One hundred years ago this month, Albert Einstein put the final polish on a new theory, one that transformed how humankind understands the fundamental nature of reality. With his general theory of relativity Einstein displaced the most famous idea in science, Newton’s theory of gravity, replacing the old idea of a force with a radically strange vision of a cosmos in which space warps and time bends.

When, four years later, during a total eclipse, measurements of starlight curving around the sun confirmed general relativity, Einstein became a global celebrity, and the first line of the catechism of science was reaffirmed: A single brute fact can destroy the most beautiful idea.

Richard Feynman once phrased that credo a bit more gracefully, writing that science gains its unique power to determine “whether something is so or not” through the commandment that “observation is the ultimate and final judge of the truth of an idea.” In one form or another, that’s what would-be scientists (and the rest of us) are told from a first-science-fair Coke-and-Mentos volcano onwards: Science advances because at every turn it is subject to the test of reality, the judgment of nature from which there is no appeal.

That’s what happened, or seemed to, when the British scientists who measured the path of starlight around the sun reported to a meeting of the Royal Society that they had observed a number that matched Einstein’s prediction and contradicted Newton. A single unequivocal observation had spoken: Light swerves along the contours of spacetime, and just like that, the 200-year-old Newtonian cosmos came crashing down.

There’s only one problem: It didn’t happen that way.

Albert Einstein had no need to wait four years for confirmation of his theory. From at least a week before he completed general relativity in its final form, he already knew that nature agreed with him. When he did his sums, what had seemed a tiny error in an obscure measurement could be completely accounted for by his theory. For him, that was enough: The general theory was the real thing.

At first glance, that’s just another example of how Feynman said science ought to work. But actually, the mystery that convinced Einstein had gone unsolved for over half a century—and no one, not even Einstein himself until the very end, had recognized the phenomenon for what it was: a decisive challenge to Newton’s whole approach. Instead, decades were spent in pursuit of a planet that by every reasonable measure should have existed, but didn’t.

The story of that missing planet begins with one that was and is very much present. A definitive analysis of the orbit of Mercury in 1859 had revealed a glitch. A tiny wobble, less than one part in 10,000 of the innermost planet’s track around the sun could not be explained by any known source of gravity within the solar system. Within the framework of Newtonian gravitation, the explanation was obvious: If every recognized body had been accounted for, then Mercury’s misbehavior could only be explained by something yet to be discovered, a planet between it and the sun.

First sight of the expected body, captured in transit across the face of the sun, came almost immediately, in December 1859. The new planet was so obviously necessary that there was no hesitation in naming it: Enter Vulcan. Astrophotography—the technique of attaching cameras to telescopes—was in its infancy, so this first observation was drawn and described, but to be confirmed, it would have to be repeated by someone else. No one did, but no matter. Professional and serious amateur astronomers would glimpse their version of Vulcan at least a dozen times over the next 20 years.

The final “Eureka!” came at the great American eclipse of 1878, when James Watson, the director of the Ann Arbor Observatory, recognized Vulcan in a small reddish object within a few degrees of the limb of the shadowed sun. Unfortunately, none of the other professional astronomers at the eight stations set up by the federal government to observe the eclipse saw anything out of the ordinary.

With that the scientific consensus came to rest: Each “discovery” had been a mistake; a sunspot, a mis-identified star, a wish. Vulcan had every right to exist. In Newton’s universe it had an obligation to be there. It wasn’t.

The next move was obvious, except no one dared make it: Could Newton be wrong? A few astronomers proposed ad-hoc solutions: Maybe the sun was fatter around the middle than believed (it’s not); perhaps there is an unseen halo of dust that could exert a gravitational tug on Mercury (there isn’t); maybe one could play with Newton’s numbers a bit to make all the sums work out (they don’t). But for the most part, for the next 30 years, Mercury’s rambles faded into obscurity. On one side, there was the most successful theory in the history of modern science. On the other, a tiny unaccountable anomaly. It was no contest.

The challenge to Newton did come, of course. In 1905, Albert Einstein published the special theory of relativity, which showed that the tick of time and the measurement of space must differ for observers in motion relative to each other. By 1907, Einstein realized that the logic of this first theory of relativity conflicted with the classical understanding of motion and gravity. For one example: In Newton’s view, the force of gravity leaps across empty space instantly, the sun’s tug grabbing earth with no time delay at all, while under Einstein’s relativity, nothing, not even force, can move faster than the speed of light.

There were other issues as well, but it was that kind of contradiction and no mere awkward observation that led Einstein to extend relativity into a theory of gravity. It would take him eight years, but finally, in November of 1915, he had got it: both the physical picture of a universe in which energy and matter deform space and time—and the mathematical framework that allowed him to calculate the paths matter-energy must take in this new cosmos.

And so, when Einstein had finally tuned his math to the point where he could calculate an actual example from the real world, he turned to the case of a planet traveling close to its star: Mercury. Sometime in the week between November 11 and November 18, he inserted the appropriate numbers and cranked through the equations. Twenty-four steps later and he had his answer. Mercury’s path, Vulcan-inspiring wobble and all, appeared on the page in all its glory—or, as Einstein wrote: “This theory agrees completely with the observations.”

With that, Einstein knew. He told one friend that on seeing Mercury drop out of his equations that he felt his heart stumble, and another that he was “beside himself with joy.” There was no need to wait for the eclipse—which is why he once said that if the British expedition had come back with the “wrong” numbers “I would feel sorry for the dear Lord. The theory is correct.”

A century on, we celebrate general relativity and Einstein’s re-imagining of how the universe organizes itself. Vulcan now rates barely a footnote to the history of astronomy. But it has its uses. Contrary to the myth of science, facts are not autonomous. They gain meaning from the frameworks within which human beings interpret them. It can be—it was for Vulcan—almost impossibly hard to see past what ought to exist to what does.

The decades Vulcan lasted as almost-real mark the distance separating our myth of scientific progress and the way science actually happens day by day. Its biography is perhaps the cleanest example of how hard it is in the midst of the fray to recognize the decisive observation, but it is hardly the only such case.

The strangeness of the geology and fossil evidence behind the theory of continental drift helped drive a half-century of resistance to the idea. Siddhartha Mukherjee documented in his book The Emperor of All Maladies how a fixation on the cure for a misconceived disease inhibited recognition of the complexity of cancer for a generation. It took decades before physicists came to grips with experiments that showed that the speed of light was constant for every observer—and even then, only the very young Einstein took that observation seriously enough to produce his first relativity theory.

In the long run, it’s true: Reality imposes a final and authoritative judgment on the rights and wrongs of any idea. In the moment, though, each moment, including ours, meaning in science emerges painfully, slowly, one fallible, historically contingent, self-deceiving and (very) occasionally triumphant scientist at a time. In other words, Vulcan’s brief brush with existence (1859-1915, RIP) is no mere curiosity. It’s a caution.