THROW-TESTING MACHINE TOR REVERSALS OF MEAN STRESS. 
■285 
5 to 11 were put into the machine and run for 30 minutes each—about half the 
length of time required to break Specimens 1, 2, and 3. Specimen 5 was carefully 
put away for three days, after which the test was completed by putting it in the 
machine and running it till rupture took place. The other specimens were treated in 
a similar manner, Specimens 6, 7, and 8 resting for five days, Specimens 8 and 9 for 
eleven days, and Specimens 10 and 11 for four months. The result for Specimen 4 is 
irregular, and is therefore rejected. 
The second set of specimens was treated at a later time. They were annealed 
before testing. All the specimens of this set had the dimensions shown in fig. 4, 
which was adopted as the standard size in all succeeding tests. 
These results show that, if a specimen is allowed to rest when the test is half 
completed, there is no appreciable recovery if the period of rest is for a few days 
only. They suggest that if the period of rest extends over some months the 
specimens may or may not recover slightly; the extent would appear to depend on 
the treatment which it has received previous to the test. 
To settle definitely the restoring effect of a long period of rest, a great many more 
experiments would have to be done, but as far as this work is concerned where the 
specimens were seldom allowed to rest for more than two days, the effect of this rest 
on the total reversals for rupture is negligible. 
Relation of Limiting Range to Periodicity of Reversals. 
Under this head are given the results of experiments to determine the variation of 
the range of stress with speed when the number of reversals for rupture is constant, 
viz., one million. 
Six bars of mild steel were purchased together, and the whole of the specimens 
for which the results given in Tables II. and III. were obtained were cut from these 
bars. Six samples, each 18 inches long, were cut, one from each bar, and were tested 
for statical breaking-stress, &c., in the Owens College Laboratory Testing Machine, 
Avliich is a Buckton 100-ton machine of the Wicksteed horizontal lever type ; the 
extensions were measured over 8 inches. The figures obtained in these tests were 
as follow :— 
Yield-stress. 
Maximum 
stress. 
Breaking- 
stress. 
Percentage 
elongation at 
maximum 
stress. 
Percentage 
elongation at 
rupture. 
Maximum . 
17-44 
24-70 
21-08 
24-6 
31-5 
Minimum . 
16-81 
22-93 
19-34 
23-0 
29 
Mean . . . 
17-12 
24-54 
20-47 
23-5 
30 
Three annealed specimens of the form used in the endurance tests (dimensions 
according to fig. 4) were also broken in the same testing machine, and the maximum 
