26 THE ROYAL SOCIETY OF CANADA 
and concave mirrors, the coelostat mirror, and consequently the whole 
system is kept shaded by a blind except during the actual exposures, 
which occupy from 30 to 60 seconds each. 
7. Precautions c and d, conditions external to the spectrograph, 
were always carefully looked after. The solar definition during the 
summer months, on the clear and bright days which only were em- 
ployed, is usually fairly good and, as undue heating of the mirrors 
was prevented by keeping them shaded for suitable intervals between 
the exposures, the definition did not much deteriorate. It is essential 
that there be fair definition to ensure that the light reaching the slit 
may be confined to a small region around the desired portion on the 
sun’s disc. Great care was taken in the relative adjustment of guide 
plate and prisms, so that when the image was kept central and the 
spectrograph rotated to the desired and previously calculated position 
angle from the E. W. line (determined by the drift of the solar image 
when the coelostat clock was stopped), the positions of the points on 
the dise from which the light was taken were accurately known. This 
is rendered much easier and more certain by the large size of the solar im- 
age (about 227 mm.), and consequently it is improbable that any errors 
can have arisen either in this regard or due to poor definition. The 
only effect of the latter would be to introduce a small amount of light 
at slightly higher and lower latitudes or at greater and less distances 
from the limb, and the effects thereby produced would practically 
compensate one another. The necessity of observing only when the 
sky is free from haze will be evident when it is realized that the effect 
of the superposed sky spectrum, which is a blend of the spectrum from 
the whole dise of the sun is, to diminish the displacement and give too 
low a value of the velocity. DeLury made some experiments on this 
effect, and found a measurable influence on the equatorial displace- 
ment only when the ratio of intensity of sky spectrum to limb spectrum 
reached about 1 to 20. As on a clear day this ratio is 1 to 100 or less 
it is evident that no error can thereby be introduced. 
OBSERVATIONAL DATA. 
8. The plates were made by the authors jointly, as to make the 
focus and illumination tests and to guide the sun’s image carefully can 
be much more easily and satisfactorily done by two than one. The 
dates of the plates used will be given in the tables of measures to follow 
to save space. 
As stated ai ove, in the 45600 region, rotation spectra of each of 
the six latitudes to be observed, 0°, 15°, 30°, 45°, 60°, 75°, with one of 
the pole, 90°, for check purposes were made on each plate, but in the 
higher latitudes 80° and 85°, three of each with one of the pole were 
