This document discusses the concept of variation in causal discovery and reasoning. It argues that causal discovery relies on the analysis of differences and variations. There are three main types of causal reasoning - ordinary, experimental, and statistical - but they all depend on the evaluation of variations, whether comparing counterfactual scenarios, manipulating putative causes, or studying associations across different levels of exposure. The foundations of this variational approach are traced back to thinkers like Mill and Durkheim, who emphasized the importance of comparing instances where phenomena do and do not occur to establish causal relations. Regularities alone are not sufficient for causal learning; we need variations to detect the factors that make a difference.