In the preceding section, the kinds of Paleoindian property types that occur in the Southeast were described. In the pages that follow, the feasibility of developing comprehensive, regionwide datasets of Paleoindian property types, and conducting distributional and other analyses with them, is explored. This is followed by a discussion of specific sites and assemblages by property type category.
To date, information is available to permit the generation of maps illustrating the occurrence of all fluted points (Figure 3), as well as two distinctive southeastern Paleoindian projectile point categories, Cumberland and Suwannee/Simpson (Figure 4). While there are admittedly many problems with these data, such as varying levels of recording from state to state, differing levels of geological visibility for these materials, inclusion of nonfluted points in Florida, where Suwannee and Simpson forms are counted, and so on (as discussed in Anderson and Faught 1998), the maps still probably accurately reflect the geographic distribution of these artifact types on the southeastern landscape.
When examining the incidence of all fluted points in the Southeast, major concentrations and voids can be recognized (Figure 3). Large numbers of fluted points have been recorded in the Tennessee River Valley of northern Alabama, in portions of the Ohio and Cumberland drainages, and in parts of Florida, North Carolina, South Carolina, and Virginia. Some of these clusters reflect single, extremely rich sites or localities, while others, such as those in the major river valleys of the Midsouth, reflect large numbers of points from both sites and isolated finds. Over 1000 fluted points have been reported from four counties in northern Alabama alone, for example, from numerous locations both along and away from the Tennessee River (Futato 1982).
Several major point clusters occur along the eastern seaboard and in northwestern Florida, that may indicate regions where settlement may have continued onto the then-exposed continental shelf. Comparatively few points, in contrast, occur in portions of the Gulf Coastal Plain, lower peninsular Florida, and in the lower Mississippi River Valley (see also Dincauze 1993a). Sampling considerations aside, these areas may have been less attractive to fluted point-using populations. Population levels or intensity of use of these areas, accordingly, is unlikely to have been comparable to that in areas of dense artifact concentration. The Appalachian Mountains stand out as a particularly noticeable void in the otherwise densely covered East, indicating use of this area by fluted point using peoples was comparatively minimal (Lane and Anderson n.d). Preservation conditions, notably the exposure of cultural materials on thin soil horizons, may have led to remains being severely eroded or masked in mountainous areas, meaning site formation processes will have to be carefully evaluated. Butzer (1991), for example, has argued that early occupations may have been present in these areas, but their sites have since be largely lost to erosion.
Examining the distribution of specific Paleoindian projectile point types is difficult, because this kind of information has not been systematically recorded across the region. Two exceptions exist, however, for the highly distinctive Cumberland and Suwannee/Simpson types, which are invariably identified in Paleoindian survey projects. Figure 4 shows the distribution of these point types, based on a sample of 348 Cumberland, 490 Suwannee, and 51 Simpson points, with the latter two categories combined. Both groups have very tight, spatially restricted distributions. Cumberland points are found within and near the Cumberland and Tennessee River drainages of northern Alabama, Tennessee, and Kentucky, while Suwannee points occur primarily in central and northern Florida, south Georgia, and western South Carolina. Save for isolated outliers, the size of these concentrations are similar, roughly 300 km in maximum extent, a figure that may provide clues about the range of the people making these artifact categories. That is, these distributions may represent the regular use areas of distinctive later Paleoindian cultural traditions.
The distributional evidence suggests a number of things about Clovis and post-Clovis Paleoindian occupations in the Southeast (Anderson and Faught 1998:176-177). First and foremost, large numbers of people using fluted point technology were present in the region, and, given the evidence for a gradual evolution from fluted to unfluted forms in a number of areas, these occupations were highly successful. Given the impressive density, numbers, and diversity of fluted points present, in fact, it is possible that fluting technology could have originated in and spread from this region. Pre-Clovis populations, in contrast, may have been too small and scattered for diffusion to have been a significant factor, which may account for the spotty and disparate nature of the Early Paleoindian record. Second, since fluted points are common in some areas and absent in others, this suggests that groups making these artifacts were themselves unevenly distributed over the landscape. Group ranges, while extensive, appear to have been present only in certain areas. Third, if Clovis reflects either the movement of an initial colonizing population, as now seems unlikely, or the radiation of a technology, movement appears to have proceeded in a leapfrog manner (e.g., Anthony 1990; Fiedel 2000). Fourth, concentrations of fluted points appear to represent staging areas, where initial populations settled and grew, and where subsequent Paleoindian subregional cultural traditions emerged, characterized by distinctive and spatially restricted point types (see also Anderson 1990a; Dincauze 1993b). The occurrence of Cumberland and Suwannee/Simpson points, in areas where concentrations of fluted points also occurred, certainly suggest such a possibility. The presence of staging areas, furthermore, would provide locations on the landscape where radiating groups could return to in the event of problems, such as catastrophic accident or even a lack of suitable mates, and know that other people would be present (cf., Anderson and Gillam 2001, Moore and Moseley 2001). Finally, given that large areas of the continental shelf along the Atlantic and Gulf coasts were exposed and habitable during the Paleoindian era, artifact concentrations near the modern coastline probably represent settlement systems that continued into areas now submerged.
Early Paleoindian Assemblages and Diagnostics in the Southeast
At Cactus Hill, a true blade industry that includes small blades, polyhedral blade cores, retouched flakes, and abrading stones was found stratigraphically below a well defined Clovis occupation. The assemblage was documented in two separate parts of the site, in excavations by two different teams of researchers, led by Joseph and Lynn McAvoy and Michael F. Johnson (M. Johnson 1997; McAvoy and McAvoy 1997). Two unfluted lanceolate/triangular bifaces were also found that McAvoy and McAvoy (1997:136) have called Early Triangular. Seven quartzite flakes and three quartzite blade cores were found in and near an amorphous hearth-like scatter of white pine charcoal that yielded an AMS radiocarbon determination of 15,070±70 rcbp (Beta 81590), suggesting a very early occupation (McAvoy and McAvoy 1997:167). Three additional early dates [16,670±730 rcbp (Beta 97708), 16,940±50 rcbp (Beta 128330), and 19,700±130 rcbp (Beta 128331)] and two anomalously recent dates of 9250±60 rcbp (Beta 93899) and 10,160±60 rcbp (Beta 92923) have also been obtained on charcoal from the pre-Clovis levels (McAvoy et al. 2000). The 16,670±730 and 16,940±50 rcbp dates are from hearth areas, while the 19,700±130 rcbp date is near the base of the dune, below the cultural levels. The overlying Clovis assemblage is well defined, with numerous points and tools, and a hearth-like scatter of Southern pine charcoal from the same level has been radiocarbon dated to 10,920±250 rcbp (Beta 81589) (McAvoy and McAvoy 1997:124, 167, 169).
