An Analysis of Chipped Stone Tools from the Middle Fort Ancient Wheeler Site

1

AN ANALYSIS OF CHIPPED STONE TOOLS FROM THE MIDDLE FORT ANCIENT WHEELER SITE IN MERCER COUNTY, KENTUCKY

By Sophia Jamaldin and Derek Reaux

Introduction In 2013 the University of Kentucky field school spent two weeks excavating at the Wheeler site in Mercer County, Kentucky. The initial impetus for conducting work at this site was to determine if it contained a palisade. Based on an examination of Fort Ancient villages with palisades, Martin Raymer had determined most were located along the periphery of the Fort Ancient Area. Based on this work, he concluded that there was a high probability that a palisade was present at the Wheeler site. While the field school did not find the evidence to support or refute this hypothesis, the presence of residential, trash disposal, and mortuary areas within the village were documented, and it was determined that the site contained a mound. The Wheeler site was first surveyed and documented in 1960 by John Carter when he was a University of Kentucky undergraduate Anthropology student. He excavated two 1 x 1 m units, where he was able to identify sub-plow zone features, a possible structure, and midden deposits. He also recovered diagnostic Fort Ancient ceramics and chipped stone tools. In 1985, Jim Railey, then with the Kentucky Heritage Council, nominated the site for listing in the National Registrar of Historic Places under the name Mercer Village. In 1998, Gwynn Henderson further analyzed the materials recovered from the initial Carter excavation. Based on the diagnostic artifacts recovered, Henderson was able to attribute the site to the early Middle Fort Ancient subperiod (Henderson 1998). Mercer Village was renamed the Wheeler site in 2013 after the current owner of the land who allowed the field school to excavate there. Radiocarbon dates obtained on charcoal recovered from the residential area and the mound from the 2013 field school excavation indicated that the site was occupied sometime between cal. A.D. 1206-1275, confirming Henderson’s assignment of the site to the early portion of the Middle Fort Ancient subperiod. “Fort Ancient” refers to the cultural tradition that inhabited the middle Ohio River valley, which includes parts of southern Indiana, southern Ohio, central and eastern Kentucky, and western West Virginia between A.D. 1000 to 1750. Fort Ancient peoples are characterized as small-scale, sedentary, non-hierarchical farming societies. Their villages appear to have been independent, autonomous settlements, and no evidence exists in the Fort Ancient region for the construction of large platform mounds or the establishment of a settlement hierarchy, which might imply a greater degree of regional political centralization (Pollack et al. 2002:206). During this same period, various chiefdom-level farming societies collectively known as “Mississippian” inhabited the major river valleys of western Kentucky. These chiefdoms are characterized by settlement hierarchies consisting of regional mound centers with associated smaller settlements. Fort Ancient and Mississippian interaction always figures prominently when researchers have considered external agents of Fort Ancient cultural change (Pollack et al. 2002:206). This notion especially pertains to the findings at the Wheeler site, where discovery of Mississippian style pottery and Mississippian style wall trench architecture was found in context with Fort Ancient material. The Wheeler site is located in Mercer County, Kentucky, along the southwestern edge of the Inner Bluegrass region of the state. It is situated along a floodplain of the Salt River, which directly borders the site on three sides. Based on geophysical survey and the excavations of the 2013 Field School, this site was a permanent village settlement with a mound that contains a stone pavement. It is important to note that directly across the Salt River exists another Fort Ancient site known as Dry Branch Creek, which was excavated in 1998 by Wilber Smith Associates (Pope et al. 2005).

2

During the field school, six excavation blocks were opened, labeled Trenches 1-6 (Figure 1). These trenches ranged from 1 x 2 m to 1 x 12 m. Trench 1 was a 1 x 12 m trench oriented north-south that was also expanded in the east and west. A 1 x 2 m and a 1 x 3 m crosscutting the trench east-west in order to reveal the extent of the stone cap. These trenches revealed the limestone cap measured 6 m on a side and was surrounded by a series of burials and features. Trench 2 (1 x 3 m) was placed west of the mound area and contained the edge of a possible midden feature. Trench 3 (1 x 10 m) and Trench 4 (1 x 5 m) were placed within the midden ring. Within these trenches portions of house basins and a range of features, including hearths, pits, postmolds, and burials, were documented. Trenches 5 and 6 also were placed within the midden ring. Trench 5 (1 x 4) contained multiple pit features and a portion of a wall trench that could represent another house basin. Though no notable features were documented in Trench 6 (1 x 2 m), it did yield a high density of diagnostic artifacts.

Figure 1. Trench Locations at the Wheeler Site.

The Wheeler site is intrinsically complex and excavations from the 2013 field school yielded a significant amount of features and artifacts. For the purpose of this report, the formal and informal chipped stone tool assemblages were analyzed as a part of a joint independent study by University of Kentucky undergraduate Anthropology majors Sophia Jamaldin and Derek Reaux. The initial research objectives and questions that were proposed for this project include:

Learning how to properly analyze chipped stone materials for archaeological research. Learning how to properly utilize data gathered from analysis into a formal report.

Do the chipped stone materials from the Wheeler Site exhibit evidence of Mississippian interaction, considering that some of the ceramics and wall trench architecture found at the site point to some level of Fort Ancient Mississippian interaction?

3

Is there any clear difference in the chipped stone tool artifact assemblage recovered from the mound area relative to the village area?

How do the chipped stone typologies fit into the known categories of Fort Ancient typologies?

Can we further connect the Dry Branch Creek site and the Wheeler site through their chipped stone lithic assemblages?

What chert sources did the residents of the Wheeler site utilize in their production of chipped stone tools?

