Correspondence analysis (CA) A multivariate statistical method that estimates the position of a set of assemblages along an underlying gradient or dimension on the basis of the relative frequencies of types in each assemblage. The spacing of the assemblages relative to one another on the gradient is a function of how similar their relative frequencies of types are. Similar assemblages are close to one another. Very different assemblages are far apart. DAACS uses correspondence analysis to seriate assemblages and build intrasite chronologies. Assemblages that are close along the gradient estimated by CA to another are assigned to the same phase. The idea here is that the underlying gradient estimated by CA is correlated with time. When relative frequencies of each type are plotted against the estimated positions of the assemblages, the type frequencies should describe a unimodal, "battleship-shaped" curve.