56 Transactions of the Boyal Society of South Africa. 
elements which pertain to individual stars. Thus we arrive at the, at first 
sight paradoxical, conclusion that, while the first drift stars are distributed 
at random throughout the sky, the very irregularities pertaining to the 
second drift point to some kind of structural unity in this drift. 
Before attempting to explain this further I wish to draw attention to 
an entirely different method of investigation by which the phenomena I 
have been discussing may be examined. 
I have concerned myself hitherto only with the visible motions of the 
stars transverse to the line of sight, as derived by the older methods of 
measurement. The introduction of the spectroscope into astronomical 
research has opened up vast new fields into which, so far as they relate 
to the chemical and physical constitution of the sun and stars, it is not 
my purpose to enter to-night. What I wish rather to emphasise is the 
value of this instrument as a supplement to the older methods in relation 
to the geometrical astronomy of position. 
In accordance with the principle laid down by Doppler, the wave- 
length of light received from a source which is either receding from or 
approaching a receiver will appear to be modified by an amount dependent 
in a known manner on the velocity of approach or recession. If the 
receiver takes the form of a spectroscope which permits by any means, 
direct or indirect, of the measurement of the wave-lengths, and the 
normal wave-lengths of the lines under examination are independently 
determined by laboratory investigations, the difference between the ob- 
served and the normal wave-lengths will thus afi'ord a means of 
measuring the velocity of approach or recession of the source of light. 
Of the precautions necessary to ensure precision it is not my purpose 
to speak to-night. The large spectroscope of the Cape Observatory, which 
we owe to the munificence of the late Mr. Frank McClean, was from the 
outset constructed with due regard to these precautions, so far as they 
could be foreseen, for the purpose of determining with the greatest 
accuracy attainable the radial velocities of stars. The instrument has 
been already successfully used, and its capabilities have been established 
in an investigation of the aberration constant of light as depending on the 
apparent variations in the radial velocities of stars resulting from the 
earth's orbital motion. 
From a relatively short series of observations discussed by my colleague 
Dr. Halm, this constant has been derived with a precision not inferior to 
that attained by the best series of older observations, and the capabilities 
of the method are yet far from exhausted. 
At the present time the instrument is being devoted to a series of 
observations of all such stars as are accessible in the southern skies, 
whose spectra present sufficiently pronounced features to admit of 
measurement, primarily with a view to ascertaining what evidence can 
