Brian Hayden argues that our analysis of Keatley Creek stratigraphy and dates offers inaccurate conclusions. Although our data demonstrate that the village appeared late and was somewhat unstable, Hayden continues to support a model of early emergence and high stability. Hayden offers no new data to support his position. Consequently, we argue that there is no reason to reject our revised view.

This paper provides an analysis of radiocarbon dates acquired during earlier and recent field seasons at the Keatley Creek site, southern British Columbia. Results indicate that early occupations predating 1900 cal. B.P. occurred, but were not likely associated with population aggregation and large housepits. The aggregated village appears to have emerged by approximately 1700 cal. B.P. and was abandoned at approximately 800 cal. B.P. A break in the occupational sequence is recognized at 1450-1350 cal. B.P. and one other short break may have occurred shortly after 1250 cal. B.P. Peak socioeconomic complexity appears to have been achieved between 1350 and 800 cal B.P. Climatic warming may have provided a selective environment favoring population aggregation and intensification during this time. The final abandonment of the Keatley Creek village appears to have been part of a regional phenomenon suggesting the possibility that climatic factors were important in this case as well.

Sullivan and Rozen's (1985) debitage typology has been proposed as a method for measuring the effects of variation in lithic reduction by describing “distinctive assemblages.” This is in contrast to many traditional analytical methods oriented toward identifying the effects of lithic reduction techniques on individual flakes. Debate over the use of the typology has focused primarily on the ability of the typology to accurately measure variation in lithic reduction behavior, and secondarily on the role of experimental studies in archaeology. In this paper I present an analysis designed to estimate the reliability and validity of the typology. An experimental design is developed to permit data collection with minimal analyst induced random or systematic error. Principal components analysis and the coefficient theta demonstrate that the typology provides reliable or replicable results when applied to debitage assemblages of similar technological origin. Further principal components analysis suggests that the instrument is of limited utility in recognizing effects of variation in reduction activities associated with highly vitreous lithic raw materials. A means of expanding the typology and increasing its accuracy in archaeological pattern recognition is presented.