The classic diffusion creep model predicts mass (atomic) transfer from grain boundaries that are normal to the high (compressional) stress direction to boundaries that are normal to the direction of low stress (Coble, 1963;Herring, 1950; Nabbaro, 1948). This process produces grains with anisotropic shapes that are often responsible for mineral lineations and foliations in deformed rocks (Passchier & Trouw, 2005). Based on experimental studies on fine-grained olivine aggregates, we recently reported the finding that diffusion creep is in fact diffusion-controlled (accommodated) grain-boundary sliding (GBS) creep (Maruyama & Hiraga, 2017a;, which corresponds to the current view of the majority of material scientists (Rust & Todd, 2011). In contrast, geologists commonly identify GBS creep and diffusion creep based on equiaxed grains and anisotropic grains with a shape-preferred orientation (SPO), respectively, in highly deformed rocks (Paterson, 2001;Vernon, 2004).Our series of experimental studies on fine-grained olivine aggregates revealed the formation of either equiaxed or anisotropic (tabular) grains via static grain growth, depending on experimental conditions (Maruyama & Hiraga, 2017a;Miyazaki et al., 2013;. Tabular grains develop interfaces (i.e., grain boundaries) parallel to particular low-index crystallographic planes. Directions parallel and normal to the boundaries correspond to long and short axes of the tabular grains, respectively. Preferential GBS occurs on low-index-plane boundaries, which promotes grain rotation to align the boundaries with the shear direction, resulting in crystallographic