MERGE TO GYRENE. 523 we have first mentioned ajjpears to have been, there are five passages (or vomitoria), by which the spectators entered, and two communicating with some place beneath the front of the stage which are so much blocked up with rubbish, occasioned chiefly by the fall of the roof, that we could not explore them to the end. These passages descend very abruptly towards the centre, and appear to communicate with the same point, or with each other ; they have been arched with blocks of stone, ranged longitudinally, and are of very good construction. We were able to go down tliirty-two feet in one of them, after some little trouble in clearing the entrance ; but the impediments which then presented themselves were too serious for our time and resources. A casual observer would not have been aware that there were any passages in this theatre by which the spectators entered, so much was the whole building covered with soil and vegetation ; and it was only on close examination, that some appearance of the arched roofs which covered them was discernable ; and we determined upon excavating in the same line below.

It soon appeared, that passages really existed ; and we succeeded in clearing one of them sufficiently to determine the fact beyond dispute. W e found that the roofs descended with the seats, some of which they probably supported, but the floors appear to have been level ; or, at least, the inclination is so slight (if there be any) as not to be ascertained by the eye. In the course of this excavation we found that some of the rows of seats w'ere hollow ; and were in hopes of discovering a further confirmation of the circumstance mentioned by Vitruvius, that the Greeks were in the habit 3X2