«By James R. Keron Graduate Program in Anthropology Submitted in partial fulfillment of the requirements for the degree of Master of Arts Faculty of ...»
12. The Neutral village trend line is best approximated by a logarithmic equation. The R-Squared value is high at.7857 (Figure 14, Table D-2).
13. The trend line of all the Neutral cabins is more problematic. The R-Square numbers for all equation types are low indicating variability in the data that is not a simple function of the distance (Table D-2).
14. The R-Squared values for Neutral cabin sites are higher if the two outliers are removed. However, there is no good reason to eliminate these.
15. There is a very distinctive difference between the regression line for the Glen Meyer period and that of the Neutral villages. These have been plotted together in Figure 16.
Onondaga Chert Distance Decay The analysis of the use of Onondaga is somewhat more problematic. The analysis, proceeding similarly to that of Kettle Point chert, involved creating a table that has the percentage of Onondaga chert and the distance from the source (Appendix D, Table D-1). The percentage is drawn from the data in Appendix F. Locating the source of Onondaga is not so easy as it outcrops in various locations along many miles of Lake Erie shoreline. As all of these outcrop locations are in the same general direction from the study area, the initial analysis selected a point on the shoreline east of Port Dover.
The UTM for this point was 4,740,000 north and 570,000 east. The distance to each site was calculated as the distance between this point and the UTM grid reference for each site. This information was then plotted as a scatter plot with the distance as the X-axis and the percentage as the Y-axis. Again, each of the four time periods was plotted as a distinct set of points so that changes over time would be apparent. The results of this analysis is shown in Figure 17.
This diagram, however, shows no obvious distance decay in any of the periods which, at least in theory, presents a problem. It is unreasonable not to expect some form of distance decay thereby suggesting either a problem with the data or some other confounding factor. One of the factors which has been shown to distort distance decay functions in the past has been ready access to water transport (Hodder and Orton 1976;
Luedtke 1976). Janusas (1984) in her analysis of the distribution of Kettle Point chert actually used the seeming anomaly of high percentages of Kettle Point chert on Petun sites to infer that water transport was a factor. With the Onondaga chert sources located along Lake Erie it is not out of the question that chert was transported in bulk along the lake shore by canoe. Also, Ellis (personal communication 2003) reports that there are secondary deposits of Onondaga chert along the Lake Erie shoreline and Kenyon (1980) actually maps out secondary deposits of Onondaga-like material along the Lake Erie shoreline and inland.
Consequently, in an attempt to locate a distance decay function, the Asource@ was arbitrarily moved to Port Stanley with UTM coordinates of 4,723,000 north and 482,500 east producing Figure 18. Again there is no obvious distance decay function. However, it should be noted that the orientation of the study area while ideal from the perspective of sources to the east and west is not very good for a source to the north or south. The study area has a long east west orientation but is shallow in a north-south direction leaving little room for even observing a distance decay function. Most of the sites occur within a range of 27-40 km of Port Stanley making identification of a southern distance decay function problematic. Furthermore, there is an implicit assumption in conducting distance decay that the source is localized. If the source is more dispersed, distance decay can be problematic.
Morphological Variation This next phase of the analysis examines the morphological flake types across the various chert sources and through time looking for patterns. In order to facilitate the analysis, all of the raw flake counts from Appendix F were brought together in a single spreadsheet along with the information on cultural period and site type from Appendix A.
The raw flake counts were combined by cultural period giving three tables (Appendix D Tables D-3, D-4 and D-5). These tables break down the five common flake types against the four chert sources plus unidentified chert. The vertical columns are the percentages of all flakes types for that particular chert source. As the uncommon flake types (bipolar, unifacial retouch and ventral unifacial retouch) are not included, the columns will sum to something slightly under 100%. Also, given that there were only three Uren sites in the sample, these were combined with the Middleport substage sites to create one table for the MOI. Across the top of the table the total percentage of all flakes identifiable as to chert source for that time period are shown. As discussed earlier, unidentified flakes are excluded for that calculation. The flake type percentage column includes all flakes including unidentified. One other issue encountered during this analysis was that it seemed, especially for the Brian site, that the numbers being used here were out of synchronization with the data being used in the intra-site spatial analysis in the preceding chapter. Upon closer analysis it was found that the earlier surface collection from Brian differed from the CSP sample. As it seems reasonable that the CSP collection representing one intensive systematic collection of the site would be the most representative of the site as a whole, it was decided to exclude the earlier material from this analysis for the Brian, Cassandra, Drumholm and Dorchester sites. The CSP sample from Mustos constitutes the entire collection from that site. In the case of Brian this reduced the percentage of Onondaga chert by about 5% as it was more common in the original sample. This result again underscores the issues around how representative a given sample can be of each site. Anything short of a systematic surface collection of the entire site will be somewhat suspect.
Examining these Tables D-3 through D-5 leads to the following observations.
1. All forms of chert can arrive as nodules from any chert source. Each chert source in each period shows the presence of decortication flakes and shatter.
