On the Normal and Abnormal Development of the Feather

The feather is the most elaborate of all the derivatives of the Malpighian layer. The production of so complex a structure from so simple a layer of cells is a challenge to the developmental anatomist and has been studied for many years. Historically two main views of the fundamentals of feather development may be distinguished. In 1889 Davies published a very full account in which he described the growth as being essentially a simple lengthening, by increment at its base, of an ectodermal cylinder surrounding a mesodermal core. He was supported in this view by Strong (1902). Both of these authors thought of the barbs developing in the ectodermal cylinder of the protoptile as being each added to from below by the incorporation of cells always in the same generator of the cylinder, these barbs thus being parallel, with no rachis. In the ordinary feather (teleoptile) each barb rudiment was regarded as growing by the incorporation into it from below of cells in generators of the cylinder lying progressively, from tip to base, nearer and nearer to that unique ‘dorsal’ generator which will become the rachis. Thus, on the later splitting of the cylinder along a ‘ventral’ generator (opposite to that of the rachis) the barbs would slope toward the rachis, from above downwards, as, of course, they do in the familiar feather.

Because of the morphology of certain feather defects, and of the shapes of the areas on some feather vanes showing a change in colour resulting from the injection of hormones (particularly of oestrone) into the bird, another view of feather development was advanced by F. R. Lillie in 1932. He supposed that there was a real tangential movement of actual tissue during the formation of barb ridges, from the ‘ventral’ to the ‘dorsal’ (rachis) generator of the cylinder. The tissue which he supposed to move tangentially in this way he considered to be made up of foci for the development of barb ridges, which formed barb distal to themselves all the time until they arrived at the rachis and became incorporated into it. This, the so-called theory of ‘concrescence’, has been modified from time to time (Lillie & Juhn 1932; Lillie & Wang 1941, &c.).’Espinasse in 1939 supported the views of Davies and Strong in his interpretation of the reactions of developing feathers to hormones and to injury.

It has been established by the work of Lillie & Wang (1941, &c.) that the activity of the ectodermal cells which results in the formation of a feather is initiated and to some considerable extent influenced by the dermal papilla, a clearly recognizable structure lying beneath this ectoderm and in close association with it, first to be seen in this location (Lillie, 1942; Sengel, 1958) after about 6 days of incubation. If the dermal papilla in an adult is destroyed, the follicle regresses and no further feather is produced. Moreover, the orientation of feathers produced is determined by the orientation of this dermal structure.

Until now all theories have been concerned with events, notably barb formation, in the living unkeratinized base of the ectodermal feather cylinder which has been termed the ‘collar’. However, this collar is lost with every plucked feather. Because of this loss such theories did not account satisfactorily for the repetition of certain chimaerical patterns in feathers growing from artificial combinations of papillar rudiments. Neither did they really account for the repetition of normality in series of feathers growing from a normal papilla. The events responsible for the repetition of normality, of some chimaerical feathers, and of feathers with certain anatomical faults, cannot be located in the collar. They must occur in the persistent papilla.

We propose to consider the production of such feathers in three stages :

• (1) The healing of a papilla from which a pin-feather has been plucked, resulting in

• (2) the establishment of a dome of ectoderm whose base is a new collar;

• (3) the transformation of this new collar into a new feather and its consequent loss.

The present paper reports observations, and advances arguments, which show, we believe, that all the experimental results we have obtained, as well as all those we have found in the literature, can be accounted for only if the persistent papilla is understood to give rise in a particular way to a series of transitory collars, each of which in turn, by a particular mode of growth, becomes a feather.

The birds used were Brown Leghorn capons and Light Sussex and White Leghorn hens.

The developmental potentialities of the collar were studied in tissue culture. Small pieces of the collars of pin-feathers were taken from different levels, always proximal to any visible formation of barb ridges, and from different generators, and cultured in a semi-solid medium consisting of 50 per cent, arterial fowl plasma and 50 per cent, chick embryo extract, undiluted (Cohen, 1959). In successful cultures the explant appeared to be of pure, or nearly pure, ectoderm; keratinization did not early supervene, and infection was not serious. These conditions were realized in about a third of the attempts made. In these cases sheets or scrolls of epitheliocytes grew out from the explant and showed a significant degree of organization which could be recognized with certainty as barb formation.

The mode of the transformation of collar into cylinder was also studied by examination of the distribution of mitoses in the bases of young pin-feathers incubated for 4 hours in 0·04% colchicine in Tyrode’s solution. (Attempts to arrest mitoses in feather collars and papillae by injecting the drug into the living bird were unsuccessful.) Sections and split flattened preparations (Lillie & Juhn, 1932) were stained in orcein or Heidenhain’s iron haematoxylin. Alternate 10-µ sections (in which inferior umbilicus was included) were counted, and a ruled eyepiece made possible the division of the collar into 20-µ zones from the plane of separation from the papilla.

The minute anatomy of the papilla was studied in dissections of follicles and in serial sections of papillae from which pin-feathers had been plucked and which were beginning to produce new collars.

