188 
PROFESSORS G. D. LIVEING AND J. DEWAR ON 
The instruments. 
The goniometer used was constructed for us by Hilger, and has a circle of 
18 inches diameter, graduated at intervals of 5' by Simms. Fractions of 5' are read 
by a microscope with a micrometer eye-piece fixed to the arm which carries the 
telescope. 
The telescope and collimator have each an object-glass consisting of a single lens of 
quartz Lf inch diameter and a focal length of 18^ inches for the sodium yellow light, 
but not more than 16 inches for the highest rays measured. The sliding tubes of both 
telescope and collimator are graduated in fiftieths of an inch, and alterations of focus 
were made on both instruments at the same time, so that the rays falling on the 
grating might always be nearly parallel. The graduation of the sliding tube was also 
used for ascertaining the distance of the photographic plate from the object-glass of 
the telescope. This was necessary for computing the corrections of the angular 
measure, as explained below. The collimator is furnished with a quartz lens, of 
3 inches focal length, in front of the slit, movable to a greater or less distance, but 
retained by guides so that its axis may remain coincident with that of the collimator. 
This lens was placed about 6 inches in front of the slit, and the source of light at the 
same distance beyond it, so that its image was focussed on the slit. 
The measurements were all made by means of photographs taken on Wratten and 
Wain Wright’s instantaneous dry gelatine plates. The plates (2f inches by 1 inch) 
were held in a small slide attached to a tube which fitted the telescope in place of the 
eye-piece, and thus the plate could easily be turned about an axis perpendicular to its 
plane and coinciding with the axis of the telescope. This turning of the plate about 
is a matter of no small importance, as it enabled us to avoid the errors which would 
have arisen from measuring;' the distances of the lines from the irregular edge of the 
plate, as will be seen when the mode of measuring the photographs is described. The 
plates were retained in one position in the slide during exposure by three springs, of 
which two pressed against two edges of the plate and the other against its back. 
The grating was ruled on speculum metal by Chapman with Rutherfurd’s 
machine, and has a ruled surface of rather more than If inch in each dimension, with 
17,296 lines to the inch. It is an excellent grating, but, of course, has the faults 
which belong to the particular machine by which it was ruled. The definition, when 
it has not been exposed to variations of temperature, is very good, but it has one 
inconvenience for our present purpose, which is, that the focus for the same ray in the 
spectrum of the same order does not fall at quite the same distance from the object- 
glass of the telescope on the two sides of the normal. The explanation of this has 
been given by Cornu (Comptes Rendus, lxxx., 645), who has shown that it is due 
to a systematic variation in the distance between the ruled lines. As the method 
employed by us depends upon taking angular measures of the position of the ray 
on both sides of the normal, and any shift of the focussing tube between the two 
