Those figures are capture counts, not a census. They describe how many individual phascogales came into the hands of ecologists during a defined survey effort, not how many live inside the fence. That distinction matters for a species this cryptic, where detection can swing on weather, timing, and luck. But AWC’s ecologists read the jump as a signal that the population appears to be rebounding after a punishing stretch of years — an inference drawn from capture rates rather than a headcount.

Mt Gibson sits on Badimia Country, in a landscape where the Wheatbelt’s cleared farmland gives way to remnant woodland and shrubland. The phascogale is one of the mammals that vanished from much of that country during the last century, and AWC returned it to the sanctuary behind the predator-free fence nine years ago.

The dry years, and the rain that followed

Before the record, there was a trough. AWC describes the sanctuary as having been “incredibly dry” through the 2023–2024 drought, with the region logging one of its worst heat waves and just 171 mm of rain — the lowest rainfall on record for the Wheatbelt, according to AWC Field Ecologist Aliesha Dodson.

For a small insectivorous marsupial, that kind of year is not an abstraction. Rainfall drives plant growth, plant growth drives invertebrate abundance, and invertebrate abundance is the food supply. When the rain fails at the bottom of the chain, animals at the top of it feed less, breed less successfully, and raise fewer young. Capture numbers in 2024 — 28 individuals — came out of that squeeze.

Then the pattern flipped. AWC reports the Wheatbelt received 401 mm of rain after June 2024, followed by 319 mm in 2025. Two comparatively wet years stacked on top of the driest on record, and the sanctuary’s vegetation had the moisture to respond. AWC’s interpretation is straightforward: as conditions improved, phascogale numbers appeared to recover, and the 2026 survey picked up the difference.

It is a useful reminder of how tightly small mammals in semi-arid Australia are wired to rainfall. Populations here do not hold steady; they contract hard in drought and expand when the seasons allow. The 48 captures at Mt Gibson are best understood as the upswing of that cycle — measured, this time, inside a fence where the usual predators are not also taking a share.

What a record capture count does and does not mean

AWC names three contributing factors for the 2026 result, and the weather is only one of them.

The second is method. Survey timings were refined, which can meaningfully change how many animals a given trapping effort encounters. Shift the survey window and you can shift the count without the underlying population changing at all — one reason ecologists are careful about reading capture totals as abundance.

The third is the fence. Mt Gibson’s 7,800-hectare enclosure excludes feral cats and foxes, and the red-tailed phascogale is a species that feral cats hunt readily. Dodson notes that capture numbers this high are rare, and that the predator-free area is “doing exactly what it was designed to do” — holding a space where a small, vulnerable marsupial can ride out a drought and then multiply when the rain returns, rather than being picked off during the recovery.

So what the 2026 number does say is that more phascogales were detectable, and catchable, inside that fence this year than in any survey since reintroduction. What it does not say is how many phascogales are in there. No one is claiming 48 is the population, and no one is claiming the population doubled. Capture rates are an index — a tracked signal, compared against the same site’s earlier numbers — and the honest version of the news is that the index has risen sharply from a drought low to its highest point in nine years of monitoring.

Three sanctuaries, one on the west coast

Mt Gibson is one of three sanctuaries owned or managed by AWC that protect red-tailed phascogales, and the only one of the three on Australia’s west coast — which gives the site weight beyond its own fence line. The others sit inland in New South Wales: Mallee Cliffs National Park, which AWC manages in partnership with the NSW National Parks and Wildlife Service, and Scotia Wildlife Sanctuary, where AWC reintroduced the species last year.

Spreading a species across widely separated, individually protected populations is a hedge against the kind of year 2023–2024 turned out to be. A drought that flattens one region does not necessarily flatten the others; a single catastrophic event in one fenced area does not take the whole recovery with it. Mt Gibson’s role in that arrangement is to carry the western population, in the region where the western population now depends on protected habitat.

Which brings the September 2026 result back to its actual scale: 48 red-tailed phascogales captured in one survey, inside one 7,800-hectare predator-free fence, up from 28 captured in 2024 and the most since the animals were put back nine years ago. Not a population total — a capture count, taken in the same place, the same way, year after year, and now reading higher than it ever has.

Guinness World Records has now made that career official, and the number they attached to it is what makes this more than a nice human-interest note.

What Guinness actually confirmed

The record is precise. Guinness World Records lists it this way: “The longest career as a language teacher (female) is 67 years and 142 days, and was achieved by Beverly Hannett-Price (USA) in Beverly Hills, Michigan, USA, on 22 January 2026.” She was 90 when the count closed.

Most of us describe our working lives in rough decades and job titles, not as a single unbroken figure counted to the day. That framing is what makes this interesting. It takes something we normally experience as a blur of years and pins it to an exact length.

There’s a second number Hannett-Price offers herself, and it’s one we keep coming back to. She told Downtown Publications, “I’ve not only taught consecutively but I’ve only missed five days since I began in 1958.” Five days across roughly 67 years points at what the record is really made of: showing up, over and over, for a very long time.

What 67 years in one profession looks like from the inside

To see why this stands out, it helps to know how most teaching careers go. They tend to be short. The Learning Policy Institute reports that every school year, about one in seven US public school teachers either changes schools or leaves the profession.

The early years are the leakiest. Research by University of Pennsylvania’s Richard Ingersoll, reported by Education Week, found that 44 percent of new teachers leave within five years. That figure is debated. A federal study from the National Center for Education Statistics, summarised by the NEA, put it closer to 17 percent. Either way, the picture holds: a lot of people leave in the first few years.

Against that, staying a decade is a real commitment, and staying 67 years is a category of its own. Hannett-Price broke a series of barriers at Detroit Country Day School, where she became the school’s first female Upper School teacher, first female department chair, and first female dean. She has no interest in stopping. “I have no plans for retirement,” she said. “I love this school. I love teaching.”

The students who passed through her classroom

The list of former students is eye-catching. It includes Tony- and Emmy-winning actor Courtney B. Vance, actor Robin Williams, and former Microsoft chief executive Steve Ballmer.

Vance told WXYZ-TV that “she really taught me how to write and to put wonderful thoughts together, and because of that, we’ve stayed in touch.”

What we take from a career measured this way

Perhaps the most useful way to read this record is to strip away the exceptional parts. The celebrity students are incidental. The exact day count is a fun accounting trick. Even the record itself is really just recognition, after the fact, of something she was doing anyway.

What’s left is continuity, which is perhaps both the least dramatic quality a career can have and the hardest to keep up.

Her own advice, offered at the surprise assembly where students celebrated her, points the same way. “In life, be the best version of yourself you can possibly be because there will never be a better you,” she told the room. It’s the sort of line that could sound like a poster on a classroom wall. Coming from someone who has backed it with nearly seven decades of walking in the door, it reads differently.

That pairing of numbers is the point. Reintroductions of large raptors often run into the same infrastructure that helped empty the landscape in the first place; releasing birds into a corridor with lethal pylons can mean feeding a population sink. The Jaén work, published in Raptor Journal and available through reference-global, tracks what happened when the release programme and the line retrofit were carried out in the same area: a breeding population rebuilt from local absence, and a formerly dangerous stretch of line that tagged birds used repeatedly without dying on it.

From local breeding absence to releases

The red kite (Milvus milvus) had become extinct as a breeding species in Jaén in the early 2000s. That is a local, provincial loss, not the disappearance of the species — red kites persist elsewhere in Spain and across parts of Europe — but for Jaén it meant a raptor that had once nested there stopped doing so, and stayed gone.

Reversing that required birds on the ground. Following a habitat suitability assessment of the area, 186 individuals were released between 2021 and 2025. Of those, 181 were fitted with GPS transmitters, which is an unusually high proportion and the reason the rest of the study’s numbers exist at all. Transmitters turn a release into a dataset: where birds go after leaving the release site, how long they survive, which parts of the landscape they use heavily and which they avoid, and — when a bird stops moving — where and often why.

The monitoring also produced a mortality picture that the paper does not soften. Among monitored red kites there were 80 mortality events, of which 72 had an assigned cause. Natural causes accounted for 54.2%, with predation dominant (n=34). Anthropogenic causes accounted for 45.8%, led by poisoning (n=9) and electrocution (n=8). Reintroduced raptors die, and a substantial share of those deaths trace back to human activity. Any claim about mitigation has to sit inside that context rather than replace it.

The 48 pylons and the 802 fixes

One power line in the release area had previously been associated with elevated raptor mortality — a known black spot rather than a hypothetical risk. It was retrofitted with anti-electrocution devices across 48 pylons.

Electrocution on medium-voltage distribution lines is a well-understood mechanism. A large bird perching on a pylon crossarm can bridge the gap between a live conductor and an earthed structure, or between two conductors, using its own body. Retrofits work by removing the opportunity: insulating conductors near the pylon, moving perching positions away from live parts, or adding devices that make the dangerous spot unusable or harmless. The engineering is not exotic; the difficulty is usually deciding which pylons out of thousands to treat, and then paying for it.

The GPS data then provided an exposure record. The study reports that 21 reintroduced kites were tracked within the treated line area, producing 802 recorded locations across that corridor. None of the recorded fatalities was associated with the retrofitted structures.

Two limits matter for reading that result correctly. First, it is specific to that line: the zero applies to the 48 retrofitted pylons, not to the programme as a whole. The eight electrocution deaths recorded among assigned causes occurred elsewhere, on structures that had not been treated — which is, in a blunt way, the contrast that gives the retrofit result its meaning. Second, absence of recorded deaths on a set of structures is an observational finding, not a controlled experiment; 802 fixes from 21 birds is a real exposure record, but it is a record from one line in one release area over a few years, not a universal guarantee.

What it does establish is that tagged kites were genuinely present on and around structures that used to kill raptors, and that presence did not convert into carcasses at those structures. For infrastructure mitigation — a field where the counterfactual is usually invisible, because you cannot easily count the birds a retrofit saved — documented use plus documented absence of mortality is among the more concrete forms of evidence available.

What 22 pairs in 2026 actually measure

The breeding count rose quickly. Jaén had three pairs in 2023 and six in 2024, then 13 in 2025 and 22 in 2026. From a starting point of no breeding pairs at all, that is a population establishing itself and then compounding, as released birds reach breeding age and settle.

It is worth being clear about what that curve does and does not show. Twenty-two pairs is a provincial breeding population rebuilt from zero inside a few years — a meaningful conservation outcome for Jaén and for Andalusia. It is not evidence of species-wide recovery, and the study does not claim it is. Nor does the retrofit result mean the threat landscape has been solved: poisoning remained the leading anthropogenic cause of death among monitored birds, and predation was the single largest assigned cause overall. Mitigating one hazard does not remove the others.

The methodological contribution is the combination. Releases supplied birds; habitat assessment guided where; GPS tags measured survival and space use rather than assuming them; and the retrofit addressed a specific, previously documented mortality source inside the area where those birds were going to live. Reintroduction and infrastructure mitigation were treated as one problem instead of two.

The locked figures, then: 22 breeding pairs in Jaén in 2026, after 186 releases between 2021 and 2025, with 181 birds GPS-tagged; 48 retrofitted pylons; 21 birds and 802 fixes in the mitigated corridor; no mortality recorded on those structures. Not a finished recovery — a measured one.

It is also at maximum entropy. Shake it, and the entropy goes down. Leave it alone, and the second law of thermodynamics carefully, patiently, puts it back into two neat layers.

That example is not mine. It belongs to the physicist Dan Styer, who used it in a 2019 paper about a teaching error he thinks has done real damage, and once you have seen it you cannot go on saying that entropy means disorder.

What the word is actually counting

Take a deck of 52 cards and shuffle it. The number of orders that deck could be in is 52 factorial, which is 8.07 followed by 67 zeros. For scale, that is roughly 10 to the 50th times the number of seconds that have elapsed since the Big Bang. Any well-shuffled deck is, with overwhelming probability, in an order no deck has ever been in before.

Boltzmann’s formula, the one carved on his gravestone in Vienna’s Zentralfriedhof, is S = k log W. The W is that count: the number of microscopic arrangements that would look identical from where you are standing. The k is Boltzmann’s constant, 1.380649 times 10 to the minus 23 joules per kelvin, and since the 2019 redefinition of SI units it is exact by definition rather than measurement. The idea is Boltzmann’s from the 1870s; the equation in that form is Planck’s, around 1900. Boltzmann did not live to see it on his own headstone.

So entropy counts indistinguishable arrangements. Not messiness. And the phrase doing the quiet work in that definition is “from where you are standing,” because whether two arrangements count as identical depends on which differences the observer can resolve.

The card example needs one caveat, which is that 52 factorial is a combinatorial count rather than a thermodynamic entropy. It is the right intuition wearing the wrong units.

Why disorder is the wrong word

Styer’s line in “Entropy as Disorder” is the cleanest statement of the problem I have found: “Entropy is a precise, measurable quantity. Disorder is a matter of opinion.”

The oil and vinegar is one counterexample. Phase transitions are another: water freezing into ice releases heat into its surroundings, and the total entropy goes up while the water becomes visibly more ordered. Frank Lambert spent years arguing the point in chemistry education, most memorably in a 2002 paper titled “Disorder: A Cracked Crutch”, and proposed energy dispersal as a replacement metaphor. That replacement is itself contested, and Styer and Lambert do not fully agree with each other, so the safest lay definition remains the counting one.

