Also Puyumadamuk ‘blood’, Paiwandjaq ‘menstrual blood’, ma-djaq ‘menstruate’. This comparison presents an apparent contradiction, since it is in competition with *daRaq ‘blood’, a form that is widely attested in both Formosan and Malayo-Polynesian languages. The simplest explanation would appear to be that *daRaq was found in PAn in the meaning ‘blood’, and that *damuq is a later innovation spread by borrowing among Formosan languages. However, with the possible exception of Puyumadamuk, which may be an irregularly transmitted loanword from Paiwan, reflexes of *damuq show a geographical distribution within Taiwan that cannot easily be explained by borrowing. To reconcile these problems I propose that both words occurred in PAn, but with different meanings: *daRaq ‘blood’, *damuq ‘menstrual blood’. If this solution is correct Paiwan shows a reversal of meaning, reflecting *damuq in the meaning ‘blood’, and *daRaq in the meaning ‘menstrual blood’.