188 REPORTS ON THE STATE OF SCIENCE.—1913. 
gradually ceases to increase, and the ‘ cyclical permanent set’ gradually tends 
to disappear ; with such disappearance the elastic limits become adjusted 
to the stresses, and the material recovers its elasticity. Recovery during 
cycles of equal + stress was observed by Bairstow (No. 4), the width of 
the hysteresis loop being seen to decrease with repetitions of the same 
stresses. Recovery under equal stress alternations is usually masked in 
fatigue tests by the circumstance that the primitive elastic limits are 
further apart than the adjusted limits ; but some tests of Rogers (No, 62) 
of heat-treated steel appear to show adjusted limits higher than those found 
by static tests on the same treated material. 
Recovery during repetitions of stress is difficult to explain. The heal- 
ing up, which occurs with rest after a single overstrain, is not by itself a 
sufficient explanation, for the stresses succeed each other too rapidly in 
alternating stress tests for any material healing up to take place in any one 
cycle. It has been shown (No. 82) that adjustment of elastic limits (and 
therefore recovery) occurs not only with slow repetitions of two cycles per 
minute, but also with 800 cycles per minute ; and, of course, the existence 
of safe ranges of stress, one of whose limits may be outside the primitive 
elastic limits, is a fact known since the time of Wohler. It may be con- 
jectured that recovery during cyclical stressing is a slow continuous action 
due to change in the material between slipping cleavage planes. The 
slowness of this action, in tests at laboratory or workshop temperatures, 
still obtains at higher temperatures ; butit appears from the experiments of 
Unwin* (No. 91) ‘and Howard (No. 47) that the resistance to fatigue was 
somewhat greater at 400° to 500° F. It may be noticed that the energy 
corresponding to the hysteresis loop, which may cause considerable rise of 
temperature of the test piece, is generated at the slipping cleavage surfaces, 
which is the very locality where increased temperature will have its 
greatest effect. 
The small increase of resistance to fatigue mentioned above may 
possibly result either from a tendency to create a more extended elastic 
range (due to recovery at the higher temperature, in which case the 
adjusted elastic limits would be further apart) ; or merely froma greater 
healing tendency counteracting the disintegrating action of the to and 
fro slipping, but not leading to any extension of the elastic range ; or from 
both these two together : the three suggested alternatives being, of course, 
different aspects of the same thing. ‘Lhe first of the three seems impro- 
bable from some experiments of Bairstow (No. 4) on alternate boiling and 
overstrain of a specimen previously subjected to repetitions of stress in his 
machine. The considerations concerning temperature and recovery in 
static tests, also, are in accordance with this view. In short, it seems 
probable, though not quite certain, that for large limits of temperature 
the rapidity of (or rather slowness), or degree of tendency to, recovery 
is somewhat affected, but not the extent of the elastic ranges of iron and 
steel. 
It should be noticed that Howard in a further paper (No. 48) found the 
resistance to fatigue much increased for tests carried out at 400° to 600° F. 
Whether the increase of 100° F. between the experiments of Howard’s 
papers No. 47 and No. 48 corresponds to some critical change in tempera- 
* Unwin attributes the increased resistance rather ‘ to the annealing effect each 
evening when the bars were left to cool.’ 
