If more hours reliably produced more output, the discovery would have been made where output mattered most. It was. The result went the other way. During the First World War, Britain needed shells more desperately than any organisation has ever needed anything, and its munitions factories ran punishing weeks on the same logic every overworked doctoral researcher runs today: the front is waiting, so work more. The state, to its credit, measured what happened. A century later, the economist John Pencavel organised those records, observations on munitions workers, most of them women, into one of the most instructive findings in the economics of work [1]. The relationship between hours and output is non-linear. Below a threshold, output rises with hours, as intuition says. Above it, each additional hour buys less than the last. At the longest weeks in the data, additional hours were buying almost nothing at all. The factories that needed output most learned that hours and output are different quantities. They learned it under existential pressure, from their own records. Research, a century on, still mostly runs the pre-war model.
Table Of Contents:
- Sofia logs sixty hours and calls it a good week
- What the hours model actually costs
- The energy model: three currencies, not one
- Focus time: put your best work where your best hours are
- The focus budget: two and a half to four hours, and that is the day
- Sofia’s week, re-denominated
- A word for the PI whose lab glorifies hours
- FAQ
- References
And before anyone files this under wartime history: the finding keeps replicating, on modern workers, in this decade. Iceland ran two large public-sector trials between 2015 and 2019. Around 2,500 workers moved from 40 hours to 35 or 36, with no pay cut. Productivity and service levels held or improved in most workplaces, and the result went national [2]. The UK then ran the biggest four-day-week pilot to date, in 2022: 61 companies, around 2,900 people. Revenue stayed broadly stable. Burnout fell. A year later, nine in ten of those companies were still running the shorter week [3]. Now, you could dismiss all this because it is not a controlled lab study. I would argue the opposite, and this is my personal view: real-world evidence of this kind is stronger than an artificial lab environment. A company answers to customers and shareholders. It lives under real pressure to deliver value and return on investment. No lab study carries that pressure, and, let us be frank, neither does most of academia. So when a business tries a shorter week, finds its people more productive and less burnt out, and then keeps the shorter week, that is not sentiment. That is a hard-nosed economic verdict, renewed every quarter, and some of those companies went on to shorten further. The shell factories reached the same verdict under existential pressure. The companies reach it under market pressure. Only academia is still waiting for more data.
So “how many hours should a PhD student work?” turns out to be the wrong question. I know, it is the question of this article, because that is what was on your mind when you landed here. But this article will not answer it with a number, and the munitions data makes the refusal respectable: the number is not where the productivity lives. So I am going to give you something better. And the right question you should ask yourself is the one the rest of this page answers. What does your work actually run on, and how do you allocate it? The answer is energy, not time. The difference between managing the two is the difference between Sofia’s week and the week she could be having.
Sofia logs sixty hours and calls it a good week
Sofia is in the fourth year of her PhD, and her metric is hours. Sixty is a good week. The badge is visible presence: evenings in the office, weekend commits, first in after conference travel. Look closely at a Tuesday inside those sixty, though, and the accounting changes. The morning’s two sharp hours produced more than the afternoon’s five foggy ones. The evening’s three were spent re-reading a methods section her eyes kept sliding off. And re-doing on Wednesday what tired-Sofia did on Tuesday night has become a quiet, recurring line item. Sofia is not lazy, and she is not even wrong that effort matters. She is running a munitions factory on the pre-war model: counting the hours in, never the shells out. The model is costing her the very output it is supposed to prove.
What the hours model actually costs
The costs run on two ledgers. The first is the output ledger the factories discovered: fatigued hours produce less, produce errors, and produce rework. The marginal evening hour is often worth a fraction of the morning one, and sometimes worth less than zero once Wednesday’s corrections are billed. The second ledger is the one I keep writing about on this site. Levecque and colleagues [4] found work organisation and workload among the strongest organisational predictors of the psychological distress that affects one in two PhD students. The mechanics of that pathway, strain compounding like debt, with the interest paid in sleep, composure and attention, are the subject of my article on PhD burnout. Hold both ledgers at once and the hours badge inverts. A sixty-hour week that produces less than forty hours of value while accruing interest on the other ledger is not dedication. It is bad accounting, worn proudly.
The energy model: three currencies, not one
Replace the single currency of hours with the three your work actually runs on, and the week reorganises itself. This is the accounting behind the central model of this site, the one everything here connects back to: The Academic Athlete. What follows is that model applied to the working week.
The floor: a sixteen-hour day, anchored by sleep. Everything starts from an inversion I advocate hard, when I write about PhD burnout and also in every training we run: sleep is the fixed block, and the day you are actually planning has sixteen hours in it, not twenty-four. I hold this as a strong opinion. I own it. And research backs me up all the way. Every judgement about how many hours you can work is unreliable while chronically underslept, because sleep-deprived brains overestimate their coping. That is how sixty-hour weeks come to feel sustainable, right up until they are not.
