Pliny, 2 112 (or 2 109)
Dionysodorus is certainly less worthy of confidence; but I cannot omit this most remarkable instance of Grecian vanity. He was a native of Melos, and was celebrated for his knowledge of geometry; he died of old age in his native country. His female relations, who inherited his property, attended his funeral, and when they had for several successive days performed the usual rites, they are said to have found in his tomb an epistle written in his own name to those left above; it stated that he had descended from his tomb to the lowest part of the earth, and that it was a distance of 42,000 stadia. There were not wanting certain geometricians, who interpreted this epistle as if it had been sent from the middle of the globe, the point which is at the greatest distance from the surface, and which must necessarily be the centre of the sphere. Hence the estimate has been made that it is 252,000 stadia in circumference.
In fairness to the Romans, I could probably find popular "science" books today with a similar level of stupidity.
I was wondering how seriously we've considered storing useful information to improve the chance of rebounding from a global catastrophe. I'm sure this has been discussed previously, but not in sufficient depth that I could find it on a short search of the site. If we value future civilisation, then, it may be worth going to significant length to reduce existential risks.
Some interventions will target specific risky tech, like AI and synthetic biology. However, just as many of today's risks could not have been identified a century ago, we should expect some emerging risks of the coming decades to also catch us by surprise. As argued by Karim Jebari, even if risks are not identifiable, we can take general-purpose methods to reduce them, by analogy to the principles of robustness and safety factors in engineering. One such idea is, to create a store of the kind of items one would want to recover from catastrophe. This idea varies based on which items are chosen and where they are stored.
Nick Beckstead has investigated bunkers, and he basically rejected bunker-improvement because the strength of a bunker would not improve our resilience to known risks like AI, nuclear weapons or biowarfare. However, his analysis was fairly limited in scope. He focused largely on where to put people, food and walls, in order to manage known risks. It would be useful for further analysis to consider where you can put other items, like books, batteries or 3D printers, in an analysis of a range of scenarios that could arise from known or unknown risks. Though we can't currently identify many plausible risks that would leave us without 99% of civilisation, that's still a plausible situation that it's good to equip ourselves to recover from. What information would we store?
The Knowledge, How to Rebuild Civilisation From Scratch would be a good candidate based on its title alone, and a quick skim over i09's review. One could bury Wikipedia, the Internet Archive, or a bunch of other items suggested by The Long Now Foundation. A computer with a battery perhaps? Perhaps all of the above, to ward against the possibility that we miscalculate. Where would we store it? Again, the principle of resilience would seem to dictate that we should store these in a variety of sites. They could be underground and overground, marked and unmarked at busy and deserted sites of varying climate, and with various levels of security. In general, this seems to be neglected, cheap, and unusually valuable, and so I would be interested to hear whether LessWrong has any further ideas about how this could be done well.
Further relevant reading: Adaptation to and Recovery From Global Catastrophe, Svalbard Global Seed Vault (a biodiversity store in the far North of Norway, started by Gates and others).