June 30, 192 1] 



NATURE 



559 



Such a distribution of density would be unstable, 

 and it can be proved that nuclei would form at 

 .approximately equal distances, around which the 

 matter of t'he arms would condense. In this way 

 it is possible to explain the nuclei and condensa- 

 tions which are observed in the arms of the spiral 

 nebulae. It is also found possible to calculate the 

 amount of matter which will condense around each 

 nucleus ; the mass of each is found to be of the 

 order of magnitude of the known masses of the 

 stars. 



In this way I have been led to conjecture that 

 the spiral nebulae are whirling- masses of gas 

 which, owing to their rapidity of rotation, throw 

 off gaseous stars much as a "Catherine-wheel" 

 firework throws off sparks. If so, the condensa- 

 tions in the arms of these nebulae are stars in the 

 process of birth. Dynamically the mechanism is 

 almost identical with that imagined by Laplace as 

 resulting in the birth of systems of planets and 

 satellites, but on a far more stupendous scale. 

 The final product of the chain of events we have 

 been considering must be some type of star- 

 cluster — perhaps a globular star-cluster, or pos- 

 sibly an '* island-universe " similar to our galactic 

 system. The difficulties in the way of an exact 

 mathematical investigation into the history of the 

 ejected gas as the filaments condense around 

 nuclei and as these form stars and begin to move 

 as detached bodies are enormous. On the other 

 hand, the determination of the final steady states 

 possible for a system of stars created in this way 

 is quite simple. There is found to be only one 

 type of final steady state possible for a system of 

 stars created out of a rotating mass of gas, and 

 this shows exactly the features presented by 

 the system of stars of which our sun is a member. 

 The system of stars will be of a flattened shape, 

 symmetrical about the plane of greatest cross- 

 section (the galactic plane in our system) ; the 

 velocities in any small region of space will not 

 be distributed at random, but will show a prefer- 

 ence for two opposite directions ("star-stream- 

 ing ") ; these directions will be parallel to the 

 plane of symmetry and perpendicular to the radius 

 to the centre of the system. This last direction 

 is that given by Charlier for the direction of 

 " star-streaming " in our system. CXir system 

 passes all tests for having been born out of a 

 spiral nebula the plane of which was what is 

 now the plane of the Milky Way; indeed. Easton 

 and others have claimed to find traces of the 

 two spiral arms still surviving in the distribution 

 of stars in this plane, as though the final steady 

 state had not yet been reached. 



Let us now turn to a study of the lives of in- 

 dividual stars. To the naked eye the stars appear 

 as mere points of light of varying brightness. 

 The telescope adds little except possibly differ- 

 ences of colour. The spectroscope appears at 

 first to add a wealth of new information, but a 

 detailed study of stellar spectra discloses the un- 

 expected fact that all stellar spectra, apart from 

 a few exceptions, fall into one single linear series. 

 Photographs of the spectra of all stars, in which 

 NO. 26q6. VOT.. 107! 



varying exposures have been made to compensate 

 for varying brightnesses, can be arranged 

 uniquely in a consecutive order in which each 

 spectrum differs only imperceptibly from its neigh- 

 bour. All the complicated diversities of stellar 

 spectra appear to be determined, in the main, by 

 one single variable. This is believed, with good 

 reason, to be the temperature of the star's surface. 



Positions on this linear series are specified by 

 reference to six selected points denoted by the 

 letters B, A, F, G, K, M in this order. The 

 order given is that of decreasing surface tempera- 

 ture. Stars having B-spectra are of bluish colour 

 with a surface temperature of 10,000° C. or more. 

 Stars of type M are red with a surface tempera- 

 ture of only 3000° C. Our sun is of type G, 

 with a surface temperature of about 6000° C. 



We might also arrange the stars in order of 

 brightness. The distances of many stars are 

 known, and for these we can calculate the "abso- 

 lute brightness " or "luminosity " — i.e. the amount 

 of light emitted as compared with our sun. Since 

 the masses of the stars are all approximately the 

 same, it might be expected that the order of 

 " luminosity " would prove to be substantially the 

 same as that of surface temperature, but this does 

 not prove to be the case. Eight years ago it 

 was found by Hertzsprung and H. N. Russell 

 that the red M-stars fell into two widely different 

 classes, one class having abnormally high lumin- 

 osity, and the other abnormally low. The ratio 

 of luminosities in the two classes is of the order 

 of 10,000 to I, and since the surface temperatures 

 are the same, this ratio must imply a correspond- 

 ing ratio in the areas of the radiating surfaces. 

 Thus the two classes of M-stars must have volumes 

 in a ratio of about 1,000,000 to one, for which suffi- 

 cient reason they have been designated " giants " 

 and "dwarfs." From a comprehensive discussion 

 by Russell, recently confirmed by Adams and Joy, 

 it is clear that the demarcation between "giants " 

 and " dwarfs " extends, although with diminished 

 intensity, through the types K, G, and F, while at 

 types A and B the classes coalesce. 



Lately Shapley, by determining the distances of 

 the globular clusters, has greatly increased our 

 knowledge of stellar luminosities, and has calcu- 

 lated the individual luminosities of 1152 giant 

 stars in clusters. If we plot the logarithms of 

 the luminosity (or the absolute magnitude) against 

 spectral type as in Fig. 2, the vast majority of 

 Shapley's 1152 stars are found to lie within the 

 belt marked "giants," while of the stars previ- 

 ously discussed by Russell and by Adams and Joy 

 nearly all lie either within this belt or within that 

 marked "dwarfs." In this diagram a few typical 

 stars have been marked. The stars o Ononis and 

 our near neighbour Lalande 21.185 ^^^ examples 

 of giant and dwarf red stars. The diameter of the 

 former has recently been found by direct 

 measurement to be about 300 times that of our 

 sun, corresponding to a density of the order of 

 at most one-thousandth of that of atmospheric 

 air; the latter has a luminosity only 0009 times 

 that of the sun, and probably a mean density com- 



