Statistically Orientated

Alun Salt is a blogger and scholar whose abnormally interesting work borders on many of my own abnormal interests. He has a new paper at PLoSOne, “Astronomical Orientation of Ancient Greek Temples.” Alun studies the orientation of 41 Greek temples on Sicily and glances at temple orientations elsewhere. He plans this as the first in a series. 40 of his 41 temples faced the eastern half of the sky.
This is quite a distance away from my expertise but that won’t keep me from making a few observations.
With the observed data being 6.09σ away from a random set of orientations, I think it would be perverse not to see cultural forces at work in determining temple orientation. So does Alun. When one narrows the definition of east from half the sky to the portion of the horizon over which the sun rises, the results become even more statistically significant even as the number of eastern facing temples drops from 40 to 38. But given the less clear situation on the Greek mainland, I think Alun’s suggestion to apply the same statistical methods to the orientation of Greek temples among the Black Sea colonies and in Hellenistic Asia is important.
Alun also makes an abnormally interesting pitch for science journals and open access publication,

Archaeologists and historians do not usually have basic statistical analysis as part of their tool-kit, and so it is possible for papers without such statistical underpinnings to appear archaeological and classical journals because the basic problems simply are not recognised. Therefore research using statistical tools needs to be published either in a forum used by statistically minded scholars, or else in an open-access environment where they can access and critique the work. [reference omitted]

Alun explains why he chose PLoS specifically on his blog.
While I think Alun was correct in choosing a scientific journal rather than a more traditional archaeological journal, I worry that if this becomes a major trend the problems of the “failure of interdisciplinarity” that Alun correctly laments will only be exacerbated. Archaeologists need to learn to deal with and critique this kind of analysis. Many younger ones and a few older ones can but far too many cannot.
Alun’s paper outlines a few additional areas of research. I worry that he missed an important one. While I think Alun has done an abnormally interesting job with the statistical analyses, I think the nonconforming temples in the data set deserve more study. Why don’t they conform? Alun is correct when he concludes, “It is significance in the distribution of the set, rather than the alignment of any individual temple that matters in finding evidence for an overall preference. I believe discussion above does show there was a significant preference for easterly orientations in the alignment of ancient Greek temples.” But the non-conforming temples cannot be studied in aggregate and they may tell every bit as interesting story as those that conform. In fact, it is just possible that they may hold clues to the reason that the majority of temples face easterly.
I also found it understandable but a bit disconcerting that Alun needed to spend so much time explaining basic statistical analysis. That goes back to the “failure of interdisciplinarity.”
Go read Alun’s paper and see what you think. Its open access. Also, take a look at his photo of The Wall of Temple B, Naxos and his whole collection of temple photos.