196 REPORTS ON THE STATE OF SCIENCE.—1913. 
constituent by the incipient fracture is noted.* Further, the fatigue of 
steels which had been variously heat-treated on systematic lines is simi- 
larly studied. This helps to throw light upon the overheating of steel &c. 
Remarks on the Phrase ‘ Crystallisation through Fatigue.’ 
From views which I have elsewhere expressed (No. 63a) as to the 
microscopic nature of strain effects it will doubtless be expected that I do 
not endorse the use of this hackneyed phrase. Twinning and the recrystal- 
lisation of polyhedric steels might, however, be regarded as admissions of 
the possibility of recrystallisation after straining, and therefore possibly 
after fatigue. But my view is that the expression arose through‘ the 
crystalline appearance which is well known upon the fracture of defective 
iron, and was later sometimes found on fractures of relatively brittle steel. 
I always find evidence that when such ‘crystalline ’ fractures are obtained 
they can also be obtained without fatigue; and, further, that metal which 
gives a fibrous or silky fracture does not develop ‘crystalline’ fractures 
by fatigue. 
Note on Stress Alternation Curves for Bending Tests on Rotating Bars. 
By EK. M. Even. 
Wohler’s rotating cantilever experiment showed that n (the number 
of rotations to fracture) depended on f (the maximum stress). For two 
materials ‘ Phoenix Iron’ and‘ Homogeneous Iron’ a fairly definite curve 
can be obtained by plotting the stress f against the number of alter- 
nations n, the curve extending in the case of the Phcenix Iron from 
n = 50,000 to n = 20,000,000, and in the case of the Homogeneous Iron 
from n = 3,000 to n = 4,000,000. 
The other materials experimented with by Wohler appeared to obey 
similar laws, but the results were much more irregular. 
Later rotating beam experiments than Wohler’s on steel, iron, and 
copper confirm the form of /, n curve given by these wrought-iron tests, 
but the more modern experiments have usually only carried the /, n curve 
up ton = 10% The form of the f, n curve suggests that there is a limiting 
value of the stress f below which fracture cannot be caused by any number 
of alternations; this limiting stress may be called f,. Values have been 
assigned to this limiting stress by Wéhler, but I cannot see any reason for 
thinking that the values he gives are correct. 
In practice, material does not have to withstand an indefinite number of 
alternations of stress, but the useful life of some machinery may involve 
some hundreds of millions of alternations of stress. In solid rotating 
beam tests the resistance of a material to 10° alternations (f,,8) would 
appear to be considerably lower than the resistance to 10® alternations 
(f108)- 
},, the limiting stress, is the value of f where the /, n curve is horizontal ; 
as far as I know it has not been reached in any solid rotating beam test. 
Alternating stress tests in direct tension and compression on recipro- 
cating weight-testing machines with a presumably uniform distribu- 
tion of stress over the cross-section of the test piece show f, n curves, 
which although in many cases rather vague in form are more nearly hori- 
* For a review of this question see 78, 
