( 836 ) 
hydrate evaporates and melts. Then it will be clear at the same 
time that the earlier observations of DE ForcrAND *) and of CarLLeETET 
and Borper ®) refer to this three-phase curve, as they have determined 
pressures which were necessary for the formation of hydrate; then 
it becomes also clear that this pressure as Dr ForcrAND points out, 
is independent of the concentration of the mixture, which is, indeed 
required for this three-phase equilibrium by the phase rule.*) [ have 
indicated their observations by the figures Cand £ both in the table 
under SZ, and in the P-7-projection of fig. 2. My observations 
appear to lie between them. The pretty large deviations between the 
three lines must, in my opinion be ascribed to this, that the formation 
of the hydrate in case of compression, and the melting in case 
of expansion takes place slowly; in the first case we tind too 
low, in the second too high pressure. | have made my observations 
of these three-phase equilibria by slowly heating at constant 
pressure, and by determining the temperature at which the trans- 
formation S~L+G occurred. This method of working appeared 
to yield more accurate results than that in which the transformations 
were observed for isothermal change of volume. 
Whereas the equilibria on the other three phase curves establish 
themselves spontaneously at constant temperature, and the deviations 
rarely exceed 0.1 atmosphere, I think I shall have to estimate the 
accuracy of the observations on the lines SL,G and SL,L, at 01 
10)/0; 27, -): 
In the /P-2-section the gasphase ( and the liquid phase ZL, lie 
on the three-phase-curves on the hydrogen sulphide side; now it 
appeared from some preliminary experiments that the liquid phase 
L, lies near the water side, in other words that the region of non- 
miscibility extends over almost the full width of the figure. 
Finally a few remarks may be made on the composition of the 
hydrate. It is not to be derived with certainty from the literature 
on the subject. Dr Forcranp, who has made numerous analyses of 
hydrate, considered successively H,5.15H,O, H,S.12H,0, and H,5.7H,O 
the most probable formula. He justly ascribes the very bad agree- 
ment of the analysis results to the fact that the hydrate easily 
1) pre Forcranp. Cr. 94, 967 (1882); De FORCRAND and VILLARD Cr. 106, 
849 (1888). 
2) GAILLETET and BoRDET. Gr. 95, 58 (1882). 
3) DE FoRCRAND. Gr. 94, 967 (1882). 
+) Also the observation of SZ, Lj took place by the determination of the tem- 
perature at which the transformation S— Zj + L, is found with slow heating at 
constant pressure. 