The rhythm: energy arrives in waves, not a block. Attention runs in roughly 90-minute cycles. That is why the 90-20 time boxing I map in my post on the Pomodoro technique for researchers works as well as it does. One participant structured “a whole workday in 90-minute units (and pauses)” and made the observation that matters: it stopped them expecting four unbroken hours of concentration from themselves, and getting frustrated at the inevitable failure. A day planned as waves gets more from the same hours than a day planned as a slab. And, crucially, it plans the pauses instead of stealing them.
The peaks: your best hours are not interchangeable with your other hours. This is the currency almost nobody allocates deliberately, and it is the highest-denomination note in the wallet, which is why it gets the next section to itself.
Focus time: put your best work where your best hours are
Within your sixteen waking hours sit two or three in which your particular brain does its hardest work at something like double value. Their location is set by your chronotype, not your calendar. I wrote about the science in Night Owls. The single highest-leverage scheduling move available to a researcher costs nothing: identify your peak window and reserve it, daily, for the work that actually needs your best brain. The analysis, the writing, the thinking. Email, admin and meetings get moved to the energy they deserve, which is less. One of our participants, a night owl, described the discovery plainly: “I have started to work creative between 10 and 11 p.m. This is the most productive time for me.” Ten to eleven at night, and it is the right answer for that researcher. That is the point: peak placement is empirical, not moral. What belongs inside the protected window, and how to keep it unfragmented once you have placed it, is the subject of my article on deep work for researchers.
And here is what the reallocation does to the feeling of the week, in the words of another participant, rediscovering their own rhythm inside our training:
“It was nice to “re-discover” what I was doing intuitively 🙂 This however really helped alleviating my guilt when I was not productive 100% of the time throughout the day.”
Read the second sentence twice. It names the hours model’s cruellest product, and it is not the lost output. It is the guilt: the standing verdict of inadequacy, issued every afternoon, by a metric that demands uniform productivity from a brain that does not produce it. No brain does. Never has, in anyone. The energy model does not lower the bar. It moves the bar to where the evidence says performance actually lives. And the guilt, which was never information, only bad accounting felt from the inside, dissolves on contact. That sentence comes from a researcher who worked no fewer hours afterwards and produced more. It is the whole argument of this page in one line of feedback.
The focus budget: two and a half to four hours, and that is the day
Here is the number that surprises every class, so sit with it for a moment. The research on deliberate practice, the studies of elite violinists, athletes and chess players that Anders Ericsson and colleagues ran [5], keeps landing on the same ceiling: expert performers sustain roughly two and a half to four hours of full-concentration work per day. Depending on the person and the kind of task, that is the budget. It comes in blocks, usually two or three sittings with real recovery between them, and the best performers in the world do not get more of it than that.
Now hold that number against what researchers quietly demand of themselves. PhD students and postdocs routinely believe they should be able to crank out an eight-hour day of pure focus, and file themselves as failures when they cannot. Some are attempting sixteen. It is not possible. Not with more discipline, not with 90-minute cycles, not with perfect breaks. The ceiling is physiological, and no scheduling trick raises it. The cycles and the breaks decide how close to your own ceiling you get, not where the ceiling sits.
Which turns the hours question on its head one final time. Your two or three peak hours are not a nice slice of your workday. They are most of your focus budget for the entire day. Spend them on email and the budget is gone by ten in the morning, whatever the timesheet says about the rest of the day. The other waking hours are not worthless: they carry the shallow work, the meetings, the bench mechanics, and the recovery that refills tomorrow’s budget. But the doctorate gets written inside the budget, the budget is small, and that is not a personal failing. It is the specification of the equipment.
Sofia’s week, re-denominated
Sofia ran the audit this article amounts to: one week, energy logged against output, no behaviour change yet. The findings embarrassed the badge. Her peak, nine to noon, had been hosting her email. Her analysis had been living at five p.m., in the fog. Her Sunday hours produced almost nothing except a slower Monday. The factories found the same about Sunday shifts, a century ago. The rebuild cost her nothing but the badge. Analysis moved into the morning peak. Email demoted to the two-thirty trough. Work planned in waves, with the pauses budgeted. The sixteen-hour day anchored by an actual bedtime. Her logged hours fell to just over forty. Her supervisor, who sees drafts rather than timesheets, remarked within a month that her output had jumped, and asked, with no irony available to him, what she was doing with all her extra time.
