the magnificent ruins of Château-Gaillard would still afford sufficient proof that he concentrated all his energy, skill and experience upon the work. It would doubtless be rash to argue that all the characteristic details of the buildings on the Rock were first developed by Richard and were entirely due to his study of the Latin fortresses in Syria; but it is quite obvious that Château-Gaillard marks a turning point in the history of western fortification; and it is incredible that Richard did not profit by his experiences as a Crusader. Apart from the boldness of the work as a whole, the structure of the elliptical citadel with its series of curvilinear bosses,2 and of the circular keep with its wedge-like machicolation,3 reveals a profound practical knowledge of fortification. The scientific use of military engines and other methods of attack had been brought in this age to such perfection that the existence beneath the walls of a single 'dead angle,' or spot which could not be reached by missiles, might be the cause of disaster. The structure of citadel and keep, and to a less degree of the triangular advance work was designed to remove this defect, inevitable in a rectilinear fortress; and it was combined with a long sloping base

1. M. Dieulafoy, Le Château-Gaillard et l'architecture militaire au xiiie siècle in the mémoires of the Academie des inscriptions et belleslettres, 1898, vol. xxxvi, part i, pp. 325-386; and the remarks of Coutil, op. cit., pp. 68, 72-75.

2. This enclosing wall of the inner court is the distinctive feature of the castle. It is preceded,' in the words of M. Dieulafoy, 'by a fossé cut almost vertically out of the rock, and it consists of a chaplet of towers, or segments of circles on a chord of about three metres, which are united by strips of wall nearly a metre in width. Its tracé, elliptical in form, and its carefully conceived profils show profound knowledge. No angle mort, nor secteur privé de projectiles is to be found; the approaches and the fossé are covered by the fire of the garrison right up to the foot of the wall, and no sapper could touch any point in towers or walls, provided that the fortress was under the direction of an experienced commander' (p. 330).

(talus) from which the projectiles hurled from above would ricochet with increased force. It has been pointed out by M. Dieulafoy that Philip Augustus was finally successful in the siege of the castle because he availed himself of the single defect in its scheme of defences. This defect was the protection afforded to the besiegers by the stone bridge which connected the citadel or second court with the outer court.

No Syrian castle combined all the characteristics of Château-Gaillard. Some of them may be found in other western fortresses of this date; for example, in Pons, Etampes, Provins, Issoudun, La Roche Guyon, Ghent. Nor could anyone who has stood before his rival's work at Angers or Gisors claim for Richard a monopoly of knowledge; indeed by his patience and subtlety Philip was the better engineer of the two. Yet we must regard Richard's wars in Normandy as a continuation of his fights as a Crusader. He was the chief soldier of his age; we may be sure that during the few years of his reign the unity of the soldier's life was never broken, and that his best talk was heard in the company of Mercadier the Brabançon or Master Ivo the Balistarius. He brought back with him from the east men who had been trained in Syria. Franks born in Syria, one of them from Nazareth, were among his artillerymen; indeed, there is good evidence that he had brought back a band of Saracens to fight for him. We are told that Greek fire that terrible explosive was used during these wars, and if information

3

1. Ibid., 334.

2. "Petro de Tanentonne et Martino de Nazareth et sociis eorum Arbalistariis liiij li. per breve Regis" (Rot. Scacc., ii, 302). Cf. the reference to a Baldoin of Jerusalem (ibid, ii, 301).

3. I have discussed the evidence in the Scottish Historical Review for October, 1910, p. 104. The references are to Rot. Scacc., i, 221; ii, 301; compared with text D. of the Histoire d'Héracles in the Recueil des Historiens des Croisades, Historiens Occidentaux, ii, 196.

4. In 1195, when the French destroyed Dieppe (Howden, iii, 304).

were forthcoming the history of siege engines in these years would probably throw some light on the military results of the Third Crusade. We may think, then, that as he supervised the workmen at Andeli, Richard's thoughts often went back five or six years to the siege of Acre or to his visit to the great Syrian fortress of Margat.

III.

Although the story of warfare which it has been possible to compile in previous chapters is in the main a story of sieges, the evidence is much too scanty to enable any modern student to follow in detail the wars between Philip Augustus and the dukes of Normandy. Indeed, it is probable that only a handful of persons were kept aware of all the military operations: only now and then did definite facts become matter of common knowledge. We are able to see from the terms of treaties and from casual letters enrolled by officials that, all along the Norman frontier, there was constant building, attacking, and destruction of fortresses,1 of whose existence in most cases a piece of disordered ground, or perhaps a popular tradition, is the only record.2 Fortunately, however, we are better informed about the organisation of the castle and its importance in time of war.

