Students of folk-lore will note that the tale in this form includes features not found in the majority of the versions, but representing well-recognised folk-tale formulæ. Thus the Life Token is here—incomplete—the maiden gives the fisherman ‘something’ to be given to his wife, his horse, and his dog (obviously a fisherman does not need a horse and a dog—these two features do not belong to each other); the wife has three sons, the horse three foals, and the dog three pups. Horse and dog ought rightly to play a part in the story, but in this special variant they do not appear, though in another they are mentioned in a subordinate rôle. The Grateful Beasts and the External Soul are equally well known in folk-tale, though again, as a rule, in a different connection. But the tournament is lacking; and after examining many variants of the tale, I have come to the conclusion that this feature belongs exclusively to the continental versions. Horses and dresses are found in the insular forms, but, so far, I have not found a single instance of the tournament. On the other hand, no continental variant appears to contain the sea-maiden episodes.

If we now summarise the leading incidents of the various groups, we shall find them somewhat as follows:—

1. Hero—King’s son. Herdsman or shepherd. Fisherman’s son turned herdsman.

2. Slays three giants and wins three castles in which he finds three steeds of different colours with dresses or armour to correspond. The horses are occasionally winged.

3. Appears at a Three Days’ Tournament in these dresses, and thus wins the hand of a princess.

(Incidents 1, 2, 3, which combined correspond to Le Petit Berger, form the shortest version of our story, but probably not the most primitive.)

4. Rescues the princess from an ‘Otherworld’ prison. Form of imprisonment varies, but the ‘rescue’ is most generally found in company with the tournament.

5. Rescues princess from a monster. Here the conflict generally lasts three days, the three disguises are employed, and the tournament is often absent.

6. Is robbed of the credit of his deed by a cowardly rival. This, which is most generally found in combination with 5, is also sometimes found in a modified form combined with 4, and is often lacking altogether.

7. Is carried off by a mermaid, to whom he had been promised before his birth. This appears to be confined to the Celtic group collected by Mr. Campbell.

If the reader will refer to the various examples I have given above, he will see that these seven incidents represent what we may call the perfect skeleton of our story (to use a simile often applied by Mr. Campbell), though the bones are differently placed in different versions.

But, having summarised them, we also become aware of a very curious coincidence. Out of these seven incidents, six are found, and found more than once, in the earlier forms of the Lancelot story. Thus dropping out incident 2, the winning of the armour, to which I know no good parallel, we find that Lancelot was a king’s son (incident 1), which, in itself, of course counts for little, but is of value in combination with other features (Lanzelet—Prose Lancelot); that he appears at a tournament, three days running, in different armour, the colours of which correspond with the prevailing colours of the folk-tale—green, red, white, or black, red, white (incident 3) (Lanzelet—Prose Lancelot); that he frees a princess (queen) from an Otherworld prison (incident 4) (Charrette—Prose Lancelot—Lanzelet, modified form); that he slays a monster (apparently a dragon), and is robbed by a cowardly rival (incidents 5 and 6) (Morien). A second version of the False Claimant story is found in Le cerf au pied blanc. Finally, when a child, he was carried off by a water maiden, meer-wîb (incident 7) (Lanzelet—Prose Lancelot).

Now these are characteristics which, in their ensemble, he shares with no other Arthurian hero. True, Gawain visits the Otherworld, but he does so rather in the character of lover of the queen of that world than as rescuer of one confined within its precincts. In the Dutch Walewein alone, so far as I know, is his rôle definitely that of the deliverer. But none of the other incidents belong to his story. So, too, Tristan is the hero of a very fine version of the Dragon Slayer and False Claimant story, and it is moreover probable that the Morien version has borrowed certain details from the Tristan, but he too can claim no share in the other incidents. The close correspondence, point by point, with a folk-tale of so widespread and representative a character, is, I submit, a peculiarity of the earlier Lancelot story, which is of extraordinary interest as throwing light upon the genesis and growth of Arthurian legend.

