The earliest reference to North American corner-tang knives appeared in the late nineteenth century in Thomas Wilson’s (Reference Wilson1899) volume titled Arrowpoints, Spearheads, and Knives of Prehistoric Times. The asymmetric shape of the corner-tang knife was described as a “lopsided form” or “peculiarity” (Wilson Reference Wilson1899:941–942), with the suggestion that it may have been hafted for use as a scraping or cutting implement. Between the 1910s and 1930s, further attention was drawn to their distinctive morphology and unusual characteristics by Moorehead (Reference Moorehead1910), Brown (Reference Brown1926), Hough (Reference Hough1927), and Renaud (Reference Renaud1932). However, it was not until Patterson’s (Reference Patterson1936, Reference Patterson1937) comprehensive synthesis and classification that the corner-tang knife was established as a unique and somewhat rare oddity that became widely distributed across the Plains from its origin in central Texas. Since then, dozens of researchers have reported on the utilitarian function or ceremonial and trade value of corner-tang knives (e.g., Belz Reference Belz1964; Black and McGraw Reference Black and McGraw1985; Broehm and Lovata Reference Broehm and Lovata2004; Davis Reference Davis1975; Fischel Reference Fischel1938, Reference Fischel1939; Kraft Reference Kraft1993; Loendorf Reference Loendorf1964, Reference Loendorf1974; Marshall Reference Marshall1963, Reference Marshall1972; McCormick Reference McCormick, Bonnichsen and Oliver1981; McReynolds and Chandler Reference McReynolds and Chandler1984; Quigg Reference Quigg2011; Renaud Reference Renaud1942; Stephenson Reference Stephenson1944). However, they remain an enigmatic “biface” within the context of Great Plains prehistory. Corner-tang knives, in their various manifestations, have been identified from southern Texas, east through Mississippi, Missouri, and Illinois, to as far north and west as Montana.
As previously speculated by Reeves (Reference Reeves1969), corner-tang knives could be reasonably expected to occur in the Northern Plains of Canada. To that point, there are several known corner-tang knives from southern Saskatchewan (Orchard Reference Orchard1942:55, 76). One was reported by Mr. Wilson of Caron, and two were reported by Mr. Gunn of Pense. One diagonal variant was also reported in the Lew Dewalt Collection, recovered near McCord (Robsinson Reference Robsinson1945, Reference Robsinson1948). Nathaniel Wowchuck (personal communication 2026) identified a mid-back corner-tang knife from Oxbow in Jack Trusty’s private collection. There is also a large backed variant reported from the northeastern extent of the Great Plains near Melita, Manitoba (Robsinson Reference Robsinson1948). As with the majority of corner-tang knives from the United States, the foregoing specimens from Saskatchewan and Manitoba were recovered on the surface with uncertain provenience; none have been reported from excavated contexts in Canada until now.
In the summer of 2020, a Historic Resources Impact Mitigation excavation was conducted by Circle CRM Group Inc. at the multicomponent site of DjOn-8 (de Guzman et al. Reference de Guzman, McKay, Peck, Belanger and Wright2020). It is a large campsite situated along the southwestern edge of Elkwater Lake in the Cypress Hills Interprovincial Park in southeastern Alberta, Canada (Figure 1A–C). The campsite is attributed to the Middle (7500–1500 BP) to Late (1350–250 BP) prehistoric periods based on the recovery of several diagnostic projectile points, with a projected age range of approximately 3600–250 BP. A broken corner-tang knife was excavated at the site and likely coincides with the Pelican Lake (3600–2800 BP) or Besant (2100–1500 BP) occupation. Here, we present multiscalar evidence on the first corner-tang knife reported from Alberta, together with its archaeological context and complementary obsidian sourcing results. We then discuss the broader implications of this tool type for understanding the movement of precontact peoples, goods, and ideas across the North American Great Plains.
Study area: (A) subdivision of the North American Great Plains; (B) Northern Plains, Cypress Hills Interprovincial Park, and probabilistically identified obsidian sources for artifacts from Locality 5 (Obsidian Cliff, Wyoming; Bear Gulch, Idaho); (C) DjOn-8 overview. Photo Credit: Esri World Imagery and Esri Topographic. (Color online)
Figure 1 Long description
A) A thematic regional map of North America showing the Great Plains area. Labels include Canada and United States. The Great Plains is shown as a distinct region, with internal labels Northern Plains, Central Plains and Southern Plains. B) A thematic locator map labeled Cypress Hills Interprovincial Park. The map area is labeled Northern Plains. Two point markers are labeled Obsidian Cliff and Bear Gulch. C) A satellite style overview map labeled DjOn-8 and Elkwater Lake. A north arrow is shown at the lower right. A scale bar labeled 0, 50 and 100 km appears near the lower right, alongside a representative fraction label 1:100,000.
Context
Corner-Tang Knives
Early descriptions of corner-tang knives identified them as corner-notched bifaces with a tang opposite the cutting edge, which were used as hafted implements (Hough Reference Hough1927). Patterson’s (Reference Patterson1936, Reference Patterson1937) synthesis established six variants of the form based on the placement of the tang in relation to the blade, including the base, diagonal, back, mid-back, bifurcated, and two-tang knife (Figure 2A–F). Our compilation of available data demonstrates that all six corner-tang variants are found across 18 US states, and 843 (or 849, including the five from Saskatchewan and one from Manitoba, in Canada) can be attributed to a specific locality or county. There are four outliers (i.e., Mississippi, South Carolina, Pennsylvania, and Oregon) that are well removed from the main areas of concentration across the Great Plains (Figure 2G; Supplementary Table 1; Adams Reference Adams1992; Barnier Reference Barnier1971; Beckes and Keyser Reference Beckes and Keyser1983; Belz Reference Belz1964; Benedict and Contreras Reference Benedict and Contreras2009; Boshart Reference Boshart2018; Broehm and Lovata Reference Broehm and Lovata2004; Campbell Reference Campbell, Fox and Hester1976; Creel Reference Creel1989; Davis Reference Davis1975; Feyhl Reference Feyhl1972; Fischel Reference Fischel1938, Reference Fischel1939; Foster Reference Foster1961; Hagen Reference Hagen1962; Hall Reference Hall1981; Hoffman Reference Hoffman1961; Jelinek Reference Jelinek1960; Kelley Reference Kelley1959; Kraft Reference Kraft1993; Loendorf Reference Loendorf1964, Reference Loendorf1974; Marshall Reference Marshall1963; McCormick Reference McCormick, Bonnichsen and Oliver1981; McReynolds and Chandler Reference McReynolds and Chandler1984; Mitchell and Orchard Reference Mitchell and Orchard1984; Moore Reference Moore2012; Mulloy Reference Mulloy1958; Patterson and Hudgins Reference Patterson and Hudgins1983; Pearson and Krefting Reference Pearson, Krefting, Bonnichsen and Young1979; Perttula et al. Reference Perttula, Bruwelheide and Owsley2020; Quigg Reference Quigg2011; Renaud Reference Renaud1932, Reference Renaud1942; Russell Reference Russell1974; Stephenson Reference Stephenson1944; Taylor Reference Taylor2006; Thompson Reference Thompson1949; Wood Reference Wood1971; Wyatt Reference Wyatt2002). Most specimens, however, are located in central and southern Texas, as well as in Nebraska and Wyoming, with a smaller cluster near the Oklahoma panhandle.
Corner-tang knife types (after Patterson Reference Patterson1936, Reference Patterson1937; Fischel Reference Fischel1938): (A) base (Type 1); (B) diagonal (Type 2); (C) back (Type 3); (D) mid-back (Type 4); (E) bifurcated (Type 5); (F) two-tang (Type 6); (G) distribution of corner-tang knives by county in North America, probabilistically identified obsidian sources for artifacts from Locality 5 (Obsidian Cliff, Wyoming; Bear Gulch, Idaho), and precontact trade trails across North America (after Oetelaar Reference Oetelaar2021). (Color online)
Figure 2 Long description
The map of North America illustrates the distribution of corner-tang knives, obsidian sources and precontact trade trails. The density of corner-tang knives is marked with purple shading, concentrated in central and southern Texas, Nebraska and Wyoming, with a smaller cluster near the Oklahoma panhandle. Four outliers are noted in Mississippi, South Carolina, Pennsylvania and Oregon. Obsidian sources are identified at Obsidian Cliff, Wyoming and Bear Gulch, Idaho, marked with triangles. Trade trails are depicted across the map. Inset illustrations show six corner-tang knife types labeled A to F: base, diagonal, back, mid-back, bifurcated and two-tang. The map includes a scale bar and a legend indicating density per 1,000,000. The map is oriented with a north arrow and includes county boundaries.
Corner-tang knives are believed to have originated around the Balcones Escarpment and Edwards Plateau in central Texas (Kraft Reference Kraft1993; Patterson Reference Patterson1936). They have been found in association with Marcos Points, a Late and Transitional Archaic period corner-notched point type dating to 3000–1000 BP (Mitchell et al. Reference Mitchell, Chandler and Kelly1984; Turner and Hester Reference Turner and Hester1985). They have also been found alongside Castroville, Ensor, Frio, Edgewood, Marshall, and Montell projectile points (ca. 4000–1400 BP; Saner and Tomka Reference Saner and Steve1998; Waters et al. Reference Waters, Keene, Prewitt, Everett, Laughlin and Stafford2021). Fairland and Kent points have also been found embedded in skeletal remains associated with the Group 2 burials at the Ernest Witte Site (41AU36), where a young female was interred with two well-crafted corner-tang knives. Bone sampled from the Group 2 burials yielded dates of 2400 to 1600 BP (Hall Reference Hall1981). Of the two types, Kent points often appear earlier in the record and are also associated with Middle Archaic occupations in the region (Kraft Reference Kraft1993; Turner and Hester Reference Turner and Hester1985).
