Gauss introduced a procedure for calculating least squares estimates a
nd their precisions. Yule introduced a new system of notation adapted
to correlation analysis. This paper describes these formalisms and com
pares them with the matrix and vector space formalisms used in modern
regression analysis.