Why dividing the hours evenly is the one plan that fails
The usual plan is arithmetic: forty hours available, twenty days left, two hours a day. It divides the work evenly but does nothing about memory. Each topic gets one long sitting and is then untouched for a fortnight, which is roughly the interval over which most of what you understood quietly disappears.
Forgetting is fastest immediately after learning, not spread evenly across the days. The steep part of the curve is the first twenty-four hours. A topic you had completely on Monday is noticeably thinner by Tuesday and thinner again by Friday, and a ten-minute review placed inside that window holds far more than a second long sitting three weeks later.
So the schedule is built from intervals rather than slices. The same forty hours, arranged as day 0, 1, 3, 7, 14 and 30, leaves more material intact on the exam morning than forty hours in equal daily blocks. You are not working more. Each review simply lands while the memory is still fading rather than after it has gone.
How the intervals are chosen and squeezed
The gaps come from the SM-2 family of spacing intervals, the same scheduling logic behind flashcard software: a first pass on the day you learn the topic, then the next day, then after three days, a week, a fortnight and a month. Each interval is roughly double the one before it.
Those intervals assume open-ended time, and an exam does not offer that. Any review that would land after the date in “Exam date” is pulled back inside the window, so twelve days out you get day 0, 1, 3, 7 and 11 rather than three reviews and an awkward gap. The pattern compresses; it is not cut short.
One slot is fixed. Whatever the topic count and however short the run-up, every topic is given a short pass on the day before the exam, because that is the cheapest place to top up recall. It is a pass, not a first reading — twenty minutes of retrieval, not a fresh attempt at the chapter.
Making it survive contact with a real week
Count the sittings before you commit. Twelve topics with five reviews each is sixty separate sittings; across three weeks that is about three a day, or an hour daily at twenty minutes each, on top of the first pass through each topic. If that will not fit around lectures and practicals, cut topics rather than cutting reviews.
When you miss a day, do not restart the sequence. Take the missed review at the next opportunity and carry on from where the schedule stands. A late review is worth a great deal more than a skipped one, and the intervals tolerate a day's slip without the whole plan needing to be rebuilt.
A review also means something specific. Close the notes and try to reconstruct the derivation, the definition or the worked steps, then check what you got wrong. Re-reading feels fluent and returns almost nothing, and that false fluency is exactly why an even-hours timetable feels productive right up to the paper.