important guide, as it marked a definite stratum helping to define the different changes that had taken place. The small bath-house, from its method of foundation and the evidences of subsequent alteration, was clearly earlier. The large block of building to the west must also have been in existence prior to the making of these defences, arid it was presumably abandoned at that time. Two of its walls had been cut through by the ditch, while the end of one of them was detected passing beneath the foundation of the rampart. These walls, like the rest of the building of which they form part, but unlike the early bath-house, lie on river cobbles.

Both the early bath-house and the western block, then, seem to have been older than the ditch and rampart. What was their relation to one another? Did they belong to the same period? It was unfortunate that a complete plan of the western block could not be obtained. The walls lay at no great depth, and at the west end their scanty remains could be traced without much difficulty. But as they came near the ditch, portions had disappeared entirely, and to the east of it, with the exception of the two walls in question, no part of the building could be traced at all. In such bath buildings as those at Silchester, the rooms properly belonging to the bath are approached by a spacious pen style and a large apodyterium. An entrance courtyard is also to be seen in the plans of more than one of the military bath-houses in Germany. In this case, however, the block, although at first sight it seemed to suggest a courtyard entering from the west, proved to be marked by none of the peculiar features of the bath. There were no signs of hypocausts or of the debris of tiles and plaster work, and no remains of apsed apartments. Moreover, the building lies in a different alignment from the baths themselves. No prolongation of its incomplete walls indicates that the two were ever joined together. The block to the west, in fact, appears to have been built for some separate purpose. It is possible that the walls cut through by the ditch formed part of a corridor, by which access was obtained to the bath buildings. Of these walls, the one lying to the north was traced for what appeared to be its full length. It terminated on the edge of a large pit or well (LVII) which lay partially beneath the cobbling. It was built of sandstone, three or four courses of which remained at the end. There was no sign of any return.

Although it was very evident that the bath buildings had undergone considerable alterations at various periods, the exact nature of these changes