ABSTRACT
Before actual data can be fit to a model, extraneous effects manifested in the trace must be removed, such as the shift in baseline as a result of the change in heat capacity of the sample during the transformation (see section 3.7.2). It may, for some device designs (e.g. post-type DTA), be difficult to purify the instrument output to represent only the latent heat from the transformation because of random baseline float. Hence, the data set fitting a particular model is a necessary but insufficient criterion for guaranteeing that the model describes the measured phenomenon.
