In summarizing, the first issue of note is the statistical significance (p < .01) of the
numerical discordance of males and females. This discordance can only be hypothetically addressed.
If the difference is related to some social mechanism whereby extra women and girls were being adopted into the society, one possibility would be polygamy, where the local population needed to seek outside the community for women. However, the early Copts were monogamous. Another possibility is persecution of Christians in the early centuries AD by the Roman Empire. In this scenario, women and children were potentially being sent here to stay with relatives to avoid the larger urban centres, where persecution was more intense. Thus, if these women arrived in this manner and were related to local people, then the hypotheses outlined herein would not be significantly impacted. However, Brown (2013) using Smith’s Mean Measure of Divergence, has demonstrated that the pre-Christian population is statistically part of the same deme as the K2 Christian population, so that a large influx of unrelated women seems unlikely or at least we would not be able to evaluate this proposition.
More likely is a mechanism that removed men from the population. It is unlikely that the mortuary program is singling out a significant portion of the men for alternate treatment, although there is a possibility that some status males are being buried elsewhere in the cemetery in an unexcavated area. The removal of 30% of the male population for special treatment does, however, seem excessive. A more likely explanation is that many of the men were dying elsewhere and not being returned for burial. Given the patriarchal nature of the society, it is likely that the men would be traveling to more distant locations than the females and thus more likely to die elsewhere. In support of this scenario, Haddow (2012) notes the male involvement in the caravan trade. If true, the ratio observed here does suggest that travel in those times carried an increased risk of dying. Whether it is disease, crime or religious persecution, the chances of not returning home at some point seems to be high.
Regardless, the disparate sex ratio does not invalidate the kinship versus accretionary hypotheses on the distributions within the cemetery. It may, however, make testing for a patrilocal interment more difficult by reducing the male sample size.
The second issue arising from the above analysis is a tendency for male burials to cluster together primarily at a range of five metres. The best indication of this tendency is the Proximity Count statistics in Table 5-1. The tendency for male clustering is also evident in both the LDA at the five metre distance and in the Nearest Neighbour calculations. What are the implications of this clustering towards evaluating our hypotheses? First, it should be noted that the size of this cluster spike is fairly small at five metres. It is not the case that major sections of the cemetery are reserved for males and others for females. In the accretion model, these small clusters would be highly unlikely, but they do tend to suggest that something might be taking place at the family level so, if anything, these micro-clusters of males are more in line with the kinship hypothesis. What it does mean for the analysis of the distributions of the discrete traits is that clustering of a discrete trait in the males could show as being closer than expected by chance. Regardless, if the males are all related in these clusters, it can be used to evaluate the kinship hypothesis.
However, the other possibility for the presence of male clusters in K2 is a confounding effect, since their absence may not be random. If males tended to be missing more often within some families than others, the results of the statistic would show as an apparent clustering of females when, in fact, it may have been evenly distributed if all of the missing males were buried with their families. As will be shown below, this factor does seem to be relevant.
When it comes to the smaller sample of individuals where the cranial traits are being evaluated (see Figure 5-5), some of the male clusters in the larger sample are not as well- defined. The results of the Proximity Count run against the actual sample being analyzed are shown in Table 5-4. In this case none of the values are significant.
Table 5-4: Proximity Count - Males and Females Small Sample
Table 5-4
Males n = 71 Females n = 107
Distance Count p Count p
3 31 .87 77 .83
5 98 .32 206 .77
7 180 .34 396 .57
10 310 .92 790 .18
15 598 .97 1569 .13
While part of this is a result of the reduced sample making statistical significance less likely, the nature of the sample also eliminates or reduces the size of the male clusters identified in the larger sample, particularly the male cluster in the southwest, and also the northwest, leaving a more even distribution of males.
In sum, it appears that the two main hypotheses are not significantly impacted by either the reduced number of males in the cemetery or by the relative distribution of males and females in the smaller sample. Therefore, these data can be used in the analysis of the distribution of discrete traits. However, the uneven distribution of missing males is a confounding factor.