page 19. two of the groups you see deviated toward the chinese are austro-asiatic speaking tribes. one of whom, the santhal, have a presence in bangladesh and bengal as a whole. it seems likely that a lot of bengalis are indo-europeanized santhals.

here’s the rub for genetics: who were the tribal people in eastern bengal? it may be a substantial proportion were austro-asiatic types who were the substratum across much of southeast asia. so one need not posit recent ancestry:

http://upload.wikimedia.org/wikipedia/commons/6/6f/Se_asia_lang_map.png

For instance, if you had photographs of a sample of individuals volunteering for such a study and had a random group of people rate pairwise similarity of their faces on a scale of 1 to 10, or even ask a question like how similar (or ease of “passing for”) a “European”, “African” or “East Asian” does individual X look, on a scale of 1 to 10?

Would you then might see people’s social categorizations match such genetic groupings (though I’m not familiar enough with stats to know how’d it be done)– if no one has done it, it would definitely be interesting.

On a related note, I know the old-school craniometric groupings of
“races” are considered obsolete, but have you seen any studies that have actually used morphological-type data (eg. nasal breadth, head width etc.) as variables for a multivariate anaylsis like PCA, with genetic maps, and looked at the similarity? It’d be insightful too to look at where “looks can be deceiving”. An analogue would be those phylogenetic trees where zoologists compare morphological with molecular characters.

Sorry if my comments are a bit wordy.