The developmental potentialities of the papilla were further studied by transplanting papillae, from the top of each of which a pin-feather had been plucked immediately before the operation, as in the work of Lillie & Wang (1941), leaving the mesoderm exposed at the top of the papilla only. Mesodermal papillae which were partially or completely stripped of their epithelial coats (Cohen, 1959), truncated papillae (from which the whole distal end had been removed), and papillae which had been split sagittally, were also implanted. Feathers which grew from all these papillae, through several feather generations, were examined and filed and compared with one another and with the illustrations of feathers given by other authors.

In two of the Brown Leghorn capons an area of the neck was subjected to X-irradiation at a time when most of the follicles in the chosen area contained pin-feathers. The two birds each received a dose of 2,000 r over an area of some 6 sq. cm. No filter was added, the estimated half-value layer being equivalent to 1·25 mm. Al. (anode voltage 80 kVp., current 70 mA., F.S.D. 10 cm.).

The material reported on here was in part that considered in a different connexion and illustrated in a previous paper (Cohen, 1959) and in part new. Much of the material, and some part of the arguments presented in this paper were submitted to Hull University in a Ph.D. thesis by one of us (J. C.).

All the collar fragments chosen for explantation were taken from positions proximal to the region of visible barb formation and consequently had no barb ridges. Development of barb ridges occurred in about half the satisfactory cultures (79 of 164), both in those cases in which fibroblasts had previously made their appearance and in those where the only outgrowth was of epitheliocyte sheets and scrolls. It was considered that in these latter cases no mesoderm had in fact been explanted. In many cases, therefore, barb ridge formation occurred in epithelium when it was not under any mesodermal influence. These barb ridges appeared normal in that they produced barbs with barbules and were in many cases normally pigmented. The orientation of the explants reported on here was certain, since one boundary of each included a length of the torn lower edge of the plucked pin-feather while the other edges had been cut.

In one of the satisfactory cultures taken from the ventral generator of the pin-feather an almost normal ‘ventral triangle’ developed (Plate 1, fig. F; proximal is to the left) orientated correctly with respect to the pin-feather from which it had been taken. However, in all other cases from anywhere round the collar, even including the ventral generator, the orientation of the barb ridges which appeared was at random. The barb ridges appeared both in the explants themselves and in the areas of outgrowth, one or both situations occurring in any individual culture which showed any barb formation at all. In many cases two or more areas of barb ridge formation could be seen and the orientation of the barb ridges in any area might be at any angle to that in any other area (Plate 1, fig. E). In this last culture, and in some others, the whole of the explant and its outgrowth eventually became transformed into barbs with barbules. That is to say that these collar explants did not give rise to feather, but became feather.

We believe that all the feather structures which occurred in our cultures were the result of such a process of transformation, whether all the tissue in any one case was involved or not. We think that the production of feather in vivo is to be understood in terms of such a transformation.

For mitotic counts in longitudinal sections an arbitrary volume was chosen in each pin-feather such that it always contained the whole of the inferior umbilicus and thus the longitudinal axis of the pin-feather. These volumes in different feathers were orientated at random in a radial sense. No difference in counts was found in volumes of differing orientation. Counts were made in alternate sections of the ectoderm of this volume from the plane of separation up to the level at which keratinization made counting impossible. It will be realized that each of these sections contained two longitudinal sections of the ectodermal collar and cylinder, diametrically opposite to each other, both of which were counted. Counts were also made in the ectoderm of whole split flattened preparations.

Table 1 gives figures for the average of the numbers of mitoses counted in 20-µ zones proximo-distally. Keratinization usually begins about the 6th to 8th zone. Below this point there is little proximo-distal disparity between the zones in frequency of mitoses. This must mean that the transformation of the original collar, when no barbs were present, is of a very particular kind. Even when the apical part of the feather is being produced by transformation of the extreme distal part of this original collar, the rest of the collar is lengthening by intus-susceptive division. As the distal part is pushed distally, so it transforms. Thus while transformation of distal collar involves cells already forming their part of the original collar (or their immediate descendants), transformation of progressively more proximal collar involves the progressively more remote descendants of the cells which formed the original collar.

In vertical sections of a growing feather in situ (Plate 1, figs. A, B) it is observed that around the apex of the papilla the epidermal layer which invests the mesodermal papilla is reflected outwards to form the collar, this collar being in effect the proximal end of an epidermal cylinder which opens more distally to give definitive feather. When a growing feather is plucked the break occurs between the ‘papillar ectoderm’ and the collar. By ‘papillar ectoderm’ we mean simply that ectoderm which can be seen to remain investing the dermal papilla when a feather has been plucked, if the situation is examined in fresh unfixed material, and can be recognized if such a papilla is sectioned (Plates 1, fig. D; 2, fig. G).

