(234)
The 45 fiducial mark projectors are hidden among the light baffles in the
body of the camera. Each projector consits of a lamp, a reticle, and a lens
whose focal length is about 2 cm. An auxiliary glass focal plane calibrating
plate containing an etched cross for each collimator cross and fiducial mark is
placed on back of the camera. The glass plate has large holes through which
one can adjust the lenses and projectors.
The adjustment of the nine-lens camera consists of (1) spacing each lens-
mirror unit at exactly 45° intervals, (2) inclining each mirror exactly 19°,
(3) flattening the mirrors within optical tolerances, and (4) placing the fiducial
projectors in their correct positions. The operations are performed at a special
laboratory twenty miles from Washington where the vibrations of heavy
traffic, elevators, street-cars, etc. are minimum. The laboratory consists mainly
of a steel framework for supporting the camera, and a group of optical
collimators projecting their crosses at predetermined angles toward the camera
and its mirrors. The angles of the collimators are set with a one-second
theodolite placed at the camera station.
The position of the center lens is adjusted so that the cross from an
illuminated vertical collimator set beneath the camera coincides with the center
cross of the calibrating plate.
Three optical collimators are set at 0°, 45° and 90° relative to the center
of the camera. With-the camera level, if one chamber of the camera is set with
the 0° cross in coincidence with the corresponding cross of the calibrating
plate, then one cross from the 45° collimator, and one from the 90° collimator
should also be in coincidence. In addition, five other collimators also show in
the same chamber as the 90° collimator. If coincidence is lacking, the angular
setting of the mirror of the 90° chamber can be adjusted by means of three
of the nine screws that fasten the mirror to the cone (the other six being loose
at this stage), shifting the lens, and raising or lowering the lens. Thus it is
possible to set each mirror and lens at a 45° spacing as well as each mirror at
a 19° inclination by analyzing the direction and magnitudes of the discrepancies
of the collimator crosses.
It is generally not possible to manufacture a relatively thin piece of
polished metal rigid enough to remain optically flat. Even though the mirrors
are 5 mm thick, they are never flat when unsupported. Consequently the
mirrors are flattened by means of the remaining six screws that fasten the
mirror to the cone, while the mirror is observed through an interferometer
system. This system consists of an observing telescope which also projects a
beam of monochromatic light, an optically flat cylinder of glass half-silvered
on the side which is placed near and parallel to a steel mirror, and an inter
vening mirror to bend the optical axis so that the observation can be made
from a convenient position for adjusting the screws. The observer sees a
pattern of interference fringes which he reduces to a minimum by turning the
six screws in or out a very slight amount.
A root-mean-square error of 0.02 mm is the residual of the collimation
crosses observed with a microscope. This discrepancy corresponds to about
20 seconds at the perspective center. The root-mean-square of the residual
collimator error is 4 seconds: thus the errors in setting the collimators has a
negligible effect on the errors of orientation. Moreover, small residual errors
of camera adjustment can be removed during the transformation-calibration