Biological evolution involves the appearance of new traits, new phenotypes and new processes but the mathematical understanding of such novelties is still lacking. In this paper, I argue that the notion of a new possibility should be and can be a fundamental notion in biology. This idea has been used in several models but it still lacks a proper mathematical, theoretical and epistemological framework. The notion of new possibility should be analyzed carefully since, like for the notion of probability, points of view a priori and a posteriori cannot be conflated. Moreover possibilities are virtual and it seems at first sight that adding possibilities by a retrospective theoretical move does not change the state of an object. In order to define precisely the notion of new possibilities, I will first discuss a positive notion of possibility which pertains to properties that are properly defined theoretically. This will lead us to a more precise discussion than the set-theoretic language to describe possibilities and enable us to define a robust notion of new possibilities.