After establishing these research goals, the samples that would yield the data to address our questions were selected. After pulling all chipped stone artifacts from the collection as well as viewing the artifact catalog sheets, it was determined that the entire assemblage of projectile points, bifaces, bifacial fragments, cores, and modified debitage, including polished flakes, or essentially the complete assemblage of chipped stone tools would be examined. The non-modified debitage was not analyzed due to time constraints. To expand the sample, bifaces and projectile points that were surface collected by John Carter in the 1960s were analyzed. The sample analyzed consisted of 986 formal and informal chipped stone tools (Table 1). Of these, 936 were recovered from excavated context and 50 from the surface. The analyzed sample consisted of 68 projectile points, 32 complete bifaces, 134 biface fragments, 695 modified flakes, three hoe flakes, and 54 cores. Additionally, 9852 pieces of debitage were recovered from excavation but were not analyzed for this report. Raw Material Procurement The proper identification of raw material resources is important because it provides understanding of the first stage of chipped stone tool production, which is raw material acquisition. Besides a single piece of debitage made of quartz, chert was the primary raw material source used for chipped stone tools at the Wheeler site. Chert, often referred to as flint, is an opaque rock composed of silica that occurs in nodules and massive beds within geological formations (Pope et al. 2005:5-3, 5-4). To identify chert types, variation among characteristics, such as color, grain, luster, and microfossil inclusions, were considered. Most raw material identifications were confirmed through the use of comparative collections from the William S. Webb Museum at the University of Kentucky, as well as advice from archaeologists who were more familiar with the local resource types in the region. This allowed for further insight into the practices of Fort Ancient knappers regarding the efforts they made to acquire this raw material, whether it was through local procurement or through long distance exchange. The presence of non-local types in the assemblage could be indicative of interaction with groups in other regions. Due to the proximate location of the site and accessibility of the curated collection at the William S. Webb Museum of Anthropology, the Dry Branch Creek site report was the primary source for this raw material analysis (Pope et al. 2005). The Kentucky Geological Survey’s digitalized map information service was utilized o determine what types of raw material were local to the region. A 50 km radius was used to define chert as locally available. From this we identified nine different possible local chert outcrops: Salvisa, Curdsville, Grier, Tyrone, Boyle, Muldraugh, Harrodsburg, St. Genevieve, and St. Louis.

4

Table 1. Chert Raw Material Data for Formal and Informal Tools.

Chert Type Projectile

Points Bifaces Biface

Fragments Modified

Flakes Cores Hoe

Flakes Total

N= % N= % N= % N= % N= % N= % N= % Salvisa 24 35.3 10 31.3 37 27.6 253 36.4 19 35.2 0 0.0 343 34.8 Curdsville/Grier 20 29.4 9 28.1 37 27.6 243 35.0 18 33.3 0 0.0 327 33.2 Tyrone 2 2.9 0 0.0 3 2.2 8 1.2 1 1.9 0 0.0 14 1.4 Muldraugh 5 7.4 4 12.5 9 6.7 24 3.5 3 5.6 1 33.3 46 4.7 Boyle 9 13.2 6 18.8 13 9.7 47 6.8 7 13.0 2 66.7 84 8.5 Ste. Gen/St.Louis 4 5.9 1 3.1 2 1.5 17 2.4 0 0.0 0 0.0 24 2.4 Unidentified 4 5.9 2 6.3 33 24.6 103 14.8 6 11.1 0 0.0 148 15.0 Total 68 100.0 32 100.0 134 100.0 695 100.0 54 100.0 3 100.0 986 100.0

Raw Material Descriptions: Salvisa is a Lexington Limestone chert that has a texture varying from medium to fine-grained and a moderate to dull luster. Colors are variable and include mottled grays, brown and yellow, with medium to light mottled grays the most common. Some samples contain light blue white corals, calcite, or other microfossils in a dark or light matrix. Subangular cobbles of Salvisa chert up to 13 cm in length are known to occur in gravels in Dry Branch Creek in the immediate vicinity of the site. Similar-sized cobbles with secondary and primary cortex are also known to occur in the Salt River within 10 km of the site to the south (Pope et al. 2005:5-4). As in the Dry Branch Creek report, the categories of Curdsville and Grier were combined because they outcrop together within the Lexington Limestone formation, and because they exhibit similar visual characteristics. Curdsville/Grier cherts vary a great deal with a fine to coarse-grained texture and luster that can be dull, moderate, or semi-vitreous. Colors are variable but typically include a range of light to medium grays and gray-browns. Interiors tend to be grainy or bedded with areas that are homogenous, and other areas in the same nodule contain fossiliferous bands in alternating shades of gray and brown. Calcite crystals are common in Curdsville, while a variety of microfossils characterize Grier. Nodules or pods of Curdsville/Grier chert up to 16 cm in length are known to occur in gravels and stream terraces of the Salt River within 10 km of the site (Pope et al. 2005:5-5). Tyrone chert is a fine to very fine-textured chert with a semi-vitreous luster, it is typically a very light gray to white colored chert. This chert type is also characterized by white, gray, or black banding. Samples of Tyrone chert are known to occur in the Dix and Kentucky River drainage systems in central Kentucky, south and east of the Wheeler site (Pope et al. 2005:5-5). Boyle chert has a very fine to medium fine-grained texture, and a semi-vitreous luster. There is variation in color, modeling, and opacity in this chert type. Colors include white, gray, tan, brown, olive, pink, red, and blue. A distinctive feature of this chert is the occurrence of white microfossils scattered throughout the silica matrix. Boyle chert is known to occur in terrace, slope, and ridgetop settings in the Salt, Chaplin, Rolling Fork, Green, Dix, and Kentucky River drainages. Pebbles of Boyle chert up to 9 cm in length are known to occur within 10 km of the site (Pope et al. 2005:5-5). Muldraugh chert has a medium coarse-grained texture and a dull luster. The chert is typically mottled. Colors include light brown, yellowish brown, gray-white, bluish gray-white, and pink gray-white. White microfossil inclusions are common. Muldraugh chert is known to occur in slope and ridgetop settings of the Salt, Chaplin, Rolling Fork, Green, Dix, and Kentucky Rivers. Pebbles of Muldraugh chert up to 12 cm in length are known to occur in the Salt River within 10 km of the site (Pope et al. 2005:5-5, 5-6).

