A reservation should perhaps be made with regard to the latest elements in the poems. Although definite evidence is wanting, the beginnings of the Greek alphabet may quite probably go back to the ninth century, and it may very well be that the poems had not then attained their final form. But it should not be assumed that the alphabet was introduced simultaneously throughout the Greek world. Some districts may have acquired it generations before others, and Aeolis (including even Chios and Smyrna) was probably not one of the more advanced. Further, we must admit that in all probability its use was at first very limited. From all analogies we should expect that it was employed for inscriptions, correspondence, etc. for a very long time before it was made to serve any literary purpose. Such apparently was the case with the alphabet of ancient Rome and with the Runic alphabet almost throughout its history. Unless the conditions in Greece were quite exceptional we should not expect the alphabet to come into contact with heroic poetry for a considerable time.
The theory that certain portions of the poems were written down from the beginning presupposes of course a use of writing quite different from that which is brought before us in the story of Bellerophon. This theory[343] is, and necessarily must be, bound up with another theory, which we have already discussed, that these portions date from a period not earlier than the seventh century. Our discussion has led us to the conclusion that the evidence for the latter theory is unsatisfactory. But, apart from this, it is inconceivable that the 'literary' portions, which are said to amount to several thousand verses, should contain no reference, direct or indirect, to the use of writing. In Beowulf we find only one direct reference to writing (v. 1694 f.)—an inscription in Runic letters such as had long been in use. But in the Christian additions or 'interpolations' we meet with three examples of the verb scrifan (forscrifan, gescrifan—from Lat. scribere), which of course is indirect evidence for the use of writing, though in a different language[344]. If large portions of the Homeric poems were really of literary origin the authors could scarcely have failed to betray themselves by usages of this kind, even though they deliberately avoided all mention of writing.
In addition to this general consideration we have to take account of the linguistic difficulties discussed above. We have seen that Hesiod's works show almost the same form of language as the Homeric poems, although the Boeotian dialect was quite different from anything spoken on the other side of the Aegean. Did Hesiod, who lived in the eighth century, probably before the Ionian conquest of Smyrna, really employ the 'impure Ionic' in which his poems have come down to us? He himself says (W. and D. 650 ff.) that he had never crossed the sea except (once apparently) to Chalcis. Presumably then his knowledge of heroic poetry was derived either from Boeotia, where he lived, or from Cyme in Aeolis, from whence his father had emigrated. But Cyme never came into Ionian hands. Are we to suppose then that either here or in Boeotia poets were already employing as their vehicle a form of language which according to the theory under discussion owed its existence (whether in ordinary speech or only in poetry) to certain political changes in a third district—changes too which had hardly begun much before the time of Hesiod? This hypothesis seems to me quite incredible[345].
On the other hand if Hesiod's poems have undergone a change of dialect since their original composition may not the Homeric poems have passed through the same process? In this case of course such a change would come about quite naturally if Smyrna or Chios was the original home of the poems. But here we must notice a curious feature in the 'epic dialect' to which we have already referred. Except in Aeolic forms—and this exception deserves to be remarked—the poems almost always preserve initial h-[346]. This is a characteristic which the 'epic dialect' shares with western or European Ionic but not with the language of the Asiatic coast[347]. Its presence raises a distinct difficulty in the way of supposing that the poems were Ionicised in their original home.
Now if the Homeric poems had been written down in Aeolic and preserved in literary form we can hardly doubt that they would have retained their original dialect, just like the works of Alcman, Alcaeus, and Sappho. It must not be argued that Ionic was the proper language of the epic, Doric and Aeolic of the lyric; for if the Homeric poems had become generally known in Aeolic nothing could have prevented this dialect from becoming the language of the epic also. We may assume then that they were not transmitted in written Aeolic. But the same argument really militates against the theory that they were written in Asiatic Ionic[348], though in this case the difference of dialect is less striking.
A form of language practically identical with that of the epics appears in the remains of several poets of the seventh and sixth centuries, such as Archilochos and Solon. The only noticeable difference is in the proportion of non-Ionic forms which they use. Now since these authors belonged to quite different districts we must, if we are to trust our evidence, infer the existence of a kind of literary language at this time. Its difference from Asiatic Ionic is of course comparatively slight, and in the texts which have come down to us it is not always carefully observed[349]. Still there is a sufficient amount of regularity to show that it was generally recognised.
The true home of this literary language must not be sought in Athens, but rather in Euboea or the Cyclades; and we may probably attribute its spread, in part at least, to the influence of the poems of Archilochos. But it certainly affected the writing of Attic for some two centuries, and there can be little doubt that it was thoroughly domiciled in Athens at quite an early date—probably in the seventh century. I cannot see any objection therefore to supposing that it was in Athens that both the Homeric and Hesiodic poems acquired those peculiar linguistic characteristics which we comprehend under the term 'epic dialect.' This of course brings us back to the story that the poems were collected or written down by order of Peisistratos. The evidence for the story is late, and its truth is hotly contested by many scholars. But at all events it has the merit of providing a satisfactory explanation of the linguistic phenomena.
Certainly, if we may judge from the analogy of other peoples, heroic poetry would not by any means be among the first species of literature to get committed to writing. There is no reason whatever for supposing that anything of this kind was written down in England before the eighth century, i.e. at least a century after the language was first applied to literary purposes, in Aethelberht's time. The same remark seems to be true of Ireland, Germany and the North, while Bosnian heroic poetry is being written down only in our own generation, and not by natives of the country. It is only natural therefore to expect that the Homeric poems would be written down according to an orthography which was already well established.
This orthography no doubt represents more or less truly the form in which the poems were recited in Athens at the time. But does it also represent the form in which they were recited at (let us say) Sicyon? That is a question upon which we have no direct evidence. But it is worth noting that in Doric and other non-Ionic states we find a number of ancient inscriptions in hexameters or elegiacs which contain epic words and forms. We may note especially the Gen. sg. ending -οιο, e.g. κασιγνετοιο, οδοιο, Αραθθοιο, and above all false imitations of epic forms, such as Τλασκιαϝο[350]. But in all these inscriptions Ionic characteristics are conspicuously absent[351]. Does this mean that wherever the heroic poems were introduced the rhapsodists tried to accommodate them, as far as possible, to the language of the district? But if so, is it possible that the Ionic element in our texts is wholly due to the rhapsodists? A Chalcidian or Naxian or Athenian audience would certainly experience at least as much difficulty in following a purely Aeolic Iliad as a southern English audience would have in listening to a purely Northumbrian or Mercian Beowulf. On the other hand an Iliad Ionicised in Chalcis or Naxos would be easily intelligible in Athens[352].