I think you are right that in cases where there is an equal number of notes in each column, the offset can be interpreted both ways, as long as the overall music spacing is disregarded. As I understand it, what will determine the axis in these cases is whether or not there is any compensation for the horizontal space lost by the vertical shift of the fastest moving voice. If not, then this must be interpreted as a shift of the column containing the fastest moving voice and vice versa.John Ruggero wrote: ↑09 May 2017, 14:06 And I guess I am now lost, because in the Bach example I posted originally from the BGA, it seems clear to me that the "main axis" is intended to be the soprano, tenor and bass and the alto is offset to the right. (I am assuming that the "main axis" is where the greater number of layers fall together and the "offset" would be where the fewer fall. If there are equal numbers of each, then the left column might be the "main axis". I hope that is right.)
Following the principle I've advocated above, I prefer the Henle, because it keeps all similar note values aligned in both measures. Truth be told, I think aligning the quarter notes in the second measure is not exactly a must, because of the tie-in in the bass. Then again, it does give a more even spacing, which is a plus.John Ruggero wrote: ↑09 May 2017, 14:06 Here are three other editions' takes on this measure. To me, the first two offset the alto. The Henle, which is something like OCTO.'s, offsets both alto and bass. (Incidentally, the various MS don't offset at all!)
It's kind of hard to tell, but it looks like the Bischoff lacks sufficient space compensation, which makes the quarter and eight notes take up almost the same amount of space. This is wrong in my estimation, even if it retains a more even spacing in the bass.
The uneven whole notes in OCTO's version makes that one unacceptable to me.