The experiment was the third of five studies in a paper led by Nicholas Light with Philip Fernbach, Nathaniel Rabb, Mugur Geana and Steven Sloman. Across the studies, the team compared three things: how strongly people opposed the expert consensus on a contested scientific issue, how much they actually knew, and how much they believed they knew.

This is one research group’s work, not settled consensus. But it sits on top of a decade of related research, some of which held up and some of which did not, and that history says as much about forming opinions from evidence as the headline finding does.

What the bet measured

The first two studies set up the pattern. According to PsyPost’s report on the paper, 3,249 US adults recruited through online survey platforms were each asked where they stood on one of seven issues on which scientists broadly agree: climate change, genetically modified foods, nuclear power, vaccination, evolution, the Big Bang and homeopathic medicine. Those who fully agreed with the scientific consensus were excluded, leaving 2,133 participants who rated their own understanding of the issue on a seven-point scale running from “vague” to “thorough.” Then they answered 34 true or false questions on general science, items such as whether the center of the earth is very hot.

People who were more opposed to the consensus were more likely to claim a thorough understanding. They also tended to score worse on the test.

A self-rating costs nothing, so the third study added stakes. Participants could bet on their ability to score above average on the knowledge test. Those with greater opposition to the consensus tended to earn less, and the authors attribute the shortfall to knowledge overconfidence: they bet on a performance they did not deliver.

Two further studies, run in July and September 2020, found similar patterns for willingness to take a COVID-19 vaccine, before one was publicly available, and for attitudes toward mitigation measures such as mask wearing and social distancing. In the September study the pattern held after the researchers accounted for political identity. In all, the five studies recruited more than 5,600 participants, before exclusions.

The same gap appeared in an earlier study of GM foods

The 2022 paper extended work the same group had published in Nature Human Behaviour in 2019. In that study, led by Fernbach, more than 2,000 adults in the United States and Europe rated how well they understood genetically modified foods and then took a true or false test on science and genetics. More than 90 percent expressed at least some opposition to GM foods. As opposition became more extreme, test scores fell and self-rated understanding rose.

The paper’s abstract puts it plainly: “Extreme opponents know the least, but think they know the most.” The pattern repeated in representative samples from the United States, France and Germany, and in a separate study of attitudes toward gene therapy.

It did not repeat for climate change.

Light’s team found the same exception in 2022. “The main caveat is that although this pattern of effects appears to be fairly general, we don’t find it for all issues,” Light told PsyPost. “One notable example is climate change.” The researchers also saw some evidence that on more politically polarized issues, the link between opposition and low knowledge was weaker. One plausible reading, which the authors of the 2019 paper raised, is that on polarized topics people’s positions follow their group more than their grasp of the details.

What the studies do not show

These are correlations. The design cannot say whether knowing less leads people to oppose a consensus, whether opposing it leads them to stop learning, or whether something else drives both. The samples were recruited online, and the knowledge test measured general science literacy rather than expertise on each specific issue.

The findings also describe tendencies across many people. They do not mean that any particular person who disagrees with a scientific consensus is uninformed, and plenty of people who accept the consensus could not explain the science behind it either.

The authors draw a practical conclusion of their own. Because the strongest opponents tend to be the most overconfident, they wrote, “fact-based educational interventions are less likely to be effective for this audience.” Someone who already feels they know all there is to know has little reason to listen.

A neat fix for overconfidence that failed to replicate

If overconfidence props up strong opinions, puncturing it might soften them. In 2013, Fernbach and colleagues reported exactly that in Psychological Science. In three small online experiments, Americans who were asked to explain step by step how policies such as a flat tax or cap-and-trade would work rated their own understanding lower afterward, and their positions became modestly less extreme. Simply listing reasons for a view did not have the same effect.

It was an appealing result. Then other researchers tested it.

In 2021 Jarret Crawford and John Ruscio published three preregistered replications, meaning the methods and analyses were fixed in advance, with 306, 405 and 343 participants. People again reported understanding the policies less after explaining them. Their positions were no less extreme. The authors concluded that explaining “increases people’s recognition of their ignorance but is unlikely to increase their political moderation, at least under these conditions.” A 2024 study in Political Psychology by Richard Walker and colleagues, with 5,139 people in Hong Kong, went further and found extremity was higher when people set out their understanding of a policy.

The research on overconfident opinion was, in other words, corrected by the same process it was studying. A plausible idea was put to a test, the test was repeated, and the results proved mixed: a 2022 study found reduced extremity for policies framed around consequences but not for those rooted in protected values.

Expert opinion gets tested too

That process applies to professionals as well as the public. In 2013 Vinay Prasad, then at the US National Cancer Institute, and colleagues reviewed 2,044 original articles published in the New England Journal of Medicine between 2001 and 2010, of which 1,344 concerned a medical practice, reporting in Mayo Clinic Proceedings. Of the 363 that tested an established medical practice, 146, or about 40 percent, found the practice was no better than, or worse than, the alternative it was compared with. Another 138 reaffirmed the practice, and 79 were inconclusive.

The reversals included stenting for stable coronary artery disease, which for most patients did no better than medication, and hormone therapy for postmenopausal women intended to protect the heart, which did worse than no treatment. The figure describes one journal over one decade, and Prasad noted that the reversals were not the final word on each practice. It is not a rate for medicine as a whole.

What it does show is that confident professional opinion, held by people with real expertise, was overturned again and again when someone ran the trial.

What evidence asks of an opinion

Read together, these studies point in two directions at once. The gap between feeling informed and being informed was widest among the people most sure they were right, at least on several of the issues tested. And science itself kept revising its claims, about medicine and about the psychology of opinion, whenever a careful test disagreed with what people expected.

Basing an opinion on science, on this evidence, means more than citing a study that agrees with it. It means checking how much is actually known, how strong that evidence is, and being prepared to hold a view a little more loosely when the next test comes in.

That ranking comes from a single question. People imagine a ladder, from 0 (the worst possible life for them) to 10 (the best), and say which step they’re on. It’s a good question. But you can put yourself on a solid seven and still wake up exhausted every morning. You can have wonderful friends and be worried about rent.

And then there’s money, the question everyone asks sooner or later: does money buy happiness?

For years, the answer you’d hear was a number. A 2010 study by Daniel Kahneman and Angus Deaton suggested that day-to-day happiness stopped rising at around $75,000 a year. Later work disagreed, and in 2023 the researchers on both sides looked at the data together. Their answer was that for most people, happiness kept rising with income, and it only flattened out for the least happy.

A happiness score can’t tell you how you’re really doing

Now a new study tries to look at all of it at once. It’s called “Global mental wealth”, and it asked nearly 54,000 people in 92 countries not one question but ten: about energy, calm, relationships, meaning and more, plus questions about their income, savings, debts and the money they grew up with.

To give you a sense of scale: the last time these ten questions were asked across many countries was in 2012, in 21 European countries. The lead author of that study was Kai Ruggeri, a professor of health policy and management at Columbia University. This time, he went back to it on a much bigger scale: 92 countries, 50 languages and 261 researchers.

It’s a huge piece of work, and it has been accepted for publication in the journal Social Indicators Research. I’d bet these numbers will be quoted a lot once the final version is out. Before that happens, I asked Ruggeri what the results actually say.

I didn’t have to look far to ask. I’m one of the study’s 261 authors (number 126, so not exactly the headline name), and I helped bring the survey to Georgia, where I’m from. I’d also worked with Kai before, on a 61-country study of how people make financial decisions, published in Nature Human Behaviour. But before we get to his answers, here’s what all those people were actually asked.

Ten questions, 50 languages, one awkward word

The core of the survey is ten short statements (all ten are listed in the paper), rated from 1 to 7.

• “I feel full of energy.”
• “I receive help and support from people I am close to when I need it.”
• “I feel what I do in my life is valuable and worthwhile.”

Between June and August 2025, people answered them, mostly online, in 113 versions across 50 languages.

Among 261 authors, three of us adapted the survey for Georgia, and I have to admit, even simple words needed so much work. To give you one example, the Georgian word for “household” isn’t one people really use, so our version asks about your family instead, with “household” in brackets.

Across all countries, the highest-scoring statement was the one about support from close people (5.05 out of 7), followed by meaning (4.93). The lowest was energy (4.20), as in “I feel full of energy.” Even that lowest score is above the middle of the scale, so it doesn’t mean people are miserable. But only about one in five people gave their energy top marks, against nearly two in five for their relationships.

That’s also why one number can hide so much.

In our data, Denmark and Norway gave exactly the same life satisfaction score (6.2), yet their profiles across the ten measures looked clearly different.

Why does my own life look better than the world?

The authors read that pattern as people feeling fairly positive about their own lives while worrying about things outside their control, like political and economic instability. The survey doesn’t ask about society directly, so this is mostly interpretation. I asked Ruggeri what could be behind it.

“It is somehow equally compelling and unsurprising that people largely view their own circumstances as more positive than society at large,” he told me. “Generally, it relates to the idea of control and perception: whether my personal situation is ideal or suffering, I have some ability to impact it, and I am acutely aware of all aspects – good and bad. For society, I mostly am an observer of outcomes.”

And then he gave the example I keep thinking about: “when a company lays off 1,000 people, it makes the news; but if a company hires 1,000 over a one-year period, that is less likely to get noticed.”

It’s such a simple example, and it explains a lot.

Young adults came out lowest. Is it age, or this generation?

You might expect the usual story, where life satisfaction dips in middle age and climbs back up later.

The study found no such U-shape. Instead, young adults scored lowest, and well-being mostly rose with age.

On a short screening questionnaire, only 14.7% of 18- to 25-year-olds showed no signs of depression or anxiety, compared with 23.6% of everyone. That gap is hard to ignore.

The paper also notes that retirees now score about as well as working adults and students in most countries, which wasn’t the case in 2012. I asked Ruggeri whether that’s about age or about this generation.

“What’s far more concerning is just how widespread the low scores for young adults is,” he said. He listed a few possible reasons, from young people being “overburdened and given limited space to develop at their own pace” to social media and “extreme/perpetual peer comparisons.” Then he pointed to one thing in particular: “Our own data mainly points toward future outlook: messaging toward young people, especially in the face of AI advances, implies careers and economic opportunities are going away, whereas future outlook used to be a primary driver of positive mental health in young people.”

More money helps, but each step buys less

On money, the study found something a little uncomfortable: well-being kept rising with income, with no clear point where more money stopped helping. So I asked Ruggeri what that means for someone choosing between a higher salary and more free time.

He was careful. “There are major differences between economic groups,” he said, so people under more financial pressure probably need more money to reach their happiest point, while for higher earners “the marginal gain… may be generally less, even if it continues.” Going from no income to an average income, and from average to above-average, has big effects on mental health, he explained. Going from very wealthy to extremely wealthy may still help a little, but it takes a much bigger jump in money for a smaller gain.

Here my own small part of the study comes back in. The funniest part of translating the survey turned out to be the money.

The English version asked people for their household income for the year, in brackets that went all the way up to “Over $200,000”. In Georgia, nobody thinks about income per year, so we switched to monthly. And then we had to shrink the numbers. A lot. Our top box ended up as “over 4,000 GEL” a month. That’s about $17,500 a year at the 2025 average exchange rate, which would land you in the second-lowest box of the American version.

That’s not a typo. So when Kai says the first steps up the income ladder matter most, I think of those brackets. For most people back home, those are still the steps that count.

What you can, and can’t, take from the numbers

If you want to look around yourself, there’s a public Global Mental Wealth dashboard where you can explore the results by country and group. It was built by Sofia Pelica, a PhD candidate in psychology at Vrije Universiteit Amsterdam and the study’s second contact author. Ruggeri’s advice is worth reading before you do.

“Those data are meant for general patterns – definitely explore those!” he said. “But it’s most critical never to infer any specific finding.” His recommendation: “only note and take away the general, high-level patterns. But avoid any specific country or group comparisons.”

The result that surprised me most was Georgia’s, maybe because it’s my own country and I know it from the inside. In the paper, it had the highest share of people above the depression screening cutoff, 46.72%. But Ruggeri’s advice applies to me too. The sample was recruited mostly online and through the research teams’ own networks, so it isn’t representative of any country. The questions were answered in 50 languages, and a short screen flags symptoms, it doesn’t diagnose anyone.

Mental health is not just the absence of illness

My last question to Ruggeri was a bigger one: what do we, as a society, get most wrong when we talk about mental wealth and well-being? Here’s how his answer began.

“We say ‘mental health’ but we mean ‘mental illness’. Some might think that’s just semantics, but it genuinely matters,” he told me. When the conversation shifts to depression, anxiety and other conditions, “it almost universally shifts to treatment, whether it’s clinical, pharmaceutical, or institutional.” That narrows what we think we can do, and it mixes up grief tied to a sudden event with long-term conditions like clinical depression. What mental health should mean, he said, is “having a sense of purpose, having good quality relationships, feeling like what we achieve matters.”

“These cannot be solved by yoga and meditation nor by medicine (alone).” — Kai Ruggeri, Columbia University

The data shows this too. People with no signs of depression or anxiety still varied a lot in how well they were doing, and plenty of them scored fairly low.

This isn’t only about how we talk. For decades, psychology itself focused mostly on what goes wrong with people. That’s why, in 2000, Martin Seligman and Mihaly Csikszentmihalyi called for a “positive psychology”: a science of what makes life worth living.

We do the same with resilience. We tend to look for it only in the hard times, as if people need something terrible to happen before they find out how strong they are. But you don’t need a crisis to ask how quickly someone bounces back. One of the survey’s ten statements does exactly that.

Maybe the better question isn’t what people have survived. It’s what gets them out of bed on a normal Monday morning.

There’s much more in this study than I could fit here, and the final paper should be out soon. I’m sure I’ll get back to this conversation. Until then, I hope the next time someone asks how you are, you think about it for a second longer before saying “fine”. I’ll be over here, trying to do the same.

Now someone tells you that one is called “bouba” and the other is called “kiki”. Which is which?

You probably did not have to think. Bouba is the round one. In a study of people who speak 25 different languages, about 72% made the same call.

It is strange when you stop to think about it. Both words are made up. Nobody taught you what they mean. And still the answer arrives before the question is finished.

Where could the match come from?

One guess comes from 2001, when the neuroscientists V. S. Ramachandran and Edward Hubbard made the two words famous. Say “kiki” slowly. Your tongue taps the roof of your mouth in a sharp, quick way, a little like the jagged line looks.

