Januaey 4, 1907] 



SCIENCE 



picture of it is somewhat like the follow- 

 ing: 



1. Conceive a single negatively charged 

 electron — whatever that may be — placed 

 inside a mass of positive electrification — 

 whatever that may be. On the basis of the 

 Zeeman effect, we may imagine this elec- 

 tron to be revolving about the center of 

 the positive charge ; and we may assume its 

 rate of revolution such that it is in equi- 

 librium under the first-power-of-the-dis- 

 tance law. 



But even if we did not have the Zeeman 

 effect to suggest rotation we should be com- 

 pelled, as Jeans^ has shown, to introduce 

 rotation, on the basis of Barnshaw's the- 

 orem, to secure stability. Any accelera- 

 tion of this electron which is periodic will 

 produce a periodic radiation of energy. 

 Precisely such an acceleration is here pres- 

 ent in the familiar rw^ centrifugal accelera- 

 tion which is periodic when we consider 

 radiation along any one fixed direction. 

 The frequency of this acceleration deter- 

 mines that of the radiation just mentioned. 

 The electromagnetic effect is roughly that 

 of an alternating displacement current. 



2. Let us next suppose that instead of a 

 single corpuscle we have a large number 

 distributed throughout the same orbit. 

 Their radiation is now almost nil, the vec- 

 tor sum of the accelerations being zero and 

 the electromagnetic effect being roughly 

 that of a steady current. 



If the ring does not contain so many of 

 these self -repellent corpuscles as to become 

 unstable we have a simple type of a non- 

 luminous and, during stability, non-radio- 

 active element. 



Imagine now that these electrons are the 

 same for all elements; then one element 

 differs from another mainly in the number, 

 disposition and character of the rings 

 which surround the central attracting 



'Jeans, Phil. Mag., Z, 425, 1901. 



charge, the number of corpuscles being 'of 

 the same order as the atomic weight.'* 



Such is a rough sketch of the normal 

 Saturnian atom. The beautiful manner in 

 which this structure permitted J. J. Thom- 

 son to infer the same periodicity in electro- 

 chemical properties as that contained in 

 Mendele Jeff's table is already familiar to 

 you. But for the present inquiry this ex- 

 quisite achievement of Thomson 's is merely 

 an 'aside.' So also is Drude's elegant con- 

 nection of electrical and thermal conduct- 

 ivities in metals. Likewise his explanation 

 of the Hall effect. 



We come now to the question which is 

 fundamental to all spectroscopic theory, 

 namely, under what conditions does a gas 

 atom become radiant. 



This question may be asked and answered 

 in two different senses: 



First, one may inquire as to the labora- 

 tory conditions necessary to produce lu- 

 minosity in a gas; the corresponding an- 

 swer is threefold: either a high tempera- 

 ture, thus obtaining, in some rare cases, a 

 heat spectrum, or secondly, a rapid chem- 

 ical change as in fiames, or thirdly, an 

 electric field as in the arc, spark and vacu- 

 um discharge. 



Again, one may ask what is the differ- 

 ence between the internal conditions of a 

 radiant and non-radiant atom. So far as 

 I am aware, this latter query has never 

 received an answer which is definite or 

 based upon indisputable experimental evi- 

 dence. However, the Zeeman effect points 

 to rotation in the luminous source and sug- 

 gests the revolving electron as the light- 

 giving body; but it is difficult to see how 

 one electron could give rise to more than 

 one line in the spectrum. Not only so, but, 

 since the electrons are the same for all ele- 

 ments, it is clear that the electrons alone 

 can not emit characteristic spectra, the 



" J. J. Thomson, PMl. Mag., 11, 774, 1906. 