5

Harrodsburg chert is a medium coarse-grain chert with a dull to semi-vitreous luster. Colors include tan, orange-tan, olive-brown, and reddish-brown. Harrodsburg chert is a cryptocrystalline, consisting predominantly of microfossil fragments in a silica matrix. Harrodsburg chert is known to occur in ridgetop setting of the Chaplin and Rolling Fork Rivers. This chert was not identified within the Wheeler site assemblage (Pope et al. 2005:5-6). Ste. Genevieve chert has a fine to medium-fine texture and moderate luster. Colors include shades of light to medium blue, and olive gray to yellowish gray concentric zones just beneath the cortex. Inclusions are rare. Ste. Genevieve chert is known to occur in ridgetop settings along the Green and Dix Rivers (Pope et al. 2005:5-6). St. Louis chert has a fine-grained texture and a dull to semi-vitreous luster. Colors include blue, olive, white, mottled grayish-blue and tan, dark greenish-gray to light olive gray, and pale brown. St. Louis chert is known to occur in ridgetop settings of the Salt, Chaplin, Rolling Fork, Green, and Dix Rivers. For the purposes of this study, Ste. Genevieve and St. Louis cherts were combined into a single category due to an inability to separate these chert types as visually they appears to be very similar (Pope et al. 2005:5-6). Any chert that could not be classified was labeled “unidentified.” This category included items too small to identify, those that were too burnt to identify, and cherts that could not be assigned to a known source. It is likely that the majority of the unidentified items were local cherts that could not be confidently categorized. Of all the chipped stone specimens analyzed, 34.8 percent were derived from Salvisa, 33.2 percent from Curdsville/Grier, 8.5 percent Boyle, 4.7 percent Muldraugh, 2.4 percent Ste. Genevieve/St. Louis, and 1.4 percent Tyrone. The remaining 15.0 percent were classified as unidentified (Table 1). Formal and Informal Chipped Stone Tools Projectile Points: Projectile points are defined as a hafted bifacial tool thought to have been attached to a shaft and used as spear point, dart point, or arrow point (Crabtree 1972:86). For the purposes of this study, the only items assigned to this category were diagnostic projectile points (Tables 2 and 3). All non-diagnostic projectile point fragments were classified as biface fragments. To categorize Fort Ancient triangular projectile points, Jim Railey’s 1992 classification scheme was utilized. Accordingly, metric data, such as length, upper width, middle width, lower width, maximum thickness, and weight, were recorded. In addition to these attributes basal shape and raw material type were recorded, and through visual inspection each projectile point was assigned to one of five types. The curated collection from the nearby Dry Branch Creek site was utilized as a comparative collection to help classify projectile points and raw material identifications. Out of 68 projectile points examined, 11 were non-Fort Ancient triangular points and 57 were classified as Fort Ancient triangular points. Of the 57 Fort Ancient points, 70.2 and 25.9 percent were classified as Type 5 and Type 2 Fine Triangulars, respectively. The remaining two points were classified as Type 3 Fine Triangulars (1.8 percent) or Type 4 Fine Triangulars (1.8 percent) (Table 3). No Type 6 Fine Triangulars were identified within the Wheeler site projectile point assemblage.

Table 2. Provenience Data for Formal and Informal Tools.

6

Provenience

Trench Size (m²)

Projectile Points* Bifaces

Biface Fragments

Modified Flakes

Hoe Flakes Cores Total Percent

Trench 1 17 6 1 16 122 2 13 160 16.2 Trench 2 2 4 1 12 84 0 5 106 10.8 Trench 3 10 11 2 26 129 0 6 174 17.6 Trench 4 5 7 2 31 167 0 11 218 22.1 Trench 5 4 9 7 23 150 0 15 204 20.7 Trench 6 2 4 5 17 43 1 4 74 7.5 Surface - 27 14 9 0 0 0 50 5.1 Total 40 68 32 134 695 3 54 986 100.0

Table 3. Projectile Point Type and Provenience Data. Provenience Type 2 Type 3 Type 4 Type 5 Other Total

Trench 1 2 0 0 4 0 6 Trench 2 2 0 0 2 0 4 Trench 3 4 0 1 5 1 11 Trench 4 1 0 0 4 2 7 Trench 5 2 0 0 7 0 9 Trench 6 0 0 0 4 0 4 Surface 4 1 0 14 8 27 Total 15 1 1 40 11 68

Type 2 Fine Triangular (n=15) The Type 2 Fine Triangular (Flare Based) (Figure 2), refers to points with incurvate sides, and/or markedly flaring bases with convex or straight basal margins. These Type 2 points dominate early Fort Ancient subperiod assemblages but continued to be manufactured well into the Middle Fort Ancient subperiod (Pollack et al. 2012; Railey 1992:156-158). Of the 15 Type 2’s, 11 were associated with excavated context, and five were recovered from the surface. Type 2 points were manufactured from Salvisa (n=7), Curdsville/Grier (n=4), Muldraugh (n=2), and Ste. Genevieve/St. Louis (n=2). Two-thirds of Type 2 projectile points have straight bases (n=10) and one third exhibited convex (n=5) bases. The five complete Type 2’s range in length from 26.4 mm to 42.6 mm, with a mean of 32.6 mm. The upper width range is 5.8 to 9.8 mm, with a mean of 8.4 mm. Middle width ranges from 11.2 to 13.9 mm, with a mean of 12.7 mm. Lower basal width ranges from 16.8 to 18.9 mm, with a mean of 17.7 mm. The maximum thickness ranges from 3.6 to 5.3 mm, with a mean of 4.1 mm.

7

Figure 2. Type 2 Fine Triangular Points.

