Somewhere in every place you've ever lived there's a door that shuts fine in February and fights you in August.
It's never the front door, which would at least be dignified. It's the bathroom, or the one closet that holds the vacuum and the wrapping paper and a tote bag containing four other tote bags, and it swings shut with a light push all winter, and then some morning in late July you give it the same light push and it stops an inch short and bounces back at you with what can only be described as an attitude. So you push harder. Then you do the thing where you get a shoulder involved. And it goes, eventually, with a sound like a cork.
(The tote bags are the part I can't get past, incidentally, because there is no moment at which a person decides to start a collection of tote bags, there is only a canvas bag from a bookshop that seemed too nice to throw out, and then another one from a conference somebody made you go to, and then one with a cartoon lemon on it that you have never once carried anywhere in public, and at some point the newest one gets folded into the second-newest one for tidiness and that is the exact instant the collection begins, unwitnessed, in a closet you can't currently open.)
The thing you say out loud, to whoever is standing there, is that the building is settling.
Buildings do settle. Not seasonally, though, and not back again in the other direction come January, which yours does, reliably, every year, and has since before you moved in.
What's happening is that wood never stops trading water with the air around it. The cell walls take up moisture when the air is damp and give it back when the air is dry, and the board swells and shrinks accordingly. It happens almost entirely across the grain rather than along it. A wide board can move a few millimeters between a dry winter and a humid August. Which doesn't sound like anything at all until you remember that a door is hung with a gap around it measured in millimeters. Carpenters know this and size the gap for the wet season on purpose. Nobody tells the person who rents the apartment.
Rude.
So the door is a hygrometer, which is the word for an instrument that measures how much water is in the air. It also keeps people out of the bathroom. Two jobs.
And you already read it correctly, which is the bit that gets me. You have never once in your life thought about cell walls, and you still know, in the same way you know it's Tuesday, that the sticking door means the air is heavy today. You got the right answer. You just filed it under the wrong explanation and moved on with your morning, because nobody's grading this.
Anyway.
People have been doing that with the weather for as long as there's been weather. A great deal of what came out of it got compressed into rhyme, because rhyme survives being repeated by tired people and prose doesn't.
Red sky at night, and the part of it that's real
Red sky at night, shepherd's delight. Red sky in the morning, shepherd's warning.
Sailor in some versions. It's very old, it's in Matthew's gospel in slightly different wording, and it's the one that gets trotted out whenever anybody wants to argue that the old ways knew something.
Which, in this case, they did.
Start with the color, because the color is not the point but it's where everyone starts. Sunlight is all the wavelengths together, and air molecules scatter the short ones (the blues) much more efficiently than the long ones (the reds), which is the whole reason the sky overhead is blue at noon. When the sun is near the horizon its light has to cut a much longer slanting path through the atmosphere to reach your eye, so most of the blue has been bounced away somewhere else by the time it arrives, and what's left is the red end.
That much happens at every sunset everywhere and predicts nothing. Every sunset is reddish. That's just geometry.
The forecast lives in what makes one sunset a deep dirty red rather than a pale one, and the answer is particles. Dust and aerosols in the lower atmosphere scatter the long wavelengths forward. The air gets loaded with them when it's descending and stable and trapped under an inversion, which is exactly what air does inside a high pressure system. High pressure also means the western horizon is clear enough for the light to get through in the first place. Low pressure does the opposite. It lifts air, it makes cloud, it washes the muck out with rain, and it puts a wall of cloud between you and the setting sun so there's nothing to see. Pale sunset, or no sunset.
Now the actual mechanism, which has nothing to do with optics at all.
In the mid-latitudes, weather moves west to east. It just does, driven by the prevailing westerlies, and it does it with enough consistency that you can bet on it. So when you stand facing a red sunset, you're looking directly at the piece of sky that will be over your head tomorrow morning, and it's telling you it's clear and stable and full of settled dust. At dawn you're facing the opposite way, into the east, at weather that has already gone past you. A red sunrise means the clear air is leaving and whatever comes next is arriving with cloud.
Which is a person reading the incoming half of the sky and reporting back on it in a rhyme, with none of the vocabulary I just used available to them.
What the rhyme can't do
It can't leave the neighborhood.
The saying holds in the belt of prevailing westerlies, roughly 30 to 60 degrees of latitude, in both hemispheres. Inside about 30 degrees of the equator the trade winds run the other way, easterly, and weather arrives from the east. Stand on a beach in the Caribbean watching a spectacular red sunset and you're admiring the sky that has already finished with you, while the thing that's actually coming is behind your head, unlit, doing whatever it likes. Gorgeous sunset. Useless. In that part of the world the rhyme isn't merely weak. It's aimed backwards.
Local geography breaks it too. Enough coastline, enough mountain, enough of a sea breeze that runs on its own schedule, and the general westward-to-eastward drift gets overruled by something regional that doesn't care about your rhyme.
