lithic analysis

by Tara L. Potts and Philip J. Carr

Jump to: Lithic Analysis and Organization of Technology - Culture History
Lithics and Reconstructing Cultural Lifeways - Flakes, Tool Use, and Other Lithics - Conclusions
- References

Lithics and Reconstructing Cultural Lifeways

An organization of technology approach is used here to inform the analysis of the lithic assemblage with the goal of addressing questions of prehistoric lifeways at the Madison Park site. The chipped-stone assemblage recovered from the Madison Park site consists of 15,969 artifacts larger than ¼ inch: 82 bifaces, 114 hafted bifaces, 449 cores, and 15,324 pieces of flake debris. (Note that some subsequent discussions include artifacts smaller than ¼ inch and therefore use different counts.) Specific analyses were conducted for each of these artifact classes and inferences are drawn with a consideration of each artifact class in relation to the entire assemblage. The entire life cycle of the stone tool—from raw material procurement, to tool production, use/reuse, and discard—is examined here and this cycle is used here to organize the discussion.

Chipped-Stone Raw-Material Procurement
Lithic raw material identification is based on macroscopic characteristics, such as inclusions, texture, cortex, and color. The lithic type collection at the Center for Archaeological Studies (consisting of geologic samples of raw materials from the greater Southeast concentrating on Alabama and Mississippi) was used for comparative purposes when identifying lithic materials. A formal geologic survey to determine raw material distributions in central Alabama has not been published. Table 6-2 lists the raw materials found in the Madison Park site chipped stone assemblage. Unsurprisingly, the two materials locally available for the production of chipped stone tools overwhelmingly dominate the assemblage: quartz (n=54,120) and Knox chert (n=9,687). Other materials include Coastal Plain chert (n=241); Hollis quartzite (n=75); Tallahatta chert, also commonly referred to as Tallahatta agate (n=59); Tallahatta sandstone (n=3); Fossiliferous Bangor (n=1); and unidentified materials (n=14).

Table 6-2. Lithic raw materials at Madison Park.

Raw Material







Knox chert



Coastal Plain chert



Hollis quartzite



Tallahatta chert (“agate”)



Tallahatta sandstone



Fossiliferous Bangor









Local Materials. Raw materials are considered local if they are available within 10 km of the site. The two most abundant and locally available raw materials are quartz and Knox chert. Quartz is available in gravel bars of Sevenmile Creek adjacent to the Madison Park site, but no formal survey of this source was undertaken. The Knox chert formation is located at the northeastern tip of Alabama, along the Tennessee and Coosa rivers approximately 265 km from the site. However, the Knox cores (Figure 6-9) recovered at Madison Park are small (¼- to ½-inch wide) and highly weathered, suggesting transport south, down the Coosa River to central Alabama. Inhabitants of the Madison Park site likely procured Knox chert nodules from the Alabama River, approximately 10 km away, near the confluence of the Coosa and Tallapoosa.

Figure 6-9. Typical Knox chert cores from the Madison Park site: (a) FS 390; (b) FS 521; (c) FS 503; (d) FS 506; (e) FS 591.

To better understand raw material availability at Madison Park, four raw material samples were taken along the banks of the Tallapoosa and Coosa rivers. Two samples were taken from a gravel bar of the Tallapoosa River, approximately 3 km southeast of the confluence of the Tallapoosa and Coosa rivers and about 8 km from Madison Park. One sample was collected near the shore and another approximately 30 meters from the shore. In these instances, all rocks and gravel on the surface of a 1-meter diameter circle were collected. Each sample was passed through 1-inch, ½-inch, and ¼-inch nested screens. Materials from the 1-inch and ½-inch screens were examined using bipolar percussion to assess the quality of the raw material each cobble contained. Quality and consistency were compared subjectively to the raw material choices made by the original inhabitants at Madison Park, as evidenced by the cores reduced at the site and bifaces manufactured there. Approximately, 80 percent of both samples are small, water-rolled pebbles that pass through a ¼-inch screen. The sample near the shore yielded 108 cobbles from the 1-inch screen. Of these, all are low-quality quartz and are inconsistent with the characteristics of the material recovered at Madison Park. Of the lithics from the ½-inch screen (n=604), all are quartz and only eight are examples of the quality used for stone tools at Madison Park. The sample collected 30 meters from the shore yielded 121 low quality quartz cobbles from the 1-inch screen. The ½-inch screen yielded 768 quartz cobbles, but only 27 were considered knappable quality.

