100 THE ROYAL SOCIETY OF CANADA 



temperature variation obtained. It depends upon the speed of 

 calibration and the type of auxiliary circuit used. This temperature 

 correction reached as high as 0.11 ohms in one case. In the table 

 given below, these corrections have already been applied. 



After the successive deflections have been obtained, and corrected 

 for this temperature effect, it is possible to reduce the calculation of 

 the final corrections to a very simple arithmetical process. Table I 

 which is explained in detail below, illustrates this method, and Table II 

 gives a list of the corrections obtained in our case. a„ is employed to 

 101 

 2a„ 

 represent a„ TrTT m what follows. 



In Table I, n represents the slider reading. d„ is the deflection 



obtained for the reading n. k varied from to during the 



calibration, and kd„ represents the difference of the value of R n +i + 



R n+2 from that of F-~ . For ai we have ai= k~t\ — 5006= ' »„.. 



+ .56 = .82 in the case of this calibration. The remaining values of 

 a„ can then be obtained from the preceding column by means of the 

 equation a„ = kd„ — a n -\. a n is given by subtracting the value of 



101 101 



( i n ) from a n . i "is obviously the algebraic sum of the column 

 ^ 101 ' 101 



for a n divided by 101, and in the above case this amounted to — .12. 

 The figures of the next column are merely the additions up to n of the 

 preceding values. It will be noted that this forms the chief part of 

 the correction and for many purposes would give a sufficiently accu- 

 rate calibration. The values for a M+ i + a„+ 2 can be computed immedi- 



e 3 



ately from the column for a n . Now v n = — — r- (a„ + i+a„ +2 ) and 



in order to make it possible to calculate the figures more rapidly the 

 'next column is introduced in which — is added in each case to \ of 

 the corresponding value in the column for a M+ i + a M+ 2 



(g = ai + a 2 - 2000c = .54). 

 To get v n it is then only necessary to subtract the values in this last 

 column from the corresponding values in the preceding one. Finally 



it 

 to obtain the total correction, it is necessary to add v n to lOOOw 



n 



4 2 a„. 



1 



A comparison of the figures given in Table II with those obtained 

 by Prof. Callendar (loc. cit.) for the same instrument in 1894 shows 



