This looks like a retouch job, I can't think of a way these sharp changes in darkness without visible change in stroke width could have been produced in one go. If it wasn't for 'q', I could argue for a curvelet-based distance cipher ("Johannes Hasenbard" cipher, You are not allowed to view links. Register or Login to view.) where each stroke is added separately (e.g., 'e' is put on the vellum first and then converted into 'y' or 's' later, likewise the left leg of k/t/p/f is added first and then converted to the full character later). However, once you have drawn a vertical stroke that goes well below the baseline, you can only convert it to 'q' in Voynichese, as far as I can see, so it's unclear why the top of 'q' could be darker than the descender in this case.
Another possibility, related to out of order ciphers, is retouching to make the order of writing less obvious. E.g., first the scribe would put symbols in the order of encoding and then rewrite them with fresh ink to make them appear written sequentially.
There are similar patterns on f26r, which is available in the MSI set, I wonder if it's possible to detect two separate outlines there under the dark ink.
Another possibility, related to out of order ciphers, is retouching to make the order of writing less obvious. E.g., first the scribe would put symbols in the order of encoding and then rewrite them with fresh ink to make them appear written sequentially.
There are similar patterns on f26r, which is available in the MSI set, I wonder if it's possible to detect two separate outlines there under the dark ink.