R. A. Fisher's account of the decline of inverse probability methods during the latter half of the 19th century identifies Boole, Venn and Chrystal as the key figures in this change. Careful examination of these and other writings of the period, however, reveals a different and much more complex picture. Contrary to Fisher's account, inverse methods--at least in modified form--remained theoretically respectable until the 1920's, when the work of Fisher and then Neyman caused their eclipse for the next quarter century.