1912-13.] Application of the Law of Torsional Oscillation. 185 
when cold. It was raised to a temperature of 530° C., and, on plotting, 
it was found that, with a = 0, an s-shaped curve was obtained ; but, with 
a = 20, all the points after the first five oscillations could be brought to 
lie on one straight line, the values being : 
a = 20 
n= 1-7 
6 = 826. 
The slope of the line has thus been practically doubled. It was found 
that no value of a could be got to cause the first five points to lie on 
one line. 
Platinum-Iridium . — Length = 30 cms. Diameter = 1 mm. Two lines 
were found in this case over a range of 90 oscillations, the values being : 
A. a = 40 B. a = 4=0 
n = ' 90 n — ‘ 96 
6 = 301 6 = 362. 
The doubling occurred at the value *66 for log y. These results agree 
fairly closely with those got for commercial platinum as described formerly, 
the values for this being : 
A. a = 28 B. a = 30. 
rc = - 95 n = 1*05 
6 = 341 6 = 429, 
and the doubling points occurred at a value of log y = *80. 
Pure Platinum . — Length = 30 cms. Diameter = 1 mm. A range of 
100 oscillations was observed in this case, and not only was one line 
insufficient to accommodate all the points, but even two lines could repre- 
sent only up to 50 oscillations, and a third line was necessary. These are 
shown in Diagram I., and the values obtained were : 
A. a = 25 B. a = 75 C. a = 75 
n= 1-75 n=‘ 60 n = '50 
6= 1136 6 = 275 6 = 243. 
Repeated experiments gave exactly similar values, the values of n being 
1*85, *60, and *50 respectively. This is the only wire yet experimented 
with which, over a range of a hundred oscillations, required three straight 
lines to cover the observed points. 
Assay Silver . — Length = 30*5 cms. Diameter = 1 mm. A change was 
