The geographical distribution of the different species of rat flea is one of the most important factors governing the spread of plague.

Xenopsylla cheopis is the most effective transmitter of both epidemic and epizootic bubonic plague in all parts of the world.

The relative immunity of certain parts of India and Ceylon from plague is to be primarily attributed to the prevalence of an inefficient plague-carrying flea on the rats of these localities, viz. Xenopsylla astia.

Climatic conditions play only a secondary part in limiting the spread of the disease in these relatively plague-free localities.

Plague infection can be maintained in such localities by the importation of cheopis-infested merchandise.

Climate does not exert an equal effect upon the transmitting power of different species of flea.

Numerous attempts to transmit plague from rat to rat by means of X. astia at room temperature during the plague season in Colombo have all failed. X. astia, however, can carry plague from rat to rat under climatic conditions favourable to plague transmission. Continuous transmission of plague from rat to rat has been obtained with X. cheopis in Colombo under exceptionally unfavourable climatic conditions.

Both plague and X. cheopis have a very irregular distribution within limits of Colombo City. The coefficient of correlation between two periods of human plague incidence and percentage of cheopis to total rat fleas of all species in 14 districts of the city is ·95, probable error ± ·02 for both periods.

There is a close relation between premises infested with X. cheopis and premises where plague-infected rats are found.

A three-fold relationship exists between the amount of imported grain, the prevalence of cheopis on rats, and the incidence of plague in the localities where grain is stored in Colombo.

The measures so far devised for combating plague among Oriental populations, tolerant of all forms of animal life, have not proved conspicuously successful.

Due recognition of the importance of the flea species factor should lead to marked improvement in plague preventive measures in many parts of the world.

A rat flea survey of all the principal plague areas, especially the ports, becomes urgently necessary.

The exports from localities where X. cheopis is found to be prevalent upon the rats should be considered actively dangerous when the locality is plague infected, and potentially dangerous even if it is plague free, since the continual importation of cheopis into a locality where this flea is not indigenous may eventually result in the extension of plague infection to a new area.

I agree with Glen Liston and Norman White, that the export of produce, especially grain, from plague-infected ports should be controlled by international agreement.

The limitation of imported cheopis and plague to definite zones in such localities as Colombo greatly facilitates the task of plague prevention, since all available resources can be concentrated simultaneously, or successively, on the infected areas.

In such instances it is possible to mark out the potentially plague-infected zones by the simple and accurate method of surveying the rat fleas.

Ten years have now elapsed since I tentatively put forward the suggestion that the spread of epizootic plague in the East Indies might be governed by the flea species factor. At first I adopted a cautious attitude towards my own hypothesis. I believe that the evidence now available warrants its acceptance by epidemiologists without further hesitation.

If so, the discovery is but further testimony to the essential unity of science in its bearings on the welfare of the human race, for it is the natural outcome of the purely zoological researches of Rothschild and Jordan on the systemics of the Siphonaptera.

Summarising our results, and considering their probable significance, we should draw the following conclusions.

Our observations do not suggest that the dissemination of a bacteriophage is likely to have been the cause of the sudden decrease in the excretion rate of B. aertrycke, which was observed among the population at risk during the epidemic previously recorded (Topley, Ayrton and Lewis, 1924).

Our results are completely at variance with those recorded by d'Herelle in the case of fowl typhoid, a disease which shows many points of similarity to enteric infection in mice. The observations we have recorded do not suggest that the presence of the bacteriophage will, in itself, prevent the epidemic spread of infection, check an epidemic when it has once started, or appreciably reduce the mortality among the population at risk. In the light of these results we cannot share d'Herelle's optimism as to the probable results of the wholesale administration of the lytic principle in a public water supply. We do not suggest that the Twort-d'Herelle phenomenon has no significance in the epidemic spread of disease, but it does not appear to us that the problem is the simple and straightforward one envisaged by d'Herelle.

The results of the last experiment of this series are in substantial agreement with d'Herelle's experiments on barbone in cattle, so far as they concern the action of the lytic filtrate in causing an immunising response, though we have not obtained such strikingly uniform and satisfactory results.

We would note, in conclusion, that we have not yet tested the effect of the administration of the lytic principle, by injection, during the course of an experimental epidemic.

Owing, in our opinion, to faulty classification and terminology of bacterial types, it is very common, especially on the Continent and, to a lesser extent, in U.S.A., to ascribe outbreaks of food poisoning to B. paratyphosus B, the common cause of paratyphoid fever. If such a conception is true, it is obvious that B. paratyphosus B can at one time cause paratyhpoid fever, at another an outbreak of food poisoning. Further, one would expect in outbreaks of either condition that some cases would be of the one clinical type while others would exhibit the other. In particular in outbreaks of paratyphoid fever in which the vehicle of infection was some form of food, it is to be anticipated, on this view, that a considerable proportion of the cases would be of acute food poisoning type.

In the Goulstonian Lectures, delivered in 1919, one of us (W. W. C. T.) discussed the possibility of attacking epidemiological problems by the method of direct experiment, and reported a few preliminary observations on the spread of enteric infection among mice. Since then a series of reports have been published dealing with the investigation of particular problems along these lines (Topley, 1921 a and b, 1922 a and b, 1923, Topley and Wilson, 1923 a and b, Topley and Ayrton, 1924 a, b and c, Topley, Ayrton and Lewis, 1924 a and b, Topley, Wilson and Lewis, 1925).