FLAME SPECTRA AT HIGH TEMPERATURES. 
163 
A trial was made with various developers in order to ascertain which were the most 
suitable. The spark spectrum of cadmium was photographed on plates of the same 
kind, with an exposure of tive seconds in each case, and development was carefully 
timed. Developers containing the following reducing substances were used :— 
(l) pyrogallol, (2) eikonogen, (3) amidol, (4) rodinol, (5) hydroxylarnine hydrochloride, 
(5) hydroquinol, (6) ferrous oxalate, already prepared from potassium oxalate and 
ferrous sulphate, (7) ferrous oxalate, prepared just prior to use by mixing ferrous 
sulphate and potassium oxalate solutions kept separate. 
Some years ago a similar trial of the then existing developers was made by me and 
preference was given to hydroxylarnine hydrochloride, as prescribed by Eglt and 
Spiller, because it gave a brown deposit of silver showing under the microscope no 
structure or granulation. A commercial sample of the salt, recently purchased, which 
proved to be strongly acid, was recrystallized from hot alcohol and rendered neutral. 
It gave good results, but the image was slow in appearing. 
Freshly prepared ferrous oxalate was excellent, but best of all was hydroquinol, 
because it not only produced a dense black image with as much freedom from granu¬ 
lation as any other substance, but it also reproduced lines of feeble intensity, and it 
developed completely in three minutes as against seven minutes for hydroxylarnine, 
and four or five minutes for other substances. 
Granulation appears to be caused by a condition of the gelatine now generally 
used rather than by the nature of the developing solution as was formerly the case. 
It was decided to use sensitized plates and hydroquinol as a developer. 
Method of Measuring the Positions and Wave-lengths of Lines. 
The most convenient and simple method of measuring the s}>ectra emitted l)y 
flames is to take a photograph of the spark spectra of tin-cadmium and lead-cadmium 
alloys superposed upon the former. From the lines of these metals and those of air 
which accompany them we obtain measurements from which, by an interpolation 
curve, the oscillation-frequencies and their corresponding wave-lengths may be 
ascertained. 
The measurement of the lines is made in the same manner as the measurement of 
the bands in aVjsorption spectra, namely, by simply applying to the ])hotograph an 
ivory scale which is divided into hundredths of an inch, and by means of a lens or 
low-power microscope with cross wires in the eye-piece, reading by judgment to 
tenths of each division. To do this with the greatest accuracy it is necessary to 
have a straight line ruled down the middle of each s})ectrum, against which the edge 
of the scale is fixed in position. To rule this line a very slight nick is made in the 
jaws of the slit of the spectroscope, which admits more light at this than at any other 
point, and causes a feeble continuous spectrum to be photographed; upon this the 
* Several prominent iron lines beyond A. 3900 were used in drawing tlie curve. 
Y 2 
