We have tested the effect of spatial gradients in stellar mass-to-light ratio ( \Upsilon ) on measurements of black hole masses ( M _ { \bullet } ) derived from stellar orbit superposition models .
Such models construct a static gravitational potential for a galaxy and its central black hole , but typically assume spatially uniform \Upsilon .
We have modeled three giant elliptical galaxies with gradients \alpha \equiv d \log ( \Upsilon ) / d \log ( r ) from -0.2 to +0.1 .
Color and line strength gradients suggest mildly negative \alpha in these galaxies .
Introducing a negative ( positive ) gradient in \Upsilon increases ( decreases ) the enclosed stellar mass near the center of the galaxy and leads to systematically smaller ( larger ) M _ { \bullet } measurements .
For models with \alpha = -0.2 , the best-fit values of M _ { \bullet } are 28 % , 27 % , and 17 % lower than the constant- \Upsilon case , in NGC 3842 , NGC 6086 , and NGC 7768 , respectively .
For \alpha = +0.1 , M _ { \bullet } are 14 % , 22 % , and 17 % higher than the constant- \Upsilon case for the three respective galaxies .
For NGC 3842 and NGC 6086 , this bias is comparable to the statistical errors from individual modeling trials .
At larger radii , negative ( positive ) gradients in \Upsilon cause the total stellar mass to decrease ( increase ) and the dark matter fraction within one effective radius to increase ( decrease ) .