One sample of large cobbles was hand-collected along a ½-km stretch of the Tallapoosa, and an additional sample was hand-collected along the Coosa River near Wetumpka, Alabama. From the Tallapoosa River, 86 cobbles were collected and subjected to bipolar percussion to gauge the quality of the raw material. All cobbles are quartz and only 11 were considered knappable. The sample from the Coosa River was the only sample to yield Knox chert (n=46) and all but 15 passed through the ½-inch screen. Most of the Knox chert pebbles are high quality and few (n=10) are unknappable, mostly due to inclusions. Sixteen of the 102 quartz cobbles from the Coosa were considered high quality.   

This collection experiment, albeit limited and certainly not completely exhaustive of the banks of the Tallapoosa and Coosa rivers, sheds some light on the materials most readily available to the inhabitants of Madison Park. Quartz could be acquired from the nearby Tallapoosa River, but was probably available even closer from the banks of the tertiary creeks or streams in the region. Knox chert is available from the Coosa River in the form of small nodules. If Knox chert was needed for tool manufacture, then travel to the banks of the Coosa River (or to the Alabama River, further downstream) may have been required. However, a similar study of an upstream section of the Tallapoosa River in Elmore County yielded Knox chert in a gravel bar sample (Belcher 1987:23), suggesting more diligent search might reveal Knox chert in other locations closer to the Madison Park site.

Non-Local Materials
The remaining raw materials (Tallahatta chert, Tallahatta sandstone, Hollis quartzite, Coastal Plain chert, and Fossiliferous Bangor) are considered non-local raw materials. The closest Tallahatta formation is approximately 80 km south of the Madison Park site. Coastal Plain chert is available in the most southeastern counties of Alabama (Haywick and Carr 2004b:1). The Hollis quartzite formation is in east-central Alabama, and the closest outcrop to the Madison Park site is approximately 81 km away. However, this formation is located along the Tallapoosa River, suggesting that hydraulic transport of materials closer to the site is possible.

Importantly, none of the Hollis quartzite recovered from Madison Park exhibits weathering from transport, and our raw material study did not recover any Hollis quartzite. The Fossiliferous Bangor flake recovered from Madison Park is high quality blue-grey chert with fossilized Bryozoa inclusions (Meeks 1999:157). The Fossiliferous Bangor formation is located in northeastern Alabama, in Colbert and Franklin counties (Meeks 1999:157).

Interestingly, most of the non-local raw material (by count) comes from southern Alabama. Travel and contact with peoples to the south may have occurred to acquire this material.

Chipped Stone Tool Manufacture
A total of 449 cores and 231 stone tools were recovered at Madison Park. Each class is described in terms of technology and manufacture. In subsequent sections, tool use and discard are discussed, as well as interpretations of lifeways.

A core is defined as “a block of raw material from which flakes, blades, or bladelets have been taken, in order to provide blanks for tools” (Inizian et al. 1992:84), or a “piece of isotropic material bearing negative flake scars, or scar” (Crabtree 1972:30). A total of 449 cores were recovered from the Madison Park site, the largest non-flake artifact class in this assemblage. Cores are subdivided by number and kind of flake scars, while cortex is recorded as present/absent. A block/nucleus of raw material with a single flake removal larger than 1 cm is considered a tested pebble (n=149, 33.2%). A block/nucleus of raw material from which flakes have been removed by the application of force from two directions is a bipolar core (n=143, 31.8%). Bipolar cores are further identified by the presence of flat fracture faces, exaggerated negative compression rings, platform crushing, and a relatively small size. A bifacial core is a nucleus with a portion of the flaked surface exhibiting a bifacial edge (n=41, 9.1%). Bi-directional cores (n=33, 7.3%) exhibit flake removals from at least two different platforms in two different directions. Many bi-directional cores in this assemblage exhibit flake removals in one direction on one face and one or more flake removals in another direction on the opposite face.

