308 Proceedings of Royal Society of Edinburgh . [sess. 
the cobalt, as shown in the first three experiments, is essentially 
the same. This is well shown by tabulating the rates of change 
themselves. These quantities were calculated from the observa- 
tions by the same general method of interpolation as was used in 
calculating the numbers of Table I. They are given in Table IV., 
of which the first column contains the platinum resistances to 
which the tabulated rates of change correspond. 
Table IV. 
Platinum 
Resistance (or 
Rates of Change of Cobalt Resistance per Unit Change of 
Platinum Resistance. 
Temperature). 
Experiment 
Experiment 
Experiment 
Experiment 
I. 
II. 
III. 
IV. 
2*0 
7*02 
7-30 
10-33 
9-15 
1*8 
6-19 
7-24 
6-74 
5-09 
1-6 
5-45 
5-57 
6-63 
6-10 
1-4 
3-76 
3-58 
3-65 
3-66 
1-2 
3-58 
3*23 
2-78 
I have thought it sufficient to give the condensed numerical 
results as contained in Tables I., II., and IV. The individual ob- 
servations upon which these results are based are entered graphi- 
cally in the diagram. Curves 1 and 2 have already been mentioned. 
They show the march of resistance with temperature as measured 
on a mercurial centigrade thermometer. In curve 3, the platinum 
resistances are the abscissae, and the ordinates are the corresponding 
cohalt resistances. The points belonging to the various experiments 
are distinguished by special mark. 
In one particular, cohalt resembles iron and nickel in its behaviour. 
There is a rapid increase in the steepness of the curve at higher 
temperatures. In iron and nickel, however, this rapid increase is 
followed at still higher temperatures by a distinct decrease, the 
curves bending so as to present a concavity towards the temperature 
axis. Neither Table IV. nor the curves give any hint of such a 
tendency in cobalt. It will be seen that Experiments I. and II. 
are in fair agreement throughout; and that all four experiments 
point to the existence of a critical temperature at which the re- 
