92 REPOET OF THE SECRETARY. 



one-third as great as that of the very light system then in use. To discover the 

 cause of this a hirge number of experiments were made on the magnetic moment of 

 needles of various sizes and shapes between weights of 0.0006 and 0.0030 milligrams, 

 and on combinations of them at various distances. In the course of this investigation 

 it was again shown (as had already been accepted here) that for magnets between the 

 limits of five and fifty times as long as thick, the magnetic moment is directly propor- 

 tional to the product of the weight by the ratio of the length to the diameter. This 

 relation is, and has been for some years, a fundamental consideration in the construc- 

 tion of galvanometer needles here. But though no error appeared in this assumption, it 

 was found that the other, that a great number of magnets could be combined Avithout 

 much lo.ss within a distance apart equal to their diameter, was quite unsound. AVhile 

 this is the case for two magnets, it is not so for a larger number. They can not 

 advantageously be placed at a distance apart less than 3 diameters without a consid- 

 erable loss of total magnetic moment. This is the case whether the needles are first 

 magnetized and then approached or magnetized in position. It is slightly better to 

 keep them 4 diameters apart. This condition prevented making needle systems such 

 as were proposed, but a greater we'ight of system seemed so very desirable that new 

 devices were discussed. Six needles in a group seemed to be about the limiting 

 numl)er which could be used with advantage, so that to get more than 12 needles to 

 a system retiuired more than 2 groups to put them in. It was proposed to use 16 

 coils instead of 4 in the galvanometer, thus allowing 8 groups of needles or 48 in all. 

 This involved a deliberate sacrifice of sensitiveness. Allowing for the several coun- 

 terbalancing effects, it was computed that the sensitiveness at a time of swing of one 

 second would be only about one-half or two-thirds as great for a 16-coil instrument 

 as for a 4-coil instrument of equal resistance. But it was confidently believed that 

 the steadiness of the former and its better capacity for use at higher times of swing 

 would more than compensate this loss. 



Accordingly a 16-coil galvanometer, with needle system containing 48 magnets and 

 weighing nearly 10 milligrams, was constructed. Its sensitiveness was found to be 

 about as computed, and its steadiness is so much superior to that formerly employed, 

 both as regards drift and tremor, that it has been used since December with most 

 marked advantage for holographic work. So entirely satisfactory has it proved that 

 another 16-coil galvanometer with needle system, weighing 0.012 grams, has been 

 constructed for use on the great suspension system of which I spoke last year. A 

 large number of separate needles are weighed, measured, and their magnetic 

 moments determined before the 48 are selected, so that no opportunity is allowed for 

 a few weak ones to lower the sensitiveness and injure the astaticism of the combi- 

 nation. The system is strongly magnetized after its completion by the aid of a 16- 

 coil electro-magnet. In the making up of these needles to form the system, a new 

 process of fastening the magnets has been devised by which results very superior to 

 any heretofore obtained here are reached. 



Further experiments on the exhaustion of the air from the galvanometer case have 

 been made. It was found extremely difficult to prevent leakage of air into the instru- 

 ment, and this when it occurred caused "drift." This difficulty has at length been 

 practically overcome. It had been supposed that a pressure of 1 millimeter or there- 

 abouts would be (juite as low as would be necessary with a 10-second single swing. 

 But even with the heavy needle system now in use, the damping became excessive at 

 2.5 secondssingle swingand 0.08 millimeter pressure. The pressure was reduced to0.20 

 millimeter, and next the time of single swing was raised to 5.5 seconds, before damp- 

 ing became excessive. The very promising result was reached that the deflection was 

 proportional to the .square of Ike time of .wing up to a time of single swing of above 5 

 seconds. Further exhaustion of the air proved impossible at the time, owing to the 

 vapor pressure of the wax and grease used about the galvanometer. But by a modi- 

 fication of the outer case, it is believed that this difficulty can be overcome, and that 