The Texas knives are generally manufactured from local high-quality materials such as Edwards chert (Kraft Reference Kraft1993). Interestingly, all corner-tang knives recovered outside Texas were similarly manufactured from locally sourced materials and are therefore not considered to be direct physical trade items (Kraft Reference Kraft1993; John D. Reynolds, Notes on File at the Kansas State Historical Society, 14 CF 343, TWH#100, 1986). Rather, it is suggested that the form itself spread north and west either as an idea—through trade ties, by emulation, integration (as groups interacted with one another on a seasonal basis)—or by diffusion through other forms of group interactions such as family exchanges, communal hunting, and mass migrations (Davis Reference Davis1975; Hall Reference Hall1981; Kraft Reference Kraft1993; Spielmann Reference Spielmann1983). Many corner-tang knives from the Central and Southern Plains are associated with open campsites, rockshelters, and refuse/cooking middens, whereas several well-crafted specimens have also been found as grave goods (Gooding Reference Gooding1981; Hofman Reference Hofman, Hofman, Brooks, Hays, Owsley, Jantz, Marks and Manhein1989; Kraft Reference Kraft1993; Mitchell et al. Reference Mitchell, Chandler and Kelly1984). As a case in point, five corner-tang knives and several Marcos points were found with an isolated burial at the Rudy Haiduk Site (41KA23; Mitchell et al. Reference Mitchell, Chandler and Kelly1984). Another five corner-tang knives were also found alongside two individuals at the Silo Site (41KA102) in south Texas, which led Broehm and Lovata (Reference Broehm and Lovata2004) to surmise that these artifacts were nonutilitarian items based on their ornate nature and long, thin, crescent-shaped forms that would have likely broken upon use. For this reason, corner-tang knives may also be understood to reflect socioeconomic status or stratification during the Late Archaic.
Unfortunately, most corner-tang knives have not been recovered from reliably dated, excavated contexts. The majority have little to no contextual data given that they have been found in plowed fields and now lie within private collections (Kraft Reference Kraft1993; Patterson Reference Patterson1936, Reference Patterson1937), or they were only anecdotally mentioned in older reports. What has been extrapolated from these data suggests that most corner-tang knives are associated with secondary butchery/campsites centered on bison hunting and subsistence (Kraft Reference Kraft1994; Patterson Reference Patterson1936). Not many corner-tang knives have been recovered from Late Archaic camp/processing sites (Quigg Reference Quigg1997, Reference Quigg1998).
There are also several “outliers” that have been identified along the foothills of the Rockies or at Woodland sites to the east. For instance, a fragmentary corner-tang knife was recovered at Vail Pass Camp (5ST85) in Colorado at an elevation of 3,224 m, which represents one of the highest-altitude precontact sites in North America (Gooding Reference Gooding1981). The knife was recovered in association with materials dated to the Late Plains Archaic and Plains Woodland periods. Another notable corner-tang knife was found at Crystalsin Cave (24CB85) in Montana near Yellowstone National Park alongside evidence for sheep procurement (Beckes and Keyser Reference Beckes and Keyser1983; McCormick Reference McCormick, Bonnichsen and Oliver1981)—a specialized subsistence practice within an area otherwise dominated by bison hunting.
Several variants of the corner-tang knife have also been identified beyond the Plains proper. One diagonal corner-tang specimen manufactured from a local opalitic chalcedony was found along the Columbia River in Oregon (Stephenson Reference Stephenson1944). Other notable specimens include one from a plowed field in Davidson County, North Carolina (Adams Reference Adams1992), and a base variant recovered in Berks County, Pennsylvania (Fischel Reference Fischel1938). The hierarchical complexity of corner-tang knives would suggest ties to groups on the Central and Southern Plains, although it is possible that similar forms developed independently through technological convergence. However, radiocarbon dates for corner-tang knives north of Texas postdate those around the Balcones Escarpment and Edwards Plateau and support the idea that this technology spread rapidly northward from Texas (Kraft Reference Kraft1993). Corner-tang knives from sites along the eastern margins of the Plains are dated to approximately 2000–1500 BP (Kraft Reference Kraft1993). Sites in Montana dated from the later part of the Archaic into the early Late period (ca. 2000–500 BP; Malouf Reference Malouf1959) have yielded variants of corner-notched points associated with the Pelican Lake complex (Davis Reference Davis1975). The exact mechanisms by which corner-tang technology spread are yet to be identified, but it is reasonable to assume that traditional trails may have facilitated their movement across the Great Plains (Oetelaar Reference Oetelaar2021).
Among the northernmost distribution of corner-tang knives recovered within the United States, 18 are located in Montana, distributed across 10 counties (i.e., Prairie, Dawson, Carbon, Big Horn, Yellowstone, Glacier, Gallatin, Golden Valley, Beaverhead, and Flathead; Supplementary Table 1). As previously hypothesized by Reeves (Reference Reeves1969), the corner-tang knife could have diffused north into Canada by mobile bison-hunting groups along the foothills of the Rocky Mountains, through seasonal movement adjacent to or along interconnected traditional trails (Oetelaar Reference Oetelaar2021). There is evidence of the corner-tang knife in Mission Valley in the vicinity of McDonald Lake (Barnier Reference Barnier1971), where groups may have moved onto the Northern Plains via the foothills through what is now Waterton-Glacier International Peace Park. However, there are several other forms identified within the central and eastern portion of Montana (Feyhl Reference Feyhl1972; Hough Reference Hough1927; Patterson Reference Patterson1937). These areas lie immediately south and east of the Cypress Hills, an area that is known to have served as a regular settlement or resource stop on the Northern Plains until the 1870s (Bonnichsen and Baldwin Reference Bonnichsen and Baldwin1978; Oetelaar Reference Oetelaar2021).
Cypress Hills, DjOn-8, and the Corner-Tang Knife
The Cypress Hills extend from southwestern Saskatchewan to southeastern Alberta and form a flat-topped erosional outlier subdivided into three physiographic units, including the west, center, and east blocks (Figure 1B; Bonnichsen and Baldwin Reference Bonnichsen and Baldwin1978).
The lowermost exposed strata consist of Upper Cretaceous (Bearpaw, Eastend, Whitemud, Battle, Frenchman Formations) to Paleocene (Ravenscrag Formation) deposits, including sandstone, siltstone, shale, coal, and tuff (Furnival Reference Furnival1950; Kupsch and Vonhof Reference Kupsch and Vonhof1967). These are overlaid by Eocene to Miocene conglomerates, gravels, and sandstones associated with the Cypress Hills Formation. The former predominantly include quartzite, chert, and other fine-grained lithologies (Gryba Reference Gryba, Quigg and Byrne1976; Westgate Reference Westgate1965). Elevated segments of the plateau that escaped glaciation by the Laurentide ice sheet retain the broad, level surface typical of a peneplain (Westgate Reference Westgate1965). The surrounding topography was shaped extensively by glacial processes, meltwater, and mass wasting events, which resulted in the formation of small basins such as Elkwater Lake.
DjOn-8 is located along the southwestern margin of Elkwater Lake within the west block of the Cypress Hills, Alberta (Figure 1B–C). The site is extensive, covering approximately 119 ha and extending 1 km to the west and 200 m to the south of the lake’s current shoreline. Elkwater Lake is estimated to have formed around 9440 BP based on paleoenvironmental reconstructions (Wiseman et al. Reference Wiseman, Running and Freeman2002).
DjOn-8 was originally identified in 1966 when various lithic tools, bison bone fragments, and one pot sherd were discovered eroding from an alluvial slope near the lake (Bonnichsen Reference Bonnichsen1966). One sample of burned soil (GaK-1273) with a recorded depth of 22 cm was subsequently radiocarbon dated to 1260 ± 100 BP (Martindale et al. Reference Martindale, Morlan, Betts, Blake, Gajewski, Chaput, Mason and Vermeersch2016). Additional materials were recovered in 1979, from both surface and subsurface contexts scattered across a large area along the southwestern extent of the lake (Quigg Reference Quigg1980). Salvage operations conducted in 1980 revealed that the site comprised three distinct areas—namely, the Upper Terrace, the Lower Terrace, and the Shoreline Area (Brumley et al. Reference Brumley, Dau, Greene, Heikkila, Quigg, Rushworth and Saylor1981). Here, an extensive array of material was recovered, including concentrations of fire-broken rock (FBR), bifaces, unifaces, gravers, perforators, expedient tools, and debitage manufactured from a variety of raw materials. In addition, ash and burned sediments were suggestive of shallow hearth-like features. No suitable organic materials were found for radiocarbon dating. However, diagnostic Pelican Lake, Besant, and Washita (variants of Plains side-notched points; Joyes Reference Joyes1973; Justice Reference Justice2002; Peck Reference Peck2011) projectile points were recovered, suggesting repeated occupations during the Middle to Late prehistoric periods (ca. 3600–250 BP). This is also demonstrated at the multicomponent Stampede Site (DjOn-26), on the eastern side of Elkwater Lake, with the earliest occupation recorded thus far dated to as early as 8000 BP (Gerald A. Oetelaar, personal communication 2024).