The diagram of the anatomy of the base of a growing feather in situ given by Lillie & Wang (1941) (see Plate 1, fig. B) has, it is supposed, been derived from longitudinal sections closely comparable with that illustrated in Plate 1, fig. A. In such fixed material the base of the developing feather seems indeed to be embracing the dermal papilla. Text-fig. 1 shows diagrammatically how, in our view, the situation in life is distorted by fixation. It is natural to suppose that in plucking a growing feather from a situation as shown in Plate 1, figs. A, B, and Text-fig. 1c, the break in the ectoderm would occur in the thickness of the ectoderm and might even extend outwards on to the follicle wall (the line of break in Lillie & Wang’s diagram (Plate 1, fig. B) is presumably indicated by r.c. = regeneration cells). However, the situation as observed in the live feather follicle is very different : the base of the developing feather can be seen to be situated above the papilla, which shows a distinct ‘waist’ (see Plates 1, fig. D; 2, fig. G). It can be appreciated that in such a situation it must be that the developing feather is in reality plucked off the top of the papilla, the only ectodermal wound surface being a ring around the top of this papilla. This may be seen to be the case in Plates 1, figs. C, D; 2, fig. G, where the smooth keratinized surface of this papillar ectoderm continues from the follicle wall upwards to the transverse plane at the top of the papilla at which the developing feather had been connected : the plane of separation. It will be argued that it is by the healing over of the exposed mesoderm at the top of the papilla by the growing centripetally, in a very particular way, of the ring of exposed ectoderm here that the collar to form the next new feather is established.

The mode of this healing process was studied in 68 papillae, 5 from the neck, 34 from the saddle, and 29 from the breast. These were dissected from their follicles 3 to 20 hours after pin-feathers had been plucked from them. Before they were removed the dorsal aspect of each was marked with Nile blue sulphate which could be seen throughout the subsequent histological processing, and in addition a flap of follicle tissue was left attached to each papilla on its dorsal side, so that the correct orientation of longitudinal sections could be subsequently assured.

Considerable variation was found in the period of time elapsing before regeneration began. Six papillae presented healed-over distal surfaces after a latent period of only 7 hours, while seventeen showed no regeneration after 20 hours. Most of the papillae showed some ectodermal growth at 10–12 hours, and regeneration of a complete ectodermal dome was completed in 1–4 hours after growth began. The process of healing-over was observed directly in live papillae under a binocular microscope. Plate 3 shows views from above and from the side of such a healing papilla. Plate 2, figs. K, L, M, N show sagittal, or para-sagittal, sections of saddle and breast papillae in early and late stages of this ectodermal over-growth. It can be clearly seen that the greater contribution comes from the dorsal side (the left-hand side in the figures). In all the papillae examined maximum growth was from the dorsal sector of the ectodermal woundedge. This observation plays a major part in the development of the arguments to be presented here (pp. 238–9).

This transplantation series involved the exchange of papillae between brown saddle and black breast tracts in four Brown Leghorn capons and so constituted a repetition of some of Lillie & Wang’s work (1941, 1943, 1944). The results are summarized in Table 2. Some follicles received so-called ‘intact’ papillae, that is, papillae on which, on implantation, mesoderm is exposed only at the top; the sides remain completely invested in their coat of native ectoderm. These follicles produced, in all cases in which feathers were produced at all, feathers of donor tract structure and colour. Those follicles which had received papillae from which an attempt had been made to remove the epidermal coat (‘stripped’ papillae), all produced feathers of recipient tract structure and colour, or, in two cases, chimaerae. In these chimaerae the boundary between the two recognizable structures, of saddle and breast feathers, almost coincided with that between the two colours. Saddle follicle 17, which had received a collodion ‘stripped’ breast papilla (Wang, 1943), produced a feather of donor tract structure and colour after having produced a chimaerical feather. All these observations consistently agree with those of Lillie & Wang.

This series involved the exchange of papillae between the black-and-white neck and white flank tracts in one Light Sussex hen. The results are summarized in Table 3. None of the papillae was ‘stripped’, and flank follicles which had received neck papillae produced donor (neck) feathers only, with the normal black-and-white coloration. However, four of the five neck follicles which had received ‘intact’ flank papillae, although in the first generation producing three white flank feathers, in the second generation produced three neck feathers and a chimaera, and in the third generation produced four neck feathers. The fifth follicle produced nothing. It is supposed that local ectoderm did ultimately replace the donor ectoderm during a process of reconstitution in this case, but did not in the reverse situation.

This series involved the exchange of papillae between the black-and-white neck and the white saddle tracts in two Light Sussex hens. The results are summarized in Table 4, and some feathers illustrated in Plate 4, figs. R, S, T, U, V. All the eleven follicles (36, 38, 39, 43, 44, 45, 46, 48, 49, 50, 51) which had received ‘intact’ papillae, and which produced any feathers, produced feathers of donor tract structure, except follicle 46 which produced, after a long period of inactivity, one vestigial feather whose structure could not be analysed. Of the five follicles 37,41,42,47, 53, which had received ‘stripped’ papillae, four produced feathers of recipient tract structure and one, 53, which had had the ectoderm removed from the dorsal aspect only, produced two chimaerical feathers. The following five follicles are now considered in more detail:

Saddle follicle 41 had received a neck papilla which, as well as having been ‘stripped’, had been split sagittally for about half its length. This produced in the first generation a feather whose rachis was split almost to the calamus, each half-rachis possessing little more than a half-vane (Plate 4, fig. R, centre); the after-feather was single and normal saddle in structure. This feather was allowed to mature before it was plucked. The second generation from this follicle was a single feather of saddle structure, apparently normal, but with a double afterfeather of the same type (Plate 4, fig. R, right).

