170 COSMOS. 



mum, in so far as its diminution of light does not proceed 

 uniformly ; but, after having been for some time tolerably 

 rapid, it comes to a stand, or at least exhibits a very incon- 

 siderable diminution, which suddenly becomes rapid again. 

 In some stars it would almost appear as though the light 

 were prevented irora fully attaining a second maximum. In 

 X Cygni it is very probable that two periods of variability 

 prevail — a longer one of 100 years, and a shorter one of 8^. 



The question whether, on the whole, there is greater reg- 

 ularity in variable stars of very short than in those of very 

 long periods, is difficult to answer. The variations from a 

 uniform period can only be taken relatively ; i. e., in parts 

 of the period itself. To commence with long periods, x Cygni, 

 Mira Ceti, and 30 HydrsB must first of all be considered. In 

 X Cygni, on the supposition of a uniform variability, the devi- 

 ations from a period of 406'0634 days (which is the most 

 probable period) amount to 39-4 days. Even though a por- 

 tion of these deviations may be owing to errors of observa- 

 tion, still at least 29 or 30 days remain beyond doubt ; i. e., 

 one fourteenth of the whole period. In the case of Mira 

 Ceti,* in a period of 331-340 days, the deviations amount to 

 5'5-0 davs. even if we do not reckon the observations of David 

 Fabricius. If, allowing for errors of observation, we limit 

 the estimate to 40 days, we still obtain one eighth ; conse- 

 quently, as compared with x C^Yo^^i' nearly twice as great a 

 deviation. In the case of 30 Hydrse, which has a period of 

 495 days, it is still greater, probably one fifth. It is only 

 during the last few years (since 1840, and still later) that the 

 variable stars with very short periods have been observed 

 steadily and with sufficient accuracy, so that the problem in 

 question, when applied to them, is still more difficult of solu- 

 tion. From the observations, however, which have as yet 

 been taken, less considerable deviations seem to occur. In 

 the case of rj Aquilse (with a period of 7d. 4h.) they only 

 amount to one sixteenth or one seventeenth of the whole pe- 

 riod ; in that of /3 Lyrse (period 12d. 21h.) to one twenty- 

 seventh or one thirtieth ; but the inquiry is still exposed to 

 much uncertainty as regards the comparison of long and short 

 periods. Of (3 Lyree between 1700 and 1800 periods have 

 been observed; of Mira Ceti, 279 ; of % Cygni, only 145. 



The question that has been mooted, whether stars which 



* The woi-k of Jacques Cassiui {Elimcns d^ Astronomie, 1740, p. QQ^ 

 f)9) belongs to the earliest systematic attempts to investigate tb^ nieau 

 duration of the period of the variation of Mira Ceti. 