The reason to care about a metaphor is that this one licenses an entire genre of writing. Once entropy means disorder, and the second law means entropy always increases, you have a physics-shaped justification for the belief that everything is falling apart, and it gets applied to institutions, relationships, cities and civilisations by people who would not attempt the underlying arithmetic. There is even a published critique of the misuse of thermodynamic entropy in economics, which is a fairly good indication of how far the metaphor has travelled.

The clause that gets dropped

Here is the sentence at the centre of it. The entropy formulation of the second law holds that entropy never decreases in an isolated system.

Isolated means exchanging neither energy nor matter with anything else. A flat is not isolated. It has electricity coming in, gas, water, food, sunlight through the window, and heat leaving through the walls at a rate my landlord and I have discussed. The popular version of the second law drops the qualifying clause, and the entire misunderstanding lives in that omission.

Which means the physics is not saying your flat tends toward mess. It is saying something narrower and more interesting: local order is always purchasable, and the purchase is always paid for elsewhere as heat.

Order is not scarce. What is scarce is the energy you spend making it, and the payment is non-negotiable.

The cost is easy to see once you look for it. A conventional petrol car converts only 14 to 30 percent of the fuel’s energy into motion at the wheels, with 68 to 72 percent lost in the engine as heat. That is not sloppy engineering. Carnot proved in the 1820s that the ceiling for any heat engine is set by the temperatures it works between, and a worked example in a standard physics text gives a theoretical maximum of 47.6 percent against roughly 35 percent achieved in a nuclear plant. The waste is structural.

Twenty minutes of tidying is the same transaction, in a different currency. The room gets more ordered, my body burns glucose, heat goes into the air, and the total entropy of the room plus me plus the atmosphere goes up. The flat wins. The universe charges for it.

The word I have to say out loud

This is an analogy. I want to say that clearly and early, because the next move is the one where essays like this go wrong.

No peer-reviewed work measures entropy in housework, or in friendships, or in a marriage. When someone writes that a relationship decays without energy input because thermodynamics says so, thermodynamics is not saying so. It is being borrowed to make an ordinary observation about maintenance sound like a law of nature.

The ordinary observation is still true, and it does not need the borrowing. Maintenance and neglect are asymmetric because of arithmetic. There are very few arrangements of my kitchen I would call tidy and an enormous number I would call not, so a random change is far more likely to land in the second pile. That is the counting intuition again, and it works perfectly well without invoking a law that was formulated for isolated systems.

What routine actually looks like when it is measured

The most useful thing I know about this comes from a study that had nothing to do with entropy metaphors and quietly demolishes them anyway.

Chaoming Song, Albert-László Barabási and colleagues took three months of mobile phone traces from 45,000 anonymised users and calculated the entropy of their movement. Human mobility turned out to be extraordinarily low in entropy: the actual entropy of where people go peaked at around 0.8 bits per hour, and the theoretical maximum predictability of an individual’s location came out at roughly 93 percent, with a distribution so narrow that it barely varied between people who travelled widely and people who did not.

Ninety-three percent. The person who feels their life is drifting into chaos is, measured by where their body actually goes, running one of the most predictable systems anyone has instrumented.

Claude Shannon found something comparable in written English, which he estimated carries between 0.6 and 1.3 bits of information per letter, meaning around 75 percent redundancy. That is Shannon entropy rather than thermodynamic entropy, and the two are related by mathematics rather than being the same quantity, which is worth saying whenever both appear in one paragraph.

So the story does not go where the metaphor wants it to. The bottle on the counter is ordered and at maximum entropy. The flat gets messy because most arrangements are messy, not because the universe is pushing. And the life inside the flat, tracked by the least sentimental instrument available, turns out to run on rails.

What I find myself thinking about is not the tidying. It is that 93% and the fact that the person inside a routine is the last one able to see its shape.

A survey published in Royal Society Open Science has now put numbers on the problem. Jazmín Ramos-Madrigal and colleagues, with Hannes Schroeder as senior author, pooled low-coverage shotgun sequencing data from 318 human skeletal and dental samples taken from 33 archaeological sites across the Caribbean, collected between 1998 and 2020. The authors describe it as the first systematic assessment of how, where and for how long ancient DNA survives in the region.

The headline is an expiry date. The team fitted decay rates to the 75 samples that were directly radiocarbon dated and still held enough human DNA to model, then took the slowest-decaying of those as a deliberate best case. After about 5,300 years, they estimate, more than 99.9 percent of that specimen’s DNA fragments would have broken below 30 base pairs, the length at which a fragment can no longer be matched reliably to a reference genome. Their conclusion is that recovery from Caribbean remains much older than roughly 5,000 years will be extremely difficult.

The inner ear held the most human DNA of anything they compared

One of the two effects the survey calls strong is anatomical. The petrous bone, the dense wedge of the skull that houses the inner ear, is often the preferred sampling target in the field, and this survey shows why under conditions this hostile.

The measure is endogenous DNA content: the proportion of usable reads that map to the human genome rather than to contaminating microbes. Averaged across the 199 teeth, 86 petrous bones and 11 rib and long bone fragments that passed the team’s quality filters, it ran as follows:

• petrous bone: 17.4 percent
• teeth: 1.2 percent
• ribs and long bones: 0.01 percent

The gap is not a matter of a few good specimens. Of 86 petrous bones in the comparison, 39 exceeded 10 percent endogenous DNA. Of 199 teeth, six did. Every rib and long bone fragment performed poorly. The differences between petrous bone and the other two categories were statistically significant at p below 0.01 regardless of where the remains had been buried.

The quality of whatever DNA survived is a different matter, and here the tissues converged. Fragment length and cytosine deamination rate, the two other standard measures of preservation, showed no significant difference between tissue types, with one exception: at open sites, teeth carried significantly longer fragments than petrous bones did.

Most of the fragment lengths the study quotes are modelled median values, derived from the decay constant rather than counted off the sequencer. The two exceptions in this article are the 30 base pair mapping cut-off above and the measured average that follows. The team prefers the modelled figures because the raw reads leave out the very shortest molecules, which are lost during extraction, library preparation and sequencing. The raw measured average across the dataset was about 51 base pairs; the modelled equivalent was about 25.

The uneven benefit of a roof over the grave

Caribbean human remains come mainly out of three kinds of place: open sites, caves and semi-open rock shelters. The expectation, supported by earlier work, was that caves would preserve DNA better because they are cooler and more stable.

The survey splits that expectation in two. On endogenous content, caves and open sites came out similar, averaging 5.6 and 2.9 percent, a difference that did not clear the paper’s strict p below 0.01 threshold. Petrous bones from rock shelters did score significantly higher, but the authors attribute that largely to the exceptional preservation at one site, Canímar Abajo in Cuba.

On molecular condition the picture is cleaner. Open-site samples had shorter modelled fragments, averaging 23.6 base pairs, against 33.6 in caves and 30.6 in rock shelters, and higher damage, at about 31 percent deamination against 24 and 23 percent. So shelter helps the molecules hold together without reliably raising the share of the sequence that belongs to the person.

Underlying geology may matter too. The authors note that DNA recovery has been particularly difficult in the Lesser Antilles and Aruba, where volcanic soils tend to be acidic, and better on limestone, which holds a higher pH. They had no soil pH measurements to test this, and present it as a pattern rather than a result.

A free triage step that is right three times in four

The most immediately useful finding needs no equipment at all. Before extraction, every sample had been sorted by eye into good, meaning dense and compact, or poor, meaning brittle and chalky.

That eyeball judgement correctly predicted whether a sample would land above or below 1 percent endogenous DNA in 76 percent of cases, with similar accuracy for petrous bones at 75 percent and teeth at 74 percent. Samples rated well preserved averaged 14 percent endogenous DNA; those rated poorly preserved averaged 0.9 percent.

That matters because the analysis is destructive and the remains are ancestral. A triage step that is free, non-invasive and right three times in four is a way of not grinding up bones that were never going to yield anything.

Caribbean DNA breaks down about four times faster

To compare regions, the team estimated decay rates from the 75 radiocarbon-dated Caribbean samples that still held enough human DNA to fit a decay parameter at all, alongside 24 previously published ancient genomes from North America, Greenland and Alaska. Caribbean DNA decayed on average about four times faster than in temperate North America and about 14 times faster than in the subpolar sites. The gap largely holds up when the comparison is restricted to open sites alone.

Damage tells a similar story with a twist. Deamination at the first position averaged about 28 percent in these samples, higher than the roughly 16 to 20 percent reported for Eurasian material of similar age, but lower than the roughly 34 percent reported from Southeast Asia. Hot and humid is not a single setting, and Southeast Asia appears to be harder on cytosine than the Caribbean is.

Where the region does better than its reputation is in what remains reachable. Only 108 of the 318 samples, about 34 percent, cleared the 1 percent endogenous threshold that makes plain shotgun sequencing affordable, but 189, about 59 percent, cleared 0.1 percent, the working minimum for targeted enrichment. On the 82 samples where both approaches could be compared, 1240k capture raised recovery at targeted sites by an average of 280-fold, with individual gains running from 1.5-fold to 7,268-fold once a single outlier of nearly 13,000-fold is set aside. It helped the worst-preserved material most. Samples below 5 percent endogenous gained an average of 606-fold, while those above 5 percent gained only about 12-fold.

The ranking question stayed open

The analysis that would have ranked these factors against each other was tried and abandoned. Multivariate models combining sample type, burial context and age did detect effects, but gave poor diagnostics and results that were not consistently reproducible across subsets, which the authors attribute to an unbalanced dataset and to missing variables, so they fall back on simple pairwise comparisons. Everything above is therefore a set of one-at-a-time contrasts.

A second limit sits in the paperwork. The samples were processed across three laboratories using different extraction and library protocols, and the team tested for batch effects from library protocols and sequencing platforms and found none. A separate comparison of protocols did turn up one exception: petrous bones from open sites showed protocol-linked differences they cannot fully separate from site-level variation.

Then there is the question of what the averages are hiding. Endogenous content had a mean of 5.8 percent and a median of 0.3 percent. The paper also reports 192 samples below 1 percent, a count it never reconciles in its main text with its own figure of 108 of 318 above that threshold. Among petrous bones alone, endogenous DNA ranged from 0.01 percent to 66.5 percent and modelled fragment lengths from 8 to 63 base pairs, with that spread appearing within single sites as well as between them. The 17.4 percent petrous average describes a distribution with a long tail rather than a typical bone, and the enrichment figures skew the same way, the 606-fold average for samples below 5 percent endogenous sitting against a median of 138, and the 12-fold average above that line against a median of 11.

The regional comparisons are ratios of modelled decay rates rather than of survival times, and they rest on a decay-kinetics framework fitted to fragment length distributions. The roughly 5,000-year ceiling comes from extrapolating one benchmark specimen, deliberately chosen as the slowest-decaying sample available, which makes it a best case rather than an average expectation.

Across all 102 directly dated individuals there was no clear correlation between radiocarbon age and any preservation measure. That sounds like a null result, but the authors read it as masking: when they narrowed the analysis to 11 petrous bones from a single Cuban site, endogenous content and fragment length did decline slightly, and damage rose significantly with age. Time is doing its work. Tissue and setting are simply louder.

The point of knowing the limit

The survey’s real product is the sampling rule rather than the expiry date. It is written for a region where every extraction spends a piece of somebody’s ancestor. Favour the inner ear and tooth cementum, and avoid ribs and long bones. Caves and shelters may have treated the molecules more kindly. Look at the bone first, screen several individuals rather than one, budget for extensive screening, and treat enrichment as the fallback while remembering what it costs. Capture rescues poor material, but it raises clonality and pushes GC content from 42.5 percent to 55.1 percent, which can complicate analysing captured and shotgun data together.

The 5,000-year figure turns out to be less bleak than it sounds. Most of Caribbean human history sits inside that window, so the limit is real without being disqualifying. What the survey narrows is the guesswork that used to precede the decision to destroy, and the number of bones ground up to learn nothing.

Those seventy birds did.

Ruddy ducks (Oxyura jamaicensis) are small, stiff-tailed diving ducks native to North America, the males distinctive in breeding plumage for a chestnut body and a sky-blue bill. They had been imported into British wildfowl collections since the 1940s and bred readily in captivity. The first confirmed sighting of one in the wild in Britain came in 1952, and by 1960 a pair had nested in Avon. The Slimbridge escapees, along with others that got out of smaller collections over the following years, found the flooded gravel pits and reservoirs of lowland England to their liking.

A booming population, and a problem it created

For most of the rest of the twentieth century, the ruddy duck’s spread looked like an unremarkable naturalisation story, the kind told about little owls or Canada geese before it. Between 1976 and 1996 the British population grew at roughly 24 per cent a year, according to figures compiled by the GB Non-native Species Secretariat, and by the late 1980s the species was breeding widely across England and Wales, with smaller populations established in eastern Scotland, Northern Ireland and the Republic of Ireland. By 2000 the feral population had reached an estimated 6,000 birds, a striking figure for a species with no wild presence in the country fifty years before.

