that such was the case. While it is possible that this only means that some strengthening of the defences had rendered these ditches unnecessary, it is more probable that it indicates yet another distinct occupation. Generally speaking, such devices of fortification as the titulus and the clavicula appear to be characteristic of the forts with earthen ramparts. In the stone-walled forts with stronger gateways they must have ceased to be employed, and the natural inference is that the abandonment of these clavicula-shaped ditches, which took place at a comparatively early period, coincided with the erection of the surrounding wall. This is the more likely as we find that in the reduction of the size of the fort in the succeeding period the new defence on the west consisted of a wall. It is improbable that this wall would have been erected had the rest of the defences consisted simply of earthworks.

The second sign of change discovered towards the close of the operations was as follows. The barrack blocks in the Praetentura were traced out in the winter of 1906–7, and in the succeeding autumn two trenches were cut across the area from north to south, in the hope of finding earlier foundations. But it was not till the spring of 1968 that, during the investigation of the ditch of the early fort, the heavy foundations of a large building were discovered at the south-east angle of the fort, lying beneath the later barrack blocks. They could not have belonged to the first period, as they were carried over the ditch of the early fort, and it is therefore probable that they must be assigned to the second period.

The Phases of Occupation

It will be convenient to summarise the several phases of occupation thus indicated. First, there was erected a fort of somewhat irregular form with an earthen rampart, and having its gateways at right angles to the line of its ramparts. Second, this was succeeded by a fort more regular in form and somewhat larger in area, defended by a rampart and triple ditches, clavicula-shaped or overlapping ditches, being cut in front of at least three of the gates. Third, we have the same fort, but with these overlapping ditches filled up. It is probable that this phase of the occupation was accompanied by the building of the wall. Fourth, there came an occupation marked by a reduction in the size of the fort, and by the construction of the wall parallel to the Via Quintana. And Fifth, we find a return to the area of the second and third occupations.