DowLiNG — Sensitive Valve Method for Measurement of Capacity. 181 



III. — Application of foregoing as an iiltra-micrometer and 

 1nicro-]press^l,re gauge. 



The first trials were made with a simple form of plate-condenser, in 

 which one of the plates was movable normally by means of a micrometer- 

 screw. The two plates were 2'o cms. in diameter, and were carefully ground 

 to a plane surface. The movable plate was attached to the end of a short 

 rod, which was ground to a smooth sliding fit in a short length of brass tube. 

 This tube was screwed in in place of the anvil of a micrometer-screw gauge, 

 and a couple of springs were arranged to pull the sliding-rod so that its 

 extremity made tight contact with the end of the micrometer-screw. This 

 arrangement was screwed to the long leg of an L-shaped piece of ebonite, 

 so that the other plate, supported on a rod inserted in the shorter limb of 

 the L, was opposite the first plate. By a device which I shall describe later, 

 the two plates were made exactly parallel, and the micrometer allowed them 

 to be moved apart any desired fraction of a millimetre. 



An example of the results obtained with this rough arrangement is given 

 in fig. 5, where the abscissae give the micrometer readings in thousandths of a 

 centimetre. The points lie nearly on a straight line over the range covered ; 

 but the sensitiveness is not very great. The fact that the points lie a little 

 to one side or the other of the line is obviously due to the imperfect action 

 of the screw. 



It is easy to see that the sensitiveness can be greatly increased by 

 bringing the plates closer together, and also by enlarging the areas of the 

 plates. With plates close together and of moderate size, they form a con- 

 denser of which the capacity is much greater than that of the grid and other 

 connexions. In equation (I) we can thus regard C as wholly the capacity of 

 this condenser. G will then be given by the formula — 



C = ;i^ V (9 X 10") farads. 



And we may write (I) in the form— 



v^B.^ . . . ■. . (II.) 



if we include all the other quantities in the constant B. Now, the slope of 

 the a.c. characteristic being practically uniform, a given change in v will pro- 

 duce a corresponding change of constant amount in the galvanometer reading, 

 no matter what may be the actual value of v. Consequently, the condition 



for sensitiveness is that -rr should be as great as possible. But 



%--4 (™) 