Investigations in 1981 determined that much of the site remained intact despite encroachment from campground development (Fedirchuk et al. Reference Fedirchuk, Heitzmann and Newton1982). Additional inspections at DjOn-8 were conducted in 2004 and 2008 (Gillespie and Mirau Reference Gillespie and Mirau2009). In summarizing previous observations, Robertson (Reference Robertson2006) noted that the site is situated atop a large colluvial deposit that was subsequently buried by shallow deposits, with no depositional events registered between occupations, resulting in a palimpsest accumulation. In 2018, additional work was conducted along a series of trails southeast of the Firerock Campground. Wickham (Reference Wickham2018) recovered 99 pieces of lithic debris, including five tools, 120 pieces of FBR, 68 faunal remains, and several historic artifacts from eight localities.
In July 2020, Circle CRM Group Inc. conducted subsurface testing at Wickham’s (Reference Wickham2018) eight localities, followed by excavations (de Guzman et al. Reference de Guzman, McKay, Peck, Belanger and Wright2020). The initial shovel-testing program was conducted along or immediately adjacent to several pathways at 4–5 m intervals. Briefly, the subsurface tests and subsequent excavation at Locality 1 led to the recovery of 49 pieces of lithic debitage and 57 pieces of FBR. Four subsurface tests were conducted at Locality 2, eight were conducted at Locality 5, and 24 were conducted at Locality 8. At Locality 8, only two subsurface tests yielded artifacts: one platform-bearing flake and one piece of shatter. The subsequent 1 × 2 m excavation block at Locality 8 led to the recovery of only a secondary decortication flake and one core.
Due to poor artifact returns during the mitigation stage at Localities 2 and 8, the focus shifted to Locality 5 (de Guzman et al. Reference de Guzman, McKay, Peck, Belanger and Wright2020). Here, the subsurface tests and excavations yielded a large assortment of material. Six of the eight subsurface tests were positive for artifacts—specifically, 33 pieces of FBR, 39 pieces of debitage, eight unidentifiable faunal remains, and one large scapula (Table 1). Consequently, a 31 m2 excavation block was opened (Figure 3a). The matrix consisted of loose, unconsolidated duff followed by dark gray-brown silty loam extending to a depth of 16 cm below the surface, overlying a mottled gray-brown and yellow silty clay down to 20 cm below the surface, followed by a basal deposit of compact yellow clay down to a maximum depth of approximately 30 cm below the surface (Figure 3b).
Total Artifacts by Level, Count, and Weight from Locality 5.
Table 1 Long description
Counts and weights in grams are listed by level for fire-broken rocks, lithics, fauna, and the overall total. Level 1 records 393 fire-broken rocks weighing 6,634.34 grams, 105 lithics weighing 196.50 grams, and 20 fauna weighing 554.32 grams, for a total of 523 artifacts weighing 7,435.50 grams. Level 3 records 1,475 fire-broken rocks weighing 46,827.20 grams, 561 lithics weighing 2,483.36 grams, and 82 fauna weighing 106.41 grams, for a total of 2,123 artifacts weighing 49,418.42 grams. The row labeled, n/a lists 98 fire-broken rocks weighing 2,136.52 grams, 45 lithics weighing 996.31 grams, and 9 fauna weighing 16.72 grams, for a total of 152 artifacts weighing 3,149.55 grams. The Total row sums 6,490 fire-broken rocks weighing 164,115.28 grams, 2,595 lithics weighing 9,408.66 grams, and 293 fauna weighing 1,595.85 grams, for an overall total of 9,387 artifacts weighing 175,119.79 grams. Across all levels, fire-broken rocks have the largest counts and weights compared with lithics and fauna.
a n/a = unrecorded, from a shovel test or wall clean-up.
DjOn-8 Locality 5: (A) piece plotted artifacts at Locality 5; (B) western wall profile of excavation unit 506N 499E at Locality 5, where the corner-tang knife was found; (C) deer pelvic fragment with three cut marks from Locality 5. (Color online)
Figure 3 Long description
A plotted map of artifacts at Locality 5, with a legend indicating different types of artifacts such as debitage, faunal remains and tools, marked with various symbols and colors. The map includes a grid with coordinates along the bottom and left sides. The image B displays a cross-section of the western wall profile of excavation unit 506N 499E at Locality 5. It illustrates layers of duff, silty loam, silty clay and clay, with a labeled position for the corner-tang knife. A scale bar is present at the bottom, marked in centimeters. The image C shows a deer pelvic fragment with three cut marks, accompanied by a scale bar below it, also marked in centimeters.
Based on the horizontal distribution of materials, several artifact concentrations—or discrete activity areas—were identified at Locality 5. One is located in the north, and a second is near the south-southwest corner (Figure 3a). Most artifacts were recovered within the gray-brown silty clay situated between the bottom of Level 2 and the top of Level 3, with the majority concentrated in Level 2 (10–20 cm below surface; Table 1).
The FBR from Locality 5 consists of variably sized quartzite, granite, gneiss, and sandstone. Many of the FBR fragments are heavily crenellated (4,286 of 6,490), indicating rapid cooling, possibly from immersion in water for cooking or boiling. The remaining materials display reddening, cracking, or planar spalling (1,323 of 6,490) from thermal stress, suggestive of open-fire cooking. One hearth-like feature was identified containing a high quantity of FBR from units 500N 502E through 500N 504E, with blackened sediment concentrated in unit 500N 501/502E (Figure 3a). The blackened sediment measured a maximum of 20 cm in horizontal space and extended between 16 cm and 25 cm below surface (de Guzman et al. Reference de Guzman, McKay, Peck, Belanger and Wright2020:Figures 30–32). There is a strong likelihood that this feature represents a hearth, but definitive evidence is lacking, which mirrors previous observations at DjOn-8 (see Brumley et al. Reference Brumley, Dau, Greene, Heikkila, Quigg, Rushworth and Saylor1981).
Of the 293 pieces of faunal material recovered from Locality 5, the majority comprised fragmentary long bone segments or small unidentifiable remains, suggesting a high level of processing. As much as 57% of the faunal material was burned or completely calcined (Table 2). The majority of the material was identified to the level of Mammalia—most likely large ungulates, due to the density and thickness of the cortical bone—which are most likely associated with bison or cervid (Table 3). The large intact scapula identified during the shovel-testing program is attributed to Bos, and one pelvic fragment is attributed to the deer family. The latter is the only specimen from the faunal assemblage to exhibit cut marks on the outer surface (Figure 3C). All three are parallel and have a V-shaped kerf, with the middle one being depressed at the bottom, suggesting that considerable force was applied during butchery. One of the bone modifications is shallow and was likely induced with a more acute angle than the other two. Overall, this would suggest that the bone was sliced or hacked during initial dismemberment or after cooking (Reitz and Wing Reference Reitz and Wing2008). The pelvic fragment was excavated from unit 501N 500E, approximately 5 m southwest of the corner-tang knife and at a similar depth (Figure 3A).
Taphonomy of Faunal Remains from Locality 5.
Table 2 Long description
Counts of faunal remains are listed by level for burned, calcined, cut marks, none, and total. Level 1 has burned 3, calcined 3, cut marks shown as a blank entry, none 14, total 20. Level 2 has burned 22, calcined 88, cut marks 1, none 71, total 182. Level 3 has burned 28, calcined 25, cut marks shown as a blank entry, none 29, total 82. Row labeled n slash a has burned shown as a blank entry, calcined shown as a blank entry, cut marks shown as a blank entry, none 9, total 9. Totals row lists burned 53, calcined 116, cut marks 1, none 123, total 293.
a n/a = unrecorded, from a shovel test or wall clean-up.
Species and Representative Faunal Elements from Locality 5.
Table 3 Long description
Columns list element, NISP, weight in grams, and MNE split into left, right, axial, and indeterminate. Bos includes scapula with NISP 9, weight 719.92 g, and MNE right 1, plus humerus with NISP 1, weight 244.66 g, and MNE left 1. Cervid has one indeterminate entry with NISP 1 and weight 3.95 g, with no MNE values recorded. Mammalia (Ungulata) includes tooth with NISP 13, weight 8.06 g, and MNE axial 13; lumbar vertebrae with NISP 1, weight 23.88 g, and MNE axial 1; and indeterminate vertebrae with NISP 13, weight 13.81 g, and MNE axial 13. Additional Mammalia (ungulata) entries are rib with NISP 1, weight 27.15 g, and MNE indeterminate 1; cuneiform with NISP 1, weight 17.73 g, and MNE right 1; radius with NISP 2, weight 167.12 g, and MNE left 1 and right 1; radial carpal with NISP 1, weight 22.96 g, and MNE right 1; ulnar carpal with NISP 1, weight 23.07 g, and MNE right 1; scaphoid with NISP 1, weight 21.76 g, and MNE right 1; tibia with NISP 2, weight 30.93 g, and MNE right 2; sesamoid with NISP 2, weight 28.53 g, and MNE indeterminate 2; first phalanx with NISP 1, weight 23.90 g, and MNE indeterminate 1; indeterminate long bone with NISP 16, weight 87.24 g, and MNE indeterminate 16; and indeterminate scrap with NISP 227, weight 131.18 g, and MNE indeterminate 227. The total row reports NISP 293, weight 1,595.85 g, MNE left 2, MNE right 8, MNE axial 27, and MNE indeterminate 247.
Note: Two specimens were radiocarbon dated (scapula, Beta-566370; humerus, Beta-566371).