The excavations conducted in both areas of Cactus Hill through 1996 have been superbly documented in a lengthy, well-illustrated monograph (McAvoy and McAvoy 1997), and extensive multidisciplinary follow up work is being conducted to verify the pre-Clovis occupation (McAvoy et al. 2000). The site deposits are comparatively shallow and loosely compacted sands, however, and at least five radiocarbon dates obtained from the Clovis levels ranging from ca. 5285 to 9790 rcbp have been rejected as too late due to "downdrift" (McAvoy and McAvoy 1997:169). The age of the earliest assemblage, at several thousand years before the Clovis occupation, accordingly, must be considered somewhat tentative at present. It is clearly below the Clovis level stratigraphically, however, and as such the Early Triangular point form and associated blade industry appear to be the first well defined pre-Clovis assemblage documented in the region. Given the small assemblage sample sizes and the concerns about the absolute dating, additional research will be necessary to document the nature and true age of this occupation, and whether artifacts associated with it have utility as unambiguous pre-Clovis temporal markers. Even given these caveats, Cactus Hill appears to be the oldest well defined Paleoindian occupation in the region.
Possible fire cracked rock, disarticulated mastodon bones (possibly from butchering and burning), and probable stone and bone tools have recently been reported from the Saltville River valley in southwest Virginia in lake deposits dated from ca. 13,000 to 15,000 rcbp/15,630 to 17,950 B.P.; the results of this work have been admirably summarized in a recent publication (McDonald 2000). The Saltville area has long been known as a rich Late Pleistocene fossil and archaeological locality. In the mid-1940s a fluted point base and three tools of modified mastodon bone and tusk were reportedly found in a bone-bearing bed exposed in a drainage ditch, and by the early 1950s four fluted points were reported from the same general area, three apparently from the same bone deposits (McCary 1951:11; Pickle 1946; Reinhart 1989:158). Extensive excavations have been conducted at Saltville over the past two decades by Jerry N. McDonald (2000), who found a number of possible artifacts during paleontological investigations in a 30 x 20 m excavation area beside the main river channel. Three horizons were recognized, dating from roughly 14,500, 13,900, and 13,000 rcbp/17360, 16675, and 15,630 B.P. (McDonald 2000; see also Goodyear n.d.). Collagen from a fractured and apparently use-worn tibia of a probable musk ox (Bootherium bombifrons) was AMS dated to 14,510±80 rcbp (Beta 117541) from the lowest level; a second date on wood from the same level is nearly identical, at 14,480±300 rcbp (Beta-5701) (McDonald 2000:8, 37-46). Among the other possible artifacts recovered, where the disarticulated mastodon was found, include a small flat sandstone slab, a possible sandstone ax, a small chert chunk with possible use wear damage, a worn chert flake, and concretions that are inferred to have possibly formed from fat rendering (McDonald 2000:8, 33-34).
The middle horizon at the Saltville site included a cluster of pebbles and cobbles from a small depression, the uppermost stratum of which yielded 12 pieces of microdebitage and some fish bones. Twigs collected from a sand lens from within the block yielded a radiocarbon date of 13,950±70 rcbp (Beta 65209); two other dates from the same stratum were similar in age, 13,460±420 rcbp (SI-641) on tusk and 13,130±330 rcbp (A-2985) on wood (McDonald 2000:8, 33). Seven concretions and one prismatic column of weathered bedrock were also found in a second, mud filled depression, and appeared to have been deliberately placed in an upright position. The latest horizon was a feature in an eroded rill into the middle horizon that contained a midden-like concentration containing over 200 clam shells, over 500 pieces of small vertebrate teeth and bones, and 125 pieces of chert microdebitage, some of which appear to be intentionally produced flakes. McDonald (2000:34-36) suggests that the debris formed by people harvesting shellfish and small animals from the lake during periods of low water. If created by human actions, the site would be the oldest shell midden in the New World. The chert is reported as extra-local, and not something found in other non-cultural riverine deposits in the area, arguing for a cultural origin for some or all of the site assemblage.
At the Topper site (38AL23), located at the Allendale County, South Carolina chert quarries, in a 2x2 m test unit opened in 1998, Goodyear (1999b, 2000, n.d., Goodyear et al. 1998) found a small concentration of chert cobbles at a depth of ca. 180 cm below the ground surface, that appeared to represent a Paleoindian cultural feature, possibly a raw material cache. Follow up excavations in 1999 and 2000 encompassed 78 square meters of area, as well as over 100 meters of deep backhoe trenches opened to facilitate geoarchaeological research. A microblade industry characterized by numerous small blades, burins and burin spalls, and microblade and blade cores was found, amid a number of rock clusters, and in one area five circular organic stains that may represent cultural features, possibly post holes.
The upper 100 to 140 cm of the deposits at Topper are colluvial slopewash, and encompass Clovis through historic era remains. Below that are alluvial sands roughly a meter in thickness, where the presumed pre-Clovis industry occurs, that in turn rest on a scoured gray Pleistocene terrace. The geological age and setting of the deposits is currently undergoing extensive examination, to determine when and how the deposits formed. Two dates on humic acids obtained from beneath the gray clay were 19,280±140 rcbp (CAMS-59593) and 20,860±90 rcbp (CAMS-58432) (Goodyear 1999b:10), while OSL (optically stimulated luminescence) dating yielded dates from ca 13,000 to 14,000 B.P. at the base of the colluvium, and from 15,000 to 16,000 B.P. at the top of the alluvial sand layer. These dates suggest the archaeological remains encountered in the alluvial sands date to at least 13,000 to 14,000 B.P. (immediately pre-Clovis), and possibly between ca. 15,000 B.P. (the age of the upper part of the alluvial sands) and ca. 20,000 rcbp/23,700 B.P. (the age of the deposits below the gray clay). Archaeological fieldwork has been ongoing at the Allendale quarries by Goodyear and his colleagues for over 15 years, and is planned for the foreseeable future; detailed publications on the work, however, have yet to appear, and will likely be deferred until the newly discovered possible pre-Clovis deposits have been thoroughly explored. Topper, if verified as pre-Clovis in age, would be an important example of a quarry/occupation property type.
A number of other sites also suggest the existence of pre-Clovis occupations in the Southeast, although for the present they remain to be securely dated, yield remains in more ambiguous context, or are incompletely reported. At Coats-Hines (40Wm31), an apparent kill site in Tennessee, 10 chert tools and 24 flakes were found with the remains of a disarticulated mastodon. Butchering marks and other evidence for human modification were observed on a number of bones, and the tip of a bone projectile point was found between the ribs of the mastodon (Breitburg et al. 1996; John Broster: personal communication 2000). Four OCR (Oxidized Carbon Ratios, Frink 1992, 1994) dates have yielded an age around 13,000 B.P., suggesting a Clovis era site. A radiocarbon date of 27,050±200 rcbp (Beta-80169) was obtained from the base of the deposits, below the cultural level, and a second date, on material from within the dental cusps of the mastodon, was 6530±70 rcbp (Beta-75403) (Breitburg et al. 1996:7). An AMS date on materials from the bone bed yielded at date of 12,030±40 rcbp/14,076 B.P. (Beta 125350) (John Broster: personal communication 2000). The tools included a prismatic blade and a bifacial knife-like base, as well as gravers and scrapers. A number of other faunal remains were also found, including horse, deer, turkey, muskrat, frog, turtle, and a single first phalanx that may be from a domestic dog. The age and contemporaneity of the various remains reported as Paleoindian in age will need to be carefully demonstrated, given the location of the site by a small spring, and the great time range suggested by the existing dates for the deposition of the remains. An early kill site, possibly Clovis or even pre-Clovis in age does, however, appear indicated.
