How birds find their way

A bar-tailed godwit leaves a mudflat in Alaska in September, climbs, turns south, and doesn't land again until it reaches New Zealand. No islands, no rest, no feeding. One tracked bird covered around 11,000 kilometres in about nine days, flapping most of the way, and arrived having burned through roughly half its body weight.

It didn't have a map, and it can't have been following anyone — godwits set off in loose groups but the route is over open ocean with nothing to follow. Whatever it used, it worked to within a few hundred kilometres over a journey the length of a continent and a half.

Working out what "whatever it used" means has taken about seventy years, and a fair chunk of it still isn't settled.

How anyone knows any of this

Before you can ask how birds navigate, you need to know where they actually go, and for most of history nobody did. Aristotle thought redstarts turned into robins for the winter. As late as the eighteenth century there were serious proposals that swallows spent the cold months in the mud at the bottom of ponds, which sounds ridiculous now but was a reasonable guess when the alternative was that a bird weighing 20 grams flew to Africa.

Ringing settled it. Hans Christian Cornelius Mortensen, a Danish schoolteacher, started putting numbered metal rings on birds' legs around 1899, and the method spread fast. It's brutally inefficient — you ring thousands of birds and hear back about a handful — but each recovery is a hard data point linking one place to another, and over a century those points assembled into the migration maps everyone now takes for granted.

The modern kit is better. Light-level geolocators weigh under a gram and record sunrise and sunset times, from which position can be reconstructed to within a hundred kilometres or so, though you have to catch the bird again to get the data off. Satellite transmitters report continuously but are heavier, so they're limited to larger birds. Weather radar picks up mass movements at night. Networks of automated receiving towers track small tagged birds across whole continents.

The Arctic tern is the headline case, and it took geolocators to prove it. They breed in the Arctic and winter in Antarctic waters, and tracked Greenland birds turned out to be covering something like 70,000 kilometres a year once the looping, wind-assisted route was properly measured — far more than the straight-line distance anyone had assumed. A tern can live thirty years. Do the arithmetic and it's a few return trips to the Moon.

The experiment that split the problem in two

Here's the piece of work that organises everything else, and it's beautifully simple.

Displacement experiments are the basic tool of the field. You catch birds somewhere on their normal route, move them a long way off it, release them, and see what they do. If they carry on with the heading they had, they're following a rule. If they correct — turn towards where they'd have been going and compensate for the move — they've worked out where they are, which is a much harder trick.

In the 1950s the Dutch ornithologist A.C. Perdeck caught starlings migrating through the Netherlands, thousands of them, and shipped them to Switzerland before letting them go. The recoveries split cleanly by age.

The adults, who'd made the journey before, corrected. They headed north-west from Switzerland and turned up in their usual wintering areas around northern France and southern England. The young birds, on their first migration, didn't correct at all. They flew off on the same south-westerly bearing they'd have used at home, and ended up in southern France and Spain — displaced, wrong, and entirely consistent.

That result gives you the two-layer structure that the whole subject now uses. Young birds run an inherited programme: fly in this direction for roughly this long. It's often called clock-and-compass, and it works well enough to get a naive bird into the right general region, where it can then learn the details. Experienced birds have something more — a way of establishing position, usually called a map sense, which lets them correct for being in the wrong place.

Compass and map are different problems. A compass tells you which way is north. It won't tell you whether you're in Zurich or Amsterdam.

Three compasses, and they don't agree by accident

The compass half is far better understood, and birds have at least three.

The sun compass needs a clock, because the sun moves. To read a bearing off it you have to know the time of day and compensate, and Gustav Kramer showed in the 1950s that birds do exactly that. The test is elegant: keep pigeons for a week or two under artificial lighting shifted six hours from real local time, then release them. Their internal clock now says late morning when it's actually dawn, so they apply the wrong correction and set off around 90 degrees away from home. Shift the clock the other way and the error flips. That's about as direct a demonstration as biology offers.

The star compass turns up in night migrants, and Stephen Emlen worked it out in the 1960s with indigo buntings and a planetarium. The apparatus he used is still standard — a funnel-shaped cage lined with material that records the marks left by a hopping bird, so you can read off which direction it kept trying to leave in.