Saddle follicle 43 had received a neck papilla which had not been ‘stripped’ but which had been split sagittally for part of its length. This produced in the first generation a neck feather whose apical half was double (Plate 4, fig. S, centre). The next generation from this follicle consisted of two feathers, one of which was entirely neck in structure and colour, and the other chimaerical: saddle distally and medially but neck proximally and laterally (Plate 4, fig. S, right). All three after-feathers were normal.

Saddle follicles 44 and 45 each received a truncated neck papilla whose tip had been removed but which had not been ‘stripped ‘. Each produced two generations of neck feathers whose tips were absent or defective (Plate 4, figs. T, U).

Neck follicle 53 had received a saddle papilla ‘stripped’ of its ectodermal coat on the dorsal aspect only. This produced in the first generation a chimaerical feather (Plate 4, fig. V, centre) which may be analysed as follows :

The apex of the feather is of donor saddle structure and this region extends proximally at the edges of the vane; the entire proximal region of the feather is darkly pigmented, including most of the fluff. The median part of the vane proximally is of recipient neck structure, while the structure of the fluff does not resemble either that of the neck feather or that of the saddle feather. A very small after-feather is present, about 0·5 cm. long; the after-feather of the saddle feather is usually about 2–3 cm. in length and that of the neck feather usually 1·5–2 cm. The rachis width is about intermediate between that of saddle and neck feathers, and it is deeply pigmented down to the point of insertion, a condition which is not normally found either in the saddle or neck feathers of the breed.

In the second generation the follicle produced another chimaerical feather (Plate 4, fig. V, right) whose apex was again of saddle structure, and which had a region of neck structure diagonally across the vane. Parts of the fluff resemble that of the saddle feather, and other parts that of the neck feather, while some of the fluff was not typical of either tract. There was an abnormal after-feather 0·2 cm. long.

Some of the irradiated neck follicles had before the operation contained feathers which showed depigmentation attributable to the previous irradiation, marked * in Table 5, and some had contained normally pigmented feathers. In both sections (a) and (b) below, structure will be considered first and then pigmentation.

This series involved the exchange of papillae between previously irradiated neck tract and unirradiated breast tract in two Brown Leghorn capons. The results are summarized in Table 5, and some of the feathers illustrated in Plate 4, figs. W, X, Y. All papillae were implanted ‘intact’.

Of the twenty-nine feathers produced in the first two generations, all but one were wholly of donor tract structure. The one exception, in neck follicle 61, first generation, was a chimaera. Neck follicles 60 and 61, and breast follicles 66, 67, and 68 then produced in the third and subsequent generations feathers of donor tract structure only.

Subsequent generations from the following three follicles are now considered in more detail :

Neck follicle 58 produced in the third generation a chimaerical feather the tip of which was neck structure with the remainder of the vane breast structure. Generations 4 and 5 were of neck structure, generation 6 was a vestigial feather of breast structure, and generation 7 was a large normal breast feather. Presumably donor (breast) ectoderm persisted in the papilla but on two occasions (generations 4 and 5) failed to contribute to the collars and so to the feathers.

Neck follicle 59 produced, in the third generation, a very abnormal chimaerical feather with extremely thick barbs, a small part of which was of neck structure (see Plate 4, fig. W, centre). The neck elements in this feather were recognized by both the shape and the colour of the barbs. All four subsequent generations of feathers from this follicle were of normal neck-feather structure. Presumably host ectoderm had replaced donor on the papilla and gave a series of neck collars.

Breast follicle 68 produced, in the third generation, another feather of normal neck structure, and in the fourth generation a chimaerical feather, mainly breast structure but with some of the tips of the barbs of neck structure. After a period of apparent inactivity this follicle then produced one vestigial feather of breast structure. Presumably host ectoderm had replaced donor ectoderm during the period of inactivity.

The irradiated neck follicles 56, 57, 60, and 61 had before the transplantation contained white or mostly white feathers. Their papillae were transplanted into breast follicles 64,65, 68, and 69. Neck follicles 54, 55, 58, and 59, however, although they had been irradiated, had before the transplantation contained normally pigmented neck feathers. Their papillae were implanted into breast follicles 62, 63, 66, and 67.

A loss of pigmentation is to be observed in feathers of neck structure produced in all these breast follicles which had received previously irradiated neck papillae, whether or not these papillae had, before transplantation, produced feathers showing any loss of colour. Of the 33 feathers which were wholly or partly of neck structure growing in breast follicles, 22 displayed decoloration in recognizable neck feather, and in neck feather only. On the other hand, feathers of breast structure growing from unirradiated breast papillae transplanted into irradiated neck follicles never showed depigmentation.

This series involved the exchange of papillae between previously irradiated neck tract and unirradiated saddle tract in one Brown Leghorn capon. The results are summarized in Table 6. Eight papillae in all were transplanted of which two were ‘stripped’ and six ‘intact’.

Neck follicles 70, 71, and 73, and saddle follicles 74 and 77, all of which had received ‘intact’ papillae from the donor tract, contained, when feathers were produced at all, feathers of donor structure, though the last feather produced by saddle follicle 77 was vestigial.