A word for the PI whose lab glorifies hours
And if you lead a group, take a straight look at which model your lab rewards. The hours culture is not self-generating. It is modelled: the late emails, the weekend presence quietly noticed, the “X was in at seven” said admiringly at coffee. You are the factory manager in this analogy, and the factory data is a century old. Beyond the threshold, the hours you are implicitly demanding are not buying you output. They are buying fatigue, errors, rework, and your share of the distress statistics, paid by the people with the least power to refuse, and, among the brightest of their generation. The alternative costs one cultural sentence, said out loud and meant: I judge the work, not the hours, and I expect you to protect your recovery, because I have read what happens when you do not. The skills underneath the energy model, the planning, the rhythm, the peak placement, are what Fast Forward, our project management and personal productivity training, trains in seven weeks, with over 1,500 individuals trained at more than 30 institutions. The factories that needed output most stopped counting hours a hundred years ago. A lab that needs discoveries can afford to be at least as modern as a shell factory.
FAQ
How many hours should a PhD student work per week?
The straight answer is that hours are the wrong unit, and I hold that as a strong opinion. The evidence runs from the First World War munitions records, where the economist John Pencavel showed output flattening beyond a threshold, to Iceland’s public-sector trials and the UK’s four-day-week pilot in this decade, where modestly shorter weeks held output steady. The productive question is allocation: anchor a sixteen-hour waking day with fixed sleep, plan work in roughly 90-minute waves with real pauses, and reserve your two to three peak hours daily for your hardest work. Measured that way, most researchers produce more in modestly fewer hours.
Is working 60 hours a week normal during a PhD?
Common, yes; productive, mostly no. The long-hours weeks in the historical output data bought little or nothing at the margin, and the doctoral evidence links workload and work organisation to the psychological distress affecting one in two PhD students, the finding of a 2017 study of more than 3,600 PhD students. A 60-hour week that produces less than forty hours of value while eroding sleep and composure is bad accounting worn as a badge, and the badge is usually cultural, set by what the lab visibly rewards.
How many hours of deep, focused work are possible per day?
Fewer than almost anyone believes: roughly two and a half to four hours, depending on the person and the task. The deliberate-practice studies of elite performers, run by Anders Ericsson and colleagues, found the same ceiling across fields, sustained in two or three blocks with real recovery between them. An eight-hour day of pure focus is not a discipline problem you have failed at. It is a physiological impossibility everyone fails at, no matter how the day is cycled. Plan around the budget: peaks for the hard work, the rest of the day for everything else.
When is the best time of day to do PhD work?
When your chronotype says, not when the office does. Everyone has a two-to-three-hour peak window in which hard cognitive work runs at something like double value; for morning types it is early, and for real night types it can sit at ten p.m. and be the right answer. The move that costs nothing and pays most: put analysis, writing and thinking inside your peak, and demote email and admin to the energy they deserve.
Should PhD students work weekends?
As a system, no; as a rare exception, that is what exceptions are for. The munitions records found Sunday work so unproductive that weekly output barely benefited, because the recovery it consumed was billed to Monday, and the modern recovery research says the same about weekends spent half-working. Two properly free days are not time stolen from the doctorate. They are where next week’s capacity is manufactured.
References
[1] Pencavel, J. (2015). The productivity of working hours. The Economic Journal, 125(589), 2052–2076.
[2] Haraldsson, G. D., & Kellam, J. (2021). Going Public: Iceland’s Journey to a Shorter Working Week. Autonomy & Alda (Association for Sustainability and Democracy).
[3] Autonomy (2023). The Results Are In: The UK’s Four-Day Week Pilot. Research led by Boston College (J. Schor, W. Fan), University of Cambridge and University College Dublin.
[4] Levecque, K., Anseel, F., De Beuckelaer, A., Van der Heyden, J., & Gisle, L. (2017). Work organization and mental health problems in PhD students. Research Policy, 46(4), 868–879.
[5] Ericsson, K. A., Krampe, R. T., & Tesch-Römer, C. (1993). The role of deliberate practice in the acquisition of expert performance. Psychological Review, 100(3), 363–406.
Dr Nadine Sinclair
Dr. Nadine Sinclair is a molecular biologist, strategy consultant and co-founder of Mind Matters. A scientist by training and at heart, she conducted her doctoral research at the Max Planck Institute for Biophysical Chemistry. She has 30,000+ hours of hands-on project management experience, built across 18 years as a strategy consultant at McKinsey and as an independent consultant, working with research institutions, foundations, pharmaceutical and biotech companies and governments. She is the co-creator of the Personal Resilience Indicator and the author of On Track. Since 2018, over 1,500 researchers from more than 30 institutions have trained with Mind Matters.