The Latin word for 'feather' is penna, which is undoubtedly a derivative of the PIE root *pet(h2)- 'fly'. If we knew nothing else it would be assumed that penna was from *pet-na, but the picture is complicated by the existence of the forms pesnis (Fest. p. 222L) and pesnas (Fest. p. 228 L). This has led some scholars, e.g. Meiser 1998:118, to suppose that penna is from *petsna and that the outcome of *-VtsnV- was not -V:nV-, as one might have expected, but -VnnV-. This is not totally impossible since pullus < *putslo- (cf. pusillus) shows that at least one *-VtsRV- sequence could lead to -VRRV-. Szemerényi's idea, on the other hand, that penna is from *pēna by the Iuppiter rule is totally impossible. But I think the Festus passage at 228 suggests a solution other than the one favored by Meiser. The passage reads in Lindsay's edition:
Pennas antiquos fertur appellasse †peenas† ex Graeco quod illi πετηνὰ quae sunt volucria, dicant. Item easdem pesnas ut cesnas.
It is evident that what stands between the obels must be emended to pet(V)nas, as suggested by Mueller, since only if the form had a t in it would the derivation from Greek πετηνὰ make sense. The second sentence (item easdem pesnas ut cesnas) means that Festus' source also knew an old form with s. Thus there were two old forms floating around petna and pesna, just like *putslo- (Lat. pullus) beside *putlo- (Osc. puklo-). Thus nothing stands in the way of deriving penna from *petna. Whether *petsna would have given penna too or *pēna cannot be answered with certainty. The upshot is that I agree with what I wrote on pg. 168 fn. 3.