ON OUR KNOWLEDGE OF SPECTRUM ANALYSIS. 319 



thalHnm, sodium, lithium, calcium, strontium, and barium chlorides ; 

 spark-spectra of rubidium, caesium, thallium, sodium, lithium, calcium, 

 strontium, barium, magnesium, erbium, yttrium, cerium, lanthanum, 

 and didjTnium ; and absorption-spectra of erbium nitrate and didymium 

 sulphate. 



It is necessary that the indications of each spectroscope should be 

 reduced to the common scale of wave-lengths, if the results obtained are 

 to be compared with those obtained with other spectroscopes : but for the 

 mere purpose of identifying an alkali or an alkaline earth it is not necessary 

 to go beyond the scale of the spectroscope itself. Photographed scales, 

 giving the positions of lines directly in wave-lengths, to be used instead 

 of the ordinary scale of equal parts, have been constructed,^ but for 

 accurate work it is much the best to employ a scale of equal parts, and 

 to effect the reduction of wave-lengths separately. 



It is not very often that any other means of ignition than the Bnnsen 

 flame is employed when the spectroscope is simply used as an addition 

 to the ordinary means of chemical analysis. The employment of a higher 

 temperature, however, much extends its range even for such purposes, 

 and at the same time increases the difficulty of identification, and neces- 

 sitates more exact measurements. 



A small induction coil, actuated most conveniently by some form of 

 battery, such as the Bichromate cell, which can be kept always ready,^ 

 and a small Leyden jar — the inside coating connected by an insulated 

 wire with the one terminal of the coil, and the outside coating with the 

 other — furnish a spark of the necessary intensity. If platinum wires 

 are employed as poles, it is important that fresh wires should be taken 

 each time, since wires which have been used for any particular metal 

 often continue to give the lines of that metal with great persistency. 



Bunsen^ recommends as poles little cones of pure porous carbon, 

 impregnated with a solution of the substance under examination. 

 A further^ difficulty in the employment of the spark with the spectroscope 

 for the ordinary purposes of chemical analysis arises from the constant 

 presence of the air-spectrum. It is necessary, therefore, to carefully map 

 the spectrum of air^ as obtained with the coil and spectroscope, which 

 are to be employed, say, first with platinum wires and then with silver 

 wires as poles. In each case the brightest lines will be those due to air, 

 with the addition in the one case of the fine lines of platinum, and in the 

 other of those of silver. The fine lines given by the less volatile metals 

 are often easily distinguished from those of air by the fact that they often 

 extend only a short distance from each pole, and do not reach across the 

 whole breadth of the spectrum, while those of air are of equal width across 

 the whole breadth of the spectrum. 



The air-lines are fainter when no jar is employed, so that with 

 the more volatile metals it is easier to work with the coil without a 

 Leyden jar. 



' Bascoe and Scliorlemmer's Chemistry, vol. ii. pt. ii. p. 471. Salet, Paris Ckem. 

 Soc, May 4, 1877. 



2 Bunsen, Phil. Mag. (Fom-th Series), vol. I. p. 527. 



3 Phil. Mag. (Fourth Series), vol. 1. p. 430. 



* For other modes of procediire see Lockyer's Studies in Spectrum Analysis, 

 pp. 60 and 63. 



* Maps of the air-spectrum are given in Bunsen's paper, Phil. Mag. (Fourth 

 Series), vol. 1., and in Thal^n's Determination des Loiigcvrs d'Onde. 



