Churchill ran a world war and still went to bed every afternoon. Not a head-tilt in a chair. Undressed, under the covers, on purpose. He explained the practice to a Time-Life man over breakfast in 1946, and the International Churchill Society keeps the account: "You must sleep some time between lunch and dinner, and no half-way measures. Take off your clothes and get into bed. That's what I always do." His claim for it was that you get two days in one — well, he allowed, at least one and a half.

Point being: napping isn't what you do because you failed to sleep properly. It's a skill, and the people who take it seriously tend to be the ones with the most on the line.

The research is unusually practical about it, too. Trial after trial lands on the same finding — whether a nap helps you has less to do with taking one than with how long it runs and where it sits in your day. Get the size right and the alertness holds. Get it wrong and you spend your first twenty minutes back in a fog you did not have before you lay down.

So here's the menu. Three naps, three different jobs.

The 20 to 30 — the on-call nap

Short keeps you in the shallow end. When the phone or the tones go off you come up fast instead of clawing your way out of the deep stuff.

This is the only nap for when the job can wake you. Set the alarm at 30 as insurance. The zone to avoid is roughly 40 to 60 minutes, which is long enough to sink and too short to finish the climb back out.

How much of that is specifically about slow-wave sleep is less settled than the internet thinks. The big 2019 review of sleep inertia found that deeper sleep at the moment of waking did track with slower responses in some studies and not in others, and concluded plainly that it is difficult to synthesize a clear conclusion. What is not in dispute is the shape of the thing: the grogginess is real, it is worst right at the alarm, and it fades.

Field practice, not research: Jocko Willink's version runs shorter than anyone expects. Six to eight minutes, feet elevated above the heart — "I still do when needed." No trial behind that specific protocol; he swears by it anyway. What I'd take from it isn't the number. It's that he does the same thing the same way every time, so his body knows the drill before he lies down. That's a practice, not a trial.

The 90 — the full cycle

Down through deep sleep and back up the other side. This is the one that actually pays down debt, and it's half the protocol behind the split nap — a 90 plus a 30 that held the line as well as a single two-hour block across a simulated sixteen-hour night.

It has one rule: only when someone else catches the first call. Partner has the radio. Kids are at grandma's. Because the cost of getting yanked out of this one is not trivial, and there is a real number on it.

In a 2006 study in JAMA, Wertz and colleagues found that performance on a cognitive test immediately after waking was worse than performance after a full night of sleep deprivation. Read that again. The first few minutes off a nap can test worse than not having slept at all. It resolves — most measures return toward baseline within about thirty minutes, with full recovery taking an hour or so. But it is not nothing, and it arrives exactly when someone is handing you a patient.

So the 90 is a nap you schedule, not a nap you take. Anything that can wake you has to be somebody else's problem for that hour and a half.

The coffee nap — the weird one that works

Coffee, then immediately lie down. Caffeine needs twenty to thirty minutes to reach peak concentration in your blood, so it arrives about when your alarm does, and the two stack instead of competing.

It sounds like an internet trick. It comes out of drowsy-driving research. Reyner and Horne at Loughborough put sleepy drivers through a two-hour monotonous simulated drive and compared 200 mg of caffeine alone against caffeine followed by a 15-minute nap. Caffeine alone cut driving incidents to about a third of placebo. Caffeine plus the nap cut them to under a tenth. Same caffeine. The nap did the rest.

Best use is the obvious one: parking lot, seat back, before the drive home off nights. Yes, the parking lot. The alternative is drifting lanes at 0730.

Where they go in a real week

Before your first night shift, the 90 in the early evening — you're pre-paying, not borrowing. On a 24 or a 48, everything is the 20 to 30 until your partner takes the radio and you can afford the long one.

On rotation, the easy window moves with you. Your natural dip runs roughly four to five and a half hours after you wake, wherever waking currently lives — which for a night-shift week is the middle of the afternoon and for a day-shift week is somewhere after lunch. That's the same arithmetic the free plan builder runs against your actual rotation, so you don't have to do it at 0400 with a calculator.

And the rule that outranks the whole menu: a nap is not a moral event. You are not lazy for taking one and not tough for skipping one. Churchill knew that, and he wasn't shy about it. The man took off his clothes and got into bed in the middle of a war.

It was policy.

Sources

This is not medical advice. Talk to your provider.

You are going to be tired this week regardless. The only question on the table is whether the rest you get is the kind you chose.