1910-1 l.J An Apparatus for inducing Fatigue in Wires. 447 
3. A length of brass wire, clamped vertically between C and D, and 
fatigued for two and a half hours, gave the values 
a = 90, tz=* 76, 6 = 841. 
These are the values got for the original wire, and thus extensional fatigue 
has no effect on the constants when the wire is unheated. This was 
confirmed by further experiments. 
4. After heating to a red heat and then cooling, extensional fatigue was 
found to have the same tendency as stretching had — i.e. to tend to straighten 
the curve so as to allow of one straight line covering all the points. A 
typical experiment gave the results 
a = 30, 71 = 1-24, 6 = 524. 
This wire was fatigued for one hour after having been raised to 485° C. 
It was found that the pendulum made twenty-five complete oscillations per 
minute ; therefore, since the wire between C and D receives two pulls during 
one oscillation, the total number of pulls given was 3000. The weight 
attached to the end H of the lever was 3000 gms., and the ratio of the arms 
3 to 1, the fulcrum I being half way between the spindle P and the end C 
further from H. The leverage exerted per pull is, therefore, in this case 
9000 gms. The values got on testing this wire after heating to 485° C., 
and before fatigue was applied, were 
a = 2, n = 2*15 and 2 # 70 ; 
a then is seen to be raised to a value intermediate to those obtained from 
the fresh and heated wires. Similarly n has now an intermediate value. 
The effects of both forms of fatigue on the empirical equation as applied 
to other materials will be the object of further research. 
cause of the unusual value of n. A repeated experiment gave the same result. The wire 
was then halved, and each half tested separately. Change of length has been shown to have 
no effect on the values of n or a. 
Each half tested gave values as follows : — 
(1) a=*5, 71 = 2-00 and 2 ’80 
(2) a='5, 71 = 2*04 
for first oscillations, and curve for rest, which could not be fitted with one straight line. 
This latter half was then again halved, and the two small parts tested. Again the two parts 
were found to give different results, and the abnormality was traced to one of the quarters. 
{Issued separately May 16, 1911.) 
