But apart from details, if we look at the general plan of these four early stone castles, we shall see that it is exactly similar. It is the keep-and-bailey plan, the plan which prevailed from the 10th to the 13th century, and was not even superseded by the introduction of the keepless castle in the latter century.[1132] The motte-and-bailey type was of course only another version of the keep-and-bailey. In this primitive type of castle the all-important thing was the keep or donjon.[1133] Besides the donjon there was little else but a rampart and ditch. “Until the middle of the 12th century, and in the simpler examples of the epochs which followed, the donjon may be said to constitute in itself the whole castle.”[1134] Piper states that up to the time of the Crusades German castles do not seem to have been furnished with mural towers.[1135] Köhler, whose work treats of French and English castles as well as German, says that mural towers did not become general till the second half of the 12th century.[1136] Nevertheless, as it is highly probable that the baileys of castles were defended at first with only wooden ramparts on earthen banks, even when the donjon was of stone, it is not unlikely that mural towers of wood may have existed at an earlier period than these writers suppose. It is, however, in favour of the general absence of mural towers that the word turris, even in 12th-century records, invariably means the keep, as though no other towers existed.[1137]
That the baileys of some of the most important castles in England had only these wooden and earthen defences, even as late as the 13th century, can be amply proved from the Close Rolls.[1138] Colchester Castle had only a timber wall on the banks of its bailey as late as 1215, and in 1219 this palicium was blown down and an order issued for its reconstruction.[1139]
The arrangements in the stone donjons were probably the same as those we have already described when writing of the wooden ones.[1140] The basement was the storehouse for provisions,[1141] the first floor was generally the guardhouse, the second the habitation, of the lord and lady. Where there were three or four storeys, the arrangements varied, and the finest rooms are often found on the third floor. An oratory was probably an invariable feature, though it cannot always be detected in ruined keeps. One of Mr Clark’s most pronounced mistakes was his idea that these keeps were merely towers of refuge used only in time of war.[1142] History abounds with evidence that they were the permanent residences of the nobles of the 11th and 12th centuries. The cooking, as a rule, was carried on in a separate building, of which there are remains in some places.[1143]
Occasionally we find a variant of the keep-and-bailey type, which we may call the gatehouse keep. The most remarkable instance of this kind in England is Exeter, which appears never to have had any keep but the primitive gatehouse, undoubtedly the work of Baldwin de Moeles, the first builder of the castle. In Normandy, De Caumont gives several instances of gatehouse keeps. Plessis-Grimoult (which has been visited by the writer) has a fragment of a gatehouse tower, but has also a mural tower on the line of the walls; as the castle was ruined and abandoned in 1047, these remains must be of early date.[1144] The gatehouse keep is probably an economical device for combining a citadel with the defence of the weakest part of the castle.
We must pass on to the keeps of Henry I.[1145] There is only one in England which authentic history gives to his time, that of Rochester.[1146] But the chronicler Robert de Torigny[1147] has fortunately given us a list of the keeps and castles built by Henry in Normandy, and though many of these are now destroyed, and others in ruins, a certain number are left, which, taken along with Rochester, may give us an idea of the type of keep built in Henry I.’s time. The keeps attributed by Robert to Henry I. are Arques, Gisors, Falaise, Argentan, Exmes, Domfront, Ambrières, Vire, Waure, Vernon, Evreux, Alençon, St Jean, and Coutances. How many of these survive we cannot positively say;[1148] we can only speak of those we have seen (Falaise, Domfront, and Gisors),[1149] and of Arques, described by M. Deville in his Histoire du Chateau d’Arques, by M. Viollet le Duc in his treatise on Donjons,[1150] and by Mr G. T. Clark.[1151]
Speaking under correction, as a prolonged study of the keeps in Normandy was impossible to the writer, we should say that there is no very striking difference to be observed between the keeps of Henry I. and those built by his father. The development of the forebuilding seems to be the most important change, if indeed we are justified in assuming that the 11th-century keeps never had it; its remains can be seen at Arques, Falaise,[1152] Domfront, and Rochester. At Arques and Falaise the doorway is on the second floor, which is an innovation, a new attempt to solve the difficulty of defending the entrance. The first floor at Arques could only be entered by a trap from the second floor; at Falaise there is a stone stair from one to the other. Rochester is entered from the first floor. The basement storeys of Arques, Falaise, and Domfront are quite unlit; at the Tower the basement has had a number of loopholes, and the angular heads of those which remain suggest that they are at least copied from original lights. The main floors in Henry I.’s keeps are always of wood, but this was not because vaulting was then unknown, because the crypt, sub-crypt, and chapel of the Tower are vaulted, not to speak of many early churches.[1153] The four keeps mentioned have all three storeys, thus not exceeding Colchester in height;[1154] the Tower has now four storeys, but a good authority has remarked that the fourth storey has not improbably been made by dividing the third.
