Our understanding of human history is a battle between contingency and convergence. Do stable, long-term trends drive change? Or does history pivot on the tiniest details? We’re left to speculate between the two worldviews because we can’t experimentally test the past.
But what if you could create multiple worlds? And what if, within them, you could not just control what happens inside but also control time? Imagine the ability to play God, pressing pause at will, even rewinding and replaying key moments. That would give us a glimpse of the inner mysteries of cause and effect with unprecedented precision. We would finally know how change happens—and whether contingency or convergence reigns supreme. It’s an intoxicating thought experiment. But could it happen?
A few decades ago, a scientist named Richard Lenski realized it was possible without science fiction. Lenski, who sports an impressive Darwin-style beard, had been working as an evolutionary biologist, conducting fieldwork in rural North Carolina to study the predatory southern ground beetle. He began to wonder if experiments on evolutionary change could be run, not in the untamable wilderness, but instead in the controlled environment of a scientific lab. In 1988, Lenski launched one of the longest-running and most important experiments in scientific history.
Lenski’s experiment is elegant in its simplicity. Take twelve identical flasks, populate them with twelve identical strains of E. coli bacteria, feed them the exact same glucose broth, and let them get on evolving. Because E. coli reproduce rapidly, they pass through 6.64 generations per day. The average human generation lasts for 26.9 years, so one day in the world of these bacteria is roughly akin to 178 years of human time. It’s hard to believe, but since 1988, Lenski has directly observed evolution over 70,000 generations of E. coli, the human equivalent of 1.9 million years of change. In 2004, another remarkable scientist, Zachary Blount, joined Lenski’s lab. Together, they have long overseen twelve microbial universes, each swirling around in a flask.
I visited them so I, too, could gaze into these controlled universes. Lenski and Blount’s lab at Michigan State University is unremarkable. There are beakers, graduated cylinders, petri dishes, and white bottles of chemicals on packed shelves. Next to the door, Lenski points out a boxy incubator, set to 37°C, or 98.6°F, the same temperature as the human body. The incubator is humming as it slowly swirls and shakes a flask of microbes.
Our understanding of human history is a battle between contingency and convergence. Do stable, long-term trends drive change? Or does history pivot on the tiniest details?
Blount describes the experiment with enthusiasm. Every day, the bacteria in each of the flasks grow in an identical broth of glucose, or sugar, and citrate, better known as the “acid that gives orange juice its tang.” The tiny organisms swim in citrate, but can only eat glucose. Rather than having sex to reproduce, bacteria subdivide into two nearly identical daughter cells. Variation in the flasks, therefore, mostly comes from mutations, or little mistakes in DNA that occur during copying.
The genius of the experiment is that from one common ancestor, twelve different populations are free to evolve in identical conditions. The experiment has therefore eliminated sex, environmental change, and predators from the equation, allowing the scientists to observe evolution at its purest.
Lenski and Blount can therefore test whether contingency or convergence rules. If change is driven by convergence, then the twelve flasks should only have minor variations even over long periods. They might take a dozen different paths, but they’ll end up in roughly the same place. But if contingency dominates, the twelve populations should eventually diverge in substantial ways, as chance occurrences create microbial freaks, forever shifting evolution’s path.
Lenski and Blount also have something that most scientists do not: a time machine. E. coli can be frozen without harming it, allowing freezers to act like a pause button. To press play, just thaw the bacteria back out. From the beginning, Lenski and his team froze all twelve lines of bacteria every 500 generations, which meant they could replay any part of the experiment from any given snapshot in time. Want to create a bacterial replay starting from the day the Soviet Union collapsed or from September 11, 2001? No problem. In those twelve universes of broth, Lenski and Blount control time.
For more than a decade, the experiment seemed to be backing up the hypothesis of evolutionary convergence. The twelve cultures were different, as tiny changes were inevitable. But all twelve seemed to be mostly changing in similar ways. Each lineage of bacteria was getting incrementally better at eating glucose, becoming more “fit” in the Darwinian sense. There was a clear sense of order. The specific mutations didn’t seem to matter much. It was as though all twelve were following the same railway track, all racing toward the same destination.
