Western Vulgar Latin reanalyzed the neuter plural as a feminine singular noun (see arma and hoja).

According to Beekes (1996), there are five reasons why the word is borrowed from another language. First, cognates in other languages point to *a in Proto-Indo-European - a vowel which did not exist. Second, if one reconstructs *kh₂p- we must account for a missing full-grade **keh₂p-. Third, an incongruous a/au variation (Old English hafut versus Gothic haubiþ) cannot be satisfactorily explained by contamination and there is no reason to assume an ablaut that ended in a stop. Fourth, the suffix -ut in Latin caput is utterly alien to Italic tongues except as the perfect participle, which makes little sense. Admittedly it could be a borrowing from Celtic, but that does nothing to explain the suffix *-it in Germanic, as evidenced in Gothic haubiþ. The most parsimonious explanation is that suffixes *-ut and *-it are remnants from a non-Indo-European language. Fifth and finally, k-initial plain velars words are rare in Proto-Indo-European. Their presence should raise suspicion. On the other hand, Kroonen (2014) believes the words are of native Indo-European stock, as he assumed variation in *a/au was due to supplanting the old proterodynamic paradigm with -au- (borrowed from the oblique).

"Given this situation [above] - which has always been clear - it seems certain the word is a loan. I find it surprising that I nowhere have seen even the suggestion that the word could be non-IE." ~ R. Beekes, "Ancient European Loanwords" (1996)