The trouble, though, played out in Spain, where the ruddy duck’s closest relative, the white-headed duck (Oxyura leucocephala), had been reduced by hunting and wetland drainage to a few dozen birds by the 1970s. A long conservation effort, centred on the Laguna de Zoñar reserve in Córdoba, where the last birds had survived, had begun rebuilding the population. Ruddy ducks dispersing from Britain and France started turning up among them and interbreeding, producing fertile hybrids; the first was documented in Spain in 1991. Most of the ruddy ducks reaching Spain are thought to have originated in Britain, which left conservationists trying to save one duck species confronting a problem created, several hundred kilometres away, by another that Britain itself had let loose.

The cull, and the argument over it

Britain’s first attempt to remove ruddy ducks came in 1997, when the Department of the Environment proposed a mass cull. Animal Aid, a UK animal-rights group, publicised welfare problems from trial shootings that had already taken place, including one bird that reportedly took two hours to die and another recovered alive from the water after being shot thirteen times, and the resulting protests led the department to shelve the plan that spring. Further government culling trials ran from 1999, and their results supported the case that eradication was achievable, so ministers proceeded.

The full national programme began in September 2005, funded jointly by the UK government’s Department for Environment, Food and Rural Affairs (Defra) and the European Union’s LIFE-Nature fund, and carried out through what is now Defra’s Animal and Plant Health Agency (APHA). Marksmen shot birds at more than 140 sites across England, Scotland and Wales, mainly in winter when ruddy ducks gather on open water. The RSPB and the Wildfowl and Wetlands Trust, the charity Peter Scott founded and the one whose own collection had let the birds out in the first place, both supported the cull as necessary to prevent the white-headed duck’s extinction.

Not everyone accepted that the end justified the means. Critics argued that Britain was killing a bird that had done nothing more than adapt successfully to a landscape it had been introduced into, in order to solve a problem whose causes in Spain, continued habitat loss and hunting pressure on the white-headed duck among them, were not solely down to hybridisation. Others doubted a species so adept at hiding in reedbeds could ever be fully eradicated, or questioned whether the disturbance caused by shooting parties at wetland reserves in winter did more harm to other wildfowl than the programme was worth. The disagreement never really got resolved; it got overtaken by results.

What is left

The population fell quickly once the full programme began. From an estimated 4,400 birds in 2005, the UK total dropped to around 200 adults by July 2010, according to AEWA reporting from that period, and to roughly 100 the following year. The decline continued through the 2010s: a wintering peak of just 23 birds in 2018-19, and by 2019 only two breeding pairs remained, neither of which fledged young, the first breeding failure of its kind recorded in fifty years.

The most recent formal assessment, reported to the Bern Convention, November 2024, put the entire British population at around 11 birds, with survey estimates ranging from as few as six to as many as fifteen depending on conditions, split between two breeding pairs in south-west England. No ruddy duck-white-headed duck hybrid has been recorded in Spain since 2007, and the Spanish white-headed duck population has since recovered into the thousands. Britain’s original target of “functional extinction” for the ruddy duck by 2020 slipped to 2023 and has slipped again since; a handful of wary, well-hidden survivors have proven harder to find than the thousands of birds that came before them.

The Slimbridge collection that started it all still keeps ruddy ducks on the same site. This time, their wings are clipped.

So why does PVC defeat the usual recycling methods so completely, and what did a team at Virginia Tech do about it?

The short version: they stopped trying to melt and remold it, and turned it into oil instead.

Why PVC beats the three standard moves

Most plastic recycling still relies on mechanical recycling: sort plastics by type, clean and shred them, then melt the material and form it into something new. PVC makes that process unusually difficult, and much of the reason comes down to its chemistry. Roughly 57 percent of pure PVC’s weight is chlorine, chemically bound into the polymer chain rather than sitting loosely on top. That chlorine is a major complication.

Heat PVC and it begins to release hydrogen chloride, a highly corrosive gas. Improper thermal treatment can also produce chlorinated organic pollutants and other hazardous byproducts. On top of that, manufacturers formulate PVC with very different combinations of plasticizers, stabilizers, mineral fillers and other additives depending on what the finished product needs to do. A PVC pipe and a PVC raincoat may share the same polymer backbone while containing very different additive packages.

That variation matters when the products become waste. Pure, collected and well-sorted PVC can be recycled successfully, but mixed PVC streams are much more difficult to process.  So PVC is recyclable, but getting a clean enough and chemically compatible stream to recycle well is the difficult part.

What the Virginia Tech route actually does

The method, published in Nature in August 2026 skips sorting, melting and remolding entirely. Instead of fighting the chlorine with heat, it dissolves the plastic and pulls the chlorine out chemically.

The reaction is gentler than you might expect for something breaking apart a stubborn plastic. The dissolved PVC is treated with two chemicals, aluminum trichloride and alpha-olefins, and heated to 158 degrees Fahrenheit, about 70 Celsius, for three hours. Under those conditions three things happen at once, as the Nature paper describes it: the chlorine is stripped off.

The chlorine removal is close to total. The process removes more than 99.98 percent of the chlorine. Leftover chlorine is exactly what would make the product corrosive and useless. Strip it out cleanly and the thing that made PVC a recycling headache stops being an obstacle.

Why the product is worth more than the waste

What comes out is not a lower-grade plastic pellet. It is a thick amber oil, called a polyalphaolefin, or PAO. That family of synthetic oils is already what goes into a lot of premium engine and gearbox lubricants. The recycled oil’s thickness, friction and wear resistance are comparable to commercial PAO products, according to the study write-up. The oil meant to keep metal parts from grinding themselves down performs in line with the material engineers already trust.

“Lubricants are the silent hero out there,” one of the authors, Guioliang “Greg” Liu,  said. “We often don’t recognize they exist, but they are out there working quietly.” Turning the world’s most-avoided plastic into one of its most essential fluids is a neat inversion, and it changes the economics too. Recycling usually asks you to take a valuable material and accept a cheaper version of it. This does the opposite.

The claim Liu is willing to make plainly is a careful one. “Number one, we have proved that it is feasible to use plastic waste to make high-performance lubricants,” he told PopSci. He also frames the appeal in market terms: “these lubricants are green, and they can meet the emerging needs for sustainability by the market,” he added.

The scale check

What the team has is a lab-scale proof of concept, and a mild bench reaction is a long way from a working plant. The gap between a beaker that behaves and a facility processing tonnes of mixed, contaminated waste is where most promising recycling chemistry stalls.

The researchers did try to map that gap. They ran the numbers on a hypothetical 50,000-tonne-per-year plant and found a 22.8 percent return on $84 million in build costs, with the investment paid back in under five years. On paper that looks workable.

A tiny recycling rate on 60 million tonnes a year leaves an enormous amount of PVC with nowhere useful to go. Our read is that a route running this cool, stripping chlorine almost completely, and producing something more valuable than the input is a promising direction rather than a solved problem. The chemistry is elegant and the product is real. Whether it survives contact with a truck full of dirty pipe offcuts is the question the next few years will settle.

Three decades later that record has produced something no shorter study could. Across the full run of monitoring, 4,609 individual green turtles were tagged, yielding 1,399 measurements of the interval between one breeding season and the next. Among the gaps that were three or four years long, the four-year share rose from 28 percent to 37 percent. The authors report it as the first systematic long-term change in sea turtle breeding periodicity on record.

The work, by Jared Tromp, Mohd Uzair Rusli, David Booth and Graeme Hays, appears in Royal Society Open Science as an open access paper.

From a few hundred nests a year to well over a thousand

The first half of the story is a conservation success. Annual nest numbers at Chagar Hutang ran at roughly 400 to 500 at the start of the time series and reached roughly 1,500 to 2,000 in recent years, a three to fourfold increase across three decades. An exponential trend fitted to the counts returns an R-squared of 0.52.

That local pattern sits inside a global one. Green turtles were listed as endangered by the IUCN in 2004, and in 2025 the assessment was changed to least concern, on the strength of nesting recoveries at many sites around the world.

The turtles themselves are long-lived enough to make the record personal. More than half of the tagged animals were seen more than once, and one individual was recorded up to 79 times across 28 years.

Why counting the years between nests is harder than it sounds

A female green turtle lays several clutches within one season and then disappears for years, because the fixed costs of migrating and breeding are too high to pay annually. The gap between clutches inside a season is the inter-nesting interval, and at Chagar Hutang it has a clear mode of 10 days with a second peak at 20 days. Re-sightings at zero to seven days are turtles that crawled up the beach and did not lay, termed a false crawl.

The gap between seasons is the remigration interval, and it is the number that carries the finding. Across all years the modal remigration interval was three years, accounting for 37 percent of all recorded intervals, with four years at 20 percent and five years at 15 percent.

The difficulty is that a turtle can nest on a beach nobody is watching. Redang has other beaches, and only Chagar Hutang has a dedicated tagging programme. A female who slips away for one season and nests unobserved will appear in the data as having taken twice as long as she did. That is why the authors treat the longer intervals, five years and above, as the ones most likely to be inflated.

An index built around the turtles nobody saw

The detection problem is what decides whether the headline holds. The team measured how efficiently the study beach was patrolled by asking what proportion of re-nesting turtles were re-sighted after a single inter-nesting interval of 8 to 15 days, rather than after some longer gap. That efficiency fell over the study, from 97.8 percent in 1993 to 74.0 percent in 2024, a decline of about 0.77 percent a year. It stayed above 70 percent for most of the period.

A detection rate that falls while an interval appears to rise is exactly the pattern a spurious result would produce. So the team did not report a mean remigration interval at all. They reported the ratio of four-year to three-year intervals, on the reasoning that missing a turtle for one season turns a two into a four or a three into a six, but does not turn a three into a four.

That ratio is an index rather than an average, and it is the reason the trend survives the detection problem. It is also a narrower instrument than a mean would be. It says the mix shifted; it does not say by how much any individual turtle slowed down. It also counts intervals rather than animals. The paper plots nine individual turtles, and they show a mixture: some held one rhythm across every recorded return, others moved around, and in two cases a three-year gap became a four-year one.

A second bias runs the other way, which strengthens the result. A turtle first tagged three years before the record closes cannot yet have shown a four-year gap, so the late years of the series systematically under-record exactly the long intervals the paper reports as rising.

The three to four switch

Measured that way, the four-year share of three- and four-year intervals rose significantly between the early and late halves of the record, with a chi-squared value of 25.33. The share climbed from 28 percent to 37 percent between the first half of the record and the second. Those two percentages appear only in the paper’s abstract, which dates the early window 1997 to 2012, while the results section and Figure 3 describe the same comparison as running from 1993.

The index also tracked the nest count directly. As annual nest numbers rose, the four-year share of three- and four-year intervals rose with them, though the relationship is a loose one: an R-squared of 0.23 leaves most of the year to year variation unexplained.

One further comparison was tested and came back flat. The rare one-year returns showed no significant difference between the two periods, at p equal to 0.17. Separately, and without a test attached, the paper reports that three- and four-year gaps together made up 64 percent of all remigration intervals up to six years in the first period and 66 percent in the second. What moved, then, was the balance within that pair rather than its size.

What a longer gap costs

A female that waits four years instead of three lays fewer clutches per year on average. The paper works this through with a deliberately simple illustration rather than a measurement. If the share of females on a three-year cycle fell from 80 percent to 50 percent, mean per capita egg output for those individuals would drop by about 8 percent. Those input percentages were chosen to demonstrate the arithmetic and are not taken from the Redang data, and the 8 percent figure should be read the same way.

The authors do say this matters. All else being equal, a longer gap lowers the mean annual reproductive output of individuals and so acts to reduce the rate of population increase, and they write that the finding might have important global conservation implications. What they add is that where there are marked increases in nesting numbers, as they report here, that increase may outweigh the reduced output per female.

The harder question is what caused the shift, and in the discussion the paper is careful. The most parsimonious explanation is that foraging conditions have got worse, so each turtle takes longer to reach the body condition that breeding requires. Beyond that the evidence runs out. Green turtles graze seagrass, and seagrass abundance has been negatively affected by climate change at some sites. Green turtles have also been observed overgrazing seagrass meadows where their own numbers are high. The paper’s abstract names density dependence alone as the candidate; its discussion sets crowding and climate side by side and says the specific reason is unknown. Separating them, on this reading, is a job for other beaches rather than this one.

One alternative was checked and set aside. If the population had simply aged, older females might breed less often. But captive studies find no evidence of reproductive senescence in green turtles, and the individual records here show no systematic change in turtles’ remigration intervals over time, so the authors judge senescence an unlikely explanation.

The whales showed it first

Southern right whales off southern Australia have been reported skipping breeding years at an increasing rate, with reproductive output falling over time and the cause similarly unresolved. That observation came first. What the turtle record adds is that it was not an isolated case among long-lived marine animals that pay a large fixed cost to breed.

The test the authors propose for separating crowding from climate is comparison across sites. The same three to four index can probably be computed wherever a beach has been tagged for long enough, and they suggest checking whether the lengthening tracks the rate of increase in nesting numbers from site to site.

Chagar Hutang holds one of the longest published records of breeding periodicity for this species, covering one beach, with the four-year share of its three- and four-year gaps up from 28 percent to 37 percent. The turtles are back on Redang in numbers the 1993 patrols never saw. How long the gaps between their nesting seasons run is the measurement that is now moving.

