292 



c. G. KXOTT. 



detîiched f"r<Jin the copper rods. AMiile the oljservatioii« were being 

 made, it was noticed that the loiu'th experiment was much inferior in 

 point of regularity and steadiness to the others, a fact sufficiently ex- 

 plained l)y the tinal condition of the met;d. 



It is not sin"])rising, tlien, that there is considerable divergence 

 Ijetween the \alues of the tem])erature coefficients a« obtained from the 

 earlier experiment in oil and the later series in the charcoal furnace. 



What is surprising is, that in spite of the great alteration in 

 structure going on in the sti-ij), the general behaviour of the cobalt 

 as show^n in the first three experiments is essentially the same. This is 

 well seen from the tabulation of the rates of change themselves. These 

 quantities were calculated by tlie same general method of interpola- 

 tion as was used in calculating the resistance. They correspond to 

 the values of dy/dx if y/ and x are taken to represent respectively the 

 corresponding resistances of cobalt and ])latinum. They are given 

 in Table IV, the first column containing the \alues of the platinum 

 resistances to which the tabulated rates of change correspond. 



Table IV. 



I have thought it sufficient to give the condensed numerical results 

 as contained in Tables I, Hand 1 \'. The indi\idual observations U])on 

 which these results are based are shown iiraphicallv in the diagram. 