Tested bipolar cores (n=31, 7.2%) exhibit a flake removal and subsequent bipolar percussion. It is hypothesized that the single flake removal occurred to test the stone for quality, then the piece was subjected to bipolar forces. A block/nucleus with more than one flake removal originating from more than two platforms is termed an amorphous core (n=29, 6.5%). Twenty-three cores (4.9%) were classified as indeterminate, because the poor quality of quartz made analysis of flake scars problematic.

The majority of cores (n=380, 84.6%) are quartz (Figure 6-10) and are generally small in size (SG 1-inch, n=78, 20.5%; SG ½-inch, n=290, 76.3%; SG ¼-inch, n=12, 3.2%). The abundance of quartz cores is probably due to easy access to this raw material from Madison Park. Most quartz cores are bipolar (n=127, 33.4%) or tested cobbles (n=121, 31.8%). The number of tested cobbles suggests a winnowing process described by Shelley (1993) was practiced by the Madison Park inhabitants. That is, quartz cobbles were examined for quality at the source prior to being brought to the site. The remaining quartz cores are bi-directional (n=32, 8.4%), bifacial (n=30, 7.9%), tested bipolar (n=27, 7.1%), amorphous (n=21, 5.5%), and indeterminate (n=22, 5.8%). Informal bipolar knapping experiments were completed with the collected quartz cobbles prior to conducting the lithic analysis. In many cases, a bipolar core was completely obliterated in these experiments and produced an abundance of bipolar flakes. This suggests that the number of bipolar cores is likely underrepresented in these totals.

Figure 6-10. Typical quartz cores from the Madison Park site: (a) FS 570; (b) FS 584; (c) FS 520; (d) FS 460; (e) FS 585; (f) FS 487.

A total of 65 Knox chert cores (see Figure 6-9) were recovered from the Madison Park site. These are generally small in size, with the majority in the ½-inch screen (n=42, 87.6%). The majority, as with quartz cores, are tested cobbles (n=24, 50.0%). The second most abundant type of core in this raw material category is bifacial (n=10, 20.8%). The remaining Knox chert cores are bipolar (n=6, 12.5%), amorphous (n=3, 6.3%), unifacial and bipolar (n=4, 8.3%), and bi-directional (n=1, 2.1%). There is some similarity in the core forms present for Knox chert as compared to quartz. These similarities likely result from the local availability of the sources and the small package size of the nodules, but the lower number of bipolar cores is somewhat surprising given the size of available cobbles of Knox chert.

Only four cores of non-local materials were recovered from the Madison Park site: Hollis quartzite (n=2) and Coastal Plain chert (n=2). Hollis quartzite is generally grainy, while Coastal Plain chert is generally fine grained. The two Coastal Plain chert cores are small, retained in the ¼-inch mesh screen, while both of the Hollis quartzite cores were caught in the ½-inch mesh screen. The Coastal Plain chert cores exhibit bipolar percussion and the Hollis quartzite cores are classified as a tested cobble and bifacial core. The Coastal Plain chert cores may have arrived in some other form and were simply bipolarized at the end of their use life to produce whatever useable flakes were possible. The tested cobble of Hollis quartzite is a bit surprising, considering the potential distance it traveled, but a bifacial core is expected for nonlocal materials, as a biface is a useful form to transport (Kelly 1988).
In sum, the most common types of cores present at the site are single flake removal, bipolar, or bifacial. Given the nature of locally available quartz and Knox chert, these core forms are not surprising. Bipolar percussion is an expedient way of producing flakes from small cobbles and is often used on low quality materials, such as quartz. In an examination of expedient core technology and sedentism, Parry and Kelly (1987:288-289) suggest that:

A fairly dramatic transformation of the lithic industry occurred shortly after AD 500 in the Eastern Woodlands region of North America, between the Middle and Late Woodland periods.

