ASTRONOMY: J. STEBBINS 
261 
between centers is approximately 5 times their average radius, whose period 
of revolution is about 4 days, and whose mean density is 1 /20 that of the 
sun. Systems of greater or less separation are not so numerous, or we 
should find more of them among the eclipsing variables. 
The foregoing discussion is based upon the variables which have been 
found by visual and photographic methods, but there is abundant field 
for work in the same lines for the new forms of photometer. Here in- 
stead of examining thousands of stars and picking up a variable now and 
then, it will be better to make a special study of a few stars which from 
spectroscopic results are known to have large companions. I have made 
a start in this testing of spectroscopic binaries, and the first two stars 
observed were found to be variables. Further observations yielded 4 
variables out of 11 stars, and among the hundreds of spectroscopic 
binaries already known, there are no doubt dozens of variables awaiting 
discovery. The point to emphasize is that not only will systematic 
studies of stars which vary in light give us direct information, but indi- 
rectly we can draw far reaching conclusions about stars which do not 
vary. 
In addition to the eclipses, other causes of Kght-variation of binaries 
may be mentioned, such as elHpsoidal figure, orbital eccentricity produc- 
ing changes in the interaction of radiation of the components, tidal 
phenomena, and the effects of resisting media. There is a class of ob- 
jects Hke Polaris, whose spectral variations are accompanied by syn- 
chronous light changes, but there is the possibility that these are not 
double systems. Some day we shall probably be able to assert that all 
stars with rapid changes in their spectra, whether these be in the shift 
or the character of the Hues, must vary in light, and it will become a 
practical problem to find really constant stars to be used as standards 
of comparison. These may be found only by trial, and we shall have 
to observe stars in groups so as to eHminate the variables. No doubt 
we shall find a difference in the average amount of variation as we pass 
along the spectral series. 
This takes us to the case of our sun, which, according to Abbot, is 
a variable star. Of the variation there can not be the sHghtest doubt, 
for a single sunspot is enough to change the total light, the only question 
is how much. In direct measures of the sun's radiation, the chief dif- 
ficulty lies in the proper allowance for the atmospheric absorption, but 
this trouble may be eliminated by comparing the reflected light of one 
of the planets with the light of a number of stars. Probably Saturn 
is a good object for this purpose, as there are few markings on the sur- 
face, but Uranus would be still better on account of its slower motion, 