In this connection I have by no means forgotten the energetic protests which, in certain quarters, were evoked by Mr. Nutt’s attempt to show that the story of Perceval might in this way be connected with popular tales; and I am quite prepared to be told that tales collected in the nineteenth century are not to be trusted as indications of the sources of twelfth century romance. But in the instance before us the evidence, while of precisely the same nature as in the case of Perceval, exceeds it, both in bulk and extent. The story is not one story, but a large and well-marked group of tales; the folk-lore parallels affect not one, but many incidents of the romance. How large and how widely diffused is that story-group can only be appreciated by those who will examine the lists of variants appended by M. Cosquin to the four stories I have named above and those cited by Mr. Campbell under the heading of the Sea Maiden, and then compare these stories with the numerous examples given by Mr. Hartland in his exhaustive study of the Perseus legend. The incidents are, as I have shown, six out of a possible list of seven. If, further, we remember that the group, with all its varying forms, is connected with such pre-historic heroes as Perseus and Cuchullin, we have, I think, a sufficient answer to those critics who would reject the evidence en masse on the ground of modernity.

But supposing, for the sake of argument, that we accept the possible priority of the romantic over the popular form, what, with regard to the criticism of the Arthurian literary cycle, is the logical result? This: if the folk-tale be dependent upon a romance, that romance must of necessity be the Lancelot, as no other hero offers the same combination of incident. But a version of the Lancelot story, from which all these incidents could have been borrowed, must have been older than any form of the story we now possess. As we have seen above, the correspondence is sometimes with one, sometimes with another version; and a very famous incident of the tale, the False Claimant, only exists now in two romances, each of them preserved in an isolated and unique form. Therefore, if this be not a fully proven instance of the conversion of a popular folk-tale into an Arthurian romance, it must be a case of the development of a folk-tale from a fully organised and coherent Lancelot story in a form anterior to Chrétien. The adherents of the theory which ascribes independent invention to Chrétien de Troyes, and a literary origin to the Arthurian stories, can make their choice between these two solutions of the problem—one or the other it must be.

For myself, I unreservedly accept the verdict pronounced by Mr. Campbell upon the Sea Maiden as representative of the entire story-group. ‘Is it possible that a Minglay peasant and Straparola[47] (or we may add Hue de Rotelande and the peasants of the Odenwald and Lorraine)—neither of whom can have seen a giant, or a flying horse, or a dragon, or a mermaid—could separately imagine all these impossible things, and, having imagined them simultaneously, invent the incidents of the story and arrange so many of them in the same order?

‘Is it on the other hand possible that all these barefooted, bareheaded, simple men, who cannot read, should yet learn the contents of one class of rare books and of no other? I cannot think so.

‘I have gone through the whole Sea Maiden story, and all its Gaelic versions, and marked and numbered each separate incident, and divided the whole into its parts, and then set the result beside the fruit of a similar dissection of Straparola’s Fortunio, and I find nearly the whole of the bones of the Italian story, and a great many bones which seem to belong to some original antediluvian Aryan tale. The Scotch (insular) is far wilder and more mythical than the Italian (continental).[48] The one savours of tournaments, kings’ palaces, and the manners of Italy long ago; the other of flocks and herds, fishermen and pastoral life; but the Highland imaginary beings are further from reality and nearer to creatures of the brain. The horses of Straparola are very material and walk the earth; those of old John MacPhie are closely related to Pegasus and the horses of the Veda, and fly and soar through grimy peat-reek to the clouds.’[49]

Mr. Campbell continues: ‘What is true of the Gaelic and Italian versions is equally true of all others which I know. If examined, they will be found to consist of a bare tree of branching incidents common to all, and so elaborate that no minds could possibly have invented the whole seven or eight times over[50] without some common model, and yet no one of these is the model, for the tree is defective in all, and its foliage has something peculiar to each country in which it grows. They are specimens of the same plant, but their common stock is nowhere to be found.’[51] Were Mr. Campbell living now, may we not feel sure that to these closing words he would add: Assuredly it is not to be sought in an Arthurian romance of the twelfth century?

THE ROMANCE

So much for the present as regards our folk-tale as a whole. Let us now see what light the study of it may have thrown upon the special subject of our investigation—the Three Days’ Tournament. And first of all, I think it has definitely settled the correctness of our title. East or west, north or south, wherever we have traced our story, whatever the hero’s feat—whether the rescuing the princess from a devouring dragon, or the winning her hand at a knightly tournament—the days required to complete the task are three—neither more nor less.

