356 Proceedings of the Royal Society of Edinburgh. [Sess. 
catch this point in the case of square plates than of circular. In the 
latter, set begins at the same instant and equally for each fibre; but 
in the former, it begins only in a few fibres and gradually spreads 
to the rest. 
3. By means of a beam of light reflected from a mirror connected to 
the plate by a system of multiplying levers. This was chiefly used for 
the detection of zero errors and for obtaining permanent set curves. 
Before these permanent set curves could be drawn with accuracy, many 
months of experimenting was undertaken. It was found that, if as much 
as a drop of oil escaped from any of the connections during an investigation, 
the no-load reading was affected ; should there occur an air-lock anywhere, 
there was a chance, during the working to and fro of the piston in the 
cylinder of the tester, that a little of this air could escape and thus affect 
the no-load result ; when a certain quantity of oil had been admitted into 
the cylinder, that quantity had to be kept constant even though the flexure 
of the plate admitted more and more oil under it, thus using up the 
working fluid; the hand- wheel for moving the piston had always to be 
brought back to the same position, in this case against the stops ; the 
weight-piston had to be turned to overcome static friction; in applying 
the weights to the weight-piston, the piston was allowed to rise slowly 
under small loads, and the final load reached gradually. 
For the optical apparatus, two vertical standards were fixed in the 
mild- steel block into which the recess was cut for the plates, and a cross- 
bar was fitted to them, so that it lay right across the plate and above it. 
A flat was milled on this and a tiny groove made. A knife-edge to the 
top of which a lever was fixed rested in the groove made. From one 
end of the lever depended a round-ended bar which rested exactly at the 
centre of the plate, and on the other end was fixed a small circular mirror. 
A ray of light from an arc lamp was allowed to fall on the mirror and 
to be reflected to a screen a considerable distance away. In this way 
a very slight deflection of the plate gave a large movement of the spot, 
and the variation in position of the no-load deflection reading at the 
centre was ascertained. All the causes of irregularity were discovered 
with this apparatus, because the movement of the spot could be watched 
on the screen. The manipulation of the arc lamp, however, gave trouble, 
and, as absolute readings could be taken with sufficient accuracy with the 
micrometer, the apparatus was not largely used for direct measurement. 
