A novel evolutionary method that allows us to study the emergence of modularity for genotype-phenotype mapping in the course of Darwinian evolution is described. The method is based on composite epigenotypes with two parts: a binary genotype; and a mapping of genes onto phenotype characters. For such generalized epigenotypes the modularity is determined in the following intuitive way: The genes are divided into two subgroups; simultaneously with this decomposition there is defined an accompanying decomposition of the set of phenotype characters. We expect that for epigenotypes with modular structures the genes from one group will be mapped onto characters from the same group, that is, that the appearance of crosslink mappings will be maximally suppressed. A fundamental question for all of evolutionary biology (and also for evolutionary algorithms and connectionist cognitive science) is the mechanism of evolutionary emergence of modular structures. The presented explanatory model is an implementation of the assumption that variation in genotype is produced on a faster time scale than variation in the genotype-phenotype mapped part. Moreover, the evaluation of the epigenotype in the evolutionary algorithm is based on directly selectable properties (corresponding to the decomposition of the set of phenotype characters). The modularity of genotype-phenotype mapping emerges in the simulations.