Especially if the master did not want to make the recording, but was probably cajoled into it by a well-meaning manager or other professional who could not foresee the pernicious effect of the recording on generations of unsuspecting musicians.
I ran into the problem recently while coaching a fine group of amateurs. At issue was the reading of a rare minuscule sixty-fourth-note subdivision: whether or not to make it coincide with a triplet going on in the other voices. The players were going to a lot of trouble to distinguish the two types of subdivision; I pointed out that scholars had recently been studying just such cases and concluded that, there being no way to indicate subdivisions of triplets with rests during Schubert's time, the sixty-fourth-note remedy was as close as they could get. My suggestion that they try it was greeted with scorn by someone who said that "all" the recordings do it as a conflicting pulse.
Okay. But are recordings necessarily definitive?
Maybe the way it feels to play it one way or the other might be more convincing and more meaningful.