MOLECULES AND ATOMS 315 



N.^OS remains unchanged) and 19 (when NO 4- N0 2 is obtained). 

 There are no figures of constant density for H 2 2 , NH0 3 , N 2 O 4 , and 

 many similar compounds which are either wholly or partially decom- 

 posed in passing into vapour Salts and similar substances either have 

 no vapour density because they do not pass into vapour (for instance, 

 potassium nitrate, KNO 3 ) without decomposition, or, if they pass 

 into vapour without decomposing, their vapour density is observed 

 with difficulty only at very high temperatures. The practical de- 

 termination of the vapour density at these high temperatures (for 

 example, for sodium chloride, ferrous chloride, stannous chloride, &c.) 

 requires special methods which have been worked out by Sainte-Claire- 

 Deville, Crafts, Nilson and Pettersson, Meyer, Scott, and others. 

 Having overcome the difficulties of experiment, it is fo'und that the 

 Jaw of Avogadro-Gerhardt holds good for such salts as potassium 

 iodide, beryllium chloride, aluminium chloride, ferrous chloride, &c. 

 that is, the density obtained by experiment proves to be equal to half 

 the molecular weight naturally within the limits of experimental 

 error or of possible deviation from the law. 



Gerhardt deduced his law from a great number of examples of 

 volatile carbon compounds. We shall become acquainted with certain of 

 theln in the following chapters ; their entire study, from the complexity 

 of the subject, and from long-established custom, forms the subject 

 of a special branch of chemistry termed ' organic ' chemistry. With all 

 these substances the observed and calculated densities are very similar. 



When the consequences of a law are verified by a great number of 

 observations, it should be considered as confirmed by experiment. But 

 this does not. exclude the possibility of apparent deviations. They may 



evidently be of two kinds : the fraction may be found to be either 



greater or less than 2 that is, the calculated density may be either 

 greater or less than the observed density. When the difference between 

 the results of experiment and calculation falls within the possible errors 

 of experiment (for example, equal to hundredths of the density), or 

 within a possible error owing to the laws of gases having an only 

 approximate application (as is seen from the deviations, for instance, 



from the law of Boyle and Mariotte), then the fraction proves but 



slightly different from 2 (between 1-9 and 2-2), and such cases 

 as these, may be classed among those which ought to be expected 

 from the nature of the subject. It is a different matter if the quotient 



of =_. be several times, and in general a multiple, greater or less than 2. 



