The slope of a straight line: a phony estimator [plus comments by M. Werman]

This note discusses an estimator for the slope of a straight line as proposed by Werman and Geyzel ("Fitting a second degree curve in the presence of error", ibid., vol. 17, no. 2, p. 207-11, 1995). By computing its frequency distribution, the authors demonstrate that the estimator is a unreliable; it possesses neither a first nor a second moment. Application leads therefore to completely erratic outcomes. Werman replies that the problem is well known, and describes the methods taken to overcome it.