This speedy erection, destruction, and re-erection is enough to prove that the castles of William in York were, like most other Norman castles, hills of earth with buildings and stockades of wood, especially as we find these hills of earth still remaining on the known sites of the castles. And we may be quite sure that the Norman masonry, which Mr Freeman pictures as so eagerly destroyed by the English, never existed.[797] But the obstinate tendency of the human mind to make things out older than they are has led to these earthen hills being assigned to Britons, Romans, Saxons, Danes, anybody rather than Normans. A single passage of William of Malmesbury, in which he refers to the _castrum_ which the Danes had built at York in the reign of Athelstan, is the sole vestige of basis for the theory that the motte of Clifford's Tower is of Danish origin.[798] The other theories have absolutely no foundation but conjecture. If Malmesbury was quoting from some older source which is now lost, it is extremely probable that the word _castrum_ which he copied, did not mean a castle in our sense of the word at all, but was a translation of the word _burh_, which almost certainly referred to a vallum or wall constructed round the Danish suburb outside the walls of York. Such a suburb there was, for there in 1055 stood the Danish church of St Olave, in which Earl Siward was buried, and the suburb was long known as the Earlsburgh or Earl's Burh, probably because it contained the residence of the Danish earls of Northumbria.[799] This suburb was not anywhere near Clifford's Tower, but in quite a different part of the city. To prove that both the mottes were on entirely new sites, we have the assurance of Domesday Book that out of the seven _shires_ or wards into which the city was divided, one was laid waste for the castles; so that there was clearly a great destruction of houses to make room for the new castles.[800] [Illustration: FIG. 39. York Castle and Baile Hill. (From a plan by P. Chassereau, 1750.)]