Two submerged sites from Florida provide additional evidence for possible pre-Clovis settlement in the Southeast, both possible kill or occupation site areas. Unfortunately, neither yielded remains that could be used to unambiguously date other sites to this time level. At Page-Ladson along the Aucilla River in northwest Florida, a series of radiocarbon dates between ca. 11,770±90 and 13,130±200 rcbp (ca. 13,816 to 15,784 B.P.) were obtained from a stratigraphic level that included a mastodon tusk with possible cut marks at its base, from possible detachment (Dunbar and Webb 1996; Dunbar et al. 1988, 1989; Faught 1996:162; Goodyear 1999a:467-468). At the Little Salt Springs site, a wooden spear associated with the remains of a giant tortoise was dated to 12,030±200 rcbp (TX-2636) (Clausen et al. 1979:611). While it was originally suggested that the tortoise had been cooked, reanalysis suggests the charcoal-like staining on several fragments is due to differential oxidation (Dunbar and Webb 1996:351). A bone from the tortoise itself was dated to 13,450±190 rcbp (TX-2635), suggesting the association between the tortoise and the spear may be equivocal (Clausen et al. 1979:611). Worked bone and ivory from extinct animals has been found at a number of locations in Florida (Dunbar 1991; Dunbar and Webb 1996; Webb et al. 1984, n.d.). When worked fresh or green, such finds unambiguously indicate a Pleistocene occupation, although absolute dating is still essential to determine exactly when in the Paleoindian era they were made. Dense concentrations of worked bone or ivory might indicate a special purpose quarry/resource extraction area.
Although just outside of the southeast as defined here, a final possible pre-Clovis site, Big Eddy in Missouri, is currently undergoing excavation and may provide important information on pre-Clovis occupation in the general region. The 1997 and 1999 excavations at the site have been superbly documented in two monographs (Lopinot et al., eds., 1998, 2000). A remarkably complete stratified Paleoindian sequence was found that spans the entire continuum from Clovis through the Early Archaic. Below this, a number of unequivocal flakes (n=10) and a possible anvil stone were found (Ray and Lopinot 2000; Lopinot et al., eds., 1998, 2000). An extensive series of radiocarbon dates place these materials between ca. 12,000 and 13,000 rcbp (Hajic et al., eds., 2000:31; Lopinot et al., eds., 1998, 2000). The detailed reporting for this project, like that at Cactus Hill, serve as models for the rest of the profession.
Other sites provide less secure information about pre-Clovis Paleoindian occupations in the Southeast, or else remain to be fully published and hence evaluated. The Southeast has had its share of claims for putative pre-projectile point stone tool assemblages, invariably inferred to have great antiquity, well before Clovis occupations. The best documented presumed early lithic assemblage in the Southeast is from Alabama, the Lively (1965a, 1965b) pebble tool complex, named after its discoverer. These artifacts are now recognized as expedient cores and tools formed during initial lithic reduction activities, and most are now known to date to the Archaic or Woodland periods (Futato 1996:301; Steponaitis 1986:368).
Clovis and Related Assemblages and Diagnostics in the Southeast
At the Carson-Conn-Short (40BN190) site in Tennessee, a large number of Clovis and later Paleoindian points, together with a wide variety of tool forms, including large prismatic blades and polyhedral blade cores, formal unifaces, and bifaces, have been found on the surface and at depths of up to 70 cm over a wide area (Broster and Norton 1993, 1996; Broster et al. 1994, 1996; Nami et al. 1996). Features that have been recognized include tight scatters of artifacts as well as deflated hearths and pits, some of which appear to derive from the intentional thermal alteration of local cherts. While the vast majority of the projectile point assemblage appears to be Clovis, a few presumably later Cumberland points have also been found. The blades and blade cores are exceptional in size, and represent one of the most dramatic Paleoindian blade assemblages ever recovered. A series of OCR dates from excavation units at the site have yielded dates of from ca. 11,700 to 15,000 B.P. These may point to an earlier starting date for Clovis than traditionally assumed, although additional dating will be needed to verify the temporal placement of the assemblage.
Another site suggesting a possible early starting date for fluted points is Johnson near Nashville, where deeply buried hearths with associated fluted preforms yielded dates of 11,700±980 (TX-7000), 11,980±110 (TX-7454), and 12,660±970 rcbp (TX-6999) (Broster and Barker 1992; Broster and Norton 1992:266, 1996:292-294; Broster et al. 1991; see also Goodyear 1999a:448-449). Two of the three dates from the site have such large standard deviations as to preclude their use for dating the onset of fluted point technology, although with the third, and taken collectively, they suggest a possible starting point some time around or just after 12,000 rcbp/14,000 B.P. Early Archaic features and materials are present in the deposits overlying the Paleoindian remains, however, and the authors indicate that further corroboration of the dating is in order (Broster and Norton 1996:294).
Other sites that suggest an early onset for fluting technology include Big Eddy and Aubrey. At the Big Eddy site in Missouri, although just outside of the Southeast as defined here, a date of 11,900±80 rcbp was obtained from a level containing a Clovis assemblage, although six of the seven the other dates from the same deposits ranged from ca. 11,400 and 10,700 rcbp/13,400 to 12,850 B.P., closer to expected range for Clovis (Hajic et al. 2000:31; Lopinot et al., eds., 1998, 2000; Ray et al. 1998; the "outlying" date was 10,265±85). Additional evidence for an early initiation of Clovis technology comes from the Aubrey site north of Dallas, Texas, where two dates of 11,540±110 and 11,590±90 rcbp have been reported (Ferring 1995; Stanford 1999:291). All of these suggest Clovis could have originated around or just after 12,000 rcbp/14,000 B.P.
Terminal Paleoindian Assemblages and Diagnostics in the Southeast
The Cumberland point, common in the Mid-South, is a distinctive waisted lanceolate with a narrow, recurved blade, an expanded base with pronounced ears and a slightly to deeply indented base, fine marginal retouch, and long flutes that run the length of the blade (Breitburg and Broster 1995; Lewis 1954:7). An age of from ca. 10,900 to 10,500 rcbp/12,950 to 12,500 B.P. or slightly later is inferred, given the resemblance to western Folsom points, which are assumed to be a related manufacturing tradition. A small fluted form, some of which may actually be extensively resharpened Clovis points, has been reported in Georgia and South Carolina, and has been provisionally called the Clovis Variant (Anderson et al. 1990:6; Michie 1977:62-65). The chronological placement of this type is unknown but, based on its resemblance to Clovis, it is assumed to be either contemporary or slightly later in age.
