1(JS THE STRUCTURE OF THE NUCLEUS. 



.002 and .003 cm. in tliaineter. To nioiint it tlie eutl of the fiber was thrust into 

 the mantle of a tlame and bent sharply at right angles, then fastened to a suitable 

 rod with cement. 



The sensitiveness so obtained was (^ = .14, aotnally less than the a = .104 of the 

 last i)aragraph. Finding no finer (juartz fibers available, I lesolved to return to the 

 silk bifilar. The experiment made, however, is snfficient to express the conditions 

 which hold for quartz fibers, as follows : Tbe needles used below weigh less than 1/2 

 f'ram. A good quartz fiber, strong enough to hold it, will therefore not be moi'e 

 than .0005 cm. in dianietei'. Consequently the torque will be reduced over 500 

 fold, and the sensitiveness of the instrument now be increased to 0-= 70. This is 

 so much in excess of the sensitiveness reached with the silk bifilar that other 

 practical difficulties (excessive dam[)ing coefficient), will intervene long before it is 

 reached. 



9. Single-Jibered ullh. — These were obtained by splitting the above silk fiber 

 into its f(mr constituents by submerging it in hot water to dissolve the glue. The 

 bifilar obtained in this way was (after desiccation) in every way entirely satisfactory, 

 except with regard to its conduction. The period of the needle increased to 60 sec, 

 its damping coefficient to 37, so that the needle was pi-actically aperiodic. The 

 heterostatic sensitiveness obtained was a = .33, over 9 times the original value. 



In using the instrument idiostatically the greatest cai-e must be taken to remove 

 all charge from the needle, as otherwise the deflections will vary as the first and 

 second powers of the needle conjointly. Indeed, the second power may often 

 relatively vanish, as will presently be shown. In the experiments the needle was 

 discharged by putting it to earth just before the measurements, but even with this 

 precaution some residual charge is liable to be evidenced by the unsymmetrical 

 deflections. 



If the needle carries a charge and if V^ = 0, d = aV^^V^ — V „/'2) = 

 aV^Vi — aVl/2. Hence, if the charge of the quadrants V^ is reversed, the 

 mean deflection will be e = aVl/'2. The law of squares must hold. If the 

 needle carries a permanent charge equivalent to Fj', this is not reversed on 

 commutation. But in the mean value of 0^ = aV^{V^ -{-V^) — aV\/2, and 

 0, = (- F2)(- F0 + F3)- F^/2, viz., = «F^/2, the law of squares still 

 holds. If, however, charge runs into and out of the needle during measurement 

 so that the sign of F3 = V'^ is also reversed partially at least, then in the mean 

 a = aF*/2 + aFg Fg the sinqile law of squares will be either accelerated or 

 retarded according as F is positive or negative. In the experiments the latter is 

 usually the case, showing that the needle leaks. The permanent charge of the 

 needle appears in the difference of deflections as d' = ~n F, FJj, and this may also 

 be made out from the data. Indeed, it has been stated that sometimes the linear 

 law may preponderate. 



The sensitiveness reached in these experiments was « = 2^/ F* = .40, when 

 the needle was charged and a = .33 when it was put to earth. This again is, on the 

 average, over 10 times the sensitiveness found in paragraidi 2 for like experiments. 



10. Neiv electrometer. — After finishing this work, I thought it expedient to 



I 



