Waxwings are gregarious birds and individuals establish no well-defined territories as do many birds. The nest itself is the only defended territory, and as Crouch (1936) has shown, the Cedar Waxwing will nest in close proximity to others of the same species. Swarth (1932:275) mentions that the Bohemian Waxwing is tolerant of the nests of other pairs near by. The extreme condition is that found in Dulus, in which the territory is not limited even to the nest, but to the individual compartment of the community nest. Phainopepla, a less gregarious bird than Dulus and waxwings, has a much more definite territory, although individuals of Phainopepla are tolerant of others of the same species; no feeding territory is established, and small flocks of birds feed together at any time of the year.

In birds whose territories lack well-defined boundaries, it would be expected that elaborate song would not have evolved, and that most of the recognition of kind and sex would be dependent upon the behavior of the birds. This is the fact; song, as such, is lacking in the three subfamilies Bombycillinae, Ptilogonatinae, and Dulinae. Waxwings utter (1) notes that serve to keep the flock together, (2) calls used by the young in begging for food, and (3) some low notes that Crouch (op. cit.:2) considered as possibly concerned with courtship. Phainopepla has various call notes, and in addition, a succession of notes which are run together. Ptilogonys utters a note which Skutch (MS) characterizes as a loud, not unmusical "tu-whip" that is used as the birds "fly in straggling parties which keep in contact by their constant chatter." Dulus is described by Wetmore and Swales (1931:349) as having only a variety of rather harsh chattering notes in chorus.

The most notable behavior pattern associated with courtship in Waxwings, in the absence of song, is the so-called "mating dance" described by Crouch (1936), and observed by me in Lawrence, Kansas, in the spring of 1948. This consists of one bird of a pair (presumably the male) hopping along a branch toward the other bird (the female), then away again, repeating the procedure for some little time. The female remains motionless until, as the male approaches, mutual fondling of the head and neck feathers takes place, or the birds may peck at each other's bill. A berry may be passed from bill to bill, although generally the berry is not utilized for food, and this can be interpreted as a nervous reaction of the birds. It may be an instance of "false feeding" as is seen in many birds, in which the female begs for food, as a nestling would beg, as a preliminary to the sexual act. I am of the opinion that these reactions are in the nature of behavioristic patterns that bring the birds into the emotional balance for copulation, as copulation follows the "dance." Sometimes, however, copulation is preceded by a "nuptial flight" around the nesting area, at which time the birds utter loud calls. Armstrong (1924:183) is of the same opinion, citing numerous instances in which nuptial flights and elaborate displays have evolved for just this purpose. The birds are then in the proper physiological balance to initiate the complicated sequence of copulation, nesting, incubation, feeding, and brooding of the young.

It would be valuable to know more concerning the life histories of the other birds considered in this paper, since behavior is inherent, and probably can be cited as evidence of close relationship or the opposite. All that I have been able to learn is that Phainopepla has a nuptial flight in which the male chases the female, and that Dulus (Wetmore and Swales, 1931:347) seeks the company of others of its kind at all times, and that two birds, presumably paired, will sidle up to one another when they are perched.

Both species of American waxwings build bulky nests, with the base or platform composed of a large amount of twigs and sticks, from which there often trails a mass of sticks and moss or string. Softer materials such as moss, plant fibers, and string, are placed inside the platform; moss is readily available to, and preferred by, B. garrula according to Swarth (op. cit.:271), and various plant fibers and string are used by B. cedrorum. The inner lining consists of soft plant fibers or down, dry grasses, and feathers. The nest is usually unconcealed in a tree either adjacent to a trunk or on a main side branch, but sometimes in a fork. Nest building by both Cedar and Bohemian waxwings is rapid, taking from three to five days, and is followed immediately by egg laying.

Nesting by waxwings is late in the season; June is the month in which the nest is usually started. This is readily explainable in Bohemian Waxwings, since adverse weather would prohibit earlier nesting in the area in which they spend the summer. Crouch (op. cit.:1) remarks that B. cedrorum possibly evolved in the far north where it was impossible for it to start nesting earlier, and that the habit has been retained. Perhaps, on the other hand, nesting is delayed until the berry crop is ripe, to insure sufficient food for the young.

