SCIENCE 



[N. S. Vot.XXV. No. 627 



And certainly in this same category be- 

 longs the fact that the spectrum of an arc 

 is modified when the arc is surrounded by 

 an atmosphere difilerent from ordinary air. 



Here also lie the profound differences 

 between arc and spark spectra of the same 

 element. 



Notwithstanding the fact that 'multiple 

 spectra' is a term which has hitherto been 

 employed to describe the Pliicker tube vari- 

 ations, I propose that we generalize it and 

 use it to describe this entire group of facts. 

 Since the name is so appropriate, let us 

 call the sixth fundamental phenomenon 

 that of 'multiple spectra.' 



7. Any competent atom must allow us to 

 infer the relations which have been proved 

 to exist between spectral phenomena and 

 atomic weights. 



8. The phenomena of line reversals and 

 absorption bands. 



9. The fact that heat alone, at least with- 

 in the range of our highest artificial tem- 

 peratures, produces characteristic spectra 

 in only a few rare instances. 



These, briefly, are the parts of the spec- 

 troscopic superstructure for which a foun- 

 dation is sought. These are the various 

 parts which it is hoped will, some day, be 

 cemented together, by a simple and general 

 theory, into a harmonious structure. 



But there is a final criterion, even more 

 fundamental than any of those which have 

 been mentioned, that such a theory must 

 satisfy, namely, this hypothetical radiant 

 atom must not in its behavior, except as a 

 very last resort, contradict any of the es- 

 tablished principles of physical science, be 

 they mechanical, electrical or chemical. 



The principle of the conservation of en- 

 ergy must be satisfied even if it is neces- 

 sary to assign an undreamed of amount of 

 energy to each atom; in like manner New- 

 ton's third law is to be satisfied, even if the 

 electromagnetic ether is called upon to fur- 

 nish the reaction. 



But even with this added criterion, the 

 preceding list of nine phenomena is con- 

 fessedly incomplete; the only object of 

 such a catalogue is to include those typical 

 fundamental facts which ought, apparent- 

 ly, to follow as immediate consequences 

 from the structure of the radiating body, 

 so soon as that structure is correctly 

 guessed. Thus Doppler's priuciple is 

 omitted on the ground of its being rather 

 a kinematic law, governing periodic dis- 

 turbances in any medium, than a dynamical 

 fact to be explained in terms of atomic 

 structure and forces. 



THE SATUKNIAN ATOM 



Having established a set of criteria by 

 which we may estimate the fitness of a 

 radiant atom, it would be interesting, if I 

 were competent, and if time permitted, to 

 pass in review some of the various atoms 

 which have been proposed in recent times; 

 such as that of Kelvin, 1884, or those sug- 

 gested by the Hertzian oscillator. 



But neither of these two conditions is 

 fulfilled, and I propose, therefore, to con- 

 sider only one atom, namely, the one which 

 by common consent, I think I may safely 

 say, comes more nearly satisfying the de- 

 mands of experimental fact than any other 

 ever devised. I refer to the atom first pro- 

 posed in a general way by Lord Kelvin in 

 his paper entitled 'Epinus Atomized,'^ and 

 afterwards profoundly modified by Lor- 

 entz, Thomson and Larmor. 



So much work along this line has been 

 done in the Cavendish Laboratory that one 

 feels impelled to call this 'the Cambridge 

 atom'; in view, however, of its structure, 

 perhaps 'the Saturnian atom' is a more ap- 

 propriate designation. 



Now as to the proper conception of the 

 normal Saturnian atom — I am not certain 

 that I know what this is, but my mental 



'Baltimore Lectures, p. 541 (Cambridge, 1904). 



