Evolution You Can Watch

or, the argument is a dish of bacteria climbing a staircase of poison, and you can run it yourself

Want to skip ahead to the giant petri dish? Jump to the experiment ↓

There is a piece of film I keep wanting to play for people. A team at Harvard built a petri dish the size of a kitchen table — four feet long, two feet wide — and poured it full of a brown jelly that bacteria love to eat. Down its length they laid an invisible staircase of poison: a wide strip at each end with no antibiotic at all, then a strip with a little, then a strip with ten times that, then a hundred times, then a thousand. They dabbed ordinary E. coli — the same gut bacteria you are carrying right now — along the two drug-free edges, bolted a camera to the ceiling, and left it running for almost two weeks. What the time-lapse shows is one of the most quietly devastating things I have ever watched something without a brain do.

The bacteria spread inward as a creeping white tide until they hit the first wall of antibiotic. There they stop. The drug is doing its job; nothing can grow into it. For a while the front just sits at the line, stalled. And then — off in one spot — a tiny finger of growth pokes out into the poison, and behind that finger a whole fan of bacteria comes flooding into the new strip. A single cell, by a copying mistake, has stumbled onto a way to survive the dose, and its descendants — who inherit the mistake — pour into ground that just killed their cousins. The fan races to the next wall, ten times stronger. Stops. Waits. Another finger. Another flood. Eleven days in, bacteria that a thousand-fold dose would have killed instantly on day one are growing across it like it is water.

Nobody bred those bacteria. Nobody designed the resistant ones or taught them the trick. There is no chemist in the dish. The antibiotic did the choosing, exactly the way a drought chooses which finch eats and a dark tree trunk chooses which moth gets seen. Each wall is a problem the bacteria have no answer to, until pure random copying error hands one cell an answer — and because that cell’s children keep the answer and the drug deletes everyone who lacks it, the answer sweeps the strip in a day. Variation, heredity, selection. That is the whole of Darwin’s idea, performed in front of a camera by something with no plan and no idea it is starring in a movie.

I lead with that dish because most people who can’t make peace with evolution have one honest objection underneath all the others, and it deserves a straight answer rather than a sermon. The objection is: I have never seen it happen. And that is fair. The fossils ask you to trust a story about millions of years you will never live through. The great branching family tree of life asks you to trust a chain of inference you can’t hold in your hands. If that is where the case rested, a careful person would be within their rights to want more.

So set all of it aside — the deep time, the museum bones, the tree. This page is about the other kind of evidence: evolution caught in the act. Fast, repeatable, filmed, measured with calipers, frozen in a freezer, happening this week in hospitals and on windowsills and in twelve flasks in a lab in Michigan that have been bubbling along since 1988. You do not have to take any of it on faith. You can watch the entire mechanism turn with nothing but your own eyes — which is the point of the giant dish, and the point of the small one I have rebuilt for you below.

The Harvard plate is real. It is called the MEGA-plate, built in Michael Baym and Roy Kishony’s lab and published in Science in 2016, and the film of it is on the internet for free. I have rebuilt its logic here so you can run it as many times as you like, change the rules, and try to break it. Press play and watch the front climb the staircase. Then — this is the experiment that matters — turn the mutation rate to zero and watch what a world with no variation can do, which is nothing at all.

Run it a few times before you read on. Two things are worth catching. First, the pause at each wall is not the bacteria “working on the problem” — nothing is trying. The pause is just how long you wait for a lucky copying error to land in a cell sitting right at the frontier. Crank the mutation rate up and the pauses shrink, because lucky errors arrive sooner; this is exactly why hypermutator strains, which copy their DNA sloppily, evolve resistance faster in real life. Second, set the mutation rate to zero — the “control” plate — and the front simply parks at the first wall forever. No variation, no raw material, nothing for the drug to select. Selection is a sieve; it can only sort what variation already poured into the jar. Hand it nothing and even the hungriest sieve gets you nowhere.

If a dish that evolves in eleven days feels too fast to fully trust, here is the opposite extreme — evolution kept on ice and audited for thirty-five years. In 1988 Richard Lenski started twelve identical flasks of E. coli and began a daily ritual that is still going: every day, move a drop of each flask into fresh food; every so often, freeze a sample. The freezer is the trick. It is a literal fossil record you can thaw and bring back to life, so whenever something new shows up, you can rewind to the ancestors and replay the tape. The cultures have now run past seventy-five thousand generations — on a human timescale, deeper than the gap between us and the first Homo sapiens.

