: 
COMPLEX STRESS DISTRIBUTION IN ENGINEERING MATERIALS. 319 
In order to avoid this source of error, Professor Dalby has designed a new form of 
test-bar, described in a section of this Report, in which the specimen is turned from a 
solid bar with collars spaced at a convenient distance apart, and the extension between 
these latter is measured by a suitable mechanism bearing lightly against the inner 
faces of the projections. Although a plane model of this test-bar does not give the stress 
distribution accurately, it shows some interesting features which are probably present 
in the cylindrical form in a somewhat more pronounced fashion. A very small extent 
of collar is stressed in a symmetrical manner, and in the neighbourhood of this section 
there is a small field of complex stress symmetrical with respect to the collar, and rising 
to a maximum value near the join of the straight contours with the curved fillets, and 
resuming a mean stress value at a short distance away. This rise of stress is what 
one would expect from the results obtained with the enlarged ends of specimens 
described earlier, but it is found here that, provided the fillet does not undercut the 
collar, a variation in curvature makes very little difference to the stress distribution, 
and therefore is quite different in this respect from the case of an enlarged end of an 
ordinary test-piece. This result seems connected with the fact that, here, the change 
in the curvature of the lines of principal stress is practically determined by the thick- 
ness of the collar, and although they expand into this latter, their curvature and the 
stress concentration produced are not influenced much by the curvature of the contour 
at the junction of the collar with the test-piece. 
Compression Tests. 
It is well known that the end surfaces of blocks of material for compression tests 
require to be faced with great care, and for some materials, like concrete, a coating of 
plaster-of-Paris over the surfaces in contact with the pressure plates of the testing 
Fic. 63. Fic. 6B. 
Stress distribution at the ends Modified arrangement to secure 
of blocks of unequal sizes when pure compression stress in the 
pressed together. central block. 
_ machine is found to improve the consistency of the results of such tests. It is possibly 
.* 
q 
not realised, however, that, in general, when two flat surfaces are pressed together, the 
pressure between them may vary a great deal, even though they be of the same 
Material. If, for example, a block A (fig. 6A) is subjected to compression between two 
larger plates B and C, there is always a considerable rise of stress at the points D and E, 
and measurements show ® that in the upper block the curve of distribution, very close 
5 Appendix 4. 
