> Except for those with floating elements, I don't see how this is possible.
Those with floating elements CAN address that problem, but floating
elements have multiple uses.
> ". . . longer or shorter . . ." is ambiguous in this context - physically? in
> focal length?
Focal length. The 14-54 is a weird beast when it comes to that. That
whole series of Olympus lenses has some super-strange traits. The 7-14
is one such lens which will freak you out when you zoom while looking
through the viewfinder. Some parts of the image are coming towards you
while other parts are moving away. It's a carnival ride.
> I imagine corners, as that's the way all the makers write their specs. I've
> always disliked that, as the wide dimension of the format as the useful
As the OMZ 18mm is such a highly corrected lens, I think that it's
usability is that of 18mm as compared to other similar wide-angle
lenses. Once we adjust other lens' images to the same linearity and
crop the results, their effective focal length is reduced. This was
especially true with the OMD line where the lesser-grade lenses may
have the same diagonal angle of coverage, but have more distortion.
This is where side-to-side (and top-to-bottom) measurements are more
valuable as they indicate what the "usable" angle of coverage is.
Diagonal assumes 100% image area too with varying H/W ratios. Cropping
to a common dimension (long edge) is the great leveler.
In all honesty, it's just so much noise. Grab the lens, and determine
whether or not it's wide enough and linear enough for you with low
levels of apparent correction. The 14-54 has visible correction in the
corners at 14mm. The 14-42 have very little, but has more bowing. The
Leica 14-50 was another lens which had distinct advantages over the
14-54, but the distortion was very visible when you placed the horizon
near the top or bottom of the image.
Themed Olympus Photo Exhibition: http://www.tope.nl/