492 TRANSACTIONS OF SECTION G. 
much data for guidance in future constructional work has been obtained, 
especially in the United States of America. 
The still more difficult problems involved in the determination of the stresses 
in joints and fastenings of complicated structures have often been investigated 
by purely mechanical measurements of strain, and the experimental investigations 
of Professors Barraclough and Gibson and their pupils upon the distribution of 
stress due to riveted joints and curved plates of boiler shells afford a notable 
example of the successful application of the measurement of small strains to a 
stress problem of great complexity. 
That ‘ science is measurement’ is here sufficiently obvious, and it seems only 
due to the memory of that great engineer, Sir Joseph Whitworth, to refer to his 
great mechanical achievements of a true plane and well-nigh perfect screw, 
which enabled him to measure changes of one-millionth of an inch, and thereby 
gave experimental investigations of strains a new impetus, which is reflected in 
subsequent work on the subject. Nor must we forget the no less important 
exposition, by Kelvin and Tait, of the scientific principles of instrument construc- 
tion which have done so much for the design of instruments for the precise 
measurement of strains. 
Mechanical measurements cannot, however, completely satisfy all our modern 
requirements, since they are essentially average values, and fail to accommodate 
themselves to many of the problems which press for solution. 
In the quest for exact experimental knowledge, the measurement of stress at 
a point becomes of paramount importance, and we may, therefore, inquire what 
further means the researches of pure science have placed at our disposal for the 
determination of stress distribution in materials. 
It is well known that many materials when tested to destruction show a 
considerable rise of temperature at the place of fracture, especially in very ductile 
materials; but Weber was the first to discover that a metal wire when stretched 
within the elastic limit is cooled by the action of the load, and this result was 
deducted later from the laws of thermo-elastic behaviour of materials by Lord 
Kelvin, who showed that tension and compression loads produce opposite effects, 
and that materials which have the property of contracting with rise of tempera- 
ture show thermal effects of the reverse kind. Although the changes of tem- 
perature produced by stress are small within the elastic range—less than 1° C. 
for most materials—yet their effect upon a thermo-couple is readily measurable 
if the equilibrating effects of surrounding bodies are neutralised or allowed for, 
so that stress distribution can be determined by thermal measurements at a point. 
The correction for such disturbing causes is usually an important factor, and is 
generally so large that experimental work is more suitable for the laboratory than 
the workshop; but if all necessary precautions are taken a linear relation of stress 
to strain can be shown to hold up to the elastic limit of the material, while above 
this point the break-down of the structure causes a rise of temperature of so 
marked a character that it has been utilised by several investigators as an indi- 
cation of the yield point. 
Experiments, upon members subjected to tension, compression, and bending, 
show that thermal phenomena afford trustworthy indications of the stress in 
materials so diverse as a rolled-steel section, a block of cement, and beams of 
stone and slate. Although no attempt appears to have been made to investigate 
stress distributions of any great complexity, it seems not unlikely that thermal 
methods of investigation will ultimately prove of considerable value. 
The transparency of metals to Réntgen rays is another phenomenon which 
has often been suggested as likely to be of service for work on stress distribution 
in materials, and Mr. Howgrave Graham and I have examined a number of 
rolled metals under stress up to the breaking point, without, however, discovering 
any change in the appearance of the material as seen on a fluorescent screen. 
Although our experiments showed no perceptible change, it is, of course, not 
impossible that an effect may have escaped our notice. 
Another and still more fascinating field of research on stress distribution is 
afforded by the doubly refractive properties of transparent bodies under stress, 
a discovery made by Sir David Brewster almost exactly one hundred years 
ago, and but rarely made use of since by engineers, although Brewster himself 
immediately saw its value for experimental purposes, and suggested that models 
