THE STKUCTURE OK 'HIE iSTUCLElIS. 1(37 



if the mean deflection is taken, it was found tliat (x = .084, so that tlie oriiriual 

 sensitiveness has already been increased 2 or 3 times. 



6. Idiostatlc data. — I next discarded the watei- batteiy and joined one pair of 

 quadrants with the needle, putting the other pair to eartli. The teni[)orary deflec- 

 tions only were taken. The experiments were varied by exclianging the charge of 

 the quadrants and by again earthing the needle (capsule), in which case it was 

 necessary to change the zero manually. The data for the earthed needle happen 

 to l)e the smoothest. Computing a from the three groups of ex[)eriments made, 

 a ^.126, a =.090, «^.138 were obtained. These differences as well as the 

 relatively lai'ge values of (x for the smaller charges (5 volts) indicate a charge in 

 the needle which was not removed, for instance, by touching the needle by an 

 earth wire. Comp)ared with the original value (« = .030) the present increase of 

 sensitiveness is 2 to 4 times larger. This is an excess of the increase (2 1/2 times) 

 in § 4. It may be noted again that if the needle carries an independent charge 

 (due to the ccmduction of the silk fiber), V^ = 0, while Fg and V^ are difl^erent ; 

 whence =: Va^i^Vz— I^^s/^)) which is a maximum if V^ = F;,. Thus, if the 

 needle carries such a charge the maximum is approached sooner or later than in 

 the normal case of no charge. All this points to the importance of insulated silk 

 fibers. 



7. Miscellaneous tests. — The use of mica discs at e in figure 2 is apt to be 

 accompanied b}^ hurtful polarization. Approaching these discs too closely to the 

 needle, the damping soon becomes excessive (aperiodic), and the slow moving 

 needle is found to be unstable. Slight gain in sensitiveness in this case is no com- 

 pensation for the irregular behavior and the narrowed limits of deflection. 

 Paraflined copper discs showed no advantage. All this contrasts strongly with 

 the steadiness of behavior when the discs e, e, are moved far apart. 



Before proceeding to test the quartz fiber suspension, I determined to increase 

 the sensitiveness by reducing the bifilar distance further. The results did not show 

 the increase of sensitiveness anticipated, proving that the torque of the silk fibei's 

 is now seriously in question and suggesting objections to fibers glued together in 

 multiple. The electi'ometer held its charge (needle) well, the readings at intervals 

 of 15 sec. being 60, 61, 62, 64, 65, etc., with the zero at 101. This is a drop of 

 about .12 per second, without special desiccation, a favorable value in view of the 

 small capacity. The sensitiveness gained is now a^ .164, exceeding the original 

 value (a = .036), 4 1/2 times. 



The effect of the vane c alone, when the quadrants are all metallically con- 

 nected, was about 1/50 of the deflection due to the quadrants. A similarly interest- 

 ing question is the relative amount of induction conveyed by the cylinder, b, and 

 by the barriers, e, e, separately. This was found by putting the barriers to earth and 

 chai-ging the cylinder, and vice versa. The results showed the effective charge to 

 reside in the equatorial parts of the capsule, nearest the edge of the disc, c, of the 

 needle. The barriers contribute about 1/6, the cylinder 5/6. 



8. Quartz fiber. — My next endeavor to inci'ease the sensitiveness consisted in 

 replacing the bifilar suspension by a quartz fiber. The one selected was between 



