«By James R. Keron Graduate Program in Anthropology Submitted in partial fulfillment of the requirements for the degree of Master of Arts Faculty of ...»
Villages vs Cabins One of the areas of interest in understanding Iroquoian chert acquisition is the differences between the various types of site. As has been demonstrated in the vicinity of the Lawson site, there are a number of smaller sites nearby that have been interpreted as agricultural cabin sites (Pearce 1996). The expectation would be that the different activities conducted at these sites would at least involve different chert use and possibly acquisition patterns. In order to evaluate this influence, it should be necessary to simply compare the collected data for cabin sites and villages. However, there are some problems. First, there is a problem with the Glen Meyer sites in that the definition used by Williamson (1985) to separate cabin sites from villages was based on size and density of scatter. Anything under one hectare was called a cabin site and anything over this amount was a village. This determination was the best possible, short of excavation, but was arbitrary and based on what was then known about later period cabin sites and villages. Further, the Calvert village (AfHg-1) would be called a cabin site by this definition although Timmins refers to it as a village. Given the Glen Meyer propensity to reuse site locations, what was once a village at one point could be rebuilt later as a cabin site as happened at Calvert. Looking at the MOI sites in the analyzed sample these are all villages with one exception so cabin sites are not well represented.
Tables D-6 and D-7 in Appendix D were produced by sorting the data into time period and then by type of site. Totals of all flakes were produced for Neutral villages and hamlets on one hand and cabin sites on the other. Again, there is a concern that sites with greater numbers of flakes would skew the results. Therefore, the numbers of flakes in all sites was prorated such that they had the same number (in effect averaging the percentages by site). This procedure provided a cross check on the observations and as above, all observations have taken this into account. The second set of tables is not included.
The following observations can be drawn regarding the difference between village sites and cabin sites.
1. There is more local till chert used at cabin sites.
2. There is more Onondaga and Kettle Point chert used in villages.
3. More primary reduction takes place in cabin sites than in villages. Decortication flakes and shatter are higher for cabin sites.
4. Biface reduction is slightly more apt to happen in villages.
5. Core trimming flakes are more often present in villages.
One of the areas of concern with this approach is the impact of the distance decay pattern evident in LOI times. If, in the selected sample, more village sites are found to the west and more cabin sites to the east this could invalidate any observations regarding the relative use of the chert source types. Examining the site locations though shows a mix of sites across the study area. All three clusters used in the analysis have both village sites and cabin sites in the sample used to derive the percentages so this concern should not be a problem.
Formal and Informal Artefact Variation This discussion focuses on the formal and informal artefacts exclusive of the debitage or waste flakes. In analyzing the tools, it quickly became apparent that a number of the sites had very small samples so any comparison would be problematic.
Furthermore, as there were twenty-one classes of artefacts recorded against five chert types, a majority of the classes have no counts at all in any given site. In order to summarize these data for analysis, first, the numbers of artefacts were totaled for all sites and only sites with more than 25 artefacts were used in the following comparison. This procedure left eleven sites for comparison. For each site, two tables were built, one for formal artefacts and one for informal. The informal category included cores (random and bipolar) and wedges. The formal artefacts included everything else. The counts of these two classes were totaled on the site summary spreadsheet and the totals brought to another spreadsheet for the artefact comparisons. The first approach to comparison involved calculating for each chert type and site, the artefact to debitage ratio. While this would have been very informative, it had to be abandoned when it was realized that for most sites this ratio would be meaningless: only a sample of the flakes had been analyzed, while all the artefacts had been. This problem made it necessary to deal with percentages as opposed to actual counts so the next attempt, involved deriving the percentage of the four chert types (excluding unidentified) and then comparing this percentage to that derived from the debitage for the eleven sites. To accomplish this comparison the two Tables D-8 and D-9 in Appendix D were produced by subtracting the percent of the each chert type for both formal and informal artefacts from the percentages derived from the debitage. The leftmost four columns are the percentages of formal artefacts of the various chert source types and the next four columns are the differences from the percentages of the four source types as found in the debitage. The final column calculates the Robinson-Brainerd coefficient of similarity between the class of artefacts and the debitage for each site. Across the bottom under the four columns of differences is the simple sum of the differences. The intent of this procedure is to look for the overall trend within that chert type across all sites.
In the following discussion, the category Aother@ chert is ignored as the numbers are too small. Looking at the total line in these two tables the following observations can be made.
1. The percentage of Onondaga chert in formal artefacts is higher than the same chert in the debitage.
2. Conversely, the percentage of local till chert in the formal artefacts is generally lower than in the debitage.
3. Kettle Point chert seems to have similar percentages in both formal artefacts and debitage.
4. For informal artefacts, the situation is reversed. A larger percentage of the informal artefacts are of local till chert than is the case with the debitage.
5. Both Onondaga and Kettle Point chert informal artefacts occur at a percentage less than the debitage.
6. Considering the coefficients of similarity, there is considerable variance by site.
In some sites the artefact classes occur in very much the same percentages as the debitage whereas at other sites they are quite different. Sites with the coefficient of similarity less than 150 are highlighted as are the largest differences by chert source type.
