It is difficult to determine how long even on an average this relative protection or insusceptibility lasts. Evidence is fairly uniform in indicating a protection of at least three months. Usually it is longer. There seems to be some basis for the supposition that a group of individuals exposed to an attack of influenza displays within the succeeding three months, or slightly longer, a relative general group immunity. If the group be considered as a whole those even who did not develop the disease previously appear to have become less susceptible. Whether we can ascribe this to the individual as a unit, or whether we must explain it by some assumption with the community as a unit, is uncertain. Is it because the exposed individuals in the group who did not contract the disease have individually received some of the virus into their systems and developed a certain immunity, or is it a much more complex phenomenon depending on greater relative dispersion of susceptibles and other communal factors? We may place the minimum period of “immunity” at from three to five months, rarely less. There is additional evidence by which we may delimit fairly closely the other extreme, that time at which individuals considered as a group no longer manifest increased resistance to the disease. The author found that 19.17 per cent. of his population contracted influenza in 1918, and 9.55 per cent. contracted the disease in 1920. Two hundred and forty individuals, or 2.4 per cent. of the entire population developed the disease in both epidemics. Out of 1,971 individuals having the disease in the 1918 spread, 240, or 12.1 per cent. recurred in 1920. This is to be compared with the total 1920 incidence of 9.55 per cent. More correctly we should separate the 1920 cases into two groups, those who had and who had not had influenza previously. The former group, 240 individuals, constitute as just stated, 12.1 per cent. of all who had had the disease previously. The second group, 715 individuals, constitute 8.9 per cent. of the 8,034 who had not had the disease in 1918–19. From these results we must conclude that _a previous attack contracted on an average of from 10 to 17 months before, conferred no protection whatever against a second attack. On the contrary, the attack rate was slightly higher in this group than in those who had not previously had the disease._ Yet another evidence of the insignificant part played by any immunity in the occurrence of influenza in individuals in 1920 is indicated by our series of 319 infants living in 1920 but who had not been born during the 1918 spread and who were presumably not immune to the disease. We have not investigated whether the mothers had had the disease in 1918. From among these 319 infants, thirty or 10 per cent., developed the disease in 1920. This is practically the same percentage as for the population at large.