Cohen-Macaulay local rings of dimension two and an extended version of a conjecture of J. Sally (Q1376309)

Let \((A,{\mathfrak M})\) be a local Cohen-Macaulay ring of dimension \(d\), embedding dimension \(N\) and multiplicity \(e\). By a classical result of \textit{S. S. Abhyankar} [Am. J. Math. 89, 1073-1077 (1967; Zbl 0159.33202)], we have \(e\geq N-d+1\) and if equality holds, the structure of the associated graded ring \(G:=\bigoplus_{n\geq 0}({\mathfrak M}^n/{\mathfrak M}^{n+1})\) is well understood and \(G\) itself is Cohen-Macaulay [see \textit{J. D. Sally}, J. Math. Kyoto Univ. 17, 19-21 (1977; Zbl 0353.13017)]. In the case \(e=N -d+2\), \textit{J. Sally} proved in J. Algebr. Geom. 1, No. 2, 325-333 (1992; Zbl 0779.13005) that \(G\) is not necessarily Cohen-Macaulay, the exceptions being the Cohen-Macaulay local rings of maximal type \(e-2\). The main open question there was about the possible depths of \(G\) and in fact Sally made the conjecture that \(\text{depth}(G)\geq d-1\) and gave strong evidence for this to be true. In Commun. Algebra 24, No. 13, 4249-4261 (1996), we proved this conjecture and also described all the possible Hilbert functions of \(A\). But very often we know that a given algebraic variety does not lie on a quadric or more generally does not lie on any hypersurface of degree less than a certain integer \(t\). In this case the bound given by Abhyankar is no more sharp, and \textit{J. Elias} and \textit{G. Valla} [J. Pure Appl. Algebra 71, No. 1, 19-41 (1991; Zbl 0733.13007)] gave the right bound, involving the initial degree \(t\), and studied the extremal case. It turns out that if we let \(h:=N-d\), then \(e\geq{h+t-1\choose h}\) and, if equality holds, \(A\) has maximal Hilbert function and \(G\) is Cohen-Macaulay. If instead we have \(e={h+t-1 \choose h}+1\), then \(G\) is not necessarily Cohen-Macaulay, the possible exceptions being again the Cohen-Macaulay local rings of maximal type \({h+t-2\choose t-1}\). In section 3 we prove that, also in this more general setting, \(\text{depth}(G)\geq d-1\) and describe all the possible Hilbert functions of \(A\). This gives a complete solution to the extended version of Sally's conjecture we refer to in the title. One of the more interesting aspects of our proof concerns what our methods show about the relationship between certain numerical characters of a Cohen-Macaulay local ring of dimension two, a topic to which section 2 is devoted.