258 



Dr. Collie and Prof. Ramsay. Behaviour of [Feb. 13, 



electrodes ; and yet, in order to avoid contamination of the gases, 

 they mnst have sealed joints. Closing with india-rubber corks 

 appeared to work well in batterer's experiments ; but then he had 

 large quantities of common gases at his disposal. In our experiments 

 this was unfortunately not the case. Our apparatus was, therefore, 

 constructed on the plan shown in the annexed figure. A is a tube 



Fig. 1. 



To 6 apply 





















D 







HHHH 





A 



about 15 cm. in length, and 8 cm. internal diameter. A fixed 

 platinum electrode, fused into a knob at the end, was sealed through 

 the glass at B. The wire at C was attached to a long spiral of very 

 thin copper wire, which was soldered to a platinum electrode with a 

 knob. In order to be able to alter the position of this electrode, it 

 was enclosed in a thimble of glass, which contained a roll of soft iron 

 wire, as shown at A. The position of this electrode could be altered 

 by means of a magnet ; and if close to A or to C, it moved on tapping 

 the tube gently, for the copper spring extended, or shortened itself, if 

 compressed or pulled out beyond a place of rest in the middle of the 

 tube. Gas was introduced into the vacuum tube through D ; a 

 phosphorescent vacuum having previously been made by means of a 

 Topler's pump ; the tube was washed out once or twice with the gas, 

 before each experiment. 



The current was an intermittent one from one of Apps' large 

 ind action coils, driven by a storage cell giving about 8 volts. The 

 hammer of the coil was not altered in position during the ex- 

 periments. 



It was found impossible to make measurements of the extreme 

 possible length of a spark. After the coil had been running for some 

 time, an isolated spark generally passed, even when the electrodes 

 were far apart. But consistent measurements could be obtained 

 when a rapid discharge began to occur on bringing the electrodes 

 together. It was found possible to obtain constant readings of dis- 

 tance to within half a millimetre ; and to form an accurate estimate 

 of the intensity of the rain of sparks, as contrasted with intermittent 

 discharges. 