In north central Florida and immediately adjoining areas, two seemingly closely related point forms are found that are assumed to date to this interval, the Simpson and Suwannee types (Bullen 1975). Both are large waisted lanceolate points with broad recurved blades and narrow, straight to slightly expanding stems, a concave base, and faint-to-pronounced ears. Separating these forms is exceedingly difficult, since the type descriptions and illustrated specimens exhibit appreciable morphological overlap, with some of the separation apparently based on subtle differences in haft morphology and resharpening (Bullen 1958, 1962, 1975:55-56; Daniel and Wisenbaker 1987:53; Goodyear et al. 1983; Simpson 1948:11-15). While basal thinning has been reported for both forms, Suwannee's apparently also exhibit lateral marginal thinning in basal areas on some specimens (Goodyear et al. 1983:46). Because of this ambiguity, some investigators refer to these forms as "Suwannee/Simpson" while in the Georgia fluted point survey, the type Simpson was used to refer to fluted and Suwannee to nonfluted waisted and eared lanceolate points otherwise meeting the type criteria (Anderson et al. 1990:8). The presence of basal thinning and even occasional true fluting on some Simpson/Suwannee-like forms suggests an appearance some time around or shortly after 10,800 rcbp/12,900 B.P., contemporary with late Clovis or appearing soon thereafter. A long temporal occurrence, however, is also indicated by stratigraphic evidence from sites such as Harney Flats, Silver Springs, and Wakulla Springs Lodge, that suggests these types continued until they were replaced by side notched forms, which are thought to appear ca. 10,200 rcbp/11,850 B.P. (Daniel and Wisenbaker 1987:37-40; Goodyear 1999a:465-467; Jones and Tesar 2000; Neill 1958). The near complete absence of Dalton points in Florida, furthermore, has led researchers to speculate that Suwannee/Simpsons are representative of a contemporaneous subregional variant, or even a somewhat distinct cultural tradition (e.g., Anderson 1990a, 1995a; Dunbar 1991). Accordingly, Suwannee/Simpson point manufacture may be a distinctive, thousand or more year tradition of point manufacture in the Florida area. Stanford (1991:9), in fact, thinks they may be contemporary with or older than Clovis. Resolving possible chronologically sensitive variation within this tradition should be a subject for future research.
Another apparent subregional terminal Paleoindian projectile point variant is the Beaver Lake type, a distinctive waisted lanceolate with a recurved blade, an expanded base with weak to pronounced ears and a slight to deeply indented base, slight to extensive basal thinning on some specimens, and fine marginal retouch (Cambron and Hulse 1964:9; DeJarnette et al. 1962:47, 84; Justice 1987:35-36). Many specimens appear to be unfluted Cumberland points, and the type has a similar center of apparent popularity in the Mid South from northern Alabama through Kentucky, although Beaver Lake points appear at least occasionally over a much broader area, in the eastern and central Southeast. The age of the type is currently unknown, although like Cumberland it is thought to have appeared soon after Clovis ended, ca. 10,800 rcbp/12,900 B.P. Another form, the Quad point, is a waisted lanceolate essentially identical to the Beaver Lake points in most respects, save for the presence of more pronounced, large lobed ears, and a fairly deep basal concavity (Cambron and Hulse 1964:98; Justice 1987:35-36; Perino 1985:310; Soday 1954:9). Some Quads exhibit pronounced basal thinning, sometimes to the point of appearing fluted. Whether Quad and Beaver Lake points are contemporaneous with Cumberland or later in time is unknown, although the presence of basal thinning rather than fluting suggests some or all of these forms may be later in time. Many Quad and Beaver Lake points, particularly the latter specimens, would probably be classified as Suwannee/Simpson if found in Florida.
The terminal Paleoindian point form that occurs most widely over the Southeast is the Dalton, with subregional morphological variants found everywhere save perhaps parts of Florida. Dalton points are lanceolates with straight to slightly excurvate lateral margins on the blade, sometimes with serrations and beveling; straight to slightly expanding bases with slight to more typically deeply indented bases that may be fluted, basally thinned, or simply retouched; and weak to pronounced shoulders with well ground basal and lateral margins (Bradley 1997; Chapman 1948:138; Goodyear 1974; Justice 1987:35-44; Perino 1985:97). Some exhibit evidence for extensive resharpening, and extreme cases may have pronounced bevels and incurvate blade edges, and may resemble drills. Distinct morphological variants have been recognized in a number of parts of the Southeast, such as Breckenridge, Colbert, Greenbrier, Hardaway, Nucholls, and some of the varieties of San Patrice (i.e., vars. Hope and St. Johns) (Ensor 1987; Goodyear 1974, 1999a:440-41; Morse 1973, 1997a, 1997b).
Dalton points are common across the Southeast, and in some areas are found in numbers an order of magnitude or more greater than the point types that presumably preceded them (Anderson 1990a:199). Distributions have never been examined at a large scale, however, to see where distinctive variants and point concentrations occur on the regional landscape, although excellent distributional analyses, employing hundreds of sites, have been conducted in the central Mississippi Valley and particularly in northeast Arkansas (e.g., Gillam 1996a, 1996b, 1999; Morse 1971b, 1973, 1975a, 1975b, 1977; Redfield 1971; Schiffer 1975a, 1975b). Detailed attribute based analyses using large samples of points from across the region should be able to resolve the existence, similarities and differences, and spatial occurrences of morphologically distinct forms, which perhaps should be classified as subregional variants or varieties.
Dalton forms appear to have a long temporal occurrence in the Southeast. Goodyear (1982; 1999a:440-441) argued for a range of from ca. 10,500-9900 rcbp (ca. 12,500-11,250 B.P.) for Dalton, based on an inspection of radiocarbon dates, stratigraphic occurrence, and associations. The radiocarbon plateau that occurs in this interval means that as many as 2250 calendar years are actually represented. This is quite a long time, which might explain the appreciable diversification in morphology that is observed, probably as populations became increasingly localized (see also Ellis et al. 1998:159). Some Dalton points are characterized by true fluting, others have pronounced basal thinning scars, and still others exhibit little more than fine retouch of the basal margins. It is tempting to suggest that the fluted Daltons are earlier than basally thinned forms, which in turn are older than unfluted forms. Both fluted and unfluted Dalton points were found at the Sloan (3GE94) site in northeastern Arkansas (Bradley 1997; Morse 1975a; Morse, ed., 1997), however, suggesting an overlap of thinning strategies occurred, and that any such trends may be more apparent at the assemblage or regional level, then at the level of individual artifacts.
Given the presence of true fluting on some northeast Arkansas Dalton points, and a paucity of intermediate types, it is possible that Dalton evolved directly out of Clovis in the central Mississippi River valley. If so, the starting date for this form may have to be pushed back, perhaps to 10,800 rcbp/12,900 B.P. or so, at least in this area. Located along the continent's greatest river system, in an area that was almost certainly fabulously rich in exploitable natural resources, the central Mississippi River valley likely would have been settled quite early in the colonization process, and would have been an area of impressive population growth and technological innovation. Over 1000 Dalton sites have been found in this area, together with evidence for an elaborate interaction network stretching for several hundred kilometers along the river, demarcated archaeologically by ceremonial Sloan "long blades" (Gillam 1996b; Morse 1997; Walthall and Koldehoff 1998). Like the Florida Suwannee/Simpson types, Dalton points, at least in some areas, may have existed for a long time.
