1. Introduction: We have little understanding of the true rôle of parasitic insects in the maintenance of the natural equilibrium chiefly because few efforts have been made to interpret the available data.2. This is especially true concerning the problem of host relations which is in many ways of fundamental importance.3. To solve the problem of host relations we ought to be able: (a) to deduce from a study of the host the composition of its parasitic fauna; (b) to deduce from a study of the parasite the composition of its host list.This would necessitate (1) the definition of the “typical” or “ideal” parasite of any given host; (2) the definition of the typical or ideal host of any given parasite; (3) the establishment of a correlation between such definitions and morphological or systematic data in such a way that conclusions could be drawn concerning species whose habits had not yet been the subject of direct observation.The object of the present paper is the examination of this problem.4. The problem of the host relations is simply the problem of a particular kind of animal association. From analogy with other problems of this type, we may deduce that the idea of formulating a definition of the “typical parasite” of a given host is fallacious. A consideration of the data in regard to the parasitic fauna of various hosts confirms this as does experimental evidence. The attraction exerted by the host on the various members of its fauna is not necessarily identical or even similar.5. The only feasible method of attacking the problem is therefore the analysis of the behaviour of parasites with the object of defining their “typical hosts.”6. An experimental study of the host relations of Melittobia acasta Walk. was conducted in the laboratory and an attempt made by the analysis of the characters of the various species accepted or refused by the parasite, to formulate a definition of the normal host.7. The nature of the hyperparasitic and gregarious habits are discussed in connection with the habits of Melittobia. An effort is made to discuss the true causes of the phenomena.8. A study of the variations in oviposition habit in relation to the particular characteristics of various hosts shows that these variations, though complex and very difficult to explain in relation to the efficient causes involved, are easily explained if considered in relation to the object of the behaviour.9. A general definition of the host of Melittobia acasta was formulated on the basis of the observations and experiments performed.10. Attempts to simplify the definition and discover for the phenomenon of attraction some simple stimulus, completely failed.11. A critical examination of the definition of the normal host of Melittobia shows that though it indicates the general trend of behaviour in the species studied, it is at once too narrow to include all possible hosts and too broad to exclude all species which might be refused.12. A detailed study of the extensive data available on the host relations of the Tachinid Compsilura concinnata Meig. confirms the above results and shows that the typical host cannot be defined in relation to either morphological or physico-chemical characters.13. It is shown further (a) that the fauna of systematically related parasites though sometimes similar are not necessarily so and may be very different; (b) that parasites so similar as to be morphologically inseparable in most or all stages of development may have different host relations.14. A scientific treatment of host relations along ecological lines whether (a) by the establishment of an empirical correlation between systematic position and habitat or (b) by the definition of the parasite in terms of its habitat is thus impossible.15. It follows that the acts of the parasite are not strictly determined if considered simply in relation to the physico-chemical and morphological properties of the host.16. The reason for this is the fact that the choice of hosts is primarily a phenomenon of the psychological order which can be understood only by analogy with a similar process, such as the choice of food, in regard to which direct knowledge can be obtained by introspection.17. A comparison along these lines leads to the conclusion that the choice of hosts, like the choice of food, is not a phenomenon which can be referred to any constant and definite efficient cause exterior to the organism; in certain cases the substances selected differ entirely in their properties and are identical as causes of behaviour only in that they are all agreeable to the organism concerned. In other words, the cause of behaviour is one only upon the psychological plane. Both food and hosts may be considered in relation to their suitability in relation to a given organism, but again the only category in which all types of food and hosts can be grouped is that of assimilable substances. The cause of behaviour considered from this angle is one only upon the physiological plane.18. Pursuing the analysis of the process of food or host selection one finds, that this process occurs as if the suitability of substances as food or as hosts, obviously not in itself perceptible, were deduced by the animal from the perceptible qualities, though we have no reason to suppose that any conscious judgment occurs in the animal mind. This conception which corresponds exactly to the classical idea of instinct is the only one which seems to fit all the facts of the case.19. An apparent objection to the views here advanced, based on the fact that the choice of the habitat in many animals seems to have been expressed with sufficient adequacy in terms of simple factors having certain quantitative values owes what force it possesses to the fact that the factors considered are those affecting the tactile sense in various ways, this sense being that in which the qualities of the object perceived affect the sentient subject most acutely so that his sensory impressions of pain and pleasure, ultimate sources of his actions, are more nearly measures of the qualities perceived than is the case with other senses. It is to be noted further that the objects of the tactile sense are essentially less varied than in such senses as taste, smell and sight, and that many stimuli classed as gustatory or visual are in reality tactile.In view of the undeniable adaptive power possessed by organisms, it is however doubtful whether tactile impressions really constitute a valid measure of tangible qualities, or that movements induced by tactile pleasure and pain can be considered for a given organism as due to certain constant values of these qualities as measured by physical instruments. The results obtained up to the present do not constitute by any means a sufficient answer to this difficulty and have in any event but little bearing upon the problem of host selection which there is no reason to consider as determined by impressions of the tactile order.20. There is therefore no reason to suppose that the accumulation of data on the host relations of entomophagous parasites will ever lead to the discovery of laws of parasite distribution permitting us to correlate the properties of parasites with those of their hosts so that we can predict in advance the host list of a given parasite or the parasitic fauna of a given host. The data collected may suggest interesting hypotheses or facilitate greatly operations of the practical order, but they have in themselves, only an empirical significance. The laws underlying the problem of host relations are not capable of expression in scientific terms nor discoverable by scientific methods.21. From this it follows that in problems involving the practical utilisation of entomophagous parasites, as for example, when these species are transferred from one country to another, we have no means of predicting results because it is impossible for us to foretell the behaviour of the parasite with regard to the possible host insects of the new area. It would therefore be futile to make the execution of such projects dependent upon investigations in the native home of the parasite and designed to supply advance information as to the results obtainable. Such studies, while no doubt of great interest and value, cannot lead to any certain conclusions in regard to the question at issue, which experiment alone will solve.