• Ralph's chart is good because it shows you the effects of diffraction (coloured lines), also combined with (I assume) typical lens limitations for each format, though I don't know what lenses he tested with.

Anyway, the point is that the size of your diffraction circles of confusion on the film is constant with respect to f-stop, independent[1] of the lens' focal length. The reason is that the ANGLE the light diffracts at is a function of the aperture's absolute size but with longer lenses, the light has further to travel between aperture and film. Say you use f/8 at 50mm for a 6.25mm aperture, the light diverges at some angle theta for 50mm, resulting in a circle of confusion diameter of x50=theta*50. Say you shoot the next frame at 200mm f/8; the aperture is now 25mm. The light diverges at an angle of 0.25*theta (because the aperture is 4x larger) but travels 200mm (4x as far), so you get x200 = (0.25 * theta) * 200 = x50.

jp498: no, that is not correct. Yes there is always diffraction, but it actually spreads out as you squeeze the aperture - think of a spray nozzle on a hose as you squeeze the handle. It's not like there's a core of undiffracted light plus the diffracted light, one of which is changing in quantity. There is a just a single blur of light (gaussian distribution) and it changes size with aperture.

[1] to within the small-angle approximation of sinx = x and making both thin-lens and unity-pupil-magnification assumptions.