This is a brilliant post, thanks a lot!
MRC Important correction: I didn't memorise that piece (the Beethoven A major Sonata for cello and piano) over the train journey, I learnt it. That is, with the help of a silent keyboard (I got some funny looks from fellow passengers!), I put in the fingerings and worked on the most difficult passages so that at the end of the journey I was ready to run through the piece in the first rehearsal with the cellist.
I don't have anything resembling a photographic memory. What I do have is a lot of patience, good powers of concentration and good sight-reading skills (acquired in my teens). It is my proficiency in sight-reading, together with the ability to stay focused under stress, that has allowed me to jump in at the last minute on many occasions.
That's very interesting. I find it really surprising how different different people's methods are. My piano teacher doesn't have a photographic memory either, but he trained himself to be able to see the score like a map while playing pieces, to know exactly which line of which page you're on. I've tried this, and it might be possible to train, actually.
For me there is no doubt: the experience of athletes, musicians and any other people who need to produce an accurate "performance" on command, together with the results of numerous studies, shows that cramming is not as efficient as practice spaced over time.
What do you mean by "efficient" though? Do you think you would have spent less total time on the Beethoven A minor if you learned it over a period of a few weeks? Do you think it would have been learned better? How do you make that call?
What is your approach when you have more time? Do you work on say 100 measures a day in parallel for a week? Do you sight read multiple times? Or do you perfect measures like johnstaf outlined, 12, 23, 34, combine?
Remember that, even when you only have a few hours, taking frequent breaks will get you further than practising solidly through those few hours. This is compellingly illustrated by the first study that Molly Gebrian quotes in the first video that twocats posted in the thread about neuroscience-based learning. The group of surgeons that practiced a certain surgical process for three solid hours did not learn it as well as the group that practiced for only 40 minutes before taking a 20 minute break. Even though the second group only practiced for a total of two hours, they performed much better than the group that had practiced for three hours.
Try to keep calm and concentrated. If you sense that your concentration is flagging, better take a short break, even if you've only been practicing for 20 minutes.
I guess the question then remains: How do you determine when to take a break? If you get into a "flow" state for 2-3 hours and don't feel a lapse in concentration, do you keep going, or do you throw in a break there? Do you sometimes push your ability to concentrate by keeping at it? Or do you always use your internal sense of when you feel like you're losing concentration?