«By James R. Keron Graduate Program in Anthropology Submitted in partial fulfillment of the requirements for the degree of Master of Arts Faculty of ...»
Critics of the lithic reduction stage methodology have noted several problems with the method and experimental replication of the types. Shott (1994: 77) notes three problems. First, while the attribute configuration of the flake is used to assign it to the particular type, it is only inferred that the flake derived from that stage of reduction.
Experimental results have shown that such assignation is not unproblematic (Ahler 1989:
88). Second, Shott (1994) criticizes the method because everyone uses a different typology making comparison of assemblages (or more properly comparison of published reports of assemblages from different authors) impossible. This criticism is not of the method per se but symptomatic of most published reports in archaeology where archaeologists assert their individuality (or maybe ethnicity) by using different analytical structures even if none are required. He does go on to include in this criticism the fact that different people will categorize a flake differently even with the same typology which is a major problem with the comparability of analyses (see Keron 2003 for an experimental demonstration). Third, he notes that the typologies based on this approach
do not take size into account (Ellis 1984 being an exception). Sullivan and Rozen (1985:
756) expand on Shott_s second criticism noting that typologies which might seem comparable are often not since, while the same _types_ or more properly names of types are used, the definitions are different. For example, while the Anames@ primary, secondary and tertiary decortication flakes are used, the definitions of the percentage of cortex required to distinguish one from the other vary. Also words such as Aprimary@ and Asecondary@ have different meanings in different typologies. Further, they introduce another criticism that the reduction process is best viewed as a continuum and not a series of discrete stages.
Stage typologies dominated until the mid 1980s when new approaches were defined and published. One of these, which has generated significant interest and controversy over the years, was that of Sullivan and Rozen (1985). Here they noted problems with stage typologies similar to those of Shott (1994) and, building on the point that the assigned types may in fact not be diagnostic of the particular reduction stage, proposed defining a series of four types that were not at all linked to stages but were Ainterpretation free@. The four types were defined by a simple set of attributes, basically, the presence or absence of a single interior surface, the presence or absence of a point of applied force and whether or not those flakes with a point of applied force were broken or not. The four types are called Adebris@ (no interior surface), Aflake fragment@ (no point of applied force), Abroken flake@ (point of applied force but not complete) and Acomplete flake@ (point of applied force and unbroken). It is necessary to analyze the assemblage at a site and then relate the frequencies of the various types to activities that were conducted at the site. In general, high frequencies of complete flakes and debris are taken to be indicative of core reduction while high frequencies of broken flakes and flake fragments are indicative of tool manufacture. This generalization is derived through a factor analysis of a series of sites where the two pairs of types tended to co-vary and more cores were present with high frequencies of complete flakes and debris.
The method has attracted considerable interest and criticism in the literature in what can best be described as a mixed reaction. The most obvious criticism is that their assignment of type frequencies to various reduction activities is just as arbitrary as the assignment of types in stage typology (Amick and Mauldin 1989a). They also argue that inferring activities directly from the archaeological record is circular and that experimental studies are required to validate the assumptions thus building Binford_s (1981) middle range theory. Amick and Mauldin (1989b) published an edited volume of experimental studies several of which focused on Sullivan and Rozen=s typology. Of these, Mauldin and Amick (1989b), Baumler and Downum (1989), Tompka (1989) and Prentiss and Romanski (1989) reported results contradictory to Sullivan and Rozen_s analytical construct while Ingbar et al. (1989) found that it was useful in their study. On the theoretical level Ensor and Roemer (1989) challenge the nature of the Ainterpretation free@ categories and then go on to reject Sullivan and Rozen_s criticisms of stage typology as unfounded and also their claim that the reduction sequence is a continuum, noting that knappers will go through a number of stages from hard hammer percussion to soft hammer percussion and pressure flaking. Most importantly, they note that the assignment of one flake to a reduction stage is not incontrovertible proof that that stage was executed but that there is a probabilistic relationship between the assignment and the actual stage occurring at the site in question. More recently, Prentiss (1998) developed an experimental design which showed the SRT to be reliable in that it could be replicated but that its validity was poor with highly vitreous raw materials. With the exception of use by the authors themselves (e.g. Sullivan 1987), the SRT method has attracted much more attention from a methodological and theoretical perspective than it has been used in practical applications.
The final methodological approach to debitage analysis is known as mass analysis (Ahler 1989) and is very different from the preceding two approaches which both require an analyst to examine each individual flake and assign it to a type. With mass analysis this labour intensive process is replaced by size grading an entire assemblage using a series of nested screens of progressively finer gradation varying from one inch through.5,.25 and.125 inch screens. A group of flakes is deposited in the uppermost layer and a consistent shaking process is applied allowing the flakes to settle down to a screen where they are too large to drop further. From this point the count of flakes and the total weight of the flakes in each layer is recorded. The theory behind this approach is that knapping is essentially a reductive process so that the further progressed the activity, the smaller will be the flakes. Further this can also be related to various techniques of knapping in that percussion techniques generate larger flakes than does pressure flaking. The technique has several advantages not the least of which is the rapidity of performing the analysis when compared to the tedious one-at-a-time approach used by the other methods.
Furthermore, smaller flakes resulting from pressure flaking can be included in the analysis as these are normally not even recovered with other methods. The method is consistent in that results can be replicated with a good degree of accuracy. Problems with the method include, similar to the other methods, the concern that the particular pattern recorded may be difficult to link to the behaviours that are being sought. A second unique problem is the difficulty of working with mixed samples. The method works reasonably well in experimental contexts in discriminating core reduction from tool manufacture but is much more difficult to apply when these are mixed as occurs on most archaeological assemblages. In general Ahler_s method, while allowing mass processing of assemblages has not attracted much attention. Further, while it can measure flakes smaller than.025 inches, this assumes that these flakes have been recovered. However, screens that small are rarely used in Iroquoian archaeological excavation and, of course, not at all in a CSP which will be biased against smaller items.
