But there are other words in which this cannot be the case, for instance, ἄδελφε, πόνηρε, μόχθερε from ἀδελφός, πονηρός, μοχθηρός. Here the accent is the old Aryan vocatival accent. Again, in πατήρ, πατέρα, Sk. +pitấ+, +pitáram+, in μήτηρ, μητέρα, Sk. +mâtấ+, +mâtáram+, in θυγάτηρ, θυγατέρα, Sk. +duhitấ+, +duhitáram+, the radical accent was throughout on the suffix +tár+, nor would the rules of the rhythmic accent in Greek prevent it from being on the antepenultimate in the accusative. The fact therefore that it is retracted on the penultimate and antepenultimate in the vocative, shows clearly that we have here, too, the last working of the original Aryan accentuation. The irregular accent in the nom. sing. of μήτηρ, instead of μητήρ, is probably due to the frequent use of the vocative (an explanation which I had adopted before I had seen Benfey’s essay), and the same cause may explain the apparently irregular accentuation in θύγατρα, by the side of θυγατέρα, in θύγατρες, and θύγατρας. Similar vocatives with retracted accent are δᾶερ, nom. δαήρ, εἴνατερ, nom. εἰνάτηρ, γύναι, nom. γυνή, σῶτερ, nom. σωτήρ, ἄνερ, nom. ἀνήρ, Ἄπολλον, nom. Ἀπόλλων, Πόσειδον, nom. Ποσειδῶν, Ἥρακλες, nom. Ἡρακλῆς. We have thus established the fact that one feature of the primitive Aryan accentuation, which consisted in the very natural process of placing the high accent on the first syllable of vocatives, was strictly preserved in Sanskrit, while in Greek and Latin it only left some scattered traces of its former existence. Without the light derived from Sanskrit, the changes in the accent of vocatives in Greek and Latin would be inexplicable, they would be, what they are in Greek grammar, mere anomalies; while, if placed by the side of Sanskrit, they are readily recognized as what they really are, remnants of a former age, preserved by frequent usage or by an agent whom we do not like to recognize, though we cannot altogether ignore him,--viz. chance.