Mr. Hartland, to whom I referred the point, remarks that the unvarying tendency in certain families of folk-tales, notably those of Oriental origin, is to crystallise a small but indefinite number into three. Now Mr. Campbell, as we have seen, detects a likeness between the flying horses of the Sea Maiden tales, and the horses of the Veda, and Mr. Joseph Jacobs, in a note appended to another tale,[52] quotes a further remark of the same writer, to the effect that the many-coloured horses of Indian mythology may account for all the magical horses of folk-tales. So if our tale, as a whole, did not come from the east, it seems possible that this particular incident may have done so.[53]

Yet in so far as the tournament form is concerned, it is, of course, possible that certain literary versions of the story might have been affected by the ordinary customs of the day. Anyway there seems to be a fairly close correspondence here between fact and fancy. Niedner, in his work on Das Deutsche Turnier,[54] remarks that the tourney proper was generally held on a Monday; the knights assembled on the previous Saturday; Sunday morning was spent in mustering those present and arranging the opposing factions; while the afternoon was devoted to the encounter known as the Vesper-spiel, preliminary to the grand struggle of the morrow. Thus the ordinary duration of such a meeting might be reckoned as three days.

But it is clear that there might also be three distinct encounters on as many separate days, as in the folk-tale. Professor Kittredge, in his article, ‘Who was Sir Thomas Malory?[55]’ notes a very remarkable and pertinent instance taken from the life of Richard Beauchamp, Earl of Warwick. When that nobleman was Governor of Calais, hearing of a great gathering of knights, to be held in the neighbourhood, ‘he cast in his mynde to do sume newe poynt of chevalry’; and under the several names of The Grene Knight, Chevalier Vert, and Chevalier Attendant, sent three challenges to the French king’s court. These being accepted, he appeared the first two days in differing armour, the third ‘in face opyn,’ on each occasion overthrowing his antagonist.[56] The days in question are given by Rous as January 6th, 7th, and 8th; the year he does not mention; but Professor Kittredge, by a process of elimination, arrives at the conclusion that it must have been either 1416 or 1417. It is, of course, obvious that this feat must have been suggested by the romances. It is, I think, equally obvious that the three days of the romances were not at variance with actual practice. As to the version of the folk-tale there can be no question. The correct number is three—neither more nor less.

It is, of course, also clear that the occurrence of the tournament in the folk-tale must be subsequent to the institution of tournaments as part of the ordinary chivalric and social conditions; but the tale itself must be earlier, as is witnessed both by the archaic nature of the rescue incident and the magical nature of the horses. Trials of skill in horsemanship are known to all stages of society; and the original form of this special incident was doubtless something of this kind. In the Odenwald variant referred to above, the hero has to perform the feat of carrying off on his spear a ring suspended from a beam, and to hang it up again in returning. This is here supposed to form part of the tournament; but it seems most likely that in earlier forms the trial of skill by which the hero was tested and identified was simply some such feat of skilled horsemanship.

Nor do I think that we are to see the influence of romance, rather than of custom, in this transformation. Neither of the poems in which the incident approximates most closely to the folk-tale form, the Lanzelet and the Ipomedon, appear to have been particularly popular (certainly not the former), judging from the number of manuscripts in which they have been preserved, while the ‘Tournament’ form of the folk-tale is found all over Europe. It is much more reasonable, surely, to conclude that the episode has been borrowed, as so many others have been borrowed, from the stores of popular tradition than to hold that in this case popular tradition has been modified by the influences of a literary cycle.

But is it not as clear as daylight that all this immense body of evidence absolutely and finally disposes of any claim on the part of Chrétien to be first in the field? The four days of Cligés rule that romance, as a source, out of court at once and for ever. Further, not only is that version demonstrably secondary in itself, but definitely secondary to and dependent upon the Lancelot versions. These correspond with the prevailing colours of the folk-tale—black, red, and white, or green, red, and white.[57] The one is the version of the Prose Lancelot, the other of the Lanzelet. Chrétien not only gives one day too many, but manifestly does so in order to combine the two versions which he, in common with us, knew, and gives both green and black—two colours which are found together in no single version of all the dozens I have read.

There is a possible ‘clerical’ explanation of the existence of two versions of the Lancelot tale. Noir in the manuscript may have been read vair, and a copyist writing from oral dictation may thus have substituted vert. But in the face of the green, red, and white of the very primitive Celtic variant given by Mr. Campbell, and confirmed by the Greek parallel, I think it more likely that the three colours of the Lanzelet represent the older form. But inasmuch as in romances, which, like the Arthurian, were supposed to correspond in some measure to the conditions of real life, a green horse would be an impossibility, while yet horse and armour should correspond, black—perhaps under the influence of the Perceval story—would take its place. Both were represented in the folk-tale, and it may be that the version of the Prose Lancelot and of the Ipomedon simply represents ‘the survival of the fittest.’

