ON IMPACT WITH A LIQUID SURFACE. 
141 
Otherwise, we adopted the plan of “ numbering the phenomenon ” by means of a 
ticket which was photographed simultaneously. 
The Duration of the Spark. —Although it was not necessary for the success of our 
experiments that the spark should be excessively short, yet y/e cannot doubt that its 
effective duration was less than three-millionths of a second. This was ascertained 
by photographing by its means a cardboard disc, 22 centims. in diameter, roughly 
graduated round its edge with pen and ink, and kept rotating by means of a small 
electromotor at a rate of between 52 and 54 revolutions per second. This rate of 
rotation was ascertained by smoking the back of the disc and touching it v/ith a 
style attached to an oscillating tuning-fork, or with the fork itself, and the result 
was confirmed with a second fork. The edge of the disc was thus found to be 
moving at about 36'5 metres per second (or about 78 miles per hour) during 
exposure, yet no trace of motion is apparent in the photographs. A motion of one- 
tenth of a millimetre during illumination would correspond to less than tln-ee- 
millionths of a second, and would have prodnced in the photographs a blurring of 
three-fourths of one-tenth of a millimetre which, with a lens, would certainly be 
visible. This interval of less than three-millionths of a second bears to one second 
just about the same ratio as a day to a thousand years. 
Although the outside limit thus obtained is 30 times greater than the effective 
duration of the spark employed by Professor Boys with his flying bullets, yet we 
think it worth while to mention the result, for it shows how excessively short is the 
exposure necessary for taking even very detailed “objective views ” as distinguished 
from shadow photographs requiring no camera. 
It may be mentioned here that the illuminating spark-gap was sometimes as much 
as 3'3 centims. wide ; that the two jars were of not quite equal capacity, the lesser 
consisting of a set of jars and presenting an area of about 4000 sq. centims. and a 
mean thickness of dielectric of about 0‘3 centim., while the potential difference was 
such as would make a spark leap across from P to Q when'the distance betvreen 
them was about 3 centims., P and Q being spheres of 2 centims. in diameter. 
The Accuracy of the Timing. —The interval between the release of the drop 
and the production of the illuminating spark is liable to slight variation, chiefly 
from two causes—(i.) irregularity in the potential difference between the two 
terminals P and Q, on account of which the spark would leap through varying 
distances to meet the timing sphere before it reached the line of centres, and 
(ii.) want of consonance in the rates of demagnetization of the two magnets. Thus, 
after long running, one magnet would get hotter than the other, and again an 
alteration in the strength of the magnetizing current was found to shift slightly the 
stage of the splash revealed by the spark. Nevertheless, after allowing the current 
to run for a few minutes, and after taking a few preliminary discharges, a condition of 
affairs was reached so steady that changes in the height of fall of the timing-sphere 
corresponding to intervals not greaffer than 2 or 3 thousandths of a second, produced 
