238 PROFESSOR E. G. COKER ON 
This differed somewhat from the brass tube previously experimented upon, in showing 
a practically uniform coefficient of expansion under every load below and above the yield 
point (fig. 4). The mean value of the coefficient was found to be 00001121, the maximum 
deviation being for the two lowest loads, and amounting to nearly 2°4 per cent., while 
for the remaining loads the deviation was less than 1 per cent. The general accuracy 
of the results was checked by comparison with the known values of the coefticient of 
expansion at atmospheric pressure, and the agreement is sufficiently elose to make it 
clear that the observations were accurately taken, having regard to the fact that the 
seamless tubes experimented upon had been subjected to exceptional treatment in the 
process of manufacture, and were probably in a very different physical condition from 
solid bars of the same material rolled or cast in the ordinary way. | 
We may conclude from these experiments that there is practically no difference ‘in 
the linear expansion of brass and steel within the range of stress up to the yield point 
of the material, and that for brass there is probably an increase in the coefficient beyond 
the yield point, but that there is no increase for steel. In this part of the work I was 
greatly assisted by Mr Cuartes M‘Kercow, Demonstrator in the Civil Engineering 
Department, M‘Gill University, who kindly undertook the major part of the work of 
observation, and who also rendered me very able assistance in the experimental work 
detailed below. 
4. Toe Benaviour or [RON AND STEEL UNDER TENSILE STRESS. 
In applying the thermal method of measuring stress, the most convenient arrange- 
ment of apparatus consists of a thermal junction or pile of the necessary delicacy con- 
nected to a galvanometer giving a sufticiently wide range of readings for the small 
difference of temperature produced. In nearly all testing laboratories the presence of 
iron in large quantities makes it necessary to choose a galvanometer which is not 
influenced by the proximity of iron, and this was especially necessary in the present 
case, since in the M‘Gill University testing laboratory the main testing machines have, 
in the course of time, become magnetised, owing to the subsidiary mechanism being 
operated by electric motors. These difficulties are easily overcome by the use of a 
D’Arsonval galvanometer, the field of which is very uniform, even in the neighbourhood 
of large masses of feebly magnetised iron. The galvanometer coil was specially wound 
for me by Dr Tory, and was of approximately the same resistance as the thermopile 
used in the majority of the experiments, so that the arrangement was as sensitive as 
possible. ‘To avoid short-circuiting of the pile when in contact with the metal specimen 
under test, it is convenient to insulate it therefrom, and I have found a thin sheet of 
paper, as suggested and used by Jouve, the most convenient. The connections of the 
galvanometer to the thermopile were made by soldered joints, which were afterwards 
wrapped in paper to insulate them from one another, and then tied together and lagged 
with cotton-wool. 
