Exceptions Prove What?

You know the saying "the exception that proves the rule". If you are a non-British European like myself, chances are that that this expression exists in your native language as well. To name just the tree that come to mind immediately:

  • German: Die Ausnahme bestätigt die Regel
  • Swedish: Undantaget som bekräftar regeln
  • French: L'exception qui confirme la règle

The problem is that these treat the word prove as meaning to give proof for or confirm. It can mean that, for sure, but it can also mean to put to the test! Swedish even has the word pröva, with presumably the same origin as prove, which means exactly that.

Is it presumptuous to think that this would have been a better translation? After all, it does not really make any sense that an exception should count as evidence for a rule. It should diminish our credence in it, a good counter-example can completely disprove a rule. It makes however perfect sense to think that an exception tests a rule. In fact I was happy when I realized this misunderstanding some while ago because I never liked the expression before.

Nevertheless, I wonder if there is some language-thing going on here that I am missing. Can it be a simple mistranslation (which inverts the meaning of the saying!) that made its way into the other languages? If so, why did the expression stick anyway? Does it appeal to some paradoxical mindset or Straussian subtext?

Interlude: Five minutes pass, with me being annoyed that what I just wrote does not feel right. Until I finally look it up.

Here is the actual meaning: By pointing to an exception that is explicitly part of the rule, the rest of the rule can be implied. Like a sign that says "no parking from 9-5" would tell you that it is allowed the rest of the time. Or the sign outside work saying "smoking allowed", thereby passive-aggressively telling smokers to not do so anywhere else.

This make some sense. But as far as I remember, this is not how the expression is commonly used. The paradoxical meaning is dominant, in the context of real counter-examples. But maybe I am wrong about that, too.