446 Proceedings of the Royal Society of Edinburgh. [Sess. 
1. Wires were fatigued in this manner for periods varying from five 
minutes to three and a half hours, and subsequently tested. The values got 
were as follows : — 
Time of 
Best Value 
Value of n. 
Calculated 
Fatigue. 
of a. 
Value of b. 
5 mins. 
100 
•78 
1000 
1 hour 
180 
•78 
834 
2\ hours 
Ol 
^2 55 
70 
'78 
560 
100 
'76 
484 
The value of n is seen to remain constant over this range of fatigue, 
whilst that of a oscillates about 100, the values of a when large being more 
variable than when small, since in trying to find the best value of a , values 
of the latter differing by ten or twenty units may make very little difference 
in the slope of the line, or the ability of the points to lie along it. b seems 
to be the only constant affected, although it was seen in the previous paper 
that the values of b varied greatly. There appears to be a gradual fall in 
value here, however. Rotational fatigue, then, has little or no effect on the 
constants of the equation. Further work will be directed to ascertaining if 
this fall in the value of b with increase of fatigue is real. 
2. Rotational strain on heated wires. It was shown also in the previous 
paper that after brass wire had been raised to a temperature above 375° C., 
then two values of n were obtained, one value extending for one range of 
oscillation, and another for the remaining part. It was shown that these 
values of: n increased to a constant value. Fatigue, again, is found to have 
no effect on the constants a and n, as the following results will show : — 
Time of 
Fatigue. 
Best Value 
of a. 
Value 
1st Part of 
Curve. 
s of n. 
2nd Part of 
Curve. 
0 mins. 
0 
2 '52 
3 '00 
5 ,, 
2 
2 '50 
3 '20 
10 „ 
1 
2'51 
3T0 
15 „ 
0 
2'53 
3'02 
30 „ 
0 
2'81 
3-00 
60 „ 
0 
*336 
3'03 
120 „ 
0 
2'56 
3'03 
180 „ 
2 
2'59 
3*07 
The behaviour of b after the wire has been heated and fatigued will be 
the object of further work. 
* This value, 3 '36, got for the line lying along the points of the first oscillations, is quite 
abnormal, and the wire used in this experiment was examined to discover, if possible, the 
