156 
ASTRONOMY: W. S. ADAMS 
The columns in the tables designated by A and B refer to the deter- 
minations by Adams and Miss Burwell. The final values are the means 
for the two observers. The measured parallaxes are taken from a va- 
riety of sources. Y. indicates Yale determinations; K., the values com- 
piled by Kapte)ai in Groningen Publication No. 24; Sch., the results of 
Schlesinger; R., those of Russell; vM., of van Maanen; S., of Slocum; M., 
of Mitchell; J., of Jost; and F., those of Flint. Where relative parallaxes 
are given the values have been reduced to absolute measure by making 
suitable corrections for the parallaxes of the comparison stars. The 
tables are arranged according to spectral type. 
The comparison of the computed and the measured parallaxes shows 
an excellent degree of accordance for most of the stars. There are, 
however, occasional large discrepancies. Of these the most serious is in 
the case of 5 Eridani. The spectrum observations give a much smaller 
parallax than is found by the Yale observers. A striking case of agree- 
ment, on the other hand, is that of e Eridani; this parallax was computed 
before it was known that a measured value was available. A star which 
should prove of exceptional interest is Boss 6129. From spectrum ob- 
servations we have obtained a parallax of -f 0!'23: no measured value 
has been published but the star is on the observing programme at sev- 
eral observatories. 
The average deviation, taken without regard to sign, between the 
observed and the computed values of the parallaxes in Tables I and 
II is 0r024: it is Of 026 for the stars of Table II alone. There seems to 
be no marked systematic difference between the observed and the com- 
puted parallaxes; the former average somewhat larger, but this is due 
mainly to a few large discrepancies. 
There are 25 stars with measured negative parallaxes for which we 
have made spectrum determinations. The largest value for any one 
of these stars as computed from the line intensities is -fOfOS; the aver- 
age value for all is +0!'03. The spectrimi method, of course, gives no 
negative parallaxes. 
It seems reasonable to conclude from these results that the method 
of computing absolute magnitudes and parallaxes from the variation of 
the intensities of Hnes in stellar spectra is capable of yielding results of a 
very considerable degree of accuracy. Especially in the K and M type 
stars of low luminosity, the line variations are so great that such stars 
may be recognized from a mere inspection of the spectrum. Stars, for 
example, like 61 Cygni, Groom. 34, and Kriiger 60 bear very evident 
marks of their intrinsic faintness in the remarkable intensity of the low 
temperature calcium lines in their spectra. At first thought it might 