That there were two versions a closer study will, I think, make evident. Probably those who have followed the argument and illustrations closely will have already detected what hitherto I have left unnoted, that the version of the Ipomedon stands in a much closer relation with certain forms of the folk-tale, i.e. the Petit Berger or a group, than is the case with either the Cligés or the two Lancelot versions. In the Ipomedon alone the prize of the Three Days’ Tournament is the hand of the princess. And not only is there agreement in this, the leading, feature, but there is also a curious correspondence in minor details. Thus, both in the poem and in the folk-tale, the hero, in the character of a servant, has already won the princess’s love. In both she is bitterly disappointed at his apparent failure to compete. In the folk-tale she sends each evening to ask why the shepherd-lad has taken no part in the tourney, receiving each time the answer that he was unwell, but would do his best to appear on the morrow. In the poem, each evening Ipomedon sends word to the princess that it is he who has gained the tourney, but that he is leaving the country immediately, and will not be present on the next day. Thus the heroine, in each case, is kept in uncertainty as to the intentions of her lover.

If we add to this the correspondence with the Odenwald variant already pointed out,[58] and the fact that in the Ipomedon alone the hero is wounded on the third day—a feature found not only in the Odenwald story but in several variants of Le Prince et son Cheval—it becomes clear that if there be a doubt as to the source of the Cligés or the Lanzelet, the Ipomedon version must repose, directly or indirectly, upon the folk-tale.

But, as we have seen, it is precisely the evidence of the Ipomedon which leads us to connect the story with Walter Map, and the romance ascribed to him, the Lancelot. What, then, are we to conclude? I think the only satisfactory interpretation is that which I have suggested above, that there were two versions of the story; in one of which the hero was represented as winning, and probably wedding, the princess; in the other the incident, whatever its original form, had already been so far modified as simply to provide an effective setting for his first appearance at Arthur’s court. This is indeed what we find in the Lanzelet; and the general tone of that poem, wherein the hero wins the hand of no fewer than four ladies, and certainly weds three of them, shows that there would be no initial improbability in postulating another and more primitive form of the story.

To return to Cligés. The dramatis personæ of the tournament episode should be considered. The hero of the adventure does not compete with any number of knights, but is each day confronted with a chosen champion. These are, as I have already shown, Segramor, Lancelot, Perceval, and Gawain; and so far as the first three are concerned they appear here, and here only, their names, even, being otherwise unmentioned throughout the six thousand seven hundred and eighty lines of the poem.

To any one thoroughly familiar with the Arthurian romances, the juxtaposition of these three names is extremely significant. The adventure itself is elsewhere assigned to Lancelot. The hero with whom the Lancelot story in its earlier stages is most closely associated is Perceval; Chrétien himself here introduces Perceval as a famous knight, with whose renown Cligés was already familiar, and ranks him above Lancelot. One of the best-known adventures ascribed to Perceval is, as we have already shown, one in which the three colours, black, red, and white, figure, and in which he overthrows Kay in a manner curiously akin to other versions of the tournament episode. But previous to overthrowing Kay he had vanquished Segramor, who was the first to attack him. Is it not evident that Chrétien, like the authors of the Ipomedon and the original Lanzelet, was here reminded of the blood-drops adventure? If it be asked why introduce Segramor instead of Kay, we may recall the fact that while Cligés is represented as nephew to the Emperor of Constantinople, Segramor, as the Merlin tells us, was son to that potentate. Chrétien may have introduced him as less known in connection with this than Kay, who is never once named in Cligés; but I think it more likely that it was his parallelism to the hero, as well as his connection with Perceval, which determined his appearance.

But with regard to the latter, there is another point which deserves mention. In that section of the Peredur which does not correspond to any section of the Conte del Graal we find the hero, released from prison by the daughter of his jailer, attending a warlike tournament, in which each day he carries off the prizes; but there is no change of armour, and the days appear to be four instead of three. Previously to this he has also appeared three successive days at a tournament; but overcome by the beauty of the empress, of whom he is enamoured, he remains gazing at her, instead of taking part in the contest, until the third and final day. These passages are deserving of note, as they appear to me to show direct contact between the Perceval and Lancelot stories, and in this instance the borrowing appears to be on the part of the earlier story. Not only is Lancelot released from the prison of the Lady of Malehault to attend a tournament, thus corresponding with the one instance, but when he arrives on the spot he behaves in precisely the same manner at the sight of Guinevere as is recorded of Peredur with the empress. I do not feel able to accept the tournament as a real part of the Perceval story, no other feature of any version of the Perceval ‘Enfances’ corresponding with the formulæ of the group in question; yet the correspondence of detail between the two stories is so undeniable that contact of some sort, direct or indirect, there must be, and I think in this case we must hold that the Peredur has been influenced by a version of the Lancelot akin to that preserved in the prose redaction.

