During the transformation of manganite to pyrolusite the b-axis must contract about 15 %. There is evidence that the shrinkage takes place partly by contraction of the whole crystal, but electron-microscope investigations of natural pyrolusites and of manganite oxidized in the laboratory show that the reaction also involves the formation of lamellar pores about 85 Å apart, parallel to (010) of the original manganite. Thus, although pyrolusite is tetragonal, the a- and b-axes are not microscopically identical, a fact which could explain why, when natural pyrolusite is treated hydrothermally at 200 and 300 °C and 1000 bars, only one orientation of manganite is formed (Dent Glasser and Smith, 1968). The pores and the accompanying distortion of the structure act as a ‘memory’ of the parent phase.