Up to now, I required MFIT = 2 for all tracks which are used for Ds single tagging. To see how many Ds events I lose, I repeated the single tagging for the phi pi and kstar k mode, using groups of 3 tracks where at least one track doesn't have MFIT = 2. The other cuts where the same as before. The plots show histograms of the 2-c fit chi square.
There seem to be 2 real phipi events (those with small chi2). However, there is only background in the kstar k mode.