To return to Cligés. Taking into consideration all the evidence, the importance and widespread character of the folk-tale, the closer correspondence of both the Ipomedon and the Lanzelet to the popular form, and the peculiarities of the Cligés version, it becomes, I think, impossible to doubt that this latter, so far from being the source of the Lanzelet, is, as submitted above, not merely posterior to, but distinctly dependent upon a form of that story. And if we admit this, must we not also admit that here, at least, Chrétien did not understand the character of the material with which he was dealing, and that in this instance he certainly deserves the epithet which Professor Foerster asserts we would wish to apply to him, that of ein verschlechternder Ueberarbeiter? The phrase, be it remembered, is Professor Foerster’s, and not mine; but so admirably does it suit the present question, that I can only say, ‘I thank thee, friend, for teaching me this word!’ Chrétien was not dealing directly with popular tradition, but taking it at second-hand after it had already been modified and worked over in romantic form. To put it tersely, in the Three Days’ Tournament we have a folk-tale theme intelligently adapted by the authors of the Ipomedon and the Lanzelet, and misunderstood and ‘muddled’ by Chrétien.

THE BEARING ON THE LANCELOT STORY

But the interesting problems connected with this episode are not all solved when we have determined the ultimate source of the story, and the position to be assigned to Chrétien’s version. As we have seen, there is strong ground for believing that the French poet knew two versions of the Lancelot story; is it not possible that one of these versions may have been the lost French source of the Lanzelet? The ‘setting’ of the Cligés tournament, in which the hero makes his first appearance at Arthur’s court, corresponds with that of the Lanzelet; and, as we have remarked above, in the Erec we find not only the name of Lancelot, but also that of the enchanter Mauduiz, who appears nowhere save in U. von Zatzikhoven’s poem. Professor Foerster’s opinion is that we must consider the German Lanzelet as ‘die möglichst getreue Wiedergabe eines französischen Originals’; and on this point at least, I, for one, am quite prepared to agree with him. Whether, after a real study of that poem (with which I strongly suspect he had only a superficial familiarity), the learned professor will desire to maintain his opinion is another question! But, granting that the German version correctly reproduces the French original, the nature of the work—a loosely connected collection of independent tales, of marked folk-lore character—points to a period of evolution anterior to Chrétien’s well-knit and elaborately polished literary productions.

Then, again, there arises the question, Granting the existence of a Lancelot romance previous to Chrétien, could Walter Map have been the author? On this point it is not easy, with the material at our disposal, to express a decided opinion. Map and Chrétien were certainly contemporaries, but in neither case do we know the date of birth. Map died in 1209, therefore we may suppose he was not born long before 1140; a later date is scarcely probable, as he was a student at Paris in 1154, and at the court of Henry II. before 1162.[59] We do not know when Chrétien wrote the Erec, but it was almost certainly some time in the decade 1150-60. That Map should have been the author of a Lancelot poem earlier than the Erec is quite possible, but, perhaps, not very probable; but there would have been ample time for him to write one before the Cligés. Thus, while I think it highly probable that Chrétien borrowed from Map in the latter poem, I would reserve my opinion as to the former. Of the probable character of such a work we can gather some idea from Map’s undoubted literary remains; De Nugis Curialium offers abundant proof of the writer’s taste for popular tales and traditions. Had he lived in the nineteenth-twentieth centuries, instead of the twelfth-thirteenth, Map would undoubtedly have been a prominent member of the Folk-Lore Society.[60] His Lancelot poem might have been a short episodic romance of folk-tale character, a Three Days’ Tournament story, or it might have been a collection of such episodes, like the Lanzelet, i.e. its character would probably be popular rather than literary. I should myself have felt inclined to decide for the Lanzelet source, were it not for the evidence of the Ipomedon, which appears to presuppose a version closer to the original folk-tale.

Another point to be borne in mind in connection with the Cligés, and one to which I have already drawn attention,[61] is the peculiar geography of the poem, which is distinctly Anglo-Norman rather than Arthurian; the tale is obviously composed of originally independent themes; and whatever may have been contained in the book of the Beauvais Library, I think it is at the least possible that part of Chretien’s material came to him from insular sources.

