Some delay in penetrating the mud was caused by a bed of oyster shells, which had to be cut through by one edge of the cylinder. In consequence of some irregularities of the surface of the rock, the cylinder at first deviated considerably from an upright position; and it was necessary to use the pneumatic apparatus to gain access to the rock, and excavate it. The height of the annulus below the dome was such that it was not quite filled by the mud when the cylinder rested on the bottom. The work of getting the mud out of the annular space was much facilitated by the division of it into compartments. By February 1855, the cylinder had been sunk to its full depth in an upright position, and it then rested everywhere on the rock, its lowest point being 87 feet 6 inches below high water. Much trouble was given by a spring of water issuing out of a fissure in the rock, in one of the compartments, but the flow was stopped by driving close sheet piles into the fissure. The rock in the annulus was dressed, and the space filled by a ring of granite ashlar masonry which was built to a height of about 7 feet all round. The state of the work at this time is that represented in the section of the cylinder, (Pl. V. p. 218). The rock consisted of greenstone trap, so hard that tools could with difficulty be got to work it. When the ring of masonry was completed, it was expected that the bottom might be sufficiently water-tight to act as a coffer-dam, and allow of the mud being taken out from the central part of the cylinder, below the dome. But the pumping power was not at first sufficient for this purpose, and it was thought that it would be necessary to employ the pneumatic process in this space also. However, by rapid and incessant pumping the water was lowered so as to allow of the mud and rock being excavated, and the masonry in the central space built without having again recourse to the use of air pressure. The leakage water was conveyed to two wells, formed of cast-iron pipes built into the masonry, from which the water was pumped. The inner plates of the annulus were cut out, and the work in the centre which consisted of granite ashlar set in cement was thoroughly bonded into the ring of masonry already built. When the work was carried up to the level of the dome, both the dome and the internal 10-foot cylinder were cut out and removed. When the building had been carried up some height, the pump wells were filled with cement concrete, and the influx of water stopped. Finally, about the end of 1856, when the masonry was completed to the cap of the pier, the upper part of the great cylinder was unbolted and taken ashore, it having been made in two halves with that object. Thus the most difficult part of the undertaking was successfully completed.[107]