Precision Method of uniting Optical Glass. 281 



through the furnace, and examined after passing through the 

 piece of glass or silica under test, completed the apparatus. 

 By this means the effect of unequal heating caused by rise or 

 fall of temperature could be observed, and if necessary the 

 actual stresses set up in any part of the test piece could be 

 calculated at any time during or after the experiment. 



Three plates, all of a hard crown glass, were put into 

 optical contact, the centre one being about 10 mm. thick, 

 and drilled so as to give a cell suitable for the comparison of 

 two liquids, as in the Rayleigh interference-refractometer. 

 The outer plates were, of course, merely windows of about 

 3 mm. thickness. The cell was heated without pressure in 

 the electric furnace at the rate of 2° C. per minute to 460° C, 

 and after two hours at this temperature it was cooled at 

 4°-5° G. per minute ; the arrangement of the furnace per- 

 mitted continuous inspection of the cell. During heating, at 

 200°-300° C, minute imperfections in the contact grew 

 rapidly, one side of the cell coming almost completely away. 

 When cold it was found to be out of contact over parts of the 

 surface, the rest of the surface being in contact, but covered 

 with spots. 



Rapid temperature-changes, such as would easily have 

 separated glass in ordinary contact, had no effect ; water also 

 failed to lift the surfaces apart ; and when a thin knife-blade 

 was forced in from a corner where the two surfaces were 

 apart, the portion of the window out of contact was broken off 

 without moving the window as a whole. The cell was then 

 compressed in a vice in such a way as to apply large 

 shearing stresses at the interfaces, with the result that 

 corners of the windows out of contact broke away, while 

 the rest, in contact, remained unchanged. The cell then 

 appeared as shown (tig. 5, PI. V.). 



The spots referred to were small round areas where the 

 surfaces were just out of contact. Each developed from a 

 minute flaw in the optical contact of the surfaces, and when 

 examined microscopically very minute black particles could 

 be seen in each, and from the system of Newton's rings 

 apparent in them during the heating they appeared to contain 

 a gas at high pressure — by the same means the pressure was 

 observed to diminish slowly after a spot had spread outwards 

 to one edge of the cell ; and, finally (in a later experiment), 

 when the temperature was raised above the annealing-point, 

 the walls of each spot were slightly but definitely blown out, 

 bo as to appear convex from the outside. It therefore 

 appears that the spots are due to the decomposition of minute 



