13 
evidence as to the accuracies currently obtained. Each 
of the numerous laboratories now making MTF measurements 
can probably repeat their data on any one lens with sat 
isfactory precision, and so long as they continue to test 
lenses of the same type the results will indicate the 
relative qualities of the lenses in an adequate way. Thus 
the requirements of production control can be satisfied. 
But if measurements are made on the same lens in another 
laboratory, probably using a basically different apparatus, 
the results may differ widely. Also, if any laboratory 
tests lenses of very different types, e.g, changes from 
very large aperture to very small aperture lenses, there is 
no guarantee that the results will correspond to a measure 
ment of the true MTF in each case. In fact, the test des 
cribed in Appendix I revealed discrepancies as great as 
- at quite low spatial frequencies when one lens was 
measured by several laboratories, each of which estimated 
its accuracy as 3% or better. 
It is unfortunately true that so far no satisfactory 
techniques have been developed for the complete proving and 
qualification of MTF apparatus over the whole range of con 
ditions in which it is likely to be used. It is not easy to 
devise suitable checks and calibrations for this purpose. 
Diffraction limited lenses can be used, but this is restr 
icted to small apertures and monochromatic light. Another 
possibility is the design and very careful construction of a 
simple (e.g, plano-convex) lens, whose MTF could be calculated 
and for which all the aberrations would be very accurately 
known. Within the measuring apparatus itself much can be done 
to isolate sources of error. The shape and modulation of 
sinusoidal masks, their scattering characteristics, uniformity 
of illumination over masks, slits, and photocell apertures, 
all require careful checking. The linearity of amplifiers 
and recorders is another possible source of error. The 
spectral response of the photocell detector should be measured 
rather than assumed from representative data. In techniques 
that enlarge the spread function with a microscope objective, 
the abberrations of the objective, and their combination 
with the aberrations of the lens being tested require examin 
ation. Similar remarks apply to collimators and folding mirrors. 
Problems of vibration and air turbulence are more serious than 
in the less sensitive resolving power measurement. Problems 
associated with the coherence of the illumination have not 
received enough attention. 
Overall, the aim is to ensure that each technique 
genuinely measures the OTF and MTF. Unless this is achieved 
different laboratories will not agree, however, consistent 
each of them may be, and reliable standard data for lenses 
will not become available to the prospective purchaser.