In April 2014, Nadine Gravett tranquilized two female elephants and fitted them with actiwatches. These small devices—the scientific version of Fitbits—record movement, and researchers can use them to measure how well volunteers are sleeping. They’re usually worn around the wrist, but that’s not an option when your subjects’ limbs are literally elephantine. So Gravett had to implant them in the females’ most mobile appendages—their trunks.

The skin around the middle of the trunk is so thick that the implants went unnoticed, and quietly recorded the animals’ movements for a month. By analyzing their data, and looking for five-minute windows when the trunks were still, Gravett could deduce when the elephants were asleep. And she found that they slept for just two hours a day on average—the lowest duration for any animal thus far recorded.

“Sleep is such a weird behavioral state,” says Paul Manger from the University of Witswatersrand, who led the study. “For animals, the main things in life are eating, reproducing, and not being eaten—but when we’re sleeping, those things fall away. Sleep supersedes a lot of our survival instincts. We do know a lot about it in lab animals, but we don’t know a lot in exotic species.”

Manger has been studying sleeping animals for almost two decades. He began with, of all things, the platypus, which turned out to get more rapid eye movement (REM) sleep—the type in which dreams occur—than any other animal. From that quirky start, Manger went on to study sleeping dolphins, whales, hippos, echidnas, cats, and antelopes.

But most of that research involved captive animals, which enjoy plentiful food and an absence of predators. For those reasons, they sleep much more than their wild cousins. For example, in the 1980s, scientists found that captive three-toed sloths sleep for 16 hours a day—a result that earned them a reputation for, well, sloth. In 2008, when other researchers recorded the brain activity of wild sloths, he found that they only sleep for 10 hours a day. Manger also wanted to take the science of sleep into the wild. And he started with elephants.

As a very rough rule, and for reasons that are still unclear, bigger mammal species tend to sleep less than smaller ones. Captive elephants reportedly sleep for just 3 to 7 hours a day, but for their size, you’d expect them to get even less. Proving that is hard, though. Elephants can sleep standing up, so it’s difficult to eyeball whether they’re awake or asleep, especially if you’re tracking them through the bush at night. Brain activity sensors would give better answers, but an elephant’s anatomy makes it exceedingly risky to surgically implant such devices. So Manger and Gravett settled for the actiwatches instead. They also fitted satellite collars onto the animals to track their whereabouts later.

As Manger suspected, the two females got just over two hours of sleep a night, and even that isn’t unbroken—it occurs in four to five bursts, spread throughout the night. On most nights, they slept standing up; on some, they lay down. They weren’t fussy about their choice of sleeping spots, and the distance they traveled during the day had no effect on how long they slept.

“Under natural conditions, animals should show shorter sleep times than in the lab, where food is always available and there are no predators,” says Isabella Capellini from the University of Hull. “It's very nice that we start getting estimates of sleep outside the lab.”

Of course, Gravett and Manger only studied two elephants, and both were adult matriarchs—the individuals who bear the burden of leading the herd. It’s possible that males, youngsters, or lower-ranked females might get more shut-eye, but it’s at least clear that in this particular demographic, sleep time is superlatively low.

The gray whale is much larger than an elephant, but still sleeps for nine hours a day. A dolphin can sleep with one half of its brain at a time, allowing it to stay constantly alert for days at a stretch—but each half still gets at least four daily hours. Oddly, the next shortest sleeper seems to be the domestic horse, which gets just three hours a day, followed by the domestic pony, which gets slightly more. For now, it appears that the African savannah elephant is the shortest sleeper around. (The longest seems to be the little brown bat, which gets an incredible 19 hours a day.)

Weirdest of all, Gravett and Manger saw that for a few nights of the month, their two elephants didn’t sleep at all. Why?

The duo did their study in Botswana’s Chobe National Park—one of the few places where elephants are hunted by lions. The two tagged individuals were both adult matriarchs who were taking care of young calves, so they may have stayed up all night to avoid harassing lions. Similarly, they might also have been trying to flee from poaches, or bull elephants in heat. Whatever the reason, the two females never seemed to make up for their sleepless nights—at least not by sleeping longer.

“The remarkably short amount of sleep in wild elephants is a real elephant in the room for several theories for the function of sleep,” says Niels Rattenborg from the Max Planck Institute for Ornithology. Some scientists have argued that sleep evolved to give animals a chance to reset their brains, ready for a new day of learning. Others suggest that sleep provides an opportunity to clear out toxins that accumulated during the day. And yet others say that sleep allows animals to consolidate the memories that they have created while they were awake.

But if any of these ideas are right, how do elephants cope with such little sleep? “The hypotheses about restorative functions start to go out the window,” says Manger. “You can’t say that these are general things that apply to sleep across all mammals.” The idea about memory consolidation becomes especially shaky: it’s meant to happen during REM sleep, and Manger’s elephants only seemed to get REM sleep every three to four days. How do they remember anything at all, much less develop their apocryphally long-lasting memories?

It’s possible that elephants have developed special adaptations that allow them to cope with sleeplessness, and Suzana Herculano-Houzel from Vanderbilt University wonders if those adaptations allowed them to become jumbo-sized in the first place. “They must eat 17 to 18 hours a day, and I like to think that they were only viable if they could afford very little sleep first,” she says. “We know that sleep is indispensable, but we don’t know exactly what regulates the amount of sleep that different animals require. One difficulty is getting good data on how many hours different animals sleep—and the elephant, being an apparent extreme, is a very important case study.”

Manger agrees that we’ll only really understand the nature and evolution of sleep if we study a wide range of species—especially in the wild. He plans to continue doing so, although of late, he has been struggling to find the time. “I have a three-year-old, so it’s been a while since I’ve had a good night sleep,” he says.