284 Professors Ayrton and Perry on the Ratio of the 



(fig. 2) the same in plan. The plate is strengthened by stout 

 brass ribs D F, E G (fig. 2), and AB,BC (fig. 1). L, L, L 

 are three chemically cleaned and paraffined ebonite levelling- 

 screws, the ends of which are thinned to a blunt point so as to 

 allow extremely little surface leakage, and by means of which 

 the plate ABC can be adjusted parallel to the lower plate, 

 H J K (fig. 1), which is shown In plan (fig. 3) a§ LHP. 

 This plate is also strengthened by stout brass ribs underneath, 

 LN,MP (fig. 3), and H J. JK (fig. 2). This lower plate, 

 by means of hole-slot and plane, rests on three chemically 

 cleaned and paraffined ebonite levelling-screws, I, I, by means 

 of which its upper surface is made to exactly coincide with 

 the top of the guard-ring Q R, S 1 (fig. 1)^ and UVWX 

 (fig. 3). This guard-ring is rigidly soldered to the upper 

 edges of the brass box b b, bb (figs. 1 and 3), three projections 

 on the sides of which support, with hole-slot and plane, the 

 levelling-screws L, L, L. Into the bottom of this box screw 

 the ebonite levelling-screws I, I. Small vessels containing 

 calcium chloride (not shown in the figure) are placed inside the 

 brass box, bb,bb, to keep the ebonite in the neighbourhood of the 

 ebonite levelling-screws I, I quite dry, in order to avoid the possi- 

 bility of surface-leakage. In the earlier experiments the space 

 between the edges H K of the lower plate and of the guard -ring 

 R S (fig. 1) was very small; but afterwards, to avoid the pos- 

 sibility of leakage across by sparking or otherwise, this was 

 increased to 2*5 millimetres, and the area of the lower plate 

 thus reduced to 1323*14 square centimetres. The errors 

 arising from the surfaces of the condenser-plates not being 

 true planes were practically eliminated by capacity-experi- 

 ments being made with successive adjustments of the con- 

 denser-plates, a different set of points in the upper plate being 

 each time brought to the fixed distance from the lower one. 



4. The Galvanometer. 



The galvanometer employed was one constructed some time 

 back by Messrs. Elliott from our own design. It had a resist- 

 ance of 19,970 ohms at 21° # 9 C. In ordinary use, when fitted 

 with an astatic combination, four magnets being used top and 

 bottom, one Daniell's cell would give, through a resistance of 

 600 megohms, a deflection of 130 scale-divisions on a scale 

 about 1^ metre apart. But this arrangement, with its alumi- 

 nium vane, had far too much damping for being used ballisti- 

 cally. We therefore commenced by removing the vane and 

 weighting the lower set of needles with pieces of brass so as 

 to give it a barrel shape ; but if the brass was light we found 

 there was too much damping, and if heavy too little sensibility : 



