The Greater Flamingo, Phoenicopterus antiquorum, feeds by filtering chironomid larvae, seeds, etc., from mud; the Lesser Flamingo, Phoeniconaias minor, has a much finer filter, previously undescribed, by which it feeds on the blue-green alga, Spirulina platensis, and diatoms. The two flamingoes can therefore feed in the same lake without competing for food. Data from observations on living birds, from detailed anatomy of the bill, and from analyses of stomach contents of birds feeding in localities with known ecology, have been used to elucidate the process of filtration in these species, seen in Kenya, and in Phoenicopterus ruber, Ph. chilensis, Phoenicoparrus andinus and Ph'parrus jamesi, of the New World. The bill structure is of two distinct types, though always bent in the middle, with the lower jaw large and trough-like and the upper small and lid-like. The shallow-keeled upper jaw of Phoenicopterus is as wide as the lower; the gape is at the side, crossed by distal outer lamellae curling downwards to prevent its closing completely. The inner surface carries a concave lamellated area on either side of the keel, but does not touch the similar area on the convex, inflexed border of the lower jaw. The deep-keeled upper jaw of Phoeniconaias, and Phoenicoparrus, is much narrower and lies flush between the rami of the lower; the dorsal gape is crossed by horizontal, distal outer lamellae. Internally, the upper jaw has a deep keel that is flat-sided, so that in section it is an equilateral triangle. The flattened, inflexed borders of the lower jaw turn downwards for 15 mm parallel with the sides of the keel throughout the length of the gape; the central tongue-groove is thereby much constricted. These flat lamellated areas, three times the size of those of Phoenicopterus, are in close contact across the gape. Bills of the three species of Phoenicopterus are much alike; those of Ph'naias minor and Ph'parrus andinus are similar in form and size; that of Ph'parrus jamesi is shorter, with a narrow upper jaw and consequently reduced lamellated areas. Measurements show much individual variation. The lamellae are horny thickenings running transversely across the lamellated areas; they form smooth ridges, serrated ridges, or separate triangular platelets, fibres of which may separate to form a fringed edge projecting towards the tongue. Their variations in form, size and spacing, in different parts of the bill and from species to species, are described and illustrated with figures and tabulated measurements. Outer lamellae, near the margins of the upper jaw, are large leaflets or hooks projecting across the gape: vertically, to form the filter in shallow-keeled bills; horizontally, to form the excluder in deep-keeled bills. They are fringed, in species with fringed inner lamellae. Outer lamellae of the lower jaw are smaller and less distinctive but subdivide the filter in Phoenicopterus. They are integrated with the inner lamellae in other genera. Inner lamellae are smaller and more closely spaced than outer. On the lower jaw of Phoenicopterus they form smooth or serrated ridges; on the upper jaw they form platelets, diagonally alined to leave channels leading inwards and backwards from gape to tongue. In Phoeniconaias and Phoenicoparrus most inner lamellae on both jaws form small, tall, closely set platelets, also diagonally alined, making a velvety pile on the flat lamellated areas. The platelets of Ph'parrus andinus are smooth, those of Ph'naias minor and Ph'parrus jamesi are fringed; all form fine-meshed filters, small in area in the last-named. The tongue and palate in Ph. antiquorum are larger than in Ph'naias minor; they have a similar armature of spines directing food towards the gullet, and a transverse epiglottis. The tongue is large and fleshy; the distal half, packed with oily fat, has only the thin hypoglossus muscle below the broad lingual cartilage; the other muscles are straight and confined to the base, and aid its fore and aft pumping action. The bill is reputed to be rich in tactile organs; Herbst corpuscles occur under the dorsal surface of the upper jaw of Ph'naias minor, but are rare under the inner lamellae. Their position in Phoenicopterus is uncertain. A large olfactory foramen suggests that there may be a well-developed 'oral' sense to be investigated. Ph. antiquorum feeds in bottom mud, whereas Ph'naias minor sweeps near the water surface. Both use a 'jig' movement to stir particles off the bottom; in Ph. antiquorum this is also associated with eating mud. Birds in captivity show the jaw movements and the forceful tongue pulse, expelling jets of filtered water four times a second. Records of the kinds and sizes of food organisms, mostly based on stomach contents, are tabulated for all species. In Phoenicopterus the kind of food varies widely (contrary to the belief that Ph. ruber eats only the gastropod Cerithium); but the size is between 1 and 10 mm in length. Claims that they also feed on algae are dubious; but Ph. antiquorum and Ph. ruber can survive, and feed their young, on organic mud. In East Africa healthy specimens of Ph'naias minor always feed on blue-green algae or diatoms, with dimensions between 0<latex>$ \cdot $</latex>02 and 0<latex>$ \cdot $</latex>1 mm. Entomostraca or water-mites are probably eaten by Ph'parrus andinus, and algae of unknown dimensions by Ph'parrus jamesi, but the evidence is inconclusive. Analysis by British Standard Sieves shows that 80% of the grit from gizzards of Ph. antiquorum measures more than 0<latex>$\cdot $</latex>5 mm, and from Ph'naias minor less than 0<latex>$ \cdot $</latex>5 mm. To identify the structures used in filter-feeding, their sizes are compared systematically