Worked solution: Wantzel's algebraic impossibility proof for doubling the cube (1837)
Eisenstein's criterion is a quick divisibility test: pick a prime number and check that it divides every coefficient of the polynomial except the leading one, and that its square does not divide the constant term. If such a prime exists, the polynomial cannot be split into smaller integer-coefficient pieces — a bit like how a padlock with one very specific, non-repeating key cannot be picked apart into two simpler locks.
For the prime does exactly this job almost by accident: the polynomial's only 'hidden' coefficients (besides the leading ) are , and divides all of them but does not divide .
Eisenstein's criterion states: for an integer-coefficient polynomial , if some prime divides every coefficient but not the leading coefficient , and does not divide , then the polynomial is irreducible over .
Apply this to , with coefficients . Take : it divides , , and , but does not divide the leading coefficient ; and does not divide . All conditions hold, so is irreducible over (this reasoning appears already in Wantzel 1837, §III, in an equivalent form predating Eisenstein's later general statement of the criterion in 1850).
Since is the minimal polynomial of and has degree , . This is exactly the number needed for the final step: it must now be compared against the list of allowed degrees from Step 2.
- Eisenstein's criterion
- A sufficient test for irreducibility over : if a prime divides every coefficient of a polynomial except the leading one, and does not divide the constant term, the polynomial cannot factor into lower-degree integer-coefficient polynomials.