Once measurements are taken, generally need to fit the data to find parameters such as critical exponents, using a multi-parameter fit to data which usually has large errors. This is very difficult!

For example requires a four parameter fit, and often have so much freedom that a wide range of values are possible. Usually need to assume , and get from other measurements.

Above formula is only true ``asymptotically close'' to .
How close is that? Not always clear.
May see ``apparent'' or * pseudo-critical* exponents,
which change or disappear as the size of the system is increased and
we move closer to .

There may also be corrections to the asymptotic formula, which are hard to distinguish from a small power-law exponent. In this case may need input from theory.

Paul Coddington, Northeast Parallel Architectures Center at Syracuse University, paulc@npac.syr.edu