Populations, Predictors, and Functions

Multilinear regressions Multipredictor nonlinear correlations Quantitative property-property relationships and quantitative structure-property relationships 

Composite material References Further Reading Wehsites Exercises 

The number of isomers of paraffin is very large see table 5.1. We see that the iso-paraffins are not as well investigated as the normal paraffins. We have the boiling points of all three isomers of pentane, but not the 75 isomers of deeane. It is inevitable that we have to resort to estimations. 

When we have obtained a good correlation for normal paraffins, we would naturally want to know if we can extend this to the branched paraffins, and onward to the population of all the saturated hydrocarbons (by including the cyclic paraffins), and onward to the population of all hydrocarbons (by including olefins, acetylenes, and aromatic compounds), and then onward to the population of all organic compounds (by including compounds with heteroatoms, such as O, N, Cl). A correlation that applies accurately to a larger domain is more useful than one that works only for a smaller domain. 

Another example is polychlorinated biphenyls (PCBs), which have 10 hydrogen atoms that can be substituted by chlorine atoms. There are three types of site the four a sites near the bridge between the two phenyl fragments, the four p sites farther 

---