As regards the Lanzelet, we know that the source of that poem came from England, and elsewhere[62] I have pointed out that a curious allusion to England (not as is more usual to Britain) seems to make it probable that the French original was written in this island. If we couple with this the authorship and evidence of the Ipomedon, and the persistent attribution of a Lancelot romance to Walter Map, we have, I think, a strong presumption in favour of an early insular version of that story.

While this study was in the printer’s hands I came across the following allusion to the slaying of a dragon by Lancelot; it occurs in the Auchinleck Manuscript version of Sir Bevis of Hampton (Cxxx):—

This allusion is the more interesting as, saving in the case of Morien, to which I have already referred, I have nowhere found this special feat attributed to Lancelot; certainly it does not occur in the whole extent of the Prose Lancelot, nor is it ever alluded to in that romance. Yet, if my theory of the evolution of the Lancelot legend be correct, such a combat ought certainly, at one time, to have formed part of his story. The evidence of this Anglo-Norman romance, supported as it is by the independent testimony of Morien, is therefore especially welcome; I am inclined to think that it strongly increases the probability of a definitely insular version of the story, differing in some respects from the continental, having existed at the time the ‘Sir Bevis’ was written.

Nor would the existence of such a version be, as Professor Foerster asserts, incompatible with the continental origin of the character;[64] to assert as much is really to stultify his own arguments. Does not the whole system of Professor Foerster rest upon the hypothesis that the character of Arthur, indisputably of insular origin, underwent development upon continental ground? The fact that what he roundly denies of Arthur he asserts emphatically as natural for Lancelot throws a flood of light upon the ex parte character of this distinguished scholar’s methods!

If we take into consideration the character of the elements composing the early Lancelot story, a character which, be it remembered, is not a question of suggestion but a matter of proof, we shall become clearly aware that the material for development existed on both sides of the Channel. I believe myself that Lancelot was of continental origin, but I recognise clearly that if the source and development of his story were such as I suppose them to have been, that continental origin was a matter of accident, not of necessity; and if some other scholar should bring forward arguments to prove that the story had its rise on insular rather than on continental ground, I shall be quite prepared to reconsider the question.

So far as the evidence I have now collected is concerned, it looks as if the development of the early Lancelot story might thus be sketched:—

a. Lai (presumably Breton), relating theft of king’s son by water-fairy, amplified by

b. Bringing up of youth in Otherworld kingdom, peopled by women only (source, general Celtic tradition, possibly Gawain legend).

c. His entry into the world (Perceval legend).

d. Introduction of adventures of Sea Maiden story, a being the point of contact, and suggesting the development, which may have been as follows:—

d^a. Winning of magic steeds and armour.

d^b. { Rescue of princess from monster, and False Claimant story; or

d^c. { Rescue of princess from Otherworld. As we have seen (p. 25), it would be quite possible for these to be combined.

d^d. Appearance at Three Days’ Tournament.

It would seem not improbable that it was the independent existence of incident d^c in the popular tale that led to its coalescing with the Arthurian legend. As I have elsewhere pointed out,[65] the character of the Guinevere abduction story is in itself so primitive that it may well have formed part of the earliest stratum of Arthurian tradition. The variants are of such a nature as to indicate that they arose at a period when the real meaning of the story was still understood, and carefully retained. The tale must therefore be far older than any extant literary version.

If we admit the suggested hypothesis—that the hero of the Lancelot lai became through the ‘mermaid’ incident identified with the hero of the Sea Maiden story—the character of that story, and the immense popularity to which its wide diffusion testifies, would give us a solid working hypothesis to account for the choice of Lancelot as Guinevere’s lover. The similarity of the stories led to his identification with her rescuer, and that step once taken the recognition of him as her lover was—given the social conditions of the time and the popularity of the Tristan story—a foregone conclusion.[66]

But this evolution, so far as we can tell, took place on both sides of the Channel. Thus, while I have found no single insular version which gives the Tournament episode, I have equally found no continental variant which contains the mermaid. Yet it is the latter (mermaid) which appears to form the point of contact between the folk-tale and the lai, while it is the persistent recurrence of the former (the Tournament) which has given us the key to disentangle the complicated evolution of the story.

