8 
accuracy will be required for the specification of 
lenses by MTF alone, and also for research on image 
quality. There is evidence that in the present state- 
of-the-art even 1C$ accuracy may not be obtainable in 
the measurement of lens MTF's. The lens specification 
problem is considered in Section 6. 
6. Specification of Lens Performance by MTF's 
Most of the current use of MTF's, as discussed in 
previous sections, is concerned directly or indirectly 
with system design. Since the final system performance 
is still commonly specified by resolving power there has 
been no great pressure to study and thoroughly evaluate 
the current methods for measuring MTF's. The MTF, invol 
ving purely physical and objective measurements, is pot 
entially capable of greater accuracy than resolving power 
and hence is of great interest to the user of photo- 
gramraetric equipment, notably aerial lenses. At present 
there are considerable difficulties in obtaining standard 
ization of resolving power measurements between different 
laboratories, and particularly between laboratories in 
different countries. These difficulties arise partly 
because of the inherent errors which reduce precision and 
accuracy of the measurement, due to subjective differences 
of interpretation, statistical fluctuation of values due 
to emulsion grain, and the problems of controlling numerous 
variables (e.g. target contrast, development, spectral 
balance of light) which affect resolving power. In addition, 
the particular emulsion used in one country may not be avail 
able in another, or may become unavailable at some time, thus 
rendering cross-comparisons meaningless. These problems 
suggest, of course, that MTF measurements could be of iimnense 
value in enabling the quality of lenses to be assessed acc 
urately on a common basis in units that would not involve 
subjective errors, would be free from physical errors, and 
would have permanent meaning. On the other hand, unless it 
can be shown that MTF measurements can indeed be made with 
greater accuracy than resolving power, as well as with 
greater precision, * the case for applying this presently 
more expensive technique would not be good. Judgement would 
have to be used; thus it might be concluded that the obvious 
advantages of doing without emulsions altogether would be so 
great as to justify accepting the method even if the accur 
acy were no greater than say ± 10$. Also, the MTF does, of 
course, provide more information than a resolving power test, 
but it would be hard to justify MTF measurements in which the 
accuracy was worse than 10$. 
* Precision expresses the spread about the average in 
successive measurements of a quantity; accuracy is concerned 
with departures from a "true'* value, knowledge of which depends 
on analysis of systematic errors. 
Ml 
I I