XX.] ^lETHOD OF VARIATIONS. 447 



tant instance of investigations of tlie same kind, showing 

 the effects of interference of light undulations of all 

 magnitudes at a single view. Herschel gave to all such 

 opportunities of observing directly the results of a general 

 law, the name of Collective Instances,^ and I propose to 

 adopt the name Collective Experiments. 



Such experiments will in many subjects only give the 

 first hint of the nature of the law in question, but will not 

 admit of any exact measurements. The parabolic form of 

 a jet of water may well have suggested to Galileo his views 

 concerning the path of a projectile ; but it would not serve 

 now for the exact investigation of the laws of gravity. It 

 is unlikely that capillary attraction could be exactly 

 measured by the use of inclined plates of glass, and tubes 

 would probably be better for precise investigation. As a 

 general rule, these collective experiments would be most 

 useful for popukir illustration. But when the curves are 

 of a precise and permanent character, as in the coloured 

 figures produced by crystalline plates, they may admit of 

 exact measurement. Newton's rings and diffraction fringes 

 allow of very accurate measurements. 



Under collective experiments we may perhaps place 

 those in which we render visible the motions of gas or 

 liquid by diffusing some opaque substance in it. The 

 behaviour of a body of air may often be studied in a 

 beautiful way by the use of smoke, as in the production 

 of smoke rings and jets. In the case of liquids lycopodium 

 powder is sometimes employed. To detect the mixture of 

 currents or strata of liquid, I employed very dilute solutions 

 of common salt and silver nitrate, which produce a visible 

 cloud wherever they come into contact.^ Atmospheric 

 clouds often reveal to us the movements of qreat volumes 

 of air which would otherwise be quite unapparent. 



Periodic Variations. 



A large class of investigations is concerned with Periodic 

 Variations. We may define a periodic phenomenon as one 

 which, with the uniform change of the variable, returns 



' Preliminary Discourse, &c., p. 1 85. 



' Philosophical Magazine, July, 1857, 4th Series, vol. xiv. p. 24. 



