108 G. H. KNIBBS. 
(a) Arrange the contact pieces on either side of the test-piece 
and parallel to its axis, with the mirrors at the same end'—as 
shewn in Fig. 3 in Prof. Warren’s paper—and so clamped that 
the stems of the mirrors shall be vertical in all directions when 
the test specimen is horizontal, or horizontal and parallel to one 
another when it is vertical. 
(b) Set the handle-bars BB’ Fig. 3, above mentioned, parallel to 
the longitudinal axis of the test-piece. This will fix the prisms at 
right angles thereto, if the handle-bars have been adjusted as 
described in § 6. 
(c) Set each scale parallel to a plane, passing approximately 
through the pivots of the mirror, and at right angles to the stem 
and prism, and let the scale be also approximately parallel to the 
test-piece. The centre of the object lens of the telescope, and the 
longitudinal centre line of the scale, should lie on either side of 
this plane, and while being placed as near as possible to it, should 
be equidistant from it. This last disposition of the apparatus 
eliminates any errors which might arise from pivot errors. 
_ (d) So arrange the zero of the scale that a plane perpendicular 
thereto, containing the zero line shall pass through the centre of 
the mirror, the stem and the prism; and let the telescopic sight 
line be as nearly as possible coincident with this plane and direct 
to the centre of the mirror. Then by rotating the mirror about 
its axis and about its pivots PP,’ the zero of the scale may be 
seen, and the small adjustment thereto may be perfected by means 
of the slow motion screw. The scales are set in opposite way 
that is the reading runs say from left to right in one scale, from 
right to left in the other, this of course is not essential, nor is it 
necessary that the initial reading be zero; but the arrangement 
facilitates the application of the correction, see § 5. 
The correction will reduce the scale reading, that is to say, the 
error of the scale is always one of excess, with the exception bs 
SIE PRE MIRON I wt 
1 This is necessary so that the errors due to longitudinal movement 
and to rotation of the test-piece shall be eliminated. 
fie Zea gee era 
Peete te as 
