage, together with the stratigraphic separation between the cultural depositions, indicate a lack of continuity between the two complexes.

The Gault Assemblage at the Gault Site, specifically the projectile points, represents a regional manifestation within a number of possible contemporary patterns (Supplementary Materials). As evidence for the complexity in the early occupation of the Americas increases (1, 2), a more elaborate framework (9) for these early human occupations is required.


Supplementary material for this article is available at

Table S1. Sample OSL ages and dose rate data.

Table S2. Radon loss effects on each Gault OSL sample.

Table S3. Summary of the Gault Assemblage.

Table S4. Summary counts of the Gault biface assemblage.

Fig. S1. A representative decay curve (upper), growth curve (middle), and plot showing sensitivity changes through the SAR cycle (lower) for Gault 11-18.

Fig. S2. Equivalent dose (De) distributions for all samples displayed in a histogram (left) and radial plot (right).

Fig. S3. Location of the Area 15 excavation block and the excavation grid and profile.

Fig. S4. Relationship between the stratigraphy and cultural components in Area 15.

Fig. S5. Results of the geoarcheological analysis of the sediments in Area 15.

Fig. S6. Andice projectile point (left) with representative diagnostic notching flakes (right).

Fig. S7. Backplots of northing (blue) and easting (red) profiles showing the elevation of diagnostic Andice notching flakes and the cultural components discussed in the text.

Fig. S8. Limestone bedrock of Area 15, with three sets of flutes scoured into the limestone (discussed in the text).

Fig. S9. Pollen data from Boriack Bog and the NGRIP and GRIP ice core record as compared to stratigraphic units at Area 15.

Fig. S10. Gault Assemblage stone tool types and frequency (see table S2).

Fig. S11. Gault Assemblage projectile points.

Fig. S12. Principal components analysis of the Gault Assemblage stemmed projectile points and the Gower and Uvalde types.

Section S1. Area 15 stratigraphy

Section S2. Context of early dates in North America

Section S3. Gault Assemblage in Area 15

References (4385)

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: We would like to thank S. Howard who provided help in obtaining the samples and J. Martin and K. Reynolds who assisted with sample preparation. We thank Shannon Mahan who conducted the high-resolution gamma spectrometry. We would also like to thank B. Nash who assisted in producing fig. S7, M. Shoberg who conducted the use-wear analysis, and D. Madsen who provided insightful comments on an early draft of this article. We would also like to thank the three anonymous reviewers who provided helpful comments and feedback. Funding: Support for this research was provided, in part, by NSF grant 0920549 to Texas State University, San Marcos; the Gault School of Archaeological Research; the Natural Sciences and Engineering Research Council of Canada to McMaster University, Ontario; and other private contributions. Author contributions: M.B.C. conceived the project and directed the fieldwork and analysis and together with T.J.W. wrote most of the manuscript. K.R., W.J.R., and A.K.-Z. conducted the OSL analyses and wrote the relevant sections. N.V. and E.R.P. analyzed the lithic assemblage and produced Fig. 3. C.D.F. and A.G. analyzed the site stratigraphy with the assistance of S.J.A. who analyzed the Bell-Andice component. Competing interests: The authors declare that they have no competing interests. Data and materials availability: The Gault Site archeological materials are stored with the Prehistory Research Project at Texas State University, and the dates presented here are available in the Supplementary Materials. All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

Stay Connected to Science Advances

Navigate This Article