§ 5 OF ARTS AND SCIENCES. 37 1 



but nevertheless sufficiently hard to resist any change of 

 shape induced by the sudden pressure of impact. In the 

 case of gases and vapors, it is known that this yielding 

 may be disregarded; but in liquids and solids we have no 

 right to assume that the laws founded upon this assumption 

 will be rigorously true. It must not be attempted, without 

 a special examination, to apply them to mercury, or any 

 substance whose coefficient of expansion is so widely diffiir- 

 ent from that of a gas. A cursory glance at the tables of 

 expansion will be sufficient, however, to show that the 

 densest bodies have, in general, the least coefficients of 

 expansion, so that the existence of such substances as 

 mercury and the heavy metals, with their very small 

 coefficients, encourages us to think that in such light and 

 expansible liquids as ether, alcohol, etc., there may still 

 be sufficient distance betvs^een the molecules to enable us 

 to disregard their compressibility. 



The ratio of the two products, JLD and MeT, which 

 should be unity for an elementary substance containing 

 only one atom to the molecule, I shall call the principal 

 ratio. This ratio has not been, determined for a single 

 elementary substance. 



From the mean results tabulated in § 3, we find for the 

 principal ratio of alcohol the value 2.17; for ether, 0.74; 

 for bisulphide of carbon, 0.89, and for turpentine, 0.93. 

 These numbers are not probably accurate within five or 

 even ten per cent., as two determinations of the modulus 

 of elasticity alone are apt to differ by ten or twelve per 

 cent. 



For bisulphide of carbon, we may adopt the graphical 

 svmbol 6" = C = 6", the symbol ■5=6', or S = S. 



w // w // 



c c 



We shall find, in either case, one atom of carbon united 

 to two atoms of sulphur; but in the first case the two sul- 



