1205 



of water per molecule of sulpliuietted liy drogen. In (he nl>ove-inen- 

 tioned determinations this content had fallen to 5,1 — 5,5, and the 

 conclusion was obvious that the li-ne watei' content would be lower: 

 HjS . 5HjO was therefore Ihe most probable formula on the ground 

 of these experiments. 



2. As direct analysis yielded dubious results, and the formula 

 H,S . 5HjO can only be considered probable on account of the observed 

 disturbance — there were no indications pointing' to a second disturb- 

 ance in opposite sense — we have tried to find a method of ana- 

 lysis that yielded more certain i-esults. 



Indications to a definite formula that did not rest on direct ana- 

 lysis are the following: 



n. ViLLARD deems the formula H^S.tiHjO probable on account of 

 the analogy with other gas hydrates, for which he has drawn up a 

 formula M . GH^O '). This analogy must certainly be supported by 

 a closer proof, before it convinces us of the accuracy of the said 

 composition. 



b. ViLLARD could seed the (wo liquids at temperatures at which 

 the hydrate can form, with the hydi-a(e of N,0 and this leads him 

 to the conclusion that the hydrate of sulphuretted hydrogen will 

 possess the same water content '). This reason, too, makes him 

 consider the formula H^S . GHjO probable. 



c. De Forcrand makes use of a rule holding for three-phase lines, 

 which is analogous to that of Trouton for liquid-gas equilibria '). 

 This rule may be represented as follows. When the three-phase line 

 of a dissociating compound, which splits up in(o solid-gas, reaches 

 a vapour tension of one atmosphei-e, the quotient of the heat of 

 transformation and (he absolute temperature has the value 30. He 

 gives some examples for this rule, and then a|)plies it to determine 

 the quantity of wafer in gas hydrates. That this i-ule is, however, 

 dangerous appears already sufTicienfly from the fact that on appli- 

 calion to the hydrate of snlphui- dioxide the conjposition SO, . 8H,0 

 was found, whereas on the strength of Bakhuis Roozkroom's and 

 ViT, lard's analyses it could be concluded with great probability that 

 this wafei'-content is too high '). His rule, likewise, leads to H^S . 6H,0. 



As in our opinion the indirect methods have not yet yielded certain 

 results either concerning the composition of (he hydiate, we have tried 



1) ViLLARD. Ann. chim. pliys. (7). 11. 289. (1897). 

 ^) De Forcrand. G.r. 135. 959. (1902). 



2) Bakhuis Roozeboom. Rec. 3. 29. (18S4); Vh-lahd. Ann. cliim. pliys. (7). 11. 

 289. (1897). 



78 

 Proceedings Royal Acad. Amsterdam. Vol XXI. 



