42 



KNOWLEDGE 



[Febbuabt 1, 1898. 



without explanation at present. The facts, too, that these 

 divisional differences are practically traceable only amongst 

 the Orion stars and those of Secchi's first type, and that 

 no stars are found of Division c in Group XIV., whilst 

 seven are recorded as being intermediate between Divi- 

 sions a and c, point to the classification in this direction 

 being neither so perfect, nor so directly the effect of 

 simple causes, as the cross arrangement into groups. 



An interesting relationship, which Miss Maury mentions, 

 suggests that m the case of Divisions a and /' the differ- 

 ences between them may possibly be of a mechanical 

 nature rather than one of temperature or constitution. 

 She points out that the two spectroscopic binaries X, Ursie 

 Majoris and /3 Aurigic, though really of Division a, appear 

 as members of Division b at that particular point of their 

 orbit when the relative motion in the line of sight of the 

 two members of the system is sufficient to widen the lines 

 of their composite spectrum, but not to separate them into 

 pairs. It is clear, therefore, that the existence of a large 

 number of close binaries might explain the occurrence of 

 Division h spectra, provided that these several pairs were 

 composed of stars not very unequal in magnitude, of the 

 same type of spectrum, and with relative motion in the 

 line of sight such that their lines were widened but not 

 separated. 



We already know, by direct observation, of binary systems 

 in which the periods vary from five and a half years up to 

 many centuries. The Algol variables and the spectroscopic 

 doubles have similarly revealed to us the existence of 

 systems with periods ranging from a few hours to a few 

 weeks. We may be perfectly assured that there are 

 other systems with periods of an order intermediate 

 between these, not of weeks or of years, but of months. 

 And such, under the special conditions mentioned above, 

 would give us /' division spectra. In cases where the two 

 components were of different types we should have a 

 "composite" spectrum. It is possible, therefore, that 

 Division h and " composite " stars are but different pre- 

 sentments of the same relationship — a binary system of 

 two not unequal stars far too close for optical resolution. 



The researches of Darwin on tidal evolution, and of See 

 on that of double stars, lead us to the conclusion, since 

 double stars tend to widen with age, that these very close 

 binaries are yet in an early epoch of their life history. 

 The fact, therefore, that the Algol stars and those of 

 Division h are most plentiful in the Orion and first type 

 groups is a confirmation of Miss Maury's conclusion that 

 these are early forms of spectra, and seems better to accord 

 with the facts than Lockyer's view, which places the Sirian 

 stars midway in the evolution. 



The test, of course, of the truth of the suggestion will 

 be that a prolonged watch of Division h spectra will sooner 

 or later show in some instances a gradual change into 

 Division a. 



It is worth remembering that there may be a yet earlier 

 stage of double star evolution : where we have a single 

 star in rapid rotation, the separation into two distinct 

 bodies not having as yet taken place. Such rapid rotation 

 would produce a widening and a haziness of the lines — 

 a " Division h " spectrum, though differing in character 

 from that of the close binaries. This would not be 

 periodic in its character, and so not demonstrate itself by 

 the test just mentioned. 



Division c stands on a different footing, and appears to 

 point to a real difference of constitution. The stars, 

 however, of this division are so few in number tliat the 

 progress of the groups cannot be followed out with anything 

 like the distinctness of Division a. 



Annexed is a copy of Miss Maury's Table I., which 



shows at a glance how the stars observed are distributed 

 amongst the various groups and divisions. The numbers 

 in the lirst column refer to Secchi's types ; " designating 

 Orion stars, ' ' composite spectra, /- bright-line stars. The 

 last column gives the grouping of the Draper Catalogue. 

 Under the heading " Division " the sub-heads nc and ah 

 indicate forms intermediate between Divisions a and <• 

 and a and /• respectively ; the sub-headings c h, and ah, 

 <(c signify spectra which cannot certainly be assigned 

 to either division, owing to the faintness of the star or 

 the imperfections of the photographs. Peculiar spectra 

 are ranged under the sub-head P. 



CLiaSIFICATION OF SrECTBA. 



Beside the two tables given above, very complete tables 

 are supplied of the wave-lengths of the lines found In 

 the different classes of spectra, Fomalhaut being taken as 

 the representative of the first type stars, and the star 

 H.P. 1311 for the fifth type. The solar lines are cata- 

 logued from the D lines of sodium to the line p of hydrogen 

 far in the ultra-violet, and the lines of Division c stars 

 from ^ to K of the hydrogen series of lines. The complete 

 catalogue of the 681 stars arranged in order of E.A. is 

 supplemented by tables in which they are arranged in 

 order of spectral group, and copious notes add much 

 important information as to the details of Individual 

 spectra, whilst a minute description of the classification, 

 group by group, occupies the longest chapter in the work. 



In view of Prof. Ramsay's striking discovery of helium, 

 one naturally looks eagerlf to find the place accorded to 

 that spectrum in this classification. The work was, 

 however, too far advanced at the time when the helium 

 spectrum was revealed to us for it to be taken account of 

 in the actual classification. All that could be done was 

 to add a supplementary note. In this we are given a table 

 in which the helium hnes are compared with those of the 

 Orion stars, and are told that all the series of both helium 

 and parhelium are represented in them. It appears, 

 further, that nearly all the lines of the first subordinate 

 series of both helium and parhelium are very strong in 