If that is the reason, the match lives in speech, and it belongs to people. But the evidence is awkward. Three-day-old chicks made the match in a 2026 study, which ScienceBlog covered in February, and chicks do not talk. Chimpanzees and gorillas did not in a small earlier test.

So a team at the University of Parma tried an animal that lives among human speech but does not use it. They tried dogs.

What they did

The study was published in Proceedings of the Royal Society B on 14 October 2026 and tested 70 pet dogs.

Each dog faced two screens, one showing a round shape and one a spiky shape. A speaker played “bouba” or “kiki”. The owner held the dog still for 20 seconds while a camera filmed.

It is worth pausing on the word “looking”. The team did not track the dogs’ eyes. An AI program worked out which way each head was pointing, and a dog counted as looking when its head faced a screen.

For 19 of the dogs, the owners wore headphones playing rock music. They could not hear the word, so they could not nudge their dog without knowing it.

What the dogs did

Dogs looked more at the round shape when they heard “bouba”, and more at the spiky one when they heard “kiki”. The dogs with headphoned owners did the same.

The team shared its open data, and it holds a detail the paper does not spell out. In the first group of 51 dogs, every single dog spent a bigger share of its looking time on the round shape during “bouba” trials than during “kiki” trials.

Now look at how big “longer” is. The dogs only looked at the screens for about 6 of the 20 seconds. On “bouba” trials, the round shape got about 3.4 seconds and the spiky one about 2.6. On “kiki” trials it flipped, with about 2.1 seconds for round and 3.6 for spiky.

That is a shift of about one second. Small, but it went the same way for every dog.

Their first glance did not follow the sound. The paper says it went to both shapes about equally. The pattern only appeared over the full 20 seconds.

The spiky twist

The dogs also had a pull of their own. In the first group, they were quicker to look at the spiky shape, whatever sound was playing.

Here is what the Parma press release says, translated from the Italian:

• Dogs look more at the round shape for “Bouba” and at the spiky shape for “Kiki”.
• This supports the idea that the effect is part of a wider set of sound-and-sight matches shared across species, and not a purely human trait tied strictly to language.
• In children and chicks, the effect seems strongest for linking round shapes with “Bouba”. Dogs may have a different built-in leaning.
• For now, no one can say if the difference comes from the species, from each dog’s experience, or from the objects dogs usually deal with.
• Future studies will check if the match shows up early in life, how widespread it is across species, and whether living with people made it stronger.

What it does not show

This does not mean the dogs understood anything. “Bouba” and “kiki” mean nothing, and that is the point of using them.

The group was also narrow. These were medium to large dogs with medium-length snouts, tested in one lab in Parma. The paper says more studies are needed to learn how the effect begins and what it is for.

Try it yourself

You can run a small version at home, and you do not need a dog.

1. Draw one round, cloud-like shape and one sharp, spiky shape.
2. Ask a friend or family member, one at a time so nobody copies: “One of these is called bouba and the other is kiki. Which is which?”
3. Say nothing else.
4. Count how many pick round for bouba.

I tried it at a family gathering while writing this piece. I asked 27 people, and all 27 said the round shape was bouba.

That is higher than the 72% from the 25-language study, and it is worth keeping in proportion. A family party is not a lab, and I cannot rule out that people overheard each other. Your own group may split differently. The match is common, not universal.

Then think about the dog. If a chick and a dog both seem to link “bouba” with something round, maybe words did not invent the link. Maybe they borrowed one that was already there. The dogs do not prove it. They make it a little harder to dismiss.

One evening she looked over, and every spider was hanging upside down from a single thread of silk. Completely still.

“I had never seen this,” she told Science.

That evening started a six-year hunt for the answer to a simple question: do spiders sleep?

The twitching eyes

Her first clue came in 2022. Her team filmed newly hatched jumping spiderlings, which are so small and see-through that you can watch the inside of their heads. About every 17 minutes, the tubes behind their big eyes darted around for roughly 77 seconds. Their legs curled and their bodies twitched. It looked a lot like REM sleep.

But a twitchy eye is not proof of sleep. A commentary on that paper pointed out that nobody had run the usual tests. A sleeping animal lies still in a typical pose, is harder to wake, and makes up for sleep it loses.

Ten spiders, 80 eyes

The new study, published in PNAS, ran those tests on a different species, the Australian jumping spider Portia fimbriata. It grows to about 10 millimetres and hunts other spiders. Co-author Nadja Geiger calls her “famously clever” and says she sleeps upside down in a little web.

Science describes Geiger switching off the lab lights, ten spiders in front of her, each one with eight eyes staring back, before leaving the cameras to watch the night. Then came the slow part. The team had no brain recordings, only video, so it needed long recordings and huge amounts of footage to read the spiders’ behaviour.

Here is what the spiders did:

• They keep a clock. Activity dropped at nightfall, and the rhythm held without the usual light cues.
• They have two stages. A long quiet stage was broken by REM-like bouts, each about 83 seconds, with eye movements, twitching and leg curls. A spider had about 11 a night.
• They are harder to wake. After a loud sound, awake spiders reacted in about 1.7 seconds. Resting spiders took 2.7. Spiders in a REM-like bout took 4.3, and in four of the 10 tests in the open data, they did not react at all.

The night of the flies

To see if they would make up for lost sleep, the team put 25 live fruit flies into each spider’s enclosure for one night. The flies buzzed around, and the spiders sometimes tried to shake them off.

The spiders then slept about 70 minutes longer over the next 24 hours, from 13.6 to 14.8 hours. Their REM-like time rose from about 9.4 to 16.9 minutes, almost 80% more. They did it through more REM-like bouts, not longer ones. In total, 11 of the 13 spiders spent more time in the REM-like state.

The data the team posted online shows how that extra time split between day and night. Most of the extra sleep came by day, about 43 minutes, as if the spiders had a lie-in. Most of the extra REM-like time came the next night, about six of the 7.5 extra minutes.

Not the first to catch up

Lots of animals catch up on lost sleep, including fruit flies, jellyfish with no brain, and octopuses. Now we can add a jumping spider. After one bad night, it spent 80% more time in its REM-like sleep.

That makes it harder to argue that REM-like sleep needs a big brain. Nobody has counted a Portia brain, but estimates are below a million neurons. A human brain has roughly 85 billion. Columbia neuroethologist Alex Winsor, who was not part of the study, says the old view treated invertebrates as “little robots.”

How it compares with us

Adults spend 20% to 25% of the night in REM sleep. A rough sum from the paper’s numbers puts the spider at about 15 minutes of REM-like sleep (11 bouts of 83 seconds) in a night of nearly 11 hours, around 2%. That is only a rough guide, because spider REM-like sleep is judged from behaviour and ours from brain waves.

The human story started with a father and a son. In 1951, Eugene Aserinsky attached electrodes to his eight-year-old son and watched the pen jerk as the boy’s eyes moved under his lids. William Dement then showed in 1960 that people kept from dreaming made up for it later.

What we still don’t know

This was a small study. Thirteen spiders gave usable data on the flies, they came from one captive-bred stock, and the team disrupted their sleep for one night rather than depriving them of it. They also did not record brain activity, so the REM-like label rests on behaviour alone. Some spiders did not follow the trend, as two of the 13 had less REM-like time afterward.

The bigger question is the one Rößler cannot answer yet: are they dreaming? Even the definition of a dream is hard to pin down, but she wants to try. “That’s the dream,” she says.

Rößler spent a pandemic watching spiders on a windowsill and noticed they were doing something strange. Six years later, we know that spiders really do go to sleep.

Unger, a physician, decided to test the claim with the hands available to him. For 50 years, he deliberately cracked the knuckles on his left hand while leaving his right as a comparison.

When he reported the result in 1998, neither hand had developed arthritis, and he could see no difference between them. It was a satisfying answer to a family warning. Answering the same question for everyone else would take more than one unusually persistent man.

What Unger actually did

In his letter to Arthritis & Rheumatism, Unger described cracking his left-hand knuckles at least twice a day for 50 years. That added up to a minimum of 36,500 cracks, not an exact tally.

His right hand was not entirely silent. Those knuckles cracked “rarely and spontaneously,” he explained. So the comparison was between a deliberate daily habit and occasional cracking.

“There was no arthritis in either hand, and no apparent differences between the two hands,” he wrote. The letter does not say how he checked: it describes no X-ray and no formal examination.

The account was a short letter to the editor, not a full research paper. Unger paid for the experiment himself, called his result preliminary, and said a larger group would be needed to confirm it.

He also wondered whether other parental beliefs, such as the importance of eating spinach, deserved the same scrutiny. The tone was playful, but the question underneath was fair: what evidence supported a warning repeated with such confidence?

Why one man can’t settle it

The journal printed a tongue-in-cheek reply from physician Robert L. Swezey alongside Unger’s letter. Swezey had asked RAND statistician John Adams to “review” the design. The mock assessment described a two-arm trial with no randomization and an obvious problem with blinding.

“Blinding would only be possible if the investigator didn’t know left from right,” it observed. Beneath the joke was a real limitation: Unger knew which hand he had cracked, so his expectations could colour how he judged them.

Comparing his own two hands did control for age, genes and much of his lifestyle. But both hands still belonged to one person, whose experience could not stand in for people with different inherited risks, injuries or jobs.

He also chose which hand to crack rather than leaving it to chance. Hands differ in workload and injury history, so the two sides were not necessarily equal starting points.

Most importantly, repeating an action thousands of times does not create thousands of participants. Unger had 50 years of persistence, but his sample size was still one.

What the bigger studies found

In a 1975 paper, Robert Swezey and his son, Stuart, surveyed 28 older patients about whether they had cracked their knuckles. They found no link between the habit and wear-and-tear changes in the joints at the base of the fingers. This was the paper Unger cited, although his letter dated it to 1973.

A 1990 study by Jorge Castellanos and David Axelrod examined 300 people aged 45 and older, 74 of them habitual knuckle crackers. Arthritis was no more common among the crackers, but they had more hand swelling and weaker grip. They were also more likely to do manual work, smoke and drink alcohol, which makes it hard to blame cracking alone for poorer hand function.

Then, in a 2011 study, Kevin deWeber and colleagues looked back at 215 adults aged 50 to 89. All had had a hand X-ray: 135 showed hand osteoarthritis and 80 did not. Knuckle cracking was not linked to osteoarthritis, even when the researchers counted how many years people had cracked and their total lifetime exposure.

DeWeber, a sports medicine physician at PeaceHealth Southwest who led the study, told me by email that it was a retrospective case-control design. He said it was able to “harness the power of large numbers to make a more convincing argument that [knuckle cracking] wasn’t associated with hand arthritis.” It also used X-rays, which he called “the gold standard measure of arthritis,” and checked the joints of the hand one by one.

The consistent finding is narrower than “cracking can never harm you.” These studies did not find that habitual knuckle cracking makes hand osteoarthritis more likely.

What we still don’t know

DeWeber is careful not to overstate his own result. “No study has proven conclusively that [knuckle cracking] doesn’t harm joints,” he said.

He pointed to a weak spot in his own study. Everyone in it had already had a hand X-ray for some reason, and the researchers did not know what hand problems led to those X-rays. That makes it a selected clinical group, not a random slice of the population. Participants also had to recall habits stretching back decades, which a survey cannot capture as reliably as records kept along the way.

The strongest test would be a randomized trial. DeWeber noted that researchers would need to start with very young children and follow them for many years. No one has done that.

Grip strength is also unsettled. He described the results as mixed and the studies as small. The 1990 study linked cracking to weaker grip, but a study presented at the Radiological Society of North America’s 2015 meeting found no immediate difference. That study examined 40 healthy adults, 30 of them habitual crackers, and found no pain, swelling or loss of function right after cracking. It was not designed to look at decades of the habit.

There is one real, if rare, risk. Doctors have reported a small number of ligament and tendon injuries after unusually forceful attempts to crack a joint. Pulling a finger hard enough to injure it is a different question from whether everyday cracking causes osteoarthritis.

So what is the pop?

In a 2015 MRI study, Gregory Kawchuk and colleagues at the University of Alberta filmed ten finger joints from a single volunteer as they cracked. Like Unger’s experiment, it rested on one person’s hand. They saw a gas-filled cavity form in the synovial fluid, the liquid that lubricates the joint, as the joint surfaces pulled apart. Because the cavity was still visible after the crack, they argued that its formation, not its collapse, made the sound.

A 2018 mathematical model offered another explanation: a bubble that partly collapses could make the sound while leaving a cavity behind. The two accounts still compete, and neither, on its own, says whether repeated cracking harms the joint.

A warning that outlasted its evidence

The best available evidence has not found that habitual knuckle cracking causes hand osteoarthritis, which puts Unger’s personal result in line with the larger studies. His experiment is memorable because he turned an everyday warning into a comparison he kept up for half a century.

A smaller question still sits underneath the habit, and it is the one deWeber says he would most like answered. Do uncomfortable hands make people crack their knuckles, or does cracking create a feeling that only more cracking can relieve? Unger could compare his two hands, but he could not untangle that for everyone else.

It does not.

Instead, it curls the back half of its body into a loop, shoves off with its tail, and takes to the air for a split second. Then it lands and starts rolling away like a runaway wheel. Left side, right side, left side, right side. All the while, its head stays up and steady, like a tiny driver steering a tiny car.

Then it often dives under the nearest thing it can find and disappears.

Meet the dwarf reed snake of Borneo: lighter than a nickel, shorter than a school ruler, and a lot more dramatic than it looks. A team led by Yosuke Kojima of Toho University filmed it at 120 frames per second to see how it pulls this off, and published the results on 19 October in Current Biology. The fastest moment they recorded was 3.69 meters per second, about 8.3 mph. For a snake this small, that is a sprint.

How do you film a rolling snake?

Very carefully, and with sandpaper. The researchers worked with 13 dwarf reed snakes from Kubah National Park in Sarawak. They ranged from about 4 to 10 inches long, and the heaviest weighed just over 4 grams, according to the paper’s supplement.

Each snake went onto a track 2.2 meters long and 20 centimeters wide, lined with sandpaper, while two cameras recorded every move. The “danger” was a gentle tap from a human hand. No real predator was involved.

Every one of the 13 snakes rolled. Six reed snakes from a different genus, tested as a comparison, never did.

A wheel without an axle

The paper says the back of the snake rolls on the ground “like a wheel.” But a real wheel spins around an axle, and a snake does not have one.

Here is what the cameras saw instead. The snake rolls on one side of its body, then swings an arched section down past the side of its head. On the next turn it does the same thing on the other side, and the two sides take turns all the way down the track.