Type 3 Fine Triangular (n=1) Type 3 Fine Triangular (Coarsely Serrated) projectile points (Figure 3) are distinguishable from other specimens by their coarsely serrated lateral margins. Sides are straight and nearly parallel, and bases are either straight or convex. Type 3 points are primarily associated with Middle Archaic sites (Pollack et al. 2012; Railey 1992:158). One Type 3 projectile point was recovered from the surface. It was manufactured from Tyrone chert. Metric data was not gathered for this point because only the medial portion of this point was present. Although the base was broken off, the lateral margins of the point exhibited the most diagnostic aspects of a Type 3, which are its coarsely serrated lateral margins, which is why this point was considered diagnostic and categorized as a Type 3 Fine Triangular point.

Figure 3. Type 3 Fine Triangular Points.

Type 4 Fine Triangular (n=1) The Type 4 Fine Triangular (Short, Excurvate) triangular points (Figure 4) are generally less than 25 millimeters in length, lateral margins are usually excurvate, and bases are primarily convex or straight,

8

but some specimens have a concave base. Based on contextual evidence Type 4’s are primarily associated with the Late Fort Ancient subperiod (Cook and Comstock 2014; Pollack et al. 2012; Railey 1992:158-161). The Type 4 projectile point recovered from an excavated context was manufactured from Curdsville/Grier chert. It had a length of 20 mm, an upper width of 7.9 mm, a middle width of 12.1 mm, a lower basal width of 15.3 mm, and a maximum thickness of 4.3 mm. The base of this point exhibited a straight shape.

Figure 4. Type 4 Fine Triangular Points.

Type 5 Fine Triangular (n=40) The Type 5 Fine Triangular (Straight Sided point) (Figure 5) is the most common Fort Ancient projectile point found and is associated with Early to Late Fort Ancient sites (Pollack et al. 2012). These points exhibit straight lateral margins that range from nearly parallel to basally expanding. Basal corners are either angular or somewhat rounded, and bases are either straight or slightly convex. Of Railey’s five types, this type is most similar to the classic Mississippian Madison Fine Triangular projectile point (Railey 1992:161-163). Of the specimens for which basal shape could be determined, most (n=11) exhibited a straight base, with only two exhibiting a convex base, and one a slightly concave base. Of the 40 Type 5 projectile points recovered from the Wheeler site, 26 were associated with excavated contexts and 14 were recovered from the surface. Type 5 projectile points were primarily manufactured from Curdsville/Grier (n=14), Salvisa (n=14), and Boyle (n=6) cherts. A few were manufactured from Tyrone (n=1), Muldraugh (n=1), or Ste. Genevieve/St. Louis (n=1). The remaining three specimens could not be assigned to a particular chert type. As with the Type 2’s, Curdsville/Grier and Salvisa were the most dominant chert type used in manufacturing these points. The following metric data was obtained from the 14 complete Type 5’s recovered from the Wheeler site. They range in length from 22.2 to 48.4, with a mean of 35.8 mm. Upper width ranged from 6.0 to 9.8 mm, with a mean of 8.2 mm. Middle width ranges from 10.4 to 14.6 mm, with a mean of 12.2 mm. Lower basal width ranges from 13.2 to 22.5 mm with a mean of 15.8 mm. The maximum thickness ranges from 3.8 mm to 6.9 mm, with a mean of 4.7 mm.

9

Figure 5. Type 5 Fine Triangular Points.

Type 6 Fine Triangular (n=0) Lastly, Type 6 Fine Triangular (Concave Base) projectile points are distinguished by their concave basal margins, excurvate or straight lateral margins, and narrow to medium basal widths (Railey 1992:163-165). Type 6 is primarily associated with the Late Fort Ancient subperiod. No Type 6 projectile points were recovered from the Wheeler site. Non-Fort Ancient Projectile Points: In addition to a large number of fine triangular projectile points, 11 non-Fort Ancient projectile points (Figure 6) were recovered from site. They consisted of Lowe Flare base points (n=2), Merom Trimble points (n=2), a Raddatz point (n=1), and a Late Archaic Stemmed point (n=1), and unidentified/non-diagnostic points (n=5). It should be noted that one fine triangular point was categorized as unidentified because it could not be assigned to any of Railey’s Fine Triangular types. We will return to this point later in this paper during the examination of possible indicators of Fort Ancient and Mississippian interaction. It is not uncommon at Fort Ancient sites to find non-Fort Ancient points, such as earlier Archaic and Woodland projectile points. It may be possible that the Fort Ancient peoples were finding earlier projectile points, and either simply collecting them or recycling them for personal use. These non-Fort Ancient points were manufactured from local cherts, being Curdsville/Grier (n=1), Salvisa (n=3), Boyle (n=3), Muldraugh (n=2), Ste. Genevieve/St. Louis (n=1), and Unidentified (n=1) (Table 1). Refer to Table 2 for provenience information. However no further data was gathered outside of provenience and raw material type because they were not the focus of this analysis.

10

Figure 6. Non-Fort Ancient Projectile Points.

Bifaces: A biface is an artifact bearing flake scars on both faces (Crabtree 1972:38). Bifaces can range from rough blanks or preforms all the way to finished tools, such as projectile points. For the purpose of this analysis, whole bifaces were tools that could not be classified as projectile points. Because functional analysis, such as use wear analysis, was not performed, bifaces and bifacial fragments were not officially assigned function. However, arbitrary categories were created based on visual attributes to gain further understanding of the chipped stone assemblage at the Wheeler site. The bifaces that were functionally assigned were done so with an affirmation of more experienced lithic analyst, as well as using comparative collections and a review of existing literature (Pope et al. 2005; Pollack and Schlarb 2009). In addition to provenience (Table 2), data was collected on raw material (Table 1), and basic functional classification. For the purpose of this study, complete bifaces (Figure 7) and biface fragments (Figure 8) were separated into two categories. Of the 32 complete bifaces, 18 were associated with excavated contexts, and 14 were recovered from the surface. A number of complete bifaces were identifiable by possible function, and are presented below: Drills (n=3) Drills are bifacially shaped tools characterized by a pronounced, parallel-sided projection the length of which is at least one third of the total length of the piece. Drills typically have a rhomboid to circular cross section (Pope et al. 2005:5-20). The three drills recovered from the Wheeler site had straight bases and had circular cross sections, and were likely effective for use as perforators. The three drills were manufactured from Grier, Salvisa, and Ste. Genevieve/St. Louis cherts.

