Va/Hyui-'Densities of some Carbon Compounch. 493 



ot* a is that oiven by Regnaiilt, viz. -^-^ —1 = «(/>— ^j>o)7 <ii^^l 



represents the deviation of the compressibility o£ the gas 

 from Boyle^s law for increase of ])ressure of one atmosphere. 

 The error in the atomic weights of oxygen, hydrogen, nitro- 

 gen, and carbon thus determined is estimated not to exceed 

 1 part in 5000. M. Berthelot also discusses deviations from 

 Boyle's law exhibited by easily liquefiable gases, namely, 

 carbon dioxide, nitrous oxide, hydrogen chloride, acetylene, 

 phosphoretted hydrogen, and sulphur dioxide, making use 

 for this purpose of van der Waals'' modification of the simple 

 gas laws. From the previously determined densities of these 

 gases he calculates their molecular weights, using as a basis 

 the molecular weight of oxygen taken as 32 ; whence by 

 subtraction the atomic weights of hydrogen, carbon, nitrogen^ 

 sulphur, and chlorine are deduced. 



Another method of applying M. Berthelot^s procedure is 

 this : — Determine the density of the gas ; determine also its 

 compressibility at low pressures, preferably below one atmo- 

 sphere ; assuming the compressibility to be a linear function 

 of pressure, calculate the value of the product pv at zero 

 pressure : the ratio of the pv'^ will be the ratio of the den- 

 sities when equal volumes contain equal numbers of mole- 

 cules. The molecular weights, and consequently the atomic 

 weights, follow as a matter of course. The method can also 

 be applied graphically, by plotting as ordinates the pressure, 

 and as abscissas the values of p^;/T ; where the curve cuts the 

 line of zero pressure the theoretical value of pv/T has been 

 reached. 



There appeared every prospect that the determination of 

 the density of vapours should, if carried out with pure sub- 

 stances, lead to results as interesting and important as those 

 deduced by Daniel Berthelot. And if the estimations were 

 made with pure substances the atomic weights of the elements 

 contained in the compounds might be determined with ex- 

 treme accuracy. Moreover, the method of deducing atomic 

 weights from density may be regarded as a " statical^' method, 

 in contrast with the usual analytical methods, which, on 

 account of the transformations which the elements are made 

 to undergo during the determination, may well be termed 

 " dynamical.^'' 



After some preliminary attempts to apply a modification of 

 Dumas' method it was discarded in favour of a modification 

 of Gay-Lussac's ; this method has the advantage that while 

 densities are being determined compressibilities may be 

 simultaneously estimated with the same sample of material, 



