438 
Proceedings of the Royal Society of Edinburgh. [Sess. 
assume values approaching these latter. Attempts were made to test this 
point, but no appreciable effect could be got, the wire resisting effective 
extension under the pulls that could be given it. 
Zinc Wire. 
This wire, of diameter 1*6 mms., was tested in its original condition, 
and found to obey the law over a range about one-sixth of that of the 
wires already mentioned. The oscillations in this case decreased very 
rapidly, and it w r as only with great difficulty and after many attempts 
that a satisfactory series of readings was obtained. The readings had to 
be taken consecutively, and after a dozen of these had been got, the oscilla- 
tions had so far decreased as to render further readings of remaining values 
of little accuracy. On graphing the values in the usual way, it was found 
that the first twelve points lay on a straight line, with the constants 
a = 0, n = *60, b = 419. 
After this point there was extreme inaccuracy, although there was a 
marked tendency to a decrease in the value of n. 
Silver Wire. 
Diameter of Wire used = 1T0 mms. 
In this case the law was found to hold over the usual range, and one 
straight line included all the points when a was taken equal to 65, the 
result being 
a = 65 , 7i — * 45 , 6 = 166 . 
Tin Wire. 
Diameter of Wire used = 1*07 mms. 
The lead ring used with the preceding wires elongated the tin wire to 
breaking point, and a light brass ring, to the bottom of which a thin brass 
tube was attached, was substituted. Even with this oscillation elongation 
occurred, a wire of 1 foot increasing by '5 cm. in 24 hours. As in the 
case of zinc, the oscillations died down with extreme rapidity, and great 
difficulty was experienced in getting readings. The readings had to be 
taken at the end of every oscillation, and, as with zinc, only a dozen readings 
could be got with any degree of accuracy. Here, again, the points are found 
to lie on a single line until the oscillations had become so small as to 
prohibit exact readings. Again the tendency was noted for n to diminish 