Desertion of the nest is not uncommon in waxwings, despite the tolerance to other animals that is shown by the birds. A new nest may suddenly be begun before the first one is finished, and all the materials from the first nest may be removed, or the nest may be abandoned before it is completed. The eggs may be left at any time up to hatching, and the young may be deserted, especially in the earlier stages of development.

The very large and bulky communal nest of Dulus is not radically different from the nest of waxwings. In the absence of sufficient nesting sites, a pair of gregarious birds such as Dulus could combine their nest with those of other pairs, retaining for their own territory only the nest cavity, and in this way communal nests might have evolved. The nest of Dulus is communal probably because of the lack of suitable trees for nesting sites, and only incidentally does this type of nest afford better protection from natural marauders. Large numbers of Palm-chats work together in the construction of the nest platform, and both sexes probably take part in the work.

In Phainopepla the nest is built mostly by the male (Merriam, 1896; Myers, 1908), although the female does some of the work, especially in the shaping and lining of the nest. In this genus, the nest is usually a compact structure, but exceptional nests are of considerable bulk. The nest is commonly placed in a fork near the main trunk of a tree, in a conspicuous location, and generally is 10 to 20 feet from the ground. In shape and location, the nest closely corresponds to that of Bombycilla, but the materials used for a base are stems of annual plants, whereas Bombycilla uses more woody twigs. The finer materials used by Phainopepla are more readily obtainable in the ecological association inhabited by Phainopepla than would be heavier twigs such as Bombycilla uses.

Waxwings are typically frugivorous; berries are the staple food. The birds are known to catch insects, especially in the spring and summer, and their insect gathering technique has been likened to that of Tyrannid flycatchers. Nice (1941) experimented with a young captive Cedar Waxwing and found that it had a decided preference for red or blue berries, and that meal worms were utilized as food only when the birds became educated by other captive birds of other species as to the food value of the worms. Post (1916) indicates that the food given to the nestlings of Cedar Waxwings is entirely animal for the first three days, and that a mixed diet of berries and insects is subsequently offered.

In feeding of the young, regurgitation of partly digested food does not take place, according to Wheelock (1905). Rather, the adults "store" food in the form of berries in the expanded esophagus or crop, feeding them whole to the young. Digestion is an unusually rapid process, involving merely minutes for the passage of berries and cherries. This is correlated with a short intestinal tract, which is unusual for a frugivorous bird. Nice's (1940) experiments with Cedar Waxwings revealed that cherries would pass through the digestive tract in 20 minutes, blueberries in 28 minutes, and chokecherries in 40 minutes. Heinroth (1924) states that berries pass through the digestive tract of Bohemian Waxwings in the space of a "few minutes." This rapid digestion is obviously adaptive, since the value of the food is slight and therefore large quantities of it must be ingested; the large seeds would hamper further ingestion until they were eliminated, since they seem not to be regurgitated.

Members of the subfamily Ptilogonatinae are both insectivorous and frugivorous insofar as available data show, although again there is relatively little information available concerning them. Skutch (MS) has found that the Guatemalan Ptilogonys cinereus catches insects by repeated sallies into the air from a perch, after the manner of flycatchers. He notes also that the birds feed on berries of Eurya theoides and Monnina xalapensis. It is well known that Phainopepla catches insects when these are available, and its liking for berries is so apparent that in parts of its range, it is known as the "pepper bird," since it frequents pepper trees (Schinus molle) and feeds on the small red berries. The preserved specimens of Ptilogonys and Phainoptila available for this study contain only berries in the digestive tract. Dulus feeds mostly, if not wholly, on plant food. According to Wetmore and Swales (1931:349), berries, fruits, and parts of flowers are eaten.

A critical analysis of the skeletons provides evidence that aids the student in estimating which differences are merely the result of habits developed in relatively recent geological time as opposed to those which owe their existence to more ancient heritage. Stresses caused by the action of different sets of muscles can apparently stimulate changes in bones to meet new needs, and the evidence from genetics is that such mutations in wild birds are minute and cumulative, rather than of large degree and of sudden appearance. Once adaptive mutations have occurred, if genetic isolation from one source or another accompanies it, a new population different from the parental stock may become established. Study of the skeleton of any species of living bird may indicate those characters identifiable as modifications fitting it to a particular environment. If no distinguishing characters are discovered that may be attributed to environmental factors, such a species can be spoken of as generalized; the inference then is that such a species is not modified for a single, particular ecological niche.

