28 THE SAN FRANCISCO EARTHQUAKE ANE) ElRE\ 
roof trusses collapsing. Nearly every wall was moved slightly, but 
the brickwork was generally very good, and apparently had cement in 
it. The exhaust house had three intermediate walls, 18 inches thick 
at the top. The north wall and the next one fell into the building, 
the side walls being pushed out 6 inches. The building had wooden 
roof trusses and the north truss cracked at the center mortise. The 
floor settled badly around the condensers. The gas holder collapsed 
from the sudden release of the gas due to a break in the mains. The 
trestle pier extending into the bay also collapsed. 
Most of the structures built on piles along the bay suffered consid- 
erable damage, especially the frame sheds on the wharves. The 
Union Ferry Building (PI. XLVI, A), the terminus for all the fer- 
ries plying on the bay, is built on piles. It was more seriously dam- 
aged by the earthquake than would appear at first sight, and barely 
escaped the fire also. It is of interest to consider in some detail the 
behavior of this structure under the action of the earthquake. The 
ten-story tower was so seriously damaged as to require the removal of 
the masonry walls, and will probably have to be rebuilt. This tower 
consists of a steel frame which was inclosed with heavy sandstone 
walls backed with brick for several of the lower stories, and with 
sheet metal above. The floors above the masonry part Avere of stone 
concrete reenforced with expanded metal between I beams. The brick 
walls Avere badly shattered, and a large section Avas thrown out just 
below the clock on the west front, while on the east front a large mass 
fell through the skylight onto the upper story of the main building. 
There Avere but few cracks in the north and south Avails. The steel 
time-ball staff Avas badly bent, indicating a considerable movement of 
the top of the tower, probably resulting from the first shock. The 
steel work Avas severely racked, the greatest damage being just below 
the middle. Some of the diagonal braces Avere sagged, having been 
stretched beyond their elastic limit (PL XLVI, B). In the southeast 
corner of the third story the bottom and top loops of one of the 2-inch 
square diagonals Avere pulled apart, and several of the rivets in the 
angle connection were sheared off; in the southwest corner also the 
top diagonal loop pulled apart, the diagonal being considerably bent ; 
in the nortliAvest corner seven of the eight f-inch riA r ets in the angle 
of the connection for the diagonals were sheared off (PI. XLVII, B) ; 
and in the northeast corner all but 2 inches of the top angle Avas 
sheared, and the north diagonal Avas bent, the loop having been pulled 
apart. On the second floor, in the northeast corner, the top east 
diagonal connection pulled away, shearing the rivets; in the south- 
east corner the top angle of the diagonal connection sheared for 
about 2 inches, and four of the 1-inch rivets holding the cover plate 
of the southwest column (PL XLVII, A) were sheared, as were also 
