https://www.jstor.org/stable/2313748

I've read hundreds of papers in my mathematics career, and they – good and bad – almost always blend into eachother and turn into a diffuse fog in my memory, where they're identified by "that good one by Johnson-something-I-think.. no the 90s one I think".

There are very few exceptions. Kac's drum paper is one. It's so nicely written!