Toward the end of the Paleoindian era, side-notched points occur widely over the Southeast. These have been variously described, with named types including Bolen in the Florida area (Neill 1963:99), Early Side Notched in the Tennessee River Valley of northern Alabama (Driskell 1994, 1996), Hardaway Side-Notched in North Carolina (Coe 1964:67), Kessel Side-Notched in the West Virginia area (Broyles 1966), San Patrice vars. St. Johns and Dixon in the Louisiana and southern Arkansas area (Duffield 1963; Thomas et al. 1993a:35-36), and Taylor in South Carolina (Michie 1966:123). A range of from ca. 10,000 to 9000 rcbp/11,450 to 10,200 B.P. is traditionally assigned these forms, although there are indications that they may appear somewhat earlier, around ca. 10,200 rcbp/11,850 B.P., based on dates in this time range obtained at Page Ladson, Florida and Dust Cave (1LU496), Alabama (Chapman 1985:146-147; Driskell 1994:25-26, 1996; Dunbar et al. 1988; Driskell, however, believes the side-notched forms postdate 10,0000 rcbp/11,450 B.P. at Dust Cave, with the points in earlier strata intrusive). An early appearance is also suggested by the association of Bolen and Suwannee points at Harney Flats in Florida, although the cooccurrence is thought due to the presence of a stabilized land surface and, hence, compressed or conflated stratigraphy (Daniel and Wisenbaker 1987:37-40). Individual side-notched points found in surface or mixed excavation contexts in the Southeast cannot be unequivocally placed in a Paleoindian or Early Archaic time level, however, since side-notched points also occur later in time in a number of parts of the region, in particular during the Middle Archaic period in the case of the Big Sandy type in the Mid South (Kneberg 1956:25; Perino 1985:36), the Cache River Side-Notched type in eastern Arkansas (Cloud 1969:119; Morse and Morse 1983:110), and the Godar type in Illinois (Perino 1963:95). Hafted unifacial tool forms that appear to be contemporaneous with these early side-notched forms include Edgefield scrapers (Michie 1968a, 1972) and Albany scrapers (Webb 1946).
San Patrice varieties found in the Louisiana-southern Arkansas-east Texas area reflect the evolutionary continuum observed elsewhere in the Southeast, from lanceolate Dalton-like forms to side- and then corner-notched forms of the Big Sandy/Bolen and Palmer/Kirk clusters. Thus, San Patrice var. Hope is clearly a local Dalton equivalent, while var. St. Johns, var. Dixon, and var. Keithville represent later side- and corner-notched forms within the same cultural tradition. Points designated San Patrice are common in east Texas (Prewitt 1995:128), Louisiana, and contiguous portions of Arkansas and Mississippi. San Patrice var. Hope points are lanceolates with deeply indented bases and weak shoulders and (in some cases) side-notching and/or extensive resharpening of the blade area (Duffield 1963). In all but the most early-stage specimens, the shoulders are the widest part of the artifact. Basal and lateral margins are ground, and the base may exhibit thinning resembling fluting on some specimens. The variety appears closely related to classic Dalton points from the central Mississippi Valley (Ensor 1987; Morse 1973; Morse and Morse 1983). At the Big Eddy site in Missouri, a San Patrice var. Hope point is associated with a date of 10,185±75 rcbp (AA-26653) (Ray et al. 1998:77; see also Lopinot et al., eds., 1998, 2000). San Patrice var. St. Johns and classic Dalton points were also found in this stratum, which yielded dates from general level fill of 9450±61, 10,400±75, 10,340±100, 10,430±70, and 10,336±110 rcbp (Hajic et al. 2000:31; Lopinot et al., eds., 1998, 2000).
San Patrice var. St. Johns points have more varied, flat to indented basal morphology, and pronounced side- to corner-hafting notches; with extensive resharpening and basal attrition these notches increasingly resemble corner-notched forms (Duffield 1963). An additional side notched San Patrice variety, var. Dixon, has recently been proposed, characterized by pronounced side-notches, and replacing the type Edgewood var. Dixon (Thomas and et al. 1993a:35-36). A fourth San Patrice variety, var. Keithville, is characterized by a straight base and corner-notching rather than side-notching (Duffield 1963; Thomas et al. 1993b:46; Webb et al. 1971). Neither the side-notched var. Dixon nor the corner-notched var. Keithville forms exhibit evidence for pronounced basal thinning resembling fluting, something more commonly observed on the presumably earlier Hope and St. Johns varieties. San Patrice are assumed to be contemporaneous with Dalton and initial Holocene side and corner-notched horizons, with an estimated span of from ca. 10,800 to 9000 rcbp/12,900 to 10,200 B.P. San Patrice var. Hope is assumed to be earlier (probably pre-10,000 rcbp) and vars. St. Johns, Dixon, and Keithville later (post 10,200 rcbp/11,850 B.P.), reflecting an evolution from lanceolate to first side- and then corner-notched forms.
Finally, in the western part of both the Southeast and the lower Midwest, lanceolate projectile points occur in low incidence that resemble classic Great Plains Paleoindian fluted and Plano-like forms, such as Folsom, Scottsbluff, Midland, Agate Basin, and Angostura. These artifacts indicate movement or interaction occurred between these two regions, although the nature of this behavior is not well understood at present (Anderson 1995a; L. Johnson 1989; Munson 1990; Wykoff and Bartlett 1995). Folsom points occur in small numbers east of the Mississippi River (Munson 1990), and are usually described using that type name, although some are reported using local names, such as the Sedgwick type in northeast Arkansas (Morse and Morse 1983:63; Gillam 1996a:406). Folsom points are, however, extremely rare to the east of the Mississippi River, with no more than a few dozen specimens known. In some places where they are present, such as in western Illinois, their distribution appears to parallel the occurrence of Late Pleistocene grasslands, suggesting they represent an eastern extension of their inferred Plains adaptation (Munson 1990). When Plains tradition forms are found in the East they are assumed to have the same age as those in the West. Care must be taken when identifying these types, however, since some "Plano" forms found in the East are actually early stage manufacturing rejects of common fluted or unfluted forms, or else later forms like the Guilford Lanceolate or Brier Creek Lanceolate types (Coe 1964:43; Michie 1968b).
