The skull is a developmentally complex and highly integrated structure. Integration, which is manifested as covariance among structures, enables the skull and associated soft tissues to maintain function both across ontogeny within individuals and across the ranges of size and shape variation among individuals. Integration also contributes to evolvability by structuring the phenotypic expression of genetic variation. We argue that the pattern of covariation seen in complex phenotypes such as the skull results from the overlaying of variation introduced by developmental and environmental factors at different stages of development. Much like a palimpsest, the covariation structure of an adult skull represents the summed imprint of a succession of effects, each of which leaves a distinctive covariation signal determined by the specifi c set of developmental interactions involved. Covariance evolves either by altering the variance of one of these sequential effects or through the introduction of a novel covariance producing effect. Either way is consistent with the notion that evolutionary change occurs through tinkering. We illustrate these principles through analyses of how genetic perturbations acting at different developmental stages (embryonic, fetal, and postnatal) infl uence the covariance structure of adult mouse skulls. As predicted by the model, the results illustrate the intimate relationship between the modulation of variance and the expression of covariance. The results also demonstrate that covariance patterns have a complex relationship to the underlying developmental architecture, thus highlighting problems with making inferences about developmental relationships (e.g. modularity) based on covariation.2007 Tinkering: the microevolution of development. Wiley, Chichester (Novartis Foundation Symposium 284) p ••-•• I mean by this expression (correlated variation) that the whole organization is so tied together during its growth and development, that when slight variations in any one part occur, and are accumulated through natural selection, other parts become modifi ed. This is a very important subject, most imperfectly understood, and do doubt wholly different classes of facts may be here easily confounded together.