Standardized core forms nearly vanished in the Late Woodland and Mississippi periods, and formal tools became uncommon. Only small pressure-flaked arrowpoints were still produced in any quantities, while other formal tool types (microblades, large bifacial hoes, etc.) appear to have been manufactured by a few specialists. The overwhelming majority of chipped stone tools in Late Woodland and Mississippian domestic contexts were expedient unretouched flakes, frequently struck from bipolar cores.

They go on to suggest that the shift to expedient cores and flake tools is due to a shift in mobility and availability of raw materials (Parry and Kelly 1987:301). At Madison Park, locally available quartz and Knox chert were brought to the site and stockpiled, as evidenced by the number of tested cobbles, for future use in flake production.

 Bifacial Tools

Bifacial tools involve flake removals from two sides and “must, in all cases, be situated on the same part of an object, coming from the same edge” (Inizian et al. 1992:76). For all bifaces, size grade, weight, reduction technology (hard hammer reduction, soft hammer reduction, and retouch), cortex amount, and failure type are recorded, as well as length, width, and thickness. Failure type is based on definitions and figures in Johnson (1981). Finally, additional metric attributes are recorded, such as width, length, thickness, and blade length. Bifacial tools with a definable haft area are classified as hafted bifaces and a greater number of attributes are recorded. Hafted bifaces are discussed under a subsequent heading.

Formal stone tools, such as bifaces and unifaces, are traditionally analyzed using a typological classification. A common biface typology distinguishes Stage 1, Stage 2, Preform, and Projectile Point. These types are usually characterized as follows: Stage 1 bifaces are generally large and thick with some cortex remaining and are evidence of the earliest stage of biface production. While this description succeeds in providing a general understanding of the type, it fails to adequately define that type. Further, the overall typology is based on an assumption of a staged approach to projectile point manufacture that should be demonstrated, not assumed (Shott 1996). Also, employment of the type “Projectile Point” presupposes an aspect of use that is never proven. As Ahler (1971) demonstrated a number of years ago, artifacts in the same morphological class as true projectile points were potentially used for a variety of prehistoric uses, such as cutting, boring, and scraping. Quite simply, form does not equate with function.

Rather than use a staged typology, a technological analysis focuses on recording specific attributes. Some researchers use a simple paradigmatic classification (e.g., Dunnell 1971) that clearly defines each attribute, and classes are then formed empirically by the intersection of attributes. The attributes used here for bifaces and hafted bifaces are defined in Appendix I.

Production failure types noted in this assemblage are lateral snaps, perverse fractures, incipient fracture planes, outre passé, and lateral hinge fracture. A lateral snap is a transverse fracture that basically cuts the biface into two pieces in an almost straight line (Johnson 1981:47). In contrast, force traveling through a biface during rotation results in a perverse fracture, a spiral break through the biface (Johnson 1981:46). Bifaces also break during manufacture because of inclusions or incipient fracture planes within the material. When force is applied to the stone, the biface will fail along these planes creating a distinctive break (Johnson 1981:48). A lateral hinge fracture occurs during manufacture, a flake terminates at almost a right angle along the lateral edge of the biface (Johnson 1981:44). An outre passé manufacturing failure is a reverse hinge fracture that occurs when the force from impact rolls back towards the platform causing the biface to break apart (Andrefsky 1998:18). Haft snaps can result during either manufacture or use.

