Knowledge in lost tomes isn't completely unheard of. For example, the Riemann-Siegel formula (an important formula about the Riemann zeta function) was found by Siegel in the 1930s when Siegel was going through old papers of Riemann's from the 1850s. This sort of thing was more common in the Middle Ages where Greek mathematical works helped jump-start our understanding. Sometimes also today in sociology and economics, useful data sources are found in unexpected places (Fink and Stark's work on early religion in America in part relied on discovering that the old US census data contained a lot more about religion than anyone had previously realized). But that's not really the same thing since it is just data, not theory. I think the RS formula is probably the best modern example of this sort of event.