The walking catfish (Clarias batrachus) had been brought in early that decade for the aquarium and fish-farming trade. Its first wild populations trace back to breeding stock escaping, either from a fish farm in northeastern Broward County or from a truck moving fish between Dade and Broward counties.

Whichever version you prefer, the outcome was the same. A tropical fish that could move across land found itself sitting on top of one of the most connected networks of canals in North America.

What makes this fish worth a second look is not that it escaped. Plenty of species escape. This one could walk to the next escape.

Fins as legs, wet nights as highways

The name is not a marketing flourish. The Florida Museum of Natural History describes how it works: “They use their pectoral fins to stay upright and wriggle in a snake-like movement to flop across land, usually during wet weather.” Collection manager Rob Robins, writing for the museum, adds that the fish is “named for its habit of moving over land, alternating the use of the left and right pectoral fins in tandem with body twists and flops.”

Florida’s own wildlife agency reaches for a more military image. The FWC species profile notes the catfish is “noted for ability to breath air and make short overland movements by pulling themselves along with their pectoral fins much like an infantry-man scooting under barbed wire.” The air-breathing part matters as much as the fins. A fish that can gulp air survives a trip across a road that a normal fish could not, and it can wait out water so low in oxygen that its competition dies off.

Put the two traits together and you get a species that treats a rainy night as an open door. When the ground floods and canals merge, the catfish can leave one body of water, cross the gap on its fins, and drop into the next. The canals did the long-distance work. The fins handled the last few meters the canals could not.

Three counties, then twenty

The spread is where the numbers get interesting. According to the USGS, in 1968 the fish was found in just three south Florida counties. By 1978, roughly a decade later, it had spread to 20 counties across the southern half of the peninsula.

That is the old picture: a fast-moving frontier, a species doing something novel enough that fish scientists tracked it county by county. The museum’s own collection holds the record, with 168 collections totaling 1,170 specimens documenting the fish’s spread in Florida between 1968 and 2017. The dread that came with it was just as vivid. Early on, the fear was that this air-breathing invader would eat its way through native fish and hollow out Florida’s waterways.

That fear did not pan out. The FWC now states that “early accounts that this fish would eliminate native fishes were erroneous, and it has not had major detrimental effects.” The Florida Museum’s own materials note that the ecological impact remains poorly understood, and that even the species identification as Clarias batrachus may not be correct. The picture sits somewhere between “harmless” and “unknown,” not a clean verdict either way.

The 1970s frontier vs. the flooded driveway

In the 1970s, the walking catfish was a story about a boundary moving across a map. Today it is a story about a driveway. The fish is no longer pushing into new territory so much as reappearing, predictably, whenever the weather cooperates.

When Hurricane Debby flooded Florida in August 2024, making landfall as a Category 1 hurricane, the catfish did what it has done for sixty years. Gulfport resident Luis Bardach filmed one on his driveway and offered the flattest possible narration, telling CBS News: “Very weird catfish, I guess, on our driveway during a tropical storm.”

The scale of the road-crossings has long been part of the lore. Robins, describing the fish for the museum’s 100 Years exhibit, writes that “so plentiful is the Walking Catfish in some locations in South Florida, that catfishes traversing roads on rainy nights have been run over by cars by the score.” “By the score” is colorful shorthand, not a head count, but it captures the texture of the thing: a fish common enough to become roadkill on dry land.

What sticks after the frontier stops moving

What we take from the walking catfish is less about the drama of arrival and more about what comes after it. The apocalyptic version of this story, the one where a foreign fish wipes out a native ecosystem, mostly did not happen. Neither did the reassuring version where an introduced species fades away. Instead you get a third outcome that is easy to miss: it stayed, without disaster.

The fish stopped being news around the same time it became a fixture. Its numbers dropped, its range held, and it settled into the background of south Florida life, surfacing on roads and driveways every time the water rises. A species that arrives with a bang and then simply persists is probably the more common outcome than either extreme the headlines reach for.

Sixty years on, the walking catfish is neither a monster nor a footnote. Just another thing that turns up on the pavement after a storm, wriggling on its fins toward the next stretch of water, the way it has since the ponds first spilled over.

Reports. Have continued. No frequency, no mechanism, no claim about cause. Ten substances were named: citalopram, escitalopram, fluoxetine, fluvoxamine, paroxetine, sertraline, desvenlafaxine, duloxetine, milnacipran and venlafaxine. The wording is public, and so are the minutes behind it.

Note: This is not medical advice. If any of this is relevant to your own medication, that conversation belongs with the person who prescribed it, and stopping an antidepressant on the strength of an article is a genuinely bad idea.

A label warning is not a finding

The most common misreading of the 2019 decision runs in both directions. One camp treats it as regulatory confirmation that the condition is real, common and drug-caused. The other treats the absence of a frequency estimate as evidence that regulators found nothing. Neither is what happened.

Regulators operate on a precautionary standard that is deliberately lower than a scientific one. A signal reaches the threshold for a label change when the reports are numerous enough and consistent enough that patients should be told, whether or not causation has been established. That is the whole design. The label exists to inform consent, not to settle a question.

Other agencies followed. Australia’s TGA updated its warnings in 2024. The UK’s MHRA announced a review of 28 antidepressants in December 2025, which is still running, and reporting an outcome for it would be inventing one. New Zealand’s Medsafe looked at the same evidence in September 2019 and wrote the sentence that has aged best: the prevalence, it said, is unknown.

Seven years later it still is.

Why the obvious study cannot be run

The design that would answer the question is easy to describe and impossible to execute. Take people with no sexual dysfunction, randomise half of them to an antidepressant and half to placebo, measure sexual function with a validated instrument at baseline, during treatment and for several years after discontinuation, and count.

No ethics committee will approve randomising healthy people to a psychiatric medication for years. Randomising depressed people fixes the ethics and destroys the measurement, because depression itself impairs sexual function, so the baseline is already confounded before anyone takes a tablet.

David Healy and Dee Mangin set out the full set of obstacles in a 2024 paper, and the striking thing about their list is how little of it is anyone’s fault. Patients rarely raise sexual symptoms unprompted. Clinicians rarely ask. Withdrawal effects in the first weeks after stopping mask exactly the window that matters. And because no MedDRA code for the condition has been adopted, pharmacovigilance databases cannot be searched for it in the ordinary way, which means the reporting system that exists to detect signals like this one cannot count this one.

The one incidence estimate, and what it actually counted

There is a single formal incidence figure in the literature, and it is a proxy.

Ben-Sheetrit and colleagues went through nineteen years of records from Israel’s largest health maintenance organisation, identifying 12,302 men aged 21 to 49 who met a strict set of inclusion and exclusion criteria. Of those, 866 had been prescribed an SSRI or SNRI; the remaining 11,436, who had not, served as controls. Among the 866 who had taken the drugs, the rate of later being prescribed a drug for erectile dysfunction that persisted was 0.46 percent, or one in 216.

What that number counts is a prescription for a PDE-5 inhibitor. It therefore excludes women entirely. It misses genital numbness, loss of orgasm and loss of desire, none of which produce that prescription. And it captures only the men who raised the problem with a doctor at all. The authors say in their own paper that it is likely an underestimate.

Both of the readings circulating online get this wrong in mirror image. It is not a finding that one in 216 people who take an antidepressant will develop persistent sexual dysfunction, because the denominator is male users and the numerator is one specific prescription. It is also not a finding that the condition is vanishingly rare, for the same reason. The paper is a floor built out of the only countable trace the health system happened to leave.

The court case that gets reported as a verdict

In May 2024 a petition was filed asking the FDA to add a similar warning to US labelling, and when it was not acted on, litigation followed. Csoka v. FDA was dismissed on 31 March 2025.

It was dismissed on standing. A court concluded the plaintiffs had not established the legal basis to bring the claim, which is a procedural ruling about who may sue and says nothing whatsoever about whether the condition exists. It gets cited as though a court weighed the science and found against it. No court has weighed the science.

What was never measured on the way in

The most useful methodological point in this whole area is not about the persistent form at all. It is about how sexual side effects were counted during treatment, because that failure shaped everything downstream.

The registration trials mostly relied on spontaneous reporting: side effects were recorded when patients raised them. When Angel Montejo and colleagues asked 1,022 outpatients directly, using a structured questionnaire, 59.1 percent reported sexual dysfunction, with wide variation by drug, from 72.7 percent on citalopram down to 3.9 percent on moclobemide. That study was open-label with no placebo arm, and depression impairs sexual function on its own, so the absolute figures carry real uncertainty. The gap between asking and waiting to be told does not.

A 2026 meta-analysis of thirteen randomised trials covering 5,941 participants puts orgasmic dysfunction at a relative risk of 3.28 against placebo, with high certainty, while the effect on desire was not statistically significant. That is on-treatment evidence and belongs strictly in the on-treatment conversation, not the post-discontinuation one.

Which brings the argument to its sharpest point, made by the psychiatrist Ronald Pies and quoted secondhand in discussions of this literature: clinicians almost never document baseline sexual function before prescribing. If nobody wrote down what was normal for a person beforehand, then years later, when that person says something has not come back, there is no record to compare against. Not for the individual, and not, multiplied across a population, for the field.

What the absence of a number does to people

Three mechanisms are under discussion, involving serotonin receptor desensitisation, neurosteroid depletion and epigenetic effects on dopaminergic circuitry. All three are hypotheses. There is no biomarker, no imaging finding and no human mechanistic evidence for any of them. A group led by Healy published consensus diagnostic criteria in 2022, which no manual and no regulator has adopted and which have not been validated against an independent sample. No professional psychiatric body has issued a position statement, and that silence is reportable as silence, not as rejection.

There are also clinicians who think the evidence base is far too thin to support the attention it gets. The psychiatrist George Dawson has argued at length that the reporting is unreliable and the implied frequency unsupported, and it is worth being precise about his position: he disputes the quality of the evidence, not the existence of anyone’s symptoms.

What sits underneath all of it is a group of people carrying a symptom that appears on a regulatory label, has consensus criteria nobody has adopted, has one incidence estimate built from a proxy nobody thinks is adequate, and cannot be quantified by any study that could ethically be run. The label acknowledges them. The literature cannot yet count them. Those two facts have coexisted for seven years, and there is currently no design on the table that would end the standoff.

If this is your situation and it is heavy to read about, the person to raise it with is your prescribing clinician, and support from a professional is worth more than anything an article can offer.

But perhaps that reading is wrong. What happened in Midtown this summer is closer to a warning about a single project than a verdict on the whole idea, and the difference matters. New York has bet a real chunk of its housing plan on this kind of work.

We should be upfront about our limits. We are not structural engineers, and this is our reading of the reporting, not a technical assessment. The investigation into what actually failed is still open. Treat everything below as an attempt to make sense of an unfinished story, not a conclusion about who is to blame.

What actually happened at 235 East 42nd Street

The building is the former Pfizer headquarters, a tower dating to the 1960s. Nathan Berman’s MetroLoft Management, working with David Werner, is turning it into roughly 1,600 apartments across two structures. It would be the largest conversion of its kind in the country. On July 7, 2026, two columns that hold up the building buckled on the 21st floor, and the floors above began to sag. No one was hurt.

The failure sat beneath a significant addition: extra floors being built on top of the tower. Berman told the Wall Street Journal that the added weight from widening the upper floors likely caused the damage. Engineers quoted by the Associated Press pointed in the same direction, saying the failure likely resulted from the additional loads placed on the existing structure.

There is also a sharper claim on the record. GACE, the engineering firm responsible for the structure, has alleged that required reinforcing steel was never installed. The city has not settled the cause, and the investigation continues.

Why converting a building is not the same as leaving it alone

An office tower left as an office is a solved problem. It has been standing and doing its job for decades. Turning it into apartments changes almost everything the frame has to handle. You cut new holes for plumbing and light, you move where the weight sits, and sometimes, as here, you pile new floors on top.

That is genuinely harder than ordinary construction, and a conversion brings risks a straight renovation would not. But harder is not the same as unsound. Ben Schafer, a structural engineering professor at Johns Hopkins, told the AP that the incident probably does not undermine the profession’s grasp of the work: “I don’t think it really brings into question our understanding of how to do something like this.” He was one of several engineers offering theories, and his is one view, not a settled finding. Still, it points at the distinction we think this whole episode turns on: a project can be badly done without the idea itself being impossible.

The city’s response, and what it was checking

New York’s Department of Buildings moved fast and wide. It ordered the owner to bring in the firm Thornton Tomasetti for a forensic review, then ran a citywide sweep tied to the Pfizer project team. That sweep covered 180 jobsites, of which 24 were office-to-residential conversions. Along the way the city issued stop-work orders elsewhere: a full halt at GFP Real Estate’s 222 Broadway conversion and a partial stoppage at 750 Third Avenue.

The 222 Broadway halt is worth pausing on, because the two sides tell it differently. A 222 Broadway spokesperson said the order was “issued over a Department of Buildings reporting timeline discrepancy in March for since-addressed work, not safety issues of any kind.” That is the developer’s account. The same reporting gives the city’s version less charitably: contractors did not tell the DOB about cracked concrete beams.

