In the early 1880s Hermann Ebbinghaus set out to measure something nobody had measured: how quickly we forget. He had no volunteers, so he used himself.
The experiment
Ebbinghaus memorised lists of meaningless syllables - combinations like "DAX" or "BOK" - so that nothing he already knew could help. After intervals ranging from minutes to a month, he relearned the lists and timed how long it took.
If relearning was quicker than learning the first time, something had been retained. He called the difference "savings", and plotted it over time. He published the results in 1885.
What the curve shows
The famous result is a shape, not a single figure. Memory drops steeply in the first hours and days, then the decline slows and flattens. Much of what will be forgotten is forgotten early; what survives the first stretch tends to last.
In 2015 researchers Jaap Murre and Joeri Dros repeated the experiment closely following Ebbinghaus's method, and found a curve with a very similar shape. More than a century on, the basic finding held up.
The numbers to be sceptical about
Articles often claim you forget 50% of new information in an hour, 70% in a day, 90% in a week. Treat these with caution.
The original data came from one person learning nonsense syllables, measured by relearning time. Real material behaves very differently. A story you understand, a word you can connect to others, or a fact that surprises you fades far more slowly than "BOK". How well something was learned in the first place matters enormously too.
The honest summary is qualitative: forgetting is fastest at the start, and without review, a lot goes.
What flattens the curve
Reviewing, at the right times. Every review sets the memory back towards full strength, and the next decline is slower than the last. This is the basis of spaced repetition: a first review soon, later ones at widening gaps.
Recalling instead of rereading. Actively pulling something out of memory strengthens it more than looking at it again. A quick self-test beats a re-read.
Meaning. Material that connects to what you already know sticks better. Learning where a word comes from, or using it in your own sentence, gives it more hooks.
Sleep. Memories formed during the day are consolidated during sleep, which is one reason a review in the evening and another a few days later works well.
What this means day to day
If the steepest drop happens early, the single most valuable review is the first one, and it should not wait a week. Checking something you learned in the morning again the same evening catches it before much of it has faded.
After that, each successful review earns a longer gap: a few days, a week, two weeks, a month. You end up spending almost no time per item, because you are only reviewing each one when it genuinely needs it.
A kinder reading of the chart
The forgetting curve is usually presented as bad news. It is more useful as instructions. Forgetting is normal and predictable, which means it can be planned around. The fact that you forgot something is not a sign that you are bad at learning. It is a sign that it was due for review.