Are technical/scientific books from pre-1925 particularly useful for self-learning today? I'd imagine for most disciplines, the knowledge has progressed and possibly changed course since then and it may be more outdated than not.

It might depend on the topic. Classical mechanics? I'm not sure that there is any fundamentally new knowledge since 1925 in that field. What's different is that people have 100 more years of figuring out how to explain it well.