To examine any trends through time, two more tables (D-10 and D-11 in Appendix D) were produced that show the average difference for each of the three time periods. The average difference is shown as well as the average coefficient of similarity for the sites shown in these two figures.
Looking at these tables the following observations are evident.
7. During the Glen Meyer period, the percentage of chert types in the formal artefacts is very similar to the same percentage in the debitage.
8. There is a drop in similarity between formal artefacts and debitage frequencies in the MOI and the Neutral period.
9. The shift to the MOI and Neutral period is accompanied by a greater percentage of Onondaga chert being used for formal artefacts and a corresponding drop in the occurrence of formal artefacts of local till chert.
10. Kettle Point chert use is similar across all periods.
11. There is a comparable drop in coefficients of similarity from the EOI to the MOI.
12. Unlike the formal artefacts though, the coefficient of similarity between chert sources of informal artefacts and the debitage rebounds during the Neutral period exceeding what it was during the Glen Meyer period.
13. During the Early and Middle stages, there is a tendency for more informal artefacts to be made on local till chert than occurs in the corresponding debitage.
14. For informal artefacts both Onondaga and Kettle Point chert occur less often than in the debitage.
15. The preceding two trends are exaggerated a little from the EOI to the MOI.
16. During the LOI, informal artefacts occur in similar percentages to the debitage.
Some caution should be attached to the preceding observations. There are not many sites included and the sample sizes are small. However, the observed differences do seem to be plausible.
Change in Lithic Industry Through Time It is also possible to attempt to quantify the changes over time in the overall lithic industry. Looking at chert source usage a pattern of highly fluctuating use of Kettle Point chert through time has been demonstrated. However, substitution of one chert type for another does not necessarily change the reduction industry per se or the uses to which the end products were put. Ideally, to explore this possibility the comparison should include both formal and informal artefacts as well as the debitage. Given the low artefact counts in most sites, sampling could well drive spurious results so while this would certainly be informative, it is beyond the capability of the data acquired for this study. Use of published material is clearly out of the question given the varying definitions of artefact types, the highly questionable practices used to assign artefacts to a specific type, erroneous assignments to a type and the failure to even recognize some artefacts between various investigators. A good example of spurious conclusions that can arise by comparing published reports was the comparison of the differences between Pickering, Glen Meyer and Uren lithic industries by Wright (1992). While the coefficients of similarity reveal real differences between the three components, the differences are the result of varying archaeological analytical techniques not the underlying cultural units.
Consequently we are left with the debitage broken down by flake types. In order to compare these the coefficient of similarity was calculated for each pair of the three periods based on the percentages shown in Tables D-2, D-3 and D-4 in Appendix D.
The results are illustrated in Table D-12 in Appendix D with the following observations.
1. The debitage industry is highly similar for all time periods.
2. There is a slight change through time.
Site Differences from Period Average While the preceding sections focused on the differences between the averages of several sites both by period and by type of site, this approach fails to take into account the meaning of individual site variation within the same class of sites. The intent of this next section is to identify several sites that are most removed from the normal pattern and to explore how these sites differ from the others. This analysis does not relate directly to any of the behaviourial hypotheses and is best considered as exploratory. To identify deviant sites, three different analytical approaches were taken, and five sites were then selected that consistently ranked high on two or more of the analyses. To some extent the selection criteria and the assumptions used to flag deviant sites are arbitrary and not based on statistical significance. However, the intent was to select five of the most deviant sites and to explore the differences so this approach is warranted.
First, the differences between each site and the average of all sites of the specific period were compared. Any situation where a particular site deviates significantly from the average probably carries information regarding the chert acquisition and use patterns occurring at that site. Accordingly, in the original spreadsheet, a portion of which is shown in Appendix F, containing the summary for each site, the average for the particular time period was added and the differences between the site and the period average computed. One of the problems quickly encountered concerned situations where the actual counts from the specific site were very low. For instance AOther@ chert generally has very low counts and consequently the presence or absence of one flake can make a significant shift in the percentage. Consequently it was necessary to discount the differences that occur with respect to AOther@ chert. Similarly if the total number of flakes from the site was low, any comparison from that site would not be valid and consequently, a number of sites had to be ignored. For this analysis the number deemed to be worth comparing was set arbitrarily at 100 flakes. The next issue in the comparison was to determine what differences were of sufficient magnitude to be of interest. This magnitude was set arbitrarily at 10% for the total chert source percentages and the total of flake types. In looking at the combinations of chert source and flake types, this limit was increased to 15% since the reduced numbers made sampling error a real problem. With this in mind a set of notes was made recording every area exceeding these levels. For Neutral sites the distance decay of Kettle Point chert creates some significant differences.
While noted, these were not considered except where a site was significantly different from others nearby (e.g. Laidlaw and Skinner).
Second, a table was built calculating the Robinson-Brainerd Coefficient of Similarity for the total flake types from the site. Generally there were quite high coefficients of similarity for most sites. This table was examined and any coefficients less than 160 were highlighted. Then any sites that were frequently different were identified. This table is included in Appendix D as Table D-13.