The result that most complicates the easy narrative came when the DOB wrapped its first wave of sweeps. Commissioner Ahmed Tigani said the city “did not find any hazardous structural issues posing an imminent danger to public safety.” Across the sweep, the city issued 65 violations across 38 sites, none of them for immediately dangerous structural problems. On the conversions specifically, Tigani reported that the city “inspected 24 office-to-residential conversion projects, which saw a lower rate of violations than other sites we inspected.”

Neither finding should be read as a permanent all-clear. The commissioner’s “no imminent danger” is a snapshot from the first phase, and the root-cause investigation is still open. The “lower rate of violations” comes from a sample of 24 sites, so it is the DOB’s early read rather than proof that conversions are inherently safer. The DOB also went out of its way to say it found no evidence the failure was “in any way inherent to office-to-residential conversion projects.” That is a “no evidence found,” not proof the conversion had nothing to do with it. The cause remains undetermined.

A single failure is not a verdict on the whole idea

So what does this tell us about the wider push to turn empty offices into homes? Less than the shutdown headlines probably imply. The question is never whether something goes wrong, but whether the system notices and reacts.

On that measure, the response looks less like panic and more like oversight catching up. A column buckled, nobody was hurt, the city investigated, swept 180 sites, found and paused a project where cracked beams had gone unreported, and turned up no evidence the underlying idea is flawed. That is roughly what you would want to happen.

The stop-work orders are not proof that conversions can’t be done safely. They are closer to a safety net doing its job. The more useful worry is not whether to convert buildings, but whether every developer files what they are supposed to file before the inspector shows up.

Adam Brumm and colleagues report in the peer-reviewed Science Advances paper Betel nut sucking is the earliest recorded evidence for the habitual use of a neuroactive plant drug that the older person lived between 25,000 and 16,000 years ago. A second person, found 33 kilometers away, lived between about 7,600 and 6,300 years ago.

This is research reporting, not medical advice. This is one study, not settled consensus. The sample contains only two individuals, and the evidence cannot establish how common the habit was, why either person began, or what the experience felt like.

The older remains were only part of a jaw

The Late Pleistocene person, labelled Maros-LBB-1a, was excavated from layer 4a at Leang Bulu Bettue, a cave and rock-shelter complex in South Sulawesi. Only a right upper-jaw fragment and three molars survive. Dental wear placed the person in middle-to-old adulthood, at least 35 years of age, but the fragment was too incomplete for a secure sex estimate.

The 25,000-to-16,000-year range comes from the archaeological layer, not a direct date on the jaw. Researchers had constrained that layer using several methods, including uranium-series dates on stalagmites formed within the sediments. The remains were described in an earlier study; the new paper re-examines their distinctive wear and chemical residues.

The second individual, CPL-R2a, was a middle-to-old adult male from Leang Cappalombo 1 in South Sulawesi’s limestone uplands. His more complete skeleton included the skull, jaw, teeth and hands. Charcoal immediately beneath the skull and a shell from the same archaeological layer place that context between roughly 7,600 and 6,300 calibrated years before present.

Both people predated farming on Sulawesi. CPL-R2a was associated with the Toalean technocomplex, a localized hunter-gatherer tradition, while Maros-LBB-1a lived thousands of years earlier around the last glacial period.

Grooves made the habit visible

Ordinary chewing flattens tooth surfaces in recognizable ways. These teeth instead carried circumferential grooves, including severe wear around the older person’s molars and around molars, premolars and canines in the younger person’s mouth.

The shape suggested that a hard spherical object about 10 to 20 millimeters across had repeatedly been held between the teeth. Different grooves indicated that it was moved among several positions, sucked and probably gently gnawed. Making depressions that deep in enamel and dentine would have required a long-running habit, not one exposure.

The researchers considered whole, dehusked areca seeds the best fit. Wild Areca triandra is native to South Sulawesi and produces seeds within the inferred size range. Unlike the flatter wear caused by routine mastication or grooves sometimes attributed to using teeth as tools, the pattern wrapped around multiple tooth surfaces.

Neither set had the red-brown stain associated with chewing betel quid. The team instead proposes prolonged sucking of whole seeds.

The wear supports repeated behavior by two people. It does not establish that every member of either community, or even many of them, shared the habit.

Chemistry strengthened the identification

Shape alone could not name the object. The team therefore soaked intact teeth and fragments in acetonitrile, methanol or both for at least a week, then analyzed the extracted residues with liquid chromatography and high-resolution mass spectrometry.

The procedure was deliberately minimal. No tooth was cut, ground, drilled or acid-etched. Solvent blanks accompanied the runs, and the authors report that modern areca seed and betel leaf had never previously been analyzed in that laboratory or on the newly purchased analytical column.

One intact molar sample from Maros-LBB-1a tested positive for arecoline. Three small fragments from the same person did not produce that diagnostic result, although they contained other plant-related compounds. In CPL-R2a, one intact worn tooth was positive for arecoline while another sample was not.

Those mixed detections are important to report. Residue does not distribute evenly through a mouth or survive uniformly for thousands of years. The positive results identify exposure in each person; they do not turn every compound found on the teeth into evidence of the same plant or meal.

Guvacine, guvacoline and arecaidine were not detected. The paper notes that slaked lime drives formation of some of these compounds in a prepared quid, so their absence, together with the lack of staining, is consistent with whole-seed sucking without lime. Consistent does not mean chemically unique, which is why the unusual wear pattern carries so much of the interpretation.

ScienceBlog previously covered arecoline and a possible arecaidine derivative in 4,000-year-old dental calculus from Thailand. At the time, that was described as the earliest direct biomolecular evidence for betel-nut use in Southeast Asia. The Sulawesi teeth move the securely evidenced history much further back and propose an earlier, simpler method of taking the plant.

The lab test asked whether sucking was enough

Modern betel quid usually combines processed areca seed, slaked lime and the leaf of a different plant, Piper betle. Some preparations also contain tobacco. Lime changes the alkaloids into forms that are more biologically available, raising a reasonable question: could an intact seed held in saliva release active material without that preparation?

The team incubated three kinds of intact and broken areca seeds in a physiological salt solution. These included dried and fresh Areca catechu and wild A. triandra. The liquid was kept overnight for 21.25 hours at 36 degrees Celsius, adjusted for acidity and then applied to frog egg cells engineered to express human M1 and M3 muscarinic acetylcholine receptors.

Liquid exposed to intact dried and fresh seeds produced transient electrical currents in those cells, a response consistent with arecoline activating the receptors. A. triandra produced smaller responses. The experiment supports the premise that prolonged contact with an intact seed can release a physiologically active compound even without grinding and lime.

It was a receptor assay, not a trial of betel-nut sucking in people. It did not recreate prehistoric saliva, measure a dose entering the bloodstream or establish intoxication. It also cannot reveal whether pain relief, alertness, flavor, social custom or another motive sustained either person’s behavior.

“Oldest” describes evidence, not the first use

Even the younger end of the older jaw’s date range predates the roughly 13,000-year-old evidence for brewing at Raqefet Cave in Israel. On the current archaeological record, that makes the Sulawesi teeth the oldest known evidence of habitual use of a neuroactive plant drug.

That superlative needs boundaries. Archaeological firsts mark the oldest material evidence found, dated and interpreted so far. They do not recover the first time a person experimented with a plant. Leaves, fruit and soft tissue decay, while many ways of consuming them leave no durable trace.

The two individuals are also separated by at least 8,000 years and lived at different sites. Their similar wear suggests the practice appeared more than once in preagricultural South Sulawesi, but it cannot demonstrate continuous transmission across all the intervening generations.

This is where earlier ScienceBlog reporting on dietary traces in ancient dental calculus offers useful context. Teeth can preserve molecular evidence long after artifacts and plant remains disappear, but a detected molecule still has to be interpreted alongside wear, burial context, contamination controls and plausible alternative routes of exposure.

Ancient use does not make areca nut safe

The result should not be read as advice to use areca nut. The International Agency for Research on Cancer classifies areca nut as carcinogenic to humans, while arecoline on its own is classified as possibly carcinogenic. Tobacco is not required for the areca-related hazard.

The ancient mouths already show a separate form of harm. In CPL-R2a, wear exposed roots and opened pulp cavities in several places, conditions likely to have caused chronic pain and infection. Both individuals had poor periodontal health.

The authors suggest that arecoline’s analgesic effects may have made sucking useful for tooth pain, even as the hard seeds worsened the damage. That possible cycle could help explain persistence, but it is an interpretation. The study did not diagnose addiction, observe pain relief or recover a person’s reason for using the seed.

Anyone concerned about present-day areca-nut use or changes in the mouth needs advice from a qualified clinician. An archaeological study cannot guide individual care.

The next evidential step is a larger set of securely dated remains. Repeated chemical detections, direct dates where preservation permits and the same distinctive grooves at more sites would show whether betel-nut sucking was a durable regional tradition rather than two rare personal histories.

That quiet manoeuvre is the subject of a French field trial that has drawn attention as one of the more concrete results to come out of the fast-growing field of agrivoltaics, the practice of combining solar power generation with farming on the same land. Spring frost is one of the costliest weather risks in fruit growing, and the trial suggests that solar panels built to tilt on command can do more than generate electricity. They can also act as an overnight shield against the cold.

Why blossom frost is such a persistent problem

Stone fruit trees such as nectarines flower early in spring, often before the last frost has passed. A clear, still night is the dangerous scenario. With no cloud cover to trap heat, the ground, leaves and flowers radiate their warmth straight out to space, and their surface temperature can drop several degrees below the surrounding air. A weather forecast reading of two or three degrees above zero offers no guarantee that a blossom won’t freeze.

Growers have long fought this with brute force: wind machines to mix warmer air down from higher layers, orchard heaters, or overhead sprinklers that rely on the latent heat released as water freezes on the buds. All of these work, to a point, but they are expensive to run, use large amounts of water, fuel or electricity, and do nothing for the orchard on any other night of the year. That has made a passive, dual-purpose alternative an appealing target for researchers.

How a solar panel becomes a frost shield

The system involved, developed by the French company Sun’Agri in partnership with the national agricultural research institute INRAE, uses motorised panels that already tilt through the day to manage sunlight and shade for the crop below. On nights when frost is forecast, the panels are commanded into a horizontal position directly over the canopy instead of parking edge-on or retracting.

The physics is the same reason a cloudy night is milder than a clear one. A flat panel overhead intercepts the long-wave infrared radiation that would otherwise escape upward from the leaves and flowers, and it re-radiates some of that heat back down. The trees lose less heat to the night sky, so their tissue stays a fraction of a degree warmer than it would in the open. It’s a passive effect that depends on the panel being large enough and close enough to substitute for the cloud cover the sky isn’t providing.

What the trial actually measured

The results come from a peer-reviewed paper published in May 2024 in the AgriVoltaics Conference Proceedings, based on frost events recorded during the 2022 and 2023 bloom seasons in a nectarine orchard in France. The authors, led by Gerardo Lopez and Perrine Juillion of Sun’Agri, compared trees under the tilting panels with an uncovered control plot during the same frost nights.

Sensors mimicking the temperature of flower buds recorded increases of between 1.61 and 1.69 degrees Celsius under the panels compared with the open plot, close to the 1.7-degree figure widely quoted for the effect. Ambient air temperature near the trees rose by a more modest 0.27 to 0.47 degrees, which is a useful reminder that the protection is happening at the level of the flower tissue itself rather than warming the whole orchard atmosphere.

During the 2022 frost event, the only one of the two seasons’ frosts that produced measurable flower damage, 35 percent of flowers in the unprotected control plot were injured. Under the panels, injury dropped to under 10 percent.

That is a substantial difference in a crop where blossom damage translates directly into a smaller fruit set later in the season.

Reading the result honestly

It covers one orchard, one crop, and two frost events across two seasons — a single-site result, not yet a multi-year programme. A degree and a half of protection matters most in marginal frosts, where the difference between light freezing and none at all determines the crop; in a hard freeze, no amount of panel shielding is going to save a nectarine flower. The paper itself is conference proceedings rather than a journal article with full independent peer review, which is a lower evidentiary bar than a placed study in an agricultural science journal.

There are also costs the frost figures don’t capture. Tiltable agrivoltaic structures are far more expensive to install than a fixed frost-fan or sprinkler system, and they only pay their way if the electricity they generate, plus any protection against summer heat stress and hail, is factored in alongside the frost benefit. Shading a crop for part of the day, even dynamically, can also affect fruit sugar content and colour if it is not carefully managed, an effect Sun’Agri and INRAE have studied separately in their broader agrivoltaics research on grapevines and apples.

The 1.7-degree, sub-10-percent figures are specific to this orchard and these frost nights — they aren’t a guarantee for other sites, and wider trials across more sites, crops and frost severities would be needed before growers could plan around these numbers with confidence.

Where this fits in the bigger picture

Agrivoltaics has mostly been sold on its ability to do two things with one paddock: grow food and generate power. This trial adds a third, more specific pitch, that a panel already installed for energy production can be repurposed on a handful of cold nights each spring to protect the crop it sits above, at essentially no marginal energy cost since night generation is zero anyway.

