DEPARTMENT OF MERIDIAN ASTROMETRY. 



243 



Wash. Pub. No. 10). While the theoretical part, as already stated, has 

 been prepared for some time, the practical application and computation 

 were not started until recently. As the computations are only half 

 completed no results can be incorporated in the present report. 



Dr. Albrecht is preparing an article on ''A method for rendering 

 stellar radial velocities homogeneous within each spectral type." It is 

 to a limited extent known that perfect homogeneity does not exist in 

 stellar radial velocity results, even when we confine ourselves to stars 

 of one spectral type and to one spectrograph. However, no remedy for 

 this unsatisfactory state of affairs has been suggested. Though the 

 more general problem is quite complex, the greater immediate need and 

 prerequisite is to place the radial velocities within each stellar type 

 upon a well-defined and uniform basis. The restricted problem yields 

 to a very simple and effective solution. 



The values of the radial velocity given by the individual lines are 

 plotted with velocities as ordinates and wave-lengths as abscissae, and 

 a smooth curve is drawn through the plotted points. In actual prac- 

 tice it is found that this curve deviates quite appreciably from the 

 horizontal straight line which represents the mean of all the hnes, i. e., 

 the radial velocity for the spectrogram, and these deviations measure 

 the extent to which an internal non-homogeneity exists in the reduction 

 of this spectrogram. This internal non-homogeneity — the causes for 

 which need not be dwelt upon here — together with a non-uniformity in 

 intensity which makes it impossible in practice to always employ the 

 same identical Ust of hnes on all plates of any one star or of all stars of 

 one spectral class, constitutes the principal factor which introduces 

 the non-homogeneity considered in this article. 



9 lines in 6 give Mean +0.74 km. 13 lines in c give Mean —0.37 km. 

 Fig. 2. — Composite of Individual Curves. 



In order to eliminate from the above curve deviations of an accidental 

 nature, the same process is repeated for a representative number of 

 spectrograms and a curve is constructed which is a composite of all the 

 individual curves. Such a composite curve is reproduced in the figure. 

 It is evident that the three sections into which this curve has been arbi- 

 trarily divided yield radial velocities differing quite appreciably from 

 each other and from the mean. 



