On the mechanism of nucleation and growth of (111) tabular silver halide crystals, it is proposed that twinning (nucleation) occurs by stacking fault and that the anisotropic lateral growth is brought about by the ridge trough structure on tabular grain side faces. Although the “coalescence” model for nucleation and the “rough-smooth” side face structure for anisotopic growth were proposed recently, they have not yet been supported by direct experimental evidence. We analyzed the nucleation process by a combination of light scattering and X-ray diffraction and examined the coalescence of untwinned fine grains after decomposition of gelatin. These experimental results provided evidence to support the coalescence model for twinnning. We observed by TEM imaging of ultramicrotomed sections of side faces of tabular grains that have relatively thick spacings between parallel twin planes and confirmed that on the side faces of tabular grains, both rough-smooth and ridge-trough structures exist.