But in Cligés, and in a minor degree in the Charrette, we are on different ground: the geography is that neither of Wales nor of Brittany. Here we have Dover, Wallingford, Winchester, Windsor, Southampton, Oxford, Shoreham, Bath, London; while we note a marked omission of the distinctively Arthurian localities. The Charrette opens at Carlion, which it, however, apparently confuses with Camelot.
Now this is surely significant. If Chrétien had a free hand in the arrangement of his stories, if they were really compounded of elements drawn from all sources and thus combined for the first time, why did he shift his mise-en-scène backwards and forwards in this curious manner? Why turn from the geography of Erec to that of Cligés and the Charrette, only to revert to his first love in Yvain and Perceval? Is it not most probable that in those three stories, at least, he was dealing with traditional matter, the localising of which had already been effected?
In the case of Cligés and the Charrette it seems not improbable that closer investigation may find grounds to support the theory of a possible Anglo-Norman transmission, which would account for the southern England geography.[90]
A point on which we may well lay stress is, that the independence of Chrétien as a story-teller does not stand or fall with the existence or non-existence of Anglo-Norman Arthurian poems. Their importance, in relation to Chrétien, may easily be exaggerated by those unfamiliar with the character of oral tales. If we once accept as a principle the well-ascertained fact that such stories have a tendency to fall into a set form, a fixed sequence of incident and detail, would always be related in practically the same words, and, moreover, could well contain more than one sagen motif, we shall realise that the necessity of postulating a written source as explanation for the agreement in sequence, incident, and phrase, becomes infinitely less pressing.[91]
To my mind, the correspondences between the Welsh Arthurian tales and Chrétien's three poems in question offer no proof that the former repose directly on these poems as basis; but I consider it extremely probable that many of the perplexing features of the question—e.g. the occurrence in the Welsh stories, and in translations of Chrétien's poems, of details not to be found in the best mss. of those poems—may be accounted for by copyists and translators familiar with an oral version of the tale, filling in details which Chrétien had either never heard, or had purposely omitted. If we postulate, as from the character of the stories we are justified in doing, a very widespread knowledge of those tales, apart from any written source, we shall not be surprised at the existence of a large number of minor variants; the impossibility of explaining which on purely literary grounds drives Professor Foerster and those who share his views to the unsatisfactory expedient of multiplying MS. 'families.'[92]
To sum up the considerations advanced in the preceding pages, I think we are justified in saying that the real crux of Arthurian romance is the period before and not after Chrétien de Troyes. Not that the latter period does not offer us puzzles: it does, many and great, but when we arrive at some definite and proven conclusion as to the materials with which the earliest compilers of metrical romance were dealing, we shall have made a great step towards unravelling the problem of their successors. So far, I do not think we have arrived at such a conclusion; many theories are in the field, but none seem entirely to meet the conditions of the question. My own conviction is that, whether oral or written, Arthurian romantic tradition is of much older date than we have hitherto been inclined to believe.
To arrive at any solid result in our investigations there are certain principles which we must always keep in view, e.g., if the Arthurian tradition consists (as it admittedly does) largely of folk-lore and mythic elements, it must, so far as these elements are concerned, be examined and criticised on methods recognised and adopted by experts in those branches of knowledge—and not treated on literary lines and literary evidence alone. Thus it is essential to determine the original character of a story before proceeding to criticise its literary form. To treat stories of folk-lore origin from an exclusively literary point of view is to render a false conclusion not merely probable but certain.
In every case where an oral source appears probable, or even possible, we must ascertain, from the evidence of experts in story-transmission, what are the characteristics of tales so told, and what is the nature of the correspondence existing between tales of common origin but of independent development.
The evidence of proper names is valuable only in a secondary degree, as testifying to the place or places of redaction. But the older the story the less valuable they are as indications of original source, the oldest tales having a strong tendency to anonymity. So we find that in the lais the older versions only speak of 'a king,' the later identify that king with Arthur.[93]
If we take these elementary tests, and apply them to those of the poems of Chrétien de Troyes for which a traditional origin may safely be postulated, we shall I think arrive at the conclusion that there is little ground for ascribing inventive genius to the poet, whose superiority over his contemporaries was quantitative rather than qualitative. He differed from them in degree, not in kind; he had a keener sense of artistic composition, a more excellent literary style. Given the same material as his contemporaries he produced a superior result; when the material was deficient, as in the Charrette, the result was proportionately inferior.