objection disappears. Of the four cases named by Mr. Clark in which it remains perfect, Dover and Newcastle are, it is now admitted, 'the work of Henry II.,' and therefore late Norman, while the other two are Norwich and Castle Rising. Norwich, like Newcastle and Dover, is 'more ornate than usual,' while Castle Rising, which has much in common with it, is the most highly-ornamented keep in England.' If it was built, Mr. Clark wrote, 'by the architect of Norwich keep, it must be early Norman, for Norwich was besieged, 1076, at the revolt of Earl Guader, and Harrod supposes the present keep to have been then standing; but the ornamentation of Castle Rising looks much later.' Here again, as at Arques and Newcastle, the word 'castrum has misled our author against his own judgment, and the inference he ought to have drawn was that Norwich must be later. An admirable paper on this castle was read before the Norwich Archæological Congress of 1889 by Mr. Hartshorne, who showed that its original form was that of a typical burh, and that the existing rectangular keep must, from its style, be dated 1120-1140. He claimed for it a marked resemblance to the keeps of Hedingham, Rochester, and Castle Rising; and we may note that all four were built for private lords. It is by no means clear when these noble 'towers' began to be raised in England. Mr. Clark holds that they date from '1078, when the White Tower was begun' (i. 138). But this too lightly accepted date is mere conjecture. We know, from the singular Rochester story, that Bishop Gundulf was skilled in castle-building, and we also have a curious reference to his staying with a citizen of London, when 'ex præcepto Willelmi Magni præesset operi magnæ Turris Londoniæ.' This incident must have happened between 1077 and 1087: more than this we cannot say. But if even London had to wait for its Tower till the latter part of the Conqueror's reign, it would seem improbable that provincial towns, and still less rural lords, should have had towers before it. We, therefore, lean to Mr. Clark's view that these towers were not introduced till 1078 (or even later), and that timber defences, on moated mounds, were alone available till then. Only we hold, as he does not, that, where a castle was needed, and there was no mound, the new settlers made one. When the Tower' had once been introduced, its superiority over the timbered mound must have been so great that it would always be employed where time and means permitted. But, doubtless, where the latter was already in existence, the fortress, as Mr. Clark points out, would remain a 'shell-keep.' In any case, we read : The 'The rectangular and the shell keep never occur in the same castle; and, as a rule, where there is a mound, there is no rectangular keep; the only known exceptions to this rule are at Christchurch, Guildford, Clun, Saffron Walden, Mileham, Bungay, and Bramber.' As to the placing of a rectangular keep on an artificial mound, Mr. Clark tells us (ii. 62) that 'Guildford, Christchurch, and Clun are the only recorded examples'; and of these Christchurch alone stands fairly on its summit. But the mound on which stands Norwich keep is, we believe, mainly artificial, and should, therefore, be added to the list. Bramber does not belong to the class of towers built on a mound: it has a mound and a tower standing apart in one area. We should, therefore, be inclined to compare it with Oxford, as a case in which the mound was abandoned for a tower. Mr. Clark insists that the tower at Oxford was 'not the keep'; but it was certainly the 'tur' from which the Empress Maud was let down, we read, by ropes on the night of her marvellous escape. Nor can we pass over 'the once celebrated and very strong castle of Bedford' (i. 217). It is true that the author of the Gesta Stephani'-an excellent observer in such matters-describes it as 'editissimo aggere vallatum,' but Mr. Clark has omitted to mention that he speaks of an inquassabilis turris,' which was, we believe, the old tower' of the great siege under Henry III. (1224); the tower which was mined and split asunder, and from the summit of which the besieged, in their despair, displayed at length the king's standard. After that siege the earthworks were ordered to be levelled with the ground, and three quarters of the 'old tower' to be taken down. It is then clear, in spite of Mr. Clark, that Bedford had a tower apart from the mound. Lincoln, again, had a rectangular tower about 25 feet by 40 feet' apart from the shell-keep on the great mound: it was as large as the keeps of Clitheroe or Malling, and therefore entitled to rank with them; and it has a special interest as being very probably the tower, though Mr. Clark denies it, that the mysterious Countess Lucy is known to have built. Still, in spite of sundry sports,' the general conclusion holds good that the rectangular and the shell-keep are not found in combination. We do not propose to treat at length of Mr. Clark's studies on castles of post-Norman date, because his work for the earlier period is more original and striking. But we must not pass over his valuable chapters on castles of the Early English period and those of the Edwardian or concentric type. The first great innovation was the use of the cylindrical in place of the rectangular tower, corresponding, as he observes, with the middle period of the Early English style in ecclesiastical archi tecture. tecture. Of this the best known types are Conisborough-the source of as wild speculations as the round towers of Ireland— and abroad, the celebrated Coucy, the seat of the proudest of barons. Of the quatrefoil tower, the keep at York, assigned by Mr. Clark to Richard I. or John, seems to be the only surviving example. That of Warwick is said, we know not on what authority, to have once resembled it. So energetically had castle-building proceeded throughout the twelfth century that few fresh keeps were erected in this Transition period, and attention was given rather to the strengthening of the surrounding defences. The cylindrical type was gradually applied to the mural towers of the curtains, new wards were added, and strong gatehouses erected. Thus arose the 'concentric' fortress, of which, at Dover and the Tower of London, we can trace the gradual growth. As the enceinte increased, the keep dwindled, in importance, till at the accession of Edward I. the time had come for discarding the latter altogether. It was at this date that Caerphilly, 'the earliest and the most complete example in Britain of a concentric castle,' was built by the head of the mighty house of Clare. The author's skilful monograph on this fortress was