ASTRONOMY: F. H. SHARES 
zero point. This was accomplished by making the mean brightness of the 
white stars near the sixth magnitude equal to the mean of their visual 
magnitudes in Harvard Circular, No. 170, proper allowance for spectrum 
being made for those objects not of type AO. 
The resulting catalogue of photographic magnitudes includes 617 
objects, the great majority of which are fainter than the tenth magni- 
tude. All of the latter are close to the Pole of 1900, and among them are 
included the stars of the Polar Sequence. To these special attention 
has been given, although a large number of other stars have well de- 
termined magnitudes, based on ten or more individual values. The 
very faint objects, which are relatively numerous, are naturally subject 
to greater uncertainty. 
The treatment of the data for the photovisual scale was exactly the 
same as that outHned above for the photographic scale. Although 
only 6 plates were used for the intermediate and faint stars, the pre- 
cision is much greater than one would thus be led to expect. Owing 
to the better definition of the images on the isochromatic plates, the 
probable error of a photovisual magnitude is only two-thirds that of a 
photographic result. The photovisual scale for the bright stars, which 
depends upon 122 plates and 367 individual magnitudes, is probably quite 
as reliable as the photographic scale in the same region. Final photo- 
visual magnitudes were obtained for 339 stars between magnitudes 2.1 
and 17.5. 
The average deviation of a single photographic magnitude, including 
the scale error of a plate over a range of 8 or 9 magnitudes is 0.125 mag. ; 
the corresponding photovisual result is 0.087 mag. 
About 300 stars have had both their photographic and photovisual 
magnitudes determined, and, in consequence, color indices for these 
are available. These confirm a result announced at an earlier stage of 
the work, namely, that the faint stars are all relatively red. For the 
bright stars, we have small, and even negative, color indices in large 
numbers. At the seventeenth photovisual magnitude the smallest value 
is 0.6 or 0.7 mag., the lower Hmit gradually increasing as the fainter 
stars are approached. This phenomenon is undoubtedly of very great 
significance, but its interpretation cannot at present be given with cer- 
tainty. It is evident, however, that the determination of star colors 
will form an important part of stellar investigations in the future. 
This last consideration only emphasizes the necessity of precision 
in the magnitude scales, for any relative error enters to its full amount 
into the color index, and thus vitiates the results of comparisons in- 
volving stars of different degrees of brightness. In the present case a 
supplementary investigation provided an important control. 
