This paper revisits the ethical and political questions raised by feminist debates over essentialism. Reviewing these seemingly disparate debates, I identify in them a coherent history of engagement with 'essentialism' understood, in a relatively unified sense, as the belief that there are properties essential to women and which all women share. Feminists' widespread rejection of essentialism posed a well-known problem: it undermined feminist politics by denying women any shared characteristics which might motivate them into collective action. Re-evaluating two responses to this problem - 'strategic' essentialism and Iris Marion Young's idea that women are not a unified group but an internally diverse 'series' - I argue that are both are unsatisfactory, tacitly retaining essentialism as a descriptive claim about the social reality of women's lives. However, building on Young's idea that women should be reconceived as a non-unified sort of social group, I argue for understanding women to have a genealogy. Based on a reading of Nietzsche's concept of genealogy, I suggest that women always acquire femininity by appropriating and reworking existing cultural interpretations of femininity, with the result that all women become situated within a history of overlapping chains of interpretation. Because all women are located within this complex history, they are identifiable as belonging to a determinate social group, even though they do not share any common understanding or experience of femininity. I conclude that the idea that women have a genealogy can allow feminists to reconcile anti-essentialism with commitment to a coalitional politics.