328 REPORT OF THE CALIFORNIA EARTHQUAKE COMMISSION. 
nary building sand from a creek bed, combined with various amounts of water. Some 
experiments were made with gravel, but lack of time and the necessity of completing 
the work for publication in the report of the Earthquake Investigation Commission 
prevented more extensive experiments. 
When the car was loaded with moderately dry sand containing 10 per cent of its weight 
of water or less, it was plain to direct observation that the sand was moving almost 
perfectly with the car, so long as the frequency was less than 2+ double vibrations per 
second. However, if the sand was wet locally by pouring water upon it, it was also 
very evident that the wet sand did not move at the same rate as the nearby dry sand. 
In the first place, the amplitude of vibration of the wet sand was greater than that of 
the dry sand; and in the second place, the reversal of motion was much quicker in the 
case of the former than of the latter. In the region between the wet and the dry sand, 
the difference in the relative motions of the two, causes the surface to be broken up by 
crevasses which open and close periodically. This breaking up of the surface is quite 
irregular, varying from moment to moment. 



Fig. 61. — Curves obtained on recording drum. Reduced about half. 
For a precise determination of the relative motion of the car and the sand with which 
it is loaded, it is necessary to measure the curves traced on the revolving drum described 
above. The method of doing this will be best illustrated by taking a particular case. 
Fig. 61 is a copy of one of the curves obtained on the drum. The lower sinuous curve 
is the trace made by the pencil attached to the car; the middle zigzag line is the trace 
made by the pencil attached to the block embedded in the sand; the upper line is the 
trace of the electromagnet beating seconds. In this particular case the sand contained 
