Three-point sets (Q5937497)

A planar set is called an \(n\)-point set (partial \(n\)-point set) if each line intersects the set in exactly \(n\) points (in at most \(n\)-points). It is known that \(n\)-point sets exist whenever \(n\geq 2\) is a natural number. It is an open problem whether there are some Borel \(n\)-point sets. The authors prove that no three-point set contains an arc. As a consequence they obtain that there is no \(F_\sigma\) three-point set. The proofs are done using nice elementary geometrical arguments. It is further proved that any partial \(n\)-point set can be extended to a \(k\)-point set whenever \(n\geq 2\) and \(k\geq n+2\). In particular, there is a four-point set containing a circle. The authors ask whether any three-point set is zero-dimensional. This question was answered in the positive by \textit{D. L. Fearnley, L. Fearnley} and \textit{J. W. Lamoreaux} [Proc. Am. Math. Soc. 131, 2241-2245 (2003; Zbl 1023.54014)]. As remarked in the note added in the proof, \textit{K. Bouhjar, J. J. Djikstra} and \textit{R. D. Mauldin} [Proc. Am. Math. Soc. 129, 621-622 (2001; Zbl 0959.54024)] showed that no \(n\)-point set for any \(n\geq 2\) is \(F_\sigma\).