PKOFESSOll KOPP OIST THE SPECIFIC HEAT OF SOLID BODIES. 
181 
for fixing the atomic weights of the elements than that of taking, as the atomic weights 
of the elements, the relatively smallest quantities which are contained in equal volumes 
of their gaseous or vaporous compounds, or of which the quantities contained in 
such volumes are multiples in the smallest numbers; and no better means appear 
to exist for determining the atomic weights of those elements the vapour-densities of 
whose compounds could not be determined, than the assumption that in isomorphous 
compounds the quantities of the corresponding elements are as the atomic weights of 
the latter. On this basis, and using this means, the numbers for the atomic weights 
have been determined which are contained in the last column of the Table in § 2, 
and are used in § 82 et seq. The atomic weights B=10*9, €=12, Si =2 8 , cannot be 
changed for others. That the atomic weights of tin and silicium are as 118 to 28, is 
further proved by the isomorphism of the double fluorides. But to these atomic weights 
correspond atomic heats which are far smaller than those found for most other elements. 
From the chemical point of view it is inadmissible to take the atomic weights of 
boron, carbon, and silicium * in such a manner as to make their atomic heats agree 
with Dulong and Petit’s law. In any case these three elements form exceptions to 
Dulong and Petit’s law. The sequel will show that this is the case with many other 
elements. 
93. In many compounds the regularity is observed, that by dividing their atomic 
heat by the number of elementary atoms contained in one molecule of the compound, 
a quotient is obtained which comes very near the atomic heat of most of the elements — 
that is, 6-4. This is found in the alloys enumerated in § 82, and also in a great number 
of compounds of definite proportions. A few of the most important cases may be given 
here. For speiscobalt, CoAs 2 (compare § 83), this quotient is ^=6*4; for the 
chlorine compounds, R Cl and R Cl f , the mean of the atomic heats given in § 84 is 
12*8, and the quotient —=Q'4:. Of the chlorine compounds, R Cl 2 , the mean atomic 
heat of all the determinations in § 84 is 18*5, and the quotient ^=6*2. It is also very 
near this value in the double chlorides; inZnK 2 Cl 4 it is ^ =6*2, for R K 2 Cl 6 (the 
mean of the determinations of PbK 2 Cl 6 and Sn K 2 C1 6 ) it is ~=6T. For bromine 
compounds, RBr (both here and in the following examples the means are taken of 
the determinations in § 84), H^=6*9; for PbBr 2 ^=6*5; for iodine compounds, RI 
and RI,^p=6'7, and for the iodine compounds, RI 2 , ^=6'5. 
But this regularity, though met with in many compounds, is by no means quite 
* For Begnatjlt’s observation, whether, considering the specific heat which he found for silicium, its atomic 
weight is to be so taken that silicic acid contains 2 atoms of silicium to 5 of oxygen, compare Ann. de Chim. et de 
Phys. [3] vol. lxiii. p. 30. For Scheekek’s remark, that according to the most probable specific heat of 
silicium its atomic weight must be taken so that for 1 atom of silicium there are 3 atoms of oxygen, compare 
Poggendoeff ? s ‘ Annalen,’ vol. cxviii. p. 182. 
+ In the sequel E stands for a uni-equivalental and E a polyequivalental atom of a metal. 
