If there are more than a certain number (10?) of titratable protons on a molecule, then Marvin switches from showing individual protonation states to showing the distribution of charges (i.e. all configurations with the same net charge are grouped together). That would be fine, but Marvin doesn't take into account fixed charges when labeling the curves. So, I have a molecule with a fixed positive charge on a quaternary amine. Because Marvin doesn't take into account this fixed positive charge, the calculated charges are all off by -1. The curve labeled -2 should be labeled -1, the curve labeled -1 should be labeled 0, etc.