Frequency seriation has often led to valid ordering of archaeological data, yet critiques of the technique, and in particular of the construction of unimodal popularity (battleship) curves, have appeared periodically for almost 20 yr. Problems of type definition, collection manipulation, and chronological inference have been emphasized. It is here suggested that, above and beyond these difficulties of execution and interpretation, frequency seriation is methodologically unsound: that the battleship curve concept is self-contradictory; that a type/collection frequency has descriptive value only within a collection but has no comparative value between collections; that the most commonplace kinds of ceramic variation can negate the type/collection unimodal curve; that seriation of collections to conform to the self-contradictory battleship curve provides a mechanical method by which significant changes in ceramic complexes may be obfuscated. If this basic form of frequency seriation is methodologically invalid, then more complex schemes for manipulating type/collection frequencies using the same or equivalent assumptions (for example, the Brainerd-Robinson technique and others) are also methodologically suspect.