11

Adze (n=1) An adze is a bifacially shaped tool with an asymmetrical cross section used for woodworking. The one adze recovered from excavation was manufactured from Boyle chert. General Bifaces (n=25) Any biface that could not be assigned a specific function based on visual attributes was assigned to the general biface category. However, many of these specimens had the shape of knives or potential unfinished projectile points. General bifaces were manufactured from Salvisa (n=6), Curdsville/Grier (n=8), Boyle (n=5), Muldraugh (n=4), and unidentified (n=2) cherts. Rough Bifaces (n=2) For this analysis rough bifaces were defined as specimens that were bifacially worked but retained an asymmetrical shape. Rough bifaces are a part of the first stage of chipped stone tool production where generally only initial edging has taken place (Pope et al. 2005:5-26). Both rough bifaces were manufactured from Salvisa chert.

Figure 7. Complete Bifaces.

Biface Fragments: Bifacial fragments represent any stone tool fragment that was bifacially worked but not diagnostic because they are broken. Of the 134 biface fragments recovered from the site, 125 were associated with excavated contexts and 9 were recovered from the surface.

12

Similarly to the biface category, biface fragments were assigned into arbitrary functional categories: Basal, Medial, and Distal Biface Fragments (n=50) The primary reason this category was created was to separate and identify the triangular projectile point fragments that could not be assigned to a particular projectile point type. However, additional non-triangular basal, medial, and distal fragments of bifaces were also added to this category. Of the 50 fragments, 30 specimens were thought to be fragments of Fort Ancient triangular projectile points. These basal, medial, and distal biface fragments were manufactured from Salvisa (n=12), Curdsville/Grier (n=16), Boyle (n=6), Muldraugh (n=3), Ste. Genevieve/St. Louis (n=2), and unidentified (n=11). Drill Fragments (n=6) Three drill fragments were identified within this chipped stone assemblage. These specimens were manufactured from Salvisa (n=2), Curdsville/Grier (n=1), Boyle (n=1), unidentified (n=2). Scraper Fragments (n=2) A scraper refers to a steeply retouched unifacial or bifacial tool. These are generally used for scraping and processing animal hides (Pope et al. 2005:5-24). Two bifacial scraper fragments were identified within this chipped stone assemblage. They had been manufactured from Muldraugh chert or an unidentifiable chert. Knife Fragments (n=6) The knife fragments category refers to bifacial fragments that did exhibit typical characteristics associated with projectile points. These fragments generally had parallel sides and a larger maximum thickness. These specimens were manufactured from Boyle (n=2), Salvisa (n=2), Curdsville/Grier (n=1), and Tyrone (n=1) cherts. Unidentified Fragments (n=63) Unidentified fragments were specimens that were too small or burned to assign into arbitrary functional categories. These fragments were manufactured from Salvisa (n=20), Curdsville/Grier (n=16), Boyle (n=4), Muldraugh (n=3), Unidentified (n=20). For provenience information refer to Table 2 and for raw material distribution refer to Table 1.

13

Figure 8. Bifacial Fragments.

Modified Flakes: Modified flakes are informal chipped stone tools that were manufactured for a specific task at, or shortly before they were used, and either show evidence of utilization without modification, or minimal modification through retouching (Pollack and Schlarb 2009:35). In addition to provenience (Table 2) and raw material type (Table 1), the number of worked sides and flake type utilized were recorded. Flake edges were examined under a microscope and magnifying glass to determine if any of the specimens had been intentionally retouched. Of the 695 modified flakes recovered from excavated contexts, 97.3 percent (n=676) were utilized pieces of debitage (Figure 9), with only 2.7 percent (n=19) exhibiting evidence of being retouched or shaped. The number of sides utilized on each flake indicated that 46.3 percent (n=322) of the flakes exhibited one used edge, 40.6 percent (n=282) two used edges, 12.7 percent (n=88) three used edges, and 0.4 percent (n=3) four used edges. Modified flake chert types consisted of 35.0 percent Curdsville/Grier (n=243), 36.4 percent Salvisa (n=253), 3.5 percent Muldraugh (n=24), 2.4 percent Ste. Genevieve/St. Louis (n=17), 1.2 percent Tyrone (n=8), and 14.8 percent unidentified cherts. In addition, each individual modified flake was analyzed and recorded using the flake debitage classification system created by Dr. Steve Ahler, director for the Program of Archaeological Research at the University of Kentucky. These flakes were broken into six different categories: primary flakes, secondary flakes, interior flakes, biface thinning flakes, broken flakes, and angular shatter. These categories represent the various stages of progress of the raw material from the original form to the finished stage (Crabtree 1972:58). These categories are defined as followed: Primary flakes, or primary decortation flakes, exhibit an obtuse platform, pronounced bulb of percussion, and over 50 percent of cortex on the dorsal surface.

14

Secondary flakes, or secondary decortation flakes, exhibit an obtuse to right angled platform, a noticeable bulb of percussion, and less than 50 percent of cortex on the dorsal surface, including the platform. Interior flakes, or tertiary flakes exhibit an identifiable obtuse to right angled platform, a bulb of percussion, and little to no cortex on the dorsal surface. Biface thinning flakes exhibit an acute platform, often with distinct facets and lipping on the ventral surface, a diffuse bulb of percussion, and little to no cortex on the dorsal surface. Broken flakes are distal flakes lacking the platform and bulb of percussion. Angular shatter, or blocky fragments lack diagnostic flake attributes such as a platform and bulb, but otherwise show evidence of breakage by percussion. Of the 695 modified flakes, 2.1 percent (n=14) there were primary flakes, 28.9 percent (n=201) secondary flakes, 14.1 percent (n=98) interior flakes, 24.3 percent (n=169) biface thinning flakes, 28.3 percent (n=197) broken flakes, and 2.3 percent (n=16) angular shatter.