Some parts of the skeleton, obviously, are more adaptable or plastic than others. The beak seems to be the most adaptable part. Probably this results from its frequent use; it is the part of the bird to capture the food. The long bones, meeting the environment as legs which serve as landing mechanisms or as locomotory appendages, and as wings which provide considerable locomotion for most birds, probably come next in order as regards plasticity. In these parts, then, one may look for the most change in birds, which, within relatively recent geologic times, have been modified to fit a particular set of conditions. From the beak and long bones of a species in which habits are unknown, one can infer the habits and habitat from a comparison with the skeletal features of species of known habits.

Skull.—The skulls in all three subfamilies have essentially the same general appearance and structure, the most marked differences being, as would be expected, in the bills and associated bones.

The most specialized bill is to be found in Dulus; its bill is decurved, and the associated bones are correspondingly changed for support of the bill. For example, the palatines and "vomer" are much wider, the palatines are more concave from below and have longer posterior processes than the corresponding bones in Bombycilla. Moreover, the "vomer" in Dulus and in Phainoptila is larger and heavier than in Bombycilla, and the quadrate and pterygoid bones are relatively large for support of the beak. The palatines, however, are weak in Phainoptila. In the Ptilogonatinae, with the exception of Phainoptila, the wings of the palatines flare more than in Bombycilla, but not to the extent that they do in Dulus, nor does the palatine bone present a concave appearance in the Ptilogonatinae. The premaxilla is a relatively weak bone in Bombycilla and Phainopepla, stronger in Ptilogonys, and is notably heavy in Phainoptila and Dulus, and in these latter two genera shows a sharply-ridged tomium. The maxillae connect to somewhat widened nasal and naso-lateral processes in all the genera, and the premaxillae narrow abruptly from this point forward. In the family, Phainopepla and Phainoptila show the least flaring in this region.

This flaring, immediately lateral to the antorbital plate, is common to all Bombycillids and constitutes a major skeletal characteristic useful for recognition of the members of the family, since the swelling is easily discernible both externally and on the cleaned skulls. In Phainopepla there is much variability in this character; some specimens have a narrower antorbital bridge than others. Only one skeleton of Phainopepla n. nitens was available. The flaring in the skull of this specimen is identical with that in Ptilogonys. Among the skulls of P. n. lepida in the University of Kansas Museum of Natural History, is No. 19228, a juvenile, taken 5 miles south of Tucson, Arizona. In this specimen, the flaring in the antorbital region is clearly evident and equal in amount to that in skulls of P. n. nitens, but the bird had not attained full skeletal growth. However, the flaring of the antorbital region appears to be common in the nestlings of many species of passerine birds. Other specimens of the subspecies lepida show a varying amount of flaring, the least (in the series available) being in No. 24754, MNH, in which the proportion of the skull (length divided by width) closely corresponds to that in Phainoptila; the skull of No. 24754 is long and thin, and the base of the bill is only slightly swollen. The skull of Phainopepla nitens lepida is more generalized than that of Phainopepla n. nitens, having a longer and narrower bill like the generalized Phainoptila. In Phainopepla n. nitens and in members of the genus Ptilogonys, more flaring occurs in the antorbital region.

Phainoptila, as noted above, has no great amount of flaring in the antorbital region. When more specimens of Phainoptila are examined, the base of the bill probably will be found to flare more in some individuals than in others; this would be expected if we may judge by the data on Phainopepla. The premaxilla and maxilla of Phainoptila are similar to the same bones in Dulus, and there is a well-marked ridge on the tomium (possibly for cutting flower parts). In Phainoptila, the palatines are narrower than in any other genus of the family and abut the lacrimals. The entire skull appears to be modified along different lines from those of the skull of Dulus; the skull of Phainoptila seems to be modified for a frugivorous rather than an insectivorous diet. The skull of Phainoptila probably is more nearly similar to the ancestral skull than is that of any other living species in the family. The wide gape characteristic of some members of the family is undoubtedly a modification for aiding in the capture of insects, and Phainoptila has progressed less in this direction than have other species in the family.

