If wireline log sensors are assumed to record responses which are a linear function of rock composition, sets of sensor responses can be used to estimate the mixture of components in subsurface formations. Since similar sensor responses may be caused by many different mixtures of rock, additional subsurface information must be supplied to determine the most likely combination of components. Several rules are shown to illustrate how different answers may be obtained. Once a unique solution is accepted, the mathematical model may be used to estimate the size and possible variation of error in each sensor response. A practical example is shown to illustrate how a linear programming system can readily give accurate estimates of reservoir porosity but may find a wide range of lithological components depending on the rules used for selecting a satisfactory solution.