288 
ON SMELL. 
It is possible to make, with practice, a fairly accurate 
analysis by means of the sense of smell. The method is, 
knowing the constituents of a mixture, to prepare one which 
has the same smell, measuring the proportions of the ingredients. 
The only precaution to be observed is that the smell of no 
member of the mixture be so overpowering as to mask those of 
the others. Thus I have analysed or rather synthesised a 
mixture of chloroform with ether ; alcohol with ether ; and these 
liquids with carbon disulphide, provided the latter be pure, to 
within 2 p. c. ; but I failed with members of the pyridene series. 
Yet it was possible to detect the proportions of members of that 
series to each other ; and it is not difficult, however extraordinary 
it may appear, to guess approximately the boiling point of a 
mixture of members of a series, after some practice, purely by 
its smell. 
So far as I know no theory has been brought forward to 
account for the sense of smell, and I therefore venture to supply 
this want, premising that what follows is merely a tentative 
explanation, and as such, will I hope not be too severely 
censured. 
There is a probability that our sense of smell is excited by 
vibrations of a lower period than those which give rise to the 
sense of light or heat. These vibrations are conveyed by gaseous 
molecules to the surface network of nerves in the nasal cavity. 
The difference of smells is caused by the rate, and by the nature 
of such vibrations, just as difference in tone of musical sound 
depends on the rate and on the nature of the vibration; the 
nature being influenced by the number and pitch of the harmonies. 
Let us see what evidence can be adduced for this theory. 
Among the lightest substances which have smell are sulphur- 
etted hydrogen and phosphoretted hydrogen, both of which are 
17 times as heavy as hydrogen itself. 
Prussic acid is fifteen times as heavy as hydrogen, and has a 
