PROCiRESri IN ASTRONOMY PUISEUX. 137 



The probability of a collision sometime in the globular ohisters 

 seems especially groat. They are considered, with good reason, 

 among the most curious objects in the heavens. If we suppose that 

 tlie closeness of the stars in these clusters depends on the distances 

 of the stars from tlie centers of the clusters, then we may get then- 

 real distances from their apparent distances. H. C. Plummer ^ did 

 so for M 13 ^ and found groupings which would have been predicted 

 by the theory of gases in convective and isothermal etpiilibrium. 

 This is one more fact to make us believe that in clusters as well as in 

 the nebulae the force of gravitation is absent or held in check by some 

 repulsive force. 



We indeed go yet further and ask whether the law of Xewton is 

 always appUcable among the stars relatively near for which we have 

 been able to measure the parallaxes and proper motions. We possess 

 decided evidence in favor of the affirmative from binary stars, in 

 tlie fact that theh- proper motions follow dii-ections more often par- 

 allel to the galaxy than perpendicular. But there are also motives 

 for doubt. 



W. W. Campbell, by means of his valuable catalogue of the radial 

 velocities of stars, has shown that the Orion spectrum type is always 

 associated with small velocities. This suggestion, resulting from no 

 preconceived idea, quickly underwent broader developments. It 

 has consequently become of philosophical interest. We seeni to have 

 gamed now in the old system of classifying the stars, which was 

 founded upon increasing complexity in their spectra, at the same 

 time an ascending scale for tlieii" velocities and a descending one for 

 their niasses and distances frojn the sun. Of course there are often 

 individual exceptions, and the above rules apply only when the 

 stars are averaged in groups. 



The consequences of these generalizations have been skilfully 

 followed out by J. HaUn.^ It was an advance to be able to use as 

 criteria for a classification the masses and velocities, rather than 

 the ages, temperatures, or spectrum types. The first two properties 

 are more fundamental and more apt to enter into our formuhe. The 

 existence of a correlation l^etween the masses and the velocities is 

 even more worthy of remark. It makes us wonder whether there 

 is an equipartition of energy between the groui)s of stars just as 

 there is between the niolecules of a gas in equihbrium. Such a state 

 would not have resulted under the influence of a Newtonian field of 

 force including all the stars. Such movements are rather the final 

 consequence of an mitial velocity varying widely between neigliboring 

 stars. Further, the predoniinance of yeUow stars near our sun and 

 of wliito stars farther away, the existence of an eUipsoidal distribution 

 of the trajectories m the central part of our universe, establishes 



1 Monthly Notices, vol. 71, p. 400. ^ Monthly Notices, vol. 72, p. 378. ' Monthly Notices, vol. 71, p. 610. 



