190 
PROFESSORS G. D. LIVEING AND J. DEWAR ON 
Mode of 'proceeding. 
The first thing to be done was to obtain a focussing scale for the different angles, 
including the portion of the spectrum to be mapped. This was already known 
approximately from Sarasin’s table of refractive indices of quartz, and was corrected 
by a series of trial plates taken at successive angles. 
The electrodes, between which the spark was passed, were arranged so that the 
spark should pass horizontally (the slit being vertical), and at such a height that the 
visible image formed by the lens in front of the slit on the plates of the slit might fall 
just above, or partly above, the centre of the slit. The lower half of the slit was 
closed by a shutter, so that only the lower half of the field of view of the telescope 
was illuminated. The distance of the spark from the lens had next to be adjusted, as 
the focal length of the lens for the visible rays is very different from that for the ultra¬ 
violet which were to be photographed. This was done by estimation, as there was no 
need to have the image exactly focussed on the slit, so long as the slit was in the 
middle of the image and light enough passed through. 
The telescope was then set to such an angle as would bring the line so measured 
nearly into the centre of the field, the focus adjusted, the photographic slide adjusted 
and levelled, and the plate exposed. An image of the lines was thus formed on the 
lower half of the plate. The slide was then turned round through 180° about the axis 
of the tube, so as to bring what had before been the upper side of the plate to the lower 
side and right to left, and again levelled. The plate was then again exposed and thus 
a second image of the line impressed, and one of the two images was as far to the right 
of the axis about which the plate had been turned as the other was to the left. Half 
the distance between the two images would therefore be the distance of the line from 
the centre of the field, and the knowledge of this would give the means of calculating 
the deviation of the rays producing the line from the axis of the telescope. The 
telescope was next turned to the corresponding angle on the other side of the collimator 
and the operations repeated with a second plate, but without any alterations of adjust¬ 
ment. The telescope was then moved through a small angle, generally 5' or 10', and 
the same operations as before repeated on both sides of the collimator, the second pair 
of plates being intended to serve as a check upon the first. 
Similar operations wmre then repeated at such angular intervals as should bring in 
the most characteristic strong lines of iron all along the scale. Beyond the wave¬ 
length 2327, it was found that the iron lines were too faint to produce any sufficient 
impression on the plates. For the region beyond this up to wa,vedength 2135, copper 
electrodes were substituted for iron. This being about the limit of transparency of 
calcite (the material of our prisms), was the limit of our study of spectra at this 
time. 
The measurement of the distance between the two images of a line was made by the 
micrometer above described, and to convert this distance into arc, it was necessary to 
