The rates of growth and decay of latent-image specks in the developer were calculated numerically, on the basis of the assumption that the specific rates of growth and decay of Agn specks were proportional to Fermi distribution of electrons in the lowest unoccupied level of Ag⁺_n and to Fermi distribution of electron vacancies in the highest occupied level of Ag⁺_n respectively. As the energies of the levels, the values estimated by Trautweiler were adopted. Rate equations (1-m) were set up for the growth and decay of Ag⁺_n and solved by the steady state approximation method or Runge-Kutta-Gill method under appropriate boundary conditions. Following 1)-4) were showed; 1) An unexposed pure silver halide grain was reduced very slowly in the developer (Fig. 2). Once the grain had a speck which provided the lowest unoccupied level as high as or lower than Fermi level of the developer onto the surface of the grain, the grain could be reduced rapidly in the developer (Fig. 3, 4). 2) When the highest occupied level of the speck was as high as or higher than Fermi level of the developer, the speck was bleached rapidly to Ag⁺_n by the developer (Fig. 3) and then grew very slowly to a developing center similarly to other Ag⁺_n's. 3) If Fermi level of the developer lay intermediately between the lowest unoccupied level and the highest occupied level of a speck, a part of the specks grew to the developing centers and the other part decayed to Ag⁺_n's at rather slow rates (Fig. 3, Table 1). Such a speck acted as a sublatent-image in the development. 4) There was an induction period of the formation of developing center from Ag⁺_n and its period was prolonged with decreasing n of the Ag⁺_n (Fig. 4).