Skip to content

Remembering One Long Number: Grouping, Rhythm and Rehearsal

Holding a number for thirty seconds is a different problem from memorising it. What the rehearsal loop can carry, and what reliably knocks a number out of it.

By Published Updated 5 min read

Reign Creative LLC makes IQ Test & Brain Training Games, which this article links to, so we have a commercial interest in it. How we write and check articles.

Key takeaways

  • Holding a number for half a minute is a rehearsal problem, not a memorisation problem. The techniques that build durable memories are mostly wasted here, and the techniques that keep a loop running are not.
  • The classical account describes a verbal store whose contents decay in roughly two seconds unless an active articulatory rehearsal process refreshes them, which is why you catch yourself muttering.
  • The word-length effect means that lists of longer-sounding items are recalled worse. In English, seven is the only two-syllable digit name, and it genuinely costs more loop time than its neighbours.
  • Groups of two and three digits work better than groups of four or five, because each group has to be sayable as a single unit for the rehearsal loop to carry it.
  • What breaks the hold is usually competition rather than time. Berman, Jonides and Lewis found that filled retention intervals hurt performance and concluded that the mere passage of time does not cause forgetting in short-term memory.

Holding a number in your head from the screen to the keypad is not the same problem as memorising it, and it does not want the same solution. A thirty-second hold is a rehearsal problem: the number has to stay alive in a loop that refreshes itself, and almost every failure comes from that loop being interrupted rather than from the number being too long. The fix is a grouping you can say in one breath, a steady rhythm, and thirty seconds of refusing to do anything else.

The loop you are actually using

The classical account of verbal short-term memory describes a store that holds items in phonological form and loses them to decay after about two seconds, kept alive by an active articulatory control process that supports rehearsal. Robert Hughes, reviewing and then criticising that model in the Quarterly Journal of Experimental Psychology, summarises it in exactly those terms before arguing that a perceptual-motor account fits the data better and that no discrete store is needed.

The disagreement about architecture does not change the practical picture, which is what matters here: a short verbal hold is maintained by something very like saying the material to yourself, and it lasts only as long as the saying continues. That is why you catch yourself muttering a postcode, and why stopping the muttering ends the memory almost immediately.

It also explains a detail that sounds trivial and is not. The word-length effect — Hughes describes it as "the poorer recall of a list of long compared with short words" — means the time your items take to pronounce is part of the constraint. In English, seven is the only two-syllable digit name, and it really does cost more loop time than its neighbours. You are not holding abstract quantities. You are holding sounds.

Why the chunk size is two or three

Nelson Cowan's review argues that the underlying store holds roughly 3 to 5 meaningful items in young adults, and that the unit is the chunk rather than the digit. George Miller had made the operative point half a century earlier: "since the memory span is a fixed number of chunks, we can increase the number of bits of information that it contains simply by building larger and larger chunks."

The natural question is why we do not simply build enormous chunks and be done with it. The answer is the loop. A chunk only helps if it is a single unit in rehearsal, which in practice means something you can say as one word. "Eighty-three" is a word. "Five-nought-seven-three-nine" is a queue.

So take 5 0 7 3 9 1 6 2 8 4 and hold it as 507 · 391 · 6284 — three units instead of ten items, each of them sayable. Telephone formatting conventions around the world are, in effect, a crowdsourced experiment in this constraint. When you see a number formatted for you, accept the formatting; the work has already been done.

One warning: keep the grouping fixed. Starting in pairs and drifting into triples halfway through creates a boundary error that is very hard to detect from the inside, because the digits all feel present.

Rhythm is a grouping cue

Say the three groups above with equal spacing and a pause between them, and the pauses do real work. Rhythm marks the boundaries, so a digit that drifts is more likely to drift within a group, where it is recoverable, than across one, where it is not.

This is also the reason a number read aloud to you at an unhelpful cadence is harder to hold than the same number read in groups: someone else's rhythm has overwritten the one you would have chosen. If a person reads you a long number in an undifferentiated stream, the highest-value thing you can do is repeat it back in your own grouping before you try to use it.

There is a second, subtler benefit. A rhythm is a structure you can check against. If the final group comes out three digits long when you set it up as four, you know something is missing without having to remember what.

What actually knocks it out

The intuitive model is that numbers fade with time. The evidence points more strongly at competition.