Figure 9. Modified Flakes.

Hoe Flakes: While they are not informal tools or modified flakes, the appearance of hoe flakes is unusual in a typical Fort Ancient lithic assemblage. These flakes were included in this analysis to further address the research question regarding Fort Ancient and Mississippian interaction. A hoe flake, or a polished flake, is a piece of unmodified debitage with a heavy amount of polish on the dorsal surface from repeated use in soil. This type of debitage is often created when a chipped stone hoe, used for cultivation is broken or re-sharpened. Provenience information (Table 2), weight, and raw material type (Table 1) were recorded for the few hoe flakes analyzed.

15

All of the hoe flakes (n=3) were of a relatively small size and were recovered from excavated contexts (two from Trench 1 and one from Trench 6) (Figure 10). Two were manufactured from Boyle and one from Muldraugh chert.

Figure 10. Hoe Flakes.

Cores and Tested Cobbles: Although cores and tested cobbles are not formal or informal tools, they were included in this study, because they are byproducts of the initial stages of lithic reduction, and yield valuable information on raw material acquisition and chipped stone tool production. Cores are portable and reduced to a desired shape to allow the removal of a definite type of flake or blade (Crabtree 1972:54). A tested cobble represents objects that have only one or two flake scars (Lowthert et al. 1998:51). Provenience information (Table 2) and raw material type (Table 1) were recorded for these artifacts, as well as cortex type to determine if the core was recovered from a streambed or a primary context in order to gain additional insights into raw material acquisition. Of the 54 cores recovered from excavated contexts (Figure 11), 43 were classified as cores, three as core fragments, and 8 as tested cobbles. Slightly more than two-thirds (67 percent; n=36) exhibited an alluvial cortex, with only 20 percent (n=11) exhibiting primary limestone cortex. None of the remaining seven (13 percent) exhibited any visible cortex. These cores were manufactured from 35.1 percent Salvisa (n=19), 33.3 percent Curdsville/Grier (n=18), 13.0 percent Boyle (n=7), 5.6 percent Muldraugh (n=3), 1.9 percent Tyrone (n=1), and 9.3 percent Unidentified (n=5) cherts.

16

Figure 11. Cores and Core Fragments.

Debitage: Debitage can be described as residual lithic material resulting from tool manufacture, and represents both intentional and unintentional breakage of artifacts either through manufacture or use (Crabtree 1972:58). Although the non-modified chert debitage was not analyzed as a part of this report, basic information such as count and provenience were used to further investigate additional aspects of the chipped stone assemblage from the Wheeler site (Table 4). From the field school excavation there was a total of 9,852 pieces of non-modified debitage, compared to a total of 936 chipped stone tools.

Table 4. Debitage, Formal, and Informal Tool Provenience and Frequencies.

Provenience Unit Size

(m²) Tools Debitage Ratio of Tools to

Debitage Total Percent Trench 1 17 160 1491 1:9 1651 15.3 Trench 2 2 106 419 1:4 525 4.9 Trench 3 10 174 2681 1:15 2855 26.5 Trench 4 5 218 1800 1:8 2018 18.7 Trench 5 4 204 2602 1:13 2806 26.0 Trench 6 2 74 859 1:12 933 8.6 Total 40 936 9852 1:11 10,788 100 Discussion Lithic procurement The intensive utilization of Curdsville/Grier and Salvisa indicate the peoples who inhabited the Wheeler and Dry Branch sites primarily utilized locally available resources. Based on the high percentage of alluvial cortex exhibited on chert cores, core fragments, and tested cobbles, it can be inferred that these

17

people procured the majority of their raw material resources directly out of river contexts. The authors of the Dry Branch Creek report that Curdsville/Grier, Salvisa, Boyle, Muldraugh chert cobbles occur in various sizes within 10-km of the site (Pope et al. 2005:5-4 - 5-6). The remainder of chert types found at this site outcrop in ridge top settings or other drainages within a 50-km radius of the site. After examining the lithic raw material, the possibility that the Fort Ancient people at the Wheeler site were extensively trading with their Mississippian neighbors, or traveling long distances to acquire non-local materials for chipped stone tool production has been effectively ruled out. Though a few pieces of raw material that were exotic looking in nature were identified in the assemblage, it is likely these were transformed through heat treatment or burning. Even if these materials were derived from beyond the 50-km catchment area, they would not constitute a sufficient volume of nonlocal raw materials to indicate any true form of exchange for a specific type of chert. Therefore it can be inferred that the average flint knapper at the Wheeler site likely preferred to utilize chert raw material sources that were closest to the site and easiest to obtain, and it was likely that the Salt River was the primary source for their raw material. The overall distribution of raw material types by specific tool was also examined to see if any one chert type was preferred for certain functions (Table 1). The common trend across all chipped stone tools was the dominance of both Salvisa and Curdsville/Grier chert as the preferred material. These two raw material sources commonly made up 60-70 percent of all tool categories. The remaining chert types only made up a small portion of each assemblage and fluctuated slightly between tool types. The modified flakes (14.8 percent) and bifacial fragment (24.6 percent) tool categories saw the most significant amount of unidentified raw material sources. However, these high percentages of unidentified material are primarily attributable to the relatively small size of the specimens within these chipped stone tool categories, which made identification more difficult. Based on these findings, we can conclude that there are no significant differences in chert preference between tool types, and that locally available resources were the most exploited for all tool categories. Spatial Distribution The spatial distribution of chipped stone tools and debitage were examined within the mound and village area, as well as by individual trench. When examining the overall area that was excavated on the mound (17 m2) compared to Trenches 3,4 and 5 within the midden ring (19 m2), the discrepancy in chipped stone artifact quantities between locations becomes evident. The village area, primarily Trenches 3, 4, and 5, possessed the majority of both chipped stones tools and debitage, accounting for around seventy-one percent of the chipped stone assemblage (Table 4). The majority of the bifaces and biface fragments that were assigned functions, such as drills, scrapers, and knives, also were associated with the midden ring. This likely indicates that this area was the central location for much of the domestic activities (i.e., food processing and hide working) taking place at the Wheeler site. The mound area, Trench 1, contained 15.0 percent of the chipped stone assemblage. While Trench 1 contains a significant amount of chipped stone artifacts as compared to other individual trenches, its overall size is likely the cause for this occurrence. The 160 formal and informal tools associated with Trench 1 were recovered from 17 square meters, whereas Trenches 4 and 5, which only made up 9 square meters, yielded 422 total chipped stone tools. This drastic difference in the ratio of chipped stone tools to unit size may indicate that Trench 1 was not an area of either frequent use or one where many domestic activities occurred. An interesting pattern in the mound area is the dominance of cutting tools within its assemblage. Most of Trench 1’s tool kit contained modified flakes (n=122) and bifacial fragments that were assigned to projectile points/knives (n=8). This may indicate that certain activities that required the use of cutting

