No label defined (Q4336739)

[For part I of this paper see ibid. 2, No. 2 (1993; Zbl 0847.14022).] Zariski's question (1971): Suppose \(F\) is defined by \(z^p=f(x,y)\) over an algebraically closed field of characteristic \(p>0\). Let \(\widetilde{F}\) be a non-singular projective model of \(k(F)\). Is it true that \(p_g(\widetilde{F})=0\) implies \(F\) rational? \textit{W. E. Lang} provides a counter-example to Zariski's question [Ann. Sic. Éc. Norm. Supér., IV. Sér. 12, 473-500 (1979; Zbl 0457.14015)] when the characteristic is 3. Remark (1993): The question is open for \(p>3\). Also it would be interesting to find an example of a hypersurface \(F:z^p=f(x,y)\) such that the corresponding Zariski surface \(\widetilde{F}\) has \(P_2(\widetilde{F})>0\) and \(p_g(\widetilde{F})=0\), where the degree of the defining equation of \(F\) is \(p+1\) and there are no singularities at infinity. We give an example of a surface defined over a field of characteristic 2 with the following properties: (i) \(F\) is an affine surface with equation of the form \(F:z^2=f(x,y)\) in affine 3-space. (ii) The closure of \(F\) in projective space, denoted \(\overline{F}\), has a nonsingular mode \(\widetilde{F}\) with \(p_g(\widetilde{F})=0\) but \(P_2(\widetilde{F})\geq 1\), so that \(F\) is non-rational. Our construction in characteristic 2 follows closely the classical construction of Enriques surfaces as double planes. However, everything had to be done anew in characteristic 2 in order to verify that the classical computations will still go through.