o:^ THE TOTAL SOLAR ECLIPSE OF AUGUST 29, 1886. 
331 
tlie blue and violet than those used in tlie otlier spectroscope, wliich also were supplied 
by Captain Abney. The spectroscope was in good adjustment, and the reference 
spectrum taken with it on the day of the eclipse is perfectly sharp. If the develop¬ 
ment could have been carried further without danger, a better image might have been 
obtained. The plate shows in a very striking way the displacement of actinic 
intensity towards the red in the coronal spectrum. We can recognise on the plate 
the absorption of the G band and the strongest corona line, and thus fix the exten¬ 
sion of the spectrum. The reference spectrum reaches from F into the ultra-violet, 
and is very strong between h and H. It shows, in addition, a band in the yellow or 
red, which, however, is very faint. The corona spectrum begins about F, is 
strongest between F and G, and falls off very rapidly in intensity about h. It can be 
barely traced between li and II. The lines shown on this photograph will be given in 
the annexed Table. 
I must now describe the method adopted to determine the position of the lines 
which appear on the photographs. In 1882 a reference spectrum had been taken on 
the same plate previous to totality, but that reference spectrum was found to be of very 
little use in measuring the lines. It had to be near the edge of the plate in order 
not to interfere with the spectrum of the corona; and, owing to the curvature of the 
lines, we cannot directly compare the position of a line coming from the centre of the 
slit with one from near one of the edges. In order to compare accurately two spectra 
on the same plate, they must be in contact with each other, which would be impossible 
unless part of the coronal spectrum were sacrificed. A reference spectrum taken on a 
different plate is as useful as one taken on the same plate, if precautions are taken to 
fix well the position of the plate in the holder. 
All good photographs of the corona which have been taken hitherto show the 
calcium lines coincident with H and K, and we can take these always as the starting 
point for our measurements. The distance of these lines from any unknown line will 
be the same in the centre of the reference spectrum as in the centre of the coronal 
spectrum, provided no shrinkage of the film has taken place during development. 
But, if a shrinkage has taken place, there is no reason to suppose that it is uniform 
all over the plate, and it may, therefore, be just as different in the reference spectrum 
and the coronal spectrum, if these are on the same plate, as if they are on different 
plates. Besides the H and K lines we can generally recognise some other feature in 
the coronal spectrum, consisting either of another prominence line or some well- 
marked Fraunhofer line ; and in that case interpolation becomes easy with the help of 
the lines on the reference spectrum. 
If we add to these considerations the danger of exposing to Sun light, or even 
day light, a plate which is to be used during the eclipse, merely for the sake of the 
reference spectrum, I think it will be conceded that I was justified in taking the 
reference spectrum on a separate plate. Tliat this danger, which I had pointed out 
2 u 2 
