That spacing beats cramming has been known since the 1880s. The more useful and much more recent question is how far apart — and the answer turns out to depend on something most study advice never mentions: when you need to remember it.
At a glance
| You need to remember it for… | Useful gap between reviews |
|---|---|
| A week | Roughly a day |
| A month | Roughly 3–7 days |
| A term (~10–13 weeks) | Roughly 1–2 weeks, up to a month |
| A year | Roughly a month or more |
| Rule of thumb | About 10–20% of the retention interval |
The oldest finding in the field
Hermann Ebbinghaus, working on himself in the 1880s, established both the forgetting curve and the fact that distributing practice over time produced better retention than massing it. Nearly a century and a half later, the finding has survived every attempt to overturn it.
The modern synthesis most people cite is Cepeda, Pashler, Vul, Wixted and Rohrer's 2006 review in Psychological Bulletin, which pulled together a large body of distributed-practice studies. The direction is not in doubt: for the same total study time, spread beats bunched.
What the 2006 review also made clear is that "space it out" is an incomplete instruction. The size of the benefit depends heavily on the gap, and the best gap is not a constant.
The finding that makes it actionable
The paper that turns this into engineering is Cepeda and colleagues' 2008 study in Psychological Science, memorably titled Spacing effects in learning: A temporal ridgeline of optimal retention.
They varied two things independently: the gap between study sessions, and how long afterwards the final test came. Plotting retention against both produces a ridge — for each retention interval there is a gap that maximises performance, and that optimum moves outward as the retention interval grows.
The practical summary that came out of this line of work is a proportionality: the ideal gap is roughly ten to twenty per cent of the period you need to retain the material. Need it in a week? Review about a day later. Need it in a year? Gaps of a month or more.
This explains something that otherwise looks contradictory in study advice. "Review the next day" and "review a month later" are both correct — for different deadlines.
And the finding about flashcards specifically
Most spacing research uses word lists and passages. Kornell's 2009 paper in Applied Cognitive Psychology — Optimising learning using flashcards: Spacing is more effective than cramming — went at the actual object.
It compared studying a large stack of cards spread over time against breaking the same cards into small stacks studied intensively. The large, spaced stack won. Which is precisely the opposite of what most students do: faced with a big pile, they instinctively break it into a manageable chunk and drill that chunk hard.
There is a practical lesson buried in this for parents. If your child sits down with forty cards and drills ten of them four times each, they have done the version that does not work. One pass through all forty, then again tomorrow, is better — and it is the pattern a scheduler enforces automatically.
How this produces our numbers
Our ladder is 1 → 6 → 15 → 38 → 90 days. Read against the research:
The early gaps are short because retention is fragile. A card reviewed once could be gone in days, so a one-day gap is proportionate.
Gaps widen because each success extends the horizon. Once a card has survived a fifteen-day gap, it is a different kind of memory, and a one-day gap would be wasted effort.
They stop at 90 days because a term is the horizon that matters. This is the direct application of the proportionality rule. For a KS3 child assessed termly, the retention interval is roughly ten to thirteen weeks, which puts the useful maximum gap in the range of weeks to a few months. Ninety days sits at the top of that band rather than beyond it.
That is why we do not follow Anki's default of letting intervals reach one hundred years. It is not that Anki is wrong — for an adult who needs material for a career, the retention interval genuinely is decades, and enormous gaps are the correct application of the same rule. Different deadline, different answer.
What the research does not settle
Some honesty about the edges.
The exact percentage is a rule of thumb. The 10–20% figure is a useful summary of a noisy surface, not a constant of nature. Anyone quoting it to two decimal places is overselling.
Expanding versus fixed intervals is not fully resolved. Whether gaps should grow, or simply be long, has produced mixed results. Expanding schedules have practical advantages — they concentrate effort where memory is weakest — but the evidence for expansion per se is softer than the evidence for spacing itself.
Most studies use simple material. Word pairs and facts, over weeks. Generalising to a Year 8 maths curriculum over a school year involves an inferential step, and honest practitioners should say so.
What is not in doubt is the direction: spread the same effort out, let the gaps grow as the material holds, and cap them at the horizon you actually care about.
FAQ
What is the spacing effect?
Learning is more durable when the same amount of practice is spread over time rather than massed together. It is one of the oldest and most reliable findings in memory research.
How long should the gap between reviews be?
It depends on how long you need to remember. The research suggests the ideal gap scales with the retention interval — roughly ten to twenty per cent of it. For a test in ten weeks, gaps of one to two weeks.
Why does aitutors.me stop at 90 days?
Because a KS3 child is assessed each term, so the term is the retention interval that matters. A gap longer than a term would let material disappear before the assessment it was meant to support.