THROW-TESTING MACHINE FOR REVERSALS OF MEAN STRESS. 
297 
them to a number of reversals. The author refrains from publishing the results, 
which he considers not sufficient to establish anything definite, since they vary a 
great deal; but desires to mention that the effect of such annealing appeared in 
general to shorten the life of the specimen and not to restore it, as is usually 
supposed. 
As mentioned early in this paper, complete statical tests of specimens of size shown 
in fig. 3, could not be performed, since extensions could not be measured; moreover, 
as the 100-ton testing machine had to be used, the measurements were not too 
delicate. However, two specimens which had been subjected to reversals, and must 
have been nearly on the point of fracture, showed a distinctly greater maximum stress 
than the unused specimens. 
The author desires to thank Professor Osborne Reynolds, to whom he is indebted 
for many suggestions and much valuable advice and criticism during the progress of 
the work, and also for the facilities which he afforded for the carrying on of the 
experimental work. 
VOL. CXCIX.—A. 
