354 PROF. A. FOWLER AND MR. C. C. L. GREGORY ON THE ULTRA-VIOLET 
by a metal disc having a small aperture through which the light of the arc could 
pass to the spectrograph. A constant stream of ammonia, which was drawn off by a 
water pump, was caused to flow through the apparatus. The arc was on a 200-volt 
circuit, and the current used was 1‘8 amperes. The electrodes were of copper. 
The grating was mounted so as to give a normal spectrum, and precautions were 
taken to avoid mechanical displacements. The temperature of the laboratory was 
also kept as constant as possible during the exposures, which were of about 
60 minutes’ duration. An internal shutter, detached from the other parts of the 
spectrograph, was arranged next the plate to allow of two comparison spectra being 
photographed, one before and one after the ammonia spectrum. 
A photograph of the copper arc in air, which was taken to facilitate the elimination 
of lines introduced by the electrodes, showed that the copper employed contained 
impurities of nickel and silver. In the case of very strong lines, “ ghosts ” were also 
present, and there were a few lines belonging to the second and fourth order spectra; 
these are marked by double dots in fig. V., Plate 2. 
The ammonia band occurs incidentally in an excellent photograph of the magnesium 
arc under reduced pressure, taken in 1913 by Mr. Yf. Jevons, in the 4th order of 
i? 
the 10-foot grating; the band is here somewhat confused by nitrogen, but many of 
the ammonia lines can be quite certainly identified, and the high resolution has been 
of special value in connection with very close groups near the extreme ends of the 
band. 
General Description of the Band. 
The general features of the ammonia band will be best gathered from the 
photographs reproduced in Plate 2. With low dispersion, as will be seen from fig. I., 
the band resembles a double line having components of unequal intensity, but there 
are indications of banded structure on both sides in photographs which have received 
sufficient exposure. With somewhat higher dispersion, as in fig. II., the real 
structure of the band becomes more evident ; it shows a closer resemblance to the 
ordinary type of band, such as those found in nitrogen, with the exception that the 
component lines fade off in both directions from the apparent head. 
In fig. III., taken with the still higher dispersion of the quartz Littrow spectro¬ 
graph, many of the band lines are resolved into groups of three, which cannot 
properly be called triplets on account of the variable spacing. The central maximum 
about 3360, however, remains imperfectly resolved. With the highest resolution 
employed—that of the 3rd order grating—additional groups of three are separated, 
and the central maximum is seen to consist of a great number of closely crowded lines 
(figs. IV. and V.). The secondary central maximum about 3371, which corresponds to 
the weaker component of the doublet which represents the band with low dispersion, 
is also resolved into a large number of component lines. It will be observed that 
while the central maximum degrades in both directions, the secondary maximum 
degrades only towards the red. 
