Such in the second class are
ἐ-ρέφω, whence roof.
ἐ-λεύθερος, whence liber.
ἐ-ρυθρὸς, whence ruber, rufus.
ἐ-ρετμὸς, whence remus.
This point being disposed of, how are we to account for finding φυρη, instead of φερη or φηρη?
Can it be because, in cases of Greek syllabic augment, there is a tendency to avoid reduplication, as in ἀτιτάλλω for ἀτατάλλω? In but a small proportion of the cases given in Dr. Donaldson’s table is the vowel prefix the same with the vowel following.
Can it be from that tendency of what we call comprehensively the digamma to lapse into the υ, which Heyne has observed[853]?
Or, shall we found it on the principles laid down by Bopp[854], in his Comparative Grammar, that the α has a tendency to weaken itself into υ, and that liquids having a preference for that latter vowel, influence the generation of it? the conditions of interchange between α and υ resting, as he says, upon the laws of gravity or vocal equilibrium.
It must be observed that the original vowel of the root may, in this case, have been the α which we find in φᾶρις.
It is not only α that we may find supplanted by υ. The ε suffers the same fate in the Italian Siculus, which appears as the representative of the Greek Σίκελος. Again we have, in the Latin, the kindred words furo and fera. Perhaps I am wrong in dealing thus scrupulously with the variation from ε to υ, as if capable of affecting vitally the question of identity in the root. For in examining another root (that of κεφάλη), we have seen that its derivatives appear to include the whole, or nearly the whole range of the vowels of the alphabet.
Its probable signification.