“…Using 2-transitivity, find an automorphism mapping (X i , Y i ) to (X i , Y i ) and (X j , Y j ) to (X ′ j , Y j ). As above, we decompose it into automorphisms of the forms (1) and (2) and regard it as an element of a one-parameter family of automorphisms with t = 1 (not a subgroup!). We want all the matrices to remain diagonalizable, this forbids a finite number of values of t. We do not want to break edges that were constructed earlier, so for each old edge kl, as we did in Remark 2, we express the condition that (X k and X l have no common eigenvalue)&(Y k and Y l have no common eigenvalue) as a polynomial condition on t that holds for t = 0.…”