106 THE SAN FRANCISCO EARTHQUAKE AND FIRE. 
and in this case there can be nc question that the effect was entirely 
due to the earthquake. The masonry in the tower was of an admira- 
ble quality. 
In PL XLVI, B, is shown a crack in the brick masonry on the 
inside of the tower, together with a part of a diagonal tie-rod which 
had been stretched beyond the elastic limit and was hanging with a 
noticeable sag. It will be noted that the brickwork was well bonded 
and that the joints were well filled. The mortar was much harder 
than the bricks, but both were of good quality. It is also evident 
from this view that there was no general shattering of the entire 
mass ; there was a well-defined crack, but nothing else. 
At the northwest corner of the tower, about halfway from the roof 
of the main building to the top of the masonry walls of the tower, a 
diagonal tie-rod had been fastened to the wall girder by means of a 
gusset plate, with eight rivets in it. Seven of these rivets were 
sheared under the action of the earthquake, leaving the plate hanging 
by the eighth rivet at the time I saw it ( PI. XLVII, B) . 
One detail of the bracing in the ferry-building tower, of which 
a satisfactory photograph could not be procured, was as follows: 
The wall girders at the different floor levels were utilized for the 
wind struts of the bracing. In the lower part of the tower the diag- 
onal tie-rods were fastened directly to the wind struts. The ends 
of the wind struts rested between upper and lower seats attached 
to the columns, and were, as a rule, bolted to each of those seats 
with nothing but two J-inch bolts, the idea evidently being that, as 
the struts resisted compression only, it was not necessary to fasten 
them to the columns with anything designed to resist any consid- 
erable force tending to separate them from the column. The upper 
seat was evidently designed to take the vertical component of the 
stress in the diagonal, and the lower seat to take this load in addi- 
tion to the ordinary load which the wind strut transmitted to it in 
its capacity as a girder. The bolts fastening the ends of several of 
the wind struts to their seats had been sheared, and the struts had 
almost slipped out from between the seats. 
In the upper part of the tower a different method was adopted, 
the diagonal tie-rods being fastened to bent plates that passed around 
the outside of the columns opposite the ends of the wind struts. No 
shearing of the bolts or slipping of the struts was noticeable at these 
points. It is plainly apparent that ordinary assumptions made in de- 
signing wind struts will not apply when it is desired that the bracing 
shall resist earthquakes. There is evidently a tendency for the col- 
umns at the ends of a wind strut to buckle outward, and thereby also 
a tendency to pull the strut out from between its seats. In fasten- 
ing bracing to resist earthquake shock, therefore, the struts should 
be fastened to the columns at their ends much more securely than 
