The same seven hours are worth more taken at the same time

Applied physiology series, part three. Part one read recovery as a budget and part two named the withdrawal that returns the most; this one looks at where the budget is actually refilled, before part four asks which measurements are worth taking at all.

Nine volunteers. Eight hours in bed for three nights, then four hours a night for four nights running, with bedtime held at eleven while the alarm walked back to three, which is roughly what a hard fortnight does to anyone holding down a job and an early start.

Brunner, Dijk and Borbély ran it in 1993. Slow wave sleep came through the short nights largely untouched. Stages one and two went, and REM went with them (Brunner et al., 1993).

So a night doesn’t shrink evenly. Take a fifth away and you don’t lose a fifth of everything in it, because the body triages, defends the front of the night, and charges almost the whole shortfall to the back end where REM lives. The order of a night decides what you get out of it, and total hours are a much weaker predictor than nearly everyone assumes.That’s the narrow claim. Duration still counts, and a genuinely short night is short in every compartment at once.

This is an educational and strategic perspective, not personal medical advice. The views are the author’s own and not statements by Atlas Cove Lda.

Two systems decide when you sleep, and they answer to different things. The first is homeostatic: pressure builds the longer you stay awake and drains once you’re under. The second is circadian, run by an internal pacemaker on roughly a twenty-four hour cycle, and it doesn’t care how tired you are. Borbély and colleagues set out the interaction of those two, Process S and Process C, as the arrangement that best reproduces both when human sleep happens and how deep it goes, which is a model that has survived thirty-four years and a formal reappraisal by its own authors (Borbély et al., 2016). The messenger on the homeostatic side is adenosine, which accumulates across a waking day. Caffeine sits on its receptors. Hold that thought.

Two clocks instead of one has an awkward consequence. Seven hours starting at eleven and seven hours starting at two aren’t the same purchase, because the circadian system is somewhere different in its cycle and permits different things.

A night flight and an ordinary night don’t do the same work. Same duration, different night.

The cleanest evidence comes from forced desynchrony. It’s a slightly brutal protocol: you put people on a day length their body clock can’t follow, so their sleep episodes land at every circadian phase in turn and the two systems come apart where you can watch them separately.

Dijk and Czeisler kept eight men on twenty-eight hour days for a month with no time cues. The split was clean. REM sleep carried a strong circadian rhythm whose crest sat just after the low point of core body temperature, while slow wave activity fell across every single sleep episode and showed only a small circadian wobble that didn’t track sleep propensity at all (Dijk and Czeisler, 1995).

Then comes the detail that turns this from a curiosity into something you can act on, which is that REM also responded to how long the person had already been asleep and reached its highest values of the whole protocol when the scheduled sleep episode happened to end at the hour those men would normally have woken up anyway.

The hours where REM is most available are exactly the hours an early alarm removes. Structural fact. Nothing to do with your discipline.

I assumed for years that deep sleep was the thing to defend after forty. Everyone says so. It’s where the repair happens, and the repair is what you want.

The first half of that is right. Slow wave sleep carries the largest growth hormone pulse of the day, tissue repair runs hardest there, and it’s when parasympathetic activity peaks, which ties it straight back to the autonomic account part one of this series was about.

The second half is where I was wrong. Van Cauter, Leproult and Plat pooled records from 149 healthy men aged sixteen to eighty-three. Deep slow wave sleep filled 18.9 percent of the night in the sixteen to twenty-five year olds. In the thirty-six to fifty year olds it filled 3.4 percent. The difference had been absorbed by lighter sleep, and REM hadn’t significantly dropped yet at that point. Growth hormone output fell alongside it and tracked slow wave sleep independently of age (Van Cauter et al., 2000).

Those were cross-sectional comparisons between age bands, all of them men, so nobody was followed as they aged and none of it licenses a claim about women. It still reorders the priority. By the time you’re forty-five, deep sleep has mostly already done its declining, and REM is the compartment still substantially there and still being cut every morning. I’d been guarding the part that had largely gone and spending the part that hadn’t.

Losing REM produces no symptom you’d notice. That’s most of why it disappears without complaint, and it’s what makes the trade feel free.

Wagner, Gais and Born used the night’s own structure as their instrument and tested recall of emotional against neutral material across an early-sleep window rich in slow wave sleep and then across a late-sleep window rich in REM, and the emotional advantage turned up across the late one, which they read as consistent with a supporting role for the REM that dominates the second half of a night (Wagner et al., 2001).

