I think he's talking about the fact that OCaml has separate operators for floating point (i.e. in his example, you'd use 1.0 +. 2.0).