IQOi.] on Spectroscopic Studies of Astrophysical Problems. 145 



great catastrophe, which in its dilatation and contraction was alter- 

 nately cooler and hotter. And to return to the same line, x, in the 

 siDectrum of ^ LyraB, is it not possible that we have here also an 

 oscillating temperature brought about, not as in the Nova by com- 

 pression and rarefaction, but by the periodic approach and recession of 

 two stars, and, in consequence, a temporary extinction of the line at 

 the higher temperature of pericentre ? 



This was the thought that led me to a careful and often repeated 

 measurement of the total width of the composite line H^ on all the 

 plates of the series. For the line x might be present whenever the 

 total width was greatest, although possibly hidden by the dark line ; 

 and it would be absent when the whole band was distinctly narrower. 

 The measures went very well in favour of the hypothesis until the 

 last quarter of the period was reached, as is now exhibited in the 

 abridged tabulations on the screen. At the principal minimum on 

 five dates, from 10 hours before it to 5 hours after it, the mean 

 width is 20*8 micrometer divisions between extremes 22 and 18. 

 On eleven dates between the periodic dates 0-6 and 2-13 the mean is 

 15*5 between extremes 17 and 14 ; and on the twenty-three dates from 

 2-20 to 9-14 the mean is up again to 22 divisions between extremes 

 of 24 and 18. But in the last quarter of the period the mean width 

 is down again to 16*5 divisions. Except for this fall-off the 

 demonstration would be complete. Omitting it for the moment, 

 we have the narrower width only for about two and a half days after 

 the principal minimum, indicating a temporary extinction of the line 

 X during the period of highest temperature above the dissociation 

 point. But the same indication recurs towards the end of the cycle, 

 when the temperature should be lowest. Is this a clear negation of 

 the hypothesis ? I think not ; for all the bright lines are weaker at 

 this time, and the line ic is a weaker line than its companion line y, 

 and fails more on its more refractive side ; the loss of width at this 

 time may therefore be owing only to general loss of light. 



But a more important objection has yet to be met. We have 

 seen that dark H^ bisects the broad bright line at the principal 

 minimum ; and I offered this, together with the apj)arent concentric 

 positions of the other pairs of bright and dark lines, as evidence of 

 spectral conjunction at the principal minimum, a proposition which 

 is incompatible with a foreign origin of the part I have called x. 

 The answer is found in the table of figures showing the apparent 

 oscillation of dark H^. The mean position of this line is not the 

 true position of the hydrogen line. It is quite three micrometer 

 divisions to the more refractive side. This might be accepted as a 

 probable error of a single reading, but not as the mean of 14 

 measures on different plates. The true hydrogen rcentre is well 

 within the red side part of the bright line, supposing it to overlap 

 the dark line ; and the dark line may be, what its appearance 

 suggests, a stranger, for it has not the family likeness of the other 

 hydrogen lines. It is natural to ask : What has become of the line ? 



2 H 2 



