DISRUPTIVE DISCHARGE OF ELECTRICITY. 563 
discharge of the conductors of the Holtz machine, but the deflection to the 
negative side of the original zero (530) was so considerable that we always 
brought the index back to 530 before taking another reading, by touching the 
ball of the electrometer with the hand. We have plotted the results in the 
diagram, Plate XIV. The crosses mark the difference given in the fifth 
column, and the dots with circles mark the differences given by subtracting 
the deflections from the original zero, 530. The first of these is probably the 
proper curve, but whichever of the two we take, we get a straight line having 
a small negative intercept on the axis of ordinates precisely as in the case of 
the other two liquids. The curve, instead of showing a tendency to become 
concave to the axis of abscissz, exhibits a tendency to become convex with 
respect to it, thus differing in a remarkable manner from the curve obtained 
for a gas. The equations to the two curves within the limits of observation are, 
V = 900 s — 30 = 298 s —10 nm C.G. 5S. units, and 
V = 760s —17 = 252 s— 6inC.G.S. units, 
where V denotes the difference of potential, and s the length of the spark in 
centimetres. 
The reading for a 5 mm. spark in air under the same conditions, given as a 
mean by four observations, was 119; hence the electric strength of olive oil is 
3°5. This is less than that for paraffin oil or oil of turpentine (4 and 3-7 
respectively), as we expected from the fact that we were able to continue the 
observations through a slightly greater range of length of spark. 
The difficulty of making these observations was very considerable, as the oil 
had to be freed and kept free of all solid particles (Trans. Roy. Soc. Edin., vol. 
XXVill. p. 673), and the passage of a spark also produced bubbles of gas which had 
to be removed from between the plates before another reading could be taken. 
In the case of each of the three liquids we have examined, the discharge has 
been accompanied with decomposition of the substance. Having finished the 
readings, we took a note of the behaviour of these bubbles. When the plates 
were 11 mm. apart and the upper one charged positively, the bubbles were 
repelled downwards, and when suspended in the liquid, assumed the form of 
prolate spheroids, with the long axis in the direction of the lines of force. 
They discharged the electricity by convection. When the upper plate was 
negative the bubbles appeared to be repelled less strongly. The spark when 
examined by the spectroscope showed red and green with a tinge of violet. 
With the view of ascertaining the effect of condensed gas on the passage of 
the spark, we substituted palladium wires for the platinum wires in the 
instrument described in Trans. Roy. Soc. Edin., vol xxviii. p. 682. We 