The strangest part is the head. It doesn’t tumble. It stays pointed where the snake is going, and in the fast rolls it is held up high and steady while the tail whips around behind it. Imagine running while your head spun like a wheel. These snakes roll and still keep their eyes on the road.

So how fast is 8.3 mph, really?

Here’s the honest version. The 8.3 mph belongs to one snake at its single fastest instant. The average of everyone’s best moments was 2.29 meters per second, or about 5.1 mph.

The “four times faster” part is also peak against peak. The paper found that top rolling speed was on average 4.0 times top slithering speed. It says only five snakes also slithered before or after rolling, so that comparison probably rests on a small group.

Now for the fun math, and this part is ours, not the researchers’. The supplement lists the study’s fastest snake, ID1, at 16.5 centimeters long. At 3.69 meters per second, that is about 22 of its own body lengths every second.

The top speed recorded in a study of wild cheetahs was 25.9 meters per second, and cheetahs in that study usually held their top speed for only a second or two. Cheetahs measure between 1.12 and 1.5 meters from nose to rump, so that works out to roughly 17 to 23 body lengths per second.

So the snake is not faster than a cheetah. A cheetah is far quicker in plain meters. But best burst against best burst, measured in body lengths, this snake lands in cheetah territory. Not bad for an animal that fits in your palm.

No hill needed

In 2023, scientists reported the first photographed cartwheel of a dwarf reed snake. That snake was seen rolling down a slope on a road in Malaysia, and the museum noted that these snakes can use downhill slopes to help them roll. That same snake, once caught and placed on flat ground beside the road, cartwheeled there too.

The lab racetrack was a flat lane, and the methods describe no slope. No hill, no gravity lending a hand, and the snakes still hit those speeds. So the roll is something the snake powers itself.

Nature almost never rolls

If rolling is so fast, why doesn’t everything do it? Stephen Jay Gould wrote an essay about it in 1981, and Richard Dawkins asked the same question in 1996. The usual answer is that a true wheel needs a part that spins freely, and a living body has to keep every part supplied with blood and nerves.

Animals that roll anyway are a short and strange list:

• Mother-of-pearl moth caterpillar: when threatened, it flips backward and rolls at about 39 centimeters per second, nearly 40 times its walking speed.
• Flic-flac spider: this Moroccan spider cartwheels at an average of about 1.7 meters per second, with a best of 2.
• Namib golden wheel spider: it rolls down sand dunes to get away from wasps, but only downhill.
• A beach-stranded mantis shrimp: it does backward somersaults, and for about 40 percent of each turn it rolls like a true wheel.
• Dwarf reed snake: about four times its slithering speed, on flat ground, with a steady head.

Don’t read this as a race. The caterpillar’s 40 times and the snake’s four times are measured against different things.

What we still don’t know

The short report doesn’t say how much energy the rolling costs or how far a snake can keep it up. It reports no wild snake using it against a real predator. And that 8.3 mph came from a lab track, not a chase.

So the snake is not a tiny cheetah. It is something odder: an animal that found a way around one of nature’s oldest rules, using nothing but a very clever curve.

Doctors, engineers and colonial officials had arrived in Vietnam with grand plans. They wanted to turn Hanoi into a showcase of French architecture, culture and modern infrastructure.

They called it their mission civilisatrice, or “civilizing mission.” The idea was that French rule would introduce the Vietnamese population to European ways of living, including modern sanitation.

At least, that was the justification.

In practice, much of this modern infrastructure was built for the French themselves, while the Vietnamese population continued living in considerably less sanitary conditions.

And one particular piece of infrastructure, intended to demonstrate the superiority of French engineering, ended up producing one of the most spectacular backfires I’ve come across.

It involved sewers, hundreds of thousands of rats, and a government policy that somehow made the problem even worse.

The French arrive with grand plans

Paul Doumer, governor-general of French Indochina from 1897 to 1902, wanted Hanoi to become a magnificent symbol of French colonial achievement.

Historian Michael G. Vann describes how Doumer envisioned a rational, orderly city, in contrast to Saigon, which he regarded as disorderly and dominated by commercial interests.

French colonial officials were convinced that their civilization, science, technology, education and administration were superior and should be introduced to the populations they ruled.

France had established colonial control over Hanoi in the 1880s, and by 1902, the French had spent years reshaping it.

Hanoi was already a centuries-old Vietnamese city, with its own architecture, traditions and established ways of living.

But that wasn’t the kind of city Doumer wanted to showcase.

He wanted European-style streets, impressive government buildings, and modern sanitation systems that would demonstrate what French engineering could accomplish.

The city itself was meant to become proof that colonialism was beneficial. With that, the French got to work.

The result was magnificent. For some.

The transformation was impressive, and you can still see much of the French architectural influence in Hanoi today.

The colonizers built broad, tree-lined boulevards, fancy European-style villas, government buildings and beautiful public spaces.

They also constructed something particularly important: an elaborate underground sewer system.

By the beginning of the twentieth century, there were more than 15 kilometers of sewer pipes beneath the European quarter.

Along the way, the French demolished or transformed parts of the existing city, including religious and other historic sites, to make room for their vision.

The new French Quarter looked splendid, and sanitation had certainly improved for the people living there.

Except those people were mostly the colonizers themselves.

Wealthy European residents enjoyed running water, flush toilets and modern drainage systems, while many Vietnamese and Chinese residents in the Old Quarter continued relying on public fountains and workers who collected human waste in buckets before dawn.

During the rainy season, drains in the older neighborhoods could back up, sending sewage into the streets where Vietnamese residents lived.

Vann, who extensively researched the Hanoi rat campaign, also found that Vietnamese taxpayers helped finance the city’s modernization, even though the improvements disproportionately benefited the European quarter.

So, yes, Hanoi had been transformed.

But the benefits were not shared equally.

The rats loved the new city, too

I’ve lived in Southeast Asia for 12 years and traveled extensively around the region, including Vietnam.

It’s not uncommon to see a rat or two sneaking around at night, particularly near markets or areas involved in the food trade. I see this in São Paulo, too, so this is by no means exclusive to Asia.

To me, it’s always seemed like a big-city problem, nothing more.

French administrators, however, had generally regarded rat infestations in Hanoi’s Vietnamese neighborhoods as evidence of poor local hygiene.

As long as the problem remained in those neighborhoods, it didn’t seem to concern the Europeans quite as much.

Without intending to, the French had created a paradise for rats.

The underground pipes offered darkness, shelter from predators, and plenty of opportunities to breed.

And while the French celebrated their modern infrastructure, the rodents discovered that it offered something else: an efficient transportation network.

The rats could travel beneath the streets and find their way into the European homes.

According to Vann, there were even reports of rats emerging through the plumbing and toilets.

And whose homes were these?

The French residents’ homes, of course.

The very people who had celebrated the new sanitation system as proof of European superiority were now discovering rats in their own bathrooms.

What had previously been dismissed as a Vietnamese problem was suddenly a European one.

The fear of plague

The timing made the infestation particularly terrifying. In 1894, Alexandre Yersin identified the bacterium responsible for bubonic plague during an outbreak in Hong Kong. Four years later, Paul-Louis Simond published research supporting the role of fleas in transmitting the disease.

These discoveries were changing how doctors understood plague, although some of the details of its transmission were still being debated.

The third plague pandemic was spreading internationally, with infections reaching new places through commercial and transportation routes.

The French authorities were worried that rats appearing in European homes could carry infected fleas and spread the deadly disease.

They would send workers underground and exterminate as many rats as possible.

But who would carry out this dangerous, unpleasant work?

Not the French residents themselves.

They hired local Vietnamese workers.

The great Hanoi rat hunt

In the spring of 1902, the colonial government sent teams of Vietnamese workers into the sewers to kill rats.

According to Vann, the racial hierarchy of French colonial society meant that dangerous, dirty manual labor was generally assigned to Vietnamese workers.

The rat catchers had to navigate narrow, dark underground passages, move through human waste, and handle wild animals that might carry disease.

Their compensation depended on the number of rats they killed.

Between 26 April and 1 May 1902, workers turned in 7,985 rats.

On 30 May alone, they killed a reported 15,041.

And on 12 June, the authorities recorded an astonishing 20,114 rats killed in a single day.

These figures come from colonial administrative records examined by Vann, who discovered the unusual rat-killing tallies while researching Hanoi’s history in the French archives.

The records counted the rats being killed, but they couldn’t tell the authorities how many more were still living beneath the city.

The sewers still provided a comfortable habitat for the surviving rodents, and the infestation continued.

The French had complaints. So did the Vietnamese.

On 9 June 1902, just three days before the extraordinary rat count, the resident-superior of Tonkin contacted the governor-general’s office about complaints from European residents.

Apparently, some French people were unhappy about the sight and smell of Vietnamese workers emerging from the sewers and walking through their neighborhoods carrying dead rats.

The Vietnamese men were doing the dangerous, filthy work of protecting European neighborhoods from disease, and some of the people being protected were complaining about having to see them.

On 16 June, Dr. Serez, the head of health services, responded that the rat catchers were no dirtier than ordinary sewer workers. He added that he had not personally received complaints about them.

The Vietnamese workers, meanwhile, were not particularly happy either.

Their work was difficult, unpleasant and potentially dangerous, and they believed their pay was inadequate.

During June and July 1902, they began organizing work stoppages and slowdowns.

The authorities eventually had to increase their commission from one cent per rat to two cents. By 1904, the rate had reached four cents per rat.

This wasn’t the public rat-tail bounty that would become famous later. It was the payment system used for the employed sewer crews.

Long before the broader population found ways to exploit the government’s bounty scheme, Vietnamese workers had already demonstrated that they could organize and bargain with their French employers.

The authorities increased the payments and continued killing rats. But the infestation wouldn’t disappear.

So they came up with another idea. Why not get the entire population involved?

The Vietnamese find a business opportunity

The French administration decided to expand its rat-killing campaign to the general public. Residents could earn one cent for every rat tail they submitted.

Officials didn’t want people bringing entire dead rats into government offices. Handling thousands of carcasses would create an enormous burden.

So they decided that tails alone would count as proof.

One tail, one supposedly dead rat, one payment.

With that, thousands of tails began arriving at municipal offices.

But then officials began noticing rats running through Hanoi’s streets without tails.

As Vann describes in his original research, some residents had realized they didn’t actually need to kill a rat to claim the reward.

They could catch it, cut off its tail, deliver the tail, collect their money, and release the animal alive.

The rat could then continue breeding.

And some people reportedly went even further.

According to Vann’s historical research, health inspectors discovered that people in villages outside Hanoi were actually raising rats for the purpose of collecting the bounty.

In a later interview with World History Connected, Vann also described a black market through which rats were brought into Hanoi from elsewhere in northern Vietnam.

The French had introduced a system that was supposed to reduce the city’s rat population. Instead, they had created an economic opportunity around the animals themselves.

There is one qualification worth noting. Vann’s 2003 paper cites a municipal report from 1930–31 for the rat-farming account, without establishing the exact date or location of the discovery. The underlying report would be needed to settle those details.

Some Vietnamese residents had responded creatively to an economic opportunity created by the very administration that considered itself more advanced.

The authorities thought they were paying for each exterminated rat, but in reality, they were paying for each rat tail.

Vann himself pointed out the irony of a colonial administration that encouraged capitalist development and entrepreneurship, only to find Vietnamese residents applying that entrepreneurial spirit to its own disastrous policy.

Faced with abuses of the bounty scheme, the French eventually abandoned it.

The final irony

In late 1902, French officials opened an elaborate colonial exhibition in Hanoi.

The exhibition, which continued into early 1903, was intended to demonstrate the achievements of French rule in Indochina.

An original report about the exhibition describes ambitions to develop commercial relations and display the region’s resources, progress and future.

At its center stood the Grand Palais, an impressive building intended to show what French architecture, engineering and administration could accomplish.

Above ground, the French were celebrating the modern city they had created.

Below ground, its sewers were still providing shelter and passageways for rats.

But there may have been another consequence of the exhibition.

Vann suggests that the arrival of visitors and cargo for the event may have helped introduce plague-infected rats into Hanoi, potentially contributing to the outbreak that followed. The specific route was not conclusively established.

In 1903, Hanoi experienced a serious outbreak of bubonic plague.

According to the colonial records examined by Vann, 159 cases were officially reported among the indigenous population, with 110 deaths. Among Europeans, six cases and two deaths were recorded.

The true Vietnamese death toll may have been higher, as some families reportedly concealed cases and deaths to avoid interference from the colonial authorities.

The bounty scheme itself cannot be shown to have caused the outbreak. But the enormous rat-killing campaign had not eliminated the threat.

And the rats continued troubling the French authorities for decades afterward.

In his research on France’s longer struggle against rats, historian Peter Soppelsa describes another rat-bounty problem reported in Paris in 1939.

I contacted Soppelsa, an associate professor at the University of Oklahoma, to ask whether French officials had been aware of Hanoi’s earlier experience.

“I have no evidence that Parisians in 1939 were aware of the earlier rat breeding in Hanoi,” he told me in an email on 21 October 2026.

But he also pointed me toward examples from Dutch colonial Java and Japan, including a Japanese newspaper article from January 1900 that, according to him, had already anticipated how people living outside cities might exploit urban rat bounties.

That warning came two years before the Hanoi campaign. I have not yet examined the original newspaper article, so its precise wording remains unverified.

Today, the Hanoi story is often used to illustrate a problem known as the cobra effect, in which a poorly designed incentive encourages the very behavior it was intended to prevent.

But I think there is something even more interesting here than an economics lesson.

The French arrived convinced they had the scientific knowledge, engineering skills and administrative power to transform Hanoi into a better city.

They rebuilt its streets, created modern infrastructure, and confidently measured their achievements.

Yet their prized sewer system helped create a new rat problem, the Vietnamese workers sent to solve it protested their conditions, and some residents discovered ways to profit from the government’s solution.

The administration could record exactly 20,114 dead rats on 12 June 1902.

What it couldn’t reliably measure was whether its grand plan was actually working.