A total of 108 bifaces were recovered during the investigation of the Madison Park site (Appendix I). Over 50 percent are Knox chert (n=59; Figure 6-11), and these bifaces are classified by reduction technology as soft hammer percussion (n=12), soft hammer and retouch (n=21), retouch (n=24), and undetermined (n=2). Ten complete Knox bifaces yield average measurements for length = 21.6 mm; width= 12.8 mm, thickness = 8.3 mm, and weight = 5.5 g. The relatively small size of these bifaces is consistent with the small cobble size evidenced in the core data. Of the incomplete Knox chert bifaces (n=48), a majority (n=35) are the distal end. The remaining incomplete bifaces are the proximal end (n=2), medial (n=4), lateral (n=2), and one facial fragment. The most common manufacturing failure in the Knox chert biface assemblage is lateral snap (n=23). Other types of manufacturing failures exhibited in this assemblage are perverse (n=12), outre passé (n=2), lateral hinge (n=1), and incipient fracture (n=2). One biface exhibits an impact fracture and another is a possible haft snap. Four biface fragments were caused by thermal fracture. Given the high quality of Knox chert, it is unlikely that these stones were exposed to heat treatment before knapping. These bifaces, instead, were probably discarded and exposed to heat after deposition.

Figure 6-11. Typical Knox chert bifaces and biface fragments from the Madison Park site: (a) FS 451; (b) FS 364; (c) FS 365; (d) FS 766.

Quartz is the second most abundant material used in biface manufacture at Madison Park (n=32; Figure 6-12). Nearly complete quartz bifaces exhibit retouch (n=11), soft hammer percussion and retouch (n=8), both hard and soft hammer percussion (n=3), or soft hammer percussion (n=1). Only six quartz bifaces are complete and their average measurements are as follows: length = 29.1 mm, width = 20.7 mm, thickness 9.5 mm, weight = 5.78 g. The quartz bifaces are slightly, but consistently, larger than Knox chert bifaces in each dimension. As with Knox chert, most of the broken quartz bifaces are distal ends (n=18); the remaining incomplete bifaces are proximal ends (n=2), have part of the base missing (n=2), are indeterminate fragments (n=3), or are missing the tip (n=1). Manufacturing failures identified in the assemblage are lateral snaps (n=8), perverse fractures (n=9), and incipient fractures (n=2).

Figure 6-12. Typical quartz bifaces and biface fragments from the Madison Park site: (a) FS 580; (b) FS 584; (c) FS 530.

The technological classes of Coastal Plain chert bifaces (n=12) are retouch (n=5), soft hammer percussion (n=4), soft hammer with retouch (n=2), and hard hammer percussion (n=1). Six are complete and their sizes are consistent with other bifaces recovered: length=28.1 mm; width=20.74 mm; thickness=6.62 mm, weight=5.42 g. Of the incomplete Coastal Plain chert bifaces, five are distal and one is proximal. Almost all of the incomplete bifaces exhibit manufacturing failures, including lateral snap (n=4) and perverse fracture (n=1). One Coastal Plain chert biface exhibits heat damage.

Three bifaces are made of Tallahatta sandstone, including a proximal end, a distal end, and one with the tip missing. All exhibit manufacturing failures: lateral snap (n=1), perverse fracture (n=1), and a lateral snap on the proximal and impact on the distal (n=1). One of these bifaces exhibits soft hammer percussion and two exhibit soft and hard hammer percussion. One hard hammer, indeterminate biface fragment made of Hollis quartzite was recovered, and one biface of undetermined raw material. This biface is a distal fragment and the type of percussion applied to the stone was not determined, but it exhibits a lateral hinge failure type.

Hafted Bifacial Tools
As previously discussed, a total of 121 hafted bifaces were recovered from Madison Park. In terms of technological class, the vast majority (n=89, 73.5 percent) of the hafted bifaces exhibit retouch. Twenty-one (17.4 percent) exhibit soft hammer reduction only, eight (6.6 percent) show soft hammer and retouch along the edges, and three (2.5 percent are indeterminate). Most of the hafted bifaces (n=91, 74.3 percent) were created from flakes. The remaining hafted bifaces were made from nodules (n=6, 4.9 percent) and indeterminate blanks (n=24, 19.8 percent). The hafted bifaces created from nodules are either quartz (n=4), Hollis quartzite (n=1), or Coastal Plain chert (n=1). Therefore, the most common manufacturing technique was to select a suitable flake and to reduce it by soft hammer and/or retouch to the desired size and shape. Minimal flaking went into this process and it does not appear that these tools were intended for an extended use life.

