1889.] Variability in condensing Swarms of Meteorites. 403 



swarms of meteorites, the variability is produced by the revolution of 

 one or more smaller swarms round the central swarm, the maximum 

 luminosity occurring at periastron passages, when the revolving 

 swarms are most involved in the central one. 



Fig. 1 illustrates this suggestion in the simplest case, where there is 

 only one revolving swarm, as in Mira Ceti. The range of variability 

 depends upon the eccentricity of the orbit and the periastron distance 

 of the revolving swarm. 



According to this theory, the normal condition is that which exists 

 at minimum, and in this respect it resembles that suggested by 

 Newton, namely, that the increase of luminosity at maximum was 

 caused by the appulse of comets. All other theories take the maxi- 

 mum as the normal condition and the minimum as a reduction of the 

 light by some cause, such as a large proportion of spotted surface or 

 eclipses by dark bodies. In the variables of the Algol type, where 

 the periods are very short, there can be no doubt, after the Henry 

 Draper Memorial photographs, that the eclipse explanation is the 

 true one. But in the variables of Group II, where the period is 

 about a year and the luminosity at maximum in the generality of 

 cases is about 250 times, though in others it runs up to 1600, that at 

 minimum (corresponding to a difference of six magnitudes), it is 

 obvious that the eclipse explanation no longer holds, on account of 

 period, and also that the spotted surface explanation is inadmissable 

 on account of range. If, however, the minimum be taken as the 

 normal condition, and the effects of the revolution of such a swarm 

 as I have assumed be considered, both length of period and range of 

 variability can be explained. In this class of variables the rise to 

 maximum is more rapid than the fall to minimum, and, according to 

 my explanation, the sudden increase is due to the first collision 

 between the two swarms, while the fall to minimum represents the 

 gradual toning down of the disturbance. 



Tests of the Theory. 



In the Bakerian Lecture (p. 84) I showed how this explanation of 

 variability bore four distinct tests. The first test was that Group II 

 should be more subject to variability, than any other group ; and I 

 showed that 1 out of every 7 stars of Group II are variable, whilst 

 only 1 in 659 of the stars included in Argelander's catalogue are 

 variable. Tbe other tests were : — (2) when the swarm is least con- 

 densed, we shall have the least results from collisions ; (3) when it 

 is fairly condensed, the effect at periastron passage (if we take the 

 simplest case, where there is only a single revolving swarm) will be 

 greatest of all ; (4) in the most condensed swarms there will be little 

 or no variability, because the outliers of the central swarm may be 



