352 Deformation of the Crystallised Sulphates of Potassium, $c. 



directions, a and b, for which latter the increments are nearly equaL 

 This is analogous to the optical behaviour, the refractive power being 

 altered (diminished) by rise of temperature much more in the direction 

 of the axis c than in the other two directions, in which the lesser 

 amounts of change are nearly equal. 



The amount of expansion along the direction of the axis b is approxi- 

 mately identical for all three sulphates, indicating that interchange of 

 the metals is without influence on the thermal behaviour along this 

 axis. The crystals of all three salts expand least in this direction, 

 which is, therefore, the common minimum axis of the thermal 

 ellipsoid. 



The chief of the directional perturbations previously referred to con- 

 sists of a reversal, for temperatures below 50, of the directions of the 

 maximum and intermediate axes of the thermal ellipsoid for rubidium 

 sulphate, compared with their directions in the potassium and caesium 

 salts. The maximum thermal axis is c for the two latter salts, but a 

 for rubidium sulphate. A similar reversal of the direction of the 

 maximum axis of the optical ellipsoid (the indicatrix), the first median 

 line, from c to a, occurs for the same temperatures, in the case of 

 rubidium sulphate. The maximum thermal axis is identical with the 

 first median line in all three salts. 



At high temperatures the same relations continue to hold for the 

 potassium and caesium salts, both thermally and optically. But owing 

 to the increment of expansion along c being so much greater than for 

 the other directions, the intermediate expansion along c for rubidium 

 sulphate attains equality at 50 with the expansion along a, and beyond 

 this temperature c becomes the maximum thermal axis for this salt, as 

 it is for the other two sulphates. Consequently, at 50 the crystals of 

 rubidium sulphate are apparently thermally uniaxial. At tempera- 

 tures varying 10 each side of 50 for different wave-lengths of light, 

 they have previously been shown to simulate uniaxial optical proper- 

 ties. The thermal and optical ellipsoids of revolution are not, how- 

 ever, identically orientated, the axis of the former being b and of the 

 latter a. Further, the change of direction of the maximum thermal 

 axis of rubidium sulphate from a to c is followed optically at 180 by 

 the change of the first median line from a to c, rendering the rule as 

 to coincidence of the maximum thermal axis and the first median line 

 again valid. 



A close parallelism between the linear thermal expansion and the 

 directional optical behaviour is thus found to exist. The optical con- 

 stants have been shown in a previous memoir* to exhibit a clear pro- 

 gression following the order of progression of the atomic weights of 

 the three alkali metals ; the values for the three salts are very much 

 more widely separated than in the case of the linear thermal constants, 

 * ' Journ. Chem. Soc., Trans.,' 1894, p. 63C. 



