¹ A list of these is found in Monro, Grammar of the Homeric Dialect ² 25, and Hirt, Handbuch d. gr. Laut- u. Formenl. ² 576.

² So, e.g., Brugmann-Thumb, Gr. Gr.⁴ 376 f.

³ See Monro 24 f.

⁴ The term ‘root-determinative’ does not appear below because it can only cause misunderstanding, and would lead to the conclusion that, at least when these elements originated, they were perceived as something distinct from the root itself. But the existence of these ought to mean nothing more than that the root is as unstable as every other part of the word, and that it is subject to the same associative influences as stem suffixes and personal endings and case endings. Thus the μ of Gr. τρέμω ‘tremble’ as opposed to τρέω never had any more of a separate existence or meaning than the -εσ- of the Aeolic -εσσι (dat. pl. ending opposed to -σι).

⁵ So Persson, Beitr. z. idg. Wortforsch. 523 ff.

⁶ Cf. Hirt, Idg. Gramm. 3. 236 ff.

⁷ See p. 274 for the possibility of the association of perfect with aorist forms.

⁸ However, see below for a possible minor influence of ἔ-δωκ-α on βέ-βρω-κα and μέ-μβλω-κα.

⁹ Probably also its synonym ἔ-πτηκ-α, as seen below, was of some influence.

¹⁰ It is very doubtful whether a perfect ἵληκα, existed at all. See p. 273.

¹¹ In spite of Brugmann-Thumb 551, it is difficult to interpret this perfect in any other way than as a very early example of the ‘resultative’ perfect, which designates a resulting state, not of the subject, but of the object of a verb. It was particularly easy for this nuance to develop when the subject was abstract.

¹² Similarly, the compounds ἐφ-έηκα and κατa-βέβληκα were quoted above for their influence in the spread of the κ-Perfect.

¹³ If Doric -ἕ-ωκ-α, as the old vowel gradation makes plausible, was the general Greek perfect of ἵημι, it was largely responsible for the old perfects ending in -ωκα, and must have assisted very materially the aorist ἕ-ηκ-α in causing the general spread of the κ. Nevertheless, even then, it would not be possible to get along without assuming the influence of ἕ-ηκ-α altogether, for the confining of Homeric κ-Perfects to the singular necessarily points to aorist patterns.

¹⁴ It is possible, because of δειδίξoμαι < *δε-δ ικ-σoμαι, (see Brugmann-Thumb 377), that δέ-δoικα was not a κ-Perfect at all, and that the κ belonged to the root. In this case it was itself one of the patterns of the κ-Perfect.

¹⁵ Although this perfect occurs for the first time in Hesiod, Op. 591, in the participle τετoκυίης, its vowel gradation marks it as old.

¹⁶ Cf. e.g. Wundt, Sprachpsych. ³ 1. 402 et al.

¹⁷ It is of course impossible to do more than speculate whether the Doric perfect -ἕ-ωκ-α, as suggested above, was in existence in other Greek dialects also, or whether pre-Hellenic *ε- ωκ-α had become *ε-ηκ-α through the influence of the aorist *ἐ-ηκ-α. At any rate there is no particular reason for believing that Attic etna < *ἑ-εκ-α, with the root vowel of the middle εἷμαι < *ἑ-ε-μαι, existed in the Homeric language.

¹⁸ Cf. Boisacq, Dict. etym. de la lang. grecque s. v. βáλλω.

¹⁹ That the Homeric language may not have been an absolutely complete mirror of the spoken language at its basis, is self-evident. It may well be also that if we had extant certain forms that are accidentally not preserved in Homer, we might have to revise our conclusions as to the stages in the spread of the κ-Perfect in some details. However, this would mean rather that the exposition given of the various associative connections operative in the spread of the formation was incomplete rather than false. The main outlines of the process could scarcely have been given wrongly.

²⁰ See note 15.