332 REPORT OF THE CALIFORNIA EARTHQUAKE COMMISSION. 
quarters of the surface remaining practically level during the vibration of the car. 
When the sand contained less water, the surface midway between the ends and the middle 
of the car was badly broken up by crevasses and ridges at right angles to the direction of 
motion. 
It doubtless frequently happens in river valleys, coastal plains, or “made land” that 
a very soft water-soaked subsoil is covered with a crust of more solid ground. This 
condition of affairs was imitated on a small scale. A lower layer 13 cm. thick of very 
wet sand containing 20 per cent of water was covered by a piece of oilcloth, and upon 
this was placed a layer 12 em. thick of much drier sand containing 10 per cent of water, 
and tamped into as compact a condition as possible. The block carrying the tracing 
pencil was embedded in this upper layer. If the whole load of sand had been like the 
top layer, it would have oscillated almost perfectly with the car. No such result was 
obtained with the two layers. When the car was shaken, it was apparent that there 
was still considerable freedom of motion in the lower layer. The upper layer moved as 
though it were floating on a semi-fluid mass. It rose and fell at the ends, and this 
motion extended to the middle, causing the block to rock back and forth, a result which 
was not obtained when the car contained a load of uniform consistency. The to-and- 
fro motion of the block was considerably greater than that of the car, for frequencies 
of 2 or 3 per second. For frequencies greater than 3, the amplitude of the block was 
less than that of the car. The results of this experiment are given in the tables on p. 330, 
while a plot of the same is included in fig. 62. The results, however, do not do justice to 
the possible destructiveness of such a motion. The rocking motion of the upper layer, 
as well as the violent manner in which it was broken up into fissures and ridges, seems 
to show that the destructive effect of the shaking motion of a semi-fluid mass may be 
increased when it is confined by a superincumbent layer of much more solid and com- 
pact material. 
In the last experiment with the shaking machine, the car was loaded with coarse gravel. 
The gravel consisted of water-worn pebbles of all sizes up to 2 inches in diameter. It 
contained no clay nor sand to bind the gravel together. When this load of dry gravel 
was shaken, the block embedded in the gravel moved with the same amplitude as the 
car until the frequency reached 3 double vibrations per second. With higher frequencies 
the amplitude of the block was somewhat greater. Considerable water was then poured 
into the ear, and it was‘again shaken with various frequencies. The results were similar 
to those obtained with the dry gravel, except that the relative motion of the gravel with 
respect to the car was nearly twice as great as in the case of the dry gravel. The data 
for these experiments are given in the table, while a plot of the same is given in fig. 62. 
A consideration of the meager and more or less erratic data described above suggests 
various questions and criticisms. It has already been explained why more extensive 
experiments involving other materials were not undertaken. The erratic nature of the 
experimental data is not due to the method of experimentation employed, but to the 
uncertain and varying condition of the material with which the car was loaded. If, in 
the beginning of a series of experiments, the composition of the load was thoroly uniform, 
this was no proof that it remained so. A few moments of shaking sufficed to change to 
a greater or less extent this uniformity. When the material contained a large percentage 
of water, continued shaking caused the material close up to the ends of the car to pack 
and become somewhat drier; this was also true, tho to a much less extent, of the middle 
portion. The portion midway between the ends and the middle, where the relative 
motion of contiguous portions of the load was the greatest (thus causing fissures and 
ridges to develop), noticeably increased its content of water. This development of non- 
uniformity in the consistency and composition of the load is a sufficient explanation of 
the irregularity of the results obtained. 
