1 44 PROF. E. G. COKER AND MR. K. C. CHAKKO: THE STRESS-STRAIN PROPERTIES 
The result of breaking the first three specimens showed that the extensometer 
arrangement was defective. The tiny indentations at the sides, where the extenso¬ 
meter was attached, so weakened the specimens that they all broke at one or other of 
these sections. The thinnest one naturally was affected most, so much so that it 
fractured with very little extension. This defect was partly corrected by cementing 
small fillets on to the specimen from which the extensometer clips were supported, but 
in spite of this some of the specimens fractured outside the gauge limits, showing that 
even in a ductile material the effects of enlarged ends prevents equalisation of stress 
near the change of section under any condition of load. 
Stress Optical Determination .—-It was originally intended to study the stress- 
optical properties of nitro-cellulose by analysing the light which traversed the 
material by means of a spectroscope ; but the necessary apparatus was rather 
difficult to procure, and it was convenient therefore to commence with a standard 
nitro-cellulose beam and use this for comparison with the optical phenomena observed 
in tension. The methods adopted here proved to be exceedingly well adapted for 
measurement of stress distribution beyond the elastic limit and are likely to be of 
great use hereafter. The comparison beam used is of rectangular section and is 
subjected to pure bending moment of known amount, and the stress at any point can 
therefore be calculated from the formula f = without appreciable 
error. 
It is generally assumed that the relative retardation of the polarised rays in a piece 
of optical material under moderate stress is proportional to the difference of principal 
stresses at the point, but this may not be correct and cannot be assumed to hold without 
experimental proof. Hence the stresses in the comparison beam are restricted to small 
values, so that the limit of proportionality of stress to strain is not passed in order to 
give an opportunity of examining the possibility of the law following a linear strain 
function or possibly some more complex variable. In order to make the retardation 
in such a beam sufficiently great to balance the retardation in the highly stressed 
specimens, the thickness of the beam should be large. This is most conveniently 
obtained by placing the several beams side by side, with their ends clamped and pinned 
together, as shown in fig. 3 in which several beams are so fastened together by plates 
A, to which extension levers B are also attached for supporting loads C depending 
from hangers D. This compound beam is supported on knife edges two inches apart, 
and when loaded has its central section sufficiently removed from the supports to give 
pure bending moment at the central section. The material of the beams is almost 
perfectly elastic up to and probably beyond 1600 lb./in. 2 , but they are actually not 
stressed to more than 1300 lb./in. 2 . In some cases as many as eight beams ^ inch 
thick are used in this way, and a strong beam of light is then necessary to enable a 
comparison to be made with the tension member under observation. A carbon arc is 
then used as the source of light, but when only two or three thicknesses are 
employed the light from a Nernst lamp is sufficient, but in all cases the images are 
