Two conditions have been proposed (Anderson & Dow, 1948) for the census method of constructing a mortality table to be ‘strictly correct’: (1) that at the beginning and end of the investigation, ages of lives are evenly spread over the year of age x to x + 1, or (2) that numbers and distributions of ages are the same at beginning and end.

In effect, these alternative conditions are proposed to insure that the approximate formula for the total period of exposure to risk, derived from annual census,

shall be reasonably correct. It is, however, of interest to examine the method more closely, under the hypothesis that the total period of exposure to risk can be determined with negligible error, as would theoretically be possible by holding censuses with sufficient frequency.