This is the compelling argument, which Nick made in his note on sigma as well, and which I had forgotten. So while I have to admit that this argument is compelling, I do still think that lunate sigma (which is the same glyph for both word positions) is going to cause real problems with normalization. It is not those who follow the Unicode specifications who worry me, it's those who do not. Thanks to Jim Allen and as often to Nick.“φιλοσ.,” is necessarily the abreviation of some word (like φιλοσοφία) while “φιλος.” is a single non-abbreviated word, followed by a sentence period.
For the details of the process which is described in a somewhat simplified way by Katerina Sarri, one can look to the examples in *Thompson*, An Introduction to Greek and Latin Palaeography.
PTR
