19I2-] DEPTH OF THE MILKY WAY. 3 



angular dimension in the sky, and thus parallaxes might be found 

 of systems very remote from the earth. In 1890, while a post- 

 graduate student at the University of Berlin, I developed this method 

 still further, and showed how it could be used also to test the accu- 

 racy of the law of universal gravitation in the stellar systems. The 

 spectroscopic method then outlined was brought to more general 

 form in 1895, and it at once occurred to me to point out its use for 

 measuring the distance of clusters in the Milky Way (A. N. 3,323), 

 as more certain than Herschel's method of star gauges. 



Our age is one of rapid improvement in all scientific processes, 

 and during the past sixteen years naturally much progress has been 

 made in double-star astronomy, as well as in our knowledge of 

 nebulae and clusters. On looking more closely into the spectroscopic 

 method, which in 1895 had been shown to be applicable to objects 

 1,000 light-years from the sun, and might thus include all suitable 

 double stars within this sphere, I became convinced that while it is 

 a great theoretical advance over the old method of parallaxes, it 

 still is quite inadecjuate for finding the distances of the most remote 

 objects in the sidereal universe. Accordingly in 1909 I returned to 

 the improvement of Herschel's method as the most promising, for 

 the determination of the distances of the most remote objects. Here 

 are the grounds for this decision : 



1. It was noticed, as remarked by Burnham, that revolving double 

 stars are rare, if not unknown, in clusters, and among the star-clouds 

 of the Milky Way- — ^not because such systems are not present in 

 these masses of stars, but because they cannot be separated, owing 

 to the great distances at which these masses of stars are removed 

 from us. 



2. When double stars cannot be clearly separated in the tele- 

 scope they cannot be used for parallax by the spectroscopic method ; 

 and thus the spectroscopic method, while having a wider range of 

 application than the method of parallaxes, in something like the 

 ratio of the size of the double star orbit to that of the orbit of the 

 earth, is yet applicable only to stars within about 1,000 light-years 

 of our sun. 



