70 THE SAN FRANCISCO EARTHQUAKE AND FIRE. 
furring, lathing, and plaster, the walls of the vaults were likewise of 
these materials. Where the interior partitions consisted of hollow 
tile, the walls of the vaults were of hollow tile also. Although I 
examined a great many, I did not see a single vault partitioned off 
either with metal lathing and plaster or with hollow tiles that pre- 
served its contents. I was informed by some gentlemen, who were 
apparently connected with the Spring Valley Water Company, that 
on the top floor of their building a vault walled off with hollow tiles 
had protected its contents, but that the corner of the building in 
which it was situated had not been completely gutted, so that the 
vault did not receive a severe test. 
In the Baltimore fire there were a number of vaults walled off with 
hollow tiles, and all that I happened to see during my inspection of 
the ruins in Baltimore had failed. The same thing was in evidence 
everywhere in San Francisco, and it is my opinion that this result 
could have been predicted with absolute certainty at the time these 
vaults were built, from data then available. To all external appear- 
ances, no doubt, the vaults looked like secure places in which to 
keep valuables; as a matter of fact, they were the flimsiest kind of 
shells, not capable of resisting any sort of determined attack from 
either fire or burglars. The tenant would have been better off with- 
out the vault, for in that case he would probably have carried his 
papers to some other point where they would have had a better 
chance to escape the fire. 
The only vaults I saw that came through a really fierce fire without 
damage were those built of brickwork (PL LII, A). Even these 
vaults did not always protect their contents, however. I saw a num- 
ber of them opened in which the contents had been totally destroyed. 
As they seemed to be fairly good vaults, this result was a matter of 
more than ordinary interest. I therefore carefully examined a num- 
ber of them and discovered that the fire had gained access through 
cracks due to settling, or to the earthquake, or else through unfilled 
joints, due to poor workmanship in the original construction of the 
vault. It appeared that probably the contents of the building Avere 
burning fiercely around the vault before the floor above had burned 
out or collapsed, so as to give full vent to the gases of combustion. 
Some pressure must have been generated by the great heat thus 
confined, and under this pressure the incandescent gases resulting 
from the fire found their way through the smallest and most tor- 
tuous passages in the brickwork. In several cases it was apparent 
that the contents had probably been ignited by a small tongue of 
flame (probably not thicker than a lead pencil) penetrating into the 
vault as a result of such conditions. 
A few vaults failed owing to the fact that the outer door warped 
and pulled away from the frame. Whether this warping could have 
