[This is a fast-access FAQ excerpt.]
The common misconception (which you will find in several books,
including the Dictionary of Misinformation) is that "proves" in
this phrase means "tests". That is *not* the case, although "proof"
*does* mean "test" in such locutions as "proving ground",
"proofreader", "proof spirit", and "The proof of the pudding is in
the eating."
As MEU says, "the original legal sense" of the "the exception
proves the rule" is as follows: "'Special leave is given for men to
be out of barracks tonight till 11.0 p.m.'; 'The exception proves
the rule' means that this special leave implies a rule requiring
men, except when an exception is made, to be in earlier. The value
of this in interpreting statutes is plain."
MEU2 adds: "'A rule is not proved by exceptions unless the
exceptions themselves lead one to infer a rule' (Lord Atkin). The
formula in full is exceptio probat regulam in casibus non
exceptis." [That's Latin for "The exception proves the rule in
cases not excepted."]
The phrase seems to date from the 17th century. (Anthony Cree,
in Cree's Dictionary of Latin Quotations (Newbury, 1978) says
that the phrase comes from classical Latin, which it defines as
Latin spoken before A.D. 400; but no classical citations have
come to our attention.)
[See a discussion of Cicero's use of it in 56 BC.]
Below are the five seventeenth-century
citations that we could find. 1, 3, and 4 are in the OED; 2 is in
Latin for Lawyers by E. Hilton Jackson and Herbert Broom; 5 is
in A Dictionary of the Proverbs in England in the Sixteenth and
Seventeenth Centuries, by Morris Palmer Tilley.
1. 1617 Samuel Collins, Epphata to F.T.; or, the Defence of the
Bishop of Elie concerning his answer to Cardinall Ballarmine's
Apologie 100: "Indefinites are equivalent to universalls
especially where one exception being made, it is plaine that all
others are thereby cut off, according to the rule Exceptio
figit regulam in non exceptis." [Note that figit rather than
probat is here used. Probo can mean any of "give official
approval to", "put to the test", or "demonstrate the verity of";
but figo can only mean "fix", "fasten", or "establish".]
2. The reports of Sir Edvvard Coke, Kt., late Lord Chief-Justice
of England (1658 edition; Sir Edward Coke died in 1634): "[...]
upon which Award of the Exigent, his Administrators brought a
Writ of Error; and it was adjudged, That the Writ of Error did
lie, and the reason was, Because that by the Awarding of the
Exigent, his Goods and Chattels were forfeited, and of such
Awards which tend ad tale grave damnum of the party, a Writ of
Error lieth, although the Principal Judgment was never given; in
this case, Exceptio probat regulam, & sic de similibus."
["A writ of error lieth" = "an appeal is admissible"; "exigent"
= writ of suspension of civil rights; ad tale grave damnum =
"to such great loss"; sic de similibus = "thus about similar
things".]
3. 1640 Gilbert Watts, Bacon's Advancement and proficience of
learning VIII. iii. Aph. 17: "As exception strengthens the
force of a Law in Cases not excepted, so enumeration weakens it
in Cases not enumerated." [So when Lewis Carroll wrote "I am
fond of children (except boys)", he affirmed his fondness for
girls more strongly than he would have had he written merely "I
am fond of children."]
4. 1664 John Wilson, The Cheats, To Reader: "For if I have shown
the odd practices of two vain persons pretending to be what they
are not, I think I have sufficiently justified the brave man
even by this reason, that the exception proves the rule." [The
OED (but not the other books I checked) gives the date as 1662.
As far as I can tell from this scant context, Wilson seems to be
saying, "My description of two cowardly cheats should serve to
show you the bad consequences of not being brave, and hence
convince you of the need for a rule: 'Be brave!'."]
5. 1666 Giovanni Torriano, Piazza universale di proverbi italiani,
or A Common Place of Italian Proverbs I, p. 80 "The exception
gives Authority to the Rule." note 28, p. 242 "And the Latin
says again, Exceptio probat Regulam."
To convince us that *in this particular phrase* "proves" originally
meant "tests", you will have to produce citations as old as or older
than these to support your view.