Originally posted by Lowell Goudge With a variable aperture zoom lens, it is important to shoot and meter at each focal length, because the aperture changes, and exposure at 55 mm will be different from 300 mm as an example.
That will only be the case with a legacy manual zoom lens with a variable maximum aperture, where the K10D body doesn't know either the lens's maximum aperture or the actual taking aperture and so can only use stop-down metering. In that situation, non-linearity caused by the focusing screen's optimisiation for a bright viewfinder image will cause progressive amounts of exposure error at smaller apertures.
In this case, the OP is having trouble with a modern autofocus lens that transmits the relevant data for open aperture metering to the camera, so stop-down metering isn't an issue. And with the stock screen at least, the K10D has got no issues at all metering a modern zoom with a slow maximum aperture -- it's designed to work with the 18-55mm f/3.5-5.6 after all. This forum's K10D Club is full of members happily using various modern zooms with slow apertures at the long end, with no exposure problems at all.
Assuming that the OP's lens is sending the aperture data to the camera correctly (which he hasn't confirmed yet), the underexposure can only be caused by non-linearity resulting from the optimisation of the Katzeye screen for a bright viewfinder image. Darkening of the centre split prism at f/5.8 couldn't possibly cause underexposure, but only overexposure. And the Katzeye's characteristics must be wildly non-linear indeed if it works fine with the OP's 20-40mm at f/4 but then causes two stops of error with the one stop (plus a tiny fraction) darker 50-300mm at f/5.8.
I'm afraid I'm going to have to keep stressing this distinction, otherwise in a few months time we'll have people in other threads declaring it to be true writ that split prisms cause underexposure with slow lenses -- when they don't and can't.