Whether that makes tiltable panels worth the extra capital cost compared with a wind machine or sprinkler system will depend on the orchard, the electricity price, and how many frost nights a region actually gets. For growers already weighing solar installation for other reasons, though, the French trial gives them a genuine, if narrowly demonstrated, reason to also value what the panels do after dark.

The US Centers for Disease Control and Prevention says a female Cochliomyia hominivorax can deposit 200 to 300 eggs at one time beside a wound or body opening. A superficial scratch or a break in the skin as small as a tick bite may attract her. Once the eggs hatch, the larvae enter the wound, feed head-down and enlarge it as they grow.

This article is reporting, not medical or veterinary advice. New World screwworm infestation is uncommon in people, but a suspected case requires prompt professional assessment and public-health or animal-health reporting.

The biological details have become newly urgent. After being eradicated from the United States in 1966, the parasite began moving north again through Central America and Mexico in 2023. USDA confirmed the first US animal detection in the current outbreak on June 3, 2026. Understanding the threat begins with a distinction that the blunt word “maggot” tends to hide.

“Maggot” is a life stage, not one way of feeding

A maggot is the legless larval stage of a fly, not a taxonomic group with one diet. Fly larvae occupy an enormous range of ecological niches. Many of the blowflies people encounter around dead animals or food waste feed on decaying organic matter. Others cause facultative myiasis, entering a living host opportunistically and often remaining around damaged or dead tissue.

New World screwworm is different. Its larva is an obligate parasite, meaning that feeding on a living host is not an occasional accident but a required part of development. The condition it causes is called wound or traumatic myiasis. An unrelated species, the Old World screwworm, evolved a strikingly similar way of life in Africa, Asia and parts of the Middle East.

The title’s contrast is therefore useful shorthand, but “most fly maggots” should not be read as a census of every fly species. The defensible distinction is narrower: unlike the familiar carrion-feeding blowflies and facultative wound invaders, New World screwworm larvae actively consume viable tissue and continue moving deeper into it.

A wound can be almost too small to notice

An adult female is attracted by odors from a wound or a moist body opening. She can place an egg mass at the edge of a cut, scratch, bite or surgical site, or near the eyes, nose, mouth, ears, genitals or the navel of a newborn animal. The skin break does not have to look severe before the infestation starts.

That is the context for the tick-bite comparison. It does not mean every tick bite is likely to attract screwworm. The fly must be present, the site must be accessible, and exposure matters. It means only that a very small break in the skin cannot rule out an egg-laying site in a place where the species occurs.

The 200 to 300 figure describes one batch, not a lifetime maximum. CDC says a female can lay as many as 3,000 eggs across a lifespan of roughly 10 to 30 days. Females normally mate once, a detail that later becomes crucial to controlling the population.

From an egg mass to a deepening wound

The eggs hatch within about 12 to 24 hours, according to the World Organisation for Animal Health. Newly emerged larvae push into the wound together and orient head-down. Their hook-like mouthparts tear tissue as they feed, while their rear ends remain positioned toward the opening for air.

Over roughly five to seven days, they pass through three larval stages, growing and molting as they consume tissue and wound fluid. This makes the lesion larger and deeper. The characteristic packed, burrowing behavior gave the insect its screwworm name; it does not mean the larvae are literal worms or that each one moves like a screw.

A severe wound can also become a feedback loop. Odor from the damaged tissue may attract more gravid females, which can add further egg batches. Secondary bacterial infection may compound the injury. Without treatment, extensive infestation can kill an animal and has occasionally been fatal in people.

After feeding, mature larvae leave the host and fall to the ground. They burrow into soil and pupate. Adult emergence can take from about seven to 54 days, depending strongly on temperature and humidity. In warm conditions, the entire generation can turn over in less than three weeks.

Livestock bear most of the burden, but not all of it

Any warm-blooded animal can serve as a host. Cattle, sheep, goats, pigs and horses account for much of the agricultural concern because wounds can follow ordinary husbandry, birth, fighting, thorns, insect bites and predation. Dogs, wildlife and birds can also be infested. People are less commonly affected, but human myiasis can involve wounds or mucous membranes and may be medically serious.

The infestation is not contagious in the ordinary sense. It does not pass directly from a person to another person, or from an animal to a person. An adult fly must lay eggs. An infested animal can nevertheless move developing larvae into a new location, and wildlife movement makes a biological boundary harder to police than a fence or checkpoint.

Newborn animals can be particularly vulnerable because an unhealed umbilical site offers the fly an entry point. More generally, risk follows access to exposed wounds and exposure to adult flies in an affected region. The seriousness of the parasite should not be translated into the idea that every person or animal with a cut is in danger everywhere.

Why this old livestock threat returned to US attention in 2026

New World screwworm remains endemic in parts of South America and the Caribbean. A regional resurgence that began in 2023 carried it through areas of Central America and into Mexico that had previously eliminated it. The CDC’s situation summary dated August 11, 2026 said Central America and Mexico had reported more than 209,400 animal cases and 2,500 human cases.

On June 3, USDA confirmed the first US animal case associated with the current outbreak. More animal detections followed in southern Texas and one county in New Mexico, prompting movement controls, surveillance and expanded sterile-fly dispersal. CDC activated its lowest-level emergency response on June 11 to support the wider federal and state operation.

The same August CDC summary described the US risk to the public and animals as very low and said there had been no locally acquired human infestation in the country. Both statements matter. An emergency response can be warranted because establishment would have serious consequences even while the risk to an individual member of the public remains low.

Case locations, active zones and trade controls can change quickly. USDA therefore maintains a current-status dashboard. Any fixed number in an article should be read with its date attached, not as a permanent account of the outbreak.

The fly’s one mating became the key to eradication

The reproductive behavior that helps a single female place thousands of eggs into the environment also supplies an unusually precise control point. Female New World screwworm flies normally mate only once. If that mating is with a sterile male, her eggs do not hatch.

The sterile insect technique turns this fact into an area-wide program. Screwworm flies are mass-reared, their pupae are exposed to radiation that leaves the adults able to compete for mates but unable to produce viable offspring, and enormous numbers are released across the affected region. With repeated releases, sterile males outnumber fertile wild males and the next generation shrinks.

The US Department of Agriculture credits this method with helping eradicate established New World screwworm from the United States in 1966 and eliminating a Florida Keys outbreak in 2017. The joint US-Panama production facility at Pacora can generate more than 100 million pupae per week.

Sterile flies cannot cause an infestation because their offspring never become larvae. The program is not a vaccine and it does not treat an animal already carrying maggots. It is population control conducted at a scale large enough to interrupt reproduction across a landscape.

Not every larva in a wound is a screwworm

A rapidly worsening, painful or foul-smelling wound may raise suspicion, particularly after exposure in an affected area. Other signs can include swelling, bleeding or discharge, an egg mass at the wound edge, visible larvae, or a sensation of movement. None of these signs alone identifies C. hominivorax.

Other flies can cause myiasis, and a deteriorating screwworm wound can later attract species that normally feed on damaged tissue. Confirmation therefore depends on collecting larvae and having trained specialists identify their morphology, or using an approved laboratory. The distinction affects both patient care and the public response because a confirmed screwworm case may signal that reproductive flies have reached a new area.

CDC clinical guidance calls for removing and killing every visible larva and egg in a suspected human case, sometimes through surgical removal, and checking the site again after 24 to 48 hours. Secondary bacterial infection may also require treatment. Evidence for ivermectin in human screwworm myiasis is limited to anecdotal reports, and CDC notes that controlled trials are lacking.

This is why identification and treatment are not sensible do-it-yourself tasks. Anyone who suspects a human infestation should seek immediate medical care and contact public-health authorities; a suspected animal case needs prompt veterinary attention and reporting to animal-health officials. Deep larvae or eggs left behind can allow tissue damage to continue.

The small wound is an entrance, not a forecast

The most disturbing fact about New World screwworm is not simply that a female can lay hundreds of eggs beside a tiny wound. It is that the larvae transform that opening into an expanding source of living food. Wound size at the start is therefore a poor guide to how serious an untreated infestation can become.

The balancing fact is that this is a known organism with a vulnerable reproductive system. Surveillance, rapid treatment, movement controls and sterile-fly release have eliminated it over large regions before. The 2026 detections deserve careful reporting because re-establishment would matter greatly to livestock, wildlife and occasionally human health, not because casual exposure makes infestation likely.

A tick bite can be enough of an opening where the fly is present. It does not make infestation inevitable. The difference between those two statements is the difference between understanding a parasite’s biology and turning that biology into fear.

The specimen shipped from Leiden was female and genetically male-sterile, unable to produce viable seed without a compatible pollen source nearby. That single detail turned out to matter enormously. Because the plant Siebold shipped to Kew cannot reproduce sexually here, every stand of Japanese knotweed descended from it has spread by vegetative means alone, broken rhizome and stem fragments regenerating into new plants. Researchers and popular accounts alike have long described the result as a single genetic clone occupying the majority of Britain’s knotweed-infested ground, and the botanical evidence for the main invasive lineage, Fallopia japonica var. japonica, broadly supports that picture.

The story has picked up some rough edges since. A smaller, less aggressive variety of the plant, Fallopia japonica var. compacta, also grows in the UK with its own separate history, and where the main clone has occasionally hybridised with related species such as giant knotweed, it has produced genetically distinct offspring including the plant known as Bohemian knotweed. Studies of knotweed populations across regions have also found measurable epigenetic differences, chemical marks that alter how genetically identical plants behave, between stands that are otherwise clonal copies of each other. So “one clone” is a reasonable shorthand for how the plant spread; individual stems can still differ at the epigenetic level despite sharing near-identical DNA.

From ornamental to national liability

What began as a fashionable garden plant is now the species most associated with devalued property in Britain. Japanese knotweed’s rhizomes can extend several metres from a visible stand and push through cracks in tarmac, brickwork and drainage, which has made its presence a routine trigger for mortgage lenders to demand a management plan before approving a loan. Local and national government bodies list it as controlled waste that cannot simply be dug up and taken to landfill, and removal contracts for a single residential infestation commonly run into the thousands of pounds.

The plant’s scale is now continental rather than local. It has been recorded across the large majority of Britain’s ten-kilometre survey grid squares, and equivalent lineages are established through much of mainland Europe, North America and New Zealand. Decades of herbicide spraying, root excavation and physical barriers have slowed individual infestations without producing a general method that reliably removes the plant from a site. That gap, a lot of practical experience but not much rigorous comparative evidence, is what a team based at Swansea University set out to close with a purpose-built field trial.

The largest trial ever attempted

Led by Daniel Jones, working with Daniel Eastwood and colleagues in the university’s Department of Biosciences and the knotweed-management company Advanced Invasives, the study ran nineteen different physical and chemical treatments across fifty-eight field plots, each measuring fifteen by fifteen metres, or 225 square metres. The plots were spread across three sites in south Wales: Lower Swansea Valley Woods, Swansea Vale Nature Reserve and Taffs Well. Treatment and monitoring ran for three years at each site, from 2012 to 2014 at two of them and 2013 to 2015 at the third, with growth tracked across 348 smaller monitoring patches nested inside the plots.

The treatments compared herbicide applications, including glyphosate delivered by foliar spray and by stem injection, against combinations of digging, cutting and physical barriers, applied at different frequencies and in different seasons. It is, by plot count and treatment range, the largest controlled comparison of knotweed management methods published to date. The results appeared in 2018 in the journal Biological Invasions.

A result that reset expectations

No treatment eradicated the knotweed from any plot within the three-year study window.

That is the study’s central finding, and for an experiment of this scale, a sobering one. Some approaches did far better than others. Herbicide regimes applying glyphosate twice a year, in summer and again in autumn, or annual autumn treatment by stem injection or foliar spray, produced the largest reductions in stem density and the area covered by knotweed growth. Digging and barrier-based methods without a chemical component performed comparatively poorly. But even the best-performing regimes left living rhizome in the ground, capable of resprouting once treatment stopped. The researchers set out to measure something narrower than eradication, which they judged close to unachievable with current tools: how far a site could be pushed toward long-term dormancy.

That distinction has practical weight for anyone managing an infested site today. It means a knotweed management plan should be built around suppression and monitoring over several years rather than a single treatment promising removal, and it explains why credible commercial contracts now typically specify multi-year herbicide programmes with follow-up inspection rather than a one-off dig-and-dispose job. It also strengthens the case for approaches that do not rely on chemicals or excavation at all. CABI, the UK’s Centre for Agriculture and Bioscience International, has spent more than a decade developing a biological control programme using a Japanese sap-sucking insect, the psyllid Aphalara itadori, which feeds specifically on knotweed and was first released into the British countryside in 2010 under strict containment trials. The original population struggled to survive British winters in the wild, which is why CABI has since released a cold-hardier strain gathered from northern Japan and nicknamed the Murakami psyllid, at UK sites from 2021, with its establishment still being monitored. It will not clear a garden overnight either, and unlike a herbicide programme it would not need reapplying every season once established, but it starves the plant from a place herbicide application can miss too — chronic aboveground feeding that, over years, drains the underground reserves a knotweed stand needs to resprout.

Whether that patience pays off is still an open question. What the Swansea trial did settle is the more immediate one: for now, managing Japanese knotweed means budgeting for suppression on a multi-year timeline. A single treatment that makes the problem disappear isn’t on offer.

