2 positions; h = 3 f and ^ — 5 f. Galileo repeats the setting of each grid point 5
times and introduces the mean as observation.
The examination of the Stereocartograph Poivilliers/SSOM Model B is carried
out by this firm according to a method which is considered best for this instrument.
For adjustment and control the plate carriers are separated from the projection
mechanism and the registration installation. The plate carriers are examined in a
photogoniometer; the remaining mechanical part of the instrument is controlled
with a view to the projective relation between the spatial direction of the rays and
the direction of their projection on xy and yz planes. No grids are used for the
production and plotting of spatial photographs. Of course also on a Poivilliers SOM
instrument it is possible — after completion of the adjustment — to carry out a grid
measurement as control of the instrument, as is done for the abovementioned
machines.
When comparing these old test methods with the Pennington proposal it will
be seen that this proposal resembles the conventional methods; the choice of hi
measurements only being much smaller than usual. This raises a fundamental quest-
ion. Mr. Pennington contends that when extensions are introduced, a standardized
test method becomes useless; the test for the applicability to all instruments becomes
too complicated and the number of measured figures so large that an analysis and
evaluation of these results would be too difficult.
However, two purposes must be distinguished. If the user of an instrument
requires a rapid working control, he will limit the number of his measurements
as much as possible. If, however, a proper control is required to discover the
various sources of error, or if the accuracy of an instrument as such is to be judged,
the number of measurements proposed by Mr. Pennington is absolutely insuffi-
cient. The problem of properly testing a plotting machine cannot be simplified
without greatly reducing the significance of this test. The best proof of the
importance of such extensions is the fact that nearly all are to be found in the
methods applied by the manufacturers.
I shall therefore indicate and explain below certain necessary extensions of
the Pennington proposal.
1) The accuracy in x and y is not only determined stereoscopically, but also
monocularly. To this end the two plate carriers are levelled, the principal
distance and zero point are determined and the coordinates of the grid points
are measured left and right.
2) It is necessary to execute the grid measurements at two or better still at three
different heights, choosing the value in such a way that the whole plotting
space is used. For a certain value of h the instrument may be adjusted in such
a manner that the errors dx and dy are practically eliminated; thus the Pen-
nington test might produce very good results, though at quite different heights
the errors might be quite troublesome.
For a control of the inner orientation it is absolutely necessary to adapt Mr.
Pennington’s proposal to industrial practice. If for a height a base length is
set, the value of which is assumed to correspond to the angular field of the
plate carrier and if the grid model is orientated stereoscopically — as proposed
by Mr. Pennington — then for instance unequal errors in the setting of the
principal distance might be eliminated by equal adjustment of tip; this latter
adjustment will show only a slight deformation of the space model. With equal