What Emlen found is the interesting bit. Birds aren't reading constellations. They're reading rotation. Raise young buntings under a planetarium sky that turns about some arbitrary star, and they'll treat that star as north for the rest of their lives. The centre of rotation is the one point in the sky that stays put, so a system that finds it doesn't care about precession, or which hemisphere it's in, or that the pole star will be a different star in ten thousand years. And it has to be learned, because the sky itself changes. Birds do this learning as nestlings and juveniles, watching the sky turn.

The magnetic compass is the third, and it isn't the kind of compass you're picturing. In 1972 Wolfgang and Roswitha Wiltschko showed that European robins don't respond to magnetic polarity at all. They read inclination — the angle at which field lines dip into the ground, steep near the poles and horizontal at the magnetic equator. So the bird's compass doesn't distinguish north from south. It distinguishes poleward from equatorward, which for a migrating bird is the more useful question anyway. Reverse the vertical component of the field in the lab and the bird reverses its heading, which shouldn't happen with a magnetite needle and does happen with an inclination detector.

Nobody has found the magnetic sense

This is the honest gap in the subject. The magnetic compass demonstrably exists, it's been measured in dozens of species, and after fifty years the receptor hasn't been identified with any confidence. Two candidate mechanisms are still in play and they might both be right.

The first is the radical pair mechanism, and it's chemistry rather than magnetism in the everyday sense. A photon hits a molecule in the retina — cryptochrome is the leading candidate, particularly the form called Cry4 — and knocks it into a state with two unpaired electron spins. The Earth's field is far too weak to move anything mechanically, but it can nudge the relative orientation of those spins, which changes how the pair recombines, which changes the chemical yield. Run that across a patterned array of receptors in the eye and you'd get a magnetic signal laid over vision.

The predictions are odd and they've held up. The compass should need light, and specifically shorter wavelengths — it does; robins orient under blue and green light and fail under red. It should be disrupted by very weak radiofrequency fields in the megahertz range, because those interfere with spin dynamics but would do nothing at all to a magnetic particle — and weak RF fields do disrupt orientation, which is one of the stronger pieces of evidence going. In 2021 a group showed that cryptochrome 4 taken from European robins is magnetically sensitive in a test tube, and more so than the same protein from chickens and pigeons, which aren't long-distance migrants. Suggestive. Not the same as showing it works that way in a living bird.

The second candidate is magnetite — crystals of iron oxide, which really do respond to a field mechanically, and which could in principle report field strength rather than just direction. That matters, because intensity varies with latitude and could feed a map rather than a compass. The trouble is that the most publicised evidence fell apart. Iron-rich cells in the pigeon's upper beak were reported as magnetoreceptive neurons and got into the textbooks; a careful study in 2012 showed they were macrophages, immune cells that accumulate iron, present in large numbers and in no consistent arrangement. Cutting the trigeminal nerve does still impair birds' ability to detect magnetic intensity, so something is being sensed somewhere. Where, and by what, isn't established.

Two mechanisms, one unproven and one embarrassed, for a sense that plainly works. That's the state of it.

Smelling your way home, allegedly

The map problem is where the arguments get bad-tempered.

Floriano Papi proposed in the 1970s that homing pigeons build an olfactory map: while sitting at the loft they learn to associate particular wind directions with particular smells, and when released somewhere unfamiliar they sample the air and infer roughly which side of home they're on. Pigeons made anosmic — by cutting the olfactory nerve, or plugging the nostrils, or applying zinc sulphate — do much worse at homing. That result has been reproduced many times.

What it means is another matter. The effects looked strong in Italy, where the work was done, and weaker or inconsistent in some other places, and critics pointed out that a bird with its nose interfered with might be stressed, disoriented or generally unwell rather than specifically lost. Supporters ran controls for that and kept getting the effect. Hans Wallraff spent years measuring actual atmospheric trace gases to check whether stable enough gradients exist to carry the information, and argued they do.

The current position is that olfaction contributes something real to the pigeon map and that nobody's certain how much, or how far it scales up to seabirds crossing oceans. Though for seabirds there's a separate and sturdier case: shearwaters and petrels have unusually large olfactory bulbs and can detect dimethyl sulphide, a compound released where plankton is being grazed. That's a way of smelling productive water from a long way off, and it's about finding food rather than finding home, but it establishes that a bird's nose works over serious distances at sea.