Jingmai O’Connor and 13 colleagues describe the specimen in a paper published September 10 in Current Biology. It comes from the uppermost Maastrichtian rocks of the Hell Creek Formation, close to the end of the Cretaceous about 66 million years ago. The authors call the exposed specimen arguably the best-preserved feather yet recovered from Mesozoic rock.

This is one study of one coprolite, not settled consensus about how all late Cretaceous birds were feathered or why some birds survived the mass extinction. Its unusually complete contents nevertheless connect feather anatomy, an extinct aquatic bird and a predator’s meal in a way ordinary bones rarely can.

A lucky break in Hell Creek

David DeMar Jr., a researcher and collections manager at the University of Washington’s Burke Museum, found the dark reddish-brown nodule in 2016 while collecting fish fossils from a rocky outcrop. It was about half the size of a golf ball. Looking through a hand lens, he saw a tiny feather exposed where the object had broken.

The setting made that observation particularly unusual. Paleontologists had searched the Hell Creek Formation for more than 150 years without reporting feathers there. Early Cretaceous sites in China and elsewhere sometimes preserve feathers beside articulated skeletons. Comparable soft tissue is almost absent from upper Cretaceous rock, leaving a thin record for the period immediately before the mass extinction.

The fossil’s discovery was therefore literal luck. An intact outer surface might have offered no hint of the material hidden within. The specimen is now displayed at the Burke Museum, according to a University of Washington account of the work.

How an unremarkable stone became fossilized dung

A feather on a reddish lump did not by itself prove that the object was a coprolite. The team studied its mineral composition and scanned it with micro-computed tomography. Thousands of X-ray slices were digitally stacked, allowing the researchers to inspect the interior without cutting the specimen apart.

The scans revealed additional feathers, tiny scales from a gar and fragmentary leg bones from a bird. The mix of digested animal remains within the phosphatic mass supported the conclusion that this was fossilized feces produced by a medium-sized theropod.

The exact predator is not known. A young Tyrannosaurus rex or the disputed animal called Nanotyrannus are possibilities discussed around the find, but the dropping cannot be assigned confidently to either. “Meat-eating dinosaur” is the narrower claim that the specimen can support.

A food chain compressed into one fossil

The bird bones resemble those of a hesperornithiform, a group of specialized aquatic birds that lived during the Cretaceous. Many members were flightless and used their feet to propel themselves underwater, occupying a role loosely comparable to modern loons.

The gar scales add another level. The paper interprets them as the bird’s last meal rather than direct prey of the theropod. A small fish would have been a difficult, low-return target for a much larger predator, while it fits the diet of a diving bird. The likely sequence is fish eaten by bird, followed by bird eaten by dinosaur.

That reconstruction remains an inference, but it is grounded in the association of scales, bird bones and feathers inside the same coprolite. ScienceBlog has previously covered how fossil dung can reconstruct ancient diets and ecological change. This specimen compresses a particularly vivid set of trophic relationships into an object smaller than a golf ball.

A modern shaft on an extinct kind of bird

The exposed example is a pennaceous feather, meaning its branching barbs form a firm vane rather than loose down. Its central shaft, or rachis, has a square cross-section and a spongy, medulla-filled interior. That design combines stiffness with low weight and is familiar from anatomically modern feathers.

The feather also has densely packed barbs and what the authors describe as a low structural index. In living aquatic birds, comparable construction helps reduce water penetration. Those features independently fit the hesperornithiform identification suggested by the nearby leg bones.

Its modern-looking construction does not make the bird a direct member of the group that survived the end-Cretaceous extinction. Hesperornithiforms sat outside Neornithes, the crown lineage from which every living bird descends, and they disappeared at the boundary. The discovery instead shows that an advanced, light and stiff feather design had evolved in a close relative before the asteroid impact.

Fossil feathers can retain information beyond outline alone. In other deposits, microscopic structures have helped researchers reconstruct aspects of color and iridescence, a subject explored in earlier ScienceBlog coverage of feather evolution. What makes the Hell Creek specimen different is its three-dimensional preservation in ordinary rock rather than a flattened film or amber inclusion.

The hidden feathers complicate the plumage

Micro-CT exposed more than copies of the visible feather. It revealed additional feather forms, including morphologies that do not occur in living birds. The researchers interpret the collection as evidence that the same aquatic bird carried modern-looking, water-resistant feathers alongside smaller, more primitive body-feather types.

That mosaic resists a simple story in which one feather design neatly replaced another throughout the body. Evolution can preserve old structures while modifying other parts of the same animal. It also means the fossil offers information about plumage diversity that the single exposed feather would never have supplied by itself.

An extinction clue, not an extinction answer

The authors suggest that plumage could have influenced which birds endured the cold interval after the asteroid impact. Body feathers provide thermal insulation. If extinct lineages retained feather forms that held heat less efficiently than the plumage of Neornithes, that difference might have mattered during an “impact winter.” Molting strategy could also affect survival if feather replacement temporarily reduced insulation or flight performance.

This is a hypothesis, not a demonstrated cause. One coprolite cannot reconstruct an entire coat, a molt cycle or the physiology of a lineage. It cannot separate plumage from habitat, diet, body size, reproduction and geography. Aquatic hesperornithiforms went extinct despite living near water, which weakens the idea that a waterside habitat alone explains survival, but it does not make feathers the sole alternative.

Genetic evidence adds another, very different view of the boundary. Previous ScienceBlog reporting on bird genomes described rapid evolutionary change around the impact. The feather fossil contributes anatomy from an extinct branch, but the broader question of survival will require several kinds of evidence to agree.

Coprolites become a new soft-tissue archive

Fossil dung has long been used to recover bone fragments, plant remains, pollen, parasites and chemical traces. The new specimen suggests that at least some carnivore coprolites may also preserve feathers, fur or other delicate structures in three dimensions.

This one was recognized only because it happened to split across a feather. Systematic CT scanning could expose comparable contents in specimens whose surfaces give nothing away. That prospect may be the most durable result of the paper: collections already sitting in museums could contain soft-tissue evidence that no one had reason to seek.

A coprolite is evidence of behavior, digestion and environment at once. In this case it also became a protective capsule, carrying the anatomy of an aquatic bird through roughly 66 million years. The find does not solve the extinction of ancient birds, but it changes where paleontologists can reasonably look for the feathers needed to investigate it.

By 1904 silt had clogged the canal carrying Colorado River water to the farms of California’s Imperial Valley, and the agricultural miracle that had been sold to settlers was dying in the fields. Engineers opened a temporary intake downstream to keep the water moving. That worked until winter. A run of floods through 1905 tore the makeshift opening wide, and the river, offered a choice between its old bed and a straight drop into a basin below sea level, went for the drop.

It kept taking it for roughly two years.

A lake nobody ordered

The numbers are hard to hold in your head. At the peak of the break, water was pouring into the Salton Sink at tens of thousands of cubic feet per second, carving fresh gorges out of desert floor as it went. Southern Pacific Railroad watched its track disappear under the rising water, and E. H. Harriman, the company’s president, went to President Theodore Roosevelt for help, as documented by EARTH magazine. Thousands of men and trainloads of rock eventually shoved the Colorado back into its channel in February 1907, per the timeline maintained by the Salton Sea Authority.

What was left behind stretched roughly 58 kilometres long and 42 wide. California had its largest lake, and nobody had ordered one.

Everyone assumed the desert would take it back. Federal officials were confident enough that in 1909 they reserved a further 10,000 acres of land under the water in trust for the Torres Martinez Band of Desert Cahuilla Indians, on the understanding it would be dry ground again by the 1920s.

Farm runoff kept it alive, then stopped

Instead the lake got a steady drip feed. Water used to irrigate Imperial Valley fields had to drain somewhere, and the only ground lower than the fields was the lake. Agricultural runoff propped up the water level for most of a century, long enough for the shoreline to grow marinas, yacht clubs and a passing association with Frank Sinatra.

Then the maths changed. Farms got better at not wasting water, Colorado River allocations tightened, and a hotter climate pulled more off the surface every year. Less in, more out, and the shoreline began walking backwards. Around 16,000 acres of lakebed have been uncovered over the past two decades, reports the University of Southern California. That’s a great deal of newly bared ground with nothing holding it down.

Efficiency is the right policy. It is also the thing draining the lake, which is the sort of knot water management keeps tying for itself.

What the wind lifts off the lakebed

Exposed lakebed is called playa, and this playa has been collecting whatever the valley drained into it for a hundred years. Fine sediments laced with selenium and arsenic, deposited by farm drainage and by local geology, sit on the surface waiting for a windy afternoon, as summarised in a Congressional Research Service report. Imperial Valley gets plenty of windy afternoons.

Working out exactly what in that dust does the damage has been harder than it sounds. A team at UC Riverside led by David Lo went looking for it by piping aerosols from the lake into exposure chambers holding mice. Their result, published in Science of The Total Environment and described by the university, was that Salton Sea aerosol switched on a distinctly non-allergic kind of lung inflammation, unlike the response to a common household mould. Mice given Pacific Ocean aerosol did nothing much at all. Lo’s suspicion is that the culprit is microbial material dissolved in the water rather than the heavy metals everyone expects.

That is one laboratory study in mice, and Lo has been careful to say it doesn’t settle what’s happening in human airways. Call it a lead worth chasing.

What it has already done to a generation of lungs

The human evidence is separate, and it is grim. Researchers at the Keck School of Medicine of USC followed 369 children in the valley for about two years, measuring how much air they could force out of their lungs and how fast. Kids living within 11 kilometres of the shoreline gained lung capacity more slowly than kids further out, at a rate the school compares to living within 500 metres of a freeway. That study was published in JAMA Network Open. Earlier work from the same group found more than one in five local children have asthma, close to triple the national rate.

Lungs do most of their growing in adolescence and then peak. Miss the growth spurt and there is no second attempt.

Hay bales, ponds and a lithium boom

California’s response has been to try to hold the dust down rather than refill the lake, because there is no spare river. Hay bales get staked across open playa to break the wind. Shallow ponds get flooded over the worst of it. State agencies running the Salton Sea Management Program have been building engineered wetlands along the southern shore, both to trap sediment and to give migrating birds somewhere to land.

Meanwhile the brine under the lakebed turns out to hold enormous quantities of lithium, and an industry has arrived to argue about how to get it out. Extraction plants mean jobs in one of the poorest counties in California. They also mean trucks, and trucks mean more dust on the same roads, past the same schools.

The accident took eighteen months. The consequences have had a century, and the children in that lung study are teenagers now, carrying capacity they were never going to get back.

That body is a honey fungus, Armillaria ostoyae, and this one covers 965 hectares of the Malheur National Forest in the Blue Mountains. Call it 9.65 square kilometres, or 2,385 acres, or roughly 1,665 football fields, the comparison Scientific American reached for. Almost all of it sits underground, and the only part visible from the surface is the trees it is killing.

How you tell one fungus from its neighbour

The number comes from fieldwork done in 1998 and 1999 by a team of researchers including Brennan Ferguson and USDA Forest Service scientist Catherine Parks, published in the Canadian Journal of Forest Research in 2003. They took 112 samples from the roots of dead and ailing conifers across about 16,100 hectares of dry mixed-conifer forest, then had to work out how many separate fungi they were holding.

Their method, somatic incompatibility, is less forbidding than the name. Grow two samples towards each other in a dish and watch what happens at the border. Strangers stop, refuse to mix and lay down a dark line between them. Cells from the same individual merge and carry on as though nothing happened.

Sixty-one of the samples merged.

Those 61 came from trees up to 3,810 metres apart. Four other genetically distinct individuals turned up in the same patch of forest at 20, 95, 195 and 260 hectares, which makes the big one unusual even by the standards of its own species.

The age is arithmetic, not a birth certificate

So how do you put an age on something with no rings and no birth certificate?

Age here comes from division: halve the maximum width to get a radius, then divide by how fast the fungus is known to creep through soil. Published spread rates for A. ostoyae run from 0.22 metres a year to about a metre, so the team ran the sum three ways and got 1,900 years, 2,200 years and 8,650 years. The widely repeated figure of 2,400 years sits near the cautious end of that spread. Ferguson and his co-authors called their own approach conservative, since lopsided growth from the original infection point would push the real age higher.

Whatever the true number, this is something that has been copying itself for millennia with barely any drift. Geneticists who sequenced a 2,500-year-old Armillaria individual in Michigan for Proceedings of the Royal Society B found its mutation rate to be extraordinarily low. That is one study of one fungus in a related species, so treat it as a hint rather than a law. James Anderson, who led it, has pointed out the contrast with cancer, which mutates furiously and destabilises itself. This thing does the opposite and just keeps going.

What you can actually see

Very little is the honest answer. Forest pathologist Mike McWilliams walked Atlas Obscura around the site and pointed out what to look for. He chips the bark off an infected fir with a Pulaski, exposing cream-coloured fans of fungal tissue pressed flat against the wood. Below ground it travels as rhizomorphs, black cords like bootlaces, feeling from one root system to the next. In autumn it sometimes fruits, and clumps of honey mushrooms appear at the base of dying trees. They are edible, though the Oregon Encyclopedia warns they upset some stomachs and need thorough cooking.