The following follicles are now considered in more detail:

Irradiated neck follicle 72, which had received a ‘stripped’ saddle papilla produced a series of three neck feathers as was to be expected.

Unirradiated saddle follicle 75, which had received an ‘intact’ irradiated neck papilla contained in the first two generations neck feathers as was to be expected, but in the third a saddle feather.

Unirradiated saddle follicle 76, which had received a ‘stripped’ irradiated neck papilla contained in the first generation a chimaerical feather, most of which was of saddle structure but the tips of whose barbs were neck. In the second generation this follicle produced a tuft neck feather, in the third generation a normal saddle feather, and in the fourth an extremely defective saddle feather.

In this series, again, pigmentation lack, attributable to previous irradiation, appeared only in feathers of neck structure, and all but one of the eleven completely neck feathers showed it. The one chimaera showed normal neck pigmentation of the neck elements.

The observations reported under (a) and (b) above seem to provide strong evidence that the depigmentary effect of the X-irradiation acts via the epidermis, and that this epidermis retains the effect through seven feather generations even when transplanted to an unirradiated region, in which local unirradiated melanoblasts are available but do not contribute pigment to the irradiated donor epidermis (compare Cock & Cohen, 1958).

Most of the feathers in sections (a) and (b) above which showed depigmentation carried a normally coloured patch distally. This pigmentation pattern was commonly repeated to a greater or lesser extent in subsequent generations. These feathers may, we feel, be regarded as ‘transverse chimaerae ‘in some sense. This will be referred to in the discussion, p. 241, as we have offered evidence that the occurrence of depigmentation does indeed result from an abnormality of the ectoderm concerned.

The observations reported here accord with the view, which does not appear to be inconsonant with that of Lillie & Wang (1943), that a new collar is formed from the persistent papilla before a new feather can grow to replace one that has been plucked.

Those previous workers who have been concerned with production and repetition of normality, defects, or chimaerical structure in feathers, have known the structures only of the papilla and of the feathers finally produced. They have derived intermediate stages from these structures. We propose that this derivation has in fact been inadequate in that repetition has not been explained by previous theories.

We propose to derive the structure of the collar both from a consideration of the mode of the transformation of it into a feather of known structure, and by a consideration of its production by a papilla of known constitution. We will then attempt to describe the production of feathers by various abnormal papillae in these terms.

The development of barbs in collar fragment cultures from generators other than the ventral, and the absence in these cultures of any ‘ventral triangle’, seem to us to render completely untenable any theory involving concrescence as a necessary element in barb formation, because any theory of concrescence rests upon the belief that barbs can originate only in a ‘ventral triangle’, where their tips must be located. It seems to us, therefore, that the peculiarity of the ‘ventral triangle’ is just that barb formation (a potentiality of the collar all the way round) here takes place in two directions always at an angle to one another. At present there seems to be no indication at all which would point towards an identification of the factors determining the angle at which barbs form in vivo. In the teleoptile, with its vanes and its rachis, they do, in fact, form at angles to the longitudinal axis of the cylinder.

Thus the transformation of collar into feather can, we consider, be adequately accounted for if it is supposed, as in the classical view, that the only movement of actual material which takes place in the cylinder is upward by a very particular mode of elongation of the whole of the original collar. The distal part of the feather is thus derived by transformation of the distal part of the collar and progressively more proximal feather is achieved by transformation of tissue progressively more proximal in the collar. This also follows from our interpretation of the results obtained from the colchicine experiments (see above, p. 227).

The name ‘collar’ has in the past been given both to the whole of the original dome of growing ectodermal tissue now realized to be constituted in a particular way above the papilla, and then retained to mean the proximal (living) end only of the feather cylinder into which the whole collar in the first sense is lengthening. This double usage has led to misunderstanding. We mean by ‘collar’ living ectodermal tissue destined to be transformed into feather by a particular mode of elongation and most complicated subsequent differentiation. This covers both usages given above.

We wish now to direct attention to the papilla and to the manner in which this persistent papilla may be conceived of as giving rise to a series of collars, each transitory and destined ultimately to be lost by transformation into a feather. We wish to do this because the manner in which this happens must surely determine the constitution of the new collar, and so of the new feather.

It has been shown above (p. 228) that when a pin-feather has been plucked the only ectoderm on the papilla which is not undergoing keratinization is its edge around the top of the papilla, bounding the exposed apex of the dermal papilla. This is the only possible source from which can be derived the dome of ectoderm (Plate 2, fig. J) which comes to cover the exposed mesoderm (Plates 2, figs. H, K, L, M, N; 3, figs. P, Q) and whose tip marks the morphological, but not necessarily the actual, position of the tip of the feather. The ring of ectoderm constituting the base of the walls of this dome is the collar, which by its particular mode of growth becomes the feather.

Our examination of regenerating normal papillae from saddle, neck, and breast tracts shows that the proliferation is not annular but proceeds initially and mainly from the dorsal wound-edge (p. 229 above and Plates 2, figs. K–N; 3, figs. P, Q). This we relate to the observations of Lillie & Wang (1941) that there is a gradient of inductive activity in the dermal papilla, which is high dorsally and low ventrally. This may in turn be related to the observation of Hamilton & Koning (1956) that there is a similar localization in the distribution of alkaline phosphatase. Thus the dorsal wound-margin can be seen to contribute far more to the healing of the wound than does the ventral margin, which may contribute very little. The dorsal margin produces tissue which extends over the top of the dome into the region of ventral collar, as in Plate 3, fig. Q and as represented in Text-fig. 2.

