COMPLEX STRESS DISTRIBUTIONS IN ENGINEERING MATERIALS. 297 
notch is decreased to ;'s cm., the maximum stress rises to nearly six times the 
mean stress, as fig. 4 shows, while if the radius is very large, say 10 mm., the 
maximum stress is approximately 2.2 times the mean stress. 
It seems evident from these experiments and others described below that 
an impact test on a notched specimen affords a valuable means of discriminating 
between ductile and brittle material, for whatever be the radius of curvature 
at the apex of the notch in a ductile material, it acts primarily as an indicator 
of the place where fracture is to commence, but once plastic stress begins, its 
influence in maintaining the high stress concentration observed here must recede 
in importance, although it is quite possible its form may be a factor in the 
final result. In hard and brittle materials, however, the radius at the bottom 
of the notch is apparently a dominant factor since the stress distributions 
Fig.5a.STRESS DISTRIBUTION AT THE CONTOUR 
OF A NOTCH OF THE CHARPY TYPE. 
——J000 
: SS 
X 00 
wae = 
~ ~7-- 3800 
~y4 £ 
— —W000-— 
mS SE SEE RCE AON, aoe aes new e SHH Se 
Fig.5b. PRINCIPAL STRESSES AT THE 
SECTION C.D. OF FIG.5a. 
15 74 
(6976.£.) 
observed here can suffer little change in type throughout, and when due allow- 
ance is taken of static breaking stress in tension, such materials must have 
a comparatively low impact value on account of the maintenance of the 
relatively high stress concentration. 
As a discriminating test of quality an impact test may possibly be too 
severe, and good material may be rejected by it, since the radius of curvature 
of the notch is a factor of variable importance according as the material is or 
is not ductile. In very hard materials there seems good evidence for bringing 
this factor into account in grading materials by impact tests. 
This is more particularly emphasised when tests under different conditions 
of notch form are considered, as, for example, in one of the Charpy type, in 
which all the conditions of experiment are exactly the same as before, with a 
notch 1 cm. deep rounded off at the apex by an are of 7% cm. radius, 
(fig. 5a). Measuring the general accuracy of the experimental values by a 
comparison of the average stress intensity of 282 Ib. per sq. in. recorded by 
hee machine with the integrated stress intensity curve AB of fig. 5s, 
x 
