To answer your question - and this is something that is explained more clearly in the most recent version of the Lesson 1 ellipses section, the normal vector of the circle in 3D space, and the minor axis line of the ellipse drawn in 2D space, coincide - or at least, come close enough to coinciding in the vast majority of cases that this is the principle that is taught to help students understand the relationship between the circles in 3D space, and the ellipses they're drawing to represent them. So when analyzing your drawing, you are indeed finding the true minor axis (which is a 2D element), so you can test how it orients - and therefore how the normal vector of the circle being represented would be oriented in 3D space, based on the ellipse that has been drawn.

Jumping into your critique, I'm quite pleased with your work. Your ellipses are at times admittedly a little loose, so be sure to keep engaging your whole arm and using the ghosting method when executing them, but all in all you're demonstrating a lot of variety with the orientations of your cylinders and the rates of foreshortening throughout this first section.

There's really just one issue I wanted to call out, and it's something that comes up sporadically. It comes down to the two manifestations of foreshortening - the shift in scale from the end closer to the viewer to the end farther away (basically where the far end is always smaller in overall scale), and the shift in degree (where the far end gets proportionally wider than the end closer to the viewer). You are incorporating both of these shifts, which is great to see, but the point that is sometimes missing is that these two shifts should be consistent. That is to say, if the far end is considerably smaller in scale, it should also be considerably wider in degree to match.

The reason is that these are both visual cues that work in tandem to help inform the viewer of how much "unseen" space exists between those ends. If a cylinder were to be running parallel in space to the picture plane itself - so as though it were running straight across from left to right - we'd be able to see the entirety of its length, because it would all exist in a dimension that is visible to us. If however it were to rotate towards or away from the viewer, some of that length would fall into the dimension of depth, which doesn't exist on the page. Instead, this manifests as a change in the rate of foreshortening.

If the shift in scale suggests that there is a lot of "unseen" space, but the shift in degree suggests that there is minimal unseen space, then we end up with a contradiction. It's important to keep this in mind when drawing those ellipses, so we avoid the sort of thing in our drawings that will make the viewer feel like something is off. They will notice, even if they can't necessarily figure out what the issue is.

Now, throughout your work, you do have plenty of examples of cylinders where these relationships remain consistent - but there are many others, strewn through the set where you don't quite push the shift in degree far enough. It isn't an exact science - or at least, not as we deal with it here - but just try to get in the habit of pushing that degree shift a little farther, and being more aware of the relationship between the two different kinds of shifts.

Moving onto your cylinders in boxes, you're making great progress here. This exercise revolves around developing a student's capacity to construct boxes that feature two opposite faces which are proportionally square. We achieve that using the line extensions - just as in the box challenge where the line extensions help us shift our approach and rewire our brain to better construct boxes whose edges fall into one of three possible sets of parallel lines, this exercise's additional line extensions allow us to test and improve our proportions. By looking at the extensions of the minor axis line and the two contact point lines for each ellipse, and testing how far off they are from converging with the box's own vanishing points, we can see how far off they are from represent circles. In turn, that also tests how far off the planes that enclose them are from being squares (since the plane of a box that encloses a circle in 3D space would itself be a square in 3D space).

I definitely see progress on this front in your work, but there is one issue that may have gotten a little bit in the way here. I can see that you are indeed testing your "true minor axes" (in green pen), but because they're only drawn with a fairly limited length, it becomes much harder for you to test them against the box's vanishing points. I highly recommend that when you do this exercise in the future, you extend them further, so you can more readily catch cases where they don't align correctly, and make adjustments accordingly for your next set.

Anyway, I'll go ahead and mark this challenge as complete. Keep up the good work.