DECEMBER 8, 1899. ] 
lowing it, and as nearly as possible on the 
same parallel of declination with it, first 
at a rather large hour angle when east of 
the meridian and again in the morning 
hours when west. Now it is evident that if 
the red star is lifted less by refraction than a 
comparison star which precedes it in right 
ascension, the distance between them on the 
east side of the meridian will be greater 
than it would be if both stars were white 
and less on the other, and vice versa if the 
comparison star follows it. By taking two 
comparison stars as described above we are 
able to eliminate any change in the scale 
value, any variation in the atmospheric re- 
fraction, and the difference of the two meas- 
ured distances would evidently give double 
the effect we are considering. 
Now the refraction of a star of average 
color may be represented by 
Rh =f tan z. 
The refraction of a star with light of a 
different refrangibility would then be 
R= (6+ dg) tan z. 
It will therefore be easily seen that the 
refraction correction to the observed distance 
between two stars of different refrangibility 
should receive an additional correction 
— df tan z cos (p— q), 
z,p and q having their usual significations 
of zenith distance, position angle, and par- 
allactic angle, respectively. 
To each measured distance it would then 
be simply necessary to add the term 
— df tan z cos (p— q) 
and to find the value of df. 
Now for finding of the value of dj the 
observations of each of the five red stars 
which were to be observed four times on 
each side of the meridian, would furnish 
eight equations of condition of the form 
z+oed?=n, 
where x is the necessary correction to the 
assumed value of the difference between 
_SCIENC 4. 
843 
the distances from the two comparison 
stars ; ais the value of tan z cos (p — q), and 
nm is the observed difference of the two dis- 
tances minus the assumed difference. j 
Combining the normals derived from these 
equations of condition for the observations 
of both last year and this, and solving, I 
find the following values for d? : 
Star. | Redness. dg. | Wt. 
R, 6.0 + 0.//020 + 0.//015 63.6 
R, Pos — 0.77008 + 0.//014 16.0* 
Re 7.0 —0.015 = 0.020 64.4 
Ry 8.7 —0.046 +0.018 | 45.2 
1B, 7.8 +0:008 = 0.024 | 5bai 
Average probable error 1 observation = + 0.151. 
A separate investigation was made along 
with that of star R,, which was specially 
selected because it had a close neighboring 
white star. The distances of the white 
star from the same comparison stars were 
measured on the same nights and the ob- 
servations were made symmetrically with 
respect to those of the red stars, so that the 
conditions were absolutely the same for the 
two stars. 
Similar equations for this white star gave 
df= + 0.004 = 0.022, wt. 55.6 
as compared with + 0.008 =: 0.024 for R,. 
The small values above found for d? (in 
the mean for five red stars we find d3= 
— 0.006 + 0.010), and more especially 
the fact that two contiguous stars, one white 
and the other red, give no appreciably dif- 
ferent results, afford rather forcible evi- 
dence that to my eye, at least, difference of 
color in the stars does not effect heliometer 
observations of distance. 
J. E. Keever: The Ring Nebula in Iyra. 
In order to test the capabilities of the 
Crossley three-foot reflector of the Lick Ob- 
servatory, a number of photographs were 
made of well-known celestial objects. As 
the focal length of a camera should be from 
thirty to sixty times its aperture in order 
* Not observed in 1899. 
