Come to think of it, after looking at the family-level reconstructions, I wonder why Bomhard has only three phonemic vowels.  As you [Pat] said, Dravidian data seems to challenge that theory and call for a proposal of all six vowels being phonemic.
 
Remember, Bomhard uses Diakonoff's (?) Afro-Asiatic which has the one phonemic vowel, a~@, and a reduced consonant inventory as well, taking out the uvulars.  Plus I think he favors Kartvelian more when it comes to consonantal reconstruction.  Still, I think AB's a little too conservative on his phonemes, while Dolgopolsky is probably too liberal (he has more sibilants among other consonants and seven vowels).
 
I'm probably going to "edit" the roots as far as I am able.
 
I'm also confused about the palatized and labiovelarized velars, plus why does he have only one labiovelarized uvular, /q'w/.  Most cases are justified by the occurence of palatal-velar-labiovelar stops in Indo-European and Afro-Asiatic.
 
~DaW~