Atomic Laivs of Thermochemistry. 37 



then to abandon it by giving a bond two meanings. The 

 better course for the present seems to me to be to determine 

 as carefully as possible the values belonging to all the nuclei 

 for which Stohmann has been supplying a fine store of experi- 

 mental determinations, and if no rational relationship between 

 structure and thermal ^alue in the different nuclei is dis- 

 coverable, to let the matter rest at giving the thermal value 

 for each nucleus until the rational relationship is discovered. 

 2. Haloid Compounds. — The data relate to 15 chlorine 

 compounds, five bromine, and two iodine, from which Thomsen 

 draws the conclusion that for a difference in composition 

 Br — 01 there is a difference in heat of formation amounting 

 to 7*5, while for 01 — I the difference is 19 "4, the ratio of 

 which is 2*6, which is identical with our mean value in 

 inorganic compounds. But it is possible that there is some 

 experimental error involved in this value, because Thomsen 

 finds that the thermal value corresponding to the difference 

 01 — H increases with the increasing molecular weight of the 

 hydrocarbon in which the substitution of 01 for H takes place, 

 so that in the haloid compounds in his values of p + ad he is 

 forced to represent the thermal value of CH 2 , namely v 1 + 2r, 

 by 13-96 +-31-44 in place of 14-2 + 30-0. This would be a 

 very interesting result if it could be relied on as true, for it 

 would demonstrate an influence exerted by an atom of halogen 

 on every 0H 2 group of the organic radical with which it is 

 associated, and the same action on each CH 2 , no matter how 

 many there may be ; this is so improbable that we are led to 

 suspect strongly the existence of experimental error, especially 

 as the experiments and numerical reductions are made more 

 complicated by the presence of the halogen. The peculiarity 

 of some further results with chlorine compounds is also 

 suggestive of the intrusion of experimental error : when an 

 atom of 01 replaces H in a hydrocarbon, then in the expression 

 for p + ad for the hydrocarbon, one r is replaced by the 

 corresponding thermal value of the bond of union of a 01 to a 

 ; for a single 01 this is found to be 13*5, for two CI, as in 

 CH 2 01CH 2 C1, it is 2 x 16*5, for three CI, as in CHC1 3 , it is 

 2x16-5 + 13-5, while for four 01, as in CC1 4 , it is 2 x 16'5 

 + 2x13-5. To show the uncertainty that must attach to 

 some of these numbers let us compare two of Thomsen's 

 values with the later determinations of Berthelot and Matignon 

 (Compt. Rend, cxii.) : thus Thomsen gives : — 



C (amorphous) + H + 301= CHCl 3 (gas) + 23-5, 

 C (amorphous) + 401 = CCl 4 (gas) + 20*4, 

 while Berthelot and Matignon's results are :— 



