Four new nuclear star cluster masses , M _ { nc } , plus seven upper limits , are provided for galaxies with previously determined black hole masses , M _ { bh } .
Together with a sample of 64 galaxies with direct M _ { bh } measurements , 13 of which additionally now have M _ { nc } measurements rather than only upper limits , plus an additional 29 dwarf galaxies with available M _ { nc } measurements and velocity dispersions \sigma , an ( M _ { bh } + M _ { nc } ) – \sigma diagram is constructed .
Given that major dry galaxy merger events preserve the M _ { bh } / L ratio , and given that L \propto \sigma ^ { 5 } for luminous galaxies , it is first noted that the observation M _ { bh } \propto \sigma ^ { 5 } is consistent with expectations .
For the fainter elliptical galaxies it is known that L \propto \sigma ^ { 2 } , and assuming a constant M _ { nc } / L ratio ( Ferrarese et al .
) , the expectation that M _ { nc } \propto \sigma ^ { 2 } is in broad agreement with our new observational result that M _ { nc } \propto \sigma ^ { 1.57 \pm 0.24 } .
This exponent is however in contrast to the value of \sim 4 which has been reported previously and interpreted in terms of a regulating feedback mechanism from stellar winds .

Finally , it is predicted that host galaxies fainter than M _ { B } \sim - 20.5 mag ( i.e .
those not formed in dry merger events ) which follow the relation M _ { bh } \propto \sigma ^ { 5 } , and are thus not ‘ pseudobulges ’ , should not have a constant M _ { bh } / M _ { host } ratio but instead have M _ { bh } \propto L ^ { 5 / 2 } _ { host } .
It is argued that the previous near linear M _ { bh } – L and M _ { bh } – M _ { spheroid } relations have been biased by the sample selection of luminous galaxies , and as such should not be used to constrain the co-evolution of supermassive black holes in galaxies other than those luminous few built by major dry merger events .