The popular version of that finding gives REM a good deal more than the evidence does, motor skill learning in particular, where the lighter non-REM stages have the better claim. Worth knowing if you’ve read otherwise.

Kept at its real size the point holds. Cutting the tail off a night removes something specific rather than a slice of everything, and what goes is emotional processing, which no training log and no blood panel is built to see. Week after week it shows up as a slightly shorter fuse. Almost nobody blames the alarm clock.

Everyone says keep the bedroom cool. The specifics are better than that.

Kräuchi and colleagues ran a whole field of candidate predictors against how fast people actually dropped off, and the one that won was the distal to proximal skin temperature gradient, roughly how much warmer your hands and feet run than your trunk, which beat core body temperature, the rate at which core temperature was changing, heart rate, melatonin onset and how sleepy people said they felt (Kräuchi et al., 2000).

So you aren’t simply getting colder. You’re opening the taps at the extremities and dumping heat, and the cool room matters because it gives that heat somewhere to go. Cold feet and a cold bedroom fight each other. Socks with the window open. Took me an embarrassingly long time to accept that one.

Alcohol is the clearest case, and the honest version is more interesting than the disapproving one. Twenty-seven studies pooled in a 2025 systematic review and meta-analysis found that falling asleep faster, which is the effect the drink is actually bought for, only registered at around five standard drinks, while total sleep time, sleep efficiency and night-time waking all stayed too uncertain to call and the one thing that came out clearly was a dose-response effect on REM, delayed and reduced at roughly two standard drinks and worsening as the dose climbed (Gardiner et al., 2025).

None of which says the wine at dinner has to go. It says the thing it’s bought for is largely unevidenced, while the cost lands on the compartment that’s already most exposed.

Caffeine works on the other system. It occupies adenosine receptors without switching them on, so the pressure keeps building while your ability to feel it drops away, and at a half-life around five to six hours the afternoon cup is still substantially present at midnight. Drake and colleagues gave 400 mg, about two strong coffees, at zero, three and six hours before habitual bedtime in a randomised placebo-controlled crossover, and every one of those timings cost more than an hour of sleep, with deep sleep at the six hour point falling to roughly forty-nine minutes against seventy-one on placebo (Drake et al., 2013).

Here’s the part that explains the behaviour. At six hours the instruments caught it and the participants’ own diaries didn’t. The cost wasn’t absent. It was invisible from the inside, which is a fair explanation for why people ignore this advice in perfectly good faith. Twelve participants though, and a home EEG headband rather than a laboratory rig. Hold it loosely.

Most people will check all of this against a wrist, and the devices deserve neither the dismissal nor the trust they usually get.

Chinoy and colleagues put seven consumer trackers against laboratory polysomnography across three nights in thirty-four healthy young adults, one night deliberately disrupted. Every device detected sleep well, sensitivity 0.93 or better. Specificity landed between 0.18 and 0.54. Most devices left thirty to fifty percent of both deep and REM sleep unidentified on average, usually relabelling it light sleep, and they got worse on the poorer nights (Chinoy et al., 2021).

So last night’s stage chart is an estimate, it degrades on exactly the broken nights that made you open the app in the first place, and averaging across a week doesn’t rescue it either, because a weekly figure sitting near the truth may have arrived there by cancelling errors in both directions.

This one irritates me. The least trustworthy number on the screen is the one everybody reads, and the most trustworthy one is the timestamp nobody looks at. Your device knows when you went to bed and when you got up. That’s a timing measurement, and timing is where the evidence actually is.

Windred and colleagues built a sleep regularity index out of more than ten million hours of accelerometer data across 60,977 UK Biobank participants, mean age sixty-three, then followed deaths for an average of a little over six years, and against the least regular fifth of that cohort the four more regular fifths carried twenty to forty-eight percent lower all-cause mortality risk. Then they put regularity and duration into equivalent models. Regularity won, minimum hazard ratio 0.52 against 0.69 for duration before adjustment, 0.70 against 0.76 after (Windred et al., 2024).

It’s observational. The authors say plainly that regularity might be a cause of mortality risk or simply a marker of it, nobody has shown that fixing your schedule lengthens your life, and somebody drifting into illness would produce this same signature. Read at its proper strength it still says something worth having: the best-evidenced input here costs nothing, needs no equipment, and is the first thing a calendar takes.

Now the bill. I live in Portugal, where dinner starts at an hour I was raised to consider bedtime, so holding a wake time inside a one-hour band across all seven days costs me evenings I’d rather have and I don’t always pay it. That’s the real trade. Anyone telling you it’s free hasn’t tried to hold it through a Portuguese summer.