And even where it holds, it's a probability rather than a promise. The forecast it hands you is tomorrow-ish and fair-ish, and on any given evening it can simply be wrong without the reasoning underneath it being wrong at all. Which is an uncomfortable thing about good forecasting rules generally: the better ones don't feel more certain in the moment, they just win more often across a season, and nobody experiences a season, they experience Tuesday.
Still. There's a real chain there. Particle scattering to pressure system to prevailing wind direction to what's over your head in fourteen hours. Every link in it is checkable, and every link has to hold, and down in the trade winds the last one doesn't.
A pine cone is a machine and it isn't predicting anything
Pine cones open in dry weather and close up in the damp.
True. Straightforwardly, boringly, mechanically true, and you can run the experiment on your windowsill this week.
The scales of a conifer cone are built as a layered structure of two tissues that respond to moisture differently. The inner layer swells a lot when it takes on water and the outer layer swells much less. They're bonded together, so the scale has no choice but to bend, the same way a bimetallic strip curls when you heat it. Damp air and the scales curve upward until the cone has shut itself. Let it dry out again and they bend back down and it opens.
Recent work finds the real tissue architecture messier than the neat two-layer story, with fiber bundles embedded in a matrix doing part of the work, plus noticeable hysteresis. All right. The mechanism holds.
Now the fact that should be more famous than it is. None of this requires the cone to be alive. The tissue that moves is dead. A cone that fell off a tree two years ago and has been sitting in a bowl in your kitchen next to the batteries and the takeout menus will still open and close, over and over, for years. Researchers have documented the same hygroscopic motion in fossil conifer cones. Fossil ones. Stone-old, and still working.
So the cone isn't sensing anything, and it certainly isn't forecasting. It's a humidity gauge made out of dead plant, reading the air right now, this minute, the way your door is.
The cone is not telling you it will rain. It's telling you the air is damp right now. Damp air is more often followed by rain than dry air is, so the cone gets promoted to a forecast when it's really a measurement with a decent correlation attached to it. The cone's own reason for doing this has nothing to do with you: it opens when conditions are dry because that's when its seeds fly furthest.
(The tree is not thinking about you either. Nothing in this article is thinking about you.)
Aching joints, where the evidence genuinely fights
My knee tells me when it's going to rain.
Somebody in your life says this. Possibly it's you. And I want to handle it carefully, because it's the one on this list where the honest answer is that the research is a mess and anyone telling you it's settled in either direction is overselling.
The proposed mechanism is at least plausible on its face. Barometric pressure drops ahead of a storm, and if the tissue inside a joint capsule expands slightly against surrounding structures when external pressure falls, an already inflamed joint could register that. Plausible is not the same as demonstrated, and it has proven remarkably hard to demonstrate.
On one side, a UK smartphone study called Cloudy with a Chance of Pain recruited over 13,000 people across every postcode area in the country and had them log daily pain against weather derived from their phone's location. Among the 2,658 participants who reported consistently for about six months, people with long-term pain conditions were around 20 percent more likely to report a painful day when it was humid, windy and low pressure. Humidity carried most of that, with low pressure and wind contributing less. In absolute terms it's modest: a 5-in-100 chance of a bad day becomes roughly 6 in 100. Real, though. Not nothing.
On the other side, a retrospective analysis of US Medicare claims covering more than 1.5 million adults over 65 and about 11.6 million outpatient visits found no meaningful relationship between rainfall and visits for joint or back pain. The tiny difference they did find pointed the wrong way. The authors were straight about their own instrument, noting that claims data is a blunt way to measure pain and that a real relationship could still be hiding behind it.
And a 2024 systematic review pooling eleven case-crossover studies, 15,315 participants, found no association between humidity, air pressure, temperature or precipitation and the risk of rheumatoid arthritis pain, knee pain or low back pain. One exception came through clearly: gout, where high temperature combined with low humidity was associated with roughly doubled odds of an attack.
Gout. The one condition nobody's grandmother built a rhyme about.
So what do we have. Different populations and different conditions, a painful day defined four different ways depending on whose paper you're reading, weather variables that don't line up between studies, designs that answer different questions, and a disagreement nobody has resolved. That's the genuine state of the field rather than a polite way of telling the believers they're wrong, and the reason it matters here is that this one could have a mechanism. There's a candidate, it's testable, and after all that measuring it still hasn't come out one way or the other, so the sensible position is to say so and stop.
I know that's unsatisfying. There isn't a better answer sitting anywhere. Ugh.
Cows lying down, and how a coincidence gets a rhyme
Cows lie down when rain is coming.
This one gets repeated with total confidence by people who have never stood near a cow, and it has, as far as anyone can establish, nothing behind it.
The Met Office says there's no scientific backing for it at all. Nobody has disproved a mechanism here, because no mechanism was ever offered and there's no signal sitting there needing to be explained. Stories get attached afterward. The cows are keeping a patch of grass dry, or their stomachs respond to falling pressure, and neither one arrives with anything behind it.
