It is shown that the intensity and stability of magnetization of reversed and normal Upper Tertiary lava flows from the United States of America can be summarized by average decay curves of very simple form. When first measured the intensity of normally magnetized specimens is one third greater than that of reversely magnetized specimens, and this is interpreted as being due to secondary viscous components (mean intensity 0.77 × 10−3 gauss) imposed by the recent field on a stable primary component (mean intensity 3.02 × 10−3 gauss). The differences between normal and reversed samples in low demagnetizing fields (<100 Oe) can be used to discover polarity in unoriented rock, and this could be useful in the study of dredge samples from the deep oceans, where, in order to test the hypothesis of ocean floor spreading it is desirable to relate the bottom rock polarity to the sign of the field anomaly observed at the surface. After demagnetization in alternating magnetic fields of 100 Oe (peak), the secondary magnetization disappears and the stable magnetization in normal and reversed samples is indistinguishable whether in its magnitude or its decay characteristics. This suggests that the mean intensity of the field at the time of extrusion was the same irrespective of polarity, and that the mineralogical variations (notably oxidation) affecting remanent intensity are the same in both normal and reversed rocks. The coherent decay of both secondary and primary components suggests a new method of determining intensity of the field in the geological past.