To use the types single, double, extended, and comp
           you must use an numeric coprocessor.

    Type                    Range                     Decimal Digits
   --------------------------------------------------------------------

                             -39            38
     Real            2.9 x 10    to 1.7 x 10           11 to 12

                             -45            38
     Single          1.5 x 10    to 3.4 x 10            7 to  8

                             -324           308
     Double          5.0 x 10    to 1.7 x 10           15 to 16

                             -4951          4932
     Extended        1.9 x 10    to 1.1 x 10           19 to 20

                             18             18
     Comp (integer) -9.2 x 10    to 9.2 x 10

    The ( {$N-+} ) compiler switch is used to select the way the compiler
    uses numeric values.