Skin detection is an essential preliminary step in many applications. Most skin detectors are color based. The distributions of skin and non-skin colors overlap, and so color cannot fully discriminate between skin and non-skin pixels. Skin detection is made more difficult by the need to be able to robustly detect skin in a wide range of illumination settings. Several color correction methods have been proposed to account for non-standard illumination. We measured the overlap of skin and non-skin distributions of the ECU dataset with the Bhattacharyya distance. We also evaluated the performance of a selection of color correction methods. It was found that the majority of the selected color correction methods cause a decrease in performance while the best provides a negligible improvement in performance. It was found that the Bhattacharyya distance is a useful measure of the effect of pre-processing methods on the performance of skin detectors.