Whether it deserves to be called a villain

Grand fir and Douglas-fir made up 89 per cent of the infected trees sampled. Both species crowded into these forests over the past century, as fire suppression and selective logging pushed out the ponderosa pine and western larch that once dominated. McWilliams reckons that shift helps explain the size. Ferguson’s team read it differently, arguing that the fungus predates any human fire policy by thousands of years and that burning does little to an established individual. What grew, on their account, was the visibility of the damage rather than the organism causing it.

Either way there is no removing it. Root disease of this kind has no cure at the stand scale, and the standing advice for foresters is to favour the species that shrug it off and thin out the ones that do not.

McWilliams has a caveat he likes to repeat, which is that this is the largest documented organism and there is almost certainly a bigger one somewhere nobody has thought to sample. Hard to argue. The record only exists because a handful of people walked into one specific forest with sampling kits, a GPS unit and the patience to grow 112 wood chips in petri dishes.

The work also clears up an awkwardness that has sat inside track-cycling models for years. Predictive models have simply been handed a non-zero speed at time zero as an input rather than deriving it, a choice the authors call ad hoc at first sight before showing that it falls out of a soft-impact transfer of momentum from rider to bike.

The gap between wheel and line

Held standing starts are used in the kilometre time trial, the team sprint and the individual and team pursuits, and in comparable para-cycling timed events. The rear wheel is locked into a starting gate that opens when an audiovisual countdown reaches zero. That instant defines time zero for the race.

That opening instant gets measured closely. A 2024 analysis of elite track cycling found the quality of the initial launch correlates with early split times in the team sprint, which is why a tenth of a second at the gate is worth a laboratory’s attention.

The rule on jumping the start is simple. As the paper states the practical constraint, the front wheel must not cross the start line before the official signal, and a start judge decides whether a start has to be stopped. The constraint bites on the wheel rather than on the bicycle.

That distinction is what makes the rest possible. Because the gate is placed by officials and bicycle frames differ in length, a rider’s front wheel often sits some distance behind the line at the start. The team measured offsets of 6 and 14 centimetres on their three riders, and note that the range seen in competition runs from zero to about 10 centimetres. Speed built up inside that gap is legal, on condition that the wheel is still behind the line when the signal comes.

The three riders were filmed at 120 frames per second, which resolves the trace to 8.3 milliseconds, and with spatial calibration to sub-centimetre positions. Only one of them arrived at the signal already moving. Starter 1, whose wheel sat 6 centimetres back, was already doing 0.4 metres per second when the signal came and reached the 30-centimetre mark in 0.23 seconds. Starters 2 and 3 both had larger 14-centimetre gaps to work with and both began from a dead stop, reaching 30 centimetres in 0.35 and 0.40 seconds. The gap alone was not enough. The phasing of the body motion against the gate release was decisive too.

Inside the Swiss Timing gate

Before the riders could be modelled, the gate had to be. The Swiss Timing device used at UCI World Cups, World Championships and Olympic Games clamps the rear wheel with pneumatic actuators fed from a 200-bar nitrogen cylinder through a regulator, with 6 bar the standard competition setting.

The team put a force sensor between the caliper jaws and ran the gate at a range of pressures. Clamping load rose with regulator pressure on what the paper calls an affine fit, of about 43.4 newtons per bar, and at the competition setting they measured a normal load of about 278 newtons. Those are two separate facts, and this piece keeps them separate: the appendix’s own tabulated loads scale on that slope alone, which at 6 bar would fall short of 278, so what follows rests on the measured value, which is the one the paper’s model runs on. Then they secured an instrumented bike in the gate and had someone pedal against it until the wheel slipped free, which returned a mean effective friction coefficient of 1.33 across seven slip-release trials, all of them run at lower pressures than competition.

Multiply that coefficient by the competition-pressure clamping load and the gate resists forward motion with about 370 newtons before it opens. Time-resolved traces showed the gate starting to move immediately after the official signal and finishing its release in about 0.06 seconds, after which the clamping force falls to effectively nothing. In practice its grip is gone within the first few centimetres of travel.

The body is the first engine

Three things push or hold the bicycle in that opening fraction of a second: the gate braking it, the drivetrain driving it, and the rider’s mass moving relative to the frame. The third dominates the first fifth of a second.

Each rider’s centre of mass was reconstructed frame by frame from digitised joint positions using published sex-specific segment mass fractions, and it follows a consistent path. Back first, then hard forward. Starter 1 pulled his centre of mass furthest rearward before the signal and reached the highest peak forward speed relative to the bike, about 2.0 metres per second, against roughly 1.6 for Starter 2 and 1.4 for Starter 3.

The propulsive moment is not the lunge itself. Force reaches the bicycle from the deceleration that ends it, when the rider’s forward-travelling mass is arrested against the machine. Peak centre-of-mass velocity is not the quantity that matters on its own. The paper’s term is the relative acceleration together with its timing. Multiplied by the rider’s mass, it becomes a force on the bicycle, and for Starter 1 that force peaked at about 1800 newtons, against the roughly 370 newtons the gate can hold. Peak values for all three riders arrived within the first 0.2 seconds after the signal.

Pedalling, measured with an instrumented crankset sampling at 200 hertz, delivered peak forces the paper describes as of the order of a few hundred newtons, and settled into a near-plateau after about 0.2 seconds. The chain takes over once the gate has let go and posture has stabilised. It is not the dominant term while the gate is still releasing.

A tenth of a second in either direction

With the gate characterised and the three force terms measured independently, the model reproduces the measured velocity traces without being told the starting speed. It matched Starters 1 and 3 closely. For Starter 2 it predicted a slightly late onset and correspondingly lower early speeds than the video showed.

That independence from an assumed start speed is what makes the optimisation worth reading. Its baseline is Starter 1 at his 6-centimetre offset with the timing the model says is best for it, which puts his speed at the signal at 1.34 metres per second, against the 0.4 metres per second he actually managed. From there the team perturbed each model parameter by 10 per cent and compared the resulting speed at the signal. Increasing the explosiveness of the centre-of-mass motion by a tenth raised it by 10.45 per cent. A tenth more crank torque bought 1.49 per cent. A tenth more bicycle mass cost 0.75 per cent, and a tenth taller gearing cost 1.49 per cent. A tenth more rider mass bought 1.49 per cent, and a tenth more grip from the gate cost 2.24 per cent.

Timing swamps all of it. Initiating a tenth of a second later dropped the modelled speed at the signal by 91.04 per cent. Initiating a tenth of a second earlier raised it by 39.55 per cent, and the paper is explicit that this row is a sensitivity case rather than a recommendation, because at a 6-centimetre offset it would very likely put the wheel over the line before the signal.

The team also ran the limiting case with the wheel effectively on the line, where any movement before the signal would be a false start. The model says the best legal onset there is about 0.03 seconds later than what Starter 1 actually did, and that it reaches 30 centimetres in 0.22 seconds with effectively no speed at the signal. That is a modelled figure, and it comes out 0.01 seconds under the 0.23 seconds he actually recorded on video with 6 centimetres of room. Run the same optimisation at his real 6-centimetre offset and the model returns 1.3 metres per second at the signal, which is the sensitivity table’s 1.34 to the precision this second table prints it, and 0.18 seconds to 30 centimetres. The paper’s reading is that he had been using only part of the distance the rules allowed him.

The measured core underneath all of that is narrow. Three riders from one national programme, two men and one woman, on one gate, filmed during their own training sessions. All three were adults who consented, no minors were involved, and the study was exempt from formal ethical review as an observational one with no intervention. The gate characterisation was made on the device itself and does not depend on the riders, although its 370 newtons is a clamping load measured at competition pressure multiplied by a friction coefficient measured well below that pressure. Every statement about how riders behave rests on three people, one of whom the model fits imperfectly.

Numbers that are not measurements

The most eye-catching figures in the paper are not its own measurements. The comparison rows showing British and Australian starters already doing 1.2 and 2.0 metres per second at the signal, and reaching 30 centimetres in 0.14 and 0.11 seconds, are drawn from a companion analysis of the 2022 World Championships by the same group, which the paper lists as under review and not yet published.

The Australian starter reached that 2.0 metres per second at the signal from an offset of only 4 centimetres, smaller than Starter 1’s 6. The authors treat it as an athlete-specific observation rather than something the model predicts from Starter 1’s parameters, and note that matching it from a smaller offset would take a bigger and more precisely phased body impulse than Starter 1 produced.

The percentages are outputs of the model, not repeated trials. The sensitivity table asks what the model does when one parameter is nudged while everything else is held fixed, which is not the same as putting three riders through a tenth of a second of mistiming and watching what happens. The team is similarly careful about the offset result, declining to read it as a universal monotonic law and noting that very large offsets would depend on gate geometry, frame size and whether a rider can load the gate without losing balance. Aerodynamic drag and rolling resistance are left out of the model entirely, which is defensible over 30 centimetres and would not be over a kilometre.

What the framework does offer is portability. A held start followed by a release, with a compliant coupling between athlete and apparatus, describes BMX gates, athletics blocks, short-track skating and swimming block departures in the same terms as a velodrome. Each would need its own gate measured before the decomposition means anything, and the paper is clear that each needs its own instrument: a block’s reaction in athletics, a hinge’s dynamics in BMX. That part is measurement, not modelling. The team has posted its data and code.

Earlier events should have been a warning. A Favelle Favco tower crane collapsed in Sydney in 2012 and one in Melbourne in 2016, and the manufacturer later issued a safety bulletin recommending voluntary fixes. According to the forensic investigation New York commissioned, none of those measures had been done on the crane that failed in Manhattan.

What follows is a plain walk through what went wrong at 550 10th Avenue, why a hydraulic line becomes a fire hazard, and what the city changed.

What actually happened at 550 10th Avenue

On the morning of July 26, 2023, crews at a 45-story tower going up at 550 10th Avenue were pouring concrete using a Favelle Favco diesel-powered tower crane. The crane was lifting a 14,000-pound load when a fire broke out on the crane’s deck, roughly 500 feet in the air.

The hydraulic hose failure

The city ran a multi-year investigation with its Department of Investigation and the engineering firm Gilsanz Murray Steficek. Their 968-page report concluded that the fire was likely caused by a hydraulic hose connection near the engine. The hose came loose from its fitting and sprayed a fine, flammable mist of hydraulic oil onto a hot surface, which set it alight.

The report says the fire was likely caused this way, not proven beyond doubt, and we’ll keep that hedge. It also noted that, “Required inspections that could have identified signs of defects in the hydraulic oil hose were not performed.”

Buildings Commissioner Jimmy Oddo was more confident, saying “we now know why this incident happened, and we have already implemented new changes to help prevent it from happening again.”

The boom collapse and the damage next door

Once the deck was burning, the fire climbed. It reached as high as 900 degrees Celsius and damaged the luffing rope, the line that holds the boom at its working angle. With that rope gone, the 165-foot boom fell onto 10th Avenue, hitting the neighboring 555 10th Avenue building on the way down.

The human toll could have been far worse. About 11 people were hurt, mostly with minor injuries.

Why a hydraulic line is a fire risk in the first place

This is the part that surprises people who don’t work with heavy machinery. Hydraulic oil under pressure isn’t a slow, dripping fluid. When a pressurized line lets go, the oil comes out as a fine spray, almost like an aerosol. Land that mist on a hot enough surface and it catches easily. A diesel-powered crane has plenty of hot surfaces around the engine, which is exactly where this connection failed.

The other ingredient is age. A company mechanic told investigators the failed hose had last been replaced in 2011, although investigators found no written record of that replacement. The report says the hose may even have been original to the crane.

Rubber hoses and their fittings don’t last forever, and the point of regular inspections is to catch a hose that’s wearing out before it fails under pressure. When those inspections don’t happen, a part with a limited lifespan keeps running past it.

What the DOB investigation changed

The most direct result is a new equipment standard. The city now requires new fire-safety measures on every diesel-powered tower crane in New York, built around fire detection and suppression.

Detection is the first half. At 550 10th Avenue the fire started on a deck hundreds of feet up, where someone on the ground has almost no chance of spotting it early. A detection system is meant to flag a fire fast, while seconds still matter and the boom’s key lines are still intact.

Suppression is the second half, aimed at exactly what happened here: a fire near the engine feeding on sprayed fuel and oil. The rule is written for diesel-powered tower cranes specifically, because that mix of a combustion engine, hot surfaces, and pressurized flammable oil is what turned a hose failure into a boom collapse. Deputy Mayor for Operations Jeff Roth called the whole episode “a stark reminder of the need for constant vigilance and improvement in our safety standards.”

That mandate wasn’t the only change. The DOB also tightened licensing for smaller lifting equipment, requiring operators of devices rated 2,000 pounds or more to hold a DOB license with over 3,000 hours of experience. It also issued a violation to the crane’s user, Cross Country Construction. As the Buildings Department and Department of Investigation frame it, the findings have already informed improved crane safety protocols.

We’re not safety engineers or crane inspectors, and this is reporting on a public investigation, not guidance for your site. If you need to know how the requirement applies to a specific crane or job, the Department of Buildings’ own materials on the case are the place to start, along with whoever holds your equipment’s inspection records.