2000A&A...360...85M

The data from the surveys by Franx et al. (1989AJ.....98..538F), Peletier et al. (1990AJ....100.1091P), Goudfrooij et al. (1994A&AS..104..179G), completed by our own measurements, have been used to build a sample of ellipticals with the full array of U-B, B-V, B-R and V-I colours, and corresponding gradients. A number of objects with data for only 3 colours are also considered. Photometric profiles were re-calibrated in a coherent colours system, and a seemingly successful attempt was made to improve gradients by minor corrections to adopted sky background values. ``External" probable errors are obtained for the revised data by comparisons between the various sources. These have poor statistical weight in U-B and V-I. An ad hoc ``standard" gradient Δ_s is defined as a weighted average of the four gradients in the above listed colours (or only three). The correlations of the various gradients to Δ_s are considered, and their slopes Δ_U–B/Δ_s,... are compared to the corresponding calculated ratios for current hypothesis regarding the origin of colour gradients in ellipticals, i.e. population gradients and large scale dust distributions. Both the ratios for the U-B and V-I colours are not compatible with the explanation in terms of diffuse dust but tend to agree with the explanation in terms of metallicity variations. The correlation between the gradient Δ_U–B and the standard one Δ_s might be worse than could be explained by observational errors. Both population variations and possibly complex dust dustributions should be invoked to explain the complications of colour gradients.