ok...been studying Gleb's illustrations... Please advise of incorrect interpretations and conclusions, if you have time.
Top figure of a symmetrical source/knife layout...flat and test optic (paraboloid) are square. Left knife edge image - knife horizontal and entering the beam from below. The knife is testing the tangential aberration of the image. The test optic (I like to say the image) appears undercorrected. Lots of Foucault testers are made with a lateral source and knife edge offset, with a vertical knife edge. So this figure applies to that test configuration. May lead to an overcorrected figure, eh? Right knife edge image - knife vertical and entering the beam from the right. The knife is testing the sagittal aberration of the image. The knife appears to be outside the "best focus", and the ROC of the central portion of the image is too short. The image would appear to be overcorrected. This is the configuration of the slitless Foucault tester. So there you have it. Depending on the configuration of the Foucault tester, the image could appear either overcorrected of undercorrected thanks to astigmatism. I presume this applies equally well to the Bath depending on the source/return beam offset and the fringe tilt direction. I presume this gibes with the differing curvatures of the tangential and sagittal focal planes.
Bottom figure of an off-axis configuration...flat and test optic not square. Flat adjusted to return the beam on top of the off-axis source. This mimics an off-axis stellar image, except for doubling the coma, right? Left knife edge image - knife horizontal from below. The knife is testing the tangential aberration of the image. The image is a mess and if observed on a mirror being figured would cause gas. Needed correction is asymmetric. Top of image too long, Bottom of image too short. Right knife edge image - knife vertical from right. Top of image too short. Bottom too long. More gas pains.
I used to get paid for thinking this hard. You guys spit it out like cheap beer (Keystone).