Between the penthemimeral caesura and the bucolic diaeresis : 7TE7TVU fLEVO�, 8EOELKEAO�, fLEyaA�Topo�, SOUPLKAUTOV, 8EOELKEAOV, fLEya8vfLou, fLEyaA+ TOPL, fLEya8vfLWL, fLEyaA�Topa ; between the feminine caesura and the b uco 1· IC d· laereSlS : ap7JLOV, . . , • , afLufLovo�, afLufLova, . , . ,
cppova, SdCPLAE ; between the bucolic diaeresis and the end of the line :
" 8 , , " � " • �
U70 EO� 7TOLfLEVL 7TOLfLEva opxafLo�
• 8 ' "
avn EOLO, OPXafLE
VI. The device of using an epithet alone to fill a distinct part of the line would in all likelihood never have come into being without the generic epithet. But once it had become current, once poets had grown familiar with the several possibilities of its use, they began to treat specific epithets in the same way, choosing those which, like the majority of generic epithets, extend between a caesura and the bucolic diaeresis, or between the latter and the line-end. Such a use in Homer is that of N7JA�LO�, which can apply to Nestor and to his sons :
'P 349
'P 5 14
W� El7TciJV N/{1Twp NT)A�LO� wp ev;' XWP'T)L
TWL 3' ap' e7T' :4vTlAoxo� N'T)A�LO<; tAauEv i1T1TOV�.
We can compare the latter line with another in which the poet pre ferred to fill the same space with an epithet qualifying the horses instead of using SdCPLAO�, as he often does in similar situations :
VII. The poet makes use of essentially the same device when he comes upon a hero's name itself able to fill the space between the penthemimeral caesura and the bucolic diaeresis, and inverts the usual word-order, putting the epithet in the first half of the line :
dMa ucpm<; KpaTEpo<; ..::I'OJL1/3'T)<; egEVap"gE 3� TChE y' :4TpE{3'T)<; :4yal'.E!-,vwv egEVap,g€. These lines may be compared with :
We now arrive at that area of formulary diction in which it becomes more difficult to distinguish between what is certainly traditional and what could be a single poet's original creation. Noun-epithet formulae designed for the expression of some more or less particular idea do not by definition make up sets of analogous cases, and may even not conform to standard measures. One might be tempted to conclude from this that they are due to Homer's I originality. But a closer examination of the circumstances under which epic diction developed will show that the singularity of a formula is no proof that it is not traditional.
{3. Formulary diction and the operation of analogy
We must once more return, as we have done so many times already, to the relation between the ideas of epic poetry and the words which express them. It was the constant tendency of epic diction to make the expression of the ideas of heroic poetry as simple and as easy as possible, and to this end it employed the means of analogy, only abandoning it when the complexity of the ideas to be expressed made its use no longer feasible. In more definite terms : the bards, always trying to find for the expression of each idea in their poetry a formula at once noble and easy to handle, created new expressions-in so far as the result was compatible with their sense of heroic style-in the simplest way possible : they modified expres sions already in existence. To this process are due all the series of formu lae which we have so far examined. In each of these series it would be pointless to look for the original or the oldest formula. But in every case there must have been an original expression from which the series was produced by the system of imitation we call analogy. Analogy is perhaps the single most important factor for us to grasp if we are to arrive at a real understanding of Homeric diction. To understand the role of analogy in the formation of epic language is to understand the interdependence of words, ideas, and metre in heroic poetry. It is to see to what extent the hexameter and the genius of the bards influenced epic style. And finally, it is to recognize that there are limits beyond which analogy could no longer advance the simplification of the technique, so that some formulae remained more or less unexampled.
Philologists have long recognized that formal associations I bring about changes in the spoken language and thus constitute one of the most important causes of the continual modification of language. They were more reluctant to recognize that this linguistic process found in the literary language of epic poets a realm in which it could operate in a fashion unknown to any spoken language ; that the desire of bards to possess 'alien' forms for the embellishment of their style and also to have forms adapted to the mould of the hexameter was in large part realized by the means of analogy. Ellendt, who with Diintzer was the first to explain the influence of metre on Homeric forms in a methodical fashion, does not stress the role of analogy, the importance of which was not yet fully perceived by the philologists of his time ; but most of the examples which he cites are cases of analogy. To say, for example, that the in fluence of metre determined the masculine ending of an adjective where we should expect the feminine, or that, for the same reason, an adjective has sometimes two endings and sometimes three, is to say that the poet
[86-7]