If, in sampling from an existent population, we wish to know the regression ofuonv, a line is fitted by least squares, minimizing the sum of the squared residuals ofv(ory). If there is no error in measuring the independent variateu, regardless of whether the dependent variate has been measured without error asvor with error asy, the regression is an estimate of the true regression. If, however, the independent variate is measured with error inx, the regression obtained is a biassed one. The bias will characterize the fitted line, and will be present whether the sample is a random one of the entire population or has been taken at preassigned selected values ofx.