Sources and further reading

1. Michael G. Vann (2003). Of Rats, Rice, and Race: The Great Hanoi Rat Massacre, an Episode in French Colonial History. French Colonial History, 4, 191–203. Original archival research.
2. Michael G. Vann (2018). Interview with Marc Jason Gilbert. World History Connected, 15(2).
3. Ivan Franceschini and Michael G. Vann (2020). The Great Hanoi Rat Hunt: A Conversation with Michael G. Vann. Made in China Journal.
4. Peter Soppelsa (2021). Losing France’s Imperial War on Rats. Journal of the Western Society for French History, 47.
5. T. Butler (2014). Plague History: Yersin’s Discovery of the Causative Bacterium in 1894. Clinical Microbiology and Infection, 20(3), 202–209.
6. Exposition d’Hanoï 1902–1903: Rapport général. Contemporary exhibition report, digitized by the Conservatoire national des arts et métiers.
7. Peter Soppelsa, personal email correspondence, 21 October 2026. Original reporting on rat-bounty examples in France, Dutch colonial Java, and Japan.

So I started asking. Over cups of tea and over the garden fence, I put the same question to anyone born in the 1960s and 70s who’d humour me: did your childhood make you tougher? What surprised me was how similar the answers were, from people with completely different backgrounds. Then I went through the actual research papers to see whether it holds up.

Home when the street lights come on

Nobody I spoke to described being neglected, as such. They were fed, clothed, and sent to school. They were just left to get on with it.

M.A, a close family friend, put it perfectly:

“I remember having a lot of freedom. Mum and dad pretty much left us to our own devices – come home when the street lights turn on. We didn’t have mobile phones back then, so we weren’t constantly in touch. There was a level of trust between parents and children that today’s world lacks. My parents don’t know half of the things I got up to as a teen! We made our own mistakes and we had to learn from them. I do believe that my generation were exposed to a lot more than younger people today. This is probably why a lot of us have a ‘get on with it’ attitude, although perhaps not as strict as our own parents.”

The data backs him up. A 1990 report with the rather ominous title One False Move found that in 1971, 80% of 7 and 8-year-olds went to school on their own. By 1990, it was 9%. By 2010, only 25% of primary school children were even allowed to travel home alone.

Kids born in the 60s, like my mum, made their own way to school. Someone my age was far more likely to be walked to the gate. And kids today? Many are tracked on an app from door to door.

What all that freedom taught them

The psychologist who’s argued loudest that this matters is Peter Gray of Boston College. His TEDx talk is well worth a watch:

His point is that the unsupervised part is what counts:

“Play is where children learn that they are in control of their life… Play is where they learn to solve their own problems.”

With no grown-up to referee, kids practise self-regulation and problem-solving hundreds of times over. In a 2023 paper in the Journal of Pediatrics, Gray and two colleagues argue that a decline in children’s freedom to “play, roam, and engage in other activities independent of direct oversight” is “a primary cause of the rise in mental disorders”.

A 2015 review of 18 studies on “risky play” also found links with physical activity and social skills. But – and this is the bit most write-ups skip – its authors rated that evidence “very low” quality. And Gray’s team admits most of the studies “cannot, by themselves, prove causal direction.” The pattern is there. Proof isn’t.

Emotional calluses

When I spoke to my mum, she didn’t talk about playing out. She talked about distance.

“Coming from a working-class family in the north of England, our parents didn’t talk to us much. If they did, it was practical – ‘When did you last eat?’ type of stuff – nothing to do with emotions or how we were feeling about school. They were in survival mode, they didn’t have time to pander to us kids. But in a way it also made us a lot more independent and resilient. I grew up having to sort out my own problems. I couldn’t run to mum or dad every time I faced an issue.”

In 2010, Mark Seery and colleagues published “Whatever does not kill us”, following 2,398 American adults. People who’d been through lots of adversity did worse – but so did people who’d been through none. Those with two to four adverse events came out best.

That sounds like my mum’s emotional calluses. But I don’t think it quite covers how her generation were raised:

• Childhood isn’t singled out. The events covered whole lives, and nothing measured how people were parented.
• Hardship isn’t the same as independence. Losing a loved one is a very different challenge to walking to school alone at seven.
• It relies on memory. The authors admit “recall or reporting biases” and say the design “cannot establish causation”.

So for me, it supports the idea that a life with no struggle isn’t the healthiest – not that a distant upbringing builds calluses. My mum didn’t think so either. She raised her own kids differently, “because I often felt there was a cold distance with my parents.”

“We repeat what we know”

Our next door neighbour, S.C, is about as quintessentially English as they come. Yet listen to how similar it sounds:

“I grew up in a ‘children should be seen and not heard’ type of family, and that type of upbringing stays with you for life. My parents were strict but they always made sure we had everything we needed. But in terms of being emotionally open with us, it was completely different to how parenting is done today. Arguments were had behind closed doors, and if I ever asked what was wrong, I was told it’s ‘grown-up stuff’.”

And then she said something I think most people can relate to:

“I think the lack of learning how to emotionally regulate from my parents played a part in my marriage and how I parented my own kids. We repeat what we know. Did it make us more resilient? Yes. But it also made some things harder in life, like managing relationships or setting boundaries.”

Being able to endure hard things isn’t the same as knowing what to do with your feelings – and that skill is largely learned from parents. In a 1996 study of 56 families, John Gottman and colleagues found that children whose parents practised “emotion coaching” – noticing feelings, naming them, and working through problems together – were better at calming themselves and more socially skilled by age eight. It’s a small, “preliminary” study, but it echoes S.C.

What the viral version leaves out

You’ve probably seen the claim that five to eight times as many young people today have serious anxiety or depression as in the 1950s. It comes from a 2010 study led by Jean Twenge of questionnaire scores from around 77,000 American students between 1938 and 2007. The rise is real, but a few things get lost:

• These aren’t diagnoses. The authors did consider whether people have simply become more comfortable admitting to symptoms, but when they tested it, that didn’t explain most of the rise.
• The biggest rises weren’t in depression. The scales that climbed most were hypomania, schizophrenia, and paranoia.
• The rise includes the 60s and 70s generation. Scores rose “steadily between 1938 and 2007”, so the supposedly durable generation were already scoring higher at university than students from the 1930s and 40s.

Experts don’t agree on what’s driving today’s numbers, either. Jonathan Haidt blames the switch to a “phone-based childhood”; Gray’s team finds “little support” for screens being a major cause; and Candice Odgers, reviewing Haidt’s book in Nature, warns against drawing two trend lines and assuming one caused the other.

Final thoughts

So, did the 60s and 70s accidentally produce one of the most emotionally durable generations in modern history?

What you can fairly take from this: children back then really did have far more independence, and a reasonable body of psychology suggests unsupervised play, problem-solving, and manageable hardship help build a sense of control over your own life.

What it doesn’t prove: there isn’t one single study that ranks generations on emotional durability – the case is built from several strands of research pointing the same way. The links are correlational, and “durable” isn’t the same as emotionally healthy.

That last point is what my mum, M.A, and S.C were all getting at. Every one of them made it clear things softened with their generation. My mum wanted to close that “cold distance”. S.C realised she’d carried the silence into her own marriage. Even M.A admitted his generation weren’t as strict as their parents.

Maybe the calluses were real. But the people who grew them didn’t necessarily want their kids to need them. What they seem to want – and what a lot of us in our 30s are figuring out now – is a middle ground. Enough freedom to fall over and get back up on your own, and enough warmth that you know someone will listen when you get home.

When the street lights come on, of course.

I was standing at the glass of the orangutan enclosure when one of them came right up to the window. She was watching me just as intensely as I was watching her. But what struck me was the warmth in her eyes – honey-coloured, curious, and almost knowing. I’ll never forget the expression on her face.

Monkey World was set up in 1987 to take in primates that had nowhere else to go – chimpanzees used as props by holiday photographers on Spanish beaches, monkeys from laboratories, and animals from the illegal pet trade. Several of its orangutans were smuggled out of Borneo as babies and kept as pets in Taiwan – one was found living in a cage in someone’s kitchen, another confiscated from an amusement arcade.

Even as a child, I felt sad seeing them behind glass, even knowing it was a far better life than the one they’d come from. Because somewhere in the back of my mind, I knew this wasn’t where they were supposed to be.

So when I read about a new study on how wild orangutan mothers raise their young in the forests of Borneo, I was hooked. Because it turns out those mothers go to remarkable lengths for something we’d never have guessed – making sure their babies have friends to play with.

“It just looked like a really long, hard day”

Orangutans are the most solitary of the great apes. A mother usually spends her days alone in the canopy with a single infant, and she’ll care for that one baby for years. Which is lovely for the bond – but not great if you’re a youngster who wants someone your own age to wrestle with.

Maria van Noordwijk, a Fellow at the Max Planck Institute of Animal Behavior and senior author of the new study, has spent more than two decades watching these families at the Tuanan research station in Indonesian Borneo. I asked her what first made her suspect that the days these mothers met up weren’t just chance encounters.

Part of it, she told me, is simply how much she knows about a mother’s normal routine:

“Most of the time mothers are alone with their offspring, so we have a lot of data on their ‘normal’ activity pattern. When two mothers are together they often just sit and watch while their offspring play and are ready to interfere, especially when their infants are still very young.”

And that’s what puzzled her team. Unlike most primates, orangutan mothers don’t groom each other or seem to get anything out of each other’s company:

“So we wondered why they would even stay and travel together (sometimes hours or even more than a day) once they encounter and their offspring started playing. One may expect feeding less means resting more, but instead they traveled more, thus expending more energy while having a lower intake.”

She also noticed that mothers met some neighbours far more often than others – and that this could change when one of them had a new baby. “Thus mothers do not come together with ‘friends’ or for their own benefits,” she said. They come together for the kids.

Looking back at those early observations in the Max Planck press release, she summed up what these meet-ups looked like from the mothers’ side: “It just looked like a really long, hard day for them.” The infants, meanwhile, were having a ball – orangutan kids, she says, have a “play button” that only mum can switch off when she decides it’s time to move on.

So how do you prove an ape planned it?

You can’t exactly ask an orangutan whether she meant to bump into her neighbour, so the team, led by first author Odd Jacobson, went to the data – and there’s a lot of it.

Over 15 years, researchers at Tuanan followed 31 mother-infant pairs from the nest they woke up in to the nest they built that night, racking up around 32,000 hours of observation along with GPS tracks of where they went.

When they mapped it all out, the first thing that stood out was that neighbours with babies of a similar age shared a lot more space than you’d expect – even after accounting for whether the mothers were related, or how much fruit was around. But they didn’t move house to do it. Instead of shifting or stretching their home ranges, they simply spent more time in the patches of forest they already shared with that particular neighbour. And when the team looked at the days around each play session, they found mothers heading towards their play partner’s core area beforehand, then heading back home again afterwards.

I asked van Noordwijk how confident she was that this was deliberate, rather than two hungry mothers ending up at the same tree. Her answer came down to the forest itself:

“In some other study sites females may be attracted to the same big fruiting tree and thus can rely on ‘chance’ to expose their offspring to playmates. However, in the peat swamp forest of Tuanan there are hardly any trees big or conspicuous enough to attract multiple orangutans at the same time.”

The analysis also showed the meet-ups weren’t happening at rich food sources. And once they’d found each other, she said, the mothers might travel together for a while, “visiting multiple feeding locations while their offspring play on and off.”

Her conclusion was clear: “We conclude the encounters are deliberate because of the directed travel before/toward and after/away from overlap areas, in addition to the selectivity of which neighbouring mother-offspring dyads are encountered.”

How would a solitary ape even know where to go? As Jacobson put it in the press release, orangutans “might spend most of their day alone, but they are still embedded in a social network. They know exactly who their mostly life-long neighbors are, and they know who had a baby when.”

The price of a play date

What really got me is that these meet-ups cost mothers something – and they get nothing back.

On days their infants played, mothers travelled further and spent a smaller share of their time feeding. They kept travelling more the day after, too. For an animal living in a peat swamp where food is patchy and energy is precious, that’s a genuine sacrifice.

The researchers suspect two reasons: mothers keep a watchful eye while their babies play (which distracts them from eating), and finding and staying near the right playmate means extra travel.

And that’s what makes orangutans unusual. In many animals that live in loose groups, young ones end up playing as a side-effect of their mothers gathering for food or company. Orangutan mothers, by contrast, don’t seem to get any direct benefit at all. They travel further, eat less, and change how they use their space – all, as far as the researchers can tell, for their babies’ development.

As van Noordwijk put it in the press release: “On top of providing food and safety for their young, orangutan mothers are tasked with supplying social resources too.”

I couldn’t read that without thinking about how many parents juggle exactly this – driving across town for a birthday party, arranging the school-gate play date, rearranging their week so their kids get time with their friends. Study co-author Meg Crofoot, director of the Max Planck Institute of Animal Behavior, made the same connection: “We evolved to raise children in large groups, but modern life is increasingly pushing us into a solitary lifestyle. Orangutans remind us of what we lose when we isolate ourselves.”

What the study doesn’t tell us (yet)

“Play dates” is a lovely way to put it, and the pattern is convincing. But it’s worth knowing a few things the headline version skips:

• They’re rare and short. Play happened on just 6% of the days a mother and infant were followed – roughly once every 20 days – and only for around 20 minutes in total on those days.
• It’s a pattern, not an experiment. The evidence comes from where mothers moved, not from testing their intentions. The authors themselves say future work needs to separate movement aimed at finding company from ordinary foraging more directly.
• Not every part of the trail is equally strong. The heading-towards-a-partner pattern on the day before play held up, but the extra travel on that day didn’t quite meet the researchers’ own bar for confidence. The extra travel on play days and the day after did.
• Family still matters. Infants whose mothers had no known relatives nearby were rarely seen playing at all – so it seems kin familiarity comes first, and the “planning” builds on top of it.
• We don’t know if it pays off. Nobody has yet shown that these play dates help infants survive or thrive later in life. That’s the next big question.
• It’s one population. All of this comes from Bornean orangutans at a single site.

None of that takes away from what the team found. It just means “mothers arrange play dates” is shorthand for something more precise: mothers seem to adjust where they travel to make play with suitable peers possible, and they pay for it.

The forest where it all happens is on fire

After our interview, van Noordwijk sent me a follow-up email. It wasn’t about the study. It was about the forest.

“This month, peat forest fires have reached the Tuanan study area in Central Kalimantan. So far about a third of the study area is severely affected by fires that are still ongoing.”

The long-term studies needed to understand how orangutans organise their lives, she explained, are being severely challenged by forest fires, especially in El Niño years. And this is an El Niño year.

According to CORE Borneo, the nonprofit that runs the Tuanan station with Universitas Nasional and Rutgers University, the fires have now reached camp, burning just 500 metres from the research station. Peat fires are especially hard to fight because they smoulder underground and can flare up again without warning. It isn’t the first time – fires in 2015 and 2019 burned more than 160 hectares, around 15% of the study area.

