With the advent of high dynamic range imaging and wide gamut color spaces, gamut mapping algorithms have to nudge image colors much more drastically to constrain them within a rendering device's gamut. Classical colorimetry is concerned with color matching and the developed color difference metrics are for small distances. For larger distances, categorization becomes a more useful concept. In the gamut mapping case, lexical distance induced by color names is a more useful metric, which translates to the condition that a nudged color may not cross a name boundary. The new problem is to find these color name boundaries. We compare the experimental procedures used for color naming by linguists, ethnologists, and color scientists and propose a methodology that leads to robust repeatable experiments.