English.
III. That it is English, or at least Anglo-Norman.
Literary.
IV. That it is very mainly a "literary" growth, owing something to the Greek romances, and not to be regarded without error as a new development unconnected, or almost unconnected, with traditional sources of any kind.
The Celtic theory.
The first explanation is the oldest. After being for nearly half a century discredited, it has again found ardent defenders, and it may seem at first sight to be the most natural and reasonable. Arthur, if he existed at all, was undoubtedly a British hero; the British Celts, especially the Welsh, possess beyond all question strong literary affinities and a great literary performance, and Geoffrey of Monmouth, the father of the whole story, expressly declares that he took it from a book written in the British tongue. It was natural that in comparatively uncritical ages no quarrel should be made with this account. There were, even up to the last century, I believe, enthusiastic antiquaries who affirmed, and perhaps believed, that they had come across the very documents to which Geoffrey refers, or at worst later Welsh transcripts of them. But when the study of the matter grew, and especially when Welsh literature itself began to be critically examined, uncomfortable doubts began to arise. It was found impossible to assign to the existing Welsh romances on the subject, such as those published in the Mabinogion, a date even approaching in antiquity that which can certainly be claimed by the oldest French texts: and in more than one case the Welsh bore unmistakable indications of having been directly imitated from the French itself. Further, in undoubtedly old Welsh literature, though there were (v. supra) references to Arthur, they were few, they were very meagre, and except as regards the mystery of his final disappearance rather than death, they had little if anything to do with the received Arthurian story. On the other hand, as far as Brittany was concerned, after a period of confident assertion, and of attempts, in at least doubtful honesty, to supply what could not be found, it had to be acknowledged that Brittany could supply no ancient texts whatever, and hardly any ancient tradition. These facts, when once established (and they have never since been denied by competent criticism), staggered the Celtic claim very seriously. Of late years, however, it has found advocates (who, as usual, adopt arguments rather mutually destructive than mutually confirmatory) both in France (M. Gaston Paris) and in Germany (Herr Zimmer), while it has been passionately defended in England by Mr Nutt, and with a more cautious, but perhaps at least equally firm, support by Professor Rhys. As has been said, these Neo-Celticists do not, when they are wise, attempt to revive the older form of the claims. They rest theirs on the scattered references in undoubtedly old Welsh literature above referred to, on the place-names which play such an undoubtedly remarkable part in the local nomenclature of the West-Welsh border in the south-west of England and in Cornwall, of Wales less frequently, of Strathclyde and Lothian eminently, and not at all, or hardly at all, of that portion of England which was early and thoroughly subjected to Saxon and Angle sway. And the bolder of them, taking advantage of the admitted superiority in age of Irish to Welsh literature as far as texts go, have had recourse to this, not for direct originals (it is admitted that there are none, even of parts of the Legend such as those relating to Tristram and Iseult, which are not only avowedly Irish in place but Irish in tone), but for evidences of differential origin in comparison with classical and Teutonic literature. Unfortunately this last point is one not of technical "scholarship," but of general literary criticism, and it is certain that the Celticists have not converted all or most students in that subject to their view. I should myself give my opinion, for whatever it may be worth, to the effect that the tone and tendency of the Celtic, and especially the Irish, literature of very early days, as declared by its own modern champions, are quite different from those of the romances in general and the Arthurian Legend in particular. Again, though the other two classes of evidence cannot be so ruled out of court as a whole, it must be evident that they go but a very little way, and are asked to go much further. If any one will consult Professor Rhys's careful though most friendly abstract of the testimony of early Welsh literature, he will see how very great the interval is. When we are asked to accept a magic caldron which fed people at discretion as the special original of the Holy Grail, the experienced critic knows the state of the case pretty well.[66] While as to the place-names, though they give undoubted and valuable support of a kind to the historical existence of Arthur, and support still more valuable to the theory of the early and wide distribution of legends respecting him, it is noticeable that they have hardly anything to do with our Arthurian Legend at all. They concern—as indeed we should expect—the fights with the Saxons, and some of them reflect (very vaguely and thinly) a tradition of conjugal difficulties between Arthur and his queen. But unfortunately these last are not confined to Arthurian experience; and, as we have seen, Arthur's fights with the Saxons, except the last when they joined Mordred, are of ever-dwindling importance for the Romance.
The French claims.
Like the Celtic theory, the French has an engaging appearance of justice and probability, and it has over the Celtic the overwhelming advantage as regards texts. That all, without exception, of the oldest texts in which the complete romantic story of Arthur appears are in the French language is a fact entirely indisputable, and at first blench conclusive. We may even put it more strongly still and say that, taking positive evidence as apart from mere assertion (as in the case of the Latin Graal-book), there is nothing to show that any part of the full romantic story of Arthur, as distinguished from the meagre quasi-historical outline of Geoffrey, ever appeared in any language before it appeared in French. The most certain of the three personal claimants for the origination of these early texts, Chrestien de Troyes, was undoubtedly a Frenchman in the wide sense; so (if he existed) was Robert de Borron, another of them. The very phrase so familiar to readers of Malory, "the French book," comes to the assistance of the claim.
And yet, as is the case with some other claims which look irresistible at first sight, the strength of this shrinks and dwindles remarkably when it comes to be examined. One consideration is by itself sufficient, not indeed totally to destroy it, but to make a terrible abatement in its cogency; and this is, that if the great Arthurian romances, written between the middle and end of the twelfth century, were written in French, it was chiefly because they could not have been written in any other tongue. Not only was no other language generally intelligible to that public of knights and ladies to which they were addressed; not only was no other vernacular language generally known to European men of letters, but no such vernacular, except Provençal, had attained to anything like the perfection necessary to make it a convenient vehicle. Whatever the nationality of the writer or writers, it was more likely that he or they would write in French than in any other language. And as a matter of fact we see that the third of the great national claimants was an Englishman, while it is not certain that Robert de Borron was not an English subject. Nor is it yet formally determined whether Chrestien himself, in those parts of his work which are specially Arthurian, had not Map or some one else before him as an authority.
The theory of general literary growth.