Bison, elk, and deer were known to have resided and been hunted historically within the Cypress Hills (Bonnichsen and Baldwin Reference Bonnichsen and Baldwin1978), and they have been previously identified at DjOn-8 (Bonnichsen Reference Bonnichsen1966), as well as to the east of Elkwater Lake at the Stampede Site (DjOn-26; Gerald A. Oetelaar, personal communication 2024). The Cypress Hills are considered to have been a refugia from the surrounding prairies, and they would have readily offered access to water, wood, sheltering topography. They may have even served as a bison wintering ground (Bonnichsen and Baldwin Reference Bonnichsen and Baldwin1978; Hildebrandt and Hubner Reference Hildebrandt and Hubner2007).
A total of 2,595 pieces of lithic material was recovered from Locality 5. Approximately 97% of the material consists of variably sized debitage (platform-bearing flakes, decortication flakes, flake fragments, and shatter) struck from quartzite, various cherts, chalcedony, petrified wood, siltstone, argillite, and probably some Knife River Flint (KRF; Table 4). In addition, 19 pieces of obsidian were recovered from Locality 5, including nine platform-bearing flakes, nine flake fragments, and one piece of block shatter. All obsidian artifacts were excavated from Levels 2 (n = 15) and 3 (n = 4) across 11 units. These include 499N 503E (n = 1), 499N 506E (n = 1), 501N 500E (n = 1),
Locality 5 Lithic Debitage by Count, Category, and Raw Material.
Table 4 Long description
Counts are listed by lithology across categories labeled PDF, SDF, CRF, PBF, BRF, BF, RRF, FF, and BS, with a Total column. Column totals are PDF 13, SDF 43, CRF 1, PBF 795, BRF 6, BF 1, RRF 25, FF 993, BS 634, and overall Total 2,510. Chert totals 503, with PDF 4, SDF 8, PBF 198, BRF 2, RRF 18, FF 193, and BS 80. Chalcedony totals 133, with CRF 1, PBF 59, RRF 2, FF 57, and BS 14. KRF Brown Chalcedony totals 107, with PDF 1, SDF 3, PBF 40, BRF 3, BF 1, RRF 3, FF 49, and BS 7. Argillite totals 71, with PBF 26, FF 40, and BS 5. Petrified Wood totals 49, with SDF 2, PBF 12, FF 16, and BS 19. Obsidian totals 19, with PBF 9, FF 9, and BS 1; Quartz totals 21, with PBF 7, FF 7, and BS 7. Granite totals 3, with PBF 1 and FF 2; Mudshale totals 1 with FF 1; Sandstone totals 12 with FF 5 and BS 7; Siltstone totals 12 with PBF 3, FF 8, and BS 1.
a PDF = primary decortication flake.
b SDF = secondary decortication flake.
c FF = flake fragment.
d PBF = platform bearing flake.
e BRF = bifacial reduction flake.
f RRF = retouch/resharpening flake.
g BF = bipolar flake.
h CRF = core rejuvenation flake.
i BS = block shatter.
501N 501E (n = 3), 501N 502E (n = 1), 502N 500E (n = 1), 503N 499E (n = 1), 507N 499E (n = 5), 507N 500E (n = 3), and 508N 500E (n = 2; Figure 3A). Therefore, it may be suggested that the majority of obsidian pieces are penecontemporary with the corner-tang knife, which was excavated from the base of Level 2 (Figure 3A). There are no known geologic obsidian sources on the Northern Plains, given that outcrops are typically found within or west of the Rocky Mountains;—for example, in British Columbia and south through the western United States (Allan et al. Reference Allan, Kristensen, Wowchuk, Wowchuk, Armstrong and Allan2025). Therefore, obsidian is considered an exotic material and would have necessitated the development of long-distance exchange networks or the movement of people for acquisition.
Nine cores were recovered from Locality 5, seven of which were made on quartzite, whereas the other two were manufactured from argillite and siltstone. A total of 76 tools were recovered from Locality 5—including 30 expedient tools, 24 scrapers, 14 projectile points, seven bifaces, and one graver—the majority of which were manufactured from chert and probably some Knife River Flint. However, similar brown chalcedony is known to occur within the South Saskatchewan River Valley, the Souris gravels of Manitoba, several areas in Montana (e.g., Smith River quarries), and the White River outcrop in South Dakota (Steuber Reference Steuber2018; Thomson Reference Thomson1994). Without microscopic or geochemical analyses, it is not possible to reconstruct exchange patterns involving Knife River Flint from sources in North Dakota’s Dunn and Mercer Counties (Kirchmeir Reference Kirchmeir2011).
Especially noteworthy is that a broken “point base” identified as a corner-tang knife was recovered from unit 506N 499E at a depth of 20 cm below surface near the northern extent of the excavation block (Figure 3A). The same unit also yielded 19 platform-bearing flakes, 12 flake fragments, five chunks of block shatter, and 71 pieces of FBR from depths ranging from 0 to 30 cm below surface. All but five artifacts from this unit were excavated from Levels 2 and 3. The artifact distribution across multiple levels suggests vertical mixing, which is also observed in other units (Table 1), supporting the interpretation that the assemblage represents a palimpsest accumulation (Robertson Reference Robertson2006).
Only two of the 14 projectile points from Locality 5 are complete, and five others retain diagnostic elements despite being fragmented. Two specimens are similar in form to Plains Triangular points (Figure 4A–B), which are associated with the Old Women’s Phase (1100–250 BP) in Alberta (Bubel et al. Reference Bubel, McMurchy and Lloyd2012) and dated between 450 and 250 BP in Saskatchewan (Boser Reference Boser2022). Both points were excavated from Level 2 less than 2 m apart (i.e., 500N 500E, 500N 502E; Figure 3A). One point is attributed to the Besant Phase (2100–1500 BP) and was excavated from the top of Level 3 in unit 505N 500E (Figures 3A and 4C; see also Bonnichsen Reference Bonnichsen1966; Martindale et al. Reference Martindale, Morlan, Betts, Blake, Gajewski, Chaput, Mason and Vermeersch2016). Two other points were excavated from Levels 2 and 3 in units 500N 503E and 499N 503E, respectively (Figures 3A and 4D–E). Both may be associated with the Country Hills Complex, which is an Early prehistoric point type (Peck Reference Peck2011). However, a Pelican Lake (3600–2800 BP) assignment cannot be ruled out based on the points’ techno-typological characteristics (see Peck Reference Peck2011) and is therefore favored considering the projected age range of the site.
Five diagnostic projectile points from Locality 5: (A–B) dorsal (left) and ventral (right) sides of Plains Triangular points; (C) dorsal (left) and ventral (right) sides of Besant point; (D–E) dorsal (left) and ventral (right) sides of Pelican Lake points. (Color online)
Two samples were submitted for radiocarbon dating in an attempt to refine the site’s age. One sample was taken from the scapula recovered from unit 515N 505E (Level 2), and another was taken from the humerus recovered from unit 502N 501E (Level 1). Radiocarbon analysis of the scapula returned an age of 80 ± 30 BP (Beta-566370) calibrated at the 2σ confidence level to 1810–1926 cal AD (70.5% probability; 140–24 cal BP) and 1690–1730 cal AD (24.9% probability; 260–220 cal BP). Radiocarbon analysis of the humerus returned a similar age of 90 ± 30 BP (Beta-566371) calibrated at the 2σ confidence level to 1807–1928 cal AD (69.2% probability; 143–22 cal BP) and 1684–1732 cal AD (26.2% probability; 266–218 cal BP). Given the shallow depths at which archaeological materials were recovered and the recent dates obtained on the bone samples, the latter are considered intrusive.
Overall, Locality 5 is interpreted as a campsite associated with animal processing that likely incorporated a refuse area. The horizontal artifact distribution indicates a minimum of two activity areas, and the minimal sedimentation between occupation episodes indicates repeated use of the area, resulting in a palimpsest accumulation spanning the Middle to Late prehistoric periods (Robertson Reference Robertson2006).
Materials and Methods
Lithic Analysis
The corner-tang knife was measured following Davis (Reference Davis1975:129), and it was weighed, analyzed, and classified following Patterson (Reference Patterson1936). Measurements were taken using a Fowler 54-101-150 digital caliper and recorded in mm to two decimal places, with the average reported here. The raw material from which the specimen was made was identified by comparison with a reference collection encompassing all three major rock types—including 21 subtypes from 85 regional sources—that are commonly encountered at archaeological sites in British Columbia, Alberta, Saskatchewan, and Manitoba. The identification of raw material attributes and the classification of variably translucent cherts as chalcedony follow Luedtke (Reference Luedtke1992).
Use Wear
The corner-tang knife was subject to use-wear analysis to characterize and interpret surface abrasions resulting from friction with other materials. Independent examination was completed as a blind test by one of us and Shawn Bubel at the University of Lethbridge. The specimen was first dry-brushed, then examined using an Olympus SZ61 stereomicroscope with a magnification range of 2.5–4.5× and photographed with a mobile device mounted to the lens. Surface modifications were interpreted in terms of kinetic motion and the properties of the worked material (Dubreuil and Savage Reference Dubreuil and Savage2014). Definitions of microchipping, polishing, and rounding/smoothing follow Vaughan (Reference Vaughan2017). The location and type of surface modifications were used to reasonably exclude possible natural, confounding sources of use-wear formation.