No marked advance is observable in the masonry of these keeps. Arques is built of petit appareil; Falaise of small stones in herring-bone work; Domfront of very small stones rudely coursed; Rochester of Kentish rag mixed with flint rubble. Both Falaise and Domfront have plinths of superior masonry, but there is always the possibility that these plinths are later additions. The voussoirs of the arches at Falaise, Domfront, and Rochester are larger than the rag or tile voussoirs which are used at Colchester, the Tower, and Langeais. At Rochester and Arques provision is made for carrying the water-supply from the well in the basement to the upper floors, a provision of which there is no trace in the older keeps.[1155]
As Robert de Monte says that Henry I. built many castles in England as well as in Normandy, we naturally ask what other English keeps besides Rochester may be assigned to him. It appears to the writer that Corfe and Norwich keeps may very likely be his. Both were royal castles in his time, and both were originally wooden castles on mottes.[1156] Both these castles have forebuildings, and neither of them have floors supported on vaults.[1157] Corfe has very superior masonry, of larger stones than those used in the keeps known to be Henry I.’s, but wide-jointed. At Norwich only a very small piece of the original ashlar is left. Corfe is extremely severe in all its details, but quite corresponds to work of Henry I.’s reign.[1158] Norwich has a great deal of decoration, more advanced in style than that to be seen at Falaise, but still consistent with the first half of the 12th century. Neither keep has the least sign of Transition Norman, such as we seldom fail to find in the keeps of Henry II. Moreover, neither of them figure in the Pipe Rolls of Henry II., except for repairs; and as Stephen in his harassed reign can hardly have had any money for building stone keeps, we may with some confidence ascribe these two keeps to Henry I.
A few words should be given to the castle of Gisors, which contains in itself an epitome of castle history. The first castle, built by William Rufus in 1096, was undoubtedly a wooden castle on a motte, with a stockaded bailey below it; certain portions of the present bailey walls rest on earthen banks, which probably belonged to the original castle, and show what a much smaller affair it was than the present one. Henry I., Robert de Monte tells us, strengthened this castle with a keep. Probably this was the shell wall which now crowns the motte; the smallness of the masonry (stones about 5 inches high, rudely dressed and coursed) and the slight projection of the buttresses (9 inches) agree with much of the work of his time. There would be a wooden tower inside.[1159] The chemise or shell wall is pierced by loopholes, a very unusual arrangement; they are round arched, and of very rude voussoirs.[1160] Inside this shell there is a decagonal tower, called the Tower of Thomas à Becket, which is almost certainly the work of Henry II.,[1161] as its name would indicate; the chapel of St Thomas is close to it. A stair turret of the 15th century has been added to this keep; its original entrance was, as usual, a door on the first floor, but a basement entrance was built afterwards, probably in the 13th century. Philip Augustus, after he had taken this castle from John, added to it one of the round keeps which had then become the fashion, and subsequent enlargements of the bailey converted it into a “concentric” castle, of which the motte now forms the centre.
There is one keep which is known to be of the reign of Stephen, though not built by him, that of Carlisle, built by David, King of Scotland, in 1136,[1162] a time when he thought his hold on the four northern counties of England was secure, little reckoning on the true character of his great-nephew, Henry, son of Matilda. There is no advance to be seen in this keep on those of Henry I., except that the walls are faced with ashlar. The vaulting of the basement is pronounced by Mr Clark to be very evidently a late insertion.[1163]