Occasionally, however, finished-looking Plano-like forms are found much further to the east, such as at the Smith Mountain site (44PY154) in Virginia, where a quartz unfluted lanceolate point described as "Plano-like" was found at a depth of three meters below the surface, in the Roanoke River floodplain (Childress and Blanton 1997; Goodyear 1999a:451). An AMS date of 10,150±70 rcbp (Beta 93017) was obtained from charcoal found in the same 2.5 cm vertical level and 50 cm horizontally from the base of the point (Childress and Blanton 1997:12). Scottsbluff, Agate Basin, and Eden-like Plano point forms have also been found in a number of locations in the upper Great Lakes and into Quebec, the Maritime provinces, and into the extreme northeastern United States, sometimes in association with presumably Early Archaic side notched forms (Ellis et al. 1998:160-161). At the Varney Farm site in Maine, Eden-like points have been dated to 9410±110 rcbp (Beta-79658) (Petersen 1995:211; Spiess et al. 1998). The Miller Lanceolate found at Meadowcroft Rockshelter resembles Plano forms, although its dating is appreciably earlier (Adovasio et al. 1999:28). The age and cultural relationships of Plano-like forms found in Eastern North America with assemblages much farther to the west is currently poorly understood. If Clovis could have originated (or been extremely early) in the eastern Woodlands, however, it is entirely possible that lanceolate Plano-like forms could have originated there as well.
Resource Distribution: Caches
The Hawkins Cache, also from northeast Arkansas, was a single cluster of 40 Dalton tools, including 18 points, 11 preforms, 2 grooved abraders, 3 adzes, 3 utilized flakes, 1 chisel, 1 end scraper, and 1 backed blade. These artifacts were found by an avocational archaeologist, who said they came from an area under a square meter in extent amid a much larger Dalton site scatter that was interpreted as a base camp (3LW89) (Morse 1971a:19). No human remains were found in association, although their preservation was considered unlikely due to high local soil acidity. The assemblage, which was described in detail by Morse (1971a), was considered to be a man's toolkit used for a variety of tasks, including hunting, butchering, and the working of wood and bone. Whether Hawkins was a ceremonial burial cache or utilitarian cache will probably never be known. Morse (1997a:2) later noted that the context of the find was probably inaccurately reported to him at the time, and speculated that the cache may have actually come from a burial site like Sloan. A possible candidate site, 3LW505, had, in fact, been looted about this time nearby. The Hawkins cache materials resemble some of the artifact clusters found at Sloan and may well have come from one or more burials. Given the ambiguous nature of its recovery, had tiny bone fragments like those found at Sloan been present, they may not have been noted or recovered.
A number of other Dalton caches have been found in the Central Mississippi Valley, in an area extending some 700 km from northeast Arkansas to south central Illinois. The scattered occurrence of Sloan points over this area, in caches and as well as isolated finds, many of them made on Burlington chert from the Crescent Quarries near the Missouri-Mississippi confluence, has been used to infer the existence and extent of a possible Late Paleoindian prehistoric ceremonial and alliance network (Walthall and Koldehoff 1998). The existence of adzes in Dalton toolkits has long been thought to reflect appreciable woodworking skills, including for the manufacture of dugout canoes (Morse and Goodyear 1973), and watercraft are inferred to have been the way these groups were linked together (Walthall and Koldehoff 1998:261; see also Engelbrecht and Seyfort 1994 and Jodry 1999 for discussions of the evidence for the probable use of watercraft by Paleoindian populations). An analysis of edge wear on Dalton adzes from Sloan indicates that these tools were likely used to work charred wood (Yerkes and Gaertner 1997:63-66).
Utilitarian caches of raw material and tools have been described previously, and appear to have been fairly common on Clovis and later, post-Clovis Paleoindian sites. The Adams mastodon site (15Hr18) in Kentucky is one possible candidate for a submerged meat cache (Walters 1988). At this site, the disarticulated remains of an adult animal were found in pond deposits, together with a number of limestone rocks. A number of the bones exhibited cut marks, and it is possible that the rocks were used to help keep the remains submerged.
Resource Distribution: Kill Sites and Bone Beds
The least ambiguous southeastern "kill" site comes from the Wacissa River, where a Bison antiquus skull was found with an apparent projectile point fragment embedded in the frontal bone (Webb et al. 1984). Whether the animal actually died of this wound and was consumed by Paleoindians is unknown, but the association is unequivocal, and certainly indicates the species was targeted. The disarticulated remains of a young male mastodon were found at the Coats-Hines site in western Tennessee, together with a number of stone tools. The materials are in former pond/stream deposits, yet the presence of stone tools suggests it was not a submerged meat cache but a butchering area. Bones or teeth from a number of other species are reported from the same deposits, including:
If these species can be shown to have been exploited by the Paleoindian people who dismembered the mastodon, this site may provide some of the first evidence for a diversified Clovis age (or earlier) Paleoindian subsistence strategy in the Southeast. Equally important, the remains suggest the presence of domestic dog and the hunting of white-tailed deer, aspects of life well documented in subsequent periods. Dunbar and Webb (1996:351-352), in contrast, note that evidence for the use of white-tailed deer in Florida does not proliferate until after the megafaunal extinctions, although they also take care to point out that the presence of a possible mammoth bone digging stick suggests these people were likely generalized foragers.
Florida's rivers and springs have produced numerous examples of tools made on green bone and ivory from now-extinct Late Pleistocene animals, points and foreshafts, anvils, abraders, awls, and digging tools, providing additional evidence for hunting or scavenging (Dunbar and Webb 1996). Ostensible butchering/cut marks on bones from extinct species are fairly common in Florida, although many specimens remain to be professionally described (Bullen et al. 1970; Dunbar and Webb 1996:351). Similar remains may be present in submerged contexts in other parts of the region.
The giant land tortoise (Geochelone crassiscutata) found in Little Salt Springs, if actually speared, would clearly be a Paleoindian kill site; another example of this species that is thought to have been intentionally killed was found in the Hillsborough River (Clayton 1981; Dunbar and Webb 1996:352). The Little Salt Springs specimen was found on a ledge in direct association with the remains of numerous other extinct and modern fauna, including ground sloth, bison, rabbit, rattlesnake, wood ibis, freshwater turtles, land tortoise, and an immature elephant (Clausen et al. 1979:610). Later Paleoindian and Early Archaic occupations at the site, dated to ca. 9000 to 10,000 rcbp/10,200 to 11,450 B.P., reportedly yielded the remains of white-tailed deer and other game. How many of these remains may represent animals that fell into the sinkhole and were trapped there, rather than hunted and processed by local populations, is unknown. The Late Paleoindian assemblage at Little Salt Springs also included a "nonreturning oak boomerang" as well as "a socketed antler projectile point with the tip of the dart shaft still in its base and the basal portion of a carved oak mortar" (Clausen et al. 1979:611; the mortar was dated to 9080±250 rcbp [TX-2594]). A variety of implements were apparently used to kill game, and fairly extensive plant processing would also appear indicated given the presence of the carved wooden mortar. Numerous crudely sharpened stakes or pins were found driven into the ground near the edge of the drop off to the lower part of the sinkhole; two were dated, yielding determinations of 9645±160 rcbp (I-6460) and 9500±120 rcbp (TX-2460) (Clausen et al. 1979:611). The function of these stakes is unknown, although use as a barrier or a tie-off point for raising water from the cenote has been suggested (Clausen et al. 1975a:31). The investigations at Little Salt Springs, like those at Warm Mineral Springs, have not been reported in detail, rendering interpretation difficult.
