Thermocliemical Constants. 59 



cidences were a matter of mere chance, whereas in the other 

 two they are only | and \ as numerous as they should be ; 

 but it must be remembered that we are only taking the actual 

 data expressed in a limited number of ways, such as Thomsen 

 happened to have selected, and that these same data might 

 be expressed in numerous other ways, giving us a much 

 longer list of numbers. 



With Nos. 5 and 6, where the quantities are equal, the 

 fallacy of their connexion may be well shown thus : — The 

 two quantities in No. 5 are selected from the two pages (403, 

 404) on which the sulphur data exist : can we find numerical 

 equality as striking by taking any other two pages of data? 

 Taking p. 402, which contains iodine data, and p. 403, which 

 contains sulphur data, we find : — 



K, l,0 3 (cryst. salt) =124,490, 



S, 3 , H 2 "(liquid acid) .... =124,560, 



S0 3 ,aq. (heat of dissol. of solid S0 3 ) = 21,320, 



I, Cl 3 (heat of form, of solid) . . = 21,490, 



S0 2 aq., O (solution) = 63,630, 



HI, 3 (anhydrous) = 64,000, 



S 2 7 H 2 , aq. (heat of dissol.) . . = 54,320, 

 H, I, 3 , aq. (heat of form, and dissol.) = 55,800. 



Four instances of approximate numerical equality between 

 quantities representing reactions which are altogether dis- 

 similar. 



Nos. 7, 8 and 9 refer to the formation of metallic com- 

 pounds : they are the only ones of this nature quoted by 

 Thomsen, and bear but a very small proportion indeed to the 

 hundreds of values for metallic compounds given in vol. iii.* 



Of those cases which deal with differences between the heat 

 of formation of similar compounds, Nos. 11 and 12 have already 

 been discussed, and it is only necessary to point out further 

 that these three instances form but a minute fraction of the 



* The heat of formation of the oxides of iron is partially dependent 

 on that of the chlorides and on the heat of neutralization of the respective 

 oxides by hydrochloric acid. It is possible that there is an error of 

 4,000 cah or more in the determination of Fe.,0 3 3H 2 0, GHC1 (vol. i. p. 410); 

 since the difference between this quantity and the heat of neutralization 

 of the same oxide by sulphuric acid is smaller by this amount than in 

 any other known case. It is more probable, however, that the error is in 

 the heat of neutralization by sulphuric acid, for the indirect determina- 

 tions of it involved in the precipitation of iron sulphate by a barium salt 

 (pp. 410, 417) give similarly anomalous numbers. Some recent deter- 

 minations of these latter quantities (Fay, ('hem. News, lvii. p. 86, and 

 Pickering, ibid. lvii. p. 7C>) tend to show that this anomaly does not always 

 exist with iron sulphate. It may have been due to the presence of 

 colloidal ferric oxide in the solution. 



