In a previous post I mentioned the discovery of an Etruscan form numasianas which seems to match the Praenestine Fibula's from NVMASIOI. I further argued that "the form numasianas supports Numasios and makes it certain that Numasios is unconnected with numerus," but an email from Michiel Driessen, the author of the best etymology of urbs (JIES 29, 2001, p. 41-68), has made me reconsider at least the second half of this claim. He points out that the medial a of the Etruscan form may be the result of weakening and points to the parallel of the praenomen Avile <— avil 'year' which also appears as Avule, and—significantly from Caere—as Avale (Cr. 3.23, 5th cent.) The aryballos in question dates to the second half of the 7th century, but weakening is found in the archaic period, so we cannot exclude the possibility that the a is a weakening product. Note too that 7th century forms of the NumVs- name are found with a medial e-vowel, e.g. Numesiesi (Ta 3.1). Further since the name numasiana- is a derivative in -na and since the base is well attested as ending in -sie, we would expect the last three syllables to be -siena. Hence the second a of numasianas is a weakening product and this strengthens the case for the first a being unoriginal. Thus there is no real objection to deriving the Etruscan forms from an Italic, probably Sabellic, source *nomesiyo-.
So is numasianas totally uninformative in regard to NVMASIOI? I would say not, since it gives a possible source for the always problematic a of the Praenestine form, i.e. a re-borrowing from a form with Etruscan weakening.