maybe this can help you
This sounds like you're asking about a binary floating
point format, though I'm not positive because of the
vocabulary you use. I interpret "characteristic" to mean
"exponent" and "significand" as "fraction" (or "mantissa").
If that's true, then you've described:
sign(1) + binary fraction(20) + binary exponent(10)
Note that there are other possible interpretations (for
example, a base-16 exponent).
Since I'm assuming a binary exponent, we can use the same
trick that IEEE 754 floating point standard does, and assume
an "implicit" leading "1" to the fraction portion. This
trick means that the fractional part will be between
the following numbers (in base 2):
1.00000000000000000000 (1 followed by twenty zeros)