Integral subschemes of codimension two (Q1304904)

The even liaison (or linkage) classes of codimension two subschemes of \({\mathbb P}^n\) have a common structure, first shown by \textit{E. Ballico}, \textit{G. Bolondi} and \textit{J. C. Migliore} [Am. J. Math. 113, 117-128 (1991; Zbl 0754.14032)]. [For curves in \({\mathbb P}^3\) this was also shown by \textit{M. Martin-Deschamps} and \textit{D. Perrin}, in greater detail: ``Sur la classification des courbes gauches'', Astérisque 184-185 (1990; Zbl 0717.14017).] This structure has been extended and applied in many papers, and the one under review is a very nice example. The basic idea of this structure is that there is a well-defined subset \({\mathcal L}^0\) of an even liaison class \(\mathcal L\), called the set of minimal elements, whose elements all have the same postulation and cohomology and lie in a flat family, and every other element of \(\mathcal L\) can be obtained from a minimal one by performing a sequence of basic double links and then a deformation through schemes all in \(\mathcal L\) which preserves cohomological dimensions. (A basic double link is a pair of links which in effect adds a complete intersection to the original element, thereby increasing the degree and other invariants.) This structure is called the Lazarsfeld-Rao property. In this paper the author's goal is to describe the integral elements of an even liaison class \(\mathcal L\). He defines domination in an even liaison class by saying that \(Y\) dominates \(X\) (not necessarily minimal) if \(Y\) can be obtained from \(X\) by a sequence of basic double links and then deformation. He then first considers a property weaker than integrality, namely that the scheme lie in the smallest degree complete intersection (of height two) allowed by the Hilbert function. He characterizes the elements of \(\mathcal L\) which deform to a subscheme having this property, and describes the structure of the set of such elements, which is similar to the Lazarsfeld-Rao property and is phrased in terms of domination. Then the author gives necessary conditions for a scheme in \(\mathcal L\) to be integral, in terms of numerical information about the scheme and the minimal elements. Furthermore, he shows that if \(Y\) satisfies these necessary conditions, \(Y\) dominates \(X\) and \(X\) is integral, then \(Y\) deforms with constant cohomology through subschemes in \(\mathcal L\) to an integral subscheme \(\bar Y\). He concludes with some examples and applications.