Atomic Laws of Thermochemistry. 55 



metals do not vanish. By analogy it was assumed that (0) 

 and (N) are both zero, though from a mathematical point of 

 view it does not matter whether they are zero or not ; for in 

 (R) +(S)+/(RS), to assume that (R) is zero is the same as 

 merging it in/(RS). The values determined for/(RS) from 

 Thomsen's data for organic compounds are as follows : — 



/•(OC) = 14. /(C:C) = 14-2. /(O;C)=0. /(CH) =15. /(CH))=18-5 and 15. 



flC:0) = 53. /(ON)=0. /(0:N) =0. /(0;N)=0. /(OH) =29-5. 



/(O0) = -10. /(ON)=0. /(0:N T ) =0. /(NH) =3'8. /(SH) = 2«5. 



/(OS)=0. /(0:S) = 9. /(N-N)=0. 



The association of the bindings C"0 and C:0 in one atom 

 of C causes an evolution of 15 ; and the association of two 

 bindings C:0 in one atom of C causes an evolution of 30 ; 

 the conversion of tetrad to dyad C causes an evolution of 15. 

 The introduction of a second N:0 binding into an N atom 

 connected with C or with connected with C causes an 

 evolution of 10. 



The first point to be noticed in these results is that the 

 functions /(RS) cannot be taken to have the form "^(R)^(S), 

 as we found to be the case when R is a metal and S a halogen : 

 in fact the data are too few to give any reliable cue to the 

 form of the functions. But as regards their numerical values, 

 the second point to be noticed is that zero occurs eight times, 

 zero standing for a value too small to be assigned in the 

 present state of thermochemical analysis. The third point is 

 that the value 15, or nearly 15, occurs six times; while the 

 value 29*5 is nearly twice 15, and 53 nearly three and a half 

 times 15. Of the remaining values one is 3*8, which is almost 

 exactly a quarter of 15 ; another is 18*5, which exceeds 15 by 

 almost a quarter of 15; and the rest are —10, 10, 2*5, and 9. 

 Thus it appears that nearly all the values could be expressed 

 as multiples of a quarter of 15, that is of 3*75, or nearly 3'8. 

 Now at the end of Part I., in discussing the values of (R) for 

 the metals, it was found that on dividing them by 3' 8 the 

 results suggested that (R) for the metals is a multiple of 

 3*8, with some indication of law amongst the multiples. 

 By an accidental coincidence it happened that in studying 

 /(RS), when S is a halogen and R any metal, we found 

 /(RCl)-/(RBr)+/(RCl)-/(RI) or 2/(RCl) -/(RBr)-/(RI) 

 to be a series of multiples of 3*8, with indications of law 

 among the multiples. There is, therefore, rather strong evi- 

 dence for the hypothesis that atoms, in combining chemically 

 with one another, give forth integral numbers of a certain unit 

 of heat, which may be called the atomic thermochemical unit. 



