It is well known that between the years 1087 and 1089 the celebrated architect, Gundulf, Bishop of Rochester, built a new _stone_ castle for William Rufus, "in the best part of the city of Rochester."[617] This castle, of course, was on the same site as the present one, though the splendid keep was not built till the next reign.[618] But if what we have maintained above be correct the castle of Gundulf was built on a different site from that of the castle of William. Nor are we without evidence in support of this. What remains of the original Norman wall of Gundulf's castle (and enough remains to show that the circuit was complete in Norman times) does not stand on earthen banks; and this, though not a proof, is a strong suggestion that there was no earthen bank belonging to some previous castle when Gundulf began his building.[619] But further, Mr Livett has shown in his paper on _Mediæval Rochester_[620] that in order to form a level plateau for the court of the castle the ground had to be artificially made up on the north and east sides, and in these places the wall rests on a foundation of gravel, which has been forcibly rammed to make it solid, and which goes through the artificial soil to the natural chalk below. Now what can this rammed gravel mean but an expedient to avoid the danger of building in stone on freshly heaped soil? Had the artificial platform been in existence ever since the Conquest, it would have been solid enough to build upon without this expense. It is therefore at least probable that Bishop Gundulf's castle was built on an entirely new site. It seems also to be clear that the Boley Hill was included as an outwork in Bishop Gundulf's plan, for the castle ditch is cut through the Roman wall near the south gate of the city.[621] Mr Livett remarks that King John appears to have used the hill as a point of vantage when he attacked the city in 1215, and he thinks this was probably the reason why Henry III.'s engineers enclosed it with a stone wall when they restored the walls of the city.[622] Henry III.'s wall has been traced all round the city, and at the second south gate it turns at right angles, or nearly so, so as to enclose Boley Hill.[623] It is probable, as Mr Livett suggests, that the drawbridge and _bretasche_, or wooden tower, ordered in 1226 for the southern side of Rochester Castle,[624] were intended to connect the Boley Hill court with the main castle. In 1722 the owner of the castle (which had then fallen into private hands) conveyed to one Philip Brooke, "that part of the castle ditch and ground, as it then lay unenclosed, on Bully Hill, being the whole breadth of the hill and ditch without the walls of the castle, extending from thence to the river Medway."[625]