COMPLEX STRESS DISTRIBUTIONS IN ENGINEERING MATERIALS. 411 
this line of reasoning must be seriously at fault. Alteration of the mechanical 
properties of the wire, such as raising the yield-point stress, can hardly be assumed, 
because load-extension diagrams for the wire, taken on a specially built machine, 
showed little difference between wire when it was new and after it had been bent 
over a pulley many times. 
The usual wrapping test for wire demands that it shall be closely wound round 
itself and be re-straightened without failure. If the radial diameter is unchanged 
the wire must elongate on the outside and shorten on the inside 50 per cent. lf a 
bar of steel be taken large enough to be marked off by circumferential lines and others 
8 
oe 
: 
~~ 
: 
+0:007— O0048 
< Strain in bent | Strain in bent wire = Strain in restraighlened 
wire under no tension. at 50/bs tension wire under 5O0/bs ténsion. 
Fig. 24 
parallel to the axis, after it is bent it can be seen that changes of length of this order 
actually occur, but the percentage elongation at fracture of such steel in tension 
is probably about sixteen times that of this wire, for which the value is about 1-6. 
In a recent paper (Lngineering, June 15, 1923) Sir Alex. Kennedy shows that 
when metals are tested by bending to yield the calculated stresses are greater than 
those obtained from tension tests, particularly when the region of maximum stress 
is reduced. He offers alternative suggestions, either that the maximum stress under 
bending actually exceeds the tenacity of the material, or that the accepted formule 
must be radically modified to allow for an altogether different distribution of stress 
intensity. 
The further results given here appear to indicate that the boundary of an 
unexplored region has been reached, that in which the,strains vary from point to 
point, and are not entirely elastic. 
