We have attempted to determine why samples of the Mio-Pliocene Purisima Formation are not amenable to thermal demagnetization. Analyses of the changes in direction, intensity, susceptibility, color, and saturation magnetization as well as examination of individual magnetic grains using a scanning electron microscope have all contributed to the development of a consistent hypothesis for the fate of iron during thermal demagnetization. In this particular case, we believe that oxidation-exsolution of ilmenite and then direct dissolution of nearly pure magnetite result from progressively higher heating levels. Our approach should be useful in understanding the origin and nature of the magnetization of many lithologies, including those for which thermal demagnetization is the method of choice. ¿ American Geophysical Union 1988