of some of the principal fxed Stars, &c. 431 
broken in putting it together, so that I never was able to use 
it. As each apparatus for this adjustment is quite independent 
of the other, no possible inconvenience can attend their appli- 
cation, as either may be employed alone, at the option of the 
observer. But as any verification requiring many bisections 
is objectionable, I would in general certainly prefer Mr. 
Trough-ton's method, and only have recourse to the other, 
when there was reason to suspect that some alteration had 
taken place to render it necessary. 
One more circumstance respecting the instrument remains 
to be noticed : when the divisions were first examined by 
opposite readings, i",25 was the greatest possible error which 
was to be apprehended, and o",j the mean error ; but in its 
journey it seemed to have suffered some very small derange- 
ment in its form : this was discernible both from examining 
the opposite readings ; and by deducing the error of collima- 
tion by zenith stars, and comparing it with that found by an 
horizontal object, there was constantly perceived a difference 
of 3" between the error of collimation deduced from y Draconis 
and by an horizontal object ; and this quantity was very uniformly 
distributed through the intermediate arc. In what particular 
manner the observations would be affected by this derange- 
ment I will not venture to decide, but I think it most likely 
that it has only rendered the instrument rather less accurate 
than it was originally, as is above stated. I have before ob- 
served the great advantage the circle possesses of showing 
the amount of its own errors. These may be determined with 
great certainty by examining the errors of collimation as 
deduced from different stars. This method is founded upon 
the supposition that half the difference of the two extreme 
mdcccvi. 3 K 
