The

not surrendered till June of the following year. eastern boundary of Caux, therefore, was at this time represented by a line drawn from Dieppe through Arques and Bellencombre to Rouen. Beyond this line, except in the valley of the Seine, Philip was supreme. Drincourt, the great castle of Bray, otherwise Neufchatel-en-Bray, was his, also Radepont and the Andelle valley and the whole of the manor of Andeli outside the walls of Château Gaillard. During the siege of Château Gaillard the king distributed his favours in these districts. Nicholas of Montigni, a Norman, received lands near Drincourt.2 Peter of Moret, a Frenchman, was entrusted with Radepont.3

It is more difficult to draw the line of division between the French and Norman governments in the district lying between the Seine and the Risle. This bailiwick was farmed in 1203, and its accounts were presented at the exchequer at Michaelmas. The fortified ducal manor at Moulineaux on the Seine, and Montfort and Pont Audemer on the Risle, were certainly under John's control; but there is some evidence that Philip had begun to push forward from Vaudreuil and to drive a wedge between Neubourg and Pont de l'Arche into this important country.4 All the lands south of a line drawn from

1. Cart. Norm., no. 74, p. 14. These places belonged to William the Marshal, and were dealt with in his agreement with King Philip at Lisieux in May, 1204. The Marshal promised to hand them over immediately to Osbert of Rouvrai, who in turn would undertake to hand them over to Philip on June 24th.

2. Cart. Norm., no. 68, p. 293; Actes, no. 797, p. 181; at Anet, November, 1203. Later gifts in 1206-7, Actes, no. 984, p. 226.

3. Cart. Norm., no. 184, p. 297; dated by Delisle, October, 1203, Actes, no. 790, p. 179.

4. The value of this evidence turns on the identification of Landa, granted by Philip in October, 1203, to Raoul de Louvain, with La Londe, north of Pont de l'Arche (Actes, no. 786, p. 178: and Index, p. 609). Le Prévost identifies it with lands at Canappeville (Cart. Norm., no. 252, p. 301).

Montfort to Pont de l'Arche were, with the exception of Neubourg, lost to the duchy. Beaumont and Conches had fallen in the previous spring. After the fall of Château Gaillard, if not before, Neubourg also must have surrendered. A brief study of the map will show the reader that when Philip advanced in May 1204, east and west Normandy were practically cut off from each other.

West of the Risle, however, the valleys and highlands of western Normandy were as yet hardly touched by the French. It is true that in the south the continuity of the March from Alençon to Verneuil must have been broken, for nothing is heard of Moulins and Bonmoulins, of l'Aigle or Breteuil after John's departure for England. The reason is obvious. By means of his conquest in the Evrecin and the valley of the Risle, Philip could attack the fortresses on the Itun and in the forest country from two sides. Only Verneuil stood out, self-supported and for the time impregnable. Beyond the March, on the other hand, Philip was able to do little until he could advance in full force from the Seine. Before he left John had, according to the Marshal's biographer, strengthened the fortifications of the Touque valley in the north-east at Bonneville and Trianon.2 To the south were Lisieux and Falaise. Argentan and Domfront were sufficient to keep back the barons of Maine.

In the west a little group of strongholds, Avranches, Mont-Saint-Michel, Pontorson, Saint-James, Vire and Mortain protected the Breton frontier. Within this strong cordon stretching from Bonneville to Pontorson, the administration may reasonably have been expected to hold out, even if Rouen and the Seine were lost. It still would control Caen, the centre of government, Barfleur and Cherbourg, its links with England, and the wealthy domains of the Côtentin and the Vau de Vire, as yet hardly touched by war.

1. i.e., on the north from Conches and Breteuil; on the south from Dreux and through Maine.

2. Guill. le Maréchal, iii, 174.

John and his advisers in fact seem to have anticipated that concentration in the centre and west might be necessary. His fortification of the valley of the Touque was criticised in military circles on the ground that it was unwise to prefer the Touque to the Risle as a line of defence. Although this criticism, as it has come down to us, is faulty and ill-informed,1-John seems, for example, to have paid special attention to Pont-Audemer on the Risle 2-it would seem to be a survival of some discussion upon the most suitable system of defences in the case of defeat in the east. This view is strengthened by the statement of William the Breton, that, after the fall of Château Gaillard, Pont de l'Arche and Moulineaux on the Seine, and Montfort on the Risle were destroyed by the Normans.3 In the west the year 1203 had been spent in careful inspection, restoration and garrisoning of the fortresses. The bailiff Richard of Fontenai had expended large sums, largely consisting of escheats and the proceeds of a tallage on the Jews, in the payment of knights and men-at-arms, and especially in elaborate additions to the fortifications of Mortain. One of John's last acts, before he sailed, was to give to this great official the control of Mont-Saint-Michel."

