Tim Radford at The Guardian:
Isaac Newton set it up for Albert Einstein: he calculated a system of heavenly motion that governed the entire measurable cosmos. He then added a challenge: a theory, he wrote “that agrees exactly with exact astronomical observations cannot fail to be true.”
He didn’t live to find out quite how much frustration that claim would give his fellow astronomers, who identified Uranus, and then from the behaviour of Uranus inferred the existence of another planet, and finally identified Neptune. They relied on Newton’s predictions, which were spot on and self-evidently right, all the way to the edge of the solar system – except for one tiny little niggling detail about the planet closest to the sun.
Mercury’s orbit precesses around the sun at a rate that cannot be fully accounted for by Newtonian mechanics. The discrepancy is very small, but it isn’t an observational error. And since Newton’s theory could hardly be wrong, the only answer that made sense was that there must be another small, invisible companion affecting the orbit of Mercury. So sure were astronomers that this companion planet must exist, they even gave it a name: Vulcan.
more here.