Recent experiments on MnF₂ (King and Rohrer, 1976, 1979) suggest Q=1.56+or-0.35 for the amplitude ratio determining how the critical lines meet at the spin-flop bicritical point, in serious disagreement with the original prediction of Q approximately=2.5>2.2 for universal Heisenberg (n=3) bicriticality. The authors report extensive two-variable series calculations using constrained partial differential approximants, which yield Q=2.34+or-0.08. In addition it is shown that the natural least-squares fitting procedures may be strongly biased even for data of high precision.