A replication experiment tested the hypothesis that the size range of waste flakes from biface manufacture decreases from initial to final reduction stages and may be used to estimate biface reduction stages in prehistoric flake samples by comparison with replicated flake data. All waste flakes from the replication of twelve projectile points were separated into four experimentally defined stages, and were then sieved into ten size categories. The size distribution was based on the cumulative distribution function of flake frequency by size category, and was accurately modeled by the Weibull distribution. Constrained least squares analysis successfully assigned most single and multiple stage test samples to their correct stage of reduction. Based on careful replication, this method should allow the rapid, systematic estimation of biface reduction stages in unbiased prehistoric flake samples. The Weibull distribution may potentially serve as a screening device to help assure that a particular prehistoric flake sample represents the unbiased remains of biface manufacture. See also 21-189. -- AATA