In the normal case, of course, there is no way in which we can distinguish by inspection, and without reference to the position in the feather, between different parts of it derived from different parts of the collar. In the case of an artificially contrived situation, however, where a composite papilla has been made up by uniting the dorsal half of a breast papilla with the ventral half of a saddle papilla, we can distinguish by inspection between feather derived from the one constituent and that derived from the other, because they are of different colours and structures.

Feathers derived from such composite papillae have been called ‘longitudinal chimaerae’ by Lillie & Wang and described by them as having ‘a breast apex, followed by a series of barbs at each side, the marginal halves of which will be formed by saddle ectoderm and the central halves by breast ectoderm’. They further say that the proximal part of the feather ‘appears to be influenced by both components’.

In Text-fig. 2 the development of these feathers is pictured in terms of our views. In diagrams a, b, and c a papilla is seen from above. The ectoderm is represented as healing over from the dorsal as we have shown it normally does (Plates 2, figs. K-N; 3, figs. P, Q). Diagrams A, B, and c show the same papilla from the ventral aspect. In diagram c the ventral half of the new collar can be seen to be constituted distally of ectoderm dorsal in origin, which has grown over the top of the dome, and proximally of ectoderm which was always ventral. Dorsally, of course, the whole of the new collar is of ectoderm which has always been dorsal. As the feather has come into existence by a lengthening of the collar, the pin-feather has the dorsally derived and the ventrally derived ectoderms related to each other in ventral view as shown in diagram D, and the feather resulting from the splitting of this pin-feather down its ventral generator will have these tissues, of different origin, distributed as in diagram E. Where the dorsally and ventrally derived constituents of the feather can be distinguished by colour and structure, as in this case of a feather grown from a composite papilla by Lillie & Wang, the feather has in fact the derivatives of the two constituents, breast and saddle, distributed in it as shown in diagram E. At the same time the papilla, after the loss of this feather, will have its original double composition with breast and saddle components related as they were in the original composite papilla. We therefore believe that our suggestions above as to normal feather growth do for the first time account satisfactorily for the feathers found to grow from such composite papillae.

We now turn to a consideration of the feathers derived from composite and abnormal papillae of kinds other than this.

First we take the exact converse of this papilla, namely one made up of saddle dorsally and breast ventrally. Feathers derived from such a composite papilla have been called ‘transverse chimaerae’ by Lillie & Wang (1943), and described as having ‘a saddle apex, followed by a series of intermediate barbs with breast margins of gradually increasing extent and saddle centres correspondingly diminishing until the entire lengths of the barbs is breast, and the remainder of the feather also’. From their table 4, however, it appears that, as in their ‘longitudinal (marginal) chimaerae’ (their table 3), the base of the feather in fact contains both components. It seems doubtful whether the differences between these two cases, then, are really sufficient to justify calling the first ‘longitudinal’ and the second ‘transverse’ chimaerae. Such differences as there are appear to us to be satisfactorily explained by attributing them to differences in the exact shape of the overgrowing dorsal contribution to the distal part of the ventral collar during the healing of the papilla (Text-fig. 2). The shape of this will determine the slope or the curve of the boundary between the two kinds of feather, saddle and breast, in the finished product. According to the shape of this boundary those barbs involved will have had different origins along their lengths. Such barbs are described by Lillie & Wang.

So far our observations of regenerating papillae have suggested that saddle and breast papillae may in fact differ in the shape of the normal dorsal overgrowth and so we believe will contribute differently to the ventral collar. This we suggest may well account for Lillie & Wang’s observations that conversely constituted composite papillae produce feathers which are not the exact converse of each other. Each kind will show repetition in generations of feather as pictured in Text-fig. 2.

We consider that the irradiated feathers described above (pp. 233-7) and also those illustrated by’Espinasse (1959) can be explained in terms of our views. Some of these feathers have normal colour at the tip and continued down the vane medially, the rest of the vane being without pigment, and this condition may be repeated. The distribution in successive generations of pigmented and unpigmented areas in these feathers, therefore, is directly comparable with the distribution of dorsally derived and ventrally derived feather in Lillie & Wang’s chimaerae discussed above. Since in Lillie & Wang’s examples the differently coloured feathers were derived respectively from dorsal and ventral parts of the ectoderm of a composite papilla, we suggest that in these irradiation feathers it was the ventral ectoderm in the papilla that was affected by the irradiation while the dorsal ectoderm remained unchanged. The transplantation experiments reported on p. 237 provide independent evidence that it is indeed the ectoderm which is changed, whatever the nature of this change may turn out to be. These feathers may therefore be regarded as ectodermal chimaerae, comparable with the chimaerae of Lillie & Wang, and understood in terms of Text-fig. 2.

