Ma · Mathematics

Proofs of Major Theorems Later Found to Be Wrong

Published By the 9to5Science editorial team

At a glance

  1. 1. Set the standard

    Count a published proof only when a specific flaw was identified years later.

  2. 2. Start with Kempe

    His 1879 proof of the four-color theorem was challenged by Heawood in 1890.

  3. 3. Separate near misses

    Wiles’s gap and Lamé’s failed approach were exposed too quickly to fit the time criterion.

A famous proof that was later found to be wrong is Alfred Kempe’s proof of the four-color theorem. Kempe published it in 1879, and Percy Heawood showed in 1890 that a key step did not work. That gives a clear interval of about 11 years between publication and the published identification of the flaw.

There are fewer unambiguous examples than a long list might suggest. “Unnoticed” is hard to establish: a proof may have been challenged privately, a gap may be a matter of rigor rather than a fatal error, or a later author may have repaired an argument without showing that the original claim was false. The examples below distinguish the strong case from well-known near misses.

Kempe’s proof of the four-color theorem

The four-color theorem says that every planar map can be colored with at most four colors so that regions sharing a boundary segment receive different colors. Equivalently, every planar graph has a proper vertex coloring with at most four colors.

Kempe’s proof appeared in 1879. Its approach used what are now called Kempe chains: connected chains of vertices colored with two selected colors. By swapping those two colors along a chain, one can sometimes free a color for a vertex that needs to be colored. This is a useful method, and Heawood later used related reasoning to prove the five-color theorem.

The problem was not the general idea of swapping colors, but a claim about how the swaps could be used together. In a particular configuration, the relevant chains can intersect in a way that prevents the color changes Kempe’s argument requires. As a result, the argument does not establish that every map can be colored with four colors. Heawood identified this flaw in 1890, eleven years after Kempe’s paper.

The flaw disproved the proof, not the theorem. The four-color theorem is true. It was eventually proved by Kenneth Appel and Wolfgang Haken in 1976, using a computer-assisted analysis; later proofs have also been developed.

EventDateWhat happened
Kempe publishes his proof1879The argument is presented as a proof of the four-color theorem.
Heawood publishes his criticism1890He identifies a failure in a key use of Kempe chains.
The theorem is proved1976Appel and Haken give a computer-assisted proof.

The dates measure the interval between the published proof and Heawood’s published criticism. They do not prove that no one privately suspected a problem earlier.

Why the list should be kept short

Several famous episodes are sometimes grouped with Kempe’s, but they do not meet the “wrong for at least a few years” test as cleanly.

Wiles’s first proof of Fermat’s Last Theorem

Andrew Wiles announced a proof in 1993. A gap was found during scrutiny, and Wiles and Richard Taylor repaired it; the corrected proof was published in 1995. This is a major example of a proof needing repair, but the gap was found within months, not after several years of acceptance.

Lamé’s proposed proof of Fermat’s Last Theorem

In 1847, Gabriel Lamé announced an approach based on factorization in a ring of cyclotomic integers. Ernst Kummer pointed out a serious obstruction involving unique factorization. The issue surfaced promptly, so this is not a case of a proof remaining unnoticed for years.

These cases are useful comparisons, but including them as qualifying entries would change the question from “proofs with long-undetected errors” to the broader “proof attempts that failed.”

Common mistakes when describing historical proof gaps

  • Calling the theorem false. A failed proof only shows that the argument does not establish the result. Kempe’s proof failed; the four-color theorem did not.
  • Treating a gap and a counterexample as identical. A flaw in an inference may invalidate a proof even when no counterexample to the theorem exists.
  • Claiming nobody noticed the flaw. Historical sources usually establish when a flaw was published or documented, not the first moment anyone suspected it.
  • Counting every incomplete early argument. Older standards of exposition and rigor differed. A later proof that fills in omitted details does not automatically mean the earlier argument was known to be wrong.

For a carefully supported short list, Kempe’s proof is the standout example. Other proposed entries should be checked for the publication date of the proof, the date and nature of the flaw, and whether the theorem itself remains true.

References

  1. On the geographical problem of the four colours — Alfred B. Kempe, American Journal of Mathematics, 1879
  2. Map-colour theorem — Percy J. Heawood, Quarterly Journal of Pure and Applied Mathematics, 1890
  3. The Four-Color Problem — Oystein Ore, 1967

Spotted an error? Send a correction. Read how we write and check articles in our editorial policy.