1. The incidence produced during 1926 by two epidemics of diphtheria, in a remote and unsalted country district, was equivalent to 30 years of the London endemic morbidity.2. The cases arose in two distinct groups. An autumn outbreak with a definite concentration of infection on the school age, and probably disseminated by droplet infection, followed by an epidemic in the succeeding spring which was more severe, probably milk-borne, and in which the frequency of cases was higher among infants and adults than school children.3. The effect of a long drive over bad roads caused a deterioration of the diluted toxin which was used for a Schick test investigation. The use of this toxin led at first to false conclusions regarding the immunity of the children, since it only caused reactions in the most susceptible members of the community.4. As a result of retesting with a fully potent toxin it was estimated that the figure of 40 per cent. susceptible which was found with the weak toxin would have been about 80, if full strength toxin had been used.5. The attempt to immunise actively those children who had reacted to the weak Schick toxin was found to have failed in 72 per cent. of a group which was retested 7 months later. The prophylactic course was identical with one which had succeeded in inducing Schick immunity, within 4 months, in 100 per cent. of the susceptible children in the Dunedin orphanages.6. The slow rate of immunisation at Kaitangata is attributed to two causes, (a) the low original herd immunity of the district, (b) the selection by the weak toxin of only the most sensitive reactors which were almost certainly the most “unresponsive” members of the total Schick positive population.7. By a comparison of the rate of response to a small dose of toxin-anti-toxin mixture, in those positive Schick reactors (a) who had, and (b) who had not, received previous prophylactic injections, it was shown that the principle of “primary and secondary” antigenic stimuli held good for human subjects, as well as for experimental animals.8. Careful examination of the known data led to the conclusion that the air-borne autumn epidemic had increased the herd immunity of the population, in spite of the fact that the subsequent milk-borne spring outbreak produced as high a morbidity and a more severe type of diphtheria.9. The practical lessons which may be learnt from this study are: (a) That unless a special buffered diluent is used, the toxin used for Schick testing must be freshly diluted near the place where it is to be used, (b) Active immunisation of remote unsalted communities against diphtheria requires more time, and more intensive courses of prophylactic, than areas where diphtheria has been endemic for some years, (c) In such places with a low original herd immunity, it is essential, even more than in endemic centres, never to omit retesting those who have been inoculated, in order to be certain that any attempt made to induce active immunity to diphtheria may be successful. (d) Estimations of the relative efficiency of diphtheria prophylactics, which are based on the rapidity with which samples of children become immune, are worthless unless all the observations have been made in the same environment on groups having the same original herd immunity, (e) An immunity, good enough to withstand droplet infection in a day school environment, may be broken down by massive doses of diphtheria bacilli in milk.10. The study of outbreaks of disease in communities which are not too large to prevent a complete supervision of the whole population and environment, enable one to visualise imaginary mechanisms of infection very much more easily than in large populous areas, where so many more factors have to be left unexamined, or remain unsuspected. The hypotheses, perhaps better termed surmises, which the facts suggest, are only intended to be purely tentative and temporary. Their practical value is that they indicate to the investigator a plan of campaign and suggest what to look for, in the lucky event of his ever again meeting with a similar combination of circumstances. In epidemiology in general, as illustrated by this particular instance, opportunities arise and pass, often never to recur, and much is missed, or left undone that might have been done, if only the knowledge and hints gained during the passage of the epidemic itself had been known before its advent.