There are certain facts, which have not been put together before, which enable us to make a very probable guess as to the date at which this ancient castle was cut through by the newer city bank. The walls of Canterbury have never yet received so careful an examination as those of Rochester have had from the Rev. Greville Livett;[328] but the researches of Mr Pilbrow about thirty years ago showed that the original Roman walls included a very small area, which would leave both the motte and the Plantagenet castle outside.[329] Certain entries in the _Close Rolls_ show that the fortification of the town of Canterbury was going on in the years 1215-1225.[330] But it is too often forgotten that where a wall stands on an earthen bank it is a clear proof that before the wall was built there was a wooden stockade in its place. Now the portion of the city wall which encloses the Dane John stands on an earthen bank; so, indeed, does the whole wall from the Northgate to the castle. It is clear that this piece of bank cannot have been made till the first Norman castle, represented by the earthwork, was abandoned; and fortunately we have some evidence which suggests a date for the change. In the _Pipe Rolls_ of Henry II.'s reign there are yearly entries, beginning in 1168, of 5s. paid to Adeliza Fitzsimon "for the exchange of her land which is in the castle of Canterbury." There can be little doubt that this land was purchased to build the great Plantagenet castle whose splendid keep was once one of the finest in England.[331] The portion of the castle wall which can still be seen does not stand on an earthen bank, an indication (though not a proof) that the castle was on a new site. Henry II. was a great builder of stone keeps, but he seldom placed them on artificial mottes. It is no uncommon thing to find an old motte-and-bailey castle abandoned for a better or larger site close at hand.[332]