The tree on which the Angiosperm Phylogeny group based orders, etc., together with elaborations since, allow morphological (inc. chemical, anatomical) characters to be optimized (but cautiously) on the tree. Characterizations of well-supported lineages from genera groups upwards, and for all lineages at the family level and above, can be provided. These characterizations are part of a web-based resource that also includes literature, character definitions, synonymies, and photographs. Although only an interim solution to our changing understanding of phylogenetic relationships and to the absence of an archived database in which individual observations are linked to literature records (databases of character states would be inadequate), it provides a heuristic resource for both research and teaching that is archived and updated every six months. Characters and states can be considered in a local context, so allowing a more sensitive treatment of variation. Thus fruits need not be considered as "types", but as morphologies whose elements can be teased apart and treated independently. Interesting perspectives on the evolution of trimerous, pentamerous and bisymmetrical flowers, and of floral evolution in general, quickly become apparent - for instance, the plesiomorphic monocot flower differs in basic construction from that of both many Commelinid and core Eudicot flowers - as do differences in variation patterns within the units that we conventionally discuss as families. The questionable nature of the idea of "family variation" promulgated in ordinary texts becomes inescapable. Family characterizations become ever shorter as data are added and character states are removed to appropriate hierarchical levels, our knowledge of evolution thereby simultaneously increasing, and the many gaps in our basic morphological knowledge also become glaringly evident.

Key words: Angiosperm Phylogeny Group, angiosperms, clade characterisations, morphology, phylogeny