[1] We have tested the additivity of partial anhysteretic remanent magnetization (pARM) for suites of eight synthetic magnetites with mean grain sizes from 65 nm to 18 mm and 18 natural samples including lake sediments, oceanic and continental volcanic rocks, gabbros, and granites. In both synthetic and natural sample suites, domain states inferred from hysteresis and other magnetic properties vary from single-domain (SD) through pseudosingle-domain (PSD) to multidomain (MD). For each sample, total ARM intensity was compared with sums of partial ARMs of three different types: four conjugate pairs of parallel pARMs; four pairs of perpendicular pARMs; and one set of five neighboring parallel pARMs. In each case, the intervals of alternating field (AF) over which a steady field H was applied to produce the partial ARMs are nonoverlapping and cover the entire AF range (0-100 mT) used to produce the total ARM. Additivity of partial ARMs was verified to better than ±3% for all the samples, whatever the domain state (SD, PSD, and MD) or composition (ranging from pure magnetite to x = 0.6 titanomagnetite). The universality of pARM additivity is unexpected because its analog, partial thermoremanent magnetization (pTRM), deviates from ideal behavior as the grain size increases and the domain structure becomes MD. The different behaviors probably result from the fact that pARM is produced at ordinary temperatures over short times, whereas the most intense pTRM is produced at temperatures approaching the Curie point with significant dwell times, promoting such processes as isothermal remanence acquisition, domain nucleation, and domain wall reequilibration. Verification of the law of additivity of pARMs is an encouraging first step toward validating pseudo-Thellier and other methods of paleointensity determination that use ARM in place of, or in addition to, TRM.