SPECTROSCOPIC OBSERVATIONS OF THE ROTATION OF THE SUN. 93 
inated by a plane glass of purple colour fixed in front of the eye-piece. The grating 
is mounted upon a stand, turning on a vertical axis in such a way that the axis of 
rotation coincides exactly with the plane of the grating. This stand carries a horizontal 
circle, divided from 5 to 5 minutes, which can be read by two diametrically opposite 
microscopes. By this arrangement a differential determination of wave-lengths is made 
possible. 
The viewing telescope, which is horizontally mounted on another isolated stand, has 
an object-glass of 4°1 inches aperture and 60 inches focal length. A Cooke wire microm- 
eter is used for measuring the distances between the lines. After various preliminary 
trials with crossed and parallel wires, the pointing on the spectral lines by means of 
sufficiently close parallel wires seemed to me to yield the best results. The observation 
consists in bringing, by turning the micrometer screw, the centre of the space between 
two close wires exactly upon the line to be measured, in the same way as the meridian 
circle observer sets the division of the circle between the wires of the microscope. Care 
has to be taken, however, that the setting is made exactly on the solar limb. I have 
found that this may be done without the least difficulty. The errors of the micrometer 
screw have been repeatedly determined, with the result that the progressive error may 
be considered as negligible, but that there are indications of a small periodic error. In 
order to eliminate the effect of the latter directly from the observations, I have invariably 
observed the lines with two pairs of wires which were at a distance of exactly 14 turns 
of the screw. As is well known, the periodic error is practically eliminated from the 
arithmetical mean of the observations made with two sets of wires thus arranged. The 
value of a revolution of the screw, expressed in wave-leneths, has been determined by 
measurements of the distances of neighbouring spectral lines, by a method which is fully 
described in Professor DuNER’s paper. Such measurements will be continued in the 
future, but there can be no doubt already that the provisional value here adopted must 
be sufficiently correct, and that therefore the computed values of the velocities may not 
have to be altered in the final discussion. The value of the screw is subject to small 
variations, chiefly owing to temperature fluctuations. These have, however, been taken 
into account in the reductions in the following way :—Each set of observations yields a 
very accurate determination of the normal distances between the four lines of the group 
measured. Since the wave-lengths of these lines are known, the value of a revolution 
of the screw expressed in wave-lengths may be found by dividing the measured distances 
of each pair of lines into the difference of their wave-lengths. ‘Thus values of a turn of 
the screw are found for different points within the group, and from these the value to 
be applied in the reduction can be easily derived. 
The computation of the position angles of the various points measured along the 
solar limb requires the knowledge of the rotation of the field of the siderostat at the 
time of the observation. Obviously, what is required is the diurnal rotation of the hour 
circle of the sun, and therefore also of the pole of the heavens, round the centre of the 
field. The data required for this computation have been supplied by Cornu in his 