Sixty-three (52.1 percent) of the hafted bifaces are complete. Of the incomplete hafted bifaces (n=58), the most common manufacturing failure types are lateral snap (n=17, 29.3 percent) and perverse fracture (n=15, 25.8 percent) Of the remaining incomplete hafted bifaces, the types of manufacturing failures are haft snap (n=2, 3.4 percent) hinge (n=1, 1.7 percent) incipient fracture (n=2, 3.4 percent) lateral hinge (n=1, 1.7 percent), outre passé (n=1, 1.7 percent), and indeterminate (n=1, 1.7 percent). Four hafted bifaces (6.8 percent) exhibit thermal damage, but it is unlikely that this happened during manufacturing.

The Hamilton points (n=37) are made of Knox chert (n=27, 73.0 percent), Coastal Plain chert (n=6, 16.2 percent), quartz (n=3, 8.1 percent) and Tallahatta chert (n=1, 2.7 percent). Of these, 24 are complete and the weight for these artifacts ranges from 0.27 to 1.37 g, with an average of 0.45 g. For the complete Hamilton points, the average length is 16.8 mm, ranging from 10.4 to 31.4 mm, although only four Hamilton points were longer than 20 mm. The average blade width is 8.1 mm, with a range of 5.8 to 10.9 mm. Blade thickness ranges from 2.5 to 4.2 mm, with an average blade thickness of 3.1 mm. Finally, base width averages 11 mm, with a range of 7.9 to 14.6 mm.

Thirteen Hamilton points exhibit manufacturing failures, the most common being the lateral snap (n=6). Others exhibit perverse fractures (n=2), one outre passé, and one inclusion fracture. One Hamilton point has thermal damage, probably inflicted post-depositionally. Two Hamilton points exhibit haft snaps, but both have soft hammer percussion as the last type of manufacturing activity. Therefore, it is more likely that these failures occurred during manufacture rather than during use. Turning to technological aspects of the Hamilton points, all but five exhibit retouch as the final reduction technique. One Knox chert Hamilton is classified as indeterminate technology, and two exhibit soft hammer and retouch. Both of the latter have lateral snaps, indicating they were discarded due to manufacturing failure. Two Hamilton points exhibit soft hammer percussion only, one of quartz and one of Knox chert. The quartz Hamilton seems to have been discarded before completion for reasons other than failure. The Knox chert Hamilton exhibits a haft snap, indicating discard due to a manufacturing failure. Apart from one indeterminate blank type, all Hamilton points were manufactured from flake blanks.

Similarly, the Madison points are primarily made of Knox chert (n=31, 58.4 percent). Others are quartz (n=11, 20.7 percent), Coastal Plain chert (n=9, 16.9 percent), Tallahatta chert (n=1, 1.8 percent), and Hollis quartzite (n=1, 1.8 percent). A majority of the Madison points are complete (n=33, 62.2 percent). Weight of complete Madison points ranges from 0.16 to 2.5 g, with an average of 0.6 g. For the complete Madison points, average length is 17.2 mm, ranging from 11.3 to 32.8 mm, although only five are longer than 20 mm. Of the complete Madison points, average blade width is 9.1 mm, ranging from 5.8 to 16.0 mm. Blade thickness ranges from 1.3 to 6.9 mm, with an average of 3.1 mm. Finally, base width averages 10.8 mm, with a range from 7.2 to 16.4 mm. The most common manufacturing failure type is lateral snap (n=7), followed by perverse fracture (n=4). The remaining Madison points exhibit hinge fracture (n=1), incipient fracture (n=1), lateral hinge (n=1), and one is indeterminate (n=1). Two Madison points have thermal damage in the form of a pot lid on one face of each, damage unlikely to have occurred during manufacturing. Ten of the Madison points exhibit soft hammer reduction as the final reduction technique. Four of these are complete and do not have any use-related or manufacturing failures. The other six exhibit perverse failure (n=4), incipient fracture (n=1), and an indeterminate failure. The remaining 43 Madison points exhibit retouch as the final manufacturing technique. All but eight of the Madison points were manufactured on flake blanks.

