454 D r - Herschel’s Observations of a. Herculis. 
In order now from this table to obtain the time of the pe- 
riod, we shall first take all the successive observations from the 
16th of September till the 28th of November. They shew 
very clearly that the star has completely gone through all its 
changes. For* admitting a maximum of the light of a Her- 
culis to have been the 16th of September, we find a minimum 
the 25th of October ; and a second maximum about the 28th 
of November. The period, therefore, is of somewhat more 
than two months duration. 
But as changeable stars are subject to temporary inequali- 
ties, which will render a determination of the length of a pe- 
riod, from a single series of changes, liable to considerable 
errors, we shall now take the assistance of the most distant 
observations. By an inspection of the table, we find again the 
first maximum to have been about the 16th of September, 
1795 ; and the fourth the 14th of May, 1796. This being an 
interval of 241 days, in which four successive changes have 
been gone through, we obtain about 60 days and a quarter for 
the duration of the period. 
In confirmation of this computation, the table shews that 
our periodical star was very faint in August, 1795; bright 
about the middle of September ; faint towards the end of Oc- 
tober ; bright the latter part of November ; faint in December; 
bright in January, 1796; not observed in February; bright 
in March ; faint in April ; and lastly, bright again in May. 
This is just what should have happened according to the above 
determination, which, as we have seen, gives a period of eight 
weeks, four days and a quarter. Greater accuracy can only 
be obtained by future observations. 
