DECEMBER 1, 1899. ] 
Gr. M. T. 1896, Sept. 87 22.8 —20.1 km. 
Oe ley PPA} —19.1 
6 23 21.4 —18.9 
Oct. 5 21.0 —19.0 
Nov. 11 19.3 —20.1 
Dec. 8 16.7 —20.3 
Mean —19.6 
The agreement of these results was satis- 
factory, and gave no evidence of variable 
velocity. 
In order to test the current results of our 
work, another photograph of the spectrum 
of Polaris was obtained on Aug. 8, 1899. 
This yielded a velocity of —13.1 km., and 
led to the suspicion that we were dealing 
with a variable. Two additional plates 
were secured on August 9th and 14th, which 
yielded velocities of —11.4 and —9.0 km., 
respectively. Inasmuch asa range of 4 km. 
is not permissible in the case of such an ex- 
cellent spectrum, the star was suspected to 
be a short period variable, and further ob- 
servations were obtained, as below : 
Gr. M. T. 1899 Velocity Measured by 
August 92 0.84 —13.1 Campbell 
9 20.1 —11.4 Campbell 
14 22.8 — 9.0 Campbell 
16 0.1 —14.1 Campbell 
23 0.3 —10.9 Campbell 
24 0.8 —15.2 Campbell 
26 0.9 — 9.4 Campbell 
we — 8.6 Wright 
27 0.3 —10.6 Campbell 
27 16.2 —14.0 Campbell 
28 0.8 —14.7 Campbell 
le —14.3 Wright 
28 16.3 —13.7 Wright 
29 0.4 —12.1 Wright 
29 18.8 — 9.6 Wright 
30 0.0 — 8.9 Wright 
30 16.2 — 9.3 Wright 
* Measures of the same plate by Mr. Wright. 
On plotting these observations it became 
evident that Polaris is a spectroscopic 
_binary, having a period a little less than 
four days. The 1899 observations have 
been collected and plotted on the assump- 
SCIENCE 791 
tion that the period is 3° 23". The velocity, 
at present, seems to be included between 
—8.6 and —14.6 km., having an extreme 
range of only 6 km. The velocity of the 
binary system seems to be about 12 km. 
The determinations of velocity made in 
1896 lie entirely outside of the present range 
of values, and leave no doubt that the 
velocity of the binary system is changing 
under the influence of an additional dis- 
turbing force. I think it is certain, there- 
fore, that Polaris is at least a triple system. 
The 1896 observations were made at in- 
tervals differing but little from multiples of 
the period of the binary system, and there- 
fore fell near the same point in the velocity 
curve. Assuming a period of 3'23" +, there 
is no difficulty in selecting the epoch of 
minimum so that these six observations will 
fall on the curve satisfying the 1899 obser- 
vations. The residuals will be negligible if 
we assume the observations to fall near the 
lower part of the curve; and I have no 
doubt that future determinations of the 
orbit will definitely place them there. It 
will be seen that the velocities of the bi- 
nary system in 1896 and in 1899 differ about 
6 km. 
The Spectroscopic Binary Capella. 
An examination of six spectrum plates of 
a Aurige, obtained with the Mills’ Spectro- 
graph in 1896-7, leaves no doubt that this 
star is a spectroscopic binary. The spectrum 
is composite. The component whose spec- 
trum is of the solar type furnished the fol- 
lowing velocities with reference to the solar 
system : 
1896, Aug. 31 -+ 34 km. 
Sept. 16 + 54 
Oct. 3 +- 49 
Oct. 5 ++ 44 
Nov. 12 + 4 
1897, Feb. 24 + 3 
On the first photograph the spectrum is 
of essentially normal solar type; on the 