This is the forest where every one of those 32,000 hours of observations was made. It’s home to the critically endangered Bornean orangutans in this study, and to the mothers and infants who might one day help answer whether these play dates really pay off. As study co-author Sri Suci Utami Atmoko said in the press release, “only long-term studies like these can possibly unlock the answer.”

CORE Borneo has launched an emergency appeal to fund firefighting equipment, safety gear, and support for local fire crews. Co-author Prof Erin Vogel is currently at the field station and can be contacted for updates through the CORE Borneo website.

Back to the window

I keep coming back to that orangutan at the window. Many of the orangutans at Monkey World never had a mother to raise them in the wild – some were snatched as babies for the pet trade, others were rejected by their mothers in zoos. They never got the years in the canopy, the long, hard days of travelling, and the mother who’d walk an extra mile so they could find a friend.

So no, we can’t say these mothers sit and plan it the way we would. We don’t know yet whether all that effort pays off for their babies later in life, and we can’t assume every orangutan population does the same. But over 15 years in one Bornean forest, mothers with similar-aged babies kept drifting into each other’s patch, heading over before play and back home afterwards, and paying for it in miles travelled and meals missed. It’s a kind of mothering we’d barely noticed – giving their babies not just food and safety, but friendship.

And it’s all happening in a forest we’re only just beginning to understand. Right now, the place where these mothers make their quiet sacrifices is burning.

The orangutan I met as a child was safe, but she was behind glass. The ones at Tuanan are free, in the place they’re meant to be. I hope it’s still standing when the next generation of infants goes looking for someone to play with.

Florida Atlantic University (FAU) calls it the first quantified evidence that free-swimming sharks react to sound from that far out, which makes the study worth knowing about. Its headline went one step further: “Sharks Can Hear Sounds Nearly 250 Feet Away and Find the Source.” The study backs the distance. Whether sharks can pinpoint where a sound comes from is still an open question.

In a study in Integrative Organismal Biology, Caroline Sullivan, Edmund Gerstein and Stephen Kajiura of FAU filmed wild blacktip sharks off Palm Beach County, Florida, with a drone while an underwater speaker played low, pulsing sounds. A 10-kilohertz tone, outside the range at which sharks are known to hear, got no reaction at all.

Sound has long moved sharks

Scientists have known for decades that sound moves sharks. In 1963, Donald Nelson and Samuel Gruber reported in Science that sound could call them in. On reefs off Miami, large sharks came to low, pulsing sounds, mostly 20 to 60 hertz, that resembled struggling fish. Higher pulses and steady low tones did not work. A struggling fish makes noise, and a hungry shark follows it.

The abstract holds a surprise. It says the sharks “apparently detected and oriented to the sounds in the acoustic far field.” So the idea that sharks react from a distance is not new. What was missing was a measurement of how far. The FAU paper’s synopsis still says sharks are “presumed to lack the capacity” to detect such sounds beyond the near field.

Volume changes everything, though. In a 1978 study in the Straits of Florida and the Bahamas, Arthur Myrberg and colleagues watched silky sharks approach a sound source. When the level rose by about 20 decibels, especially suddenly, the sharks withdrew fast. The response faded with repeated tests, and fresh chopped fish weakened it in small sharks.

A study of lemon sharks the next year, as the FAU paper describes it, found withdrawal took a jump of at least 18 decibels above the background. A quiet pulse was a dinner bell. A sudden blast was a car alarm.

Setting off the alarm on purpose

The FAU team chose the alarm. They played 140 decibels, referenced to one micropascal at one metre, about 80 decibels above the quietest background levels at the site. The authors say they “deliberately presented sharks with a repulsive acoustic stimulus designed to evoke an immediate and robust response regardless of motivational state.”

Every winter, blacktips gather in the shallows off Palm Beach County, which makes them easy to watch from above. “Their abundance and accessibility made it possible for us to observe them from above without disturbing their natural behavior,” Kajiura said in FAU’s press release.

The team anchored a boat in 2 to 5 metres of water and let a speaker drift 13 to 19 metres behind it. A drone hovered 40 to 50 metres up. A light on the speaker’s float lit up whenever a sound began, so the video showed exactly when.

Each shark that swam into view got the same routine: five seconds of the 10-kilohertz control tone, five seconds of silence, then one of three low, pulsing sounds. A shark counted as reacting only if it changed course straight after the sound began. Using the boat’s length, 6.73 metres, as a ruler, the researchers measured how far the shark was from the speaker just before it turned.

Near and far, in plain terms

Here is the idea behind the whole study. Stand right next to a loud speaker and you feel the air shove you. Across the room, you only hear music. Underwater, scientists call the first zone the near field and the second the far field. It ends three wavelengths from the source.

Fish with a gas-filled swim bladder can pick up the far-field sound. Sharks have no swim bladder, so the usual assumption is that they notice sound only in the near field, close to the source. (The speaker comparison is ours, not the paper’s.)

Where the near field ends depends on pitch. Wavelength is the speed of sound in water, about 1,500 metres a second, divided by the frequency. For a 100-hertz hum that is 15 metres, so the near field stretches 45 metres, a bit under half a football pitch (our sum). The paper puts the far field at 45 metres for the lowest sound, 22.5 for the middle one and 11.3 for the highest.

What the sharks did

The sharks turned away in 165 of 209 tests, about 79% (our sum of the paper’s counts), and never reacted to the control tone in 209 tests.

Most of those reactions, 71.5% or 118 of the 165, came from the far field. The paper’s synopsis says 71.6% of “all responses (n = 209),” but 209 is the number of tests, not responses.

The farthest reactions to the three sounds were at 74.0, 66.1 and 61.8 metres, which is where “at least 62 metres” comes from. Lower sounds worked at lower volume: about 16 decibels above background for the lowest, about 45 for the highest.

What the headline skips

The headline makes two claims. The study supports the first, with some limits. The second is still open.

The study does not show the sharks found the sound. The authors read the turn away from the speaker as a sign they sensed direction, but the video cannot show how well any shark placed the sound.

It does not show how they hear it. “Ideally, particle motion would have been measured as well,” the authors write, “but this was not practical under the field conditions.” Kajiura said in the release that the findings suggest the sharks detect particle motion at a distance. That is not the same as showing it.

The lowest sound is the shakiest evidence. It drew the farthest reactions but has the biggest near field. By the paper’s calculation, only 33.9% of its reactions came from the far field, against 87.2% and 96.6% for the other two. That is roughly 20, 41 and 57 reactions (our sums). The paper notes the lowest sound leaked some energy below 100 hertz, stretching the near field, but argues the 400 to 800 hertz reactions cannot be explained that way.

The 74 metres is a record, not a limit. The drone’s frame stretched only about 75 to 94 metres at its longest, and sharks outside it went unfilmed. So 74 metres is the farthest reaction seen, not the edge of shark hearing.

It also covers one species on one stretch of coast, and says nothing about whether sound can keep sharks away from swimmers. The 1978 silky sharks got used to the noise, and in a 2019 study in Scientific Reports, white sharks showed no significant change with orca calls. Its authors concluded sound alone was not an effective deterrent.

How do they do it?

So how might a shark with no swim bladder notice a sound from 60 metres away? The authors point to the macula neglecta, a patch of sensitive hair cells in the inner ear. Close relatives of the blacktip have far more of these cells than other sharks, but nobody has counted them in blacktips yet.

That is the next job. “The next question is how their sensory system allows them to pick up and interpret these distant sounds,” Kajiura said in the release.

Sullivan, who did the work for her master’s degree, said this kind of work belongs in the ocean with wild sharks. In a tank, she said, sound bounces off the walls and produces “complex and confusing signals.” It is, she said, “like being in a house of mirrors.”

For now, the sharks have shown how far. They have not yet shown how.

As an adult, I got my first two cats, Athena and Ruski, and everything was different. Athena came on daily walks with me (yes, a cat on walks). Ruski sat on my lap for hours every day while I worked. We were a pack. After my separation, Ruski came with me and became something like an emotional support animal. We grieved the change together, and we moved forward together.

I’ve only had Ruski a few years compared to the many I spent with Raani, but there’s no contest about which bond runs deeper. So when a new Kyoto University study suggested it’s closeness, not years of ownership, that tracks with an owner’s mood, it hit home. But I wanted to know how it holds up against other research.

(Cat tax of Raani and Ruski at end of article). 

What the Kyoto study found

The study, published in iScience and led by cat researcher Saho Takagi, followed 29 cats and 19 owners for a month. Cats wore Bluetooth beacons that talked to their owner’s phone, both gave urine samples for oxytocin (linked to bonding) and cortisol (a stress marker), and owners filled in mood questionnaires.

According to the university, owners’ positive moods “were not connected to how long they have had their cats, but rather to the overall physical proximity between owners and their cats.” Cats’ and owners’ oxytocin moved together, and cats with higher oxytocin lived with owners who reported less negative mood.

The detail matters, though. A closer look at the methods shows:

• The beacons measured closeness, not minutes. They tracked signal strength when cat and owner were in contact, not total time together.
• It was one kind of positive mood. Closeness was linked to vigour and activity – feeling energetic and lively – not every positive mood measure.
• Years of ownership didn’t reliably predict mood either way. An early link to more negative mood disappeared once multi-cat households were accounted for.
• It’s small. Hormone data covered 24 cats, 16 of the 19 owners were women, and the university says the findings “only exhibit associations”.

So does other research back it up?

The studies that agree

Cats seem to lift us through activity, not just calm. In a 2023 study, Takumi Nagasawa and colleagues had 32 Japanese owners spend 10 minutes either interacting with their cat or resting. After resting, owners’ vitality and pleasure dropped. After time with their cat, they didn’t. Heart rate data pointed the same way, and the team suggested cats may work through an “arousing” effect rather than a purely calming one. That lines up neatly with the Kyoto vigour finding.

It’s the quality of contact that counts. A 2021 study in PeerJ measured oxytocin in 30 women before and after 15 minutes with their cats. Overall, oxytocin didn’t rise more than it did while reading a book. But it did rise more for women whose cats initiated contact, and with gentle petting and hugging – while it fell when cats got scratchy. In other words, simply being around a cat wasn’t enough. The interaction itself mattered.

Cats do bond with us. A 2019 study in Current Biology found about 65% of cats and kittens were securely attached to their owners – almost exactly the rate seen in human infants. So the aloof cat stereotype doesn’t hold for most of them.

The study that complicates it

The picture isn’t completely tidy. A 2003 Swiss study of hundreds of people found that cats reduced bad moods about as well as a human partner. But only partners, not cats, boosted good moods. And how attached owners felt to their cats didn’t explain the drop in bad mood.

That’s a different measure from Kyoto’s – self-reported attachment rather than physical closeness – but it’s a reminder that the bond and the benefit don’t always line up neatly.

Cats are not small dogs

The Kyoto study also found little link between owners’ and cats’ cortisol. That’s different to dogs: in a 2019 Swedish study of 58 dogs and their owners, long-term stress levels measured in hair rose and fell together, with researchers suggesting dogs mirror their owners’ stress.

Cats seem to work differently. They don’t necessarily absorb our stress, but they do seem tuned in to it – which is certainly how it felt with Ruski.

Final thoughts

I don’t think Raani and I failed at anything. She was who she was. We shared a home for years, but we never really shared space – not the way Ruski and I do.

What you can fairly take from this: across several small studies, close, affectionate contact with a cat – especially when the cat comes to you – is linked to better mood, more energy, and higher oxytocin. How long you’ve owned a cat doesn’t seem to be what matters.

What it doesn’t show: that cats cause better moods. These are small, mostly female samples, and the links are associations. Happier, livelier people may simply spend more time near their cats. And cats may be better at easing bad moods than creating good ones.

But if a cat has ever found you on the worst day of your year and simply sat with you, you probably already know what the researchers were picking up on. It’s not about how long you’ve had each other. It’s about how close you let each other get.

Raani in one of her usual funky poses

It’s got everything – science gone wrong, a natural disaster, and a monster ending.

The trouble is, when I went looking for that study, I couldn’t find it. There’s no record of a 1992 university project releasing eight pythons into a fenced enclosure. And the more I dug, the more I realised the real story is messier, slower, and a lot more human than the one that keeps getting passed around.

So I asked someone who spends his working life chasing these snakes. Ian Bartoszek, an environmental scientist at the Conservancy of Southwest Florida, told us at ScienceBlog exactly why he thinks the hurricane version has stuck around:

“I think the Hurricane Andrew story has persisted because it offers a simple explanation for something that was almost certainly more complicated.”

Where the hurricane story came from

There is a grain of truth in it – just not the grain in the headline.

When Hurricane Andrew hit as a Category 5 storm in August 1992, it destroyed a reptile warehouse in Homestead, just south of Miami. According to Live Science, one official later recalled seeing “hundreds of Burmese pythons in that warehouse”. So the popular version isn’t about a research study at all. It’s about a commercial reptile business.

And even that version has problems:

• The pythons were already there. Burmese pythons were spotted in the Everglades as early as 1979 – 13 years before Andrew – with unconfirmed reports throughout the 1980s.
• They were in the wrong place. As ecologist Dan Simberloff told Live Science, “When the population started to grow initially, in the 1990s, most of it was 20 miles [32 km] away from that facility.” Between 1995 and 2000, 11 pythons were found in the southwestern corner of Everglades National Park, miles from Homestead.
• The ‘tracking study’ is a mix-up. Researchers really do release tagged pythons in the Everglades – but as a modern tool for finding other snakes, not as the start of the invasion. More on that in a moment.

So where does that leave Andrew? Not off the hook entirely, as Bartoszek explained:

“Hurricane Andrew contributed additional animals to that landscape, but I don’t think the evidence supports pointing to the hurricane as the singular event that started the invasion.”

What actually happened

The real culprit is a lot less cinematic: the exotic pet trade.

Burmese pythons are small and manageable as hatchlings and enormous as adults, and for decades they were sold in huge numbers. Just how many depends on who’s counting, and over what period – which I found pretty telling in itself:

• Live Science puts it at about 17,000 imported into the US between 1970 and 1995.
• Bartoszek told us it’s around 100,000 since the 1970s, based on CITES import records.
• The US Fish and Wildlife Service says pet traders brought in “over 180,000” between 1975 and 2018.

Whichever number you go with, that’s a lot of giant snakes in a lot of homes – and some of them were always going to end up in the swamp, whether they escaped or were dumped by owners who couldn’t cope with a pet that kept on growing.

