# Refinement

### Weights

The refinement is a least-squares procedure in which the sum of squared residuals
(SSR) is minimized. The SSR should be weighted. Ideally, the weight of each
residual should be the reciprocal of the variance of the corresponding measurement.
For this reason the program requires an estimate of the error
an a measurement. Weights can be calculated automatically using the error
estimate(s). The root mean squared weighted residual is called sigma.
If the experimental errors have been correctly estimated and if the model is
correct the ideal value for sigma, after refinement of the stability constants,
is 1, which would signify that the data have been fitted within experimental
error.

### Output viewer

During refinement a file is created which records the progress of the refinement.
This file is named output.res, and it is over-written each time a refinement
is performed. Once the refinement is completed, the output viewer will appear.
This shows the results of the refinement and also allows the calculation of
stepwise equilibrium constants. The file output.res also contains a copy of
the data used for the calculation and it may be copied and stored with a name
chosen by the user. Such a file can also be used as a data file

#### Evaluation of results

This is outside the scope of this step by step guide and the reader is referred
to the program documentation for more details. In general terms it can be seen
whether or not the fit is reasonable or not; if the fit is satisfactory the
stability constants will have been determined.

Contents