A fundamental riddle of evolutionary developmental biology is the conservation of adult morphological patterns (Hall, 1992). Conservative patterns are either called body plans if they concern overall body design, or homologues if they concern parts of the body (Riedl, 1978; Roth, 1982; Sattler, 1984; Van Valen, 1982; Wagner, 1989a, 1989b). An adult pattern is considered conservative if it remains unchanged in spite of changes in function, as indicated by the original definition of homology by Owen, as a similarity of organs regardless of form and function (Owen, 1848). Conservation of anatomical features despite different adaptive pressures is naturally explained by developmental constraints (Wagner, 1986). However, this approach to explain the biological basis of homology is plagued by the fact that developmental pathways are often more variable than the characters that they produce (see Tab. 1) (Hall, 1992; Roth, 1988, 1991; Spemann, 1915; Wagner, 1989b). This is also true for any other application of the concept of developmental constraints. The widely held opinion that early stages of development are conservative because any early perturbation is likely to interfere with later development, is far from absolute, since a vast amount of data in comparative developmental biology speaks to developmental variation (see e.g. the examples in Tab. 1). The question then is, how can developmental constraints on adult variation be reconciled with the fact of developmental variation?