AbstractEleven color‐difference formulae were tested for medium and large color differences in a psychophysical experiment on color scaling. A preliminary report has been published and one more formula has been added since then. A third experiment is included in this report dealing with variations of the field size and background. The results were tested statistically for conformance to three different types of mathematical law (linear, exponential, and potential) and with a few exceptions, linear approximation showed the higher regression coefficient. For color differences as large as 17 ± 11 CIELAB units, this experiment suggests that the CIELAB formula is a good recommendation, but it is no better than the new proposal of Richter or the well‐known cube root and Adams‐Nickerson fo