Page:Medieval Military Architecture in England (volume 1).djvu/465

 Colchester Castle. 429 sent entrance at Colchester leads to the conclusion that it was not originally so here. It is evident not only that the present ornate doorway is an insertion, but that the outer part of the wall above has also been taken down for a considerable height, and clumsily rebuilt. This was done to allow of the insertion of a portcullis. Without this it would have been sufficient to remove the lower masonry only, but the grate required head-room when raised, and to gain this a more extensive alteration was necessary. Immediately above the dripstone are a few courses of ashlar, derived, no doubt, from the old work; but above this the wall is of very inferior uncoursed rubble, very different from the regular courses of stone and tile seen in the wall on either side. When the outer door case was inserted the wall within was also lined with ashlar, and the ring stones of the inner arch and of the entrances to the great staircase were also so cased. Above, the recess into which the portcullis was lifted is of tile, and evidently original; but the floor with the aperture or chase, and the flat back of the recess are later.- In all proba- bility there was originally a recess with perhaps a loop where the great doorway now is, and there was certainly a recess above, with probably a loop also. When this door was opened, in the latest Norman period, the original entrance on the first floor, in the north front, was probably walled up, as it still remains. It has been suggested that the great doorway replaced a postern, but posterns in Norman keeps were very unusual indeed, especially at the ground- level, and there seems no reason for supposing one here. Probably when the keep was constructed there was no encei7ite wall, but when this was built, the keep became far more secure, and as it was convenient to have an entrance from the town side, the change was made. There is something to be said for the pertinacity with which this keep has been asserted to be a Roman building. Not only is its design peculiar, but this is also the case with the material employed, and in some degree with the workmanship. The walls, though cased with ashlar on the pHnth, and though ashlar is largely employed in the quoins up to the first floor, contain great quantities of the large, square, thick tile so characteristic of Roman work. Moreover, these tiles are mostly whole, as though they had not been em])loyed in an older building, and they are laid in bonding or chain courses, single, double, triple, and even in four courses, with intervening bands of cement stone in small rudely-squared blocks. Sometimes the tiles are laid on edge, sometimes slightly incHned, with here and there a tendency to herring-bone, and the greater part of the interior dividing wall is regular herring-bone, and a very fine example of it. Most of the recesses within the walls were not originally lopped, and are round-backed and semi-domed, which is unusual in Norman keeps. Outside the building, high up in the walls, are traces of a number of apertures or perhaps only sunken spaces, about 2 feet 6 inches square. One of these is shown in an early drawing as containing a sun-dial, no doubt an insertion. It