> When I change the white point of the display the results of the following measurements of the white point don't change accordingly. So, it work as I thought: It assumes that the same RGB output should produce the same XYZ data and measurement deviation is compensated based on a reference point. The idea itself sounds good for me. (Like the ahead-back measurement road in the geodesy surveys which reduces the effect of the stochastic errors.) [Sorry for the mirror-translated patois. I hope you can understand...] But you should average at least three measures before you fix the reference point. (Or you can possibly catch the worst measure for the reference point.) And yes, it should be (automatically) disabled during the iterative manual hardware setup.