One conclusion of mine from this figure is incorrect.
Chat pointed out the data he gave me (the lower left-hand figure)
already have Z=Z2-Z1 correct.
So these two figures disagree on the sign of M_f. 
Forcing the Bnu figure to have the same sign would then require the
resulting asymmetry to be diluted by a factor of 3. Not crazy, but a tricky
thing to interpret the resulting time reversal.