The last few kilometres are just looking

None of the clever stuff is needed at the end. Once a bird is somewhere it knows, it navigates the way you do walking home from a station — by recognising things.

Homing pigeons make this embarrassingly clear. Track them with GPS over familiar ground in Britain or Italy and a lot of them stop flying straight and start following linear features, including motorways and A-roads, sometimes going as far as taking the correct turning at a junction. It's slower than the direct route. They do it anyway, apparently because a habitual route is cheap to run once it's learned.

The general shape is a hierarchy. Long-range cues get you into the region, and a learned familiar-area map takes over for the final stretch. Which is also why the young birds in Perdeck's experiment weren't stupid. They hadn't built the map yet.

The cues are stacked, and constantly recalibrated

Every compass a bird has is unreliable in some conditions. The sun's no use under thick cloud. Stars vanish in overcast. The magnetic field is distorted near iron ore deposits and wanders slowly over decades. So the system doesn't rely on any one of them.

Cue-conflict experiments are how this gets probed: put a bird's magnetic compass and its celestial cues into deliberate disagreement, using coils around the cage, and see which one wins. The answer isn't fixed, which frustrated people for years. Sometimes the magnetic reading is corrected against the sky; sometimes the reverse.

The pattern that's emerged is that twilight matters most. Around sunset there's a band of polarised light across the sky at right angles to the sun, and its geometry gives an accurate reference for true north. Several species appear to use that moment to reset the magnetic compass before departing for the night. Nothing here is a single instrument. It's a set of rough instruments that check each other, and the checking happens at the times of day when one of them is temporarily excellent.

There's also evidence that the magnetic field acts as a signpost as well as a compass. Reed warblers and thrush nightingales held in a field matching a specific location on their route respond to it — changing how much fat they lay down, as if the field signature itself is the instruction to prepare for a long crossing ahead.

The programme is written in the genes, and it can change fast

If a first-year bird's route is inherited, it should be possible to breed it, and Peter Berthold did.

Blackcaps breeding in central Europe used to split at a migratory divide, some heading south-west to Iberia, others south-east. Berthold's team put birds from the two populations in orientation cages and got the two headings. Cross them, and the offspring pointed in between. Whatever's being inherited isn't a memory or a lesson from the parents — those birds were hand-raised — it's a direction, with a rough duration attached, encoded well enough to be blended by crossbreeding.

The same species then did something more startling. From around the 1960s, a growing number of central European blackcaps began wintering in Britain and Ireland instead of Spain, a north-westerly heading that made no sense until you notice how many British gardens now put out food all winter. The route is shorter, the birds get back to the breeding grounds earlier and take the better territories, and orientation tests confirmed the new population really does prefer the new bearing. That shift happened over a few decades. An inherited navigation programme isn't a fixed inheritance from deep time; it's a trait under selection, and selection can move quickly when the payoff changes.

What's still missing

Be clear about the shape of the ignorance, because it's specific rather than general.

The compasses are solid. Sun, stars and magnetic inclination are all demonstrated, the developmental story for the star compass is worked out, and the clock-shift experiment does what it's supposed to every time. The receptor for the magnetic sense is missing, and that's a genuine hole — a sense with no known organ.

The map is worse. Nobody can currently say what set of variables a displaced adult bird is reading to conclude it's 600 kilometres east of where it should be. Magnetic intensity and inclination are plausible components, since both vary geographically, but the gradients are shallow and noisy and the required precision looks uncomfortable. Smell is in the mix, at least for pigeons. Infrasound has been proposed to explain why particular release sites confuse pigeons in ways nothing else predicts. None of these is settled.

And it's worth resisting the pull towards making this a story about intention. The godwit isn't consulting anything or deciding to go to New Zealand. It's an animal running an inherited programme, correcting it against several noisy sensory channels, and refining it with experience if it survives the first trip. Plenty of them don't.

What's remarkable isn't that a bird knows where it's going. It's that this much reliability comes out of instruments that are, individually, quite bad.