( 353 ) 



a, 

 ?iüns tend to sliow taat — is au excessively small fraction ; if it 



A 



were otherwise, there woukl be so much damping- tliat the spectral 

 lines could not be as sharp as they are. 



Now one might believe that on the sphere itself the amplitude of 

 the -Ta-vibrations were so much greater than that of tlie I'l-vibrations, 

 that the motions of the second order could produce a perceptible 



amount of light, notwithstanding the smallness of the fiictor ^-. As- 

 suming this for an instant, improbable though it seemed, and deter- 

 mining by my formulae, for the shell as well as for the solid sphere, 

 the properties of the emitted rays of light, I was led precisely to 

 CoKNu's quartet if I supposed the observations to take place across 

 the lines of force, the middle line of the quintet vanishing alto- 

 gether. This seemed very promising at first sight, but, considering 

 the radiation along the lines of force, I found that in this case it 

 ought to be the two inner lines of the quartet that remained, and not 

 the outer ones, as observation has shown to be the case. This suffices 



to banish all idea that the influence of the factor — might be com- 



pensated by a large amplitude in the sphere. We cannot but take for 

 gi'anted that the vibrations corresponding to harmonics of the second 

 order are incapable of radiating. This is due to the circumstance 

 that in adjacents parts of the sphere there are equal and opposite 

 displacements of equal charges. 



Of course, the vibrations of still higher orders will be equally 

 incapable of producing rays, and similar remarks will apply to sys- 

 tems of a totally different nature. Thus, the higher tones of a 

 sounding body whose dimensions are very much smaller than the 

 wave-length, will be very feebly heard, and it is for a similar 

 reason that the tone of a tuning fork has to be reinforced by a 

 resonance case. After all it seems very probable that the light of a 

 flame is in every case caused by vibrations in which there is a 

 variable electric moment in a definite direction, and which may in 

 so far be called of the first order, thougli they need not depend 

 precisely on a spherical harmonic. If this principle be admitted, it 

 may be shown that, along the lines of force, only those components 

 remain visible which are polarized in the direction of these lines , 

 when viewed across the field. 



§ 12. Seeking for some means by which the vibrations of the 

 second order might be made to reveal themselves in the spectruui, 



24 



I'roccediugs Royal Acad. Amstei'dam, Vol. I. 



