Gauss showed that least squares fails to produce a unique solution only whe
n the problem is indeterminate. This note considers his argument and the no
tion of indeterminacy underlying it. It also relates the argument to twenti
eth-century discussions of estimability and identifiability.