If a picture is' naturalistic', then the form of the image follows the form of the subject it represents. We know that this is true for the very many images, from many traditions of image-making, of which we have good informed knowledge. One formal approach to pictures of which we do not have inside knowledge is to suppose that the naturalistic rule also holds there: by observing the traits of the picture, we should be able to figure out the traits of the subject it represents. How secure is this reasoning? What general rules of picture-making may there be which will be instructive?

This chapter has come about because of a number of experiences which have challenged me. At a rockshelter on Mphunzi Hill in central Malawi I was contemplating the beauty of a depiction of a chameleon and marvelling at the powerful mixture of symbolic images that were juxtaposed at the site. Near the chameleon was a python and next to this a monitor lizard, all complex symbols in Chewa cosmology relating to the unified themes of rain, fertility and creation. My thoughts were interrupted by John (a locally born Malawian who was working with me) who asked:'Why do you say that thing is a chameleon? How do you know?'I was devastated. Drawing on knowledge of Chewa cosmology I was starting to see a link between the images and particular symbolic themes. This link demonstrated a connection between the art and Chewa female coming-of-age rituals. However, what was the use of this realisation if even John, a local, was not convinced that the shapes depicted the suggested subjects? The answer in my own mind was to show that the depiction had many of the attributes of a chameleon: the shape of its head, four legs, lizard-