The effect was to gain a much greater length of flanking defence; to cause assailants of the gate to pass between a longer and narrower lane, with thorny sides. From this stage effort seems to have been confined to a development of this new principle. At Pevensey a cul-de-sac is formed between two great bastions of the main wall. The gates of Cardiff and Lambessa illustrate the same motive, modified by local peculiarities. The famous gate at Trier, known as the Porta Nigra, is a well preserved illustration of the employment of a similar plan. At Silchester the same effect is secured by a very pronounced in-curving of the two arms of the main wall towards the gates. It is noticeable how the defences of the walled city of Silchester differ in many details from the precise plans employed in the forts. It is indeed a doubtful point, productive of much discussion, to what particular period the existing masonry of the walls belongs. So far as the present illustrations help, it is plain that, while the principle of defence belongs to the new order of things, yet the plans of the gates (indeed, the existence of the double gates) preserve the older method. Possibly the work may be regarded as transitional-i.e., before the end of the third century. Postern ways, so far as the examples of Ribchester and Richborough are any criterion, were protected by a turn in direction, which covers the direct access to the interior. The former was built through the ruin of a corner tower; the latter was part of the original design, and was defended from a gangway overhead. A great part of the defence of these later works is obviously concerned with the rampart walk and its contrivances. A chance observation of an outof-the-way fortress at the oasis of El Khargeh, west from Girga, in Upper Egypt, made during the past spring, has helped to throw some light upon this point; otherwise, information upon it was nil. Unfortunately, the notes made do not quite agree with some photographs taken at the time, but they are sufficiently in concord for our purpose. There, as may be seen from a glance at the photograph on the last plate, the towers stood well up at regular intervals externally, and were connected by low parapets running along the intervening spaces; the chambers upon the towers, at the level of the wall top, embraced also the width of the wall, and are entered from each direction by means of doors. It is not clear whether the towers were provided with large apertures for military purposes on their outside; but possibly it may be found practicable to send an expedition to make further enquiry during the coming winter. In observing these differing modes of expressing the principles of defence in vogue, and the changes in those principles which took place during a space of three hundred years, it is well to bear in mind the historical aspect of the cases also. It is natural to suspect that special needs of differing localities were largely responsible for changes in form. As to details of later works there is much truth in that; but the fact remains that the earlier works admitted no such deviation. The early class of forts was called for by the necessity of defending the provinces of Britain against tribes. difficult to control, living in the north and in the There slowly elaborated from that need a system of defence which received its chief stimulus. in the time of Hadrian, and was perfected under Antonines. To these days in the first half of the second century possibly belong the 6-acre forts of the type o Chesters and Housesteads and |