132 
ASTRONOMY: ST. JOHN AND BABCOCK 
These facts and others associated with them led us to compare the 
spectra of different parts of the arc with apparatus of high power, and a 
method was developed for making studies of this kind under conditions 
favorable to the detection of very small variations in the spectrum. 
Even with very stable apparatus, considerable difficulty is found with 
high dispersion spectrographs in eliminating minute instrumental dis- 
turbances sufficiently large to interfere seriously with observations of 
quantities whose order of magnitude is a few thousandths of an angstrom, 
and it was not until we made our exposures to the two parts of the source 
simultaneous that we obtained satisfactory consistency in the results. 
This was accompKshed by means of a pair of small total reflecting 
prisms arranged over the slit of the spectrograph in a 5-fold enlarged 
image of the arc, with a rotating sectored disk interposed in the path of 
the light from the brighter portion of the source. The latter adjunct 
serves to equalize the effective intensities so that the corresponding spec- 
tral images are practically identical in blackness — a necessary precaution 
when their positions on the photographic plate are to be determined by 
visual observations with a ffiar micrometer. The same auxilliary appa- 
ratus was employed for comparing the spectra of two separate arcs oper- 
ated with different current strengths, and it may be appHed to many 
comparisons requiring the highest precision. By the aid of it we have ob- 
tained photographs of the entire visible spectrum of iron and part of the 
ultraviolet, the spectrum of the Hght from near the negative pole appear- 
ing as a narrow strip extending along the center of the plate, with contigu- 
ous spectra on each side taken from the midpoint of the arc. Upon these 
plates we have measured for about 1600 lines the wavelength at the neg- 
ative pole using the same line at the center of the arc as a reference stan- 
dard. Of the Hues examined about 1300 show no determinable difference 
in wavelength, 249 are displaced toward the red and 64 toward the violet, 
the larger shifts amounting to about -|- 0.025 angstrom and —0.030 ajig- 
strom respectively. The presence upon the same photograph of Knes 
belonging to all three of these classes establishes the fact that many 
lines in the spectrum are modified in some way at the pole of the arc to 
an extent which must be taken into account when they are involved in 
precise measurements. It becomes of importance, therefore, to enum- 
erate and classify these Hues and to determine under what conditions, 
if any, they may be used as standards of reference. Also, the question 
may be raised as to whether the shifts we observe are actual displace- 
ments of the maxima of the lines or are due merely to unsymmetrical 
widening. Concerning this latter point it should be said that especial 