At Big Bone Lick in Kentucky, three Clovis points were found together with a number of mastodon teeth and bones in the early years of the nineteenth century, but whether they were directly associated is unknown (Tankersley 1985:28, 1990a:74). Other Late Pleistocene fossil localities in the Southeast have also yielded evidence of possible associations between humans and Late Pleistocene fauna, such as at Yarbrough Cave, Georgia, where a stone tool was found amid animal bones (Elliott and Martin 1991). All such sites should be carefully examined by both archaeologists and paleontologists.
One of the most remarkable assemblages of Late Pleistocene flora and fauna, some of which are directly associated with human remains, occurs in the Aucilla River basin of northern Florida. For the past two decades a multidisciplinary team of researchers under the overall direction of James Dunbar and David Webb (1996; Dunbar 1991; Dunbar et al. 1988, 1989; Goodyear 1999a:467-468; Hemmings 1998) have been examining a number of submerged locations along this drainage. The Page-Ladson site is currently the best known and reported, although excavations have occurred at a number of other localities as well, research popularly reported for over a decade in an extensive annual newsletter put out by the research team, The Aucilla River Times. The Page-Ladson excavations as well as the Dust Cave excavations in northwest Alabama have made use of extensive fine screen and flotation work, directed to the recovery of paleosubsistence remains (Dunbar et al. 1988; Goldman-Finn and Driskell 1994; Peres 1997; Peres and Carter 1999; Walker 1997, 2000). Detailed technical monographs describing the work at both localities are currently in preparation, and should provide new insight into the collection and interpretation of Paleoindian paleosubsistence remains in the Southeast. Care must, of course, always be taken to recognize and collect paleosubsistence information whenever sites of this time period are examined in the region.
Resource Distribution: Human Burials
Resource Distribution: Rock Art and other Petroglyphic or Pictographic
Resource Distribution: Quarries and Workshops
The first detailed exploration of assemblage variation in a quarry area in the Southeast was Gardner's (1974; 1977:258-259, 1983, 1989) examination of the Flint Run Paleoindian Complex in northern Virginia. A series of sites were found along both sides of the South Fork of the Shenandoah River, at or in close proximity to major jasper outcrops, and to one another. Three major site types were identified: quarries, where initial mining and reduction occurred, right at the source of the material; quarry workshop/reduction sites, located at or a short distance away from the source, where further reduction occurred to prepare material for transport; and quarry-related base camps, where the groups exploiting the quarries stayed, and where a wide range of activities occurred. Within the complex, Rudacil (44WR5) and Thunderbird both appear to have been base camps where appreciable reduction activity, characteristic of a workshop, also occurred (Gardner 1974; Walker 1974). The research prompted the idea that Paleoindian occupations in the Southeast were in large measure tethered to lithic raw material sources, with range mobility greatly shaped by the need to return to these sources periodically (Gardner 1977, 1983, 1989). This position has been widely adopted although, as we shall see in the discussion of occupations below, some of the inferences derived from it, namely that occupation sites away from quarry areas tend to be small, have not proven correct. The Flint Run Complex research, however, also convincingly demonstrated that assemblages could vary greatly at quarry areas, something that subsequent investigators have explored in detail in other parts of the Southeast.
An extended, multiyear research program at the Allendale chert quarries in South Carolina, for example, has shown that assemblage composition can vary markedly depending on fairly subtle variation in topography and local raw material quality (Goodyear 1992, 1999a, 1999b, 2000, n.d.; Goodyear and Charles 1984; Goodyear et al. 1985). Paleoindian populations apparently focused on chert boulders freshly exposed in stream beds rather than on more weathered materials on nearby hillsides. Differences of no more than a few meters vertical elevation and a few tens of meters horizontal distance appear to separate Paleoindian period quarrying, workshop, and possible habitation areas within the locality. The research has also shown that some extensively exploited channel raw material sources are now submerged, due to sea level rise and a reduction in stream gradients. Other cultural deposits and chert cobbles and boulders were buried at appreciable depths under colluvial sediments washed down from nearby hillsides. Thus, while quarry/workshop localities are traditionally perceived as being highly visible, care must be taken to ensure that a representative sample of what is actually present is recognized and collected.
Studies of Paleoindian assemblages taken collectively, employing data from multiple sites at and up to appreciable distances from quarries, have begun to occur in various regions in recent years, and have proven to be extremely important to our understanding of raw material use and, hence, technological organization, settlement, and mobility strategies (e.g., Morrow 1996; Tankersley 1989, 1990b, 1991, 1994, 1998; Tankersley and Morrow 1993). Perhaps the finest example of this kind of research conducted to date in the Southeast is that by McAvoy (1992) in southern Virginia, in the vicinity of the Nottoway River and its tributaries. Using extensive and well controlled surface and excavation data from over 100 sites, including the well known Williamson site, the author advanced a series of detailed observations about culture change over time, differing site types, differing activities that occurred on these sites, patterns of settlement movement within the study locality, and possible group territories/ranges in the larger region of southern Virginia and northern North Carolina. The importance of this study lies in its extensive presentation of primary data, its use of well documented avocational collections for serious research, and in highlighting the variability in local Paleoindian site assemblages. Ten different Clovis site types were identified (McAvoy 1992:142-144), and are listed in Table 5.
Settlement analyses, encompassing the analysis of site locational data and/or assemblage composition from a great many sites, has shown how the occurrence of lithic raw materials can, to some extent, shape Paleoindian settlement, as Gardner (1983, 1989) and others have suggested (e.g., Daniel 1998; Goodyear 1979, 1989). The distribution of fluted point and Dalton sites in northeast Arkansas, for example, has been shown to be influenced by the location of knappable chert gravels (Gillam 1996a, 1996b). This research, employing GIS technology and computerized site file and assemblage data, permits the examination of large numbers of sites simultaneously. As such, it offers the potential to resolve site types based on their environmental associations, which can then be verified and explored further with direct field work. These kinds of analytical procedures can pinpoint areas on the landscape where various site types might be expected to occur, and can be used to suggest the kinds of activities and assemblages might be expected at such sites, where only minimal information may currently exist.
While the examination of localities or districts is emerging as a particularly useful approach to understanding Paleoindian occupations in an area, a number of discrete Paleoindian quarry/workshop sites have also witnessed appreciable data collection, analysis, and reporting. These include Adams (Sanders 1988, 1990), Carson-Conn-Short (Broster and Norton 1993, 1996; Broster et al. 1994, 1996; Nami et al. 1996), Hardaway (Coe 1964; Daniel 1998), Theriault in eastern Georgia (Brockington 1971), Wells Creek Crater (Dragoo 1973), and Williamson (McCary 1951, 1975, 1991; Peck 1985). Some of these, such as Adams and Williamson, have been incorporated into larger districts, and have been used in comparative analyses employing sites from the surrounding area. The detailed reports and papers describing these particular sites are crucial sources of information about southeastern Paleoindian life. The production of such reports, in fact, is an essential part of research, and until Paleoindian sites are well published, they can contribute little to our understanding.
