This collapse contains the mud brick superstructure of wall 20023=17002, disturbed by plough activity, and so best preserved close to wall 20023. Since the collapse covered pavement 20090=20095, it should perhaps be associated with the abandonment of this room after it was reoccupied post mid-1st c. AD fire event. The collapse indicates that the stone socles of the walls were re-used and the superstructure of the walls rebuilt with mudbrick in a timber frame. This will force us to look again at the collapse phases in the other rooms to discern whether they relate to the fire event. The mud bricks found in 2017 abutting the end of wall 17002 (17086 and 17089) may, in fact, belong to this collapse. In 2017 we noted that they were high up in the matrix and may have belonged to a post-fire phase. We could also perhaps equalize 20089 with collapse contexts 17007 (the orange-brown matrix in which the collapsed bricks 17086 and 17089 were found) and 17015 (which covers walls 17012 and 17047) as well as the collapse of mudbricks 17069. 17069 was originally interpreted as evidence of reuse of the bacino in a post-fire re-occupation of the building but now I wonder if perhaps these bricks, which abut the back of the bacino (17013), actually belong to the superstructure of 17047.

The collapse appears to belong to the superstructure of wall 17002 = 20023, and may represent the abandonment of room D in Roman period phase 3. The bricks at the end of wall 17002 found in 2017 may belong to this same collapse.