agrees with Malory in LL. 1672-3969 i. e. from the scene in which Arthur comes upon Agravain and his brothers discussing Lancelot's intrigue with Guinevere down to the end,17 but in what precedes offers a unique version of the incidents in this part of the story.18 (d) Li Chantari di Lancellotto,1o an Italian poem (probably from the first half of the fifteenth century), adapted with considerable changes from the Vulgate Mort Artu and strongly contaminated with elements from the prose Tristan. This poem, which is of very poor quality, gives, also, a complete version of the Mort Arthur incidents.20

alliterative Morte Arthure (see p. 27, above) seems to have no connection with the prose romances.

17 Bruce, p. 92, Sommer, VI, 269.

18 For the proof of this, see my article in Anglia, XXIII, cited above more especially, pp. 75 ff., 87 ff.

19

The best edition is that of Walter De Gray Birch (London, 1874). It is written in ottava rima. For a detailed analysis and comparison of its narrative with that of the Vulgate cp. Bruce, RR, IV, 436 ff. It is not unlikely that this poem was based on some earlier Italian prose version of the Vulgate, rather than directly on the Vulgate, itself.

20

For foreign versions of the Vulgate Mort Artu which do not constitute separate redactions cp. my edition of that romance, pp. XXIII ff.

Hefperia, Ergänzungsreihe: 8.

29

Chapter V.

Date of the Vulgate Cycle.

We have seen that the ascription of the last three branches of the Vulgate cycle to Walter Map, which is so common in the manuscript tradition, is a fabrication, and that, consequently, we cannot accept it as affording any genuine evidence touching the date of the cycle. Nevertheless, we have in the well-known passage of Helinandus's chronicle, quoted above, a sufficient indication with regard to the downward limit of date of at least one member of the series viz. the Estoire del Saint Graal (Grand St. Graal) and hence, inferentially, of others. Since at the end of that chronicle the author speaks of King John of England as still reigning, it is indisputable that this work and hence the Estoire, to which it alludes was composed before 1216, the year of John's death. But, if the view of the evolution of

2

Cp. p. 254, above.

That this was the true terminus ad quem for the dating of Helinandus's chronicle was first pointed out in the Introduction, p. XXV, note, to the edition (1905) of that author's Vers de la Mort by F. Wulff and E. Walberg (Société des Anciens Textes Français). In my article, Arthuriana, RR, III, (1912) - particularly, pp. 185 ff. I called attention to the importance of this discovery for the dating of the Vulgate cycle.

In the passage cited above, Helinandus uses the following words: "Hanc historiam latine scriptam invenire non potui, sed tantum gallice scripta habetur a quibusdam proceribus, nec facile, ut aiunt, tota inveniri potest." Now, Brugger, Zs. f. frz. Spr. u. Litt., XXIX', 108 and XXXVI, 208, has interpreted these words as referring not merely to the Estoire, but to the whole Grail cycle of the prose-romances. As will be seen from the next sentence, one may grant the correctness of this interpretation only with the restrictions that the Merlincontinuation was not yet a part of the cycle and that the Lancelot had not yet attained its full enormous length.

In his elaborate and, in many respects, valuable discussion of the

the Vulgate cycle which is developed below is correct, only one of its branches, the Merlin-continuation, was composed later than the Estoire. We may, therefore, affirm with certainty that by 1216 the whole cycle, with the exception of this continuation, was date of the Vulgate cycle, Lancelot, pp. 126ff., Lot unfortunately overlooked Wulff and Walberg's discovery, which renders his dating of the composition of the cycle (p. 140), viz. 1221-1225 (inclusive) impossible.

It is worth recording that in a private communication, which I received from Professor Wolfgang Golther early in 1920, he expressed the conviction that the whole passage about the Grail in Helinandus is merely a late thirteenth century interpolation. He acknowledges, however, that he is unable to produce any "unmittelbare Beweise" in support of this view, and, so long as that is the case, it seems better to accept the passage as genuine.

The fact that Manessier and Gerbert in their continuations to Chrétien's Perceval use the Vulgate cycle is of no value for fixing the terminus ad quem of the date of the cycle, since the exact dates of these writers are unknown. Manessier wrote some time between 1211 and 1244. That is all we know. Gerbert wrote probably about 1225. For a fuller consideration of these matters cp. pp. 290ff., above.

The earliest MS. of any part of the cycle viz. the first division of MS. 768 (Bibl. Nat.), which contains the first part of the Lancelot dates from the middle of the thirteenth century. Cp. Lot, Lancelot, p. 135. The earliest dated MS. of any part of the cycle, MS. 342 (Bibl. Nat.), which contains the last part of the Lancelot, Queste and Mort Artu belongs to 1274.

