169 
ajloating collimator. 
On examining the above table, it appears that by taking 
the mean of a very few results, the greatest error, if the 
experiments with the wooden float be rejected, is four- tenths 
of a second, consequently the place of the horizontal point 
may be speedily determined by the use of the collimator, to 
the utmost degree of accuracy which the astronomical circle 
employed, is capable of attaining. 
The results obtained by turning the collimator round with- 
out removing the float from the mercury, might have been 
expected to have been very nearly, if not wholly free from 
error ; but as this does not appear to be the fact, and as the 
errors are all in defect, they seem to have been influenced by 
some constant cause, which, as before remarked, I believe 
to have been expansion of the stand of the micrometer in 
consequence of increased temperature. 
When the float is removed in order to transport the box 
containing the mercury to the opposite side of the observa- 
tory, the manner of replacing it, so as to occasion the least 
error, seems to be that of bringing the edge of the side of 
the float first in contact with the mercury, and then gra- 
dually lowering it. This mode of removal can be necessary 
only when the collimator is used with a portable circle ; but 
in a fixed observatory a plank should be laid, or a sort of 
railway contrived from one support to the other, on which 
the collimator should be either slid or passed along on rollers 
without removing the float from the mercury ; by this ar- 
rangement the greatest, and perhaps the only source of error 
would be avoided.* 
* It may perhaps be found preferable to have two boxes with mercury, and to 
carry the float from one to the other. 
MDCCCXXV. Z 