Berman, Jonides and Lewis, in a study explicitly hunting for decay in verbal short-term memory, report that "both conditions with a filled interval led to worse memory performance, but the difficulty of the intervening task had no effect", and conclude that the mere passage of time does not cause forgetting in short-term memory, with interference doing the work instead. They also make the observation that most self-help advice misses: rehearsal "happens so habitually that it is often not noticed and is difficult to disengage."

Put those together and the practical rule falls out. It is not the thirty seconds that costs you the number. It is the question someone asks at second twelve, the notification you read, the price you glance at. Anything verbal competes; the difficulty of the interruption barely matters. Guarding a short hold means guarding the channel, not racing the clock. Our article on delayed recall goes further into filled and unfilled delays, and the working memory guide covers the capacity that all of this sits on top of.

Practising the hold without pretending it is more

IQ Test & Brain Training Games is our own app, built by Reign Creative, and several of its memory modes drill exactly this: Number Recall and Sequence Memory show a string, take it away, and ask for it back. The seven difficulty levels change the real mechanics rather than just the label — Beginner presents three or four items with a slow reveal, and the higher levels shorten the reveal window and lengthen the sequence. A Custom Practice mode lets you set the content, the length and the reveal speed yourself, which is the setting to use if you want to isolate one variable.

Two honest limits. First, getting better at these drills means getting better at these drills; the evidence on span training does not support a claim that it enlarges memory in general, and we do not make one. Second, the drills are practice and entertainment, not assessment: they record score, accuracy, streak, response time and personal bests. The app does also include a memory test and an IQ-style test, and its Google Play listing calls their results estimates based on app activities, not clinical or professional assessments.

Within those limits the drills are genuinely useful for one thing this article cares about — finding out, for you specifically, where the group size stops helping. Run the same sequence length with pairs and then with triples and watch which one survives. If you want the methods for numbers you need to keep rather than merely hold, the comparison of number memorisation methods covers those. Otherwise, browse the memory and attention hub, the wider education and brain section, or the description of the app on the IQ Test & Brain Training Games app page.

Good to know

Frequently asked questions

How do you hold a long number for thirty seconds?
Split it into groups of two or three, say the groups once at a steady rhythm, and then protect the loop: do not speak, do not read anything, and do not let someone ask you a question. The rehearsal loop is cheap to run and easy to interrupt, and nearly every failure at this timescale is an interruption rather than a capacity problem.
Why do I forget a number the moment someone talks to me?
Because speech competes directly for the machinery you were using to rehearse. Berman, Jonides and Lewis note that rehearsal happens so habitually it is often unnoticed and difficult to disengage, and report that filled retention intervals produced worse memory than unfilled ones. An interruption does not merely distract you; it occupies the same channel.
What size should digit groups be?
Two or three for most people. The constraint is that a group has to be sayable as one unit in the rehearsal loop, and a five-digit group is a sentence rather than a word. Phone number formatting encodes this: the three-three-four pattern is not arbitrary, it is a shape the loop can carry.
Does saying the number out loud actually help?
Usually yes, for a short hold. Speaking refreshes the representation and imposes a rhythm, and the rhythm itself becomes a grouping cue. The trade-off is that you cannot say anything else at the same time, so it is a good strategy while you walk to the keypad and a bad one in the middle of a conversation.

Sources

  1. The phonological store of working memory: A critique and an alternative, perceptual-motor, approach to verbal short-term memory (opens in a new tab)Quarterly Journal of Experimental Psychology, via PubMed Central (PMC11783984) — accessed
  2. The Magical Mystery Four: How is Working Memory Capacity Limited, and Why? (opens in a new tab)Current Directions in Psychological Science, via PubMed Central (PMC2864034) — accessed
  3. In Search of Decay in Verbal Short-Term Memory (opens in a new tab)Journal of Experimental Psychology: Learning, Memory, and Cognition, via PubMed Central (PMC3980403) — accessed
  4. The Magical Number Seven, Plus or Minus Two: Some Limits on our Capacity for Processing Information (opens in a new tab)Psychological Review (1956), via Classics in the History of Psychology, York University — accessed

About the author

— The independent app studio that builds the apps covered on this site and publishes every article here. On Google Play it publishes as Reign Collective Apps.

We use AI tools to help research, draft and edit our articles. Each article lists the sources its factual claims rely on, and details about our apps come from their live Google Play listings. No article on this site has been reviewed by an independent expert unless the reviewer is named on it.

Spotted an error? Tell us — how corrections are handled is set out in our editorial policy.

Keep reading