Page:On translating Homer (1905).djvu/165

 Arnold's (it seems) does not. If we were ordered to read dat ting in Chaucer for ''that thing'', it would at first 'surprise' us as 'grotesque', but after this objection had vanished, we should still feel it 'antiquated'. The confusion of thick and tick, thread and tread, may illustrate the possible effect of dropping the w in Homer. I observe that Benfey's Greek Root Lexicon has a list of 454 digammated words, most of which are Homeric. But it is quite needless to press the argument to its full.

If as much learning had been spent on the double λ and on the y and h of Homer, as on the digamma, it might perhaps now be conceded that we have lost, not one, but three or four consonants from his text. That λ in λύω or λούω was ever a complex sound in Greek, I see nothing to indicate; hence that λ, and the λ of λιταὶ, λιαρὸς, seem to have been different consonants in Homer, as l and ll in Welsh. As to h and y I assert nothing, except that critics appear too hastily to infer, that if a consonant has disappeared, it must needs be w. It is credible that the Greek h was once strong enough to stop hiatus or elision, as the English, and much more the Asiatic h. The later Greeks, after turning the character H into a vowel, seem to have had no idea of a consonant h in the middle of a word, nor any means of writing the consonant y. Since G passes through gh into the sounds