6 ideas
Memory is measured not by what can be recalled but by how much less time or how many fewer repetitions are needed to relearn material to the same criterion of mastery. The difference between original learning effort and relearning effort, expressed as a percentage, reveals retention that is present but no longer consciously accessible, so apparent total forgetting still leaves a measurable trace.
Retention falls most sharply in the first hour after learning, with over half the savings lost within about a day, after which the rate of loss slows dramatically and what remains decays only slowly over weeks. Forgetting therefore follows a roughly logarithmic curve rather than a straight-line decline, so the timing of review matters most shortly after learning.
The same number of repetitions produces more durable retention when spread across several days than when crammed into a single session. Spacing lets each review do more work because it comes after some forgetting has occurred, while massed repetition yields diminishing returns as each additional pass adds less.
To measure memory itself rather than prior knowledge, Ebbinghaus built about 2,300 consonant-vowel-consonant syllables with no meaning and learned them in lists under fixed conditions: same time of day, a metronome-paced reading rate, and a set criterion of two errorless recitations. Removing meaning and standardising procedure turns a vague mental faculty into a variable that can be counted and compared across trials.
Stanzas of Byron's Don Juan required roughly a tenth of the repetitions needed for nonsense lists of comparable length. Meaning, rhythm, and existing associations massively compress the work of memorisation, which is why nonsense material exposes the raw cost of forming new connections.
Ebbinghaus served as the only experimenter and the only subject, memorising and relearning thousands of lists over years and recording times with his own discipline as the main control. He showed that a higher mental process could be studied with the quantitative rigour of physics, while openly acknowledging that the results rest on one mind and need replication.