Camp Creek points (n=15) occur in Knox chert (n=10), quartz (n=3), Coastal Plain chert (n=1), and an unidentified material (n=1). The average weight of complete Camp Creek points is 1.41 g, with a range of 0.80 to 2.71 g. Camp Creek point length for this assemblage ranges from 22.67 to 30.56 mm, with an average length of 26.61 mm. Average blade width is 12.08 mm, and range from 8.26 to 14.23 mm. Thickness averages 4.30 mm, with a range from 3.07 to 5.72 mm. Base width ranges from 10.06 to 15.2  mm, with an average of 12.58 mm. Nine of the Camp Creek points are complete and the remaining exhibit lateral snap (n=2) and perverse fracture (n=4). Regarding technology, a majority (n=7) of the Camp Creek points exhibit retouch, while the rest exhibit soft hammer reduction (n=4) and soft hammer with some retouch (n=4). Interestingly, none of the soft hammer reduction Camp Creek points have manufacturing or use failures. This could mean one of two things: what is considered finishing (retouch) of projectile points does not always apply to Camp Creek points, or these points were discarded for other reasons besides material failure. A majority (n=9) of the Camp Creek points were created from flake blanks, and the remaining point blanks are indeterminate.

Of the three quartz Gary points recovered at Madison Park, one is incomplete due to a perverse fracture. These Gary points were created from small nodules. Two exhibit soft hammer reduction only (one with the perverse failure), and the other exhibits soft hammer and retouch. This small sample yielded the following average measurements: weight (12.22 g), length (42.80 mm), blade width (27.67 mm), blade thickness (10.08mm), shoulder width (29.07 mm), stem length (9.15 mm), neck width (16.27 mm), and base width (12.92 mm).

The Hernando point recovered at Madison Park is Coastal Plain chert and was created from a core. This point exhibits soft hammer percussion and has a lateral snap manufacturing failure. Measurements are as follows: weight (4.12 g), length (39.28 mm), blade width (18.15 mm), blade thickness (5.41 mm), stem length (5.41 mm), neck width (9.25 mm), and base width (9.00 mm). Since the Hernando Point is not complete, the measured length does not indicate the actual length of the point.

One quartz Ebenezer point was recovered at Madison Park. This point is incomplete with a lateral snap failure and exhibits retouch. The point was created through core reduction. Measurements are: weight (2.78 g), length (23.19 mm), blade width (15.00 mm), blade thickness (5.35 mm), shoulder width (16.90 mm), stem length (8.49 mm), neck width (12.5 mm), and base width (9.61 mm). Since the Ebenezer point is incomplete, the measured length does not indicate the actual length of the point.

Of the three Bakers Creek points, two are complete (one Knox chert and one Coastal Plain chert), and one quartz point exhibits a perverse fracture. The quartz and Coastal Plain chert Bakers Creek exhibit soft hammer percussion, and the Knox chert specimen exhibits soft hammer and retouch percussion. The Coastal Plain Bakers Creek point was created through core reduction. We are unsure of the blank type of the remaining Bakers Creek points. Measurement average include weight (5.55 g), length (39.35 mm), blade width (18.34 mm), blade thickness (7.02 mm), shoulder width (19.22 mm), stem length (6.92 mm), neck width (14.74 mm), and base width (14.30 mm).
Drills. Two possible Knox chert drills were recovered during unit excavation. One is 1.0 cm long and the other is 1.5 cm long. Both exhibit soft hammer and pressure flaking scars and have lateral snaps (Figure 6-13). Since these artifacts seem to be unfinished, use wear analysis would not help determine if these artifacts are in fact drills.

Figure 6-13. Two “drills” manufactured of Knox chert: (a) FS 515; (b) FS 586.

Flake Debris  
63,519 pieces of debitage were recovered during unit and feature excavations. A majority (n=53,680, 84.51 percent) are quartz (Table 6-3). Other raw materials from the debitage assemblage are Knox chert (n=9,489), Coastal Plain chert (n=210), Tallahatta chert (n=70), Hollis quartzite (n=57), Fossiliferous Bangor (n=1), and unidentified raw materials (n=12).