The Fish and Wildlife Service says the pythons were established “by the 1980s due to releases and escapes”. A 2011 study, cited by Live Science, concluded that “the simplest explanation” is that a few snakes were released into the southern Everglades sometime before 1985. The population then grew slowly until the 1990s – and then took off.

This is what Bartoszek meant when he told Live Science he’d “circle (D): ‘All of the Above'”. He told us he meant that literally:

“Burmese pythons were present in the pet trade in South Florida well before the Andrew event, and escapes and intentional releases almost certainly occurred over a period of years… It was a perfect storm of variables that led to the establishment of Burmese pythons in Florida.”

He also pointed out that while the first Burmese python was documented in Everglades National Park in 1979, confirmation that they were breeding came decades later. In his words, establishment “was more likely the result of multiple introductions over time, followed by successful reproduction and expansion in a landscape that proved especially suitable for this species.”

The Florida Fish and Wildlife Conservation Commission agrees: “Several introduction events likely occurred in south Florida.”

Where the ‘tracking study’ comes in

So where might the idea of researchers releasing pythons have come from? My guess is a mix-up with the work people like Bartoszek do today.

His team does release pythons into the wild – but for the opposite reason. As he explained to us:

“Our radio-tracking work involves intentionally fitting wild-caught male Burmese pythons with transmitters and releasing those tagged animals back into the established invasive population. These ‘scout snakes’ lead us to breeding aggregations and reproductive adults that can be culled from the population.”

The real prize is the big females. Bartoszek told us the largest one his team has caught was carrying up to 122 developing eggs. “Not all pythons are created equal,” he said, “and these adult females are the high value targets for management efforts.”

And the scale of that work is staggering. To date, his team has removed more than 25,000 kilos of python from a management zone of just 500 square kilometres in southwestern Florida.

But he’s clear these are not how the invasion began: “The scout snakes are research and management tools operating within an invasion that was already well established, not an introduction pathway.”

I can see how the two stories got tangled. “Scientists released pythons with trackers” is true. It just happened decades after the snakes had already made themselves at home.

The part that is true

The one bit of the viral story that does hold up is the ending. There really are tens of thousands of Burmese pythons in the Everglades – and very possibly far more. Nobody knows the true number, but estimates suggest there could be hundreds of thousands, and a 2023 report concluded that wiping them out isn’t possible with the tools we have.

Around 15,000 have been reported and removed since 2000, according to the Fish and Wildlife Service. That sounds like a lot until you realise it barely dents the population.

And the damage is real. In Everglades National Park, observations of some mammal species dropped by 95% between 1996 and 2016 after the pythons moved in. Animals that used to be everywhere became a rare sight. The problem got serious enough that in 2012, the US banned importing Burmese pythons under the Lacey Act.

Which makes the origin story matter more than it might seem. If you think a hurricane did this, it’s a freak accident. If you know it was the pet trade, it’s a lesson.

Final thoughts

I get the appeal of the hurricane version. It’s neat. It has a beginning, a villain, and a twist. But as Bartoszek put it:

“The bigger lesson is that biological invasions rarely have the movie-ready origin story we would like them to have. With Burmese pythons, there were likely multiple opportunities for animals to enter the wild population.”

What you can fairly take from this: Burmese pythons now number at least in the tens of thousands across the Everglades, and they’ve devastated native mammals. The invasion came from the exotic pet trade, through escapes and releases, most likely starting with a small group of snakes before the mid-1980s. Hurricane Andrew probably added some more to the mix.

What it doesn’t show: that eight pythons were released for a 1992 university tracking study (there’s no record of one), that they escaped a fenced research enclosure (the story is really about a commercial warehouse), or that Andrew started the invasion (pythons were in the Everglades by 1979).

And that last point is the one that stuck with me most. Blaming a hurricane is comfortable – nobody’s responsible for the weather. Bartoszek didn’t mince his words on that:

“Hurricane Andrew is part of that history, but treating it as the beginning of the invasion gives the people responsible for their establishment an easier way out to blame a natural disaster.”

The Everglades python problem wasn’t an act of nature. It was decades of people buying, keeping, and letting go of animals they couldn’t handle. Which is a much less exciting story – but it’s the one we can actually learn from.

The answer, published on 25 September in Animal Cognition, is that they appear to be. The elephants looked at their rival before gesturing. They waited to see what happened. And when nothing happened, they gestured again, or tried a different gesture.

Then came the twist. When the team worked out what the elephants wanted, it came down to two messages: “move away” and “stop what you are doing”. Those are the same two ways of saying no that wild chimpanzees use.

The team logged 844 gestures across 261 exchanges, all of them conflicts. Where the other elephant gave in, the signaller wanted it gone in 134 of 168 cases, and wanted a behaviour to stop in the other 34.

How do you prove an elephant means it?

Elephants cannot say what they mean, so the team borrowed a test from ape research. The logic is simple. A reflex, like a yelp, goes off whether or not anyone is listening. A deliberate message is aimed, the way you make sure someone is looking before you wave. So an animal gesturing on purpose should check that someone is looking, pick a gesture to suit what the audience can see, wait for a reaction, and try again if nothing happens.

The elephants aced it almost every time. Before gesturing, they were looking at the other elephant in 522 of 539 cases (97%). After gesturing, they paused for at least a second to watch for a response in 477 of 531 cases (90%). When nothing came of it, they carried on in 334 of 404 cases (83%), and that counts switching to a different gesture, not only repeating the first. It is the elephant version of saying something again, then rewording it when the other person still does not get the hint.

The audience test was the shakiest, and it is where the first chimpanzee echo appears. Catherine Hobaiter of the University of St Andrews, a co-author of the elephant paper, found in 2011 that wild chimpanzees use silent gestures for attentive targets and contact gestures for inattentive ones. Elephants did something similar. When the target was not looking, they leaned toward gestures that can be heard or felt, such as ear flaps and trunk pushes, over silent visual ones. That held across the full dataset, but not when the team looked only at elephants it could identify individually.

Forty-one gestures, two ways of saying no

Across the footage, the team saw 68 types of gesture. After dropping seven that never got a reaction and 20 seen fewer than three times, 41 were left, built from 32 basic movements. The trunk and ears did most of the work, accounting for 47% and 37% of gestures. Ear flaps, which make a sound, were logged 127 times, trunk flings 84 times and trunk pushes 56.

Forty-one gestures for two messages sounds like overkill. Of the 33 types seen often enough in successful exchanges, 21 were used for both. Hobaiter put it in human terms in an email to ScienceBlog. She wrote that “you need a lot of different ways to say something ‘negative’. Just as we might say ‘Stop that’ very differently to our boss or our little brother.”

She has seen this pattern before. In 2014, she and Richard Byrne reported in Current Biology that wild chimpanzees use 66 gestures for 19 meanings. Seventeen of those meanings invite contact, such as “groom me” or “move closer”. Only two push the other animal away, “stop that” and “move away”, and chimpanzees use those two broadly, across many contexts, to negate a wide range of behaviour.

The elephants’ two goals are the same pair. Animals separated by an enormous stretch of evolutionary time faced the same problem, how to tell someone no, and landed on the same two answers. “Finding the same in elephants is particularly exciting because elephants are so evolutionarily distant from primates,” Eleuteri said in the release. Hobaiter calls it “a strong case for convergent evolution: in essence, they are using the same tools because they’re solving similar communication problems.”

Because Hobaiter is part of the elephant team, her comments are the authors’ view, not an outside one.

Ninety-one hours, mostly empty footage

The work behind the numbers was slow and patient. Eleuteri did the filming herself, following elephants along a river in Marakele and at the main waterholes in Addo. The camera’s pre-record function caught the opening seconds of each encounter. Of the 2,469 clips she collected, only 127 (5.1%, our sum) held gestures clear enough to code, which is about nine coded gestures per hour of footage (our sum). That is a lot of waiting for a few seconds of drama.

The parks told different stories. Addo, a fenced park of about 700 elephants with no natural year-round water, saw competition over water. Marakele, with about 400 elephants and a river that runs all year, saw competition over reeds. Conflict showed up in 321 of 993 clips at Addo (32.3%) but only 101 of 1,476 at Marakele (6.8%). The filming spots differed, so the comparison is rough, and the authors cannot rule out that a few aggressive elephants inflated Addo’s share. Even so, the two populations shared 26 of their 41 gesture types.

What the study cannot say

The authors call it a first step. The team spent one month at Addo and three at Marakele, and it filmed only conflicts. That matters for the chimpanzee comparison, because the elephants could only show their “no” messages. Chimpanzees have 17 friendlier ones, and nobody yet knows what elephants gesture for when they greet or play. The 80% “move away” figure also covers only the 168 exchanges where the reaction let the researchers judge the goal, not all 261. In Marakele, males made 178 of the gestures and females 48.

The biggest limit is the phrase “on purpose”. Nobody can see inside an elephant’s head. The team judged each goal from what happened next: the other elephant left, and the signaller stopped. Hobaiter was careful about how far to go. The study, she wrote, “doesn’t (yet!) test the question of whether elephants have what we’d call a ‘Theory of mind’.” The paper draws the same line. These elephants appear to aim for a reaction, but there is still no conclusive evidence that any animal tries to change what another thinks.

Why it still matters

The finding is not only a curiosity, because people can use it. As humans and elephants meet more often, Lucy Bates of the University of Portsmouth, a co-author, said in the release that understanding how elephants communicate when they feel threatened “could help us better read their intentions or recognise when they don’t want us around, giving us the chance to move away before interactions become dangerous.”

Eleuteri’s hunch appears to hold up. When an elephant flings its trunk or spreads its ears, it is sending a message, and most of the time that message is “move away”.

The footage shows how that pump runs once a sting is under way. How a sting begins is a separate question, and it is still open.

Two pistons taking turns

A honey bee stinger contains two slender lancets that slide back and forth along interlocking rails. Each carries a flexible valve inside an enlarged chamber called the bulb, which fills with venom from the venom sac. The same strokes that work the stinger deeper also drive the pump.

On the downstroke, towards the tip, each valve inflates against the chamber wall and pushes venom towards the exit. Macquarie University’s press release likens this to a parachute opening. On the upstroke the valve deflates so the chamber can refill, and that collapse may reduce how much venom is drawn backwards.

Because the two valves move in opposite phases, one is pushing while the other returns. Midway through their strokes they deform again to squeeze past each other, which keeps venom moving while the chamber refills.

Dr Sridhar Ravi, the project lead at UNSW Canberra, compares the design with a syringe in the UNSW release. A single plunger pushes fluid out when advanced but can draw it back when withdrawn. Two collapsible pistons working in opposition avoid that simple reversal.

Honey bees are not the only insects that pump venom this way. Ants also have paired valves on their lancets, while most wasps squeeze venom out with muscle around the venom sac instead, according to a 2026 study in the Journal of Morphology. What sets the honey bee apart is a stinger that keeps working after it leaves the body, powered by its own muscles and venom supply.

Footage from the study (Ramirez-Esquivel et al., 2026) can be watched on YouTube.

Watching the pump from the inside

The researchers pulled stingers from live worker bees with forceps, AAP reported, and set each one partway into the transparent gel. From there, the stinger drove itself deeper as its lancets alternated, giving the team a clear view of machinery normally hidden inside the bee and the tissue being stung.

They filmed at 100 and 300 frames per second in the gel, and at 500 frames per second without it. Honey bee stingers suit this kind of work because they are large and translucent enough to see into, and they detach from the bee on their own.

Six stingers, each from a different worker, were tracked through their full pumping sequence, with both valves followed for as long as they stayed visible, according to the study’s supplementary information.

Micro-CT scans gave the team a three-dimensional reconstruction of the internal parts. From those scans they built a scaled-up, 3D-printed working model to test how the pump should behave.

The valves themselves are not a new discovery. Ramirez-Esquivel told New Atlas that scientists have known about them since as early as 1910, “but until now we didn’t know exactly how they work.”

What isn’t settled yet

Even within the filmed cycle, one detail is unresolved. The team identified two possible ways the valves could change shape as they cross, and more work is needed to tell which is right, New Atlas reported.

The materials are another open question. According to Ars Technica’s report on the study, the team’s artificial valve only reproduced the right deformations when it was highly elastic and slid with little friction, which makes the properties of the real valves a natural next target.

The Journal of Morphology study raises a question about timing. Working on the ant Diacamma nr. indicum, its authors froze workers at the start of a sting and found both valves pushed forward together rather than alternating. They propose a cycle that begins with both lancets moving in step and shifts to alternating strokes on the return, and they say this challenges the strictly alternating model built on honey bees. Their evidence comes from stingers fixed in different positions, not from video of a live sting.

Alexandre Casadei-Ferreira of the University of Greifswald, first author of the ant study, told ScienceBlog: “Placing the stinger into the substrate, as in the PNAS videos and in most commonly used puncture methods, may bypass the initiation of stinging, which is precisely where I would expect an in-phase stage. Their results provide strong evidence for antiphase pumping once stinging is underway in honey bees, but whether an in-phase stage occurs at initiation is hard to say without an experiment designed specifically to capture that first moment.”

Whether bees share that opening stage, or it turns out to be particular to ants, has yet to be tested.

Why the pumping matters

For anyone with a stinger left in their skin, the practical point is speed. The UNSW release reports strong, regular pumping for 15 to 70 seconds, with twitching continuing for minutes. The study measured mechanics, not venom dose, so it cannot say how much venom enters a person in that time.

A 2020 systematic review in Cureus, based on two eligible studies, concluded that a retained stinger should be removed quickly and that pinching and pulling it out appeared to offer no disadvantage. On that limited evidence, there is no reason to delay while looking for something to scrape it off with.

Engineers may take note too. Two flexible pistons sharing one small chamber could inspire micropumps for moving liquids through tiny channels. Ramirez-Esquivel told AAP that people had suggested the design could help with “something like an insulin pump”, though that is an idea rather than a result of this study.

The next experiment would start earlier than these recordings, at the first moment of a sting. Filming that moment in bees as well as ants would show whether lancets that start in step give way to alternating strokes.

The released mosquitoes are males, so they do not bite, and the wild females they mate with lay eggs that never hatch. In a trial across eight clusters of homes, researchers estimated that the releases averted 2,351 cases of dengue, according to a paper in eBioMedicine published in September.

Nobody can count dengue cases that never happened, though, so 2,351 is not a headcount. It comes from comparing areas with the mosquitoes to areas without them. The paper’s own tables show which parts of that comparison hold up and which are weaker.