The mandibles vary somewhat in the shape and proportionate size of the bones. The mandible is proportionately, as well as actually, highest in Dulus. The medial condyle varies to some extent, being slightly flattened mediad in Bombycilla, and less so in the other genera. The mandible of Bombycilla narrows to the symphysis much more gradually than it does in the other genera.

The antorbital plate is large and divides the orbital chamber from the nasal chamber. The small lacrimal bone anterior to the plate articulates with the maxilla and the premaxilla. Shufeldt (1889) states that the free lacrimal ossicle might be of some taxonomic importance in the passerines, since it is found in the generalized Corvids and in nestling Turdids. I find it well developed and identical, with a double articulation and free ends, in all the Bombycillids. There is no significant variability in the family, and this is more evidence of close taxonomic relationship between the members of the family.

The size of the crania is somewhat variable, although the differences seem to be primarily those of proportion. Ptilogonatinae have long crania, whereas the crania of the Bombycillinae and Dulinae are shorter but deeper. I regard the longer cranium as primitive, and it is longest in Phainoptila. In order of decreasing relative length of the cranium, Phainoptila is followed by Ptilogonys caudatus, P. cinereus, and Phainopepla. Bombycilla garrula has the deepest cranium in the family.

The measurements of the lengths and widths of the skulls are given in Table 9. The relative length of the bill and relative width of the skull are given in Table 10. These relative measurements are calculated by using the actual measurements in Table 9 as numerators, the length of the skull from the lacrimal bone to the posteriormost end of the skull being used as the denominator. The data indicate that Phainoptila has a slightly narrower cranium.

Humerus.—Certain families of passerine birds have a noticeable variation in the characteristics of the humerus; the bone varies in length, in diameter, and in the complexity of the processes at either end. In the Bombycillids, however, the amount of variation is relatively small, and the diaphysis of the bone is somewhat twisted, especially so in Dulus. The deltoid tuberosity is variable, being shorter but more elevated in Bombycilla than it is in the Ptilogonatinae and in the Dulinae. The tendon from the pectoralis major muscle, which inserts on this process, probably finds better insertion on a higher process than on a lower but longer one.

Distally, the two major condyles and the intercondylar groove or olecranon fossa that make efficient articulation with the ulnar process, are not variable. The external condyle, however, is significantly variable in the family. This condyle is longest and most pronounced in birds in which the humerus is short in relation to the trunk, as for example in Tachycineta. In the Bombycillidae the condyle is smallest in Phainoptila, where it is a mere suggestion of a process. In the remainder of the Ptilogonatinae, the condyle is larger but rounded, and shows a double process in Ptilogonys caudatus, and a slightly pointed process in P. cinereus. The external condyle in Dulus is not specialized, being low and rounded, but in Bombycilla, it is noticeably elongated, indicating a better attachment distally for the deltoid muscle. (No measurements are tabulated for this condyle, as the percentage of error in measuring this small structure is great.) Table 1 gives lengths of humeri, and Table 2 gives lengths of the humeri expressed as percentages of the length of the trunk, a standard measurement.

The area of insertion of the deltoid muscle is elongated in those birds with shortened humeri; these birds have also greater flight power than do birds with longer humeri and therefore a shorter external condyle.

Pygostyle.—This part of the skeletal system is variable in the species dealt with, not so much in size as in complexity. It reflects, of course, the character of the caudal muscles and their size, as well as the length of the rectrices and the corresponding force necessary to hold these feathers upright and in a useful position. Firm attachment is important even in flight, because the tail is used as a rudder, and in the Ptilogonatinae as a brake. The pygostyle is most modified in this subfamily.

In lateral aspect, the pygostyles of the species of the Ptilogonatinae are similar. The crest of the bone is flattened dorsally, and has a broad anterior surface that is thin and bladelike. This is widest in Ptilogonys caudatus, and narrowest in Phainoptila, in which genus, however, the entire bone is of small size. The centrum is widest in Ptilogonys caudatus, and is progressively narrower in P. cinereus, Phainopepla, and Phainoptila. Greater width provides a larger area of attachment for the larger rectrices and also more area for insertion of the lateralis caudae muscle, the size of which varies more than that of the other caudal muscles in the different species of the Bombycillidae.

