PL> What about a long double? Shouldn't it be able to hold over 3000 decimal
PL> places???

Let's see, a 32-bit integer is accurate to 1/(2**32), ie too small for my
calculator to register.  A "long double" is 8 bytes in Watcom C
for OS/2.  Therefor it is accurate to 1/(2**64).  If you can calculate
1/(2**64), and it comes to (say):

0.0000000000000000000000000001567

Then the first non-zero digit appears in location 27, so it is accurate to
27 digits (plus or minus 1 or 2 of course).  BFN.  Paul. 
@EOT:

---