454 Proceedings of the Royal Society of Edinburgh. [Sess. 
and 5500 through angles of 45° more or less than the angle for light of 
wave-length 3450.* Thus, with lines of wave-lengths 2800 and 5500 we 
shall have in the one case a maximum intensity of the components polarised 
perpendicularly to the lines of force and a minimum intensity of the 
parallel polarised components, and in the other case the reverse effect. 
This is shown in the second and fourth columns of fig. 3 for the resolution 
of the simple triplet and quartet. Throughout the spectrum we shall have 
a gradual variation in the relative intensities of the components, each wave- 
length having a characteristic intensity ratio. Hence it is evident that 
if we had occurring close together on a photographic plate a line of wave- 
length about 2750 in the second order and one of wave-length about 5500 
in the first order, the intensity ratios in a Zeeman effect photograph would 
at once enable one to assign the lines to their proper order. 
The author has employed this method throughout in mapping the spark 
spectrum of dysprosium, and has found it to work satisfactorily. The 
spark was produced by a large Max Kohl induction coil worked by a motor 
mercury interrupter. Six large Leyden jars were placed in parallel with 
the spark, and a self-inductance connected in the spark circuit increased the 
sharpness of the metal lines while removing those due to the atmosphere. The 
electrodes were of specially purified carbon, and were impregnated before 
use with dysprosium chloride. The light was focussed on the slit by a 
quartz lens having a maximum thickness of about 6 mms, and — when 
used centrally — giving a gyratory effect equivalent to a millimetre plate of 
optically active quartz. Two photographs were taken for the purpose of 
mapping the spectrum, viz. (1) with the ordinary spectrum and having the 
iron spectrum partially superposed for comparison (see fig. 1), and (2) with 
the Zeeman effect. In the second case the source of light was situated in the 
4*5 mm. pole gap of the electro-magnet in a field of about 25,500 c.g.s. units. 
Forty-seven plates, measuring 12x5 cms., manufactured by Schleussner 
of Frankfurt a/M., were employed in each photograph, and gave the entire 
spectral band between the points A = 2100 and A = 13,000 in the first order. 
The plates have been measured with an instrument specially built to the 
author’s requirements by Zeiss of Jena, and modelled on their Abbe comparator 
microscope. The positions on the plates of all the chief dysprosium lines 
and of a large number of iron lines were measured. The wave-lengths of 
the former were then calculated, the spectrum order being determined from 
the Zeeman effect photographs in all cases where doubt might arise. Fig. 
4 shows a part of a Zeeman effect plate, magnified about three times. The 
left-hand line is in the blue-green part of the first order, and the other line 
* Gumlich, Wied. Ann., lxiv. p. 349 (1898). 