Interesting, though of little value for the dating of the cycle, is the earliest known allusion to the prose Lancelot first pointed out by P. Meyer, Romania, VI, 494ff. (1877) - viz. in a thirteenth century MS. (British Museum, Add. 21212, fol. 4) of the prologue to a French version (now lost, if it ever existed) of the Philippis of Guillaume le Breton. The Lancelot is there referred to as "li livres Lancelot Ou il n'a de rime un seul mot." According to this prologue, the version of the Philippis was executed at the instance of Gile de Flagi, of whom the last mention we have dates from 1236. Meyer was inclined to ascribe the allusion to the end of the twenties of the century, but, obviously, this is pure conjecture. It may have been penned even after 1236, for no one knows when Gile de Flagi actually died.

in existence. To be sure, the Lancelot of that date was, most likely, of considerably less bulk than the huge work of our extant MSS.

In determining the upward limit of date for the cycle, we have, at least, one important fact to guide us - namely, that the Mort Artu was composed after the pretended exhumation of Arthur and Guinevere at Glastonbury in 1191, for in this branch we have obviously combined with the old Celtic tradition of the wounded Arthur's journey to Avalon the new invention which the monks of Glastonbury first put into circulation in that year to the effect that the great king and his consort were really buried in the local abbey. It does not, however, seem reasonable to suppose that the author of the Mort Artu would have made so serious a concession to this invention, unless it had already gained a pretty widespread credence at the time that he composed his romance, and for that a period of at least ten years would be required; indeed, an even longer time would seem more likely. In view of these circumstances, it appears safe to conjecture that the composition of the Mort Artu fell somewhere in the neighborhood of the year 1205. But this romance presupposes the existence of the Lancelot (in some form or other) and the Queste, so that the composition of these latter romances would fall in the closing years of the twelfth century (doubtless, the last decade). On

4

Cp. p. 264, above.

B All sorts of efforts have been made to establish the terminus a quo of different members of the cycle-especially, by means of the sources. But in not a single instance can we fix definitely the date of these sources themselves. In connection with this question, it should be observed that Lot, Lancelot, p. 134, is wrong in limiting the beginning of Wauchier's literary activity to 1206. Brugger, Zs. f. frz. Spr. u. Litt., XXXVI, 45 ff. (1910) had already proved convincingly that this limitation (which originated with P. Meyer, Hist. litt. de la France, XXXIII, 291) was unwarranted and that his Perceval-continuation, indeed, may possibly have been written as far back as the eighties of the twelfth century. Moreover, most students who make a distinetion between Wauchier and Pseudo-Wauchier and it is the latter on which the authors of the Lancelot and Queste have drawn particularly will agree that whatever may be the date of Wauchier, Pseudo-Wauchier's date is still earlier.

--

It is not necessary to discuss again in this connection such

the other hand, the Estoire del Saint Graal, which, as we have seen, was certainly in existence by 1216, contains allusions to the Mort Artu consequently, we may place the composition of this branch with a considerable degree of confidence between 1205 and 1216. Finally, the Merlin-continuation of the cycle, which was written to connect the Estoire and the Lancelot, was, of course, later than either, and, if we may judge by its allusions to the Perlesvaus, later even than that romance. The most probable date, then, for this latest division of the cycle would be a few years before 1230.

fundamental sources of the Vulgate cycle as Chrétien's Perceval (not to mention his other works) or the French original of Ulrich's Lanzelet, since the dates of these sources are too indefinite to help us materially in our present inquiry. If the dates, however, of these romances, respectively, which we have argued for above are accepted, we should have the eighties of the twelfth century as the terminus a quo for the Lancelot the earliest member of the cycle. The same conclusion is fortified with regard to the cycle, in general, by our dating of Robert's Joseph (a source of the Estoire and Queste) and Merlin (used in the Vulgate Merlin and Mort Artu).

The Tristan romances, which are drawn upon so freely in the Mort Artu, are relatively too early to add anything to the evidence in the case. The chansons de geste, Gaydon and Parise la Duchesse, which are apparently sources (one or the other, or both) of the poisoned fruit incident in this branch were written not later than the first part of the thirteenth century, but the extant form of Gaydon is probably not the earliest, and both poems may really be productions of the latter part of the twelfth. For a detailed examination of the evidence, cp. Bruce, Mort Artu, pp. XXIXff.

The date of Raoul d'Houdenc's Meraugis de Portlesguez (a source of the Lancelot and Merlin) falls in the latter part of the twelfth or first part of the thirteenth century no one can say which. Cp. Friedwagner's edition, pp. LXIII, f. (Halle, 1897).

Lot, Lancelot, p. 165,- observes that the raising of the host (cp. Queste, VI, 189) is first mentioned in the statutes of Eudes de Sully, bishop of Paris, 1196-1208. But the practise was certainly earlier. See the eleventh century example from St. Gall, given by Karl Young, The Dramatic Associations of the Easter Sepulchre, pp. 30f. (Madison, Wisconsin, 1920).

Following the authorities on mediaeval armour who state, apparently with one accord, that horse-armour was first introduced in the