86 



SCIENCE. 



[N. S. Vol. XVI. No. 394. 



however, that this reh^tion would hold for 

 absorption by some other mode. We may 

 then feel some hesitanej^ in accepting this 

 relation of the group-velocity to the wave- 

 velocity as based on either the kinematical 

 analogy or the energy argument. We must 

 therefore fall back for the solution upon 

 direct experimental means. The signifi- 

 cance of such an experimental solution to 

 the problem of the propagation of undxila- 

 tions should not be underestimated. In- 

 vestigations of these two velocities for 

 monochromatic light, such as from cad- 

 mium or merciiry, in highly dispersing sub- 

 stances like carbon disulphide or alpha- 

 monobromouapthaline or dense glass of 

 Faraday, now seem entirely possible and 

 sufficient for the solution of the problem. 



In the case of the ether the arguments 

 which can be advanced in regard to veloc- 

 ity of light for different colors indicate 

 the same velocity for all colors. It was 

 pointed out by Arago that any difference 

 in velocity shoxild prod\;ce a coloration in 

 anj' luminous body in the i\nivei*se which 

 should vary rapidly in intensity. Thus in 

 the observations on the eclipse of Jupiter's 

 satellites they should momentarily show at 

 the instants of disappearing and reappear- 

 ing a coloration complementary in the two 

 eases. Nothing of this kind has been re- 

 corded. Again in the case of Algol, New- 

 comb estimated, from its probable distance, 

 —greater than 2,000,000 radii of the earth's 

 orbit and the time for light to reach us, 

 30 yeax'S — that a difference in time between 

 the blue and the red rays of one hour 

 would give a dift'erence in velocity of fo\ir 

 parts in a million. In the remarkable 

 changes in Nova Persei last year, its com- 

 plete spectrum appeared to be ■sdsible even 

 though the changes in its intensity were 

 far more rapid than in the case just men- 

 tioned. As no trace of coloration has 

 ever been observed this difference of time 



cannot exceed a fraction of an hour. It 

 should be mentioned, however, that in the 

 experiments of Yoixng and Forbes, the 

 velocity of blue light was apparently about 

 1.8 per cent, greater than that of red light. 

 Both Michelson and Newcomb claim that 

 tins would have been verj^ distinctly ob- 

 served in their experiments with the rotat- 

 ing mirror in the spreading out of the im- 

 age of the slit into a distinct spectrum. 

 A further instance is cited by Lord Ray- 

 leigh which may be of value. If we ex- 

 amine the position of the bands in the 

 spectrum of glowing gases, we find certain 

 harmonic relations. Now if these rays 

 had dift'erent velocities in the free ether 

 the position of the bands would be aft'ected 

 and the harmonic relations, apparently hold- 

 ing as deduced from the spectra observed, 

 would not give the harmonic relations in 

 the radiants themselves ; or, vice versa, such 

 harmonic relations between the radiants- 

 would not give harmonic distribution 

 of the bands in the spectrum. From an- 

 other standpoint it may be mentioned that 

 on any theory of an optical medium we 

 Imow that either a difference in velocity or 

 a dispersion requires incomplete transmis- 

 sion. This may be due to internal reflection 

 or to transformation into heat. The trans- 

 mission woiild also be differential. Thus 

 only a part of the light of the distant stars 

 would reach the eye and tliis would be 

 more and more colored as the distance in- 

 creased, due to the differential transmis- 

 sion. No eft'ect of tJiis kind can be ob- 

 served even in the nebiila? which are so re- 

 mote that the telescope cannot resolve them,, 

 althoiigh the spectroscope gives us unques- 

 tioned evidence of their stellar nature. 



These argiunents from the astronomical 

 point of Aiew are, however, uncertain and 

 indirect. Until we can determine to a close- 

 approximation the wave-velocities of dift'er- 

 ent coloi'ed rays in ponderable media as.- 



