In developing progeny of mammals the two parental genomes are differentially expressed according to imprinting marks, and embryos with only a uniparental genetic contribution die [1][2][3] . Gene expression that is dependent on the parent of origin has also been observed in the offspring of flowering plants, and mutations in the imprinting machinery lead to embryonic lethality, primarily affecting the development of the endosperm-a structure in the seed that nourishes the embryo, analogous to the function of the mammalian placenta 4 . Here we have generated Arabidopsis thaliana seeds in which the endosperm is of uniparental, that is, maternal, origin. We demonstrate that imprinting in developing seeds can be bypassed and viable albeit smaller seedlings can develop from seeds lacking a paternal contribution to the endosperm. Bypassing is only possible if the mother is mutant for any of the FIS-class genes, which encode Polycomb group chromatinmodifying factors. Thus, these data provide functional evidence that the action of the FIS complex balances the contribution of the paternal genome. As flowering plants have evolved a special reproduction system with a parallel fusion of two female with two male gametes, our findings support the hypothesis that only with the evolution of double fertilization did the action of the FIS genes become a requirement for seed development. Furthermore, our data argue for a gametophytic origin of endosperm in flowering plants, thereby supporting a hypothesis raised in 1900 by Eduard Strasburger.Flowering plants (angiosperms) have evolved to be one of the predominant life forms on earth with more than 250,000 extant species. An important feature, and probably one of the main reasons for this evolutionary success, is the development of an embryo along with a second fertilization product, the endosperm. The endosperm is usually triploid because a homo-diploid female central cell fuses with one of the male gametes. The exact genome dosage of typically two maternal and one paternal genomes seems to be crucial for endosperm development. Raising the maternal contribution in the endosperm has been found to result in smaller seeds comprising fewer endosperm cells and smaller embryos. In contrast, increasing the paternal input results in larger seeds [5][6][7] . These results have been interpreted in light of the parental conflict theory (kinship theory) according to which mothers and fathers have a different interest in allocation of resources to their offspring 8,9 . Hence, it has been proposed that paternal genes promote seed growth, whereas maternal genes rather reduce growth; or, conversely, that in the maternal genome growth promoting factors are inactivated.Indeed, flowering plants have been found to imprint certain genes during endosperm development, which is similar to the situation observed for the placenta of mammals [10][11][12][13][14] . Interestingly, the imprinting machinery seems to regulate itself [15][16][17] . Imprinting involves the recognition of methylated DNA and histone...