and higher order terms of spherical aberration. Table II indicates for 
each object point the radius of the annular zone which is responsible 
for the image (the "crest of the doughnut" - shown by small arrows in 
Fig. 4). The penultimate column in Table II indicates the equivalent 
F/number of the camera (a factor which determines what exposure will be 
needed to record the central"hot spot"). Finally the last column predicts 
the axicon ring spacing, the average value of which is around 4 um. Fig. 5 
shows the assembled camera, and Figs 6 and 7 show enlarged negative 
images, the ellipticity of the latter being due to the use of a flat 
photographic plate 20? off axis. 
PRELIMINARY TESTS OF CAMERA PERFORMANCE 
Sphericity and concentricity of the CENTRAX lens surfaces Were 
monitored during construction and assembly, using the F/0.75 lens of 
a Zygo interferometer. The CENTRAX is really a symmetrical triplet, which 
consists of two similar concentric meniscus lenses of fused silica, 
enclosing a glycerol solution of slightly lower index. Satisfactory 
images have been recorded using both laser and white light point 
sources, but it is also possible to use large steel balls illuminated by 
an ordinary slide projector. White light produces some variation in the 
paraxial focus error but this has little adverse effect on the central 
portion of the axicon image. 
Experiment suggests that photoelectric synchronous detector systems 
will make it possible to sense the position of these negative axicon 
images to * 50 nm r.m.s. error per setting. Visual settings to + 100 nm 
have been achieved for short periods using a suitable microscope and 
graticule. 
; o 
A one dimensional series of 21 exposures at 2. intervals has been made 
using the CENTRAX camera mounted on a Moore 1440 division serrated angle 
table, whose angular errors are known to be less than 0.1 arc seconds. 
The positions of these images were measured by setting visually on a 
measuring microscope equipped with an optical interferometer which measures 
displacement in A/8(79 nm)increments (Downs and Raine 1979). The results 
were corrected for * 130 nm of error caused by the measured departures from 
flatness of the 6 mm thick 10E75 photographic plate, and they were then 
compared with theoretical values of the form Xn ^ 7g tan (2n?) where the 
93