114 



SCIENCE 



[N. S. Vol. XLI. No. 1047 



case of other stars in this catalogue he de- 

 votes much attention to periodic inequali- 

 ties. 



It should be remarked that the absence 

 of an appreciable periodic term in the 

 proper motion does not necessarily imply 

 the non-existence of Chandler's third body, 

 since his theory does not demand any par- 

 ticular coefficient for this periodic term. 

 The only condition is that that coefficient 

 must be at least twenty times the star's an- 

 nual parallax, and thus an accurate deter- 

 mination of the latter quantity would 

 throw some light upon the present ques- 

 tion. Unfortunately no determination of 

 the parallax accurate enough for this pur- 

 pose has as yet been made. 



Starting with Chandler's inequality of 

 173 minutes, Tisserand has attempted an 

 explanation that does not assume the pres- 

 ence of a third body. He shows that if 

 Algol be slightly flattened and if the orbit 

 of the eclipsing satellite be somewhat ellip- 

 tical, the orbit itself will revolve slowly and 

 uniformly in the same direction as the or- 

 bital motion of the satellite. Consequently 

 the eclipses will occur earlier than the aver- 

 age time if the periastron point is in the 

 half of the orbit that precedes eclipse, and 

 later than the average if the periastron 

 point is in the half that follows eclipse. 

 This explanation is beautifully simple, and 

 for a time seemed to be the key to the 

 puzzle. I am able to say, not without some 

 regret, that Tisserand 's explanation is no 

 longer tenable. In his memoir the follow- 

 ing relation is established: 



Period X eccentricity = 3.1416 X the 

 inequality. In this case the period is 2.87 

 days, and the inequality found by Chand- 

 ler is 173 minutes; an eccentricity of 0.13 

 is therefore demanded, but this is out of 

 the question. A long series of spectro- 

 graphic observations made at the Alle- 

 gheny Observatory shows conclusively that 



the eccentricity of this orbit can not pos- 

 sibly be as great as 0.13, that it is more 

 likely than not to be under one fifth this 

 amount, and that therefore no inequality 

 greater than forty minutes can be plausibly 

 accounted for in this way. 



Shortly after Chandler's formula for the 

 inequality was published, the star (always 

 El Ghoul) thereafter began departing from 

 it little by little, until now the eclipses 

 occur more than an hour later than the 

 formula implies. The character of the in- 

 equality is once more in doubt, but as the 

 existence of some kind of inequality is be- 

 yond question, this does not lessen the ne- 

 cessity for an explanation. 



While the chances in favor of the reality 

 of Chandler's third body have been grow- 

 ing less and less, evidence has been steadily 

 accumulating in favor of an entirely differ- 

 ent third body in this system. Since the 

 publication in 1890 of Vogel's classic obser- 

 vations, it has been well known that the 

 radial velocity of Algol is affected by an 

 oscillation whose semi-amplitude is about 

 forty kilometers, and whose period is the 

 same as that of the light changes, 2.87 days. 

 In 1906 Belopolsky of Pulkova detected the 

 presence of another oscillation in the radial 

 velocity, the amplitude being much smaller 

 than the other, and the period several hun- 

 dred times as long. Observations made at 

 the Allegheny Observatory have confirmed 

 this discovery in an unmistakable way. 

 The period of this new oscillation is found 

 to be a little less than two years. It could 

 be explained by the presence of a third 

 body of such mass and so situated that the 

 projected distance from Algol to the center 

 of gravity of all three bodies is about two 

 thirds of the distance from the earth to the 

 sun. It is natural to inquire whether other 

 explanations are not possible, or, in other 

 words, whether the shifts in the spectrum 

 lines from which this third body is inferred 



