190 REPORTS ON THE STATE OF SCIENCE.—1913. 
certain effects of temperature. So far as the writer knows, there are no 
actual measurements giving a comparison of the actual amounts of hys- 
teresis, of either variety, at various temperatures. But it appears to be 
certain from experiments on torsional oscillations of wires that increase 
of temperature causes considerable increase of decrement of oscillations— 
2.e., increased loss of energy by increased elastic hysteresis. On the other 
hand, the effect of temperature on plastic hysteresis is complex (see article 
‘Recovery of Elasticity’ in this Report) ; the tendency is for decrease 
with higher temperature, owing presumably to increased potency of 
recovery by ‘healing’ together of displaced portions of crystals. 
These temperature effects are evidence of a difference of nature, 
and not merely of degree, between the two kinds of hysteresis. The 
question arises whether elastic hysteresis under cyclically applied stresses 
causes weakening or predisposition to plastic hysteresis. The suggestion 
of Bairstow (No. 4) that ‘ below the static yield-point, iron and steel appear 
to be capable of maintaining an unstable condition for a considerable time 
against cyclical variations of stress’ admits of a different and more simple 
explanation (see No. 66). Hopkinson, as already mentioned, found no sign 
of increase of (elastic) hysteresis with 250,000 repetitions of a range of 
stress of 28-6 tons per square inch ; and the results of experiments on resist- 
ance to alternating stress provide many instances of very long-continued 
cyclic stressing without fracture. Thus, 200 million revolutions in a 
rotating-bar machine with calculated stresses of + 40,000 lb. per square 
inch (No. 48) have been withstood without fracture by a steel specimen. 
It is interesting to know that the experiments of Hopkinson and 
Williams (No. 45) are being continued, with the general object of discover- 
ing how elastic hysteresis is related to the elastic limit. 
Speed Effect. 
The influence of high rate of alternation of stress is to increase the 
number of repetitions required for fracture, and apparently to increase the 
Wohler range (No. 43). It is pointed out in No. 43 that the range may not 
really be increased ; but that, on account of the large number of cycles 
required to fracture a specimen, the practical effect is virtually to increase 
the endurance either in range or number of cycles. 
Speed effect does not appear to become apparent at less than 2,000 
reversals per minute. (See Nos. 23, 43, 80, 65, 82, and 84; also No. 59.) 
The article on ‘Probable Causes of Speed Effect’ on p. 147 of 
No. 43 should be consulted ; and reference may be made to the article 
‘Recovery of Elasticity ’ in this Report. 
Divergent Results of Fatigue Tests.* 
Suggested Causes.—(1) Impurities (No. 3), flaws, &c. (No. 94), in- 
cipient cracks (No. 23) (such as would be left by a lathe cutting tool after a 
deep cut). The improved endurance of ground specimens and of specimens 
filed and polished, in alternate bending tests, is probably due to the re- 
moval of small surface cracks. J. B. Kommers (No. 51) states that polished 
and also ground specimens showed an increased resistance over turned 
specimens of 45 to 50 per cent. 
(2) Unrecognised stresses, due to bending in a direct stress ; to vibra- 
* See Note by Mr. E. M. Eden (p. 41). 
