104 G. H. KNIBBS. 
by suitably grasping the edges in grooves in two opposed positions, 
one with the handle toward the observer, the other with it from 
him, and rotating the handle till its direction remains unchanged, 
or is identical in either position.’ The handle then becomes a 
director, by means of which the prism can be set perpendicularly 
to the test specimen. This setting can be done with care to about 
0-1 in the 30 mm., which is equivalent to 114’: with the most 
ordinary care there is no difficulty in setting to 0°2 mm. in the 
distance mentioned is to 23’. By means of the column of values 
of 2 in Table I. it is easy to evaluate the uncertainty in the 
result, due to the uncertainty of this adjustment. For example, 
suppose the uncertainty in setting be + 20’ and the readings at 
the 1135 distance to be 0 and 400; then it is easy to see from 
the table that the error of reading would be roughly about *7 mm. 
for extension, or about 1:1 mm. for compression. This indicates 
the importance of the adjustment when extreme precision is 
desired, and at the same time points out one of the limitations of 
the instrument. The difficulty in this respect, however, can be 
met by reading extensions up to 200 mm. on the scale, and then 
reiidjusting the prism at right angles for a second series, remem- 
bering of course, that this second’ series are not, in terms of the 
original length, but of the length Z/(2+e,), practically H —e, since 
the extension is very small in relation to the original length Z. 
7. Examination of the pivot-axis of the mirror.—In order that 
no error shall be introduced by rotation of the mirror about its 
pivots, it is necessary that the line joining them should be at right 
angles to the axis about which the mirror turns, and which as 
already pointed out, should be parallel to the two knife-edges. 
This may be tested as follows :—Suitably clamping the handle, 80 
that the axis, parallel to the two knife-edges, shall be approximately 
coincident with the sight-line of the telescope, place the mirror 
face approximately at right angles to this axis or line, and rotate 
till the line joining the pivots is parallel to the scale. Then by 4 
1 A suitable instrument designed by the author for effecting this 
adjustment was exhibited. 
