His invasion is ignored; we are told simply that John died and Henry succeeded him and was crowned by the legate Gualo. There is therefore no evidence at all for this view.

It is erroneous, in reply to the third objection against the chronicle, to suggest that the abbey of Margam was too obscure to be well informed. Just as Coggeshall was in a land of royal forest and manors, near London, just as St. Albans was on one of the great roads, so Margam had special advantages for hearing strange information. Gerald of Wales speaks of its importance, its hospitality, its connection, when scarcity of corn made connection useful, with Bristol. When we turn to the Margam records we find no ignorant and secluded community, but a powerful house, favoured and harassed alternately by great neighbours who were some of the greatest barons in England and the Marches,2 an abbey which lay on the road from England to Ireland, and was twice visited by King John himself at one time under the king's special protection, favoured almost as much as his peculiar foundation, the Cistercian house of Beaulieu. The delightful studies of M. Bédier have shown us that the information and influence of a monastery depended not so much upon its general position as upon the road on which it lay, or upon what friends the abbot had. He has demonstrated that the isolated and obscure house of Saint-Guilhem-du-Désert could mould the history of a great epic cycle, because it was visited by pilgrims on their way to Compostella.5 Conversely special information could make a chronicle of

1. Opera (Rolls Series), vi, 67, 68.

2. G. T. Clark, Cartae et alia Munimenta quae ad Dominium de Glamorgan pertinent, especially vol. iii, passim (Cardiff, 1891).

3. Rot. de Liberate, 172, 229; Annales Monastici, i, 10. In his History of Margam (London, 1877) Mr. W. de Gray Birch suggested that there was some connection between John's presence at and favours to the abbey, and its chronicler's knowledge of Arthur's death (pp. 176-180).

4. Ann. Mon., i, 30.

5. Les Légendes Epiques, vol. i (Paris, 1908).

the most meagre and unpretentious range a very valuable authority. The monks of Coggeshall knew a great deal more about Richard's captivity than did many great abbeys, because Anselm, the king's chaplain, 'told us all these things as he saw and heard them.' Now is it possible to suggest the chief channel of communication open to the monks of Margam?

In reading the chronicle one or two suggestions occur to mind which must be put aside. It might be observed that the compiler seems to have been interested in Bec. He knows that Sainte-Marie-de-Pré is a priory of Bec; he notes that Hugh of Nonant, bishop of Coventry, died at Bec in 1198. Again, it is worthy of mention that in November 1203 Margam had an agent at Rome, who was engaged in securing lengthy privileges and confirmations from Pope Innocent III.1 On his journey to and from Rome the person entrusted with the business of the abbey, whether a monk or not, could acquire information which might interest his employers. But it is not very likely that this would be of unique importance. Let us approach the problem from the other direction and ask who was likely to know what happened before and after the murder of Arthur. Ralph of Coggeshall says that Arthur was entrusted to the care of Robert of Vieuxpont at Rouen; but Robert was a north-country magnate, nor does he appear in the story of the murder. He was a busy official who probably did not live constantly at Rouen.2 Two of John's companions and counsellors however were very conspicuous in Glamorgan, and both of them probably knew a good deal more than they cared to say. William the Marshal, earl of Pembroke, and William of Briouze (de Braosa) 1. Clark, op. cit., iii, 225-234.

2. Coggeshall, p. 143. He was bailiff of Caen and the Roumois in 1203, and is identified by Stapleton with the Robert of Vieuxpont who was lord of Westmoreland, and clung to John in 1216, while his brother joined the rebels (Stapleton, II, cclxiv-cclxvii; cf. Farrer, Lancashire Pipe Rolls, p. 258). After the loss of Normandy, Robert got some of Ralph Tesson's lands in Kent (Rot. Norm., p. 140). Below, p. 519.

granted privileges to or attested the charters of Margam more than once. The Marshal kept absolute silence. It is difficult to say to what extent he knew how Arthur died. He was certainly acquainted with the course of the negotiations which followed the murder during 1204–1205, since he was one of the embassy. I think that his biographer knew a good deal, and hints at Arthur's fate, but there is not a word of explicit reference to the matter in the poem which tells us so many new things. 1 Nor were the Marshal's lands in South Wales near the abbey of Margam. But William of Briouze was in a very different position. The story of his life would, if it were thoroughly known, be the most important record we could have of the personal history of John and his baronage during the first part of the reign. He was the king's constant companion during the Norman campaigns. It is well known that the official records reveal the presence of John near Rouen just about the time when, according to the Margam annals, the murder was committed.2 William of Briouze was with him at the time. About 1207 he lost the king's favour, and in 1210 John tried to exterminate him and his family. His wife, Matilda, is said to have refused to hand over her children as hostages to the murderer of Arthur, and John pursued her thereafter with a ferocity unusual even in him. The grisly story of her and her son's death by starvation in Windsor is the most awful of many awful tales. It is impossible to believe that the debts of William of Briouze were, as John said in the official account, the cause of this

3

1. There are possible hints in ii, 81, 145. For the Marshal's embassies, see vol. iii, pp. 176-178, with Meyer's notes.

2. See the itinerary appended to Sir T. D. Hardy's introduction to Rot. Litt. Patent. (1835); cf. Miss Norgate, ii, 430. That William of Briouze was present is clear from the attestations; e.g., Rot. Norm., p. 86.

