Knowledge is less a canon than a consensus

David Shaywitz:

In 1870, German chemist Erich von Wolf analyzed the iron content of green vegetables and accidentally misplaced a decimal point when transcribing data from his notebook. As a result, spinach was reported to contain a tremendous amount of iron–35 milligrams per serving, not 3.5 milligrams (the true measured value). While the error was eventually corrected in 1937, the legend of spinach’s nutritional power had already taken hold, one reason that studio executives chose it as the source of Popeye’s vaunted strength.
The point, according to Samuel Arbesman, an applied mathematician and the author of the delightfully nerdy “The Half-Life of Facts,” is that knowledge–the collection of “accepted facts”–is far less fixed than we assume. In every discipline, facts change in predictable, quantifiable ways, Mr. Arbesman contends, and understanding these changes isn’t just interesting but also useful. For Mr. Arbesman, Wolf’s copying mistake says less about spinach than about the way scientific knowledge propagates.