142 LYMAN. — THE SPECTRUM OF HYDROGEN. 
investigate light of shorter wave-length than this value it is necessary to turn the grat- 
ing, a process which necessitates a slight change in the adjustment of slits and plate. 
To check the values obtained in the above manner lines of short wave-length 
were obtained in the second spectrum. For this purpose the left hand slit was 
covered by a discharge tube without a window and the whole apparatus was filled 
with hydrogen exactly as usual. Owing to the feeble character of the second spec- 
trum only the stronger lines between X1550 and X1200 could be photographed. 
Their position was determined by comparison with first spectrum iron lines obtained 
from light which had passed directly through the discharge tube. The average differ- 
ence between the values obtained by the two methods is 0.3 Angstrom unit. 
The plates have been measured on an engine made by Wolz of Bonn after the 
design by Kayser. The screw has been calibrated and proves to be of an accuracy 
far greater than this work demands. The intensities of the lines have been estimated, 
first, by observations made under the reading microscope, and, second, by projecting 
the spectrum on a screen. The latter method has the advantage that the whole 
pectrum is before the observer at one time. The values were estimated from plates 
aken without a fluorite window. The tables are divided into two parts. In the first 
lmi 
second 
given 310 lines lying in that region from which it has been possible to 
lines due to impurities. The error should not be greater than 0.3 Angstrom. In 
beyond the transparency of fluorite ; their orig 
not absolutely known, but they are probably due to hydrogen, since they were obtained 
when the discharge tube was connected directly with the spectroscope, a condition 
under which air lines rarely occur. The error in these values should not be greater 
than one unit. The values of the iron lines are from the measurements of Exner and 
Ilaschek as given in Watts' Tables ;* the correction to 
The wave-lengths of the aluminum lines are from the measurements of Ed 
vacuum came from the same 
d Val 
The agreement between the tables and the numbers given in the " Preliminary 
Measurements" is well within the accuracy claimed for the earlier values. 
SCHUMANN'S SPECTRUM. 
In order to compare the prismatic spectrum obtained by Schumann with the values 
ot the table the twelve plates published in the Smithsonian Memoir 7 were cut out and 
pasted together. The resulting spectrum, some 127 cm. long was placed , 
6 Index of Spectra, W. M. Watts, Appendix J. 
Bemftge zur Photochemie, p. 388. 
7 Smithsonian Contributions, No. 1413. 
pon 