3. See Meyer's long note in Hist. de Guillaume le Maréchal, iii, 156; Dict. of Nat. Biogr. s.v. 'Braose,' for authorities; e.g., Rog. Wendover, ii, 49 (Rolls Series).

persecution.1 The natural supposition is that this chosen companion knew too much to be allowed to live after he and so many others had quarrelled with the king. In 1210 he managed to escape to France; in 1211 he died and he was buried at Corbeil on the eve of St. Lawrence.2 All this we know apart from the evidence of Margam.

3

Now by far the most conspicuous person in the annals of Margam, and one of the most important figures in its records, is this William of Briouze. He was lord of Brecon, Radnor, and Gower. Between 1202 and 1207 he was responsible for the administration of Glamorgan, in which Margam lay. He attests the charters of local benefactors to the abbey. 4 In the annals we are told how William of Briouze was chiefly responsible for John's accession to the throne in spite of his previous condemnation. Except the great semi-official chronicler, Roger of Howden, the Margam annalist is the only writer to mention this condemnation of John at the court of King Richard. He is interested in William's life and alone tells us that after his death in France he was buried by the exile Stephen Langton, archbishop of Canterbury. Finally, the relations of John and William were a theme of popular tradition in South Wales nearly eighty years after the death of Arthur. On February 1203 John had granted the land of Gower to William. In 1279 the earl of Warwick contested the right of William's descendant to this honour, and especially to 1. Calendar of Documents relating to Ireland, i, no. 408.

2. Rog. Wendover, ii, 59; Matthew Paris, Chron. Mai., ii, 532; Annales Monastici, v, 40, and index.

3. See John's charter of 3rd June, 1200, in Clark (iii, 177), and the extent of 1235 (iii, 381); also Rot. Litt. Pat., p. 19 (23rd October, 1202), and p. 68b (1207).

4. Clark, iii, 144, 217. In 1193 William attested a charter of John, then count of Mortain, at Cardiff (i, 33). An interesting charter of Robert, son of Wian, granted to the abbey a lease of land for six years from Michaelmas, 1197, "que videlicet festivitas Sancti Michaelis tercia secuta est captionem castelli de Sancto Claro factam per Willelmum de Brausa" (iii, 169).

5. Ann. Monast., i, 24.

the castle of Swansea, on several grounds, including the significant plea that William had extorted the original charter from John when the king was in a panic and feared that his companion was going to leave him.1 In short, the man who was most in John's confidence was William of Briouze, and if any chronicler was likely to hear about the death of Arthur and its consequences it was the chronicler of Margam.

There is another significant fact which, so far as I know, has never been noticed, but which adds an element of certainty to this view. It has often been observed that the Margam story only reappears in one place—and there with some variation in the epic, Philippid, of King Philip's chaplain William the Breton. The variations are not great, and show that the chaplain was giving the same story independently. Now it is very curious that he singles out William of Briouze, who is not mentioned elsewhere in the poem, as the spokesman of those barons who were with John near Rouen at the time of Arthur's death. John brought Arthur to Rouen (I summarise the flowery verses) and aroused the suspicions of the barons. William of Briouze declared that he would be responsible for him no longer, and that he handed him over safe and sound. After a moody seclusion at the royal manor of Moulineux, John did away with his nephew at Rouen by night.2 This comes in book vi., which with the beginning of book vii. has been shown with some probability to have been composed before 1214.3 Now part of this story at least had been well known or suspected at the French court for a long time. The abbot of Coggeshall tells us that Philip heard in 1204 that Arthur had been drowned and 1. P.R.O., K.R. Miscell. Books, vol. i, p. 478b, 8 Edw. I.; printed in Clark, iii, 532.

2. Philippid, vi, 470-564.

3. Delaborde, Notice sur Rigord et Guillaume le Breton (prefixed to his edition), pp. 70 seqq. The references to Arthur's death in William's Continuation of Rigord are brief and casual, though emphatic (ed. Delaborde, i, 253, 293).