164 Dr. E. H. Hall on the "Rotational Coefficient" 



Moreover this difference must have varied somewhat with the 

 strength of the direct current; so that the temperature read 

 can be assumed to give only a very rough indication of the 

 changes in temperature of the nickel. 



None of the numerical results of measurements made with 

 nickel before the disturbing influence of temperature was dis- 

 covered are here published. In some of the results afterwards 

 obtained, however, the effects of variations of temperature can 

 apparently be detected, as will be pointed out hereafter. 



The general method of experiment has been already suffi- 

 ciently described in previous papers. There will now be given 

 in tabular form the most important data involved in this ex- 



animation of nickel, and the values of -^ obtained. The abso- 

 lute strength of the primary current through the nickel strip 

 in any case is not given, as, by the method of experiment, 

 both the constant of the galvanometer used to measure this 

 current and the horizontal intensity of the earth's magnetism 

 at this galvanometer (this intensity being assumed to be con- 



,E / 

 stant during any one determination of ^A are eliminated from 



W . . 



the formula for -^. There will be given, however, the tan- 

 gents of the angles of deflection of the galvanometer-needle, in 

 order to show about what were the limits of variation of the 

 primary current. It may be well to state that this current 

 was what one Bunsen cell would send through — say, six or 

 eight ohms. It will be seen that there are variations of about 

 6 per cent, in tan a ; and the actual variations in the primary 

 current may possibly have been considerably greater than 

 this; for on March 11th, 12th, and 14th the galvanometer 

 stood in a different room from that in which it was placed for 

 the previous observations, and the horizontal intensity of the 

 earth's magnetism was probably somewhat different in the two 

 places. I have, however, as stated above, assumed that within 

 the limits of these variations the value of W is a linear func- 

 tion of the direct current. It is evident that no large error 

 can result from this assumption. 



It will be seen from the table that the experiments began 

 with the smallest values of the magnetizing force and went on 

 by stages to the highest. This is the proper course to follow 

 in order to avoid at any stage of the magnetizing force the 

 effects of a previous stronger magnetizing force. It must, 

 however, be stated that, before the series of experiments whose 

 results are here published was begun, the nickel had already 

 been several times subjected to a magnetizing force of about 

 7500, i. e. four or five times as great as the forces with which 



