les. 
at 
M 
n 
irriage 
e locked 
ide is 
vision 
ove 
e and 
le 
n the 
of 
rners 
the 
ated 
the 
19 
e 
of 
ave 
ively 
f the 
within the reticle, for a lateral offset of as much as 100 microns will cause an 
error in radial distance of only 0.1 microns in the worst case. The amount of 
parallactic error caused by the gap between the scale and the plate depends on 
the effective focal length of the objective; for the B & L Macroscope, a 100 
micron radial offset will cause an error of 0.3 microns in radial distance. 
The image as seen within the reticle will lie between a pair of millimeter 
graduations of the scale. The whole miilimeter reading is taken directly from the 
lower graduation. To obtain the fractional reading, one rotates the micrometer 
drum to translate the lower graduation until it precisely bisects the image. The 
fractional reading is then made from the micrometer drum to the nearest half 
micron. The reading is recorded against point number and the process is repeated 
for all points of to be measured. When all of the measurements have been 
performed for a given position of the plate, the measuring arm is moved out of 
the way to a nearly horizontal position and the plate holder is rotated 90° and 
replaced on the comparator. The points are again measured and their readings 
are recorded against point number. The operation is completed when the plate 
has thus been measured in all four positions. 
Despite the fact that the plate is measured in four positions, the total 
measuring time is about the same as that required when double settings are performed 
on a conventional two screw comparator. This is attributable mainly to two factors: 
(a) each setting involves only a single radial bisection, rather than bisection in 
two directions; (b) slewing is very rapid, for the microscope can be moved into 
measuring position from any point onthe plate to any other point in a matter of 
about five seconds. An experienced operator can perform the complete measurement 
of a plate containing 30 images within one hour; when double settings are required 
(i.e., a total of 8 measurements per point), the time is increased to about 90 
minutes. With automatic digital readout these times are about halved. 
CALIBRATION 
A systems analysis of the comparator shows that aside from stability of the 
plate during measuring, only three elements of the system are critical to the 
attainment of the desired measuring accuracy of one micron. They are the scale, 
the screw and the pivot. All other elements require tolerances from one to two 
orders of magnitude less demanding. For example, for one micron accuracy the 
linearity of the translation of the microscope need be good only to about +300 
microns (actually, it is ten times better than this). Likewise, the axis of the 
microscope need remain parallel to itself only to within 40:6 (again, the actual 
tolerance is more than ten times better than this). Similarly, since the scale is 
translated by at most one millimeter, the direction of the translation must be 
parallel to the axis of the scale only to within 48^, a tolerance that is easily 
bettered by a factor of one hundred. 
The scale is indeed a critical item, for each measurement will inherit the 
error of the graduation to which it is referred. This means that, in the absence of 
other errors, one micron accuracies demand a scale that is either accurate to one 
micron or else is calibrated to an accuracy of one micron. The scales employed in 
the production models of the comparator are manufactured to the following 
specifications: cumulative nonlinearity of spacing is not to exceed one micron