In considering the limited and very inconvenient accommodation afforded by a Norman keep, it should be remembered that it was not meant for a residence, save during an actual siege, and that at such times it often only received the baron’s armed tenants, and not his mercenaries. Indeed, the builders of some of these keeps seem to have mistrusted their own troops as much as they feared those of the enemy. The staircases and galleries are often contrived quite as much to check free communication between the several parts of the building as between its inside and its outside. Further, the excessive jealousy in guarding the entrance, the multiplied doors, the steep and winding staircases, the sharp turns in the passages, although they helped to keep out an enemy, or, if he got in, placed him at a disadvantage, also rendered impracticable the rapid re-entry of the garrison, so that if the court or outer ward were taken by assault, the defenders had scant time to retire into the keep, which was thus liable to a _coup de main_. Otherwise, with a sufficient and faithful garrison, and ample provision and military stores, a Norman rectangular keep was almost impregnable, so great was its passive strength. Its windows were too small or too high for their shutters to be reached by fire-balls, and its walls were too thick to be breached or mined, if properly defended from the summit. This, indeed, was the true method of defence. An ordinary loop in a thick wall, however widely splayed, admitted of but little scope for an archer, or space to draw his bow. The lower loops were entirely for air, not for defence. Higher up, with larger windows, a bow could be used with advantage, but there were no flanking defences, for the angles had no considerable projection, and the shoulders, or lateral faces of the pilasters, were not pierced. With military engines for throwing heavy stones and masses of rock from the roof much might have been effected, but in the early keeps this was not contemplated, and probably not to any great extent in the later ones. An arrow shot from a battlement 50 feet or 70 feet high would lose some of its force in the descent. Of the siege of Rochester certain particulars are on record, and the account of the operations of the besiegers is confirmed by the evidence afforded by the existing keep. Rochester keep stands but 10 feet or 12 feet from the south-east angle of the outer wall, the angle of the keep corresponding with that of the wall. The angle of the wall and part of the adjacent curtain have evidently been removed and rebuilt, with the capping tower, in a later style. Opposite and behind this newer work, the lower part of the angle of the keep has also, at some remote time, fallen away, and with it a few yards of the adjacent sides. These parts have been rebuilt in a rude and slovenly manner, and the junction of the old and new work is very evident. This keep was built about 1130, and besieged by King John for three months in 1215.