322 REPORT—1905. 
correct setting of that plate every time it is lowered. The zero of the réseau scale 
then coincides with the optical axis of the camera, independently of temperature 
changes. “Means are provided for making all the necessary instrumental adjust- 
ments. 
The image of the réseau scale being projected from the camera lens, ordinary 
commercial plates with sensible curvatures may be used without detriment to 
accuracy, since any resulting distortion in the picture is reproduced in the réseau 
and differentially eliminated in measures for which the réseau supplies the scale. 
To avoid the inconvenience which varying sizes of réseau intervals would occa- 
sion, the plate is placed in the measuring machine at a distance from the micro- 
scope objective equal to the focal lens of the camera lens, and with the film in the 
same relative position, so that runs remain practically constant. 
The parallelism of the camera at both stations is-secured by the use of two 
similar stands, on which the camera and a mechanically centred signal are succes- 
sively interchanged. The transverse telescope is reversed at the second station 
in a manner eliminating the effect of small errors of collimation or inclination. 
The principal error remaining is that of bisection, and it may be made very small 
with a suitable sharply defined signal. 
The lens adopted is Zeiss’s ‘ Protar, of 141 mm. nominal focal length, in a 
special mounting and with a permanent stop of £/36. The plates used are 
‘ Kdward’s medium isochromatic,’ 4 x 5 inches, and the exposures are made through 
a dark yellow screen optically plano-parallel. The advantage of cutting off the 
rays of shorter wave-length which form atmospheric ‘blue haze’ is that photo- 
graphs may then be taken showing sufficient detail for purposes of measurement in 
a range of distance extending to 12 km. or more. 
The camera is constructed of an aluminium alloy, and weighs with the changing 
box and a dozen plates only 14 lbs., so that it is sufficiently portable for mountain 
work. A 5-inch micrometer theodolite is interchangeable on the same stand, and 
serves to connect the ends of the base with the triangulation of the country. 
The measuring machine is essentially similar in principle to those which are 
used for the measurement of stellar photographs, except that a pair of plates in 
stereoscopic combination is measured together. Any portion of the picture may 
be made to appear in relief, and any point chosen to coincide, apparently in three 
dimensions, with an index attached to the micromete s. ‘I'hese consist each of a 
screw moving a glass scale, the intersection of which with a réseau line gives the 
distance of the point from the line considered. The numbering of the réseau 
intervals of the scale and of the micrometer heads is arranged to give at sight the 
successive figures of the reading of a plate co-ordinate to the nearest micron, or by 
#stimation to a tenth of a micron. The micrometer head is then turned back a 
fraction of a revolution until it butts against a stop, when it is again at its zero 
and the measurement of the next point may be proceeded with. 
The speed of measurement with the machine was found to be, without practice, 
twenty-five pairs of points an hour. 
The mechanical parts of the machine and of the camera were made by 
Troughton & Simms, the optical parts and the glass scales by Zeiss, and the 
réseau by Gautier. These firms spared no pains in endeavouring to satisfy the 
somewhat exacting requirements that were considered necessary for accuracy, 
and the resulting workmanship of the instrument is very satisfactory. 
The p.e. of a single bisection (mean at sight of back and forward rotations 
of the drums) with either micrometer was found to be, from the residuals of 200 
observation equations formed in the investigation of the scale and screw errors, 
+ 0°27 for a mark of best definition, For a réseau line it is + 0°48p. 
“Ordinary topographical objects being usually ill-defined and irregular, can 
seldom be bisected accurately with each eye singly, but, stereoscopically, the 
adjustment of their apparent distance with that of the index can be effected with 
great comparative delicacy, the p.e, of a stereoscopic bisection being from + 1 02 
to + 2°65, according to the class of object, or equivalent to a p.e. of bisection of 
+ 0°72 to 1°88, for each plate.’ 
The distortion of the camera lens was determined with great precision from 
