MONOCLINIC DOUBLE SELENATES OF THE IRON GROUP. 
401 
In addition to the above determinations of density with clear, transparent, 
absolutely fresh crystals, a number of further determinations were made with crystals 
in the act of decomposing, employing some that were only just showing the first 
signs when placed in the immersion liquid, and continuing with the same crystals, a 
fresh determination of their density being made each half-hour. The temperatures 
of the liquid only varied from 13°‘7 to 14°'5, and the results for 20°/4° of seven 
successive determinations at the half-hour intervals were 2'5211, 2'5238, 2'5296, 
2'5442, 2'5551, 2‘5622, and 2'5726. It is thus obvious that the crystals became 
denser as decomposition proceeded. This would appear to agree with the conclusion 
of Tops0E that the monoclinic crystals of the hexahydrated salt K 2 Fe (Se0 4 ) 2 . 6H 2 0, 
which are only stable up to a temperature but slightly above 0° C., break down at 
ordinary temperatures into microscopic triclinic crystals of the dihydrated salt 
K 3 Fe (Se0 4 ) 3 . 2H 2 0. 
Optics. 
The author’s usual procedure—including the grinding of two section-plates parallel 
to the symmetry plane for extinction determinations of six section-plates (three 
perpendicular to each of the two median lines) for optic axial angle measurements, 
and of six 60°-prisms for refractive index determinations—was not possible with this 
unstable salt. As already mentioned, only four crops of the crystals were obtained, 
and the life of any one crystal was not more than six hours, after which it had usually 
become hopelessly opaque from decomposition. Hence, very rapid methods of work 
and avoidance of the grinding of many surfaces (which become much more rapidly 
opaque than natural ones) were essential, for the time lost in grinding a plate or prism, 
which when finished was useless, could much more profitably be spent in making 
confirmatory measurements with other natural crystals. 
All the optical work, therefore, was carried out with three natural plates, tabular 
crystals parallel to the basal plane c {001}, two natural prisms formed by a pair of 
p-faces (inclined at 72° 2') and a pair of g-faces (inclined at 52° 12 r ), and one ground 
prism (of 60°) only, the latter from the largest of the crystals obtained and ground 
accurately to afford the a and y indices, the grinding of which was fortunately 
accomplished and the measurements obtained with it just before opacity supervened. 
Each of the three crystal plates was mounted with hard balsam in benzene on a truly 
plane glass plate (one of the circular ones a centimetre diameter provided with the cut¬ 
ting and grinding goniometer) and covered with a small piece of thin cover glass, the 
crystal being completely protected by the balsam and its life thereby lengthened by an 
hour or two (in one exceptional case longer). Each crystal plate was rhombus-shaped 
(see fig. l), the fourg>-faces being the main rhombus-determining edge faces. It was 
suspended by its circular glass mount-plate truly vertically from the adjusting apparatus 
of the polariscopical goniometer, so that the shorter diagonal of the rhombus (from 
front to back in fig. l), the trace of the symmetry plane b {Old}, was horizontal; it 
was also adjusted so that the centre of the rhombus-]ike c -face was in the optical axis 
