CHEMISTRY. 431 



ten, on the strength of Deville and Troost's researches, AU Cle, which 

 corresponds to a density of 9.20. According to Nilson and Petterson 

 the density at 440° is 7.789, at 758° it is 4.802, and above that temper- 

 ature, from 1117° to 1200°, it is fairly constant at from 4.247 to 4.277. 

 The value 4.600 corresponds to the molecule ARM,, which is required 

 by the trivalency of aluminum, as indicated by the periodic law. 

 (Zeitsch. Physikal. Chem., i, 459.) 



On the other hand, Friedel and Crafts, at temperatures ranging from 

 218° to 433°, find a vapor density of 9.24, agreeing with Deville and 

 Troost's determination. They get no evidence of dissociation into any 

 simpler molecules. (Compt. Rend., cvi, 1764.) 



Conformable with these data Louise aud Roux, working with alu- 

 minum methyl and aluminum ethyl, get values corresponding to the 

 general molecules AI3 Re. For the same compounds Buckton aud Od- 

 ling long ago found densities according with Al Rn. Both compounds, 

 according to Louise and Rous, decompose at high temperatures, yield- 

 ing aluminum, hydrogen, and hydrocarbons. (Compare also the data 

 obtained by the same authors by Raoult's method, as cited in the pre- 

 ceding article. (Compt. Rend., cvi, 73 and 602.) 



Altogether the evidence seems to prove the existence of AljRe mole- 

 cules, which split u]> at high temperatures into two of Al R,. 



Ferric chloride, if FcaCl,;, requires a vapor density of 11.2. If FeCls, 

 its value should be 5.6. Griinewald and V. Meyer obtain the following 

 values: At 448°, 10.487 ; at 518^, 9.569 ; at 606°, 8.383; at about 750°, 

 5 389 to 5.528, with evidence of dissociation; at about 1050°, 4.915 to 

 5.307, and similar figures at higher temperatures. At 448° it vaporizes 

 completely undecomposed, and even then its vapor density is too low for 

 Fe^Cle. That compound, therefore, they believe does not exist ; and 

 regard the lower symbol, FeCls, to be the true one. At the higher tem- 

 peratures it splits uj) to some extent into FeCL and free chlorine. (Be- 

 richte, xxi, 687.) 



To this investigation, as to Nilson and Petterson's on aluminum chlo- 

 ride, Friedel and Crafts reply. They determine the vapor density of 

 ferric chloride in presence of free chlorine, in order to prevent any de- 

 composition, aud find a figure agreeing with FCiClc. Tbeir data are 

 for temperatures from 321° to 342°. (Compt. Rend., cvii, 302. Com- 

 pare Biltz, Berichte, xxi, 2766.) 



Ferrous chloride, redetermined by Nilson and Petterson, at tem- 

 peratures varying from 1300° to 1500^, has a density of 4.292 to 3.340. 

 Theory for FeCl2, 4.375. Earlier work by V. Meyer seems to indicate a 

 more complex molecule at lower i^emperatures. (Journ. Chem. Soc, 

 Liii, 827.) 



Nilson and Petterson also have determined the vapor density of 

 chromic and chromous chlorides. For the former, at 1200'^ to 1300o, it 

 agrees closely with the molecule CrCl,. At 1065'^ a higher vahu^ was 

 found, but not at all approxini(\tiilg to CriClfi. The latter probably cau 