In the first place, it should be noticed that a castle rarely stood in isolation, but was generally a part of a definite system of fortifications. Sometimes these arrangements were of a temporary nature, as in 1184, when the earl of Arundel commanded all the frontier castles from Vaudreuil to Gisors,3 but more permanent affiliations can be traced beneath these extraordinary commands. In a few cases, a fortress relied partly upon its own strength, partly upon its intimate relation to a

1. See, for example, the treaty of 1200, quoted above p. 253. 2. Bonnard, op. cit.. passim.

3. Above p. 278.

general system of communications. Thus Verneuil, with its triple town, earthworks and artificial ponds, was not only strong in itself; it had become the centre or objective towards which the energies of central Normandy as a whole were directed for the defence of the Avre. Behind it lay L'Aigle and Breteuil, upon the latter of which converged the road from Rouen by Neubourg and Conches and the important strategic route from Lisieux by Chambrais (the present Broglie) and Lire; on either side lay the less important defences of the Avre,-Nonancourt, Tillières and Courteilles to the east, Saint-Christophe and Chennebrun to the west. The accounts of the bailiff of Verneuil, who was also the farmer, in 1198, give a picture of its importance: there was frequent intercourse between Verneuil and the fortresses in the valley of the Seine;2 the bailiff's expenses show that he supervised the administration of Tillières, and the defences of Damville, Courteilles,3 and Cintray (to the north-west). Verneuil itself was the barracks of a little host of artillerymen and other mercenaries, one or two of whom bear strange-sounding names, and of mounted men-at-arms and foot soldiers.

Such great centres as Arques, Gisors, Vaudreuil, Falaise, and Argentan were, like Verneuil, places of national importance, intimately dependent upon the administration. as a whole. Where this was not the case, a system of local grouping can generally be traced, such as is familiar in feudal history and in feudal literature. This grouping is naturally more marked in private honours, which, being on the whole more compact in Normandy than they were

1. King Richard went by this road in 1194, and King John in 1203. See above pp. 151, 243.

2. "Pro prisonibus captis in Gerra de Rothomago apud Vernolium [ducendis] et hantis et picoisis et venatione Regis et hernesio balistariorum pluries ducendis a Vernolio et Aquila apud Vallem Rodolii et Insulam de Andele." (Rot. Scacc., ii, 311-2).

3. "Pro claudendo bailio de Corteilles de petra, ccc. li." (Ibid., p. 315).

in England, could be organised as military units: the fifteen castles of the Widow Lady in the romance of the Holy Graal, or the more historical nine held by Robert of Bellême in Maine, find their counterparts in the castellariae of the honours of Mortain and in France of Montfortl'Amaury. On the ducal demesne, however, the administrative system often comprised a similar organisation of strongholds. Some of the groups were originally held by vassals or neighbours; others were due to official action; some again were more of the nature of federations, such as that which comprised the four castles of Lions, Neufmarché, Longchamp, Beauvoir,3 or the union of Moulins and Bonmoulins; others consisted of a castle with subsidiary forts, such as Falaise and Pommeraye, Vaudreuil and Louviers.5 The erection of Château-Gaillard brought with it the creation of a similar group of subordinate forts, Tosny, Boutavant, and Cléry.

We get a glimpse of the relations which were customary between the various castellariae of an honour, in an inquiry instituted by Philip Augustus into the rights of Evreux and Gaillon. 6 Until the occupation of Gaillon by the French king in 1193 or 1194,7 it was part of the honour of the count of Evreux. The chief evidence was given by Geoffrey Barket, who had been the count's castellan when Philip besieged the castle. He pointed out the fiefs and woods which did not owe service at Gaillon, distinguished

1. Mortain, Cerences, Tinchebrai, Condé-sur-Noireau.

2. Gressey, Haye-de-Herce, Richebourg, Gambais, Houdan, formed part of the châtellenie of Montfort l'Amaury south of the Avre; Dion, Etude sur les châteaux féodaux des frontières de la Normandie, pp. 363-5.

3. Stapleton, I, cxiii.

4. Above p. 109.

5. Above p. 161.

6. Cartulaire Normand, pp. 21-2; no. 120.

7. More probably in February, 1194. See the account of Philip's movements above, p. 147. For Richard's attempt to retake Gaillon, see p. 169 note.