In this paper, an evaluation of the theory formulated in part 1 for coals having 70-90% carbon quantitatively interprets these established trends: (1) weight loss is constant for coals of increasing rank through hvA bituminous, then falls sharply and becomes negligible for anthracites; (2) tar yields are nominally the same for lignites, then increase for coals of higher rank, reaching a maximum value for hvA bituminous coals; (3) tar yields diminish for medium- and low-volatile bituminous coals, although gas yields decrease even more abruptly, so the tar fraction increases monotonically for these ranks; (4) noncondensible gas yields decrease monotonically with increasing coal rank; and (5) the average molecular weights of tar decrease in proportion to the weights of the initial monomeric units in subbituminous through high-volatile bituminous coals, and continue to decrease for coals of higher rank. FLASHCHAIN also depicts the pressure dependence of yield enhancement by faster heating, for rates from 1 to 3000 K/s, and of tar yields, from vacuum to 6.9 MPa, for both low- and high-rank samples. In addition to continuous trends, the model predictions also display scatter, so that the loosely banded relation between yields and coal rank is evident. In particular, for different samples of the same nominal rank, the model represents the sensitivity of tar and total yields to measured values of H/C and O/C ratios. These aspects of parametric sensitivity also have an important practical implication. Predictions based on only the ultimate analysis and linear regressions of all other required input data are as reliable as those based on the complete set of coal-specific input data.