«By James R. Keron Graduate Program in Anthropology Submitted in partial fulfillment of the requirements for the degree of Master of Arts Faculty of ...»
5. Normal Unifacial Retouch The flake results from finishing and resharpening a scraper by detaching a flake with a blow to the ventral surface of the working edge. These flakes result from scraper manufacture and resharpening. See also Deller and Ellis (1992) and Frison (1968).
While this flake type was defined and included in the analysis here, few were identified in the collections probably because they were too small for normal recovery practices to capture.
Attributes $ Almost always complete flake $ Platform approximates the ventral surface of a uniface and is right angled. Small, circular to irregular in outline with a pronounced bulb of force $ Parallel scars on dorsal surface (old working edge) $ Use wear on working edge adjacent to platform $ Pronounced curvature $ Usually feathered termination
6. Ventral Unifacial Retouch Scrapers with ventral retouch are common on Iroquoian sites in the London area. These flakes are removed by a blow to the working edge perpendicular to the ventral surface of the scraper so that a flake is detached from the ventral surface of the scraper (see Deller and Ellis  and Frison ). In practice only the first few flakes detached in this process will be identified to this category. Any subsequent flakes detached will be very difficult to distinguish from flakes of bifacial retouch.
Attributes $ Dorsal surface is flat ( approximating the old ventral surface of the uniface) $ Platforms are right angled to the dorsal surface, faceted, show old use wear as approximate the old working edge of the scraper prior to flakes removal $ pronounced bulb of percussion and undulations as struck into a flat surface $ Flakes are markedly expanding $ Terminate in hinges or steps $ Lack curvature
7. Shatter This category derives from core reduction when fracture planes are encountered in reducing a core. Instead of producing a classic core reduction flake, portions of the core will break off along pre-existing lines of weakness encountered in the core creating blocky chert fragments. In some sites in the study area where natural chert is common in the till, distinguishing culturally created shatter from shatter produced by modern farming practices can be problematic. For a more detailed discussion see Ahler (1989) and Binford and Quimby (1963).
Attributes No clear ventral or dorsal surface No visible negative bulbs of percussion No systematic alignment of cleavage scars No orientation - distal or proximal, dorsally or ventrally Blocky fragments
8. Fragmentary Flakes This type is the distal portion of a broken flake.
Attributes $ No striking platform $ Clear dorsal and ventral surfaces $ Break termination proximally Chapter 5: Intra-Site Spatial Analysis This chapter provides an initial exploration into the intra-site patterning of the chipped lithic material in five sites. It tests the hypothesis, identified above, that, if the pattern of different Kettle Point frequencies in different middens at the Harrietsville site (Keron 1986) was evidence of the historically observed cultural trait where Iroquoian lineages controlled trade routes (Tooker 1991; Trigger 1987), then the pattern should be repeated at other Iroquoian sites.
In approaching this analysis, the simplest approach would be to do an analysis of chert frequencies from midden deposits from other sites similar to what was done at Harrietsville. However, while a number of middens have been excavated in the London area, there are not really any large scale excavations, so the sample would be very spotty.
There are however, a number of Controlled Surface Pickups (CSP) where surface material from the entire site is collected and the horizontal provenience carefully recorded primarily to determine the extent of the site. Given that lithic detritus occurs in large numbers and that the location of each piece is known, then an analysis of the chert source and flake type is capable of providing a great deal of information on internal patterning of activities within a site. Consequently, five sites that have had a CSP conducted on them were selected for internal analysis. These consist of two LOI sites, Brian (AfHf-10) and Cassandra (AfHh-65), two MOI sites, Drumholm (AfHi-22) and Dorchester Village (AfHg-24), and one EOI site, Mustos (AfHg-2).
In conducting the analysis, each flake was typed for the chert source and the flake type. This information and the catalog number of each was recorded in a spread sheet on an item by item basis. The horizontal provenience data occurred in two main forms, one where a transit had been used to record the location so that a distance and direction was recorded and the other where a compass had been used so that two directions were recorded from two known points. Both of these were turned into Cartesian co-ordinates using spreadsheet calculations available on the web site of the London Chapter of the OAS (Keron and Prowse 2001). Using spreadsheet look up functions, each flake was then matched to the appropriate find spot and the Cartesian co-ordinates added to it giving a spreadsheet that has the catalog number, the chert source, the flake type and the X-Y coordinates of the location. Further analysis was based on this file.
The next stage of the analysis examines the differences in the frequency of various flake and chert source types between different areas within the site looking for differences that are statistically significant. To conduct this analysis a Geographical Information System (GIS) was used to both map the site and perform the statistical calculations for each sub-area of the site using the file containing containing the types and location information. This procedure has the enormous advantage of greatly facilitating conducting many iterative Awhat-if@ analyses.
While the details of how this analysis was accomplished and the statistical calculations are included in Appendix B, a brief description follows. First, several GIS Amaps@ were produced showing the location of each flake of a specific chert or flake type.
For example, one map would plot all Kettle Point chert flakes and another would plot all bifacial retouch flakes. While it would have been informative to be able to drive the analysis down to the point where chert source and flake type were combined for example, all Kettle Point chert flakes of bifacial retouch, the samples were not large enough to permit this kind of analysis.
