918 
accurate comparison, at a great number of temperatures, of the 
platinum-thermometer designed by Pty with the H, and He-ther- 
mometers, together with the reductions to the thermodynamical 
scale. Their figure 1, representing only the resistance above 56°, 
shows onee more the peculiar way in which the resistance decreases 
when the temperature is lowered. 
I started my computation from the corrected indications of the 
/T,-ihermometer, designed by 477, by the authors, and the corresponding 
resistances, both taken from their Table I. The absolute temperature 
of the freezing-point was assumed to be 273.09. In Table II I have 
inserted the values of 7’ and W that have been used *). 
In order to calculate the constants we put the formula thus: 
a al 
neen br! or dr en arden eend) 
Assuming a suitable value for a, we shall find from this formula 
four linear equations for a,b,c, and d by using four temperatures. 
To begin with, a value of w may be found graphically. If another 
value of w is assumed, this changes only the left-hand side of the 
four equations (5). Thus the corresponding change of the unknowns 
is easily surveyed, and w is finally chosen in such a way that the 
formula gives the best approximation at a few other temperatures. 
It was found undesirable to use more than one hydrogen tempe- 
rature in the four equations. On the other hand we cannot do 
without one such a low temperature if we want to determine c and 
d accurately. 
From the four temperatures 7’— 170.39, 90.27, 60.57 and 20.42, 
I found, taking w——1.26 2, the values of the unknowns given 
as “first approximation” in Table ILI. From these were computed the 
remaining discrepancies at the other temperatures. A second approxi- 
mation was next found from the same equations, changing the 
constant terms so as to get small deviations at the temperatures 
170, 90, 60, and 20, at which the first approximation naturally 
fitted exactly. Thus the deviations could easily be distributed more 
evenly. The resulting constants have again been inserted in Table II, 
whereas Table II gives under the heading W, the resistances cal- 
culated therefrom, the next column showing the deviation from 
experiment. 
1) The values at the last four temperatures in Table I have not been used, 
that at 77.91 was excluded because of an obvious printer’s error in the original 
paper. This error has been corrected here after the indication of Prof. Onnes. 
