More important is the fact that no example of -u after a long syllable is preserved in Bede's Ecclesiastical History. It is to be observed that Bede seems to have scrupulously followed the orthographical peculiarities of the documents which he used. Thus in Papal letters dating from the first half of the seventh century we find such forms as Adilbercto, Audubaldi which are not used elsewhere in the work[110]. Now in the record of the proceedings at the Council of Hertford in 673, which is quoted in IV 5, we find the forms Herutford, Vilfrid, Vynfrid, Hrofaescaestir; and in the similar record for the Council of Hatfield in 680, quoted in IV 17, we find the form Haethfelth. If -u had not been lost by this time the forms used must have been -fordu or -fordus, Vilfridu(s), etc., and it is most unlikely, in view of Bede's practice elsewhere, that he would have altered them. As a matter of fact the latter of these documents contains two forms—Hymbronensium (or Humbr-) and Estranglorum—which are not used elsewhere by Bede. Again, the same author almost always writes the name of the Northumbrian king Ecgfrith (r. 671-685) as Ecgfrid (in the Nom.). That this represents the contemporary orthography is shown by a coin of that king which bears the legend Ecgfrid rex.
But further, there is a whole series of names, much used in Northumbria, which have as their second element -haeþ from earlier -haþu (e.g. Eadhaed). At the time when -u was lost here the change a > æ was clearly still operative. Will anyone suggest that this was the case after the middle of the seventh century? Again, it can hardly be doubted that the form Osuiu, if not also Osuald, represents the pronunciation[111] current in the time of that king. In Bede's History, III 29, we have extracts from a Papal letter addressed to Osuiu (v. l. Osuio) regi Saxonum. The form os- can scarcely have become current in compounds before -u was lost in the Nom., Accus. sg.
It is surely unnecessary to enter into further details. We have seen that practically the whole evidence for the proposed chronology consists of two forms, Velhisci and flodu, one of which is incapable of proving what it is meant to prove, while the other is of exceedingly doubtful value; and on the other hand that this chronology has opposed to it a large number of forms in the glossaries, in charters and in early Northumbrian authorities of various kinds. Now let us consider the various sound-changes in relation to one another.
It is admitted, and necessarily so, that the loss of -u took place before the loss of h. For the sake of convenience we may apply the terms 'Period I' to times anterior to the loss of -u and 'Period II' to the interval between the two changes. From neither of these periods have we any texts surviving, unless we are to reckon the inscription on the Franks casket to Period I. Our earliest charters, and apparently also the lost archetype of the glossaries, were composed at a later time ('Period III') when h was no longer pronounced, though doubtless often written. But before the date of the earliest extant glossary (Epinal) a further change or changes had taken place which brought about the confusion of original ƀ and f. It is clear from a comparison of the glossaries that this confusion was later than the loss of h and also that it was almost, if not entirely, unknown to the archetype. So also in Northumbrian. In Bede's History we have no instance of h preserved in an English word, whereas examples of its omission are numerous. On the other hand it is clear that Bede usually retained the distinction between b (i.e. ƀ) and f, although the latter has already largely encroached on the former in the Moore MS.[112] We may therefore constitute two subdivisions of the period subsequent to the loss of h, namely 'Period III' prior to the confusion of ƀ and f and 'Period IV' subsequent to this confusion. Now we can see clearly where to date the Franks casket, for it shows confusion of ƀ and f in wylif—according to Prof. Morsbach also in sefu. It belongs therefore not to Period I but to Period IV.
Any attempt to fix an absolute chronology is of course rendered difficult by the absence of very early texts. We may probably assume that Period IV begins more or less about 700. The Kentish charter of 679 falls in Period III. Now when -u was lost, h (whether intervocalic or postconsonantal) must have been a spirant—a fact which Prof. Morsbach seems to have entirely ignored. It is quite incredible that only a few years should elapse between that stage and the total loss of intervocalic h seen in the charter. The corresponding period in Germany lasted centuries, and I cannot conceive how the transition can have been accomplished anywhere in less than half a century. I conclude therefore that the dates which I gave in my Studies were approximately correct. If there was an error it was in putting some of the changes slightly too late. I have little hesitation now in expressing my opinion that the loss of -u took place not later than the second or third decade of the seventh century.
FOOTNOTES:
[95] Transactions of the Cambridge Philological Society, Vol. IV, Part II. The page-references are to the figures in the outer corners.
[96] Nachrichten von der Königl. Gesellschaft der Wissenschaften zu Göttingen, 1906, pp. 251-277.
[97] Prof. Morsbach's paper raises a number of questions besides those mentioned above. But I am obliged here to confine my attention to those which have a bearing on the dating of Beowulf.
[98] The application of this term to the Epinal and Erfurt glossaries (or the archetype) seems to me to be open to grave objection.