All three approaches provide a way to get at the underlying cultural processes but for further work it is necessary to select a general approach. Both SRT and mass analysis attempt to introduce greater rigor into the process by focusing on better replicability of the analysis. Despite this advantage, it is felt that it would be best to use a stage typology given the shortcomings of the other methods. The criticisms of stage typology must be considered so that more is not read into the data than can reasonably be assumed.
Besides, the stage typology approach is very much ingrained in current research in Ontario so that there is an attraction to staying with this approach as long as it can be shown to provide consistent results.
One of the critical areas to be determined was the extent to which published sources could or could not be used in the comparison phase. Use of published information would be attractive as a broader range of sites could be included but could potentially inject spurious results because of subtle differences in classification.
Accordingly, an experiment was designed whereby the author and other analysts classified the same set of 200 flakes taken from London area Iroquoian sites according to chert source and two different stage typologies. The details and results of this are published elsewhere (Keron 2003) but the conclusions were as follows as they relate to
$ An error factor between analysts of up to plus or minus 7% would need to be applied to analyses of different authors.
$ It might be possible to translate one typology to another as long as the second had fewer types and there is a one-to-one mapping from the first typology to the second.
$ The assignment of any given flake to a type is not generally agreed upon by the analysts. Generally agreement was achieved between any two analysts on only 60% of the sample.
$ More complex typologies lead to less agreement between analysts.
Given these facts, it became a choice between using a simpler typology such as that of Lennox which would allow some comparison but with a degree of error or to move to a more complex typology that would permit a deeper understanding of the reduction sequence but require all collections to be personally examined. The decision was made in favour of the latter.
One of the very real criticisms of various stage typologies is that a number of those being used are not at all well-defined. Thus, in moving into this exercise, it was deemed critical to be very clear as to the definition of the attributes of the types to be used. This presentation will induce greater reliability into the process and hopefully might allow the typology to be used elsewhere with some success. The detailed definitions of the various subtypes of debitage follow.
Debitage Subtypes In defining the typology to be used here, the first decision was whether or not to use one of the established typologies most likely that of Paul Lennox given its status as de facto standard. While that idea had some appeal, there were several open questions as to exactly what went in any given category. Furthermore, there were other variables that seemed important that would be lost by using a Alumper@ typology such as decortication flakes or unifacial retouch flakes. Thus, it was decided to proceed with yet another typology but, in doing so, to provide a detailed definition of the attributes of each defined type.
The following defines the flake types used in this thesis.
1. Decortication Flake This is the initial stage in core reduction involving the removal of the cortex from the initial block of chert. The category as used here includes White=s (1963) primary and secondary decortication flakes. Cortex as defined includes the original interface to the surrounding source matrix, or weathered surfaces or surfaces with patina. Weathered could include both crushing from action in the glacial till or smoothing through being water rolled.
Attributes $ Presence of substantial (20%) cortex located other than on the striking platform $ Striking platform has very few facets (3) $ Striking platform at approximately 900 to ventral surface $ Pronounced bulb of percussion $ Dorsal surface has low number of scars $ Generally, on average it is a larger size but there is a considerable range of variation $ Ventral surface is usually straight lacking much curvature
2. Core Trimming Flake These flakes are produced during core reduction. They can be flake blanks or attempted flake blanks where the purpose is to obtain a flake for either expedient use or manufacture into a more formal tool. They also include generally smaller flakes which were removed in the process of preparing the core for the removal of a flake blank. Flakes in this category are generally indicative of the later stages of core reduction as they lack cortex.
Attributes $ No cortex on surfaces other than the striking platform or less than 20% of the dorsal surface $ Striking platform has few facets (3) $ Striking platform at approximately right angles to ventral surface $ Bulb of percussion present $ Dorsal surface has relatively low number of scars $ Variable sizes. Larger flakes may be blanks, smaller are preparation flakes
3. Bipolar These are flakes produced during bipolar reduction when a core either becomes too small to work with freehand percussion or is relatively small to begin with. The core is placed on an anvil stone and then is stuck with a hammer. For a detailed discussion see Ahler (1989) or Hayden (1980).
Attributes $ Shattered or pointed platforms with little or no surface area $ Evidence of force at both ends of the flake (These first two are key in distinguishing this from Shatter discussed below) $ Angular polyhedral cross section $ Steep lateral edge angles $ Lack of definite positive bulb of force $ Pronounced ripple marks $ Lack of distinction between dorsal and ventral flake surfaces
4. Bifacial Retouch Flake This type results from flake removal during biface reduction. As Iroquoian bifaces are generally small these flakes are also small, being around 1 cm in size. In earlier times, bifacial cores were used but these are not found on Iroquoian sites. For a description see Deller and Ellis (1992) and Frison (1968).
Attributes $ Thin and flat transverse cross section lacking pronounced dorsal ridges $ Thin longitudinal cross section $ Frequently curved so the flake is concave on the ventral surface $ Feathered edges both laterally and distally $ High number of dorsal flake scars $ Striking platform faceted, narrow, lipped, sometimes ground $ Little or no cortex on dorsal face $ Expanding flake shape $ Small or subdued bulb of force.
$ Obtuse platform to ventral surface angle $ Acute platform to dorsal angle