I've been doing a bit of reading about Russian copyright law. (Ok so it's only on
Wikipedia..)
Here's what it says about unpublished works:
"Among the true novelties introduced by the new legislation in the area of copyrights were a publication right (a copyright granted to the publisher of a previously unpublished, uncopyrighted work with a period of 25 years from the publication), and the definition of two kinds of contracts: one for copyright transfers, and licenses for granting usage rights. Newly, gratis licences were explicitly allowed (article 1235). A subtle change concerned the calculation of the copyright term for posthumously published works,
which began newly from the disclosure instead of from the publication. (See above for the difference.)"
...and crucially

:
"For a work that was disclosed during the author's lifetime, the
copyright term of 70 years thus runs from the year the author the author died (or was rehabilitated, if the rehabilitation was posthumous), even if the work is published only later."
As for what they mean by "disclosure"...
" "Disclosure" is a concept newly introduced in the copyright law of 1993 to put an end to the ambiguities surrounding the term "publication" in the old Soviet copyright law. In Soviet copyright, publication included ephemerally making available a work, such as through a performance, a speech, or a broadcast. However, for foreign works protected under Soviet law indirectly through international agreements (in particular the UCC), the definition of "publication" laid down by these agreements (typically the "making available of copies", which excluded ephemeral reproduction and required the physical fixation of a work) was used.
The new law tried to resolve this confusion by using "disclosure" for the broader sense (making accessible of a work to the general public through publication, performance, broadcast, or any other means), and using "publication" generally only in the sense of distribution of copies of a work to the general public."
In broad terms this is all in line with international copyright law. Thus (if I'm interpreting this correctly), copyright on La Bayadere (1900 version), because it had been "disclosed" (ie. performed) during the composer's lifetime, extended to 70 years after his death (in 1917) -- thus to 1987.
I can only re-iterate that IMHO whatever "work" was required to make this version performable (if it's as "authentic" as they claim it to be), was not enough to give them full and legal copyright on the work, and thus the work is for all intents and purposes in the public domain.
But they do own the original score (in the physical ink-on-paper sense). So apart from physically keeping the scores and their orchestral parts under lock and key, I don't think that, if anyone decided to do a "reverse engineering" job (ie. use what is out on YouTube etc.) and re-assemble (as-close-as-is-possible) a full score (including lost act) that way, the Mariinsky would not have any recourse to copyright protection.
Morally, however, it shouldn't have to come to that. The "moral rights" (as defined in the same article), remains with the composer. And thus the Mariinsky (as the owners of the score) have a
moral responsibility to further Minkus's reputation by making the score as widely available as possible (ie. by GETTING IT PUBLISHED).
(If you've gotten this far,

for sticking with me!)