NORMAL SULPHATES OF POTASSIUIM, RUBIDIUM, AND CESIUM. 
467 
adopted in arranging the interference apparatus of the dilatonieter, the second of the 
methods described in the memoir concerning the latter and which is illustrated 1)y a 
special figure in the German translation of that memoir contributed to tlie ‘ Zeitschrift 
fur Krystallographie ’ (30, 530), there was no necessity to polish the surfaces of tlie 
crystal-blocks. For a crystal surface is not required to act as the lower reflecting 
surface involved in the generation of the interference bands, the upper surface of the 
compensator performing that function. It was therefore only necessary to complete 
the block by preparing a similar j^arallel surface in the same manner, separated from 
the first one by as much thickness of crystal as the particular specimen admitted of. 
The crystal-block was then cleaned from crystal dust by washing in benzene, dried 
with a clean linen cloth free from fluff, and stored in a desiccator until required for 
the observations. The thickness of the blocks employed varied, as will subsequently 
be seen from the record of the accurate measurements, from 4‘8 to 10‘7 millims., the 
former limit being in the case of the only crystal under 5 millims. in thickness. The 
great majority were from 7 to 9 millims. thick. 
In two or three cases, although only benzene had been used in the treatment of the 
crystals, the observations of expansion were vitiated by the oozing of minute traces 
of liquid, wliich proved to be mother-liquor, between the surfaces of the platinum- 
iridium tripod table and the crystal, or between the latter and the compensator. 
For in most cases the three point method of contact was impossible, owing to the 
prepared crystal surfaces being narrower in one direction than in the other, too 
narrow to take the third point but not too narrow for stable equilibrium of both 
crystal and compensator. A comparison of the results for the same direction by the 
two methods shows, however, no appreciable difference, the surfaces having always 
been absolutely clean and free from dust. Moreover, the surfaces produced l^y the 
author’s cutting and grinding goniometer are so absolutely plane, that no rolling, due 
to slight convexity of surface, has ever been observed with them. Further, the 
placing of the crystal and compensator in position on the table of the tripod was 
always effected by sliding, to minimise any intervening compressed air film. In the 
cases of oozing of mother-liquor referred to, the crystals were subsequently heated 
slowly to 105° in an air-bath, and maintained at this temperature for twelve hours. 
On repeating the observations of expansion no further disturbance occurred, successful 
determinations being obtained, and the results agreed satisfactorily with those 
obtained for the same direction of the same salt in cases where this treatment had 
been unnecessary. 
The Determinations of Linear Deformation. 
Mode of Conducting the Observations. 
The determinations of thermal expansion or contraction were made in the manner 
which is very briefly outlined for crystals at the close of the memoir concerning the 
3 O 2 