Other irradiated feathers, however, are completely unpigmented except at the tip. This condition also is sometimes repeated in successive generations of feathers. Therefore it seems inescapable that a replication of the entire pattern of normal and changed ectoderm must occur in the production of a series of collars. This necessarily implies the retention by the papilla of its own spatial organization. Further, on the dorsal side of the papilla the two kinds of ectoderm (affected and unaffected), which must be there, must contribute in turn to the formation of the collar. Thus there must be some spatial organization of papillar ectoderm which can result in a proximo-distal heterogeneity of the dorsal arc of each collar. We suggest that this can be accomplished by an undercutting of the contribution of the dorsal (unaffected) papillar ectoderm by affected ectoderm from more ventrally. This will result in an ‘island’ of dorsal (unaffected) tissue which must appear distally in the feather, and which may be repeated in subsequent feathers since unaffected ectoderm will remain dorsally in the papilla; see Text-fig. 3.

The feathers with defective or missing tips described by Lillie & Wang (1941) were produced from papillae whose distal ends had been removed entirely or on the dorsal side only. In the work reported here also, follicles received papillae whose tips had been removed entirely and each produced two generations of feathers with defective apices (see Plate 4, figs. T, U).

Here for the first time, then, we are considering cases where the dermal papilla has been rendered incomplete. See Text-fig. 4. We suppose that the inductive action of the dermal papilla normally occurs around its distal end, as all the ectoderm which will become collar can only be produced at this level. Further, Lillie & Wang showed that rachis may be induced in host follicle ectoderm by sectors of a donor dermal papilla from other than the mid-dorsal, provided that the origin of such a sector was more dorsal than that of other mesodermal papillar sectors present. This must mean that the inductive power of the distal part of the dermal papilla is in the form of a dorsal-to-ventral gradient. If the papilla is truncated, as in the cases under discussion, feathers are still produced, but with defective tips. Therefore there must be a distal-proximal inductive gradient and the truncated end, like the normal distal end, must be capable of producing only mesodermal pulp above it and not an extension of potent mesodermal papilla. Thus that edge of ectoderm which is called upon to produce the new dome after such truncation can only be affected by an inductor of lower potency, and the first ectodermal tissue produced from this edge has never been subjected to the action of a fully potent inductor. As, however, most of the first feather is normal, then full potency of the inductor must be soon attained during the establishment of its collar. Ectoderm initially produced is thus not capable of organized barb- and rachis-formation, although subsequently produced ectoderm must be so capable for normal feather to be produced, as it is. The abnormal ectoderm would be carried as an island to the top of the dome as was the normal tissue in the previous case of the white feathers with coloured tips. It will be seen that should the defective tissue reach over into ventral collar, then barb tips will be involved; however, in any case if the rachis is sufficiently defective, that length distal to the fault will fall off, carrying its barbs. The series of events described above would result in a feather with defective or missing apex.

When such a feather is plucked, another feather with defective apex frequently succeeds it. We must therefore suppose the break at plucking, or plane of separation, to occur at the level of the original truncation because all the collar is removed with the pin-feather. In the case of a feather which matures above such a papilla, the mesodermal pulp right down to the truncation level is sealed off in the feather at the closure of its inferior umbilicus.

Once the new dome with its collar has been formed, the still truncated dermal papilla returns to the condition of sub-potency which is normal to it as the base of a dermal papilla. Therefore the whole process can be repeated.

Split or double feathers may be produced from papillae which have been completely divided to different depths in the sagittal plane (Lillie & Wang, 1944). These feathers may be repeated, or subsequent generations may be single or may consist of two separate feathers. Implantation of a split ‘stripped’ (dermal) papilla, as in our follicle 41 may cause local ectoderm to invest this papilla in such a way that a split host-type feather is produced (p. 233, Table 4, and Plate 4, fig. R). Therefore it may again be said that the dermal injury is the primary cause of such feathers. A dermal papilla the whole of whose sagittal diameter has been destroyed to some depth by the cut cannot recover its initial unity in all senses, because in fact it does produce in some cases split or double feathers or, occasionally, two separate feathers, not only in the first but also in subsequent generations. We suggest that the scar tissue likely to be produced at the interface during the healing together into an anatomical whole of the two sides of the divided dermal papilla does in fact keep these two halves physiologically separate.

Our view of the development of feather from such an abnormal papilla, having a physiologically double dermal component, is represented in Text-fig. 5. The dermal papilla in such a case must have two ‘most dorsal’ sectors, each of which has been displaced laterally by the injury and subsequent healing process. We suppose each such sector to induce its own dorsal overgrowth of ectoderm, shown stippled. Should ectoderm from between these ‘most dorsal’ sectors, which lies around the disorganized sagittal plane of the dermal papilla, contribute to the medial aspects of these two overgrowths, then the double dome formed when all the mesoderm has been covered will possess areas which may only be capable of aberrant barb formation, and the median half-vanes of the double feather may be defective or absent. In the extreme case this will produce a feather with split rachis (Plate 4, fig. R). However, should the double dome be formed entirely of ectoderm which has been in contiguity with unchanged dermal papillar tissue, then a complete double feather will be formed (Plate 4, fig. S). It will be seen that the valley between the two domes extends down into the collar, which it may bisect only distally, as in Text-fig. 5, diagrams c, c₁. In other cases it may bisect the collar completely. In the latter case two feathers will result. That the top of the papilla remains anatomically single has been demonstrated in those cases (for example our follicle 41) where single rachis is produced in the next generation (Plate 4, fig. R). Subsequent transformation of a partially divided collar into a partially double feather is illustrated in Text-fig. 5 D, E. If the dermal papilla remains physiologically divided by its scar tissue after such a feather has been plucked, then series of split or double feathers will be produced (Text-fig. 5 E). If, however, at any plucking, physiological continuity of the two halves is established dorsally, a single feather will result. This re-establishment of continuity may or may not survive yet another plucking, so that later feathers may be single or double.

