Intrinsic knotting and linking of complete graphs (Q1597785)

In [\textit{J. H. Conway} and \textit{C. McA. Gordon}, J. Graph Theory 7, 445-453 (1983; Zbl 0524.05028)] it is proved that every embedding in \(\mathbb R^3\) of the complete graph on six (resp. seven) vertices, \(K_6\) (resp. \(K_7\)), is intrinsically linked (resp. intrinsically knotted), i.e. it contains a non-trivial link of two components (resp. a nontrivial knot). By making use of results and techniques developed in [the author with \textit{R. Naimi} and \textit{J. Pommersheim}, Topology Appl. 115, No. 2, 239-246 (2001; Zbl 0988.57003)] (where \(K_{10}\) is proved to be intrinsically triple linked), the present paper yields an interesting generalization of Conway and Gordon's Theorem: for every \(m \in \mathbb N,\) the author shows the existence of a suitable \(n \in \mathbb N\) (resp. \(r \in \mathbb N\)) such that every embedding in \(\mathbb R^3\) of \(K_n\) (resp. \(K_r\)) contains a link of two components whose linking number is at least \(m\) (resp. a knot \(Q\) with \(|a_2(Q)|\geq m\), where \(|a_2(Q)|\) denotes the second coefficient of the Conway polynomial of \(Q\)). \noindent Note that \(|a_2(Q)|\) results to be a convenient measure for knot complexity, since it relates knotting and linking [see \textit{L. H. Kauffman}, Topology 20, 101-108 (1981; Zbl 0456.57004) for details].