The technique of using electron-paramagnetic-resonance g values to decompose the magnetic susceptibility into donor and acceptor contributions is discussed. New data on the angular dependence of the fractional susceptibilities in tetrathiafulvalene-tetracyanoquinodimethane (TTF-TCNQ) and hexamethylenetetrathiafulvalene-tetracyanoquinodimethane (HMTTF-TCNQ) are presented which provide additional evidence of the validity of the decomposition technique. These data also illustrate how to minimize errors and indicate an uncertainty in the fractional susceptibility on the TCNQ stack of 5% and 10%, respectively, for the TTF and HMTTF salts at 293 K. New g-value measurements on the Mott insulator HMTTF-TCNQF4 indicate that the HMTTF g value is virtually identical to that in the metal HMTTF-TCNQ, providing further support for the use of molecular g values. We show that the apparent paradoxes attributed by Conwell to the decomposition technique are either independent of this method or nonexistent. © 1980 The American Physical Society.