528 



NA TURE 



[June 30, 1910 



Tikhoff studied the variables RT Persei and W Ursas 

 Majoris. Some years previously he had compared the 

 velocity curves and the light curves of the stars 

 6 Cephei and t] Aquilae, which are well-known spectro- 

 scopic binaries and variable stars. Inasmuch as an 

 accepted theory of the physical nature of stars of this 

 type is still wanting, this method must be considered 

 radically unsound. But more recently M. Tikhoff 

 has had the happier idea of comparing the velocity 

 curves as determined from lines in separate regions 

 of the spectrum. Theoretically, this would seem to be 

 the method of all those which have been suggested 

 which is the most free from objection. But it is 

 doubtful whether, among the spectrographic observa- 

 tions already made, even of the highest class, suitable 

 material exists for the successful application of the 

 method. It is certainly possible to criticise on definite 

 and practical grounds MM. Tikhoff and Belopolsky's 

 discussion of the case of ;3 Aurigae. 



The results already obtained may be tabulated 

 thus :— 



Star Range Lag Authority 



/ifi min. 



)3 Persei 680-450 ... 13 ... Nordmann 



A Tauri ,, . 30 ... ,, 



RT Persei 560-430 ... 4 ... Tikhoff 



W Ursre Majoris, 625-3S0 ... 10 .. ,, 



3 Aurigce 450-400 ... 10-20 ... ,, 



The third column, which alone requires explanation, 

 contains the retardation, expressed in minutes, of 

 some event observed in blue light over what is sup- 

 posed to be the same event observed in light of greater 

 wave-length. The event in the first four cases is the 

 light minimum of the star, while in the fifth case it 

 is the disappearance of the radial component of the 

 velocity relative to the Sun. Unfortunately, we have no 

 trustworthy determinations of the parallaxes of these 

 stars. Pritchard's values for /3 Persei and /3 Aurigae 

 are near 006", and M. Tikhoff himself has found the 

 parallax of RT Persei to be insensible. Thus we 

 can only note the qualitative agreement in the sign 

 of the Tag in all cases, w^hich suggests that blue 

 light is transmitted through space at a slower rate 

 than light of longer wave-length. Yet the results are 

 liable, even on this ground, to serious criticism, which 

 has been expressed forcibly by Prof. Lebedew. It is 

 not surprising that close inspection shows that the 

 data in the case of & Aurigee are not self-consistent. 

 But in the other cases we cannot be certain that the 

 observed event is really synchronous in its origin for 

 different qualities of light. This essential condition 

 may be nullified by the physical character of the star, 

 as, for instance, by a selectively absorbing atmosphere 

 of the occulting body. Prof. Lebedew is entirely 

 right in suggesting these criticisms, but they do not 

 prove that the medium filling space is without disper- 

 sive power ; and even if this fundamental question is 

 left open, it is to be hoped that researches will be 

 continued on the same lines, for the ingenious 

 methods of MM. Nordmann and Tikhoff bid fair to 

 extend our knowledge of variable stars in a most 

 helpful way. 



Meanwhile the line of direct investigation of a 

 possible selective absorption in space has been 

 followed. If two stars the intensities of which are Ij 

 and I2 require exposure times Tj and T2 in order to 

 register images of equal density on a photographic 

 plate, we may put (after Schwarzschild) 

 I,TiP = I.jT/. 



A. priori we might expect the same effect to be pro- 

 duced b)' the same incident energy, or /) = i. As a 

 matter of fact, a number of independent researches 

 have suggested that p is much nearer the value o'8. 

 This deficiency in the value of p has been attributed 



NO. 2122, VOL. 83] 



to the properties of the photographic plate. But il 

 occurred to Prof. Turner that the fact might have it« 

 origin in cosmic causes. He had deduced from thf 

 Greenwich astrographic results that " when the timt 

 of exposure is prolonged in the ratio of five magni- 

 tudes, the photographic gain is four magnitudes.' 

 This result, which has been reached by others in mo;e 

 or less the same form, is equivalent to the above 

 statement that p — 0'8. A number of facts connectec 

 with visual and photographic magnitudes could thus be 

 reconciled by supposing that the small particles dis- 

 tributed in space actually produced a selective scatter 

 ing in accordance with Lord Rayleigh's law. 



It seemed as if a crucial test was at hand to try this 

 hypothesis. It was only necessary for M. Tikhoff tc 

 apply his light-filters and to see whether the appareni 

 law of photographic action was the same for the 

 blue starlight which affects the ordinary plate, anc 

 for the visual rays to which the bathed plate is sensi 

 tive. The experiment was immediately made, anc 

 the first results seemed to bring a striking confirma 

 tion to the hypothesis. M. Tikhoff found that p = o'6\ 

 to 079 for the photographic rays, but that p = o'gi U 

 o'gS for the green-yellow rays. But Mr. Parkhurst 

 of the Yerkes Observatory, who has made a specia 

 stiidy of the subject of photographic photometry 

 strikes a note of warning. Lender conditions appar 

 ently similar, he has obtained p = o'8S for the ordinan 

 plate, and p = o'8i for the bathed plate with colour 

 filter. These results go in the opposite direction anc, 

 must be attributed to the different plates (SchleussnCi 

 and Cramer) and filters employed. Mr. Parkhurst con 

 eludes that ''if cosmical causes played any part in thj 

 matter they would be completely masked by photo 

 graphic effects." 



The inter-relation which has been noticed betweei 

 the problem of absorption in space and the problen 

 of sidereal structure has naturally engaged the atter 

 tion of Prof. Kapteyn, who has been the mos 

 prominent and assiduous student of the latter ques 

 tion during recent years. He has noticed that th 

 marked deficiency in the numbers of the fainte 

 classes of stars is equally apparent in all directions c 

 the sky. Unless this peculiarity is attributable to th 

 effect of general absorption^ we must suppose tha 

 the sun is situated at the centre of the universe, an 

 though such a thing is perfectly possible, it is nc 

 specially probable. Kapteyn prefers to admit a 

 absorption of light, and provisionally estimated th 

 loss required by the hypothesis of nearly constar 

 star-density as o"oi6 of a magnitude for stars with 

 parallax of 0*1". More recently he has brought foj 

 ward an interesting argument of a qualitative kinc 

 Miss Cannon's classification of star spectra distir 

 guishes between two classes which differ only i 

 showing greater or less relative absorption in th 

 violet end of the spectrum. Arcturus is the type c 

 the stars less affected, while a. Cassiopeiae is the typ 

 of those more affected in this way. If the propert 

 is not intrinsic in the stars themselves, stars belong 

 ing to the Arcturus class should be nearer to us tha 

 stars like a Cassiopeiae. Hence the former class shoulc 

 on the average, possess greater proper motions. Pi 

 to the test, 45 stars of the a Cassiopeiae division ga^f 

 an average centennial proper motion ii"4", and 2 

 stars of the a Bootis division gave 47' i". Thus th 

 idea is confirmed that the distinction is due toabsorj 

 tion in space. A number of interesting points ai 

 involved in this line of argument, and it is to h 

 hoped that it will be further tested by extending il 

 area of application. 



Quite lately Prof. Kapteyn has published 

 second and rnore extended research on the subjc 

 In this he investigates the amount of selective abso* 



