That there existed in England, at the Conquest, no castles in masonry of English work it may be too much to assert; but it may safely be said that, save a fragment of wall at Corfe, no military masonry decidedly older than that event has as yet been discovered. In 1052, when the Confessor and Earl Godwin came to terms, and the attack on London was set aside, it is stated that Archbishop Robert and his Frenchmen fled, some westward to Pentecost Castle and some northwards to Robert’s Castle; so that these places probably, like Richard’s Castle, were in Norman hands, though it does not follow that they were constructed of the material or in the fashion then coming into use in Normandy. “Domesday” mentions directly forty-nine castles as existing at the date of the survey, and of these at least thirty-three were on sites far older than the Conquest; and of them at least twenty-eight possessed artificial mounds similar to Arundel and the castles in Normandy. “Domesday,” however, is notoriously capricious both in its entries and omissions on such matters as were not included in its proper view, and its list of castles is nearly as incomplete as its list of churches. Neither were required to be noted. “Of the forty-nine castles recorded,” says Sir H. Ellis, “eight are known, either on the authority of ‘Domesday’ or of our old historians, to have been built by the Conqueror himself; ten are entered as erected by greater barons, and one by an under-tenant of Earl Roger; eleven more, of whose builders we have no particular account, are noticed in the ‘Survey,’ either expressly or by inference, as new.” The fact is, however, that although the number of castles actually mentioned may be only forty-nine, of castles and castelries (which imply a castle) there are named in “Domesday” fifty-two. The castles reputed to have been built by the Conqueror himself are Lincoln, Rockingham, Wareham, two castles at York, Dover, Durham, London, and Nottingham, of which the last four are not mentioned in “Domesday.” Exeter, also omitted, is generally reputed to be one of William’s castles, as was Stafford; which, however, was constructed and destroyed before the date of the survey. “Terra de Stadford in qua rex percepit fieri castellum, quod modo est destructum,” a very short period for the construction and destruction of a work in masonry. Mr. Pearson, who has given great attention to the subject of Norman castles in England, tabulates the result of his researches in the atlas attached to his “History.” He there enumerates, as standing in the reign of the Conqueror, forty-nine castles belonging to the King and fifty to his subjects. Of these, at least thirty-eight have mounds. He gives also a list of fifty-three belonging to private persons in the reign of William Rufus, of which at least five have mounds. Probably there were of each class many more than these. Colchester, for example, is not included, nor Farnham, nor Berkhampstede.