COMPLEX STRESS DISTRIBUTIONS IN ENGINEERING MATERIALS. 197 
zontal after 10° alternations than the curves from solid rotating beam 
tests, and it is possible that in these tests the limiting stress /, has been 
nearly reached or that /,)8 may not be much lower than /,,°. Quite lately 
it has been shown that rotating cantilever tests with hollow test pieces 
where the stress should be nearly uniform over the cross-section also give 
a curve more nearly horizontal at 10° alternations than the curve from 
solid test pieces ;* in fact, /, appears to have been reached with 500,000 
alternations. 
These hollow cantilever tests help to explain the difference between 
the form of /, n curve obtained from the rotating bar and reciprocating 
mass types of testing machine, but unfortunately this is not the only 
difference in the results of tests on the two machines. 
Note on Divergent Results of Alternating Stress Tests. 
By EH. M. Even. 
In the reciprocating weight-testing machine, rate of alternation of 
stress, number of revolutions per minute, largely affect the endurance 
strength, whereas in a rotating-beam machine the endurance strength is 
quite unaffected by speed. 
Although, as far as I know, alternating stress tests of the same material 
in different testing machines have not been published, yet it appears to be 
impossible for the two types of machine just mentioned to give the same 
endurance figures, as if they did agree at one speed they would not do 
so at another. It appears that endurance under an alternating stress or 
resistance to an alternating stress cannot at present be determined for 
any material—the values obtained will depend on the testing machine that 
is used. 
In the Reynolds-Smith endurance tests with a reciprocating weight 
machine a high tenacity steel showed a lower endurance strength than a 
steel of much lower tenacity. Such a result has never been obtained with 
a rotating-beam machine where increase in endurance usually accompanies 
increase in tenacity, and this again rather points to some unexplained 
difference in the destructive action of the two types of machine. 
Apparently either the calculated stress in one or both types of testing 
machine is not the true stress, or something else besides the intensity of the 
alternating stress affects the endurance. 
In either case the calculated values of f, and /, are not the only factors 
affecting the endurance of a piece of material. 
There are some other factors besides /, and f, which are known to 
affect endurance; the distribution of the stress over the cross-section has 
been referred to before, and the condition of the surface, and the form of 
the test-piece, are also known to largely affect endurance, but none of these 
can explain the speed effect. 
In this connection it is, I think, worth noticing that it is not at all easy 
to repeat an endurance test and obtain exactly the same result. ‘Two test 
pieces cut from the same bar of metal will not usually show the same 
endurance when tested under what are intended to be the same conditions, 
on the same alternating stress-testing machine ; a great deal depends on 
the machine and on the care taken ; but in many published tests there is a 
great want of agreement between different tests of what is said to be the 
* “Welded Joints in Iron and Steel.’ Proc. I.C.H., vol. clxxxviii. 