with those of the grit and food. For Phoenicopterus, grit provides the more exact measure, and food confirms it; for Phoeniconaias, diatoms taken from the stomach provide the exact measure. In Phoenicopterus the filter is formed across the practically closed gape by the outer hooks and leaflets on the upper jaw, aided by the intermediate submarginals above and the serrated ridges on the margin of the lower jaw below. All the spaces then have one dimension less than 0<latex>$\cdot $</latex>5 mm, except near the tip of the bill, where the minimum increases to 1<latex>$\cdot $</latex>0 mm. This is commensurate with the smallest dimensions of most of the food (theoretically, cylindrical objects could be caught by any mesh smaller than their longest dimension, i.e. 3 to 5 mm in this case). The excluder must be formed by opening the gape to about 4 or 6 mm at the bend, the shape of which prevents the tips from separating as widely as they would in a straight bill. In Ph'naias minor the fine filter is formed by fringes on the inner platelets enclosing spaces measuring 0<latex>$ \cdot $</latex>01 by 0<latex>$ \cdot $</latex>05 mm. The excluder is formed by the outer lamellae, with spaces measuring about 1<latex>$ \cdot $</latex>0 by 0<latex>$ \cdot $</latex>4 mm. The same structures give a filter measuring 0<latex>$ \cdot $</latex>005 by 0<latex>$ \cdot $</latex>05 mm and an excluder of 2<latex>$ \cdot $</latex>5 by 0<latex>$ \cdot $</latex>25 mm in Ph'parrus jamesi; the intermediate filter in Ph'parrus andinus is formed by smooth platelets with spaces measuring between 0<latex>$ \cdot $</latex>3 by 0<latex>$ \cdot $</latex>06 mm and 1<latex>$ \cdot $</latex>2 by 0<latex>$ \cdot $</latex>14 mm. This filter therefore has spaces smaller than the smallest of those in Ph. antiquorum, but linear dimensions about ten times larger than those in Ph'naias minor. The excluder is also intermediate and has dimensions between 1<latex>$ \cdot $</latex>5 by 0<latex>$ \cdot $</latex>7 mm and 2<latex>$ \cdot $</latex>5 by 1<latex>$ \cdot $</latex>0 mm. There is no direct evidence on the process of filtration within the closed bill in its inverted feeding position; but the filters and excluders already identified are correlated with the movements of the jaws and tongue to postulate three methods of feeding, differing with the kind of food and the type of bill. (i) When Phoenicopterus, with its shallow-keeled bill, feeds on whole organisms, water is actively pumped by the tongue through the stationary outer lamellar filter, while the bill is swept through the water to reach new food supplies. Raising the upper jaw admits the food through the narrow gape, acting as excluder; closing the bill re-forms the filter. The inner lamellae may help to detect and hold organisms in the bill. (ii) But when Phoenicopterus feeds on mud, the process is reversed and apparently ceases to be true filtration, for there is no filter fine enough to retain mud, which must therefore be gulped down together with admixed brine. The outer lamellae form the excluder for the coarse sand, which is then left in cones marking the feeding places. (iii) When Ph'naias minor, with its deep-keeled bill, feeds on algae, the upper jaw must not move far, for its outer hooks and leaflets form the excluder; the tongue pumps water, carrying food, inwards along channels between the diagonally alined inner platelets. Pressing the extensive lamellated areas together, on one side of the bill at a time, helps to force the outgoing water through and around the filtering fringes of the inner lamellae. The volume of water moved by each tongue stroke about fills the deep gape; it will therefore oscillate about the filters, rather than be drawn through them for long distances in either direction; small vortices will help to entangle and retain the food. This is collected from the filters by rubbing them up and down on each other, like collecting wool from 'carders'; thus it is brought within reach of bristles on the tongue. Filtration in Phoenicoparrus is presumably similar. It is suggested that the reversed use of marginal hooks as excluders, when Phoenicopterus feeds on mud, may be a clue to the evolution of the deep-keeled species, which combine this use of the marginals with use of the inner platelets as a fine filter. There is no fossil evidence upon this point and little upon the evolution of the flamingo's bill as a whole, with its many specializations. The suborder had most of its present characteristics in the Miocene, except the bend in the bill, which still appears late in individual development. The affinities of flamingoes with other birds are certainly obscured by their specialization for filter-feeding, in which they are only rivalled among adult vertebrates by the whale-bone whales (Mysticeti). Like other microphagous species, flamingoes have only a choice of size of organisms but not of kind. Since flocks are large, food requirements are enormous; their distribution is therefore strongly influenced by the search for habitats where such food occurs in abundance. This means arid localities, with brackish or alkaline waters, where the few species which can withstand the ecological rigours of the situation can multiply sufficiently, whether they be Artemia, Cerithium or Spirulina. Thus flamingoes congregate near the great deserts of the world, often at high altitudes. The ecological advantage of these habitats in providing food is offset by the damaging effects of their waters, which differ greatly in salt content and ionic balance from the bird's body fluids. The physiological importance of the part played by filtration in reducing the volume of the medium ingested with the food is stressed.