on the Atomic Weights of Bodies. 267 



atoms hydrogen, which would place it among the ternary 

 compounds. In that case the atom of hydrogen would 

 weigh only 09625. This number multiplied into 3-2936, 

 the specific heat, would give us 0*206, which is considerably 

 below the average number 0*375. But this may be owing 

 to an error in the determination of the specific heat of hy- 

 drogen. Were we to adopt 4*177 for the specific heat of 

 hydrogen the number calculated from the experiments of 

 Delarive and Marcet, the multiplication of that quantity 

 by 0*0625 would be *26l, which approaches somewhat near 

 the mean 0*375, though still at a considerable distance. 

 In the present state of our knowledge, it would be hazard- 

 ous to determine whether water be a binary or ternary 

 compound. 



Corrosive sublimate is precisely in the same predicament 

 as water. It consists only of two atoms united, and yet 

 the product of its atomic weight by its specific heat is 1*175, 

 which is nearly 0*375 x 3. In the present state of our 

 knowledge I see no other conclusion to which we can come, 

 than that when an atom of mercury and an atom of chlo- 

 rine unite to constitute corrosive sublimate, the quantity of 

 heat united to each of the constituents, while separate, be- 

 comes augmented by one-third when they unite together. 



Common salt, chloride of potassium and chloride of cal- 

 cium are also binary compounds, according to the theory 

 universally admitted by chemists ; yet the mean product 

 of their atomic weight by their specific heat is 1*587, which 

 is nearly equal to 0*375 x 4. We must conclude from this, 

 if the presently received chemical theory be true, that when 

 chlorine combines with sodium, potassium or calcium the 

 quantity of heat united to each atom, before combination, 

 is doubled. 



The third table consists of compounds of one atom of a 

 metal with 1^ atoms of oxygen or sulphur. The mean pro- 

 duct of their atomic weight by their specific heat gives 

 0*847 ; while 0*375 x 2J = 0*9375. The difference is, 

 probably, owing to errors in the determination of the 

 specific heat. No light would be thrown on the subject by 

 examining the atomic weights assigned by Berzelius. He 

 doubles his weights in order to'get rid of the fraction; and, 

 of course, doubles the nnmber of constituents. In the pre- 