The trick: eggs that never hatch

The released males carry Wolbachia, a bacterium that, according to Singapore’s National Environment Agency (NEA), lives naturally in more than 60% of insect species. In a male mosquito, it changes the sperm so that it only works with eggs carrying the same bacterium.

Zhiyong Xi, a researcher at Michigan State University, explained the rest in an email to ScienceBlog. Wild females do not carry Wolbachia. So when one mates with a released male, her eggs “die or will not hatch.” Do that twice a week and the mosquitoes thin out. (Xi’s lab supplied the original mosquito line used in the trial, according to the paper’s acknowledgements.)

I cannot tell which of the mosquitoes I see are the released ones, just by looking. But NEA warns residents that they may notice more mosquitoes on release days, males that do not bite but may be drawn to people while searching for females. That would explain what I have been seeing.

Two years, 15 neighbourhoods

The researchers picked 15 clusters of high-rise housing and, by chance, gave mosquitoes to eight of them. The other seven were left as a comparison. Everyone kept the usual mosquito control. For about two years, males were let loose twice a week, on the ground, middle and upper floors of the buildings.

It worked on the mosquitoes first. Traps in the release clusters caught about a seventh as many adult females as traps elsewhere. Then dengue cases fell too. Over the trial, the release clusters had roughly a third as many cases as the others. Before the releases, the two groups had looked almost identical.

The number that is not a count

So how do you count cases that never happened? The authors took the gap between the two groups and applied it to everyone living in the release clusters. That gives 2,351 fewer cases than there would otherwise have been. Those clusters actually reported 1,037, which means the releases cut dengue by about 70%. (That last sum is ours, not the paper’s.)

Hannah Clapham, a researcher at the National University of Singapore, said in an email that Singapore has had excellent routine vector control for decades, yet dengue is still a problem. “In this trial we have seen that Wolbachia has large benefits on top of routine vector control,” she wrote. She called it “a really excellent study that has given us invaluable information about the possible impact of Wolbachia.”

Still, hold the number loosely. About 420 of those cases come from the first six months, before the paper says the benefit was clear. Everyone knew which areas got the mosquitoes. Cases are counted by home address, and only if someone gets tested. And 15 clusters is not many, so luck may have played a part.

It’s kind of fun for me to be a data point in a study like this.

A second paper, and a quiet year

A separate paper on the same trial, in the New England Journal of Medicine, measured something different: among people tested for dengue, far fewer in release areas turned out to have it. It puts the protection for an individual person at 71% to 72%, close to the 70% we got from the first paper’s numbers.

But the benefit was not steady. In 2023, a year with little dengue, protection dipped. For one stretch the cut was only about a quarter, too uncertain to tell apart from no effect. In the next stretch it was back above 80%.

The authors say the protection weakened in that quiet spell. Clapham said the reason is “interesting and worth exploring,” but that she does not think it is “something to be concerned about.” She gave three reasons. Among “linked” cases, those close together in place and time, the benefit held even when cases were low. Mosquito numbers stayed “vastly different” between the two groups. And the benefit was still there when the quiet spell ended.

Is it safe?

NEA says the mosquitoes are not genetically modified, and the males neither bite nor spread disease.

Sorting males from females is not perfect, so a few females slip into each batch. Xi said the mosquitoes are irradiated to make those females sterile, so no Wolbachia-carrying offspring are born in the wild. NEA says that is what keeps Wolbachia from establishing itself in the wild mosquito population.

Another Wolbachia method works the opposite way. It releases males and females so the bacterium spreads through the local mosquitoes, and a 2021 trial in Yogyakarta, Indonesia found it protected people there. NEA sticks with suppression, which does not need biting females released at all. I am happy to be spared that.

So why is there still dengue?

Two reasons. The trial covered only part of the country, and NEA says it will reach more than 800,000 households, about half the total, by the end of 2026. And the method only works on one species, Aedes aegypti, so other mosquitoes that can spread dengue are untouched. NEA says clearing standing water is still the cornerstone.

Clapham said the next step is rolling the method out to as much of Singapore as possible. Vaccines, she wrote, “work in different ways to protect the population,” and the big question is how to use the two together.

Could anyone else copy this?

Making enough males used to be the hard part. Telling males from females was “a major bottleneck,” according to a 2024 Science Robotics paper from Xi’s group, which describes a machine sorter that lets one person separate 16 million males a week. Xi said sorting is no longer a bottleneck. But that paper’s field test was in Guangzhou, China, on a different species, the Asian tiger mosquito, so it says little about Singapore’s own production. The eBioMedicine paper gives no cost or weekly production numbers either, so other cities cannot yet price the idea.

Clapham was cautious about copying it elsewhere. “Globally this study adds to the body of evidence that Wolbachia is an effective control measure,” she wrote, but it “will however require much thought about what approach and how it could be used in other places.” Even so, she believes it “should be considered, alongside the vaccine, in all countries suffering from dengue.”

From the ground

What strikes me, living inside the experiment, is how little of the result I can see. The mosquitoes I notice are the ones that do nothing. The success is everything that does not happen: the bite that did not come, the fever that skipped someone. The 2,351 is made of days like that, spread over many thousands of people, and no one in it will ever know which of them was spared.

Maybe that is why the number needs a careful read. From the ground, a good result looks like an ordinary week, with a few extra mosquitoes drifting about and nothing else happening.

A 2026 systematic review of 70 grazing studies, led by Jonathan Sanderman, vice president of science at the Woodwell Climate Research Center, found no change in soil carbon across the 10 studies that met all of its quality criteria. The review appeared in Communications Earth & Environment on 9 July 2026.

Teague’s study was published in Agriculture, Ecosystems and Environment in 2011. The two papers answer slightly different questions. One compared nine working ranches at a single point in time. The other pooled published studies and sorted them by how well each could support a cause-and-effect claim.

Whether rapid rotation adds carbon to rangeland soil is still not known.

At stake is how to read that gap. Ranchers who rotate cattle quickly through many small paddocks say their land comes back, and Teague’s ranch-by-ranch comparisons have stood for 15 years as the scientific backbone of that claim, so which kind of evidence carries the argument decides whether regenerative grazing is treated as a land-repair tool, a climate tool, or neither.

In an email on 9 October 2026, Sanderman said: “One thing I am confident about is the amount of sequestration that ranchers can expect is likely lower than many people are saying or hoping.”

Multi-paddock grazing splits a ranch into many fenced paddocks and moves cattle through them in turn. On Teague’s ranches the herd stayed one to three days in each paddock, and ranches had 10 to 41 paddocks. Grass rested for about 40 days in fast growth and 80 days in slow growth before cattle returned.

Heavy continuous grazing leaves cattle in one large pasture all season at a high stocking rate. The multi-paddock and heavy continuous ranches in Teague’s study carried the same stocking rate, 27 animal units per 100 hectares.

The study compared three neighbouring ranches in each of Cooke, Parker and Jack counties, where management had been constant for at least nine years. Multi-paddock ranches had 1 per cent bare ground against 30 per cent under heavy continuous grazing. Tall grasses made up 45 per cent of plant biomass against 7 per cent.

Standing crop was 4,680 kilograms per hectare against 2,696. Sediment loss was four grams per square metre against 18. The ratio of fungi to bacteria in the soil was 3.1 against 1.2, which the paper says indicates better water holding and nutrient retention.

Soil organic matter was measured by loss on ignition, which weighs a soil sample before and after burning off its organic material. The method measures organic matter. It does not measure organic carbon directly. The difference between grazing systems was significant only in the top 300 millimetres of soil, where multi-paddock soils ran 5.72 per cent against 3.76 per cent in the top 15 centimetres.

Sanderman’s review searched the literature to June 2023 and found 70 studies with direct soil carbon measurements comparing continuous or season-long grazing with an alternative. Forty-seven failed more than one quality criterion. The most common failure was a control site that could not be shown to represent the management on the treatment site before the switch.

The 10 primary studies, with 27 observations, showed a sequestration rate of 0.03 tonnes of carbon per hectare per year, with a p value of 0.90. A secondary group of 13 studies with 25 observations showed gains. When all retained data were pooled, the mean rate was 0.29 tonnes, with a p value of 0.08.

The review’s own data sheet graded Teague’s 2011 study as secondary. Pairings of neighbouring ranches could not show that the continuously grazed control represented what the multi-paddock land had been before. The review did not re-measure his ranches. The Environmental Defense Fund funded the review, and one co-author works at the fund.

The paper gives an example of the problem from another study in its set. One alternative grazing site had been converted from cropland 29 years before sampling, while its continuous grazing control had been converted 17 years before. The difference in carbon stocks in that pairing may simply be due to the extra 12 years under perennial pasture, the paper says.

A five-year experiment in Colorado ran the comparison as a randomised trial at ranch scale. David Augustine, an ecologist with the USDA Agricultural Research Service in Fort Collins, Colorado, and colleagues paired 20 pastures of 130 hectares each into 10 blocks at the Central Plains Experimental Range in northeastern Colorado. Pastures in each pair were matched on soil, plants, terrain and past management.

One pasture in each pair was assigned at random to adaptive multi-paddock rotation. The other stayed under season-long continuous grazing. From 2014 to 2018 the team found little evidence of any effect on the abundance or production of perennial cool-season grasses, in wet years or dry. Cattle in the rotational pastures gained 12 to 16 per cent less total weight each year.

The Colorado study appeared in Rangeland Ecology & Management in 2020. It did not measure soil carbon.

Each line of evidence has holes of its own. Teague’s comparison had no before-and-after baseline and did not record cattle weight gain or profit. The Colorado trial measured no soil carbon. The review’s primary set held only 27 observations, and sorting them by study design, grazing style, aridity or grassland type showed no pattern.

Sanderman said the absence of a signal in those 10 studies “doesn’t mean that sequestration isn’t happening with adoption of regenerative grazing.” The larger problem, he said, is that “the research community hasn’t done a good job studying this question.” “It is really hard to do research at the scale of commercial grazing enterprises,” he added.

Asked about reports that land visibly improves under rotation, Sanderman said: “I’m not sure what land visibly improving means.” Trends above ground do not always match trends below ground, he said, because not all feedbacks are positive. Many ranchers who adopt regenerative practices “increase herd size which ends up removing more biomass even if the forage is becoming more productive,” he said.

The review’s literature search closed in June 2023, so grazing studies published after that date are not in its dataset.

The paper, published in Nature Neuroscience in May 2024, gives the raw figures in its Extended Data Table 1. At the dye’s peak, mice kept awake had cleared 69.7% of it and sleeping mice 52.8%. Eight mice were tested, each both awake and asleep.

That is a drop of 16.9 percentage points, or about a quarter measured against the awake figure. The 30% does not seem to appear anywhere in the paper. Franks’s 20% and the table’s quarter are close, though it is not clear how he arrived at his figure.

All three numbers agree on the direction, which is that less dye was cleared during sleep in these mice. That runs against the idea that sleep washes waste out of the brain, including the amyloid-beta that builds up in Alzheimer’s disease. Underneath the percentages sits a dispute between two labs over what counts as clearance.

The idea Franks disputes comes from Maiken Nedergaard’s lab at the University of Rochester, which proposed that cerebrospinal fluid, or CSF, recirculates through brain tissue along paths that follow blood vessels, exchanging with the fluid between cells. The brain has no conventional lymphatic system, and in 2012 the group named this the glymphatic system.

In 2013, Lulu Xie and colleagues in that lab reported in Science that a tracer injected into the CSF of awake mice got into the brain about 95% less than in sleeping mice over half an hour of imaging. The space between brain cells grew by 60% in sleep.

When the team injected amyloid-beta straight into the cortex, the sleeping mice cleared it twice as fast. Rochester’s own press release went further than the paper, saying the glymphatic system was “almost 10-fold more active during sleep”, a figure that, as far as can be told from the paper, is not a clearance measurement.

Franks’s objection is to how the earlier work got from tracer going in to waste coming out. It “was largely based on an assumption that increased tracer entry into the brain could be used as a surrogate measurement of brain clearance,” he said in an email. “I consider this a very dubious assumption.”

His team measured something else. Working in male mice, they injected a small fluorescent dye into a deep part of the brain, the caudate-putamen, and used an optical fibre in the frontal cortex to track how much of it arrived over the following hours, with dye lost along the way counted as cleared.

More dye arrived in the sleeping mice, which meant less had been removed, and under three anaesthetics the drop was bigger still. The authors concluded in the paper that their results “challenge the idea that the core function of sleep is to clear toxins from the brain.”

Nedergaard’s group says that design cannot show clearance at all. In a Matters Arising published in Nature Neuroscience in March 2025, Virginia Plá, Erik Kroesbergen, Nedergaard and colleagues said that dye moving from one part of the brain to another is redistribution, and that “Waste must be flushed out of the brain to be cleared.”

They also said the sleep recordings captured rebound sleep after the mice had been kept awake, that the anaesthetised mice spent most of the recordings recovering from a single dose, that the dye crosses the blood-brain barrier, and that one of the injections was too large for the tissue receiving it.

A reply from Franks and his Imperial colleague William Wisden, published the same day, called those objections “either misunderstandings, misleading, or simply incorrect”. The clearance experiments used 0.5 microlitres injected slowly, they said, and the larger volume came from separate diffusion experiments.

The 2024 authors say their result is for “a small dye that can freely move in extracellular space,” and that larger molecules may behave differently. The 2013 amyloid result came from a different experiment, with amyloid-beta itself injected into the cortex, so the two papers need not be read as one measurement giving two answers.

In January 2025, Natalie Hauglund and colleagues in Nedergaard’s lab reported in Cell that during non-REM sleep in mice, slow waves of the brain chemical norepinephrine make blood vessels tighten and relax in rhythm, pumping CSF into the brain. The sleep drug zolpidem dampened both the waves and glymphatic clearance.

The evidence in people is much thinner. In a 2021 Oslo study led by Per Kristian Eide, in which seven people stayed awake and 17 slept, one night without sleep impaired clearance of an MRI tracer from most brain regions, and later sleep did not make up for it.

A 2026 trial of 39 participants, with Jeffrey Iliff among the authors, reported higher morning blood levels of Alzheimer’s markers after a normal night’s sleep than after a night awake. The team attributed the difference to glymphatic clearance.

In mice, the two labs still disagree about how to measure clearance at all. Both human studies were small and in patients being assessed for suspected problems with CSF, and several of the 2026 authors, Iliff included, declare financial interests.

The disease is called Buruli ulcer, and in the Australian state of Victoria it is now reaching into the inner suburbs of Melbourne.