What there is instead is arithmetic. Cattle lie down for something on the order of half of every day, resting and chewing cud. Which means that at any random moment you look at a field, there's roughly a coin-flip chance the cows are down. So in a country where rain is also extremely common, "cows lying down" and "rain within a few hours" will co-occur constantly, forever, by the pure accident of two frequent events sharing a calendar.
And then the mind does what it does. You notice the pairing on the days it happens. You don't notice the enormous number of afternoons the cows were down and the sky stayed clear, and after ten years of that you're a person who knows something about cows.
Nope.
I want to be careful with the tone here. The failure is the single most ordinary error a working brain makes, and everybody runs it daily. You've counted the times you thought about somebody an hour before they messaged you. You have not counted the several thousand times you thought about them and your phone sat there doing nothing, because there's no version of a person who keeps that tally.
The cow lore was assembled honestly, by observant people, using the only recording device available to them, which was memory, and memory keeps the hits and bins the misses without ever asking permission. There's a longer version of that mechanism in what we've written about omens and how they get assigned.
The thing that separates the two piles
Line the four of them up and something jumps out that has nothing to do with how old they are or how confidently they're said.
The two that work encode a mechanism. Not a mechanism anybody could have explained, and that's the crucial bit. Whoever first noticed that a hard red sunset meant a decent tomorrow could not have told you a single thing about Rayleigh scattering, aerosol loading, or the general circulation of the atmosphere, because none of that vocabulary existed and would not exist for centuries. They had one instrument and it was a lifetime of evenings. What their rhyme is actually carrying is a compressed correlation that turned out, when the physics arrived, to have a chain of real causes underneath it that runs all the way down.
The two that don't work encode a coincidence somebody remembered. No chain underneath at all. Just a memory of the times it happened to line up, kept and repeated and eventually rhymed.
And you cannot tell which is which from the outside. Both kinds rhyme, both arrive with an "everyone knows" attached, and both get handed to you in the same tone by somebody who has watched a great deal more weather than you have. The cows saying is, if anything, more charming than the red sky one, and charm is worth precisely nothing here.
Which brings up something I want to state plainly, because this whole category gets discussed badly. Folk knowledge about the natural world is not a museum of quaint errors. Farming and fishing and herding communities, and Indigenous peoples with continuous relationships to particular places going back thousands of years, hold bodies of environmental knowledge that are genuinely tested, genuinely predictive, and in a great many cases more finely tuned to local conditions than anything a satellite is going to tell you about that specific valley. The whole argument I'm making here depends on that being true. The red sky rule is folk knowledge. It's excellent. It was derived by exactly the method that produced the cow story, by people with no instruments, and it turned out to be right for reasons they couldn't have accessed, which is not luck, it's what careful observation over enough seasons is capable of doing.
What I'm arguing is narrower than "traditional observation is unreliable," which it isn't. The method produces both kinds of output and delivers them in identical packaging, so the packaging can't be doing your filtering for you.
There's one filter that works, and it's a question. What would have to be true for this sign to work?
Not "is it true." That's the question everyone asks and it just gets you into an argument about whose aunt was right. Ask what the chain would have to be. Red sky: light has to be reddened by something, that something has to correlate with a pressure system, and that pressure system has to be moving toward you rather than away. Three claims, all checkable, and the moment you check the third one you also learn exactly where in the world the saying stops working, which is knowledge you did not have before you asked.
Cows: they'd have to detect something, and it would have to be something that changes reliably before rain and not at other times, and their response to it would have to be lying down specifically rather than any of the other things a cow does. Ask that out loud and you can hear it fall over. Nobody has to run a study. The sentence collapses under its own weight the second you make it say what it's claiming.
That question also tells you honestly when you've hit a case that hasn't resolved, which is what the joint pain literature is. There the chain is coherent, the claim is specific, people have gone and measured it repeatedly, and the results won't agree. Fine. "Undecided" is a real answer and there's no shame in stopping there.
Where this leaves the rest of it
This site spends a lot of its time on meaning rather than mechanism. The symbolic readings of natural things, what a storm or a river has stood for across different cultures, what people felt when a natural sign showed up at a moment that mattered. That work is worth doing and it isn't in competition with any of this.
It's a separate job, though. When a rooster's crowing is used to mark the hour it's functioning as an instrument, and the reason it works is a real circadian mechanism, which we've gone into properly over in the rooster. When the same rooster shows up in a story about vigilance, it isn't measuring anything and isn't supposed to be. Both are legitimate. They just answer to completely different standards, and a piece of lore will not tell you which one it's doing, any more than a rhyme will tell you whether it's carrying physics or a memory.
The sticking door doesn't tell you either. It swells in August and shrinks in February, it's been right about the air every day you've lived there, and you called it the building settling for years and you were never once punished for it, because a sign that works keeps working whether or not the explanation attached to it is any good.
So the two of them have been sharing one set of clothes all along, and the cost of that is plain enough to say out loud: reading nature is two completely different activities that look identical from outside, and the only thing that has ever told them apart is asking what would have to be true for the sign to work.
