CAPTAIN W. DE W. ABNEY AND DR. A. SCHUSTER 
264 
supposition that differences in the refractive indices of two lines are proportional 
to the differences of the inverse squares of the wave-length. Owing to different 
circumstances, and amongst others to the fact that the distances measured are 
not quite proportional to differences of the refractive indices, we obtain in this way 
approximate results only; but by taking a photograph of the solar spectrum with the 
same prism in the same position, we can easily determine the correction which 
has to be applied in order to arrive at the most satisfactory result. For instance, 
the distance of one image was measured from the corresponding image of C and F, both 
with a micrometer and directly with a finely divided scale. The interpolation formula 
gave 5889 by the first method of measurement and 5893 by the second, the mean being 
5891. This brings us near D, and if a similar calculation is made in the reference 
spectrum, interpolating D between C and F, a wave-length 5907 is obtained. The 
true wave-length being 5882, we see that we have to apply a correction of —15 in 
this part of the spectrum. We thus find finally 5876 for the wave-length of the 
unknown line, agreeing almost exactly with the wave-length (5875) of D 3 . The 
prominences I. and II. gave results which are practically identical. 
Similarly we find for a very faint image of the prominence II. a wave-length 5315, 
which is evidently the well-known corona line (K 1474), which has a wave-length 
5316. The image, however, is exceedingly faint. The prominence I. shows also 
two lines in the infra-red which are difficult to identify. One of them is very likely 
\=8240 ; the other has a wave-length which is certainly above X 10,000, but 
by how much we cannot tell. Two lines, \ = 447l and \=4394, are seen in the 
blue; 4471 is the line f which is always present in prominences, but what 4394 
is we do not know. Young, in his catalogue of chromospheric lines, notes that 
he saw 4394‘6 fifteen times out of 100 observations. 
Besides these well-defined prominences the photograph shows two rings, which 
are evidently due to the lower parts of the corona, and therefore correspond to true 
coronal light. The wave-length of one of these rings was measured to be 5315. 
It is due to the green corona line (K 1474); the second is coincident with D 3 . 
The ring in the green is particularly strong in the south-western quadrant, and 
hardly visible at some of the other points of the sun’s limb. The yellow ring is 
much fainter on the whole, but more uniform all round the sun. In addition to the 
rings and prominences we observe on the photograph at places a certain striped 
appearance, which is due to continuous spectrum belonging to the prominences or 
to intense parts of the solar corona round the sun’s limb. 
A curious and rather puzzling point remains to be noticed. The prominence V. in 
the F ring seems to be duplicated. Two distinct impressions appear, one above the 
other. This cannot be due to a shift in the camera, for the other prominences, some of 
which are more intense, do not show the duplication. The only possible explanation 
which has occurred to us is, that one of the images is due to the comet, and that it 
is only by accident that it appears so close to an image of the prominence. 
