1. To test the colour pattern hypothesis ofE-vector orientation in the honeybee (van der Glas, 1975) menotactic orientation to unpolarized overhead patterns, in which the colours blend in a gradual transition from one to the other (Fig. 2; stimulus field: 17.7°), has been studied by observation of bees dancing on a horizontal comb. 2. When the patterns are of colours of short wavelength, dances are well oriented with relatively small variance but the angle between the indicated feeding place and the sun's bearing is usually transposed by 180° (Fig. 11). Therefore an unpolarized colour pattern evokes directional dances specifically oriented as if the pattern provided a clue to the sun's bearing. 3. In contrast, when the patterns are made up of bee white (including UV) of different brightness levels (with gradual transition between “light” and “dark”), dance orientation is poor. Hence orientation to the coloured patterns cannot be just a response to brightness or a result of differential stimulation of R9 or the other UV receptors in different overhead areas independent of other receptor types. 4. Evidence is given that the oriented response to the experimental overhead short wavelength colour patterns does not indicate that natural colour- and UV patterns in the sky are used. 5. The fact that bees do orientate to unpolarized colour patterns along with previous data on integrational processes between different types of spectral receptors inE-perception (van der Glas, 1975, 1976), are consistent with a perception mechanism for sky polarization by means of colour coding. On the other hand, the result of the bees' transposition by 180° of the angle between the indicated feeding place and the sun's bearing cannot yet be finally interpreted with any current model for celestial orientation.