5.7.1 Floating-Point Variables
  A floating-point variable is a named location that stores a
  floating-point value. The storage space required to hold the
  value depends on the variable's REAL subtype. For exam-
  ple, each SINGLE floating-point variable requires 32 bits (4
  bytes) of storage, while each DOUBLE floating-point variable
  requires 64 bits (8 bytes) of storage.

  Note that if any integer value is assigned to a floating-point
  variable, the value is converted to a floating-point number.