Abstract: In this paper a probabilistic model for computer simulation is used to simulate an emulsion layer in which the silver halfde grains are random distributed, the latent image formation in grains during exposure and granularity measurement for the emulsion layers as well. By means of a regression analysis approach, the comparisons between the granularities which arise from a random distribution of grains at different grain spatial densities and those of the emulsion layers after exposure to light or one-quantum-threshold radiation are made. It comes to the conclusion that if at each stage of the photographic process each grain in an emulsion layer is independently involved in various reactions concerned, i.e. there is no correlation between grains, there exist no detectable contribution of photon noise to the photographic granularity.