SUMMARYUnlike many anostracans, which are microphagous filter‐feeders,B. gigas, which is the largest known extant member of the group, is shown to feed to a large extent on large food masses and to be a carnivore.The most important items of food recorded in the available samples were other anostracans belonging to a smaller, filter‐feeding species. An ability to capture and deal with such large items of food has not hitherto been reported for the Anostraca.Structural features associated with the feeding habits ofB. gigasare described, and this species is shown to be incapable of filter‐feeding. No filter is present on any limb. The armature of all the trunk limb endites and of the endopodites consists of stout spines, and the limbs make up a cage suitable for retaining the prey. The three distal endites are orientated in a different manner from those of filter‐feeding species and the endite armature differs in orientation from that of filter‐feeding forms. The anterior endite spines are directed anteriorly and interdigitate with the posterior spines of the limb in front. The arrangement is such that food can be passed from limb to limb by the spines of the distal endites on promotion and by the spines of the basal endites (including those of the gnathobase) on remotion of the limb.Special attention is directed to the relationship of the anteriormost gnathobasic spines and the adjacent spines of the limb in front.The maxillules are well suited to dealing with large food masses.Although still typically anostracan in structure and musculature the mandibles exhibit specializations. Posteriorly there is on each a large tooth which, as a result of the swing of the mandible, must pierce, and force forward the food. The crushing abilities of the mandibles are clearly demonstrated by a specimen in process of swallowing another anostracan.Such a tooth also enables a measure of transverse biting to be achieved without changing the axis of movement of the mandible and without such structural changes as have accompanied the acquisition of this accomplishment in the Malacostraca.Comparison is made between this tooth and posterior cusps in a cladoceran and in two primitive insects. The remarkable convergent similarity of members of these two great groups (Crustacea and Insecta) is pointed out.Observations on the feeding habits of other anostracans and on structural features related to feeding in certain species are recorded. Several species collect bottom material by scraping with their endopodite spines which, for this purpose, are modified as scrapers. In some species, e.g.B. packardi, this scraping is very persistent. In some species at least, e.g.B. mackini, large masses of food are passed forward. It is evident that true filter‐feeding is not involved in this process though filter‐feeding can be and probably is practised at times. Gut contents of specimens collected in nature, however, suggest that material scraped from the bottom is of great importance in several species.Another large species,B. ferox, is also incapable of true filtration, at least when large, but insufficient material has been available to enable its feeding habits to be ascertained. Scraper‐like modifications of the spines both of the endopodites and of the distal endites of the trunk limbs are shown by small individuals. The modifications of the endite spines doubtless facilitate the forcing of food towards the food groove. In large individuals these spines are replaced by stout non‐scraping spines, apparently suitable for handling large food masses.The situation inB. gigasis interpreted as a case of habit reversal following the exploitation of large size rather than a case of primitive feeding habits. It is suggested that the earliest Anostraca may well have been incapable of filter‐feeding. It is possible that the CambrianOpabiniawas such a non