The contention that the Homeric epics, and perhaps also the Hesiodic poems and the Homeric Hymns, are the products, directly or at a very short remove, of a tradition of orally improvised poetry is widely accepted as a basic premiss in Homeric criticism. The cogency of the argument depends on the frequency and characteristic use of formulae in the early hexameter poetry, and their rarity in the literature of Classical, Hellenistic, and Roman times, which is known or assumed to have (...) been composed in the study. The reasoning appears to me valid, but in some respects overstated or ambiguously stated in recent publications, and the first fault arises out of the second. (shrink)

The assumption that a particular object mentioned in the Iliad or Odyssey must be described by epithets which are consistent with each other and with the narrative has complicated every attempt to relate the evidence of archaeology to the poems. It may fairly be assumed that a modern writer wants to be consistent and that, apart from oversights, he will not use an epithet unless it is directly appropriate to the object which he is creating for his immediate purpose; but (...) this is not necessarily true of a poet who had ready to hand a rich store of phrases in meaning appropriate to the various furnishings of his heroic world, and in form adapted to the needs of his verses. As within the Kunstsprache there is a group of epithets describing the moral and physical characteristics common to all heroes and used to suit not the race or actions of individuals but the scansion of their names and the place in the hexameter, so it is natural to suppose that descriptions of things in common use will include at least some similar stock phrases appropriate to types rather than to units. Once the existence of such phrases is established, there is no difficulty in their combination, as metrical convenience requires, in composite descriptions made up of elements each of which has its own real counterpart, but which need never all have existed together in the same object or even in the same age. The following tabulations of the descriptive words and phrases applied to a few familiar objects show that such metrical clausulae are always present, though they vary in prevalence and behaviour. (shrink)