Intelligence and Miscellaneous Articles. 401 



then reproduce the modified colours, and the effect would be im- 

 perceptible. 



But the result is very different for rays produced by the combus- 

 tion of various substances in a luminous body. A substance capable 

 of producing, for instance, the ray F, which corresponds to blue and 

 to a wave-length of 486*39 millionths of a millimetre, would change 

 its position in the spectrum, owing to the motion of the star ; for 

 while the vibrating molecule which produces it would oscillate in 

 the length of time which is peculiar to it and invariable, the wave 

 itself would be lengthened or shortened by the transport of the ra- 

 diant-point, and therefore would change its refrangibility ; compa- 

 ring it to the same substance radiating near the observer, a differ- 

 ence would be found in the place of the ray. Thus, if the motion 

 were such that it ought to increase the wave-length by 40" 63 mil- 

 lionths of a millimetre, the ray F would pass into the place of the 

 ray E in the spectrum, and the colour would be green. To produce 

 a similar change, the star should have in moving away a velocity of 

 31,000 kilometres in a second, and, in approaching, a velocity of 

 about 16,000 kilometres. As the earth only traverses about 30*4 

 kilometres in the same time, the star should have in the first of the 

 two cases a thousand times that velocity. 



But our present spectroscopic instruments enable us to measure 

 much smaller intervals. Let us suppose that the spectroscope is one 

 which sharply divides the ray D (which is the case with those I am now 

 using), a difference in position equal to the magnitude of this double 

 ray would doubtless be measurable. Now the two rays D' and D", 

 according to M.Yan der Willigen, are at a distance of 4 ten-millionths 

 of a millimetre. A displacement of this amount (taking the velocity 

 of light at 300,000 kilometres in a second) indicates that the star 

 has a velocity of 304 kilometres in moving away — that is, a velocity 

 ten times that of the earth. Half as much would be sufficient if it 

 were approaching. 



These velocities are by no means enormous, but still they are far 

 from those which we must assume from the proper motions of the 

 stars. Thus the star 40 of Eridanus, which has four seconds of proper 

 annual motion in the great circle # , if it were supposed to be at the 

 distance of the stellar parallax of half a second, would only traverse 

 36 kilometres in a second. The question being very delicate, very 

 accurate means of observation were necessary for determining the 

 absolute position of the rays. 



This determination may be made by means of artificial lights 

 used with the spectrometer with a slit ; but this means is embarrass- 

 ing in practice, and presents many inconveniences which need not 

 be detailed here. After several attempts, I have found that we suc- 

 ceed more easily by introducing into the field of the telescope the 

 spectrum and the direct image of the star, and comparing them at a 

 fixed point in the field itself. Either the line F or the line E may 

 be chosen ; for they are well known from the systems they belong to 

 (hydrogen and iron lines), so that we can be sure there is no error 

 * Stmve, Cat, Dorp. p. 153. 