written so far back as 1834, and its admirable illustrations, here as elsewhere, enable us to follow his description. As 'consummate' as Château Gaillard, Caerphilly deserves attentive study, for it formed the style of those Welsh castles, so familiar a feature to the tourist, which Edward, building on new sites, erected in the Principality. They were adapted to a system of defence different from that of the Norman fortress, because relying less on mere passive resistance, and more on active opposition. The assailants had now to fight their way from ward to ward, exposed, at every stage, to the fire of the besieged. The greater space secured by the concentric system allowed also of the erection of a residence more convenient and elaborate than the donjon keep. In England, however, we can only study the adaptation of this late style to castles already existing. With military as with ecclesiastical architecture, the perfection of development was but the prelude to degeneration and decay. The demand for comfort, the elaboration of ornament, the increase of order and security, and lastly the introduction of artillery, were all unfavourable to the true castle, which became less of a fortress as it grew more of a dwelling-house. Those in the hands of the Crown began early to be neglected; and the Wars of the Roses practically closed the era of their use by private lords, though they lingered on, in slow decay, to form centres of makeshift resistance in the confusion of the Civil Wars. We We would rather specially address ourselves to the problem of the introduction of castles into England. Mr. Freeman, of course, could not neglect a subject so closely connected with the Norman Conquest: with characteristic confidence he faced and settled the question. It was in or about 1050 that was built the first castle we hear of in England: * The fortress itself has vanished, but its site is still to be marked, and the name of Richard's castle, still borne by the parish in which it stood, is an abiding witness of the deep impression which its erection made on the minds of the men of those times.'† We must quote in extenso the passage which follows, as it well sets forth Mr. Freeman's views: 'Both the name and the thing were new. To fortify a town, to build a citadel to protect a town, were processes with which England had long been familiar. . . . But for a private landowner to raise a private fortress to be the terror of his neighbours was something to which Englishmen had hitherto been unaccustomed, and for such a structure the English language had hitherto contained no name. But now the tall, square, massive donjon of the Normans, a class of buildings whose grandest type is to be seen in the Conqueror's own Tower of London and in the more enriched keep of Rochester, began, doubtless on a far humbler scale, to rear itself over the dwellings of Englishmen. . . Such strongholds, strange to English eyes, bore no English name, but retained their French designation of castles.' + We see how strongly Mr. Freeman held that the 'castle,' introduced under Edward, was something wholly new, that it was constructed of stone, and that it was a 'square donjon.' So too, the castle of Hereford, destroyed in 1055, was built, he deemed, ' according to the latest continental patterns' he even pictured the square mass of the Norman donjon, an ominous foreboding of the days which were soon to come.' § Now he was not ignorant, when he thus wrote, of Mr. Clark's theories, for he mentioned our author's conclusions that the agger or mote was commonly an earlier earthwork made use of by the builders of the eleventh and twelfth centuries,' and that its fortifications were commonly of wood,' only to dismiss them. One point, however, remains: we may certainly assert of Richard's castle, and virtually of the mota of Hereford, that neither of them was ever crowned by a square donjon.' They were both shell-keeps. Right or wrong, the Professor was at *Norman Conquest,' vol. v. p. 649. least consistent in his † Vol. ii. p. 137. view: we cannot say as much of Mr. Clark. In his introductory chapters, he writes of shell-keeps that a tolerably close examination has failed to discover, either at Arundel or elsewhere in England or in Normandy, any masonry of very early character, probably none that can be safely attributed to the eleventh century. The fact seems to be that the early timber structures, which are known to have been erected originally on the moated mounds, were found to be very defensible, and so were retained by the Norman lords until they were able to replace the timber by masonry *. . . . They seem to have been content to repair the existing works, usually of timber only, and to have postponed the replacing of them with a regular shell till a more convenient season, which in many cases did not occur for a century.'† 6 6 In this important conclusion we concur, after examining the evidence. But, in his detailed account of Richard's castle, Mr. Clark tells us that the English, under Edward the Confessor, could not but have regarded with dismay the lofty walls and towers which made impregnable a place already strong' (ii. 402). In the next paragraph, however, he admits that the masonry, of which vast fragments still remain, is apparently of rather a later date'; while, in one of his introductory chapters, he pronounces it improbable that the keep was constructed before the reign of Stephen, if so early' (i. 102). Now if, as Mr. Clark maintains, Richard fitz Scrob' did but occupy an existing earthwork (ii. 401), and if, in accordance with his theory, he fortified it in the English manner, where was the novelty that so excited the terror of the people? 6 The fact is that here, as elsewhere, the writer was under Mr. Freeman's spell. His remark that the name of Richard's Castle shows how deeply the fear of its builders was impressed upon the people' (ii. 110), does but echo the words of the historian. For our part, the name of Richard's Castle is no more eloquent of terror than that of Bishop's Castle—' the residence of the Bishop of Hereford' (i. 102)—in the adjoining county of Salop. Indeed, though Mr. Clark follows the late Professor throughout, in dealing with Richard and his castle, we believe that the whole story was a singular delusion of Mr. Freeman, and that Richard's castle is not even mentioned before the Norman Conquest. Adopting Mr. Clark's general conclusion on the subject of shell-keeps, we hold that a change in fortification did not take place till after the Conquest, when the characteristic rectangular keep was first imported from Normandy. On the other * i. 139. ti. 42. Cf. p. 47. hand, |