Counterexamples to conjectures on 4-connected matroids (Q1821108)

\textit{W. T. Tutte} proved that a matroid is binary if and only if it has no minor isomorphic to \(U_{2,4}\), the 4-point line [J. Res. Nat. Bur. Standards, Sect. B 69, 1-47 (1965; Zbl 0151.338)]. For \(k\in \{2,3\}\), \textit{R. E. Bixby} [Discrete Math. 8, 139-145 (1974; Zbl 0279.05027)] and \textit{P. D. Seymour} [Combinatorica 1, 387-394 (1981; Zbl 0489.05020)], respectively, proved that every (k-1)-element set in a k-connected non- binary matroid M is in the ground set of some \(U_{2,4}\)-minor of M. In addition, Seymour conjectured that this result is also true when \(k=4\). This paper gives a 12-element counterexample to Seymour's conjecture. Another counterexample, this one on fifteen elements, was found independently by \textit{J. Kahn} [Combinatorica 5, 319-323 (1985; Zbl 0593.05016)]. His example was obtained from a binary matroid by declaring one circuit-hyperplane to be independent and he asked whether every counterexample can be obtained in this way. The author's 12-element example shows that this is not so. This paper also provides a 16-element counterexample to a conjecture of \textit{Robertson} [Ann. Discrete Math. 8, 247 (1980] that if a, b and c are elements in a 4-connected non-graphic matroid M, then M has a circuit containing all three of these elements.