496 Mr. J. Barnes on the Analysis of 
many sources tried the bright radiation from mercury vapour 
was the best for obtaining observations on the changes 
produced in the components by external changes in the con- 
ditions. We will thus first consider the results with this 
source. Great pains were always taken to have perfect 
adjustments, chiefly with respect to the focusing of the tele- 
scope and the parallelism of the interference-plates, before 
any readings were taken. 
The vacuum-tube discharge was obtained in a Geissler 
tube with mercury electrodes of the form suggested by Runge 
and Paschen*, the capillary of which was placed directly 
in front of the slit of the spectroscope. The different tubes 
were connected to a Geryk pump and a pressure-gauge, 
enabling the pressure of the vapour through which the dis- 
charge passed to be quickly changed from a few millimetres 
to a fraction of a millimetre. 
It is rather difficult to decide what is the most advantageous 
way to record results, whether to take what appears to be 
the centre of gravity of the various components constituting 
the radiation as the position from which to measure wave- 
lengths, which is the usual way in the measurements of the 
lines obtained by means of the grating, or to consider the 
component of the greatest intensity as the standard, and record 
the wave-lengths of the other components with reference to 
this. This method is the one employed by Michelson, and 
Fabry and Perot. The latter method is nevertheless unsatis- 
factory, for I have found, even in some of the few radiations 
investigated, that there are two or more bright components 
whose intensities are equal. For want of a satisfactory 
standard, and also that the following results may be easily 
compared with those of the other investigators, their method 
has, however, been followed. In the cases where the brightest 
components are of equal intensity one of them has been selected 
for the standard. In what follows the plus sign indicates 
that the component has a longer wave-length than the 
standard, the minus sign the reverse. 
The following results were obtained after a long series of 
observations with a tube whose capillary was 0°5 mm. in 
diameter and the vapour at a pressure of 1:5 mms. The bright 
green radiation, whose wave-length is 5461, consists of six 
components ; the two brightest having about equal intensities, 
the one having the longer wave-length will be considered 
the standard. The other components have the following 
* Astrophys. Journ, xy. p. 238 (1902), 
