56 
PROFESSOR W. N. HARTLEY ON SPECTRUM PHOTOGRAPHY 
they are dissolved in water, and if necessary treated again in the same manner. 
Aluminium chloride was prepared from an especially pure specimen of ammonia alum, 
the aluminium hydroxide was precipitated by ammonia and most thoroughly washed 
with hot water, it was then dissolved in hydrochloric acid. The cobalt chloride was 
a carefully prepared laboratory specimen. This series of solutions is very complete, 
inasmuch as spectra of almost every character are represented and may be referred to. 
For instance, there are comparatively few lines in the magnesium spectrum, but these 
are of a strongly marked character and they are closely grouped together. Exceed¬ 
ingly short and some straggling and long lines occur in the zinc spectrum, while in 
cadmium we see the position of two of the lines to be so near together that there is 
some difficulty in distinguishing them. In the iron and cobalt spectra there is a 
multitude of long and short lines crowded together in groups. In arsenic and 
antimony we have examples of metalloid spectra, with a considerable amount of 
diffused rays and lines distributed with tolerable regularity throughout the whole 
ultra-violet region. 
By the juxtaposition of spectra on the same plate, the lines assignable to each metal 
were compared with those ^yielded by a solution of its chloride. With but two 
exceptions the two series of spectra are identical line for line ; in the case of iron the 
number of lines reproduced is over 600, and in that of cobalt over 500. The group of 
five fine lines constituting the most refrangible group in the spectrum of magne¬ 
sium were exactly reproduced by the solution of the chloride along with the other 
lines characteristic of the metal. The dual lines of cadmium were in like manner 
plainly seen. The one sole difference between the spectra of metallic electrodes and 
those from the salts of the metals was the greater degree of continuity of the fines 
shown by the salts. Thus all the fines discontinuous in the spectrum of iron w T ere 
continuous in that afforded by a solution of ferric chloride. The short fines of the 
metals potassium and sodium, which are weak, appear as short lines in the spectra of 
their chlorides, likewise some of the shortest fines which are at the same time strong 
fines in the spectrum of aluminium are seen as short fines in the spectrum of a 
concentrated solution of its chloride. These lines form a triplet group with wave¬ 
lengths 36127, 360F2, and 3584"5. This last fine appears to be strongest in the 
aluminium spectrum. Whether the short fines appear or not depends upon the 
amount of metal present in the solution. 
Regarding the two exceptional cases above-mentioned, they are referable to two 
distinct causes—the first to the extreme shortness of the fines, and the second to the 
presence of an impurity. Zinc is a metal with a number of fines in its spectrum so 
short that they can be described as merely dots. Their wave-lengths are the 
following: 2526-3, 252T3, 2514-7, 25087, 2490-4, 2485-9, 2427-0, and 2418-8. All 
specimens of zinc yield these very short fines, but solutions made from them do not. 
As anhydrous zinc chloride contains only 65 per cent, of the metal, and as no solution 
of the chloride used in the spark apparatus contained more than 25 per cent, of this salt, 