Protein Residue
The corner-tang knife was subject to protein residue analysis using crossover immunoelectrophoresis (Newman Reference Newman1990; Newman and Julig Reference Newman and Julig1989; Newman et al. Reference Newman, Ceri and Kooyman1996, Reference Newman, Yohe, Kooyman and Ceri1997; Royal Canadian Mounted Police 1983; but see Hendy Reference Hendy2021) following use-wear analysis. The analysis was conducted by Atlatl Archaeology Ltd. Our analytical sequence may have impacted residue preservation and increased the risk of contamination. Our results should therefore be interpreted with caution. To minimize additional contamination risks, the specimen was handled with gloves and placed in a sterile weigh boat. Dilute ammonia (5%) was pipetted onto the cutting edge, and the specimen was sonicated for five minutes. The specimen was turned over and allowed to soak in the same solution for 20 minutes, then sonicated for an additional five minutes. The solution was pipetted into a sterile plastic tube and frozen.
The residue was tested against commercial antisera, including bovine (MP/Cappel Biomedical), deer (MP/Cappel Biomedical), rabbit (MP/Cappel Biomedical), and duck (Nordic-MUbio). Agarose gels were made by dissolving 1 g of agarose in 110 mL of barbital buffer at a boil. Gel bond was prepared by adhering the hydrophobic side to a glass plate and snipping the top left corner to maintain orientation. The solution (12–15 mL) was spread onto gel bond film with a syringe, allowed to cool for five minutes, then covered in stretch wrap and placed in a refrigerator overnight. Paired wells were cut into each gel, and plugs were removed using a glass pipette attached to a vacuum line. Frozen antisera and residue extracts were thawed for 30 minutes. Antisera samples (5 µL) were pipetted into the left well, and residue samples (5 µL) were pipetted into the right well. The seventh row of each gel was used for positive controls, whereas the eighth row was used for negative controls.
The gels were placed in the central platform of an electrophoresis tank, with equal levels of barbital buffer on each side. Double-thick strips of filter paper were wet with barbital buffer and attached to each side of the gel to ensure that the other end of the paper was in contact with the barbital buffer. The electrophoresis tank was run for 45 minutes at 130V. The gels were removed from the tank, numbered on their undersides, and pressed for 10 minutes covered with a double layer of filter paper and a glass plate. The gels were placed in a container of 1 M saline overnight to remove excess residue and antisera. The gels were then placed in a container of deionized water for 60 minutes to remove the saline and then oven-dried at 90ºF for approximately 60 minutes. The gels were stained with Coomassie R250 blue, then de-stained in baths of acetic acid, alcohol, and water before examination for antigen precipitates.
Bulk Composition
Corner-Tang Knife
The corner-tang knife and one geological lithic sample from the Hand Hills were characterized twice in a nondestructive manner to obtain representative bulk compositions using an Olympus Delta Premium portable X-ray fluorescence (pXRF) spectrometer (4W Rh X-ray tube; 4 µm prolene windows; large-area silicon drift detector). The characterization was conducted by XRF Solutions Ltd. Material from the Hand Hills was selected for comparison due to its prehistoric use and proximity to the Cypress Hills (Kirchmeir Reference Kirchmeir2011; Steuber Reference Steuber2018). The aim of the analysis was to confirm the macroscopic lithological identification and rule out the Hand Hills as a possible raw material source, with the understanding that thorough regional sampling and multiscalar analyses are generally required to source siliceous rock types probabilistically (see Favreau Reference Favreau2023:6–9). X-ray intensities collected under air were converted to concentration estimates (major oxides = wt%; trace elements = ppm) using a silicon (Si) polynomial calibration based on the analysis of 29 reference standards (NIST 2710A, BCR 1, BCR 2, USGS CS, USGS OS, AGV 1, USGS G2, SiO2, KAOL, CALCITE, dolomite, DOLOMITE, DTS 2B, COQ 1, NOD A1, NOD P1, CLB 1, PYRITE, SBC 1, SGR 1B, RGM 2, GSP 2, SDC 1, QLO 1A, STM 2, W 2A, BHVO 2, BIR 1A, DNC 1A). Concentration estimates were measured using two modes (8, 45 keV) for a total count time of 60 seconds. The targeted elements included 15 major oxides reported to two decimal places (Na₂O, MgO, Al₂O₃, SiO₂, P₂O₅, SO₃, K₂O, CaO, TiO₂, MnO, Fe₂O₃, Cr₂O₃, CoO, PbO, V₂O₅, ZnO) and 17 trace elements reported as whole numbers (Cl, Ni, Cu, Ga, As, Se, Rb, Sr, Y, Zr, Nb, Mo, Ba, Pb, Bi, Th, U). Nondetected element concentrations are reported as below detection limit.
Obsidian
A total of 17 obsidian artifacts from DjOn-8 was also characterized in a nondestructive manner to obtain trace element concentrations using an Olympus Vanta C-Series pXRF spectrometer at the Royal Alberta Museum’s Archaeology Laboratory. This particular pXRF spectrometer was selected because it has been used in multiple obsidian sourcing studies (e.g., Allan et al. Reference Allan, Kristensen, Wowchuk, Wowchuk, Armstrong and Allan2025; Frahm Reference Frahm2019; McMillan et al. Reference McMillan, Amini and Weis2019), has been tested for accuracy and precision (Johnson et al. Reference Johnson, Ferguson, Freund, Drake and Duke2021), and is widely accessible and user friendly. The analysis was carried out using the GeoChem2 mode with factory settings (i.e., no user factors applied) and the Fundamental Parameters calibration (McMillan et al. Reference McMillan, Amini and Weis2019). The GeoChem2 mode varies the current and voltage of the 4-W X-Ray tube (with Rh anode) in combination with two built-in beam filters to improve fluorescence of both light and heavy elements during a single analytical run. Spectra were collected for 30 seconds on each beam for a total of 60 seconds. Each artifact was subject to three analyses, totaling 180 seconds of testing time, with the mean used for comparison. No drift was observed during the analysis. No post hoc calibration or parameter adjustments were deemed necessary. The NIST 2709a and NIST 2711a reference standards were tested during the source sampling, and periodic runs were performed with the Olympus calibration disk. Concentration estimates were within the standards’ validated ranges. We emphasize the reproducibility of our results through analyses of reference checks and the availability of artifact data. Two obsidian artifacts from DjOn-8 were determined to be too small for pXRF analysis and were therefore not considered for probabilistic sourcing.
Artifact and source data were entered into the SourceXplorer (version 1.01) online statistical tool (McMillan et al. Reference McMillan, Waber, Ritchie and Frahm2022). SourceXplorer uses linear discriminant analysis (LDA) to establish associations between geological sources and artifacts of unknown provenance. Ratio-variable plots of Nb/Y over Zr/Rb were also used for comparative analysis. Following Wilson and Pollard’s (Reference Wilson, Pollard, Brothwell and Pollard2001) provenance hypothesis, the intention was to eliminate all possible sources for each artifact until a potential source could not be eliminated. Obsidian artifacts from Alberta have been previously documented to originate from various North American geological sources (Allan Reference Allan2024; Kristensen et al. Reference Kristensen, Allan, John, Ives, Yanicki and Rasic2023). For this reason, artifact data were compared to Bear Gulch (Allan et al. Reference Allan, Kristensen, Wowchuk, Wowchuk, Armstrong and Allan2025); Obsidian Cliff (Davis et al. Reference Davis, Aaberg, Schmitt and Johnson1995); Big Southern Butte, Browns Bench, Timber Butte, and Malad in Idaho (Pink Reference Pink2023; Craig E. Skinner, personal communication 2020); Glass Buttes, Newberry Volcano, and Obsidian Cliffs in Oregon (Craig E. Skinner, personal communication 2020); and Wiki Peak, Batza Tena, and Suemez Island in Alaska (Schmuck et al. Reference Schmuck, Carlson, Reuther, Baichtal, Butler, Carlson and Rasic2022; Jeffrey T. Rasic, personal communication 2023); as well as Mount Edziza and Anahim Peak in British Columbia (James et al. Reference James, Bailey and D’Auria1996; Reimer Reference Reimer2015). After initial screening, the 17 obsidian artifacts from DjOn-8 were further compared with 633 assays on materials from five obsidian sources. These include Batza Tena (n = 40), Bear Gulch (n = 259), Browns Bench (n = 205), Obsidian Cliff (n = 89), and Timber Butte (n = 40).
Results
Lithic Analysis
Although technically a biface, the specimen is the basal end of a notched implement categorized as a diagonal corner-tang knife, or Patterson’s (Reference Patterson1936:10) Type 2 (Figure 5). The artifact weighs 2.73 g and is broken across the transverse plane through the blade, distal to the tang, leaving the entirety of the basal edge or heel intact. The specimen was manufactured from a smooth, cryptocrystalline, semilustrous, waxy, white, partially opaque to translucent chalcedony (Luedtke Reference Luedtke1992). The width of the blade from the notch on the back to the cutting edge is 26.83 mm, and the length from the heel to the break is 17.55 mm.
Corner-tang knife from Locality 5: (A) biconvex cross section; (B) dorsal surface; (C) ventral surface; (D) dorsal surface metrics; (E) diagonal corner-tang knife attributes. (Color online)
The blade exhibits a thin (5.30 mm) biconvex cross section when examined along the break (Figure 5A). From the mid-body, it tapers evenly toward the margins of the back (tang side) and cutting edge, as well as longitudinally (distally) through to the margin of the heel. The heel is slightly curved out from the tang. At the bottom of the heel, opposite the tang, the basal edge transitions onto the cutting edge as a rounded corner, giving the specimen a somewhat squared shape. The cutting edge is slightly curved and appears to be constricting toward the distal tip. Both faces display well-patterned thinning- or pressure-flake scars that extend across the body, indicating careful refinement. The heel and cutting edge are thin and exhibit careful edge retouch, creating a nonsinuous—or straight —margin.
