probably it is merely a difference in the score, whether it was the original version, or if Granados changed it later on.
Yes, probably. But since the original UME edition is from "ca. 1915", he would not have had terribly much opportunity to change it before he died in 1916.
As I wrote, that natural sign is in front of the following A, where it makes even less sense than had it been on the G.
Ah, so you did. Sorry, I sloppily misread that, and thought you had said it was on the G. Still, these hamfisted Spanish typesetters are as likely to stick symbols in the wrong place as to use the wrong ones, so I wouldn't dismiss altogether the possibility of them doing both at the same time when preparing your 1966 edition.
In any case I keep believing the G nat sounds much more idiomatic. It anticipates on the temporary shift to minor.
Perhaps, but for X to "anticipate" Y, one would have thought X should involve actually changing something, which sticking in a pointless natural sign, wherever it goes, does not do. Perhaps
it should have been a flat sign instead, and on the B. OK, while this isn't really a serious suggestion, you've got to admit that it would sound nicely idiomatic too, and
anticipate the minor, specifically the B-flats which do actually occur two bars later.
I am pretty convinced that this was what Granados meant (and that Riva and Larrocha thought so when preparing the scholarly edition).
That the two of them discussed this difference and that Riva persuaded Larrocha to change her mind is an interesting thought. If Monica is planning to purchase more volumes of this new edition, perhaps the one containing this piece should move up to the top of her shopping list, if it isn't there already. Do you think they would have been able to consult the original manuscripts, or are they likely to be lost/destroyed?