The geometric reason for the non-existence of a MOL(6) (Q2094259)

A Latin square of order \(n\) is an \(n\times n\) matrix, with each entry from \([n]=\{1,2,\ldots ,n\}\) with every row and column containing exactly one occurrence of each of \(\{1,2,\dots ,n\}\). Two such Latin squares are said to be mutually orthogonal if when they are superimposed the ordered pairs of entries are all distinct. It is a famous old problem, originally posed by Euler (his 36 officers problem, with six of each of six ranks and six of each of six regiments), whether one can have two mutually orthogonal Latin squares of order \(6\), known as a MOL(6). \textit{G. Tarry} [Ass. Franç. Paris (1900) 29, 170--203 (1900; JFM 32.0219.04)] showed that one cannot and several other proofs have subsequently been produced, as described in the paper under review. The aim of the paper under review is to show the impossibility of a MOL(6) in a geometric way. Summarising crudely, a putative MOL(6), which has 6 rows, columns, first entries of each ordered pair, and transversals, is essentially equivalent to a certain linear space LSP(24). The key notion for the paper is a tactical decomposition of this linear space (hence also of the putative MOL(6)). The author previously shows that every LSP(24) has one of two tactical decompositions: one of these possibilities is ruled out, and so it must be the other case (called TDO II): however, then the author shows, in several ways, that in fact this structure cannot be extended to a MOL(6). One of the proofs (which uses two substructures of the putative MOL(6) avoids the use of a computer. Another relies on computing the automorphism group of the putative MOL(6).