168 COSMOS. 



in proportion more considerable. No certain period of these 

 oscillations has as yet been discovered. There are, however, 

 indications of a period of 40 years, and another of 160. 



The periods of vaxiation in different stars vary as 1:250. 

 The shortest period is unquestionably that exhibited by (3 

 Persei, being 68 hours and 49 minutes ; so long, at least, as 

 that of the polar star is not established at less than tw^o days. 

 Next to (3 Persei come 6 Cephei (5d. 8h. 49m.), rj Aquilse 

 (7d. 4h. 14m.), and ^ Geminorum (lOd. 3h. 35m.). The 

 longest periods are those of 30 Hydrse Hevelii, 495 days ; 

 X Cygni, 406 days; Variabilis Aquarii, 388 days; Serpentis 

 ^., 367 days ; and Mira Ceti, 332 days. In several of the 

 variable stars it is well established that they increase in brill- 

 iancy more rapidly than they diminish. This phenomenon 

 is the most remarkable in 6 Cephei. Others, as, for instance, 

 (3 Lyrae, have an equal period of augmentation and diminu- 

 tion of light. Occasionally, indeed, a difference is observed 

 in this respect in the same stars, though at different epochs 

 in their process of light. Generally Mira Ceti (as also 6 Ce- 

 phei) is more rapid in its augmentation than in its diminu 

 tion ; but in -the former the contrary has also been observed 



Periods icithin 'periods, have been distinctly observed in 

 the case of Algol, of Mira Ceti, of /3 LyrsB, and with great 

 probability also in % Cygni. The decrease of the period of 

 Algol is now unquestioned. Goodricke was unable to per- 

 ceive it, but Argelander has since done so ; in the year 1842 

 he was enabled to compare more than 100 trustworthy ob- 

 servations (comprising 7600 periods), of which the extremes 

 differed from each other more than 58 years. (Schumacher's 

 AstTon. Nadir., Nos. 472 and 624.) The decrease in the 

 period is becoming more and more observable. =* For the 



* " If," says Argelander, " I take for the epoch the minimum bright- 

 ness of Algol, in 1800, on the 1st of January, at 18h. Im. mean Paris 

 time, I obtain the duration of the periods for 



—1987, 2d. 20h. 48m., or 59s.-416i03.-316 



—1406, '' 58s.-737i0s.-094 



— 825, " 58s.-393-l-0s.-175 



-- 751, " 58s.-154-l-0s.-039 



+2328, " 58s.-193-l-0s.-096 



--3885, " 57s.-971-l-0s.-045 



--5441, " 55s.-182-l-03.-348 



" In this table the numbers have the following signification: if we 

 designate the miniinam epoch of the 1st of Jan., 1800, by 0, that im- 

 mediately pi-eceding by — 1, and that immediately following by -|-1, and 

 BO on, then the duration between — 1987 and — 1986 would be exactly 

 2d. 20h. 48m. 593 -416, but *he duration between -{-5441 and -|-5442 



