158 Britain's Heritage of Science 



who started loading the scale are then forgotten, unless 

 someone with a taste for historical continuity happens to 

 come across the record of their work. Especially when some 

 national feeling is involved, discussions on priority may then 

 be raised, and continued interminably, because there will 

 always be a conflict between those who attach importance 

 to the intrinsic merit of an investigation and those who 

 look only on the actual influence it has had on scientific 

 thought. In the strict administration of historical justice, 

 oral expressions of opinion like that of Stokes are not 

 admitted as evidence; he himself disclaimed any share in 

 the discovery of spectrum analysis. But as a testimony 

 that the analogy of sound can be applied to the radiations 

 of light and heat, it was a distinct step, and a well ascer- 

 tained and clear pronouncement such as that which passed 

 between Stokes and Kelvin deserves to be placed on record, 

 without detracting from the merit of others. 



In order to appreciate correctly Balfour Stewart's work 

 the following consideration is important. If the foundation 

 of spectrum analysis be made to depend on such laws of 

 radiation as can be derived from the consideration of what 

 happens inside an enclosure of uniform temperature, his 

 priority is well established. He undoubtedly was the first 

 to realize the significance of studying the equilibrium of 

 heat inside such enclosures, and led the way in a direction 

 of research which has proved to be of capital importance 

 in the theory of radiation. But as regards their practical 

 bearing on spectrum analysis, too much weight has been 

 given to theoretical considerations founded on thermal 

 equilibrium. In all spectroscopic observations, the loss or 

 gain of heat is the essential factor. The step which takes 

 us from the uniform enclosure to the radiation and absorp- 

 tion when there is no equilibrium is not so simple as has 

 generally been assumed, and it is safer to accept spectrum 

 analysis as being mainly founded on experiment together 

 with such plausible theoretical analogies between sound and 

 light as were pointed out by Stokes. In this respect, the 

 work of Herschel, Talbot, Wheatstone, and Swan is of 

 greater importance in the history of spectrum analysis than 

 the theoretical work of Balfour Stewart, who, however, also 



