DECEMBER 8, 1899. ] 
reducing my measures, with the aid of 
Hartmann’s formula, I obtained a result in 
substantial agreement with his. The wave- 
length is 
For East side, 4 5303.21 
For West side, 5303.32 
Mean, 4 5303.26, 
which should not be in error by more than 
=E 0.15 tenth meters. The difference in the 
determinations for the two sides corresponds 
to a relative velocity of 6.2 km. in the line 
of sight, or a rotational velocity of 3.1 km. 
persecond. However, I regard this result as 
subject to a possible error of at least + 2 km. 
per second, partly on account of unavoidable 
errors of observations, but principally on 
acount of the ill-defined and unsymmetrical 
character of the bright line. 
The continuous spectrum of the inner 
corona was recorded out to a distance of 
2.'5 on the east side and 1.’5 on the west 
side. While the dark lines in the recorded 
comparison spectra are sharp and strong, 
there is not the slightest trace of dark lines 
in the recorded continuous spectrum of the 
corona. This radiation seems to be of coro- 
nal origin, and is not due to reflected pho- 
tospheric radiations. 
In explanation of the error in the ac- 
cepted value of the wave-length (2 5317) of 
the green coronal line which has prevailed 
for many years, it is suggested that the true 
coronal line would be difficult. to observe so 
long as the chromospheric spectrum was 
visible. Hence the observers, setting on the 
strongest chromospheric line in this region, 
at 45317, which is very conspicuous just be- 
fore and at the instance of totality, assumed 
it to be identical with the true coronal line, 
and located it at 1474K. Later, when this 
line had disappeared, rather suddenly, and 
the background had become dark enough to 
allow the line at 4 5303 to be seen, the ob- 
servers. were interested in the extent and 
other properties of the line and no further 
SCIENCE. 
845 
micrometer settings were made for fixing its 
wave-length, (Published in the Asérophys- 
ical Journal. ) 
Epwin B. Frosr: Notes on the Reduction of 
Stellar Spectra. 
The advantage was pointed out of using 
Hartmann’s interpolation formula for the 
prismatic spectrum in the reduction of spec- 
trograms taken either for the determination 
of wave-lengths or velocity in the line of 
sight. An outline was given of the pro- 
cedure adopted in the latter work with the 
40-inch refractor. Each plate is reduced by 
itself, independently of any solar or other 
plate, such as many observers have used as 
auxiliary in the process of reduction. The 
comparison spectrum from a spark between 
metallic electrodes, impressed upon each 
plate, furnishes all the data necessary for 
reduction. It is hoped that errors due to. 
effects of temperature on the dispersion of 
prisms and focus of lenses are thereby re- 
duced, and that systematic errors are also 
diminished. 
Attention was called to the favorable re- 
sults obtained from the use of titanium as a 
source of the comparison spectrum. The 
lines are numerous and well distributed 
throughout the upper spectrum. The metal 
stands third in Rowland’s arrangement ac- 
cording to number of lines in the solar spec- 
trum. The spark passes with great readiness. 
between electrodes of metallic titanium, and 
the air lines, which are annoying in case of 
the spark spectrum of iron, are not pro- 
duced. Sharp titanium lines are found to 
fall at points in the spectrum close to the 
positions of the principal lines of stellar 
spectra of Type 10, and thus facilitate the 
reduction of such spectra. 
Corrections to Determinations of Absolute 
Wave-length. 
The effect upon the absolute wave-length 
of lines in the solar spectrum of the eccen- 
tricity of the earth’s orbit seems to have 
