Dr. Herschei/s Method, &c. 167 
should be the angle subtended by the stars of the first magni- 
tude, if they were all scattered equally. For it will be found 
that the distances from Lyra to Arcturus ; from Arcturus to 
Regulus ; from Regulus to Sirius ; from Sirius to (3 Navis ; 
from Elgeuse to Canopus ; from Canopus to a Centauri ; 
from a, Centauri to Achernar ; from Achernar to a Crucis; from 
Procyon to Canopus ; from Fomalhaut to Altair ; and from 
Altair to Antares, agree sufficiently well with this hypothesis. 
It must also be remembered that a perfect equality in the 
mutual angular distribution of the stars that form the first 
inclosure, is a thing that is mathematically impossible, and 
therefore not to be looked for. This would authorize us to 
take in other intervals, such as from Arcturus to Antares ; 
from Elgeuse to Regulus ; from Achernar to Rigel ; from Rigel 
to Capella ; from Capella to Sirius ; from Regulus to Spica ; 
from Spica to a Crucis ; and from Rigel to Castor ; all which 
concur, in a great measure, to support the same hypothesis. 
But as the distribution and real magnitude of stars is not my 
present subject, what has been mentioned will be sufficient. 
A second layer of stars will be more extensive ; for the su- 
perficies of the celestial regions allotted for the situation of 
these successive stars exceeds the former in the ratio of 4 to 1. 
And on looking over the collection of stars which astronomers 
have pointed out as belonging to the second class, we find 
that their number is proportionally larger. 
A similar way of considering the stars of the third order 
might be applied, if it did not already appear, from what has 
been said of the two former orders, when strictly compared 
with the state of the heavens, that such kind of limits can be 
of no real use in the classification of stars. The hypothesis 
