304 REPORTS ON THE STATE OF SCIENCE, ETC. 
number of tests made, that measurement of the cyclic strains furnishes a check 
on the accuracy of the applied stresses as well as on the normality of the 
specimen. The columns headed ‘ Preliminary Deadweight Test’ in Table IV. 
are inserted so that this and other comparisons may be made. 
It is probable that the differences of shear stress 
4P—a and Q—a 
would have been rather larger if the number of cycles at the stresses giving 
fracture in the combined tests had been as large as the number of cycles in 
these separate tests. It would appear, then, that there is strong prima-facie 
evidence for believing that for mild steel in combined tests with phase angle 90° 
and with p,=2g, the limiting range of stress is less than the limiting ranges 
for separate tests by about 10 per cent. 
With regard to the cyclic strains (phase difference 90°) it will be observed 
that the percentage increases of the component bending and twisting strains 
measuved in the combined tests are practically the same as those measured for 
the same stresses in the separate tests; in other words, the total (i.c. elastic 
+ non-elastic) strains pertaining to the bending and twisting applied simul- 
taneously are practically the same as if cycles of the same ranges of stress had 
been applied separately. But the ranges of stress causing fracture in combined 
tests are about 10 per cent. less than the ranges that would have produced 
fracture had the bending and twisting been applied separately; thus the com- 
ponent ranges of strain at stress ranges approaching fracture are, as seen in 
Tables II. and III., much less in combined than in separate tests. Comparing 
solid specimens, the percentage decrease of total strain comes out to be 36 per 
cent. for the torsional strain, and 20 per cent. for the bending strain; and com- 
paring hollow specimens, the respective figures are approximately the same, viz. 
34 and 24. To obtain these figures, A6 and Al3 were compared respectively 
with Al4, A16, B22, B27,3 and A7, All, A15.% 
Remarks on Component Strains in Tests with Phase Angle 90°. 
An interesting point observed in the tests with phase difference 90° was the 
following. After a run under a pair of simultaneous stresses, during which 
there was cyclic non-elastic strain, an increase of one of the pair of stresses, 
say the bending stress, caused a slight contraction in the range of torsional 
strain, and vice versa. The test of Specimen B24 was specially directed towards 
the study of this effect. To obtain an enlarged effect, one of the simultaneous 
stresses was made zero and a run of several thousand cycles given. Then, 
without stopping or alteration of speed, the other stress was reimposed. 
For example, in course of test of B24, 
(a) With range of direct stress due to bending= 413-4 tons per sq. in. 
and range of torque zero, the range of bending strain increased 
from 7°35 to 7:58 cm.4 during 40,000 cycles. On adding torque to 
give +5°30 tons per sq. in. of shear stress and retaining the same 
range of bending moment, the bending strain decreased to, and 
remained at, 7°33 cm. during 12,000 cycles. ° 
(6) Again, with range of shear stress due to torsion=+6°70 tons/[J’” 
and range of bending moment zero, the range of torsional strain 
increased from 10°9 to 13°5 cm. during a run of 108,000 cycles. 
On adding bending moment to give +1195 tons per sq. in. and 
retaining the same torque, the range of torsional strain decreased to, — 
and remained at, 11°80 cm. during 25,000 cycles. 
This effect is not due to friction occasioned by the extra loading on the 
parts of the machine when the additional stress is put on. In case (b) the effect 
of friction would be the reverse of that observed; and while in (a) the effect of 
friction would be similar to that observed, the magnitude of the friction (of 
ball bearings) would not be sufficient to account for the decrease. Moreover, 
3 See ‘ Alternating Stress Experiments’ (Proc. Inst. Mech. Eng., January- 
May, 1917) for record of these tests. 
4 On image of scale reflected from mirrors. 