For one line of bacteria, one tiny change meant that everything about their future changed, all because of a random mutation, made possible by four unrelated accidents.
Then, in January 2003, a postdoctoral researcher, Tim Cooper, arrived at the lab to tend to the twelve populations, just as he’d done hundreds of times before. This time, something was different. Eleven populations looked normal, “like flasks of water with a drop or two of milk mixed in, only their slight cloudiness indicating the millions of resident bacteria.” But the twelfth was wildly different. It was partially opaque, a cloudy mixture when it should have been mostly transparent and clear. “I thought it was a mistake,” Cooper told me. “But I was pretty sure something interesting was going on.
Cooper called in Lenski.
“I thought it was lab error,” Lenski told me. “Our motto in the lab to avoid contamination is ‘when in doubt, throw it out.’” Lenski decided to restart that line of bacteria from the last frozen sample. Thankfully, with their microbial time machine, mistakes could easily be corrected. A few weeks later, the same flask turned cloudy again. Clearly there had been no mistake. Something was going on.
Perplexed, the scientists sequenced the DNA of the E. coli in that opaque flask and found something incredible. The bacteria had evolved the ability to eat the citrate they were swimming in, which shouldn’t have been possible. In the twentieth century, there was just one documented case of E. coli that was able to digest citrate. That it had now occurred by happenstance was already an important discovery. But the story was about to get much more interesting.
To digest citrate, this “freak” line of bacteria had first undergone at least four unrelated mutations that provided no apparent benefit to the population—seemingly meaningless errors. But if those four mistakes had not all occurred, in that specific order, the fifth mutation, which gave them the ability to eat citrate, wouldn’t have been possible. Five contingent mutations were stacked on top of each other, and they were utterly improbable, too. Contingency all the way down.
Just how contingent were they? To find out, Blount spent years studying the freak population. He unthawed samples of the mutant lineage at various points, using the frozen bacterial fossil record to test whether the ability to eat citrate would emerge again. After analyzing roughly 40 trillion cells over nearly three years of experiments, he replicated the citrate mutation just seventeen times. But if he went back far enough into the bacteria’s evolutionary history, the citrate mutation never arose again. It was contingency, through and through.
To us, the world appears convergent, until we realize, with a jolt, that it isn’t.
To this day, after 70,000 generations—equivalent to 1.9 million human life years of evolution—only one lineage out of the twelve has developed the ability to digest citrate. For one line of bacteria, one tiny change meant that everything about their future changed, all because of a random mutation, made possible by four unrelated accidents. The other eleven bacterial universes are stuck eating glucose, blissfully unaware that they are swimming in, as Lenski puts it, a “lemony dessert.”
Blount argues that the Long-Term Evolution Experiment provides a sophisticated logic for thinking about critical turning points in human society. Many historians, for example, say that D-Day was the key to the Allied victory in World War II. If one could experimentally test that claim, historians would follow the same research design as Lenski and Blount. Imagine you had 1,000 identical Earths and could pause them at various points throughout the war. The logic would be that if the Allied victory became far more likely with worlds that started after D-Day, the historians could conclude that D-Day was the key turning point. But if the Allies won 75 percent of the time whether the world was thawed in June 1942 or June 1944, then it would be clear that the historians were wrong. D-Day didn’t matter so much. The Allies were always likely to win.
Sadly, there’s only one Earth, we can’t rewind time, and these contingency versus convergence experiments remain possible only with microbes in a science lab. For the moment, though, it seems that Lenski and Blount—and a much larger team of researchers who have worked on the Long-Term Evolution Experiment—have resolved the contingency versus convergence debate: to us, the world appears convergent, until we realize, with a jolt, that it isn’t.
Today, the Long-Term Evolution Experiment continues under the direction of Jeffrey Barrick at the University of Texas Austin.
Excerpted from Fluke: Chance, Chaos, and Why Everything We Do Matters by Brian Klaas. Published by Scribner. Copyright © 2024 by Brian Klaas. All rights reserved.
