8 
end to the fixed slab by heavy weights. The other end of 
the plate of rock, projecting over, but not at first in contact 
with the moveable slab of wood, was lightly weighted and 
allowed time for bending. The angle of sudden fracture 
could be obtained by setting the machine at a low angle and 
forcibly bending down the lamina of rock until it touched 
both surfaces. If it yielded thus far, the angle was slightly 
increased and the experiment was repeated. 
This apparatus had some advantages but many defects. 
The most serious was that the pressure exerted at any time 
was difficult to estimate. A weight placed upon a surface 
of gradually increasing inclination exerts a diminishing 
pressure which changes appreciably even at angles of 3° or 4°. 
It was difficult to read the small deflections obtained with 
any accuracy, and the apparatus was liable to disturbance 
and accident. 
At length I tried the machine represented in the Plate.* 
Here a thin plate of limestone or other rock is screwed down 
to a block / travelling in a horizontal groove h. Upon this 
descends a vertical plate c, terminating in a hinged knife- 
edge e. We get in this way pressure applied always to the 
same line upon the lamina of rock, for when deflection begins 
the knife-edge inclines forwards out of the perpendicular at 
its lower edge without sliding over the rock specimen. By 
pushing the block / along the groove h perpendicularity is 
restored. An index m connected with e makes any devia- 
tion more apparent. When a piece of rock is to be tested, 
shot is poured into the cylinder a, which is in direct com* 
munication with the vertical plate c, and the pressure is taken 
by a steelyard or balance. The index I is set at 2ero by the 
screw if and its motion along the graduated scale enables the 
observer to record with precision a deflection of less than 
* For the construction of tins machine, and for many useful suggestions, I 
am indebted to Mr. Thomas Prince, of Bradford. 
