The three classic arguments for major expansion of peridotite during serpentinization—(1) constant composition except for hydration, (2) internal deformation, and (3) lack of Mg- or Si-metasomatism in country rocks—are considered and found invalid in specific instances, and probably in general. Plots of about 350 analyses of fresh and serpentinized ultramafites show that major changes in the ratio of total bivalent oxides (MgO, FeO and CaO) to SiO2 and oxidation of iron yield an end product essentially composed of serpentine and magnetite from either dunite or pyroxenite. Structural relations of serpentinized rocks with interlayered or included fresh rocks and minerals, such as serpentinized dunite with pyroxenite, anorthosite, and chromite, prove that no change in volume has accompanied serpentinization, on hand-specimen and outcrop scales. World-wide association of calcium-rich rodingite with serpentinite indicates: that Mg-metasomatism of the kind that forms chloritic “blackwall” with talc does not occur in the serpentine mineral stability range; that Mg, Si, and alkalies in excess of the amounts needed to form serpentine are relatively fugitive elements; and that diffusion between ultramafite and country rock must be very minor compared to fluid or gas transfer of Mg and Si. Failure to recognize the chemical changes accompanying serpentinization has led to serious exaggeration of the original pyroxene content of alpine peridotite.