float_type

float_type 
      ::= 
      "number" 
      | "float" | "real" | ( "double" "precision" ) 

This rule is called by

datatype