1909-10.] Mapping of Grating Spectra. 451 
third order from X = 2000 to X = 4000, and also bright lines in the fourth 
order from X = 2000 to X = 3000, and possibly in the fifth order from X = 2000 
to X = 2400. The method adopted would probably be somewhat as 
follows : — 
X = 4300 to X — 6000. — -In photographing this region a glass absorption 
cell would be placed before the slit containing a freshly prepared 
\ per cent, aqueous solution of sesculin to which a trace of ammonia 
had been added. 
X = 3550 to X = 4300. — A glass absorption cell with cobalt chloride 
solution would be most suitable here. 
X = 3300 to X = 3550. — Employ a quartz cell with an aqueous solution 
of malachite green or of chrome alum. 
X = 2000 to X = 2300. — A very narrow quartz cell containing a fresh 
solution of nitrosodimethylaniline in glycerine would cut out the 
visible end of the first order. The ultra-violet second order could 
then be photographed by a very prolonged exposure. 
X = 2300 to X = 3300. — There is no suitable absorbent for removing the 
yellow of the first order which overlaps this region. 
To map an unknown spectrum using the second order, one would 
therefore take, at least, the following five photographs : — (1) the ordinary 
spectrum, first, second, and third orders ; (2) the second order spectrum with 
sesculin absorption ; (3) the same with cobalt chloride; (4) the same with 
malachite green ; and (5) the same with nitrosodimethylaniline. Even with 
these numerous photographs there would still be very considerable un- 
certainty in mapping the second order from X = 2300 to X = 3300. This 
would require to be cleared up as far as possible by reference to the plates 
belonging to the first and third orders in the first photograph. The work 
would not be simple, as the reduced dispersion in the first order causes 
difficulty when dealing with lines whose wave-lengths differ but little. 
The third order plates, too, are contaminated both with the second order 
blue-violet and the fourth order ultra-violet. The second order contamina- 
tion is serious, as it is at the most photographically active part of the 
spectrum, and will include the wide-extending and intense cyanogen bands 
if carbon electrodes are used. 
The difficulty of carrying out the procedure described above is the more 
important in view of the great cost it entails when one is investigating rare 
elements. As a result, the work hitherto carried out on these substances 
has frequently been confined in the main to first order photographs taken 
between the limits X = 2700 and X = 4700. Little has been done in photo- 
