Counterexample to a conjecture of Elsner on the spectral variation of matrices (Q1611865)

The authors show by a counterexample that a conjecture in a paper of the reviewer [ibid. 71, 77-80 (1985; Zbl 0583.15009)] concerning a bound for the Hausdorff distance between the spectra of any two \(n\times n\) matrices does not hold. They are kind enough to call it a conjecture though it was claimed in the reviewer's paper that it can be proved. The counterexample, however, was not completely unexpected, as I soon after publication found an error in the (unpublished) proof.