Two types of experiments have been conducted to evaluate the characteristics of the previously reported simulation model for migrations and dispersal of prehistoric human populations. The subject area for the experiments is constructed on the computer, which is a hexagonal "land" consisting of 1261 cells with hexagonal-linkage. An "initial" population made up of 100 persons has been set on the center of the hexagonal land, then the years needed to reach the cells on the edge of the hexagonal land and the total population in each period have been calculated under the conditions determined by the six parameters of the model.
We have obtained the following results;
1. The following four conditions have been examined: A: NRR when a cell is aligned to the increasing phase; B: R (=the population size of a cell/the carrying capacity of the cell) to which a cell is aligned to the increase phase; C: R under which a cell do not migrate; and F: personal probability of migration to the designated cell when a cell is going to migrate. They have made apparent differences in the years and the size of population. But the following two conditions do not affect them significantly; D: R over which a cell does not accept migrants; and E: probability to migrate when a cell exceeds R of condition C.
2. A unique pattern of spatial dispersal has been observed under the condition, F: p=0.9. Cells migrated more frequently and dispersed more rapidly.
3. Under the combined condition, C: R=0.2 and F: p=0.9, more frequent and rapid dispersal has been observed than under the condition F: p=0.9 or C: R= 0.2. This effect is considered synergistic.
4. Applying characteristic data and their mutual relations hip to Martin's (1978) model, we have found that the approximate value of NRR=2 would be needed for human populations to migrate from Edmunton to the southern end of South America under the condition, C: R=0.2.
5. When a population consists of a few hundred or so, a random effect caused by the stochastically of the reproductive process would affect the increase or the decrease of the population. A census data from such a small population should be treated with some care of that effect.
6. The applicability of this simulation model to modern human populations is also discussed.