284 W. H. WARREN AND 8. H. BARRACLOUGH. 
The general method of making a test was as follows :—the 
specimen having been placed in position in the bath, a slight load 
(approximately 200 tbs.) was applied in order to keep everything 
“taut” and in line. Then the extensometer or autographic 
apparatus was attached, as the case might require, and the bath 
was filled with oil to within a short distance of the top. By 
means of two gas kettle boilers the oil was heated to the required 
temperature—a process occupying from some twenty minutes for 
the lower temperatures, to three hours for the highest. It was 
found possible in the majority of the tests, by regulating the gas 
flames to maintain the temperature of the oil practically constant 
at any point, the variation not being more than one or two degrees 
on either side of the mean. The oil being thus maintained at the 
required temperature, the load was uniformly increased, and the 
readings of the stress-strain apparatus were taken at equal 
intervals of load. In those tests in which the specimens were 
actually broken, the gas was extinguished a little before the 
ultimate stress was reached to avoid the risk of some of the hot 
oil being thrown into the flame by the shock. 
3. Expansion of test piece by heat.—It was easily observable by 
the gradual raising or lowering of the “floating” lever of the testing 
machine, that during the heating of the bath the initial stress, 
referred to above, was markedly increased or diminished (in the 
compressive and tensile tests respectively) by the expansion of 
the test piece, and this more especially of course at the higher 
temperatures. A rough measure of the coefficient of linear expa” 
sion of copper was obtainable by observing the variation in the 
reading of the extensometer-lever, as the temperature rose. This 
could of course be only of the nature of a rough approximation, 
since the extensometer frames could not be kept at a constant 
temperature. No alteration in the total stress applied to the test 
piece is caused by this action; it will only modify slightly the 
rate and method of application of the stress at the lower loads. 
A slight error is introduced by the expansion, inasmuch as thé 
dimensions of the specimen when being tested at a high temper 