2. Onondaga arrives in a more reduced state than Kettle Point which in turn arrives in a more reduced state than local till chert in all periods. The percentage of decortication flakes and shatter is lowest for Onondaga and highest for local till chert. Kettle Point is always in the middle. There is always the risk that the differences between Onondaga and Kettle Point chert result from the nature and size of the nodules being imported. If the Onondaga nodules were larger there would be proportionately less decortication and given the propensity for Kettle Point chert to shatter more often there could be relatively less Onondaga shatter.
On the other hand what the analysis did not capture was the relative quality of the knapping of each of the source chert types. From a qualitative perspective, Onondaga debitage is generally of greater workmanship than Kettle Point. So this observation is still reasonable albeit a little weakened by the possibility of different size nodules being imported. Also, while the Onondaga beds are relatively much thicker, if secondary sources along the lakeshore were being exploited, it would be reasonable to expect that these would be reduced in size from what originated in the original bed.
3. Onondaga chert is preferred for biface production. Local till chert is least preferred and Kettle Point is in the middle. The percentage of flakes of bifacial retouch on Onondaga is higher than Kettle Point which is higher than local till chert. Also, the percentage of fragmentary flakes shows the same pattern.
4. The percentage of unidentified flakes is relatively high for fragments. At this point this result could be an analytical anomaly. As explained earlier there are two reasons for a flake to be assigned to the unidentified category, if it was burned or if it was too small to permit identification. In conducting the analysis, it was observed that in most cases there were a lot of unidentified fragments. This result could simply be a product of the fact that, when a flake breaks, there may be several small fragments produced that are difficult to identify.
5. The amount of unidentified shatter is higher than the other chert source types.
6. In all time periods, it can be inferred that Kettle Point chert is more preferred for creating expedient tools as the core trimming type has a higher percentage for Kettle Point than the other chert sources indicative of more of the resource being used to create flake tools as opposed to more formal tools.
7. While the regression analysis looked at Onondaga and Kettle Point chert individually, summing both of these together, the Glen Meyer peoples have the best access to high quality chert, and the MOI has the most restricted access.
Given the strong distance decay function that occurs in Neutral times it is difficult to compare to earlier times. Westerly sites like Lawson have an almost obsessive access to Kettle Point chert whereas the more easterly sites do not have good access.
8. The use of local till chert is lowest for Glen Meyer and increases in the middle stage. As with the preceding observation, the Neutral use of local till chert depends greatly on site location.
9. Through time there is a tendency for chert to arrive at the sites in a less reduced state. The percentages of decortication flakes and shatter increase with time.
10. Onondaga chert fragments decline steadily through time.
11. Use of local till chert for bifacial retouch peaks during the MOI when access to higher quality chert is constrained.
AOther@ chert is low in total counts and as a percentage of the entire sample so that 12.
sampling error makes any significant observations impossible. However, it should be noted that the percentage of bifacial retouch flakes is in the same range as Onondaga and Kettle Point chert.
13. The percentage of each of Onondaga chert and Kettle Point chert devoted to biface production increases dramatically over time doubling from Glen Meyer times to the MOI and Neutral periods. Part of this increase may be due to the fact that high quality chert is less accessible than in other times and with the need for production of bifaces for arrow points, more of the high quality chert is directed to this necessary activity as local till chert is usually inadequate to fill that need.
There are some limitations with respect to the data in these tables. First, the use of percentage as a means of comparison creates some problems. For example, if a new reduction technique is added to an existing industry increasing the number of flakes in one category, then the other categories will show a reduction in their percentage when in fact no change in the activity producing those ratios occurred. However, there is no easy way around this problem and the important thing is that such a shift in reduction strategy would create change in the data and will not go unnoticed. The above observations have all been reviewed with this problem in mind and are not unduly impacted by it.
A second issue involves site locations. For the specific tables to be comparable, ideally all sites so compared should be with in a few kilometres of each other. For example comparing the use of Kettle Point chert from Glen Meyer times in the Caradoc sand plain to the use of Kettle Point chert at the Uren site on the Norfolk sand plain would be nonsensical. The difference in distance from the source would be the single dominating factor rather than any underlying cultural factors. However, the only chert where this would be problematic is Kettle Point as no decay factor could be found for Onondaga and local till chert is accessible everywhere. In general, Glen Meyer is distributed throughout the study area and the middle stage sites tend to congregate in the central area and have a good overlap with Glen Meyer sites so these numbers will be generally meaningful ones. However, the Neutral sites are mostly concentrated in the eastern half of the study area and are subject to a rapid distance decay. Thus, the average Neutral period statistics for Kettle Point chert use are not very meaningful since it will fluctuate with the size and location of the samples included in the calculation. However, the only line affected by this concern will be the one showing the average chert usage per period (Total % at the top of each table). The other percentages in that line will also be impacted by the change in this percentage as discussed above. The observations have been examined and are not dependant on this situation.
Given the way that the tables were calculated, there was a risk that one or two sites that had the highest number of flakes would outweigh other sites with much fewer flakes. For example, a site with 1000 flakes would have five times the weight of another with 200 flakes. Accordingly the site counts were recalculated prorating all sites to 1000 flakes and the tables recalculated effectively averaging the averages and reexamined. The result is that a few of the observations were impacted and were adjusted accordingly despite the fact that only the first set of tables are included herein.