442 Tlie Rev. Walter Sidgreaves [Jan. 22, 



neither sharp nor thin, but wide and ill defined, and we are more 

 dejDendent upon a mean value from many plates which are wanting. 



A more complete result is offered by the similar relative oscilla- 

 tions of the dark component H^. These measures were ready in the 

 tabulated scale readings of the prominent lines ; and the results are 

 shown in the second part of the table on the screen, in measures 

 referred to the centre of dark H^. There is a clear relative oscilla- 

 tion of the line between a well-marked minimum and a well-marked 

 maximum distance from Hy ; and the mean of the two occurs twice 

 in the light period, on the dates 0-0, 0-5, 0-6, and again on 7-4, 

 7-6, indicating the spectral conjunctions at the first minimum and 

 about 18 hours after the second minimum. 



It is only necessary to add that the apparent oscillation of dark 

 H^ is the indication of the real oscillation of bright H^. But it may 

 be well to point out the cause of the oscillation appearing in H^ and 

 not in the other hydrogen lines. In the first half of the period, the 

 fiducial point of the scale readings is the centre of that part of dark 

 Hy which is seen outside the bright line towards the violet end of 

 the spectrum ; in the second half, the fiducial point is shifted to 

 another part of the same dark line, viz., the centre of the red side 

 part. The other hydrogen lines, Hg and H^ are affected in the same 

 manner as H^. Not so H^ ; this dark line appears in the second 

 half of the period, with both edges visible, and its centre (omitting a 

 smaller change to be considered later) is practically the same at both 

 epochs ; its apparent shift in one direction is the real shift of the 

 zero point in the opposite direction. 



We have, therefore, I venture to say, a good probability in favour 

 of the elliptical orbit ; and we must now look to other details of the 

 chart, to see in what way they may or may not support the conclu- 

 sion. 



And first, for the greatest development of the bright lines, we find 

 them at their best on the plates of 1-11 and 1-14, periodic dates, 

 and again on a more recent photograph, too late to find its place on 

 th/3 chart, on 1-9 date ; while at 0-17 they are still quite weak. 

 They come out, therefore, it may be quite suddenly, between these 

 dates, or about one day after the principal minimum. It has long 

 been known that this excessive strength is associated with the epoch 

 of chief minimum, and the effect has been attributed by Dr. Vogel 

 to contrast with the weaker continuous siDCctrum at the minimum of 

 light. Contrast, of itself, is certainly enough to account for much ; 

 but it does not answer in this case to the chart. The chart shows 

 the lines strong, not at the chief minimum, but well after it. Now 

 contrast should be the same before and after, for the continuous 

 spectrum is equally weak on both sides of the minimum ; and yet on 

 the fourteen dates between the second maximum and the chief mini- 

 mum, following one another at short intervals of the series up to 

 within seven hours of the minimum, there are no strong bright lines, 

 but on the contrary they are weak and dying out. I see no answer 