Two methods of flake debris analysis, primary/secondary/tertiary (PST) and the interpretation-free method (Sullivan and Rozen 1985), are widely employed by cultural resource management lithic analysts, despite receiving considerable criticism (Carr and Bradbury 2000). The utility of the PST method has been questioned for the last quarter century (e.g., Ahler 1989b; Magne 1985; Patterson 1981; Sullivan and Rosen 1985) and has been likened to a random guess (Bradbury and Carr 1995). Despite these criticisms, it is often the only method used in the analysis of flake debris. In addition, it appears that many archaeologists using this method seem to think the terms so well known that definitions are unnecessary. In examining these definitions, Sullivan and Rosen (1985:758) found primary, secondary, and tertiary flake categories extremely unstandardized. A primary flake for one analyst could be considered a secondary flake by another (see also Ingbar et al. 1989). In essence, different analysts using this method can examine the same assemblage and come to completely different conclusions.

Table 6-3. Debitage by raw material.

Raw Material







Knox chert



Coastal Plain chert



Hollis quartzite



Tallahatta chert



Tallahatta sandstone



Fossiliferous Bangor


<. 01







Sullivan and Rosen’s (1985) “interpretation free method” has received criticism based on flintknapping experiments. This method focuses on flake breakage (complete, proximal, medial/distal, non-orientable fragment, split) and has the advantage of generally being replicable between analysts. However, the method has provoked adverse criticism of its approach to data interpretation (e.g., Amick and Mauldin 1989, 1997; Ensor and Roemer 1989). Application of the method to experimental data sets has not produced congruent results, which casts doubt on its utility (e.g., Amick and Mauldin 1997; Austin 1999; Bradbury and Carr 1995; Kuitj et al. 1995; Prentiss 1998; Prentiss and Romanski 1989). Some researchers consider this method useful for generally characterizing an assemblage or as a line of evidence to support or refute the conclusions of other analyses (Austin 1999; Bradbury and Carr 1995; Ensor and Roemer 1989). In particular, Bradbury and Carr (1995) suggest that relatively high percentages of shatter support an inference of core reduction as the focus of manufacturing activities.

Recognizing the limitations of many traditional methods of debitage analysis used in cultural resource management projects, while taking into account the limited resources available to investigate debitage, methods were employed that would most efficiently derive technological information from the assemblage. The debitage from Madison Park underwent a general mass analysis based on Ahler’s work (1989). Additionally, samples of quartz and chert flake debris underwent separate variations of individual flake analysis.

A lack of comparative flintknapping experiments using quartz pebbles as a raw material limits our ability to understand the flake debris produced during core reduction and tool production. And the nature of quartz fracture mechanics precluded a detailed analysis of flakes of this material. However, a sample of quartz flakes (n=5,132) was classified based on presence of cortex and whether specimens exhibit classic characteristics of bipolar percussion. This analysis was facilitated by informal experiments conducted in 2005 at the Center for Archaeological Studies. Quartz pebbles retrieved from the Coosa and Tallapoosa rivers were subjected to bipolar forces (Figure 6-14).

Quartz pebbles were placed on a sandstone anvil and struck with a quartz hammerstone with the intention of producing flakes. This resulted in a comparative collection of bipolar flakes for use in typological analysis. Andrefsky (2001:6-7) suggests that bipolar flakes are one of the more common debitage types in such analyses. Here, flakes were required to exhibit at least two of the following attributes to be classified as bipolar: crushing at both ends, exaggerated bulb of percussion and compression rings, flat bulb of percussion, cortex on platform/margin/distal end, or a wedge shape. Additionally, a comprehensive analysis of a sample of chert flake debris (n=702) was undertaken, following the general methods outlined by Bradbury and Carr (1995).





Figure 6-14. Bipolar experiment conducted by Philip J. Carr in 2005 at the Center for Archaeological Studies, University of South Alabama.

Copyright © 2013 by The University of South Alabama
Last Updated:
Thursday, July 23, 2015 9:01 AM