¹ These results are to take the form of a doctoral thesis entitled “Formulaic devices in Hittite and Greek epic”. Some of the arguments presented here are necessarily somewhat compressed on account of limited space. What is dealt with rather summarily here, however, is discussed in detail in the thesis.

² Published in JCS. 5, 1951, 135 ff.Google Scholar, and ibid., 6, 1952, 8 ff. Also reprinted under one cover. The page-numbers referred to here are those of the reprint. For argument on poetic character of the work, see pp. 7 ff.

³ AJA. 52, 1948, 133 fGoogle Scholar.

⁴ Harvard Studies in Class. Phil. 41, 1930Google Scholar: “The epic technique of oral verse-making,” I.

⁵ cf. Od. viii, 43 ff., and xxii, 347 ff., etc.

⁶ Parry himself spent his last years in the Balkans doing field research on this very question, a work which has been carried on by A. B. Lord and others. (See Parry-Lord, , Serbocroatian Heroic Songs, Cambridge, Mass., 1954Google Scholar; Lord, The Singer of Tales, ibid., 1960; Notopoulos, J. A., Modern Greek heroic oral poetry, 1959Google Scholar; etc.)

⁷ The verse is filled with a noun-epithet formula of the type κρείων Ἀγαμέμνων or πολύτλας δīοϛ Ὀδυσσεύς respectively.

⁸ Il. XXIV, 340 ff. = Od. V, 44 ff.

⁹ Ullikummi, II, i, 29 f.; KUB. XXXVI, 25, iv, 5 f.Google Scholar; Gilgamesh, 4, iv, 6Google Scholar, and 8, i, 18 (text according to Friedrich, , ZA. 39, 1929, 1 ff.CrossRefGoogle Scholar).

¹⁰ Il. XVI, 3; cf. IX, 14.

¹¹ Used only with the Accusative case.

¹² Noun-epithet formulae admittedly provide an exception in that they play a much greater part in Homer than in the Hittite epic tradition. Even this, however, finds an adequate explanation in the theory of the metre presented here. (See below, pp. 241 ff.)

¹³ Bowra, C. M., Heroic poetry, pp. 36 ffGoogle Scholar.

¹⁴ See above, p. 238. Space forbids detailed discussion of irregular forms, which may briefly be categorized as follows:—

(1) Spurious forms, depending on texts which, as far as representing the formulae is concerned, are inferior. In Ullikummi, Texts A and D (related texts) are consistently reliable in this respect, B and C usually unreliable though not always so.

(2) Regular by-forms, e.g. “ištanzani-ši” (“to his own mind”) for the missing indirect object in a soliloquy.

(3) Isolated omission of subject in usually reliable Ullikummi Text A (I, iii, 27).

¹⁵ Occasionally, “memiyan” (“word”) is used instead of “uddar”.

¹⁶ Two examples from related texts (KUB. XXXIII, 103, ii, 1 f.Google Scholar, and XII, 65, iii, 5) illustrate this point admirably:—

This partly answers the question whether the respective lengths of the names used for subject and indirect object condition each other and/or the remainder of the verse (i.e. inclusion or exclusion of “uddar” and/or “appa”), which would imply possibility at least of a quantitative metre. Actually, although the names employed in these epics do not as a whole vary a great deal in length, the totals obtained by adding together the two names in examples of this formula range from five to ten syllables, and there are examples of verses containing short names without “uddar” or “appa” and of verses containing long names with both. In examples of this formula, the total number of syllables per verse ranges from eleven to twenty-two; and the fact that there exist verses with every number in between answers the suggestion that verses containing the maximum number may have to be divided into two. Further, if it be argued that the above statistics are unreliable because the epics are merely Hittite translations of Hurrian originals and therefore the number of syllables in the Hittite perhaps does not correspond to that in the original Hurrian, it can be said that the number of syllables in the Hittite must have at least some bearing on that of the original, being after all based largely on the number of words employed.

¹⁷ e.g. Ullikummi, III, iii, 30–37.

¹⁸ A comparison of Ullikummi with Odyssey v—about equal in size—illustrates this. The former contains twenty-seven noun-epithet formulae with seventy-eight occurrences, the latter not less than sixty-five formulae with 131 occurrences.

¹⁹ “nepišas Istanuš” ( = “Istanus of heaven”) is used in four cases of the noun without any changes other than grammatical.

²⁰ The groups of the and types, already referred to, and their employment at the end of the verse in speech-introduction formulae provide a very good example.

²¹ This is quite evident where, for example, a noun-epithet formula is used for subject or indirect object in an occurrence of the important formula on which the arguments in this study are based. In at least one case, a noun-epithet formula is used for both (KUB. VIII, 48, i, 11Google Scholar).

²² See Otten, H., MDOG. 85, 1953, 29Google Scholar; Güterbock, H. G., Kumarbi (Zürich—New York, 1946), pp. 3, 110Google Scholar; Götze, A., Kleinasien, 2nd Ed. (München, 1957), 173 fGoogle Scholar.

²³ Güterbock himself came near to suggesting this: “Song of Ullikummi,“ p. 8: “… there is a majority of verses that can be called of normal length, with usually four stresses and about twelve to seventeen syllables.”

²⁴ I should like to express my gratitude to Professor Webster and Professor Szemerényi for reading this paper and offering a number of helpful suggestions.