In proportionate size (see Table 7), the pygostyle of Bombycilla is the smallest in the family. The dorsal spinous portion is acutely pointed instead of flattened as in the Ptilogonatinae. In Dulus, the spinous portion is extremely thin, and shows a decided curve dorsad from the centrum, and there is no flattened area anterior to the spinous portion as is seen in Ptilogonys.

The centrum in cross section varies considerably. In Bombycilla the walls are indented, with definite terminal knobs; both knobs and indentations are more pronounced in B. garrula than in cedrorum, however. The spinous portion is enlarged in both species, and the rest of the neck region is constricted (Figs. 29-35).

The centrum of Dulus in posterior aspect presents the appearance of a simple shield; little of the indentation seen in Bombycilla is present. The spinous portion is plain, with no constriction nor terminal enlargement in the neck. The centrum in Phainopepla is similar to that in Dulus, but has a small expansion at the base of the spine, the entire centrum being wider in proportion to its over-all size than in any of the other species mentioned previously. The centrum in Ptilogonys shows great width, and the spine is in a large expanded tip as in Bombycilla. The lateral edges of the centrum in P. cinereus are "winged" and in two separate halves; whereas the centrum of P. caudatus is fairly plain, its specialization being reflected primarily in breadth and flatness. In cross section of the centrum, Phainoptila is similar to Phainopepla, although, in the former, the bone is smaller in proportion to the size of the animal, and the lateral wings are more angular than in Phainopepla.

In specialization for muscle attachment, the centra of the pygostyles of the Ptilogonatinae have more area for muscle attachment than do the centra in the Bombycillinae and Dulinae; the centrum is wide, the spinous portion is long, and the bone is flattened anteriorly. The most generalized pygostyle is in Phainoptila, and that of Dulus differs only slightly. In Bombycilla the pygostyle is proportionately small, but is complex in shape; there is seemingly not the need for greatly expanded areas since the caudal muscles are less specialized in this genus.

Sternum.—The sternum in Bombycillids is typically passerine in general shape and in having a long and deep carina or sternal crest. The caudal process of the bone is broad, with the terminal ends flattened, forming dorsally a graceful V-shaped outline, whereas the outline of the posterior end of the sternum is broad and convex.

In lateral aspect, the carina is deeper in Bombycilla than in other genera of the family, and is deepest in B. garrula. In this species, the manubrium is more extended and comparatively larger than in the other species of the family. The anterior edge of the keel forms the sharpest angle in B. cedrorum. In Dulus, the keel is moderately deep, the manubrium short, and there is a distinct indented curve between the manubrium and the anterior angle of the keel.

In ventral aspect the lateral processes of the sternum tend to flare outwards in adult Ptilogonatines on almost the same plane as the rest of the bone, whereas in Bombycilla and Dulus the same process is closer to the body of the sternum. In Bombycilla the xiphoid process is more dorsal in position than in other species in the family, and in Dulus an upward curve is very noticeable. The process in these two genera is narrower than in the Ptilogonatinae, and lacks the heavy distal terminal enlargement which is apparent in Ptilogonys.

Relative Lengths of Bones.—In instances where the animals being compared are obviously different in over-all size, it is useful to express the size of a given part in relation to some other part of the same individual organism if the aim is to obtain clues as to differences in functions of the parts being compared. Differences in actual lengths of corresponding bones in two kinds of animals often, of course, reflect only the difference in over-all size of the animals. Consequently, the relative size of the part is expressed as a percentage in this paper. In computing a percentage it is well, of course, to select some relatively stable part of the animal to use as a denominator in the mathematical expression that yields the percentage. The thoracic region of the vertebral column is thought to be such a part. For example, the length of the humerus divided by the length of the thoracic region yields, in Phainopepla and Ptilogonys, respective percentages of .84 and .85. These are roughly the same, whereas the actual lengths of the humeri are 2.21 and 2.39 cm.