Rational residuacity of primes (Q611887)

This article deals with generalizations of various rational quartic reciprocity laws to higher powers. Assume that \(p \equiv q \equiv 1 \bmod 2n\) are prime numbers that are \(n\)-th powers modulo each other. Let \(K_n\) and \(K_{2n}\) denote the subfields of the field \(\mathbb Q(\zeta_p)\) of \(p\)-th roots of unity with degrees \(n\) and \(2n\), respectively, and write \(K_{2n} = K_n(\sqrt{\beta}\,)\). Then the first result is that for all primes \(q\) coprime to \(\beta\) we have \((q/p)_{2n} = (\beta/\mathfrak q)_2\), where \(\mathfrak q\) is a prime ideal in \(K_n\) above \(q\). >From this general result, rational reciprocity laws can be derived by finding explicit generators \(\beta\). One such choice is given in the second theorem of this article: assume that \(p = a^2 + b^2\) and \(q = A^2 + B^2\) with \(aA\) odd, and set \(\beta_2 = pq + b(A^2 - B^2) + 2aAB\sqrt{p}\) (this choice comes from the reviewer's proof of Burde's reciprocity law in [Reciprocity laws. From Euler to Eisenstein. Berlin: Springer (2000; Zbl 0949.11002]). For \(t > 2\), define \[ \beta_t = (q\sqrt{p} + b(A^2-B^2) + 2aAB)\sqrt{\beta_{t-1}}. \] Then the authors claim that \(K_{2^t} = \mathbb Q(\sqrt{\beta_t}\,)\). It is easily seen that the only odd prime ramified in \(K_{2^t}\) is \(p\); the authors' proof that \(2\) does not ramify is not valid (and cannot be: for \(t\) large enough, say if \(2^t \nmid (p-1)\), the prime \(2\) must ramify since there will be no abelian extension of degree \(2^t\) ramified only at \(p\); one error is the invalid congruence \(\sqrt{p} \equiv \pm 1 \bmod 4\)), and the fact that the extension should be abelian is not addressed at all. In fact it seems that with the authors' choice of \(\beta_t\), the extension \(K_8/K_2\) is a pure quartic extension, hence cannot be abelian over the rationals.