In closing, while this discussion has focused on lithic quarry/workshop property types, bone, shell, or ivory "quarry" and workshop locales might well be discovered at some point in the Southeast. The collection of old or green bone or ivory for use in tools could be considered a special type of quarrying behavior, for example, and site types or activity areas may exist where these resources were processed. At present, the most likely area where such sites may be found is in submerged contexts, either underwater or in marsh or peat deposits. The presence of such sites is certainly plausible, given the discovery in Florida of anvils, digging sticks, and a possible beamer made on mastodon bones, an awl made from extinct horse tibia, as well as numerous bone and ivory point and foreshafts (e.g., Dunbar and Webb 1996; see also Clausen et al. 1979).
Resource Distribution: Occupations
Major Clovis era occupation sites were once thought to be rare in the Southeast away from quarry areas, something thought due to the early adoption of a generalized foraging adaptation to effectively exploit the widespread hardwood forests present in the lower Southeast at this time (Lepper and Meltzer 1991:177; Meltzer 1984, 1988, 1993; Meltzer and Smith 1986). The large number of isolated finds of fluted points that occur in the region are thought to reflect extensive residential mobility, with habitation areas away from lithic source areas occupied only briefly and hence leaving behind a minimal archaeological record. This perspective was shaped, in part, because many of the best known inferences about Clovis lifeways and particularly residential camps in the Southeast, at least until quite recently, came from research conducted at quarry/workshop assemblages (e.g., Anderson and Sassaman 1996a:23-28; Gardner 1977, 1989; Goodyear et al. 1990). The inferred low visibility of Clovis residential sites in the region, however, is more apparent than real once the total record is examined (Anderson 1990a, 1995b, 1996). Habitation sites characterized by appreciable numbers of Clovis points, bifacial and unifacial tools, and other artifacts, have been documented in a number of areas. McAvoy (1992:142-144), for example, delimited several kinds of non-quarry Clovis residential sites in southern Virginia, and gave examples of sites that fell into each category (Table 5).
Major non-quarry presumed Clovis residential sites in the Southeast include Belle Mina, Heaven's Half Acre, Joe Powell, and Quad in Alabama (Cambron and Hulse 1960; Ensor 1985, 1992; Futato 1996; Hubbert 1989); the Barnett Shoals and Feronia localities in Georgia (Anderson et al. 1990; Ledbetter et al. 1996); Parrish Village in Kentucky (Webb 1951); Avery Island and John Pearce in Louisiana (Gagliano 1967; Webb et al. 1971); the Taylor and Manning sites in South Carolina (Anderson 1979; Anderson and Sassaman 1996b; Goodyear et al. 1990; Michie 1996); the Pierce (40Cs24) and Twelkemeier (40Hs173) sites in Tennessee (Broster 1982; Broster and Norton 1990); and the Conover and Greensville County sites in Virginia (McAvoy 1992). While residential sites do occur away from quarry areas in the Southeast, their recognition and precise delimitation is often difficult or impossible due to the extensive reoccupation that sometimes occurred (Anderson 1990a). Thus, while sites like Manning, Taylor, and Quad, for example, have yielded appreciable numbers of fluted points and formal stone tools, they have also produced thousands or even tens of thousands of later points and tools, rendering identification of specific occupational assemblages difficult or impossible, particularly if most of the materials derive from surface context. Carefully recording where surface materials came from within a site, however, is a way to delimit specific occupation areas, as Michie (1996:242, 267-269) demonstrated at the Taylor site, and O'Steen and her colleagues (1986:40) did at Barnett Shoals in Georgia.
Comparative analyses of eastern and southeastern Clovis site assemblages has been conducted by a number of authors, and provide some insight into the characteristics of residential assemblages of this period, and the kinds of research questions they can address (e.g., Faught 1996; Meltzer 1984, 1988; Shott 1986a). Sanders (1990:65-69), in an analysis of a dozen major eastern Paleoindian sites that included several southeastern sites, looked at the presence or absence of specific site physical attributes such as size class, evidence for single versus multiple occupations, and the kinds of functional activities that were apparently represented (i.e., quarrying, workshop, hunting camp, habitation), as well as a series of technological attributes related to reduction/manufacturing practices. Interestingly, the presence of true blades and blade cores was noted to be more common on southeastern sites than those in the Midwest and Northeast (Sanders 1990:67), a finding used to suggest a fairly early date for these assemblages (Ellis et al. 1998:159). The four southeastern sites examined, Adams, Thunderbird, Wells Creek Crater, and Williamson, were considered to be combination habitation and manufacturing sites, while most of the remaining northern sites were interpreted as hunting camps.
In this same analysis, Adams was found to be so similar to the Wells Creek Crater that the two sites could possibly "be attributed to the same cultural group" (Sanders 1990:69). The difference was that Wells Creek Crater almost certainly represented a number of separate episodes of site use while Adams appeared to be the result of a single occupation. Interestingly, even though outcrops were close at hand, at both sites extensive lithic raw material conservation strategies were employed, to seemingly maximize the utility of every piece of knappable stone. Sanders (1990:68-69) pointed out how this was contrary to the expectation that raw material use should be relatively profligate on Paleoindian sites in source areas, and that conservation strategies should increase with increasing distance from source areas (e.g., Gardner 1974:5-6 and MacDonald 1968:128-129, cited in Sanders 1990:68). He suggested instead that the assemblages might be due to the recent movement of peoples accustomed to having limited lithic raw materials into a stone-rich environment. As such, Adams and Wells Creek Crater might be the signatures of initial colonizing groups. Alternatively, raw material conservation may have been routinely practiced in all locations, since it was an essential aspect of life to highly mobile peoples critically dependent on their stone tools; this was why they were made on predictable, high quality materials, and were designed to maximize use-life (e.g., Goodyear 1979, 1989; Kelly and Todd 1988).
Paleoindian occupation sites away from quarry areas postdating Clovis are fairly common in the Southeast, although well reported examples are less common. Among the best known occupation sites are Dust Cave and Stanfield Worley Bluff Shelter in Alabama (DeJarnette et al. 1962; Driskell 1996; Goldman-Finn and Driskell 1994); Brand and Lace Place in Arkansas (Goodyear 1974; Redfield and Moselage 1970); Harney Flats in Florida (Daniel and Wisenbaker 1987); Taylor Hill in Georgia (Elliott and Doyon 1981; Ledbetter et al. 1996:273); Hester in Mississippi (Brookes 1979; McGahey 1996:371-372); the Haw River sites in North Carolina (Claggett and Cable 1982); Taylor in South Carolina (Michie 1996); and Nuckolls, Nuckolls Extension, Puckett, and Twelkemeier in Tennessee (Broster and Norton 1990, 1991; Lewis and Kneberg 1958; Norton and Broster 1992, 1993). With the exception of the Suwannee/Bolen assemblage from Harney Flats, most of these sites produced Dalton assemblages, sometimes with earlier or later Paleoindian/Early Archaic components present as well (Walthall 1998). Where later materials were present, these were almost invariably side notched projectile points.