The next step in the analysis and the point around which a number of iterations occurred for each site was to develop a map of the site that would divide it up into various zones that could then be tested for differential distribution of chert source type and flake type. Ideally, this division of the site into analytical units of area should be done using culturally defined areas such as middens or longhouses. However, as we are dealing with CSP collections, the longhouse locations are not known. A general process was developed where the site was divided up into midden areas and non-midden areas. The non-midden areas were created by grouping all points within the site together that were closest to each midden. Thus, for the Brian site, there are seven defined middens and seven interior site areas, one interior area for each of the closest middens. See Figure 7 for an example. In other cases, where there were no defined midden areas the analysis had to proceed by trial and error. In the subsequent analysis, division into cultural areas is described for each site.
Once the cultural areas have been defined for a particular iteration, a GIS script was used to process the data. For the particular attribute under consideration (e.g. Kettle Point chert flakes), this script calculated, for each individual cultural area, the frequency of occurrence (number of occurrences of the specific type divided by the total occurrences of all types), the total occurrences of all types, the range of a confidence interval based on the sample size, and a map of the various areas with the analyzed flakes plotted. The calculated data are found in the legend. Where the confidence intervals do not overlap, there is a statistically significant difference in the particular attribute being considered in its distribution over the site. Where the confidence intervals do overlap, it was necessary to take the frequency and total of all types for the area to a spreadsheet for more detailed calculations to determine statistically significant differences. The differences that were significant were then shaded on the map using a feature of the GIS.
Thus, in analyzing chert source types, for example, four maps resulted, one looking at the distribution of unidentifiable or burned chert and three for each of Kettle Point chert, Onondaga chert and local till chert. AOther@ chert was examined initially but as the numbers are very small, no statistically significant differences arose so this was subsequently abandoned. For flake types, five maps were created, one for each of the predominant flake types (decortication, core trimming, bifacial retouch, fragment and shatter). Again, for a detailed description of the GIS methodology see Appendix B.
The analysis of distribution was applied to the detritus only and proceeded in two stages: 1) the distribution by chert source type was considered, 2) the distribution by flake type was considered. Initially, it had been intended to do the same analysis with the formal artefacts/tools but the samples were too small to develop any statistically significant trends in the data. A map of the analytical units is included for each site. All observations are statistically significant unless otherwise noted. Each observation is numbered for later reference, primarily in the table in Appendix E.
The tables of the calculated results comparing the differences between each pair of areas in each site are included in Appendix C and the areas that have frequencies that are different with statistical significance are shaded in the table. The observations are based primarily on these tables.
Brian Site Observations The Brian site (AfHh-10) is a late prehistoric Neutral village of approximately 2.5 ha. A detailed CSP was conducted on the site in 1988 by Peter Timmins for the London Museum of Archaeology. While earlier and later work was done on the site, this sample forms the basis of the following analysis. As there are well-defined middens, these were used to define the cultural areas as described above. There are seven middens and consequently seven interior areas defined by their proximity to the nearest midden. The cultural areas were assigned numbers as show in Figure 7. The midden numbers are those defined by Timmins (personal communication 2002). The data used to develop this analysis is presented in Appendix C: Tables C1 through C5.
Chert Type observations During the initial part of the analysis, it appeared as though burned chert and larger pieces of debitage like shatter and cores were more common in the middens.
Accordingly, an analysis was done whereby the site was separated into two areas, one the midden areas, and the other, all non-midden areas. The chert types and flake types were run through this analysis to the effect that there was little patterning that was statistically significant with the following single exception,
1. Local till chert was more often found in the middens than other areas of the site.
However, as this leaves open the possibility that different middens were used in different fashions so that one might be high in burned chert but another not, the second analysis was done using all the fourteen cultural areas as defined above. This analysis failed to show any differences. Most of the frequencies of burned chert were in the same general range.
2. The single exception was that Area 3 has a lower frequency of unidentified chert than did Midden 4.
For the remaining analysis unidentified chert was excluded from the calculations as it does not represent a real type in the sense that Kettle Point or Onondaga chert would be. Consequently, it was not included in calculations since inclusion would distort the actual frequencies of identifiable chert. For example, two sites with identical chert use would have different percentages of identified source cherts if one site had more of the total burned than did the other. This assumes, of course, that there was no differential burning of chert. In proceeding with the analysis of chert source type using the fourteen cultural areas, the following observations were made.
3. With respect to Onondaga Chert, Midden 1 has a lower frequency than the adjacent Area 1 as well as Area 2 and 3 and their middens. Midden 1 is outside the palisade. Within the palisade Areas 1, 2 and 3 have higher frequencies but the differences to the rest of the interior are not significant.
4. The highest concentrations of Kettle Point chert are in Areas 2 and 3 and their associated middens. The areas they differ from significantly are Areas 1 and 6.
5. In general, local till chert has high frequencies over most of the site with the exception of Area 7 which had a very low frequency. Unfortunately, this area has few flakes but even with a small sample the difference to the entire rest of the site is significant.
6. As to the internal distribution of local till chert, excluding Area 7 there are two other sets of areas which differ significantly from each other. Area 2 and 3 and Midden 2 have a lower frequency than Areas 1 and 6 and Middens 1, 2, 4 and 7.
Flake Type Observations In general, there is much less variation in flake types over the surface of the site than there is for chert source types. There are only a few significant differences and these involve one area of the site being different from a couple of others. Following are the significant differences.
7. Looking at the flake types both bifacial retouch flakes and shatter showed no significant differential distributions over the site.
8. Looking at decortication flakes, Midden 1 has a frequency that is significantly lower than Midden 4 and Area 6.
9. Flake fragments are lower in Midden 3 than the immediately adjacent Area 3.
10. For core trimming flakes, Midden 2 is significantly lower than Areas 1 and 3 and Midden 4.