156 [Dr. E. J. Mills on the 



referred to takes place on a sufficiently large scale, there will 

 not only be cooling down attended by periodic evolution of 

 heat, but, if the heat be sufficiently intense, light will be 

 emitted, and the site of these operations will vary periodically 

 in brightness. Hence the star will have what is called a 

 " variable *" character ; and a star of this kind may be regarded 

 as in the process of forming its elements. It is probable that 

 all stars are, or have been, " variable " stars. 



Classification of the Elements. 



An element may be defined as one of a list of comparatively 

 simple substances whose numerics are of the form 



y= (1,2,3.. ..)15-15(-9375y. 



The position of any numeric in a classification will depend 

 upon the periodic, polymeric, or group number p = l, 2, 3 . . 

 . . . ., and upon t. I have calculated these positions for all 

 known numerics, and give them in the accompanying Table. 



The classification of the elements according to the genetic 

 method of the Table leads to some results of considerable 

 interest : — 



1. Polymers. — The only known polymers of the primitive 

 matter are As, Sb, probably Er, and perhaps Os. I have 

 placed these under £ = in a later, rather than under t= oc in 

 an earlier group, because they appear to me to belong natu- 

 rally to the places assigned to them. Their respective numerics 

 are 74*92, 119 s 96, and 165 ; all obviously multiples of 15. 

 Zr, Ru, 8m, and Pt closely approach the positions of other 

 polymers. 



2. Periodicity. — The polymers whose numerics are multiples 

 of 15 are the first or prime indications of periodicity. Other 

 periodicity must be sought within the groups, as a function 

 of t ; or from group to group, as a function of p ; or as a 

 function of p and t conjointly. 



Of periodicity within the group there is a beautiful instance 

 in Groups II. and IV. ; the numerics of 0, F, Na, Mg in the 

 former group being exactly imitated, five divisions further to 

 the right in the latter, by Ti, Cr, Mn, Fe. [It is difficult to 

 resist the impression that Ni = 57*84 (£ = 30), and that it cor- 

 responds to Al in the same imitation ; we can also see that, 

 when Si or Co is exactly known, Co or Si will be fairly within 

 our reach.] 



A similar example occurs in Groups I., II., III., YL, where 

 the respective values of £=J0, 12, 14, 18 correspond to the 

 analogous elements Li, Na, K, and Bb. Again, F, Na, Mg 



