528 



NATURE 



[June 23, 192 1 



equal images on the photograph. This is secured 

 by means of a rotating shutter, a neutral screen, 

 or the use of a grating in front of the objective. 



The purpose of (3) and {4) is to make any 

 residual errors the same for the parallax star and 

 the comparison stars, and so far as possible the 

 same on all photographs. 



The knowledge of the distance of a star gives 

 us immediately its luminosity or the amount of 

 light it emits as compared with the sun. There is 

 a very great range in luminosity even for stars of 

 the same spectral type. Now the stars have been 

 arranged in an order according to the spectra, 

 which agrees fairly well with their order in colour 

 from blue to red, and is essentially an arrange- 

 ment according to temperature. This may be 

 regarded as an extremely good first approxima- 

 tion to a classification of stellar spectra. But it 

 does not detect any differences attributable to abso- 

 lute luminosity, though presumably density and 

 gravity at the surface layer of the star from 

 which the lines in the spectrum have their origin 

 must be widely different, 



A few years ago a very fruitful investigation 

 was commenced at Mount Wilson by Adams and 

 Kohlschuter. By a close comparison of the 

 spectra of stars of the same spectral class, but 

 differing greatly in absolute luminosity, they de- 

 tected lines the intensities of which differ. Adams 

 and his coadjutors at Mount Wilson have pursued 

 this research with very great success. They have 

 found in stellar spectra a number of pairs of 

 neighbouring lines, one line of each pair being in- 

 dependent of the absolute luminosity, while the 

 other changes in intensity with the luminosity of the 

 star. They have measured the relative intensities 

 of these pairs of lines, and compared their measures 

 with the luminosities of 650 stars already known 

 through the trigonometrical determinations of 

 parallax made at Allegheny, McCormick, Mount 

 Wilson, and Yerkes. Thus they have found the 

 luminosities of stars corresponding to different 

 Intensities of the lines. They have recently pub- 

 lished a catalogue (Astrophysical Journal, March, 

 192 1) giving the luminosities and parallaxes of 

 1680 stars. 



The advantage of this method is that it extends 

 the^ range of parallax determinations beyond the 

 limit (say) 002" of the trigonometrical method, 

 the limit of the spectroscopic method being deter- 

 mined only by the capacity of large telescopes to 

 give measurable spectra. In the table a com- 

 parison is given with unpublished results at 

 Greenwich obtained by the trigonometrical 

 method : — 



Mag. at 



B 1673 

 B2897 

 B 2971 

 C 1604 



B3983 

 B4181 

 B4234 

 C 2242 

 B 4322 

 B5009 

 B 6129 



Comparison of these results, obtained by 

 entirely different methods, shows the accuracy of 

 20 per cent, claimed for Mount Wilson, and 

 + 0010" for Greenwich is reached. 



A third method is being employed extensively 

 for determining stellar distances depending on the 

 fact that the masses of stars lie within very re- 

 stricted limits. It is applicable only to double 

 stars, and depends on Kepler's third law, 

 M + m = a3/p2, where M, m are the masses, a is 

 the mean distance between the components, and 

 P the period of a double star. When P is known 

 and M + m assumed, a is found, and, further, as 

 the cube root of M + m is involved, an error in the 

 assumed mass produces a much smaller error in 

 the mean distance. Now the angular mean dis- 

 tance is determined by direct observation for all 

 double stars the orbits of which can be calcu- 

 lated. At the present time this amounts to more 

 than 150. But it has been shown by Hertzsprung 

 and Russell that for double stars which have com- 

 pleted too small a portion of their orbits for their 

 periods to be known it is still possible to obtain 

 their "hypothetical" parallax with considerable 

 probability. The method has been recently 

 applied at Greenwich to obtain the parallaxes of 

 a large number of stars, and the accordance with 

 the results found by the trigonometrical and 

 spectroscopic methods is very satisfactory (see a 

 paper in Monthly Notices R.A.S., November, 

 1920. by Messrs. Jackson and Farmer). 



I believe there is in preparation by American 

 astronomers a catalogue giving the parallaxes of 

 3000 stars, about half of which have been deter- 

 mined by two at least of these three methods. We 

 may expect that in the course of a very few years 

 the distances of all stars visible to the naked eye 

 in the northern hemisphere will have been deter- 

 mined, as well as those of many fainter stars. This 

 great accession of knowledge of stellar distances 

 carries with it a corresponding increase with refer- 

 ence to the luminosities, sizes, masses, densities, 

 and velocities of stars of different spectral classes. 



Obituary. 



, William Warde Fowler. 1847-192 i. 



WARDE FOWLER, like Arthur Sidgwick, 

 was one of the men we can least spare^ — 

 a classical scholar of distinction and a writer of 

 great charm who sympathised warmlv with the 

 NO. 2695, VOL. 107] 



aims and methods of science, and strove to give 

 them a larger place in the life of his University. 

 It would scarcely be possible to gain a clearer 

 insight into the strength and weakness of an 

 Oxford education as it was nearly twenty years 



