Electromagnetic to the Electrostatic Unit. 311 



The results exhibited in the preceding table have been 

 divided into four groups. The first group consists of seven- 

 teen values found before the condenser was taken apart to 

 measure its electrostatic capacity. During this time the upper 

 half of the shell was lifted and the ball adjusted two or three 

 times. The values found by the fast fork are more uniform 

 than the others and average somewhat lower. The second 

 group extends from May 4 to 9, inclusive, when the condenser 

 had been set up again. There were two small glass tubes 

 about 5 mm in diameter and drawn out to about 2 mm, where 

 they projected (between 1 cm and 2 cm) through the shell into 

 the space within. They had once been used to pass charging 

 wires through. The wires had been withdrawn, and it was 

 supposed that the glass tubes had no appreciable effect. The 

 holes were together only ^-gY^-th of the area of the shell, and 

 the tendency of the glass to slightly increase the capacity 

 would tend to counterbalance the decreasing effect of the holes. 

 When the condenser was set up the second time the tubes 

 were intentionally left out, and the values of group II were 

 noticed to be larger than those of group I. ISTo cause could be 

 discovered for this increase (which indicates a less electromag- 

 netic capacity), but the tubes were replaced and group III 

 taken. The mean of this group is as large as that of the pre- 

 ceding group. The tubes were now again withdrawn and the 

 holes covered with gold foil, making the inner surface of the 

 shell continuous. Group IY gave values averaging almost 

 exactly the same as group I. The circumstances were alike in 

 other respects so far as is known ; the usual variations in the 

 conditions of the observations,' as already explained, occurring 

 in all the groups. I do not think the presence or absence of 

 the tubes could affect the capacity appreciably ; they were 

 altogether too small, probably not filling over a thirty-thou- 

 sandth of the space between the ball and shell. But that there 

 was a difference in the actual capacity of the condenser when 

 groups I and IY were taken from its value when TI and III 

 were obtained seems almost certain. As yet I have not become 

 satisfied as to the cause of this difference, but it seems probable 

 that in putting the condenser together some obstruction lodged 

 between the two halves of the shell and prevented them from 

 coming completely together. Had they been separated a few 

 hundredths of a millimeter only, the difference in question 

 would be fully accounted for. The surfaces of contact are 

 very accurately ground and polished, and loosening the screws 

 does not cause them to separate, as proved by the capacity 

 remaining constant. That the low fork should give higher 

 values for v than the high one (which means a lower value for 

 the capacity), is rather unexpected and not fully understood. 

 The low fork gave only a quarter the current given by the 



