173 
a floating collimator. 
same point in its focus, and consequently, whether the colli- 
mator be placed above or below the axis of the telescope of 
the circle, so long as the cross wires continue visible, the 
image will suffer no change of position. This affords an ex- 
cellent method of discovering any want of parallelism in the 
rays ; for if on placing the collimator as much above the axis 
of the telescope as possible, without losing sight of the cross 
wires, the image appears elevated above the horizontal wire, 
or if on placing it below the axis, the image appears to have 
descended, it is a proof that the rays falling upon the object 
glass of the circle are not parallel, but that they converge, 
and consequently that the cross wires of the collimator are 
too far from its object glass, and vice versa. It is necessary 
that this adjustment should be made with the utmost care. 
It might possibly be supposed that the accuracy of the 
collimator would be augmented by increasing the length of 
its telescope, but this is not the case. It is the direction of a 
ray passing through the cross wires, and the centre of the 
object glass of the collimator, which is the subject of observa- 
tion ; aiid the direction of this ray is as definite in a telescope 
of an inch in length, as in one of ten feet focus. The degree 
of precision with which any variation in the horizontal incli- 
nation of this ray can be estimated, depends upon the length 
and power of the telescope employed to view the cross 
wires, and not upon the length of that of the collimator. 
I'here is an inconvenience however in using a telescope of 
too short a focus, as the cross wires are very much magni- 
fied, and consequently appear not so well defined ; in addi- 
tion to which, if a permanent point of reference be required, 
there might be some fear that the relative positions of the 