Around generation thirty-one thousand, one flask did something that is supposed to be hard. E. coli is practically defined by its inability to eat citrate when oxygen is around — it is a textbook line in the key used to identify the species. One population taught itself to do it anyway, and bloomed on a food source none of its eleven sibling flasks could touch. Because Lenski’s team had the frozen ancestors, Zachary Blount could replay the history and show the new trick was contingent: it only re-evolved out of samples from after a certain point, because it needed an earlier, quiet mutation to have loaded the gun first. That is evolution with receipts — not a story about what probably happened long ago, but a documented sequence you can rerun on demand.

Now the case the skeptics like to sneer at, because for a long time it was told without the proof — the peppered moth. Pale, speckled moth; pale lichen-crusted tree bark; the moth’s camouflage works and birds mostly miss it. Then the Industrial Revolution coats the trees of England in soot, the pale moths light up against the black bark like chalk, and within a few decades a dark form that was once a rarity takes over the soot-belt cities — then retreats again, just as predictably, after the Clean Air Acts scrub the bark pale once more. For years the doubters said: nice story, where is the gene? In 2016 they got it. The dark form traces to a single mutation — a chunk of “jumping DNA,” a transposable element, wedged into a gene called cortex — and by measuring how much the DNA around the insertion has been reshuffled since, the researchers dated the original event to about 1819, just as the smoke was beginning to rise. The textbook story turned out to check out down to the decade.

And if you want evolution you can put a ruler on, go to Daphne Major, a speck of rock in the Galápagos where Peter and Rosemary Grant measured finches for forty years. In 1977 a drought wiped out the small soft seeds; only big tough seeds were left; the finches with deeper, stronger beaks could crack them and the others starved — and the very next generation was born with measurably bigger beaks. Then, decades later, a wetter world and the arrival of a larger competitor flipped the pressure and drove the beaks back down. They did not infer this from fossils. They caught each bird, measured its beak in millimetres, and watched the average move year by year, in the direction the weather pointed it.

Here is the part worth saying plainly, because it is where the deep-time evidence stops asking for faith and starts behaving like the dish. Evolution makes predictions, and you can go check them. Neil Shubin reasoned that a creature halfway between fish and four-legged land animal should sit in rocks of a particular age, so he went to an Arctic island with rock of exactly that age and dug up Tiktaalik — a fish with the beginnings of a wrist — right where the theory said to look. You and every ape carry the same broken gene for making vitamin C, disabled in the same spot, which is why we get scurvy and most animals don’t. The same fossil scraps of ancient viruses are stitched into the same addresses in your genome and a chimpanzee’s. Anatomy drew one family tree of life; DNA, discovered later and independently, drew the same tree on top of it. These are separate witnesses who never met, telling the same story. That is not what a coincidence looks like. It is what a fact looks like from several directions at once.

I am not trying to win an argument here, and if you came in skeptical I would rather you stayed a little skeptical — that is the right instinct, kept pointed at the evidence. The reason I trust the engine is not that a confident person told me to. It is that I have watched it run, and you just did too. The bacteria do not need you to believe in them to climb the plate. The honest invitation is the same one this whole site keeps making: don’t take my word, take the controls. Set the mutation rate to zero yourself and watch evolution refuse to happen. Then turn it back up and watch it start. Nobody changed your mind. You ran the world and read the result.

And because this site holds to its own rule — that a thing recurring in nature is not a thing nature endorses — the dish is also a warning, and the prettiness of the demonstration should not soften it. The exact same staircase is being climbed, right now, in every hospital where we flood a body with antibiotics: the bugs that survive the dose breed the next, more resistant wave, and we are running Baym’s experiment at planetary scale without meaning to. Drug-resistant tuberculosis, MRSA, the gonorrhea strains that shrug off our last good options — these are not failures of the theory. They are the theory, working perfectly, against us. The same blind sieve hardens crop pests against the spray and tumours against the drug. The universe will keep performing this proof whether or not we are ready for what it costs. The least we can do, having finally watched it happen, is stop being surprised — and start being careful about which way we point the engine, since pointing it is the one part it leaves to us.

Things to try

If evolution never quite clicked for you — if you always nodded along politely while some part of you held back — I think it is because nobody ever let you run it. The mechanism is almost insultingly simple once you have watched it: things vary, the variations are inherited, and the world quietly deletes the ones that don’t make the cut, over and over, until what is left looks designed. No designer required, and none hiding. You have now watched bacteria climb a thousand-fold wall of poison in a dish, a moth’s mutation dated to the decade the smoke rose, a beak shrink and grow with the rain. The proof was never locked away in deep time. A lot of it is happening this week, in front of anyone willing to look.