The magnetic anisotropy energy K1, the energy to rotate the magnetization from the easy basal plane to the unique axis, has been measured in Ho2Co17 and Ho2Fe17, both by the magnetization curve and by neutron diffraction. Neutron diffraction studies have been taken to imply a K1 for Ho2Co17 a factor eleven times larger than that measured by magnetization measurements. For Ho2Fe17 the discrepancy may be as large, if one also takes K2 into account. We show that the two sets of measurements are compatible; what has obscured the interpretation of the measurements is the canting of the rare earth and transition metal moments, induced by the field and by the anisotropy torques.