e evidence showing 
ogrammetric lenses 
re any evidence of 
rast tests with air 
: has no high con- 
in England as the 
gnificant that high 
has probably made 
nd that it is more 
rs ago by Selwyn's 
ou do not want to 
kind of test under 
new language, we 
e that will be used 
. The writer natu- 
wrgument that high 
they give a better 
may convey infor- 
do this better); it 
s much more to be 
n the low contrast 
sity differences of 
0:2: + 0.02, if con- 
ast target gains by 
; much better with 
re precisely fixed. 
th cases a point of 
the same whatever 
e of the lens drops 
y the high contrast 
nore sensitive to a 
fact, however, is it 
's laboratory, some 
ness. The high con- 
graininess must be 
urves of resolution 
is more critical at 
confirmed this. 
ong, the writer has 
why they should be 
and have persisted 
ts they are an un- 
1es, yet without the 
lity is indeed used 
s revealed by reso- 
THE PHOTOGRAPHIC IMAGE, BROCK 23 
lution tests with those measured on the aerial image. This has no great value, however, 
and the exercise could probably be equally well done on the Howlett annulus. The writer 
would be prepared to wage a campaign in favour of the annulus target for two reasons. 
First, it must nearly halve the time for assessing the images. Second, it gives a low 
grading to lenses with a big separation of the radial and tangential foci, however good 
the resolution on radial or tangential lines may be. This is surely in accordance with the 
practical value of the lens. 
In the writer’s opinion, some of the dissatisfaction with the resolution test when 
comparing lenses with different degrees of astigmatism is due to the artificial attempts 
to average radial and tangential line figures, instead of accepting the lower of the two 
as the real criterion of image quality. 
Long line targets are valuable where micro-densitometry is done, but for pure reso- 
lution tests they appear to have no advantages. It has been objected that they give a 
spuriously high performance in grainy images by allowing the eye to join up isolated 
grain clumps, and this was one reason for changing to the shorter Cobbs. There now 
seems to be no point in this argument unless one seeks a direct correlation between reso- 
lution tests and the probability of detecting detail of particular shapes, which is a prof- 
itless exercise. 
424, Standardisation. 
In British experience the difficulties of standardising procedures so that different 
industrial and Government laboratories can get the same results have been more trouble- 
some than any fundamental weakness of the resolution test. Relevant factors are now 
discussed. 
Targets. Density difference of low contrast targets must be held within close lim- 
its, say 0.2 == 0.02. The material should be quite neutral over the visual range of wave- 
lengths. The line and space width should be as accurate as possible down to the smallest 
size, not worse than = 29%. The edges should be perfectly sharp under magnification 
and the rectangular shape (in the case of Cobbs) should be preserved to the smallest 
size. In Gt. Britain the Ministry of Aviation has now arranged for the commercial manu- 
facture of Cobb targets to these tolerances, using an evaporated metal technique which 
gives perfectly neutral deposits. This material also has very little scatter, hence the ef- 
fective density is almost independent of the illuminating conditions. Ordinary photo- 
graphic deposits have given trouble in this respect, and it has been necessary to measure 
the effective density in the collimator. 
Collimation. It is not easy to specify collimator performance in other than optical 
terms, playing safe by asking for the highest possible standard of correction and using 
the longest possible focal length. No trouble has in fact been traced to collimator faults 
in Gt. Britain, among some half-dozen testing agencies. 
Light-source. There is some evidence that the light reflected from average ter- 
rain in daylight has a greater proportion of longer wavelengths than the “noon sunlight” 
which is generally accepted as the photographic standard. Until this is confirmed noon 
sunlight appears to be the desirable standard. It is not easy to achieve for resolution 
testing if one desires to use exposure times similar to those used in practice. (Circa 10 
millisees). Tungsten sources of the necessary power to overcome the losses in the day- 
light filter give problems of control and heat-disposal. It then becomes a choice between 
long exposure times, with consequent change in the emulsion characteristic curve, and 
incorrect light-balance by omitting the daylight filter. In Gt. Britain the latter has been 
chosen, incorrectly in the writer’s view, but repeated checks have failed to show any dif- 
ference of results. (The Minus Blue filter is normally used, for photogrammetric cameras).