w which has been lost. Nor is this all; but in certain words sw seems to have vanished. What in our text is hoi, heos, hekuros, were probably woi and swoi, weos and sweos, swekuros. Moreover the received spelling of many other words is corrupt: for instance, deos, deidoika, eddeisen, periddeisas, addees. The true root must have had the form dwe or dre or dhe. That the consonant lost was really w, is asserted by Benfey from the Sanscrit dvish. Hence the true forms are dweos, dedwoika, edweisen, etc. . . . Next, the initial l of Homer had in some words a stronger pronunciation, whether λλ or χλ, as in λλιταὶ, λλίσσομαι, λλωτὸς, λλιτανεύω. I have met with the opinion that the consonant lost in anax is not w but k; and that Homer's kanax is connected with English king. The relations of wergon, weworga, wrexai, to English work and wrought must strike everyone; but I do not here press the phenomena of the Homeric r (although it became br in strong Æolism), because they do not differ from those in Attic. The Attic forms εἴληφα, εἴλεγμαι for λέληφα, etc., point to a time when the initial λ of the roots was a double letter. A root λλαβ would explain Homer's ἔλλαβε. If λλ[Α] ap-**