In one of our cases, a follicle (41) received a papilla which as well as having been split sagitally for about half its depth, had also been ‘stripped’, that is to say, only a dermal papilla, split, was implanted. This produced in the first generation a feather of host structure and pigmentation which was split about half-way down the rachis. The after-feather was single. This feather was allowed to mature, so that when it was plucked the inferior umbilicus had become closed off. The next feather was single all the way down but with a double after-feather. Here apparently, after the last stages of development of the previous feather, scar tissue in the mesodermal papilla established a discontinuity between the right and left ventral halves. This led to the initiation of two subrachis centres ventrally just as our Text-fig. 5 B₁, B, b, shows this to happen dorsally. This we consider compelling evidence that dorsal papillar ectoderm contributes to ventral collar not only after a pin-feather has been plucked, but also when the dermal papilla is healed-over by ectoderm at the closure of any inferior umbilicus.

The two feathers produced from neck follicle 53 (Table 4; Plate 4, fig. V) present a situation which cannot yet be fully explained. These feathers are analysed fully on p. 233. The first-generation feather produced after the implantation of a saddle papilla, ‘stripped’ on its dorsal aspect only, was a chimaerical feather which can be explained by supposing that the upgrowth dorsally of recipient neck ectoderm, to replace the saddle ectoderm which had been ‘stripped’ off, was retarded so that saddle ectoderm from the ventral side had grown over the top of the dome before meeting it. The next feather was a nearly typical saddle feather with a patch of neck structure and pigmentation on one vane, from which we conclude that for some reason the donor saddle ectoderm had increased at the expense of recipient neck ectoderm. Unfortunately, the bird died before producing another generation, which might have provided more information.

Other series of feathers, in which replacement of donor by host ectoderm has occurred cannot, we feel, as yet be explained in the terms of a general theory but must be investigated throughout their cycles. It is probable that as a result of damage and subsequent irregular wound-healing, ‘unstable’ papillar investments appear which are progressively replaced. It is perhaps significant that this phenomenon was observed in our work mainly in feathers produced from the earlier experimental papillae.

All the feathers described in this work and those illustrated by Lillie & Wang can, we believe, be understood in terms of the view of feather development we have put forward here.

1. On donne un bref résumé des idées antérieures sur le développement de la plume. On souligne ici que le collier ne doit pas être considéré comme produisant une plume, mais que lui-même se transforme dans sa totalité en plume et est ainsi perdu à chaque fois que celle-ci est arrachée.

2. On décrit la culture de fragments de collier. De nombreuses cultures ont produit des barbes et des barbules d’apparence normale sans aucun processus ressemblant à une concrescence. On en conclut que le seul déplacement de tissus intervenant au cours de la transformation du collier en plume est, essentiellement, une simple élongation, telle que la zone distale du collier devient la région distale de la plume et la zone proximale du collier devient la partie proximale de la plume. C’est la distance originelle entre le collier distal et le collier proximal qui augmente, pour donner la longueur totale de la plume. Ainsi, la structure d’une plume particulière peut être retracée de manière simple; on peut en déduire la structure du collier qu’elle fut auparavant.

3. Le collier, néanmoins, est effectivement une production de l’ectoderme papillaire, et les modalités de cette production sont considérées comme un problème topologique. Une plume présentant une anomalie distale, mais qui est normale dans sa région proximale, et provenant d’un collier anormal distalement mais normal proximalement, peut se renouveler selon le même modèle. Dans ce cas, la papille doit redevenir anormale après avoir produit la partie proximale normale de chaque collier, de façon à produire ultérieurement un autre collier, présentant une anomalie distale semblable.

4. Des résultats d’expériences de transplantation de papilles décrites ici et de ceux obtenus par d’autres chercheurs, on déduit que les modalités de la formation d’un collier à partir d’une papille impliquent une croissance préférentielle du tissu ectodermique dorsal: le tissu qui était dorsal sur la papille s’accroît au-dessus de l’extrémité et vers le bas du côté ventral du dôme produit à l’extrémité de la papille en cours de cicatrisation. Selon les faits observés, ceci se produit à la fois quand on arrache une plume naissante et quand une plume arrive normalement à maturité par fermeture de son ombilic inférieur.

5. Les auteurs prennent en considération, et expliquent selon leurs vues personnelles, des chimères ‘longitudinales’ et ‘transversales’, des plumes doubles, des plumes à extrémité défectueuse, et leur réapparition.

We wish to express our gratitude to Dr. Nowell and Dr. John for their helpful suggestions and to the University of Chicago Press for permission to reproduce three illustrations.