In 1925, a French astronomer J. Mascart achieved the gigantic work of studying more than three thousand publications in order to sum up the available knowledge about the climate variability. He published his analysis and conclusions in a book. We base on this work to tackle anew some epistemological problems. In particular we weigh up the relevance of the popperian doctrine in the case of climate change. As eventually seen, the strict framework of falsifiability does not practically operate and the most rational position seemed to stake on improving measurements and scientific networking. We qualify this position as "narrowly scientific". The case of climate change is, in fact, representative of any question of knowledge whose assessments and controls are postponed to the future, which is the case of every theory attempting to say more than available measurements.
This leads us to reconsider the narrowly scientific position, a wait and see policy which does not take into account the interests involved in contradictory representations. We discuss a more engaged attitude for research centres that is better placed to anticipate possible dangerous uses.