Abstract By adopting a set of concepts stemming from linguistic science and artificial intelligence it is possible to for-mally represent a great deal of knowledge concerning rock art with the view to building an architecture which could simulate its comprehension. Knowledge acquired so far has enabled G. Sauvet and A. Wlodarczyk: 1) to formulate the hypothesis of semiotic motivation of Palaeolithic rock art, 2) to classify the animal representations , 3) to reveal the grammar underlying the composition of series of paintings. In order to make any progress it is necessary to enrich the knowledge collected by G. Sauvet in his database of Palaeolithic rock art by adding new observations of graphic forms. We propose a semiotic theory of art hypothesising that art is a code (a system of signs) in the same way as the language of literary works. The analogy that we establish between the pictural code of graphic works of art and the verbal code of literary works allows us to compare the rules of graphic composition with the rules of grammar of literary works and to consider these rules as an interface between Form (made up of distinct and significant ele-
Stored only when both papers are in the corpus — 0 citations exist upstream.