Here is a point on which I should wish to make my position perfectly clear. I do not think that Lancelot was ab origine the hero of a variant of this popular and widely-spread folk-tale. The persistent element in the Lancelot story is, as I have elsewhere shown, his connection with the beneficent Lady of the Lake. Now the maiden of the folk-tale is a sea, not a lake, maiden, and is, further, consistently represented as of a malicious, rather than a kindly, character. True, she aids the fisher in the first instance, but she belongs to that order of beings whose gifts, apparently desirable, are saddled with conditions which turn to the undoing, rather than to the profit of the receiver. Also, her presence in the story is restricted to a small and well-marked group of variants, which apparently preserve a primitive type of the story, and are never combined with the Tournament, which recurs so frequently in the Lancelot romances.

Again this folk-tale, quâ folk-tale, does not belong to the same group as that which offers parallels to the Perceval story; yet the Lancelot story was certainly affected, and that at an early stage of development, by the Perceval. Folk-lore students are well aware of the facility with which one story-type can become contaminated by another originally distinct from it; and while I see in the common ‘folk-tale’ origin of the two legends a satisfactory explanation of the undeniable influence traceable through all the earlier stages of the Lancelot evolution, I would yet distinguish sharply between the two heroes. Perceval is a British (insular) Celt; Lancelot a continental (Breton) Celt, the development of whose story is posterior to that of the insular hero. For all these reasons I think it most probable that Lancelot was the hero of an independent, and originally short, tale, which by an accidental similarity of incident became connected with one of the most popular of known folk-tales, from which it freely borrowed adventures, and which, through the medium of one of these adventures, became later incorporated with the Arthurian tradition and developed upon romantic lines.

EVIDENCE FOR AN INSULAR VERSION OF THE ROMANCE

The whole character of the earlier Lancelot story is strongly reminiscent of a lai, and I see no reason to depart from the opinion expressed in my Lancelot ‘Studies,’ that the root of the whole wonderful growth is to be sought in such a lai.

Nor do I see reason to doubt that this lai may have been of continental origin, and at the same time have taken this most important step in development upon insular ground. I cannot agree with those scholars who appear to regard the Channel as an impassable barrier to communication previous to the date of Chrétien de Troyes, and the most facile medium of intercourse immediately after that date!

For more than a century previous, i.e. from the days of Edward the Confessor, intercourse between the English court and the north of France had been frequent and continuous; for nearly a century the kings of England had also been princes of France. When, therefore, we find, as we do, that the materials for the development of a story existed on both sides of the Channel, and that the story, in its completed form, is akin to both continental and insular variants, forming, as it were, a link between the two, and combining forms which are not known to meet elsewhere, the conclusion that the process of evolution was not confined to one country appears neither illogical nor unfounded.

I would, therefore, now suggest that we have solid grounds for supposing that the story of Lancelot, starting as a Breton lai, and brought in that form to England, became in these islands connected with a special variant of a very widely diffused folk-tale. Having borrowed from this tale certain adventures, it found its way back, in this enlarged form, to the Continent, where the story from which it had borrowed being equally well known, it underwent further development on the same lines. I suspect that here the flying horses of the Celtic tale became transformed into the normal steeds of the Three Days’ Tournament, though the colour of the armour—green, red, and white—was at first retained.

But on which side of the Channel was the final and most important step, the incorporation with the Arthurian cycle, taken? Of the various versions of Guinevere’s abduction, the Melwas story exists only in an insular text, the Vita Gildæ, and this is apparently connected with a partly lost and entirely confused Welsh tradition. The Meleagant version is by locality directly connected with Melwas; and the only extant version of the Falerîn abduction tale came from England. I submit that here again we have reasonable ground for the hypothesis that the identification of Lancelot as Guinevere’s rescuer, and subsequently as her lover, may be due to insular rather than continental development. The question is, as will be seen, by no means an easy one, and I should prefer to express no definite opinion as to the real bearing of the evidence here adduced. There are, as I have shown, indications pointing in opposite directions. The precise value and relation of these indications will be better realised as we become more familiar with what is at present a somewhat novel interpretation of the facts. In any case it will be seen that the theory here advanced only affects the earlier stages of the Lancelot story, leaving untouched the question of its development as part of the Arthurian romantic cycle. It affords us a working hypothesis which may enable us to bridge the gulf between Lancelot the independent hero (Lanzelet) and Lancelot the queen’s lover (Charrette), a gulf which has hitherto presented a problem baffling to the Arthurian student.

