Theorem. If an equation f(x)=0 with real coefficients has a non-real root a+ib (b=0), then its conjugate a−ib is also a root, with the same multiplicity.
Proof idea. Divide f(x) by the real quadratic (x−a)2+b2=x2−2ax+(a2+b2), the minimal real polynomial having a+ib as a root:
f(x)=[(x−a)2+b2]Q(x)+Rx+S
for real numbers R,S (the coefficients of f and of the divisor are real, so ordinary long division produces a real quotient and a real linear remainder). Substituting x=a+ib makes the bracketed term vanish by construction, leaving 0=f(a+ib)=R(a+ib)+S=(Ra+S)+iRb. Since R,S,a,b are real and b=0, equating real and imaginary parts forces R=0 and then S=0. So the remainder is identically zero, i.e. (x−a)2+b2 divides f(x) exactly -- meaning a−ib is a root of f as well, with the same multiplicity as a+ib.
Consequence. Non-real roots of a real-coefficient polynomial always occur in conjugate pairs; an odd-degree real equation must therefore always have at least one real root. Once a single non-real root a+ib of an equation is known, its conjugate a−ib is known for free, and dividing out the real quadratic factor x2−2ax+(a2+b2) depresses the equation by two degrees at once.