[Leica] Lens Formulas and Depth of Field (35mm Summicron formulas)

[Jeff Moore <jbm@instinet.com>]

I have a question for the experts; I'd be particularly interested in
hearing what Erwin might have to say, should he have a free moment.

[N.B.  Please excuse what may be a technically imprecise usage of the
word `sharp' below...]

I'm in the habit, whenever I acquire a new lens or body or have a
body adjusted, of shooting a series of frames of a static setup,
including some resolution test targets at staggered distances and
a few 3D objects at various distances from the presumed plane of
focus, mostly to confirm rangefinder/lens calibration but also to
acquire a few datapoints (potentially of little practical value) about
the rendering characteristics of lenses.  Some photos taken under
identical circumstances with the current 35mm Summicron ASPH and the
immediately-previous 35mm Summicron were very interesting.

At apertures from, say, f/2 to f/4 the Summicron ASPH is dramatically
sharper at the plane of focus than the previous Summicron at the same
aperture; but objects behind the plane of focus definitely grow unsharp
more rapidly with the ASPH -- and it's not just that they seem unsharp
by comparison with the sharp bits.  As you move back, ASPH out-of-focus
objects soon grow *less* sharp at the same aperture than the older
formula.

Now, I've always heard that depth-of-field is a function of focal
length, distance and aperture -- optical formula isn't supposed to enter
into it -- and while at some level this has to be just a convenient
approximation, since there are optical formulas which are demonstrably
more sharp than others, what I've seen seems more dramatic than I'd
expect given the widespread acceptance of the depth-of-field `rules'.

BUT: there's a potential explanation.  I also noticed that the
diaphragm-shaped highlights produced by specular reflections in
out-of-focus areas seemed larger in diameter at the same nominal f-stop
with the ASPH than with the pre-ASPH.  If the ASPH formula happened to
transmit light less efficiently than the pre-ASPH, and if the marked
`f-stops' of the lenses were calibrated to transmittance rather than
physical aperture size (this would, after all, be the choice with the
most utility given that people will wish to use external light meters
and non-TTL flashes and such), then the two lenses could have different
physical apertures at identical marked apertures, and would have the
depth-of-field characteristics appropriate to their respective physical
apertures.

This is now my working assumption.  Comments?

 -Jeff Moore <jbm@instinet.com>