Two weeks and no hardware will tell you where you stand. Write down when you actually get up, then compare the weekday average against the weekend, and read a gap above an hour as one instruction handed to your circadian system every Friday and reversed every Sunday. Then take your last coffee and add six hours. If that lands after your intended lights out, you’ve got something to change tonight.

The strange thing about the Ironman I raced at Cascais in October 2025 is how much of it I could account for afterwards. Training distribution to the percent. Heart rate by discipline, 143 in the water, 142 on the bike, 144 on the run. Carbohydrate to the gram. My sleep timing across that whole build sits in no file anywhere, not in the training log, not in a spreadsheet, not on a watch, and of everything I recorded that year it’s the one variable carrying the mortality evidence I’ve just spent two thousand words on.

An Atlas Cove week puts sleep timing in the same column as the training load and the autonomic data, because a variable that decides what a night returns shouldn’t be filed under background conditions. Guests leave knowing which hour of their own night is the one to defend.

Atlas Cove - The Return

Eight hours describes an opportunity. It says nothing about whether the night contained the deep sleep that repairs, the REM that processes, or the regularity that outpredicts both. Almost none of what's available here needs more time in bed, which nobody has going spare. It needs the same time, held.

This is an educational and strategic perspective, not personal medical advice. The views are the author's own and not statements by Atlas Cove Lda.

1. Borbély, A. A., Daan, S., Wirz-Justice, A., & Deboer, T. (2016). The two-process model of sleep regulation: a reappraisal. Journal of Sleep Research, 25(2), 131-143. DOI: 10.1111/jsr.12371

2. Dijk, D. J., & Czeisler, C. A. (1995). Contribution of the circadian pacemaker and the sleep homeostat to sleep propensity, sleep structure, electroencephalographic slow waves, and sleep spindle activity in humans. The Journal of Neuroscience, 15(5 Pt 1), 3526-3538. DOI: 10.1523/JNEUROSCI.15-05-03526.1995

3. Brunner, D. P., Dijk, D. J., & Borbély, A. A. (1993). Repeated partial sleep deprivation progressively changes the EEG during sleep and wakefulness. Sleep, 16(2), 100-113. DOI: 10.1093/sleep/16.2.100

4. Van Cauter, E., Leproult, R., & Plat, L. (2000). Age-related changes in slow wave sleep and REM sleep and relationship with growth hormone and cortisol levels in healthy men. JAMA, 284(7), 861-868. DOI: 10.1001/jama.284.7.861

5. Wagner, U., Gais, S., & Born, J. (2001). Emotional memory formation is enhanced across sleep intervals with high amounts of rapid eye movement sleep. Learning & Memory, 8(2), 112-119. DOI: 10.1101/lm.36801

6. Kräuchi, K., Cajochen, C., Werth, E., & Wirz-Justice, A. (2000). Functional link between distal vasodilation and sleep-onset latency? American Journal of Physiology-Regulatory, Integrative and Comparative Physiology, 278(3), R741-R748. DOI: 10.1152/ajpregu.2000.278.3.R741

7. Gardiner, C., Weakley, J., Burke, L. M., Roach, G. D., Sargent, C., Maniar, N., Huynh, M., Miller, D. J., Townshend, A., & Halson, S. L. (2025). The effect of alcohol on subsequent sleep in healthy adults: a systematic review and meta-analysis. Sleep Medicine Reviews, 80, 102030. DOI: 10.1016/j.smrv.2024.102030

8. Drake, C., Roehrs, T., Shambroom, J., & Roth, T. (2013). Caffeine effects on sleep taken 0, 3, or 6 hours before going to bed. Journal of Clinical Sleep Medicine, 9(11), 1195-1200. DOI: 10.5664/jcsm.3170

9. Chinoy, E. D., Cuellar, J. A., Huwa, K. E., Jameson, J. T., Watson, C. H., Bessman, S. C., Hirsch, D. A., Cooper, A. D., Drummond, S. P. A., & Markwald, R. R. (2021). Performance of seven consumer sleep-tracking devices compared with polysomnography. Sleep, 44(5), zsaa291. DOI: 10.1093/sleep/zsaa291

10. Windred, D. P., Burns, A. C., Lane, J. M., Saxena, R., Rutter, M. K., Cain, S. W., & Phillips, A. J. K. (2024). Sleep regularity is a stronger predictor of mortality risk than sleep duration: a prospective cohort study. Sleep, 47(1), zsad253. DOI: 10.1093/sleep/zsad253