18

tools was prevalent in the mound area. While these tools may be related to the multiple burials found within this area, little is known about the activities that took place in Fort Ancient mound areas (Milner 2004:184). Therefore it is difficult to confidently link these tools to one specific activity. Additionally, the recovery of chipped stone artifacts from the mound area could be the result of the use of backfill during its construction. Evidence of mound construction sequences were discovered in the profiles of Trench 1, which indicates that these Fort Ancient peoples were gathering soil and dumping them in this area in order to build the mound. While the areas in which these groups were retrieving the soil were not discovered, it is still possible that some of these chipped stone artifacts could have been gathered in the mound building process and redistributed within the area of Trench 1. Projectile Points: One of the focal points of this analysis was to classify the triangular projectile points and determine how comparable our assemblage was to other Middle Fort Ancient sites and in particular the nearby Dry Branch Creek site. A common trend of Fort Ancient projectile points is the gradual transition from Type 2 to Type 5 points from the Early to Middle Fort Ancient sub-periods. In accordance with this trend, not surprisingly, Type 5 points were much more common relative to Type 2 points at Wheeler. However, when comparing this data to that of Dry Branch Creek, it was found that Type 2’s comprised over 58 percent of that site’s triangular projectile point assemblage, with Type 5’s accounting for only 34 percent of the points (Table 5). Thus, though these two sites are likely connected in some way, differences in projectile point styles suggests that the Dry Branch Creek site represents an earlier occupation of the area, and it wasn’t until later that a village was established on the opposite side of the river. In comparison to other Middle Fort Ancient sites, only the Carpenter Farm site in Franklin County yielded a similar dominance of Type 5 relative to Type 2 projectile points, and this site was occupied about 100 years later than the Wheeler site (Figure 12) (Pollack et al. 2012). Therefore, it may be possible that the high percentage of Type 5’s at the Wheeler site could reflect increased interaction with nearby Mississippian groups, who were known for their straight-sided triangular projectile points.

19

Table5. Fort Ancient Projectile Point Frequencies by Site and Time Period (Pollack et al. 2012).

The presence of a single Type 3 and Type 4 triangular projectile points at the Wheeler site is also noteworthy. Research suggests that the farther a village is located from Fox Farm, where Type 3 points account for more than 40 percent of the site’s fine triangular points, the overall amount of Type 3’s begins to diminish in quantity (Pollack and Henderson 2012). While our sample is limited, the presence of a Type 3 in the Wheeler Site assemblage points to some degree of interaction with more northerly Fort Ancient groups. According to Railey, Type 4’s were thought to post-date A.D. 1400 (Railey 1992:167). However, additional research over the past 20 years has indicated that small numbers of Type 4’s may be present throughout the entire Fort Ancient sequence, with this type increasing in popularity during the late Fort Ancient subperiod (Pollack and Henderson 2012). The presence of a single Type 4 at the Wheeler site is consistent with this observed pattern. The summary of Fort Ancient projectile points will be concluded with a discussion of the triangular point that we could not assign to one of Railey’s five types (Figure 12). Initially this point was believed to be a variant of a Type 3 or Type 6 projectile point. This point has the concave base of a Type 6 projectile point, but its straight, finely serrated lateral margins are not consistent with other Type 6 attributes, as the

20

lateral margins for Type 6 points are characteristically non-serrated. It also does not conform to the Type 3 typology, as the lateral margins are finely serrated as opposed to coarsely serrated, which is the defining characteristic of Type 3 triangular points. Therefore, it was decided that this triangular point does not fall under any of the classification categories defined by Railey (1992), or those outlined within other published Fort Ancient literature on triangular point typologies. Upon further investigation of Late Prehistoric triangular points, these attributes were discovered to be very similar to that of Mississippian points commonly found in the Midwest and Southeastern United States (Lynott 1991; Lowthert et al. 1998). Its presence at the site adds to the growing body of evidence of possible Fort Ancient Mississippian interaction at the Wheeler site.

Figure 12. Non-Fort Ancient Fine Triangular Point.

