The use of French ‘à travers (de)’ for describing a movement along a curved surface: peripheral membership or anomaly?

The prototype theorie (Rosch 1978) describes classes .... (def). One of the interesting aspects of this theorie consist of the fact that categories are described as open classes which "can have indefinite margins" (MacLaury 1991). This allows to understand the existence of uses quite distant from the prototypical core.
The French prepositions 'à travers (de)' and 'au travers de' have a great number of different uses. An example of a very limited one (in time and in quantity) is the context where the moving entity (the 'figure') follows the surface of a curved reference entity (the 'ground'). This particular use (I only found tokens from the 16th Century) seems derived from a much more frequent one, where a moving entity runs across a flat reference entity, and describes consequently a horizontal trajectory.
In this paper I analyse these tokens in view of assessing if the prototype theorie offers an acceptable explanation for their emergence, or that they should rather be considered anomalies. Using the Frantext database, I will compare the tokens at hand with the semantically closest ones from the same period in order to show their fundamental differences. This use constitutes a sidestep which was almost immediately abolished, and thus does not seem to form a serious extension of the categorie.