PRKSI])ENTIAL ADDRESS— SECTION A. 37 



became evident to Kapteiju in 1904, that the motions of the 

 stars conkl not be considered to be entirely at random, and 

 at the joint meeting of the British Association for the 

 Advancement of Science and the South African Association 

 for the Advancement of Science at Capetown in 1905, Kapteijn 

 announced tliat there Avere two preferential directions of 

 motion amongst the stars.. This statement can be explained 

 in a variety of ways. If the motions of the stars could be 

 plotted on a three-dimensional diagram, with vectors repre- 

 senting the velocity and direction of motion of each star, 

 the resultant surface would be an ellipsoid with one relatively 

 long axis. This long axis of the ellipsoid would represent 

 Kapteijn's preferential directions. More simply, it might be 

 considered that there are two directly opposed streams of 

 stars, which was the view taken by Kapteijn, who called 

 the two streams Drift I and Drift II. Drift I is directed 

 towards 90°, -12°, and Drift II towards 270°, -55°. These 

 directions are affected by the motion of our point of 

 observation — the Sun — and, if tliev are corrected for this, 

 the directions are found to be 94°^ +12°, and 274°, -12°, 

 opposite to one another and in the plane of the Galaxy — which 

 is a fact full of significance. 



This discovery of systematic motions amongst the stars 

 has greatly complicated the question of determining stellar 

 parallaxes from proper motions, but has carried us much 

 further in investigations into the structure of the Universe. 



The investigation of systematic motions amongst the 

 stars has occupied us at the Union Observatory for the past 

 three years, and the portion of work we have undertaken 

 is by no means completed yet. Ordinarily the process of 

 determining proper motions is a very slow one. Meridian 

 observations of star places are made at an Observatory and 

 published as a star catalogue for a certain ei^och. The work 

 is rejieated at the same or another Observatory at a later 

 date and a catalogue published for a later epoch. By bringing 

 the places of the stars to a common epoch, the difference 

 between the two places for the same star, after all possible 

 sources of error have been eliminated, is due to the propei^ 

 motion of the star. This is a slow and tedious process, and 

 it is never advisable to determine a small quantity as a 

 difference between much greater quantities. Further, and 

 this is very important, only those stars are dealt with which 

 are bright enough to be observed Avitli meridian ii\struments. 

 The results would thus be incomplete, and only refer to the 

 brighter stars. 



It is a dift'erent matter when star photographs are dealt 

 with. Suppose that two photographs have been taken Avith 

 the same telescope of the same region of the sky, and Avith 

 an interval of some years between them. These tAvo photo- 

 graphs can be optically superposed and minutely compared 

 AA'ith one another. If there were any motion common to all 

 the stars of the region, this could not be detected, because 

 the superposition of the tAvo plates would automatically 

 eliminate this. Such a common motion could be determined 



