Why 25 minutes? Because a kitchen timer said so
It is 8:40 on a Wednesday. You have ninety uninterrupted minutes before the first meeting, twelve pages of reading, and a timer on the desk. You set it to 25, because 25 is what everyone sets it to.
At 9:05 the bell goes. You are four pages in and have just started to hold the argument in your head.
The number arrived with the authority of a finding. It is not one.
The interval is a price, not a dose
Most people read 25 minutes as a dose — the amount of attention a human can supply before the tank runs dry. Read that way, changing it feels like cheating, and every session that ends early feels like a failed measurement.
The better reading is that the interval is a price — what starting costs you today.
Twenty-five minutes is a price most people will pay on most days. Ten minutes is a price you will pay while flinching. Ninety is a price you can only afford when you are already warm and nothing is going to interrupt you.
Nothing in your biology sets that price. Your Wednesday sets it.
Where 25 actually came from
Francesco Cirillo built the technique in the late 1980s as a university student in Italy, using a wind-up kitchen timer shaped like a tomato. The story is in the technique’s own history and it is short, because there is not much to it.
The first stretch he timed was nowhere near twenty-five minutes. It was a small question he put to himself at a desk: could he study one short block without breaking off. Exactly how short is where the retellings split — some say two minutes, some say ten — and none of them traces back to Cirillo himself, so the honest answer is that nobody knows.
Twenty-five arrived later, out of trial and error in one room. No lab, no control group, no journal. Cirillo has been plainer about this than most of the people who quote him — he describes the 25-minute session as one expression of the method rather than the method itself.
The winding matters more than the number, and it is the part that got dropped when the timer moved into a phone. Turning a dial takes about a second and a half of your body agreeing to something. A tap does not.
Is there any research that points to 25?
No. There is a real body of work on how attention behaves over time, and it is worth knowing, but it does not converge on 25 or on any other single figure. Here is what it actually says.
What the vigilance research shows
In 1948, Norman Mackworth published The Breakdown of Vigilance During Prolonged Visual Search in the Quarterly Journal of Experimental Psychology. Participants watched a clock face whose pointer jumped once a second, and pressed a button on the rare occasions it made a double jump — a task built to mimic wartime radar watch-keeping.
Detection dropped by roughly 10 to 15 per cent inside the first half hour of a two-hour watch, then kept sliding more gently for the remaining ninety minutes. That decline has been replicated for seventy-odd years and has a name: the vigilance decrement.
Two limits worth stating out loud. The task was staring at a dial for a rare event, about as monotonous as work gets and nothing like drafting a proposal — the effect is clearest exactly where the work is dullest. And it describes a gradual slide on one repetitive task, not a shutter that comes down at minute 30.
What you can take from it: on repetitive work, your performance drops before you notice it has dropped, so pauses are maintenance rather than reward. What you cannot take from it is a number.
The eight seconds thing is not real
You have read that human attention is down to eight seconds, one second worse than a goldfish. It shows up in conference decks, on packaging, in worried articles about children. It has no source, and the trail is short enough to walk yourself.
The figure travels from a 2015 Microsoft Canada consumer report, where it appears with a footnote crediting a statistics website called Statistic Brain. Statistic Brain listed its own sources as the National Center for Biotechnology Information at the US National Library of Medicine, and the Associated Press.
Both were asked. Jonathan Schwabish wrote up his attempt at PolicyViz in 2016; the BBC’s Simon Maybin ran the same errand in 2017. Neither organisation could produce research matching the figure. One study loosely in the neighbourhood — Weinreich and colleagues, 2008 — is a 25-person study of web browsing behaviour that does not measure attention span at all.
The goldfish half has never had a citation of any kind.
Gemma Briggs, a psychologist at the Open University, gave the BBC the more useful point underneath: an average attention span is close to a meaningless idea, because attention is task-dependent. You already know this from the inside. You cannot hold a spreadsheet for four minutes and you can hold a film for two hours, on the same afternoon, with the same brain.
So what is actually known
Three findings that survive checking, each with its edges marked.
Unfinished tasks follow you. Sophie Leroy, in Organizational Behavior and Human Decision Processes (2009), named this attention residue: switch away from a task before it is done and part of your attention stays behind, measurably degrading the next thing you do. The evidence is short lab tasks with student and office samples, so treat the direction as sound and the size as unsettled.
We switch far more than we used to. Gloria Mark’s group has logged real computer use since 2004. Average time on one screen before switching was around two and a half minutes then; in recent years it sits near 47 seconds. That is a measurement of behaviour, not of capacity — it says what we do, not what we could do.
Boring work degrades before it feels like it is degrading. The vigilance decrement, above.
Not one of the three yields a session length. They tell you that boundaries and pauses matter and leave the arithmetic to you.
Picking your number
Start from how long it takes you to load the work into your head, because that is the part nobody accounts for. If the first fifteen minutes of every writing session are spent re-reading what you wrote yesterday, a 25-minute bell rings roughly when you become useful.
| Your situation | Try | Why |
|---|---|---|
| You are avoiding the task | 10–15 min | The price has to be low enough to pay while flinching |
| Admin, email, revision, mixed days | 25 min | Long enough to move something, short enough to agree to |
| Code, writing, design, dense reading | 50 min | Ten to fifteen minutes go on loading before anything happens |
| A protected morning, already warm | 90 min | Only when nothing will interrupt, or you are just measuring the interruption |
One rule ties the table together: if starting is hard, shorten the interval. People who are struggling reach for a longer session as a show of seriousness, which raises the price of starting at the exact moment they cannot afford it.
If you keep landing near ninety and want to know whether that number has any biology behind it, the ultradian rhythm argument is where that lives — with caveats of its own, and they are not small.
When the number is not the problem
If you have changed the interval four times in a fortnight and nothing improved, the interval is not what is broken.
The usual culprits are elsewhere: a task named too vaguely to start, a phone inside arm’s reach, or a calendar with no gap longer than the session you keep setting. Those are the mistakes that actually end streaks, and none of them care what the dial says.
And if your attention runs on a different clock — ADHD, or something that has never been given a name — external timers help a lot of people and 25 suits very few of them out of the box. Some do best at 10, some at 45. That is the method working, not failing.
Fomodo will run any interval you set, in Pomodoro or in three other modes, and it keeps the count without being asked. What it cannot do is tell a good hour from a wasted one. The daily goal is measured in minutes, and minutes are an honest but blunt unit — the app will congratulate you for sixty slow ones exactly as warmly as for sixty sharp ones. Only you know which kind you just had.
There is also a ticking sound you can switch on for one session and off for the next. It does nothing measurable. Some rooms are too quiet and you want to hear the thing running.
The right length is not the one with the best story behind it. It is the one that makes starting cheap enough to do today.
