204 REPORTS ON THE STATE OF SCIENCE.—1914. 
material to an extent likely to produce permanent set, and they can be 
shown by comparatively thin specimens. Also they do not require any 
previous investigation of the co-efficient C for the given material, or of its 
dependence upon the wave-length. 
In theory observation of the isoclinic lines is sufficient to determine 
the stress system, provided we have information as to the actual stresses 
at a very limited number of points (7). Such information is generally 
available from the known boundary conditions. - 
On the other hand, the calculations required to actually deduce the 
stresses from the isoclinic lines are complicated, and are very difficult 
to apply to cases where the data are expressed by purely empirical curves. 
The isoclinic lines are, therefore, better suited to experimental verifica- 
tion of stress distribution already known from theory, and for which the 
theoretical isoclinic lines can be calculated beforehand and compared 
with observation. They have been so used by M. Corbino and Trabacchi 
(8) using rings of gelatine to verify Volterra’s (9) theory of internal strains 
in a multiply connected elastic solid; and also by Filon (10), who used 
glass beams to verify the ordinary theory of stresses in a beam at a distance 
from points of isolated loading, and also his own theory of the distribution 
of stress in a beam near a point of isolated loading. Both Corbino and 
Trabacchi, and Filon found that their experimental results confirmed the 
predictions of the theory of elasticity (11). Carus Wilson (12), who used 
in his investigation both the isoclinic and the isochromatic lines, was the 
first to apply the optical method to discover the laws of stress distribution 
in a glass beam, doubly supported and centrally loaded. 
He gives a drawing of the lines of principal stress in such a beam, but 
does not use them further, and restricts his comparison of theory with 
experiment, to the stresses in the cross-section immediately under the load ; 
the theory with which he compares his results was originally given by 
Boussinesq (13), and treats the height of the beam as infinitely thick. Sir 
G. G. Stokes gave, in a note to Carus Wilson’s paper, an empirical correc- 
tion to Boussinesq’s theory. An exact theory of this problem has since 
been given by Filon (14). 
The use of the isochromatic lines and generally of experiments de- 
pending upon tint has this advantage, that it yields directly the value 
of the stress-difference P—Q. If this be combined with a determination 
of the direction of principal stress at each point, then considerable direct 
information is given at once, and some cases of practical importance have 
been examined by Hénigsberg and Dimmer (15). 
The determination both of P—Q and of the directions of principal 
stress may be combined in one measurement, which is very simply made 
by means of an apparatus due to Coker (16). Coker uses a thin celluloid 
plate, cut to represent an engineering structure in which it is desired to 
investigate the stresses. This is a more easily worked material than glass, 
and a lesser thickness is required, as its stress-optical co-efficient is con- 
siderable. To obtain a measure of the stress-difference at any point a 
tension member is placed in front of the strained model, in a direction 
corresponding to one of the principal axes of stress, and the colour effect 
produced in the loaded model is neutralised by applying a sufficient load 
to this calibrating member. The tensional stress T affords a measure of 
the difference of the principal stresses (P—Q) subject to a small correction 
when (P—Q) and T have different signs. 
An improved way of doing this, which saves these repeated adjust- 
