The physicist Richard Feynman, one of my heroes and inspirations, once said:

“When he [a historian] says Napoleon existed, or that the French Revolution was in 1783, he means that if you look in another book about the French Revolution, you’ll find that same date. 1789, maybe. That’s pretty accurate for a physicist to have the third decimal, that’s three figures.

The thing that he says is that he makes a kind of prediction about things he has never look at before: documents that have still to be found. He predicts that the documents, if  something is written about Napoleon will coincide with what is written in the other documents. And the question is how that is possible. And the only way to know that is to suggest that the past of the world was more organized in this sense than the present.”

I’ve often thought about my own wanderings in archives, how I have learned history from the primary sources, and how these sources build in my mind a view of the past. After you discover many, many sources that correspond with each other, it becomes more and more clear that the evidence of the past could not have been planted by some mischievous fellow in order to deceive you, that the evidence actually confirms itself again and again. The moment in which you discover a new source that confirms what you already knew from another source is a pleasing moment. But more important are those moments when we discover sources that seem to contradict what we already know. In that moment, your brain undergoes a sort of stress or reorganization and you try to make sense of the new information. Either the thoughts in your mind remain in tension, or you develop some new idea for how the past was structure, how two particular pieces of evidence might correspond.

Search for Feyman’s lecture on “The Distinction Between Past and Future.”