All hopes of a prolonged defence proved to be vain. When King Philip advanced after his triumph at Château Gaillard, Normandy crumbled away before him. Perhaps the seneschal and his colleagues had not believed that the great castle could ever fall. They may not have anticipated the irrepressible rush of Bretons, or the treachery of Louvrecaire at Falaise, or the lack of English aid. They were surrounded on all sides, and those who did not

1. Ibid. It is faulty because the writer thinks that John should have fortified Montfort, Beaumont and Brionne, whereas the two last-named places were already in Philip's hand. Above, p. 238.

2. Rot. Norm., 116.

3. Philippid., 1. vii, vv. 826-829 (ed. Delaborde, ii, 208).

4. Rot. Norm., 120-1; Rot. Scacc., ii, 546-8.

5. Rot. Scacc., ii, 548.

6. Rot. Norm., 117.

go over to the enemy took refuge in Rouen, or escaped to England.

The famous siege of Château Gaillard lasted for six months from the end of September or the beginning of October 1203 until the surrender of Roger de Laci on March 6, 1204.1 The detailed narrative devoted to the siege by William the Breton has given to it an importance in history which has been denied to other incidents of the war. The preference of history is, perhaps, not quite just; it is possible that the investment of Rouen or the defence of Loches might have taken as permanent a place in our annals if the chaplain of Philip Augustus had had an equal interest in describing them.2 Yet it cannot be denied that the operations against Château Gaillard were conducted on a scale which was especially elaborate and required all the resource of Philip's ingenuity. Antiquaries profess their ability to this day to trace the course of his lines.3

Except from the Philippid and the prose additions made by William the Breton to Rigord's chronicle, we know nothing of the inner history of the defence. The English records contain a few references to the captives; but, so far as is known, the constable of Chester had with him no literary companion, and no stories of the siege have strayed

1. William the Breton's chronicle and Philippid, (ed. Delaborde, i, 212-220; ii, 176-209). The narratives of Deville and Miss Norgate are based upon William. See also Viollet-le-Duc, An Essay on the Military Architecture of the Middle Ages, Eng. tr. (1860), pp. 90-94; Dieulafoy, in Mém. de l'Institut: Academie des Inscriptions et belles lettres, xxxvi, part i, pp. 373–378; Coutil, op. cit. pp. 84-93.

2. William the Breton hints as much in his reference to Loches and Chinon (Delaborde, ii, 225, 226).

3. See the plan in Viollet-le-Duc and Coutil. On the other hand, it must be remembered that Charles VII also invested the castle in 1449 and drew lines of circumvallation.

4. e.g., Rot. de Lib., 103, loan by the king for the ransom of the constable of Chester. There is also a notice of privileges granted to a Jew who had lent money to the constable at Château-Gaillard; Rot. Pat., 47.

into the English and Norman chronicles of the time. The garrison was, like the garrison of Vaudreuil, controlled by an Anglo-Norman baron, and was largely composed of Englishmen.1 It would be interesting to know how they behaved in a stronghold so much more complicated in its structure than any castle of Kent or Yorkshire; only a description by one who had seen from within the fall of the outwork, of the outer bailey, and, finally, of the "citadel" could solve the problems which are still unsolved. Was the castle too restricted in its dimensions? were its elaborate arrangements mutually injurious ?2 Or did the defence of the apparently impregnable citadel, with its embossed curtain wall and the scientific angles of its keep, require a knowledge which was not possessed by the constable and his followers? 3

King Philip's attack on Andeli had begun in August, when he occupied the peninsula formed by the great bend of the Seine. The defenders of the new town and of the isle withdrew from their outworks in this plain 4 and broke down the bridge between them and it. It is supposed by modern scholars that Philip, not satisfied with the protection afforded to his operations by the possession of Vaudreuil and Gaillon, ordered a ditch and rampart to be made across the peninsula from Bernières to Tosny. The

5

1. The archbishop of Canterbury was granted one of the prisoners captured at Mirebeau towards the redemption of his knights who had been captured at Château-Gaillard; Rot. Pat., 40b; April 12, 1204. 2. Viollet-le-Duc, p. 94.

3. Dieulafoy, p. 376. Dieulafoy points out that the castle contained at least one defect in the stone bridge which crossed the ditch between the outer bailey and the citadel. This bridge was fixed, and offered some protection to the engineers during the attack.

4. Above, p. 286.

5. This is Viollet-le-Duc's view, based on existing earthworks. But there is no reference to the ditch and rampart in William the Breton's description of the siege, and as the night attack by John's mercenaries was only stopped by the walls of the camp, their existence cannot be regarded as proved. Cf. above, p. 374 note.