The usual formulation of the portfolio selection problem through meanvariance analysis assumes that the variance-covariance matrix of the rate of returns on risky assets is non-singular. In view of the literature discussing the creation of riskless portfolio from carefully balanced quantities of risky securities (e.g., shares and warrants as in Black-Scholes [1]), the assumption of non-singularity may be challenged. Consequently, the proofs of the classical theorems of portfolio management may no longer be developed as originally presented. Buser [2] presents a means of using a singular variance-covariance matrix in the derivation of portfolio weights, which, although slightly flawed by a mathematical error, still provides some interesting insights about the efficient frontier. We present a corrected version of Buser's method in Section II; in Section III, we comment on some of the implications of his presentation and indicate some more precise results.