The hafting area or “tang” is well formed by the placement of two U-shaped notches that have created an expanding stem neck, with a small, detailed, convex basal stem that is slightly skewed toward the notch on the backside (Figure 5B–C). The tang is 7.68 mm wide and 2.45 mm thick, and the neck measures 6.64 mm. The notches are well crafted, U-shaped, shallow, and somewhat narrow. The notch on the heel side has a breadth of 2.72 mm, with a notch depth of 2.90 mm. The notch on the back portion (top) has a breadth of 2.67 mm and a notch depth of 3.05 mm. The notch basal juncture on the heel side is right angled and slightly round to pointed, whereas the notch juncture along the back side (top) is broken away. The specimen’s overall shape is similar to corner-tang knives found in Texas (Kraft Reference Kraft1993:Figure 6C) and is considered the most abundant type across the Great Plains.
Use-wear and protein residue on the corner-tang knife from Locality 5: (A) tang; (B) heel; (C) blade; (D) bovine antigen precipitate; (E) deer antigen precipitate. (Color online)
Use Wear
The proximal portion of the specimen is heavily ground over the complete length through the notches, around the stem, and along the heel, terminating at the juncture of the cutting edge, leaving the edge rounded or blunted (Figure 6A). Many of the pressure flaking scars along the tang are obliterated. They are smooth and round along the tang’s edge and slightly polished, suggesting that the implement was prepared for hafting or lashing materials (Quigg Reference Quigg2011; Vaughan Reference Vaughan2017; John D. Reynolds, Notes on File at the Kansas State Historical Society, 14 CF 343, TWH#100, 1986).
The heel of the knife also exhibits use wear in the form of grinding that has blunted the edge, given that it is slightly polished smooth, with a well-rounded margin (Figure 6B). This suggests that the edge was deliberately dulled with a rough material (Vaughan Reference Vaughan2017), possibly to prevent inadvertent injuries. The heel may have also been blunted to fortify or stabilize the edge, given that it often does not serve as the cutting edge. Instead, it often functions as structural support for the back of the blade when heavier cutting is required to limit stress on the distal portion.
The blade displays moderate edge damage along its margin (Figure 6C). Small step scars or microchipping present near the margin on both sides of the blade have removed minute portions of the cutting edge. The edge itself is rough and slightly jagged yet still retains some level of sharpness. Edge damage is continuous along the length of the blade and ends at the juncture between the blade and heel. Overall, the use wear along the blade suggests that the tool was utilized as a slicing implement that likely involved pushing and pulling motions (Keeley and Toth Reference Keeley and Toth1981). This could explain why the juncture of the blade and heel are rounded—so as to not impede the backstroke during use (see Mika Reference Mika2022). Several corner-tang knives recovered from the Rudy Haiduk Site (41KA23) also displayed similar use-wear patterning, which Mitchell et alia (Reference Mitchell, Chandler and Kelly1984) suggested was the result of cutting and scraping soft materials such as meat.
Protein Residue
The blade of the corner-tang knife was subjected to protein residue analysis and tested against commercial antisera for bovine, deer, rabbit, and duck. Due to the potential for cross-reactions, faint antigen precipitates were observed at the family level for Bovidae (bison, cow, muskox) and Cervidae (deer, elk, moose, caribou), with a possibility for Antilocapridae (pronghorn; Figure 6D–E). Given the geographic location and its approximate time frame (i.e., Pelican Lake, Besant, and Washita), it would appear that the corner-tang knife was used to process bison, elk, mule deer, white-tailed deer, and/or pronghorn.
Bulk Composition
Corner-Tang Knife
Only three major oxides (Al2O3, SiO2, Fe2O3) and two trace elements (Bi, U) were found to be above the instrument’s detection limits, whereas the remaining analytes are reported as below detection limit (Table 5). The two major oxide sums of 99.90 wt% and 102.96 wt% for the corner-tang knife suggest that the measurements are slightly high, particularly when considering the probability that additional elements are likely to be present, albeit in low concentrations. The lithology of the corner-tang knife can be identified as chalcedony. This is based on the elevated SiO2 concentration and low concentration of trace elements, in combination with its macroscopic characteristics (i.e., texture, grain size, luster, color, translucency; Luedtke Reference Luedtke1992). The low Fe2O3 concentration (0.01–0.02 wt%) likely accounts for the specimen’s uniformly white color, level of translucency, and absence of banding—the latter of which is characteristic of agate.
pXRF Data for the Corner-Tang Knife from Locality 5 and One Geological Sample from the Hand Hills.
Table 5 Long description
PXRF results are listed for four measurements, two labeled Hand Hills and two labeled DjOn minus 8:3394, with major oxides reported as weight percent and trace elements as parts per million. For major oxides, Na2O, MgO, P2O5, SO3, K2O, CaO, TiO2, MnO, Cr2O3, CoO, PbO, V2O5, and ZnO are below detection limit in every row. Al2O3 is 0.27 and 0.00 for the two Hand Hills rows, and 0.55 and 0.67 for the two DjOn minus 8:3394 rows. SiO2 is 95.77 and 90.95 for Hand Hills, and 102.40 and 99.21 for DjOn minus 8:3394. Fe2O3 is 0.01 and 0.03 for Hand Hills, and 0.01 and 0.02 for DjOn minus 8:3394; Total is 96.06 and 90.99 for Hand Hills, and 102.96 and 99.90 for DjOn minus 8:3394. For trace elements, Cl, Ni, Cu, Ga, As, Se, and Ba are below detection limit in all rows; Rb, Sr, Y, Zr, Nb, and Mo are below detection limit in three rows, but in the second Hand Hills row Rb is 3, Sr is 2, Y is 4, Zr is 3, Nb is 12, and Mo is below detection limit. Pb is below detection limit in all rows, while Bi is 10 and 8 for the two DjOn minus 8:3394 rows and below detection limit for both Hand Hills rows. Th is below detection limit in three rows, but is 12 in the second Hand Hills row. U is 2 and 3 for the two Hand Hills rows, and 6 and 5 for the two DjOn minus 8:3394 rows; nondetected values are reported as below detection limit.
Notes: Major oxides = wt%; trace elements = ppm; nondetected element concentrations are reported as below detection limit (bdl).
In the absence of a comprehensive reference collection for the region, the corner-tang knife cannot be sourced probabilistically with any degree of certainty. However, it is reasonable to conclude that the raw material does not originate from the Hand Hills, based on the differences in Al2O3, SiO2, Bi, and U concentrations (Table 5). This includes the macroscopic differences—most notably, the brown color of Hand Hills material (Steuber Reference Steuber2018). However, an exotic source is not envisaged due to the ubiquitous presence of similarly white chalcedonic raw materials in the region (Johnson Reference Johnson1998), some of which would have likely been redeposited by the encroachment of the Laurentide ice sheet during the last glaciation (Steuber Reference Steuber2018; Westgate Reference Westgate1965).
Obsidian
Each obsidian artifact was subjected to three pXRF assays. The relative standard deviations for each analyte were first calculated to assess the degree of precision (Table 6; Supplementary Table 2). The average relative standard deviations for Rb (1.52%), Sr (3.36%), Y (2.97%), Zr (1.11%), and Nb (4.01%) indicate high precision and reproducibility. Therefore, mean concentration estimates for the 17 obsidian artifacts were subsequently compared with 633 assays on materials from five obsidian sources, including Batza Tena, Bear Gulch, Browns Bench, Obsidian Cliff, and Timber Butte. Source separation was achieved using a bivariate plot of Rb over Zr (Figure 7A). The LDA model returned a hit ratio of 98.15%, with only one sample from Bear Gulch misclassified as originating from Timber Butte. The first two discriminant functions of the
Mean pXRF Concentration Estimates and Relative Standard Deviations (RSD) for Obsidian Artifacts from Locality 5.
Table 6 Long description
Seventeen obsidian artifact samples list mean pXRF concentrations for Rb, Sr, Y, Zr, and Nb, each followed by an RSD column, plus a Source column. Samples labeled Bear Gulch report Sr from 41.00 to 48.00, with Sr RSD values from 0.00 to 3.23. Samples labeled Obsidian Cliff report Sr from 0.00 to 3.00; Sr RSD is shown as a dash for Sr values of 0.00, and is 0.00 for Sr values of 3.00, while sample DjOn minus 8:5700 has Sr 2.50 with Sr RSD 28.28. Bear Gulch Rb ranges from 175.00 to 204.67, Y from 43.00 to 47.00, Zr from 301.00 to 321.33, and Nb from 66.67 to 71.67. Obsidian Cliff Rb ranges from 261.33 to 301.33, Y from 83.67 to 89.33, Zr from 175.00 to 180.67, and Nb from 53.33 to 60.00. Across all rows, RSD values are generally under 8 for Rb, Y, Zr, and Nb, except for the Sr RSD of 28.28 in DjOn minus 8:5700.
Note: n = 3; trace elements = ppm.