But is it not also apparent that, in the light of the evidence here collected, the theory of an Anglo-Norman Arthurian tradition, independent of, and anterior to, Chrétien’s poems should no longer be contemptuously derided? Whatever may be the eventual verdict on the evolution of the Lancelot story, the examination of the various romantic versions of the Tournament story, in the light of folk-lore evidence, has, I think, made absolutely clear to any unprejudiced critic that the Cligés version cannot possibly be the source of either the Lanzelet or the Ipomedon, but represents a version further removed from the original form, and in all probability dependent upon some variant, or variants, of the Lancelot. And if this be the case in one poem, and that the very poem in which the admirers of Chrétien assert roundly that his independence is most clearly shown, are we not justified in our hesitation as regards his other works?

In my Lancelot ‘Studies’ I showed that Professor Foerster’s theory as to the origin of the Yvain would not bear the test of strict examination; that evidence, both internal and external, could be adduced in favour of the view that the tale was but a collection of lais, put together and worked over by others before Chrétien gave the final touch which converted them into a literary whole. Before long I hope to show, what I have recently recognised as a fact capable of demonstration, that the Perceval ‘Enfances,’ so far from being the source of the other versions, is but an incomplete and inferior version of a story, which in its original and perfect form no longer exists, but is better preserved elsewhere. Erec, so far, I have not examined, but I have little doubt that the result of careful investigation will here be the same; certain it is that the initial adventure, the chase of a fairy stag, represents a superstition alive in these islands to this day. The trackers on Dartmoor claim to be able to distinguish the ‘slot’ of the fairy deer from among all others, and will solemnly warn the huntsmen of the futility of following such a trail.

Those of us, and they are many, who entertain a profound respect, not merely for M. Gaston Paris’ learning, but also for his keen critical instinct, and what I can best express as ‘sense of atmosphere,’ have hesitated, even though little evidence appeared to be forthcoming, to dismiss lightly, not to say discourteously, a theory which had the support of his authority; the foregoing pages will, I hope, show grounds for believing that an investigation, conducted perhaps on somewhat different lines to those hitherto in favour, will fully justify this hesitation.

We are only on the threshold of Arthurian criticism, and till we have thoroughly familiarised ourselves with the elementary conditions of the problem before us, it is both premature and unscientific to expect to obtain in any section of this wide field a result which can be claimed as permanent. Thoroughness is an admirable quality; but the thoroughness which consists in carefully and microscopically surveying a single part, before we have ascertained the relation of that part to the whole, is only too apt to result in throwing that whole hopelessly out of focus. The time has not yet come when a final study of any part of the Arthurian legend, based upon a comparison of all the texts, is possible or indeed desirable. The different threads that form the shifting pattern of the fabric are so interwoven that no one can as yet be disentangled beyond a certain point without injury to the whole.

Thus neither the Gawain, the Perceval, nor the Lancelot stories can at the present moment receive satisfactory and final treatment. In the advanced stages of Arthurian legendary development these three main lines of tradition have become so entangled, have crossed and complicated each other to such an extent, that it is only by following what we may call a parallel method of study that we can hope to determine their exact relationship to each other; while until that exact relationship be accurately determined, a scientific study of the cycle, as a whole, is impossible. There appear to me to be three possible lines of investigation, any one of which will probably throw light on the other two; while the results to be obtained from all three would go far towards providing a sound and scientific basis for future inquiries. These three are (a) The various versions of the Gawain Grail quest; absolutely necessary if we desire to understand the development of the Grail section of the cycle. (b) The Perceval continuations; which contain sections belonging to early and non-cyclic versions of the stories affected, combined with sections drawn from later and cyclic redactions. These texts will also throw light upon the small and interesting cycle of the Bel Inconnu, which is connected with all the three lines of tradition, and is important for all. (c) A comparative study of the various Lancelot versions, which will enable us to disentangle the earlier Perceval-Lancelot redactions from the later Galahad development.

But in this investigation there are certain principles which must be kept clearly in view. We must remember that a cycle like the Arthurian cycle, compounded largely of what we may call mythical and imaginative elements, and largely devoid of historical basis, cannot be examined and criticised on the same principles and by the same methods as can the Charlemagne cycle, where historic conditions, though modified for romantic purposes, have controlled and shaped the process of development.[67]

In this latter case an appeal to documentary evidence, and a criticism conducted largely on literary lines, is, by nature of the material to be dealt with, entirely in its place; in the former, inasmuch as the material of which it is composed belongs far less to history than to that indefinable body we call popular tradition, which never finds more than partial expression in literature, and yet maintains its character practically unchanged throughout the centuries, we must follow a different method.