Cancer atlases frequently use relative scales which rank regional mortality rates relative to the countrywide mean (if relative risks are used) or median (if percentile colour schemes are used). This method has various disadvantages which limit comparability and interpretation. An alternative is absolute scaling which is well-known from geographic atlases: the natural rank order of altitudes is' mapped into a colour scheme which is preserved over all maps of an atlas. Applied to cancer atlases this means that the different magnitudes of mortality for different cancer sites and sexes is preserved and leads to the use of different ranges of a relatively wide common colour scheme. This technique is applied to the data of the German Cancer Atlas published in 1984. Its properties are outlined and it is shown that it overcomes many of the unfavourable characteristics of the convenient procedures.