Source attribution of obsidian artifacts from Locality 5: (A) Zr over Rb (ppm) with 95% confidence ellipses for five sources; (B) linear discriminant analysis with 95% confidence ellipses for five sources. (Color online)
Figure 7 Long description
A scatter plot with Zr on the vertical axis ranging from 160 to 320 and Rb on the horizontal axis ranging from 140 to 320. The plot includes data points grouped into clusters representing different sources: Batza Tena, Bear Gulch, Browns Bench, Obsidian Cliff and Timber Butte. Each cluster is enclosed within a confidence ellipse, indicating the distribution of data points for each source. The clusters are visually distinct, with Browns Bench having the highest concentration of data points, while Timber Butte has the lowest. The image B shows a scatter plot of linear discriminant analysis with Axis 1 on the horizontal axis ranging from negative 20 to positive 20 and Axis 2 on the vertical axis ranging from negative 20 to positive 20. The plot includes data points grouped into clusters representing the same sources as in image A. The clusters are enclosed within confidence ellipses, indicating the separation achieved by the discriminant analysis. Browns Bench and Bear Gulch have distinct clusters, while Batza Tena and Timber Butte are closely positioned. Obsidian Cliff is separated from the other clusters, indicating a distinct source classification.
LDA model have an approximate trace contribution of 28.8% and 9.4%, respectively. The third discriminant function accounts for approximately 6.4% of the variance. Zr, Rb, and Y have strong to moderate positive loadings on the first axis, whereas Y has a strong negative loading on the second axis (Supplementary Table 3).
Based on the LDA model, we can predict, with over 98% accuracy, the probable sources of the 17 obsidian artifacts from DjOn-8 (Figure 7B). A total of 11 artifacts are attributed to Bear Gulch (Idaho), which is located approximately 587 km away in geodesic geometry. The six other artifacts are attributed to Obsidian Cliff (Wyoming), which has an approximate geodesic site-to-source distance of 539 km. The reported distances are likely underestimates of the total transport distances because they do not account for human or geographic factors.
Discussion
Function
Given that corner-tang knives became widely distributed across the Great Plains, it is assumed that such implements gained traction among Plains groups for processing activities related to bison hunting (see Kraft Reference Kraft1993). Apart from well-crafted specimens found as grave goods (e.g., Broehm and Lovata Reference Broehm and Lovata2004; Hall Reference Hall1981), most knives have been recovered from open campsites and oven middens, which suggests that corner-tang knives functioned as utilitarian tools (Kraft Reference Kraft1993; Patterson Reference Patterson1936). According to previous use-wear studies summarized by Kraft (Reference Kraft1993) and more recent ones conducted by Saner and Tomka (Reference Saner and Steve1998) and Quigg (Reference Quigg2011), corner-tang knives (and unifacial variants) were likely used for cutting soft materials such as meat and vegetation over prolonged periods. For instance, Quigg (Reference Quigg2011) determined that hide cutting may have been conducted with the single corner-tang implement at the Late Archaic Pipeline site (41PT185/C) near Amarillo, Texas. Given that 24 scrapers were excavated from Locality 5 at DjOn-8, it is possible that the corner-tang knife was similarly used for hide-processing activities.
Given that only the blade and tip are broken away (Figure 5B–C), it appears that the DjOn-8 specimen was not utilized for heavy-duty activities in the same fashion as the broken corner-tang recovered alongside butchered sheep remains from Crystalsin Cave (24CB85; Beckes and Keyser Reference Beckes and Keyser1983; McCormick Reference McCormick, Bonnichsen and Oliver1981; Pearson and Krefting Reference Pearson, Krefting, Bonnichsen and Young1979). Instead, the results of our use-wear analysis suggest that the corner-tang knife was used as a slicing implement that may have been used to process large ungulates such as bison, elk, mule deer, white-tailed deer, and/or pronghorn (Figure 6D–E). This is corroborated by the fact that bison (and possibly other large ungulates) and deer elements were found at Locality 5 (Tables 2 and 3; Figure 3C). However, we cannot rule out the possibility that, upon breaking, the tool was discarded and came into contact with faunal remains, leading to a false positive for protein residue. Given that neither the tang nor the heel were tested for antigen precipitates, it is also unclear whether the identified proteins on the blade are the byproduct of function or hafting, because the latter would have likely involved materials such as sinew. Finally, it must be acknowledged that the antigen results should not be interpreted as definitive because the analytical tests were undertaken only after use-wear analysis.
Hafting
Patterson (Reference Patterson1936), Bleed (Reference Bleed1986), and McReynolds and Chandler (Reference McReynolds and Chandler1984) have suggested that corner-tang knives, in their various manifestations, were likely altered for optimal use. Based on the fact that the DjOn-8 specimen exhibits a blunted and slightly polished diagonal tang (Figure 6A), it was likely hafted (see McReynolds and Chandler Reference McReynolds and Chandler1984; John D. Reynolds, Notes on File at the Kansas State Historical Society, 14 CF 343, TWH#100, 1986). On the question of breakage, bifaces or projectile points typically fracture as a result of extreme bending, compression (impact) or radial forces. When a significant bending force is applied to an implement set into a shaft, it often fails near the hafting area (Taylor Reference Taylor2006). Quigg (Reference Quigg2011) suggests that the hafting limit for a corner-tang knife would have been confined to the area immediately distal to the neck. If an extreme bending force were applied to the blade, a flexed snap would most likely occur where the blade is not supported. It may be that the DjOn-8 specimen broke in such a way, which is corroborated by the absence of radial fractures commonly associated with trampling damage (Taylor Reference Taylor2006).
As summarized by Kraft (Reference Kraft1993), several hafting methods have been proposed for variants of the corner-tang knife. The most common hafting method identified for diagonal corner-tang knives is by using a small, straight handle positioned on the tang at a 45° angle. When hafted in this way, a user’s wrist would be flexed downward at a steep angle, thereby keeping the blade parallel to the material being cut (McReynolds and Chandler Reference McReynolds and Chandler1984). To overcome straining the wrist, the user would either have to stand or kneel over, which could lead to overexertion. Another proposed hafting method for diagonal corner-tang knives is to wrap raw hide or sinew around the tang and lash it around one’s wrist (Kraft Reference Kraft1993:Figure 12). The proposed benefits of using a corner-tang knife as a handheld implement include greater precision plus a reduced risk of misplacement, particularly during evisceration. If used in this way, a break caused by a bending force would most likely occur where the thumb, index, and middle fingers were holding the implement around the medial-to-distal portion.
Hand use of the DjOn-8 knife is considered unlikely considering the overall thinness and the apparent constricting cutting edge, which is estimated to have extended to a maximum length of approximately 5 cm before breaking. Yet another, more plausible hafting strategy may have involved a curved handle fashioned from wood, bone, or antler. Such a design would reduce wrist flexion, minimize fatigue, and allow the user to more comfortably maintain the cutting edge parallel to the material being cut. Although the breakage pattern and use wear suggest that the specimen from DjOn-8 may have been hafted, experimentation is required to gain further insight.
Trade and Movement on the Great Plains
The distribution of corner-tang knives across the Great Plains provides valuable insights into the complex and evolving interactions among groups over time. These interactions would have facilitated the movement of both utilitarian and prestige materials, along with associated ideas, across considerable distances. Exchange systems in this context should not be viewed as purely economic; rather, they are embedded within broader social, political, and ceremonial frameworks that would have reinforced alliances, group identities, and status differentiation—as emphasized by Brown (Reference Brown and Beck1995) and Struever and Houart (Reference Struever, Houart and Wilmsen1972).
One may point to various lines of evidence indicating that contact was common between groups on the Plains and surrounding regions. For instance, the presence of exotic grave goods at Texas burial sites and corner-tang knives east of the Plains suggests the existence of reciprocal exchange networks (Broehm and Lovata Reference Broehm and Lovata2004; Mitchell et al. Reference Mitchell, Chandler and Kelly1984). Interestingly, corner-tang knives from Woodland Village sites in Kansas were commonly manufactured from local Flint Hills chert (Kraft Reference Kraft1993). This may indicate the presence of migrating groups from Texas or the movement of individuals acting as intermediaries in “down-the-line” exchange systems. In either scenario, corner-tang knives could have been transported as part of existing tool kits only to be reproduced using locally available raw materials. Though not mutually exclusive, the northward adoption of corner-tang knives may also be associated with the convergence of traditions on the Northern and Central Plains during the Late Woodland period, where and when the use of pottery, the adoption of horticultural practices, and the increased circulation of trade goods became particularly prevalent (Reeves Reference Reeves1983).
Evidence of westward connections are equally noteworthy. The previously noted corner-tang knife from Oregon lies near a well-established travel corridor that extended through Montana, Idaho, and Washington, thereby indicating links between Plains and coastal groups (Figure 2G). This could help to explain the presence of marine shell at the Highwood Site in Alberta (EePk-272; Brink and Baldwin Reference Brink and Baldwin1988). In contrast, a marine-shell bead from the Stampede Site (DjOn-26), east of Elkwater Lake, is thought to have originated from the Gulf of Mexico (Falzarano Reference Falzarano2014:Figure 4.3d; Gerald A. Oetelaar, personal communication 2024). Such findings underscore the extensive reach of precontact exchange networks.
