82 The Total Solar Eclipse of 1900. 
follow a series of short lines, some of which seem to belong to the flash layer ; 
but whether it be due to the shortness of the exposure, or the want of sharpness 
in the focussing, the number of lines of the reversing layer which have impressed 
themselves is small. The gradual fading out of the lines as the eclipse advanced 
is exhibited in the analytical table (p. 84), and in fig. 1, Plate vim. 
The order in which the lines disappear is for the most part the order of their 
brightness in Nos. 1 and 2 of this figure, although there are one or two exceptions 
to this rule, as, for instance, the two lines of strontium at 44078 and 44216, which 
disappear earlier than other lines of the same original intensity. The figures 
given for the intensity are eye-estimates, on a scale of 10; but, although as such 
they cannot be relied on with absolute certainty, I think that they are probably 
correct to within a unit on this scale, and that a departure from the rule, such as 
occurs in the case of these lines, indicates a real change in the spectrum. 
It will be noticed that on the last spectrum of this series, which was exposed 
from about 13 to 15 sees. after the beginning of totality, the only lines which 
remain are F, H, and K. On the first plate of the second series, taken towards 
the end of totality, there are four other lines, viz., \\4472, 4341 (H,), 4102 (Hs), 
3889 (H,), and possibly \4078. From this it may be concluded, with some degree 
of confidence, that the interval between the exposure of this plate and third 
contact was less than that separating second contact and the exposure of plate 
No. 12 of the earlier series. 
For this second series the rotation of the handle was commenced at the 65th 
second from the beginning of totality ; but until the plates were developed there 
was no means of knowing whether, at the first exposure, the light was reflected 
to the right or the left of the camera, and therefore it was doubtful whether the 
first plate, on which a spectrum was found, corresponded to the first or second turn 
of the handle. If it fell to the right, then, the slide on that side not having been 
drawn, no corresponding spectrum would appear, and the first spectrum would 
correspond to the second turn of the handle. This uncertainty renders it a little 
doubtful at what instant each plate of this series was actually exposed. 
The fact, however, that the first plate of the second series contains more 
detail than the last of the first series, indicates pretty conclusively that it was 
exposed by the second turn of the handle, and that the first turn had directed the 
light to the right where the plate was not exposed, or that this exposure (which 
lasted for about 3 secs.) extended from the 68th to the 71st second of the total 
phase. If this is the case, the others would correspond to the following intervals : 
03°°D — 7d°'d, TT — 79°, 81° — 82%-5, 84° — 85°, counting in all cases from the 
beginning of totality. 
The last photograph of all shows considerably more detail that any of the 
others, a large number of lines being found on this plate only (¢f. Table), especially 
