on Rotatory Polarization in Liquids. 93 



2 mm. apart in the mercury (see fig. 2). The spark gap 

 was about 1 cm. 



Fig. 2. 

 ft 



TO L£YDENJ/]R 



Fe 



> 



/ 



/ 

 l 





\ 









»9 



1 





W 



W/M 



7 



7 



W//M 



To determine approximately the duration of the spark, 

 photographs were taken of tinned iron wires threaded 

 radially near the rim of a rotating wooden disk, the disk 

 itself being painted black. The disk was driven by a small 

 motor to which was attached a revolution-counter. The 

 following times were observed during which the disk made 

 500 revolutions : — 15'0, 15*4, 14'8, 15*4, and 14'8 sees., 

 giving a mean of 15*1 sees., or 2000 revns. per minute very 

 nearly. The middle points of the wires, each about a 

 centimetre long, were 19 cm. from the centre of the disk. 

 A camera was focussed on the wires so as to give a magni- 

 fication of 2, and the disk having been set rotating, 

 photographs were taken by the light from the spark. The 

 velocity of the wires being 3980 centimetres per second, 

 a blur of width x cm. on the photograph will be produced 

 by a spark of duration t sees, where 



x = 2 x 3980 *. 



If the duration of the spark were 3x 10 ~ 6 sec. the blur in 

 the photographs would be '24 mm. A number of photo- 

 graphs were taken of the moving wires and compared with 

 photographs taken under similar conditions but with the 

 wires at rest. No blur, however, could be detected ; there- 

 fore the duration of the spark probably does not exceed 

 3 millionths of a second. 



Owing to the intermittent nature of the light from the 

 spark, it was not possible to use the most sensitive arrange- 

 ment of the apparatus. A half-shade angle of about 12° to 

 20° was necessary. The reading for the arc was first taken 

 and then the spark substituted for the arc and a fresh 

 observation made. The temperature of the solution was 