Likewise, the Old Copper complex in western Canada points to connections with the Great Lakes region (Steinbring Reference Steinbring1975). Copper artifacts represent one of the most definitive indicators of long-distance exchange, which began in the Late Archaic period and continued through the Woodland period (Martin Reference Martin1999). Initially valued for utilitarian purposes, copper later became an important prestige material in mortuary contexts. Its wide geographic dispersion suggests the presence of well-established trade routes, sustained demand, and shared cultural values (Brown Reference Brown and Beck1995; Martin Reference Martin1999). The clearest connection of this phenomenon on the Northern Plains is the Pelican Lake–aged burial in southern Alberta noted above, where copper, exotic toolstones, and marine shell were interred as grave goods (Brink and Baldwin Reference Brink and Baldwin1988). Across the provincial border in southwestern Saskatchewan, another well-known site is the Bracken Cairn (DhOb-3) burial, involving five individuals radiocarbon dated to 2465 ± 86 BP (King Reference King1961; Pendree Reference Pendree1980; Walker Reference Walker1982). The human remains were found to be deliberately coated with red ochre and interred with a number of grave goods, including a Pelican Lake point, one copper bead, and pieces of freshwater clam shell. These findings mirror Woodland burial sites and those in Texas (e.g., Hall Reference Hall1981; Lovata Reference Lovata1997; Patterson Reference Patterson2000; Patterson et al. Reference Patterson, Hudgins, Kindall, McClure, Marek, Nuckols and Gregg1998). Consequently, the circulation of exotic goods across vast distances likely contributed to the formation of material identities shaped by sustained intergroup interactions via long-established trade trails (Figure 2G; Oetelaar Reference Oetelaar2021).
Subsistence practices in the form of communal bison hunting may have functioned similarly to strengthen social structures on the Plains, which would have increased food production and may have served as a strategy to resist the territorial expansion of horticultural groups to the east (Walde Reference Walde2006). Evidence for large-scale communal hunting extends back to the Early Archaic period (Bamforth Reference Bamforth2011; Reeves Reference Reeves, Davis and Reeves1990). These developments likely led to rising population pressures, such that the corner-tang knife may have been adopted, duplicated, or emulated as groups maintained social and political ties and/or increasingly engaged with one another in tandem with seasonally migrating herds of bison. The spread of the corner-tang knife is therefore argued to reflect overlapping interaction spheres, likely facilitated by trail systems to reinforce trade and sociopolitical alliances, probably originating from central Texas.
In Canada, the Pelican Lake complex represents an important Late Archaic tradition characterized by distinctive projectile points, bison-centered subsistence practices, and a wide geographic distribution (Peck Reference Peck2011). Lithic sourcing studies indicate that Pelican Lake groups transported materials such as Bear Gulch obsidian over considerable distances (e.g., Allan et al. Reference Allan, Kristensen, Wowchuk, Wowchuk, Armstrong and Allan2025). This pattern suggests embedded procurement, in which materials circulated through repeated social interactions rather than formalized trade systems. At the same time, the distribution of Bear Gulch obsidian aligns with known travel corridors and exchange centers associated with the Missouri River and, more distantly, the Ohio River system (Allan et al. Reference Allan, Kristensen, Wowchuk, Wowchuk, Armstrong and Allan2025; Oetelaar Reference Oetelaar2021). Obsidian sourced from the western United States has also been identified in eastern North American contexts (Hughes Reference Hughes, Charles and Buikstra2006), including Hopewellian assemblages, which highlight the scale and complexity of exchange networks during this period. Such distributions support a “down-the-line” exchange model, in which goods move incrementally between communities rather than through direct acquisition (Struever and Houart Reference Struever, Houart and Wilmsen1972). These patterns imply strong interregional connectivity and possibly shared ideological frameworks that facilitated intermediary exchange (Seeman Reference Seeman1979). Given that high-value materials such as obsidian may have been exchanged for subsistence goods or as part of gift-giving practices, the concept or design of the corner-tang knife may likewise have spread through these interactions, whether as a functional tool or a novel form.
In Alberta, sites attributed to the Pelican Lake complex and Besant Phase correspond with the increased use of obsidian, likely transported through direct acquisition as groups gathered for large-scale communal bison hunts (Allan et al. Reference Allan, Kristensen, Wowchuk, Wowchuk, Armstrong and Allan2025). At sites such as the Yellowstone Bank Cache (24YE355), located near Obsidian Cliff, obsidian was readily available and likely stockpiled before groups moved northward onto the Plains (MacDonald et al. Reference MacDonald, Maas and Hardes2010). These localities, much like DjOn-8 and DjOn-26, may have served as key nodes of interaction among groups circulating along established travel corridors through Wyoming and Montana into southern Canada.
The expansive distribution of corner-tang knives across the Great Plains into the Cypress Hills of Canada, the eastern seaboard of the continental United States, and possibly as far south as Mexico (e.g., Muñoz Espinosa and Castañeda Reyes Reference Espinosa, María and Carlos Castañeda Reyes2017:Figure 10) underscores the form’s versatility while also alluding to the degree of interaction among precontact groups in all directions (see also Hester et al. Reference Hester, Asaro, Stross, Michel, Kerr and Headrick1991). That said, corner-tang knives remained exceedingly rare on the Northern Plains given that their utilitarian function could have been fulfilled by several simpler implements. Their broad distribution nevertheless suggests that the movement of peoples, goods, and ideas across the Great Plains was likely facilitated by intimate geographic knowledge (e.g., Ac ko mok ki’s map [Fidler Reference Fidler1801]), natural travel corridors, and long-established trade trails (Figure 2G; Oetelaar Reference Oetelaar2021).
Conclusions
Similarities in material culture are generally understood to reflect one or a combination of the following: intergroup interaction, diffusion, imitation or emulation, and/or technological convergence. Despite its broad geographic distribution, the corner-tang knife has occupied a somewhat ambiguous position in the Plains archaeological record since its initial identification between the late nineteenth and early twentieth centuries. By reporting on the first corner-tang knife recovered from an excavated context in Alberta, Canada, this study reframes the corner-tang knife from a unique yet widespread artifact type to a material marker of exchange and interaction across the Great Plains.
The knife in question is a diagonal variant found at the multicomponent site of DjOn-8 in the Cypress Hills Interprovincial Park (de Guzman et al. Reference de Guzman, McKay, Peck, Belanger and Wright2020). Based on the occurrence of several diagnostic projectile points, the site was likely occupied during the Middle (7500–1500 BP) to Late (1350–250 BP) prehistoric periods. The knife likely coincides with the site’s Pelican Lake (3600–2800 BP) or Besant (2100–1500 BP) occupation (see Bonnichsen Reference Bonnichsen1966; Martindale et al. Reference Martindale, Morlan, Betts, Blake, Gajewski, Chaput, Mason and Vermeersch2016). Use-wear analysis indicates that the knife was hafted and used for cutting tasks, whereas protein antigen precipitates suggest it was used to process large ungulates. Associated obsidian artifacts are sourced to Bear Gulch (Idaho) and Obsidian Cliff (Wyoming), more than 500 km away. The identification of a corner-tang knife together with the sourcing results establish a direct link between the site’s occupants and groups to the south. When considered alongside previously reported surface finds from Saskatchewan and Manitoba, the specimen from DjOn-8 suggests that groups on the Northern Plains engaged—to some degree—with far-reaching trade and interaction networks. Overall, the spread of the corner-tang knife across the Great Plains is argued to represent the by-product of overlapping interaction spheres probably originating in central Texas, which would have been enabled by intimate geographic knowledge, natural travel corridors, and long-established trade trails. Our results not only confirm a long-standing hypothesis by extending the geographic range of corner-tang knives northward into Alberta but also highlight the potential for consulting archaeologists to meaningfully contribute to archaeological discourse. Future investigations will be required to clarify the nature of continent-wide group interactions on the Northern Plains as it relates to the Pelican Lake/Besant vis-à-vis obsidian circulation as well as the spatiotemporal distribution of corner-tang variants across the Great Plains, particularly in Canada.
Acknowledgments
We acknowledge the traditional and ancestral territories of the Blackfoot Confederacy, the Tsuut’ina Nation, the Stoney Nakoda Nation, and the historical Métis homeland. We gratefully acknowledge Margarita de Guzman, who served as the permit holder for fieldwork at DjOn-8. Special thanks to the excavation team, including, but not limited to, Alexandra Burchill, Brooke Gerard, Scott Graham, Kristin McKay, Tyler Statz, and Shannon Wright. Thanks to XRF Solutions Ltd. and Pine Environmental Services LLC for facilitating the pXRF analyses, Tatyanna Ewald from Atlatl Archaeology Ltd. for conducting the protein residue analysis, and Shawn Bubel for helping with the use-wear analysis. Many thanks to Craig E. Skinner and Jeffrey T. Rasic for kindly sharing unpublished obsidian data. We wish to formally thank Cindy Scheer and Fynley Calder-Rasmussen from the Royal Saskatchewan Museum, along with Nathaniel Wowchuk, for helping us identify corner-tang knives in Canada. We extend our appreciation to James D. Keyser and the Montana Historical Society for facilitating access to reports. We wish to formally recognize the Government of Alberta for allocating permits that made this work possible. Finally, we thank the anonymous reviewers for their suggestions and constructive comments on an earlier draft of this article.
Funding Statement
This research received no specific grant funding from any funding agency or commercial or not-for-profit sectors.
Data Availability Statement
All data are available in the manuscript and the supplementary material.
Competing Interests
The authors declare none.
Supplementary Material
The supplementary material for this article can be found at https://doi.org/10.1017/aaq.2026.10232.
Supplementary Table 1. Distribution of corner-tang knives by county in North America.
Supplementary Table 2. Raw pXRF data (trace elements = ppm) for obsidian artifacts from Locality 5.
Supplementary Table 3. Loadings of predictor variables for the LDA model.
