400 REPORTS ON THE STATE OF SCIENCE, ETC. 
about 24 per cent. These figures refer not to alternate bending of a Wéhler rotating 
specimen, but to alternate to-and-fro bending about a fixed axis. ‘The error intro- 
duced, by calculation on the usual assumption, of the skin stresses of a Wohler test 
will be less than the above, because the magnitude of the extra-elastic strain in 
the Woéhler test is less when the bending is to-and-fro about a fixed axis. The 
magnitude of the probable error of the calculated skin stresses in a Wohler test is 
very difficult to calculate, because the whole of the material outside some radial 
distance from the axis is impaired elastically. 
A deduction of the above percentage amount from the calculated (on usual 
assumption) ultimate skin stress of a fatigue test by alternating bending will, for 
steel of the mildness of the author’s specimens, bring the skin stress down to a value 
very little greater than the elastic limit determined by increasing the alternating 
bending upon a specimen previously unstrained. This would not be surprising, 
2/2 
rn) | 
8 
O28 Tans) » 
“ am, 
tf) > Ss 
t é'n0 
S 
e 
268 
Alternating Bending 
+6 Sold Specimen AIS 
Showing effet oF raunding off (a Re } 
nergh bouhood o R_ for fhe case of | 
E'= @ 
Range of Stress 
t2 
Half - Rage ef Strom xX fQ* a 
g z 4 6 8 (o Pe 1 
in view of the preceding results for torsion, and also for results of alternating direct — 
stress. , 
A point arises here with regard to the known effect of ‘ understressing ’ in fatigue — 
tests by bending. The author is inclined to believe, in view of the foregoing results, — 
that the extent of raising of a fatigue limit by ‘ understressing ’ may be overestimated 
in a bending test. 
Wher a range of torque or bending is imposed that will lead ultimately to fracture, — 
the stresses on the specimen at the first outset will be nearly those calculated by the 
usual formula, whereas the range of real stress at fracture will be considerably less — 
than the range of stress initially induced by that same constant range of bending — 
moment. Damage may be done by the higher initial stresses, and such damage i 
may be consummated in fracture at the lower stresses in the ultimate stage of the | 
test. Some carefully made Wohler tests, which gave lower fatigue ranges when the 
whole load was put on in a few seconds, would be explained by an initial ‘ overstressing 
effect’ of this kind. If the range of bending moment is imposed by increments up 