To conclude the assessment of possible Fort Ancient and Mississippian interaction at the site, the presence of three polished hoe flakes at the site were intriguing. Hoe flakes are common at western Kentucky Mississippian sites and are rare in Fort Ancient lithic assemblages, as it is believed that Fort Ancient cultures tended to use other media, such as wood or shell, as hoes due to the lack evidence of polished flakes and chert hoes at Fort Ancient sites (Yerkes 1988:335,345). While the evidence for Mississippian and Fort Ancient interaction within this chipped stone tool assemblage is slim, the dominance of Type 5 style points, the presence of the finely serrated triangular point and the hoe flakes, coupled with presence of Mississippian ceramics and wall trench architecture leads to the inference that there was some degree of Fort Ancient and Mississippian interaction, which could have been through activities such as migration, marriage, stylistic emulation, or conflict. Lastly, though we were initially surprised to learn that that modified flakes greatly outnumber formal tools at the Wheeler site, upon further investigation we discovered that this occurrence is extremely common throughout Late Prehistoric agricultural societies. The dominance of expedient flake tools was also reported for the Dry Branch Creek site, where it was noted that most flakes discarded at the site were being produced for the use as expedient flake tools rather than the manufacture of formal tools, such as bifaces (Pope et al. 2005:5-50). This idea may be reinforced at the Wheeler site through the relatively small sizes of cores, due to the fact that only small, expedient-type tools could be created with these small chert nodules. This raises the question of why informal tools dominate Fort Ancient chipped stone tool assemblages at these two sites. Based on micro-wear analysis of modified flakes from the Dry Branch Creek site, Pope et al. (2005:5-51) concluded that modified flakes were used in the processing of both soft and hard materials, including bone, antler, and wood. A similar inference could be made for the Wheeler

21

site collection, as some flakes displayed certain features that could have been useful as possible scrapers, perforators, drills, and clearly cutting tools. It is possible that a Fort Ancient reliance on expedient flake tool technology is linked to matters of sedentism, mobility, and the division of labor (Pope et al. 2005:5-50). In conclusion, the identification of these modified flakes not only provides further insight into the types of activities and behaviors of the Fort Ancient families that lived at the Wheeler Site, it also further suggests that Dry Branch Creek and the Wheeler Site can be linked through more than just their relative proximity to each other, but also through the similarities in their material culture. Summary and Conclusion The data recovered from this lithic analysis allowed us to address, develop, and answer some of the initial research questions proposed. Although the majority of evidence of Mississippian interaction at the Wheeler site can be found through its ceramics and wall trench architecture, we were able to gather additional evidence of Mississippian interaction through the chipped stone assemblage. These findings included evidence of polished hoe flakes, a possible Mississippian fine triangular point, and an increased of amount of Type 5 Fort Ancient triangular points compared to Type 2’s. Through an examination of spatial distribution of chipped stone tools between the village and the mound area, we were able to identify a clear difference in chipped stone artifact quantity between the two areas, and suggest why this occurrence exists. We also determined that a majority of the diagnostic points recovered from the Wheeler site fit within the known categories of Fort Ancient typologies. This information also allowed us to further connect and examine the Wheeler site’s Fort Ancient projectile point assemblage with other sites, such as the proximate Dry Branch Creek site and the Carpenter site in central Kentucky, and suggest how these sites may relate temporally. Lastly, through an examination of the chert raw material recovered from the Wheeler site, we were able to conclude that these Fort Ancient peoples were not extensively acquiring or making a concerted effort to trade for exotic chert materials. Similar to other Fort Ancient sites, the Wheeler chipped stone assemblage was dominated by locally available chert raw material resources, a majority of which were likely procured from the Salt River region. Overall, this study allowed us to learn a great deal about the life ways of the Fort Ancient group that occupied the Wheeler site, and we hope that this analysis can help add to the growing body of Fort Ancient literature.

22

References Cited: Crabtree, Don E. 1972 An Introduction to Flintworking. Occasional Papers of the Idaho State University Museum,

Number 28.

Cook, Robert and Aaron Comstock 2014 Toward More Continuous and Practical Artifact Analyses: Defining and Learning From Key

Dimensions of Fort Ancient Triangular Projectile Points in the Miami Valleys. Midcontinental Journal of Archaeology. 39(3):222-250.

Henderson, A. Gwynn 1998 Middle Fort Ancient Villages and Cultural Complexity in Central Kentucky. Ph.D diss.,

Department of Anthropology, University of Kentucky, Lexington. Lowthert, William, Carl Shields and David Pollack 1998 Mississippian Adaptations Along the Barren River in Southern Central Kentucky. Kentucky

Archaeological Survey. Lynott, Mark J. 1991 Identification of Attribute Variability in Emergent Mississippian and Mississippian Arrow Points

from Southeast Missouri. Midcontinental Journal of Archaeology. 16(2):189-211 Milner, George R. 2004 The Moundbuilders: Ancient Peoples of Eastern North America. London: Thames and Hudson

Ltd. Pollack, David and A. Gwynn Henderson 2012 Fort Ancient Fine Triangular Point Classification: Twenty Years Later. Kentucky Archaeological

Survey. Pollack, David, A. Gwynn Henderson and Christopher T. Begley 2002 Fort Ancient/Mississippian Interaction on the Northeastern Periphery. Southeastern Archaeology

21(2):206-220.

Pollack, David, Martin Raymer, Donald A. Miller, Jimmy A. Railey, and A. Gwynn Henderson 2012 Getting to the Point: A reply to Bradbury et al. Journal of Kentucky Archaeology 1(2):52-64 Pollack, David and Eric J. Schlarb 2009 Chapter 5: Chipped Stone, Groundstone and Mica. In Archaeological Investigations of the Early

and Late Fort Ancient Howard Site (15Ma427), Madison County, Kentucky. Pp. 22-47. Kentucky Archaeological Survey Report No. 151.

Pope, Melody, A. Gwynn Henderson, Elizabeth N. Mills, Jack Rossen, Emanuel Breitburg, Nicholas P. Herrmann, James P. Fenton 2005 Phase III Investigations at Dry Branch Creek, Site 15Me62. Wilbur Smith Associates. Railey, Jim A.

23

1992 Chapter XI: Chipped Stone Artifacts. In Fort Ancient Cultural Dynamics in the Middle Woodland Ohio Valley, Issue 8 of Monographs in World Archaeology. Emanuel Brietburg, A. Gwynn Henderson, eds. Pp. 137-139. Madison: Prehistory Press.

Yerkes, Richard W. 1988 The Woodland and Mississippian Traditions in the Prehistory of Midwestern North America.

Journal of World Prehistory. 2(3):307-358.