A sore that starts with a bite

It is caused by a bacterium, Mycobacterium ulcerans, a relative of the germs behind tuberculosis and leprosy. At least 33 countries have reported it, according to the World Health Organization, across Africa, South America and the Western Pacific. In Australia, it has surged in Victoria.

The state recorded 5.8 cases per 100,000 people in 2024 and now routinely sees more than 350 cases a year. In one coastal town, the yearly rate has hit 770 per 100,000, or roughly one resident in every 130.

For decades, nobody could say how people caught it. In Victoria the bacteria live in native possums, which develop ulcers of their own and shed the bacteria in their droppings. A 2024 study pinned down the missing link. Researchers found matching bacterial genomes in mosquitoes, possum droppings and human patients on the Mornington Peninsula, and concluded that Aedes notoscriptus, a mosquito the University of Melbourne describes as a common backyard species, probably carries the infection from possums to people.

That raised an obvious question. If mosquitoes spread it, would killing them stop it?

Traps the mosquitoes take home

To find out, researchers from the University of Melbourne and the Doherty Institute picked a 50 square kilometre patch of Melbourne’s inner north-west where Buruli ulcer was already spreading. The suburbs included Brunswick West, Essendon and Moonee Ponds. In a randomised trial, published in Nature Microbiology, they matched 12 areas of about 80 houses each, and a random number generator split them into six treatment areas and six controls.

Between 25 January and 21 March 2024, the team set out 593 In2Care stations, a commercial mosquito trap, in the treatment areas. That’s roughly 100 per area, or about one every 400 square metres. They went into shaded spots in residents’ yards, on nature strips and on other public land.

Each station is a small tub of water that a mosquito visits to lay eggs. Inside, a gauze strip is dusted with two things. One is pyriproxyfen, a larvicide that stops young mosquitoes from developing into adults. The other is Beauveria bassiana, a fungus that infects insects and kills the adult mosquito within about 8 to 10 days.

The clever part is what happens after the mosquito leaves. Coated in larvicide, it flies on to other places it would breed, such as buckets and pot-plant saucers, and contaminates those too. The mosquitoes do the spraying for you, including in spots no one would think to check.

A drop the weather can’t fully explain

To count mosquitoes, the team set out 480 egg traps, simple buckets of water baited with alfalfa pellets, before and after the trial. In the three weeks after the stations came out, the treatment areas had about 70% fewer eggs than the control areas. Statistically, it was a clear result (P = 0.0076), and the drop was still there two weeks after the stations were removed.

However, a dry spell complicated things. Rain was a strong predictor of egg numbers, and very little fell during the trial, so eggs dropped everywhere. In the ten traps that started with the most eggs, counts fell from an average of 672 to 21 in treatment areas, and from 491 to 34 in the controls.

This is why the control areas matter. “There was definitely a drop in mosquito egg counts across all 12 areas due to the change in climatic conditions from very hot and wet to very hot and dry,” Véronique Paris, a medical entomologist at the University of Melbourne and the Doherty Institute and one of the study’s two first authors, said in an email to ScienceBlog. “However, the intervention areas still had a significantly greater drop in egg counts than the control areas.”

The paper found the same gap even among those hotspot traps. On mosquitoes, the stations worked.

But this hasn’t been the case everywhere they’ve been tried. The paper notes that the same stations failed to bring down mosquito numbers in a Florida deployment, where too few were used, and in a Hawaii trial, where there were probably too many other places for mosquitoes to breed.

Too few cases to be sure

Checking the disease was much harder. Buruli ulcer takes about 4.8 months on average to show up, so someone bitten in summer usually doesn’t notice a sore until winter. To tie cases to the trial, the researchers had to count backwards from when each patient first noticed symptoms to roughly when they were bitten.

The trial also wasn’t set up for this. It was designed to count mosquitoes, and the analysis of human cases was only worked out afterwards.

Over the eight weeks the stations were out, the researchers found no clear difference in the risk of catching Buruli ulcer. A wider count, covering infections from January to mid-May 2024, found 7 cases in treated areas and 9 in untreated ones, which the authors themselves call “not different.”

The 1-to-6 figure comes from a single day. The researchers expected the stations to work best around 21 March 2024, the last day of the eight-week deployment. Among infections traced back to that day, the treated areas had 1 case and the untreated areas had 6, an 83% drop. (The paper also highlights 8 March, when infections traced back to that day numbered zero in treated areas against 6 in untreated ones, but its summary and statistics use the 21 March figures of 1 and 6, and so does this article.)

Even that is shaky. Seven cases is too few to be sure of anything. The result only just clears the usual bar for statistical significance (P = 0.047), and the likely range of the true effect runs from a 97% drop in cases to a slight rise. That leaves room for the stations having had no effect at all.

In the stations’ favour, the same pattern didn’t show up over the same weeks in 2023 or 2025, when no stations were out. Still, the authors say a longer trial across more areas is needed to know whether the stations really prevent cases.

Put plainly, the stations cut mosquitoes. Whether they also prevented infections is still only a promising hint, not a finding.

Risks to other wildlife

If the stations do turn out to protect people, the next question is what a city-wide rollout would cost the rest of the garden.

“Both bioactives can have off-target effects,” Paris said. But putting them inside the stations, she added, “is quite targeted, as the stations are not attractive to many animals other than mosquitoes.”

The real test is where the larvicide ends up once it leaves the station. Paris said mosquitoes are unlikely to deliver enough to affect a pond. It is more likely to reach the spots where they breed, which she described as “small water-holding containers that do not support other aquatic organisms beyond mosquitoes.”

The fungus is less fussy about what it kills. Paris pointed out that it is widely used on farms and attacks many kinds of insects, not just mosquitoes. “There is some evidence suggesting that it could be transferred to insects sharing the same flowers for nectar,” she said. “There is certainly a need for further research into how a large-scale rollout could affect other wildlife.”

Until that research is done, the advice for people in affected suburbs is the familiar kind. Finn Romanes, director of Victoria’s Western Public Health Unit, recommends using repellent, maintaining fly screens and removing mosquito breeding sites around the home.

The stations clearly thinned out a backyard mosquito. Whether they belong in every backyard still depends on a question yet to be answered: what they would do to the bees and other insects that visit the same flowers.

ZSL published the report on 19 January 2026, five years after the first edition, its project page says. The Tidal Thames is the stretch of river that rises and falls with the tide, from Teddington through London to the sea.

Why, then, has the average number of species caught per net sample in the upper and middle stretches slightly declined since 1992?

The figure of over 115 is a running total of every species recorded since 1957, the report’s wording makes clear. It is not a count from a single survey, and it is not the number of species living in the river today.

Mike Elliott, emeritus professor of estuarine and coastal sciences at the University of Hull, was not involved in the report. “Yes, by most accounts and historically, the Thames can be regarded as a success story even if it literally took a ‘stink’ over a century ago to get MPs and others take notice,” he said in an email.

The report draws on 21 indicators assessed by experts from 21 organisations, the project page says. Thames Water, the water company for London, funded the preparation of the report. The acknowledgements state that the content remains independent and that Thames Water had no editorial influence over it.

Fewer species per net

The fish indicator is rated “Deteriorating” over the long term and “Data stable” over the short term, the report’s assessment shows. “Fish populations have been deteriorating in the long term, but appear to have stabilised in the short term,” the report says.

That judgement rests on Environment Agency seine-net samples. A seine net is a long net anchored on the bank and pulled round in a semicircle to enclose fish. The measure is species per sample, meaning the average number of different fish species caught in each haul.

Species per sample showed a slight but statistically significant decline in the Upper and Middle Tidal Thames between 1992 and 2023, the report says. From 2014 to 2023 there was no significant trend in either direction.

The report offers two readings of the slight decline. It “might be due to changes in sampling methods”, or it “could be an indication of pressures on fish populations either in the Tidal Thames, or further afield.” Further research is needed to determine the cause, ZSL’s project page says.

“Fish are a good indication of water quality and especially dissolved oxygen,” Elliott said. He added that “we are now going backwards because of unregulated sewage discharges”, and that a fall in the number of fish species would not surprise him.

Elliott said the peak tally of about 120 species “did include some freshwater species and some marine ‘straggler species'”. He described stragglers as fish “found by accident either washing in from the sea or down from the river”. UK, European and global analyses of how many species an estuary like the Thames should support suggest that 100 to 120 is appropriate, he said.

In a follow-up email Elliott sent his own graph of species counts over time, which he said “shows we are reaching the stable situation for species”. The report, by contrast, “gives the number of species per sample which is subject to many influences”, he said. He had “seen but not read the 2026 report in detail”.

Many fish in the estuary depend on conditions at sea or in the rivers upstream, Elliott said, “so the influences are far and wide”. The 1957 verdict followed decades of absence. From 1920 to 1964 fish were largely absent from Fulham downriver to Tilbury, the report says, citing a 1998 volume edited by Attrill.

Seahorses, sharks and seals

Juvenile short-snouted seahorses are present in the Thames, a caption in the report says. Three shark species, tope, starry smoothhound and spurdog, are named on ZSL’s project page. They are found in the Greater Thames Estuary, the wider mouth of the river towards the North Sea, not in the river through London.

Seals are surveyed from Gravesend to Felixstowe and Deal, in the estuary. The 2024 estimate was about 599 harbour seals and 2,987 grey seals, the report says. Harbour seals peaked at 796 in 2017 and had fallen to 599 by 2024, below their 2013 level.

The harbour seal indicator is rated “Deteriorating” over the short term and “Data stable” over the long term. Grey seals are improving on both timescales, with 2024 counts about 3.5 times those of 2013. Rising grey seal numbers may threaten harbour seals through competition for haul-out sites and prey, and through the risk of phocine distemper virus, the report says.

Nitrate, a nitrogen compound that can be dangerous to aquatic life, is rated “Deteriorating” over the long term and “Data stable” over the short term. Sewage effluent continues to cause gradually increasing nitrate concentrations, ZSL’s page says. The Environment Agency identifies industry and sewage as the main source in London waterways.

Dissolved oxygen, the oxygen in the water that fish need to breathe, is improving on both timescales, the report says. The Thames Tideway Tunnel, a £4.6 billion “super sewer”, was fully connected in 2025, according to ZSL’s page.

The Tidal Thames also supports 92 species of bird, ZSL’s project page says. Summer water temperature in the river has risen by 0.13C a year on average since 2007, the same page says.

The verdict was returned on 16 October 2026 at Suffolk County Superior Court in Boston. Judge Christopher K. Barry-Smith presided, according to a verdict summary from Sabatini’s lawyers. The jury also rejected Knouse’s counterclaims, the claims she brought against Sabatini within the same case, that he had harassed and retaliated against her, the New York Times reported.

Jurors found that Sabatini had proved Knouse defamed him and interfered with his ability to work, according to the Times. The Globe described the second finding as interference with his ability to make a living. The award was unanimous. According to Sabatini’s lawyers, the jury’s awards by claim were $500,000 for defamation, $3.2 million for tortious interference with contractual relations and $2.3 million for tortious interference with prospective advantageous relationships, the three sums that make up the $6 million.

“We are so happy that we got David Sabatini’s good name back,” Lisa Arrowood, Sabatini’s lawyer, told the Globe. “Now he can be regarded as the brilliant scientist he is.”

Ellen Zucker, Knouse’s lawyer, told the Globe that her client intends to appeal. Sabatini’s lawyers said they expect an appeal and post-trial motions, which are due 10 days after the verdict.

She told the Globe she regrets the verdict and does not regret the decision to pursue the case. She told the Times that Knouse “came into court and told the truth.”

Zucker told People the verdict was “disappointing” and said several women had testified about “very concerning experiences” in Sabatini’s lab. “It is disappointing that the jury’s verdict reflects none of this,” she said.

Sabatini ran a research lab at the Whitehead Institute in Cambridge for 24 years. The lab studied cell biology in order to better understand neurodegenerative diseases, cancer, diabetes and ageing.

He was forced out of the lab in 2021 after an internal Whitehead investigation. That investigation concluded he had a past sexual relationship with Knouse, a younger and less experienced colleague over whom he held a role that could influence her career, the Times reported.

The investigation also found that his lab had maintained a sexualised atmosphere, the Times reported. The Globe reported that the investigation found the relationship had violated a strict workplace relationship policy at Whitehead.

Knouse told investigators the relationship was coercive, that Sabatini was in a position of influence and that he had fostered a toxic and sexually charged work environment, the Globe reported. In her counterclaim, she asserted that he coerced her into a nonconsensual relationship and abused his position of power, the Globe reported on 18 October.

Sabatini said he and Knouse, then a Whitehead fellow, began a consensual sexual relationship in 2018 that ended when he began pursuing another woman, the Globe reported.

He denied that his lab had an improper atmosphere, the Times reported. He maintained that Knouse had schemed to get him fired and disgraced after he became interested in another woman.

The verdict came nearly five years to the day after he filed his defamation suit. The original lawsuit named Knouse, the Whitehead Institute and the institute’s director as defendants.

The institute and its director were dismissed from the case before trial. In a memorandum of decision dated 3 November 2025, the Superior Court allowed their motion for summary judgment, a ruling that ends a claim without a trial, on Sabatini’s gender discrimination claim, the only claim then remaining against them.

The same order allowed Sabatini’s claims against Knouse for defamation and tortious interference to proceed to trial. It dismissed his claims against her for intentional and negligent infliction of emotional distress, according to the order. The order also allowed Knouse’s motion for summary judgment on Sabatini’s claim alleging violation of G.L. c. 151B, described in the order as a hostile work environment claim, and on the portion of his defamation count concerning statements of personal opinion, which the order described as statements characterising his conduct in terms of Knouse’s subjective, non-provable opinion.

After the verdict, the Whitehead Institute said in a statement that all claims against it had already been dismissed. “Therefore, Whitehead was not a party to this case, and the decision and judgment only apply to Drs. Sabatini and Knouse, the parties in the case,” the institute said, People reported.

The trial lasted about three weeks. Some former members of Sabatini’s lab testified that he fuelled a toxic and sexually charged lab culture, and others praised his mentorship. The jury was made up of seven women and six men.

Sabatini told the Times he hopes the verdict is a step toward regaining his reputation. “I hope that I can get back to being judged by my science and my training,” he said.

Knouse is also on the faculty of the Koch Institute for Integrative Cancer Research, which is affiliated with MIT.

Sabatini was hired in 2023 by the Institute of Organic Chemistry and Biochemistry in Prague and later opened a Boston-area branch of that institute. He has been splitting his time between Massachusetts and the Czech Republic.