Most adaptive characteristics of redpolls (Acanthis spp.) are keyed to surviving an arctic or subarctic winter, where energy balance is the severest problem. Many features combine to produce the total adaptation that makes these tiny birds some of the most cold-hardy of homeotherms (Brooks 1965, Ph.D. thesis, Univ. Illinois, Urbana), but the presence and specialized use of the esophageal diverticulum may be among the most important. It is a partially bilobed, ventrolateral outpocketing located about midway on the neck, found in redpolls and certain other northern finches (e.g. crossbills, Loxia spp.) but absent in most other passerines (Fisher & Dater 1961, Auk 78: 528). The ventral, unilobed in galliformes and others is only somewhat analogous in structure and function. The adaptive value of the diverticulum manifests itself in three aspects: (1) energy gained, (2) energy saved, and (3) reduced predation. Extra food is stored there toward nightfall, carrying the bird through low night temperatures when it is unable to feed. I have estimated that the amount of birch seeds that could be stored in the diverticulum by Common Redpolls (A. flammea) and used at night would allow survival to -62?C, a substantial increase in cold tolerance over the -30?C that could be tolerated without that extra energy (Brooks 1968, Wilson Bull. 80: 253). A hitherto unreported use of the diverticulum is as an energy saving device. During the winter finch invasion of 1971-72 in Ripon, Wisconsin, I observed that individual redpolls hurriedly selected millet from the seed mixture at a feeder, then flew away. This produced a constant, rapid turnover of the feeding flock. There were never more than about 100 birds feeding, but mark-recapture banding data indicated that 1,000-1,500 different redpolls utilized the feeder daily. Little agonistic interaction was noted while feeding. Individual Evening Grosbeaks (Hesperiphona vespertina), however, remained for fairly long periods shelling sunflower seeds, and many intraspecific agonistic encounters were seen. Closer observation showed other differences between the species. Especially on colder days, redpolls ingested seeds rapidly without shelling them. As some individuals ceased feeding they left the flock and congregated in a dense white cedar about 2 m from my observation window. Although they appeared to be resting, plumage fluffed out in the heat-saving position typical at low temperature, they were actually regurgitating, shelling, and reingesting millet seeds. Due to the numbers processed by each bird I assume these seeds had been held in the diverticulum. The primary natural food of redpolls is birch seed, about the size of poppy seed. Typical wild feeding behavior at first appears distinctly biphasic. The birds, using wings, legs, head, and tail vigorously, spend a short time working acrobatically among the terminal branchlets in phase I, knocking seeds out of the catkins. This muscular activity consumes more energy, and because of the consequent insulation disruption and greater exposure to wind, they lose relatively more calories than in phase II. They spend a longer time in phase II gathering these seeds from the ground (snow, usually) where there is less wind, where they can remain more easily in a ball position, but where they are probably more exposed to the view of predators (on snow). Energy rather than predatory considerations thus may shorten phase I. Biphasic feeding is already more complex than the feeding of most other birds, but redpoll feeding is actually a triphasic process. In phase III cover is sought and seeds stored in the diverticulum are shelled. This advanced type of feeding behavior, to my knowledge, is unique among birds but is fairly similar to that of the red squirrel (Tamiascirus hudsonicus), another boreal treetop harvester. The two obvious benefits of phase III are that additional energy is saved and potential predators are avoided. The energy saving involves four factors. One is that seeds are undoubtedly softened by moisture in the diverticulum, making hulls easier to remove and allowing easier mastication of the hydrated edible portions in the gizzard. Slightly less energy is expended for this necessary work (though some calories are lost in heating expelled hulls and water), and more is thus available for body heat production or other activities. This hydration phenomenon is quite reminiscent of cud-chewing in ruminants, whose stomach compartments allow proper separation of processed from unprocessed food. It is probable that the need for efficient separation is the reason that the esophageal pouches in redpolls are laterally oriented. The question has been raised (Fisher & Dater op. cit.) as to why the diverticulum is not ventral as the crop is in