An analysis of the condensation process during chondrogenesis in the embryonic chick hind limb

For many years it has been believed that the first indication of chondrogenesis in the embryonic vertebrate limb is the appearance of a central blastema of closely packed mesenchyme cells with rounded profiles, which are destined to form cartilage. These cells then move apart as a metachromatic intercellular matrix is established. This classical concept of early chondrogenesis based on cell ‘condensation’ was established by the work of Fell (1925) and Fell & Canti (1935), and this phenomenon has since been described at the light-microscope level by a number of other workers (Montagna, 1945; Saunders, 1948; Ham, 1969). Grüneberg (1963) has called this the ‘membranous skeleton’ and states that ‘the first sign of skeletal development is the formation of condensations …

These consist of closely-packed cells in which the nuclei are almost in contact with each other.’ However, recently reports have been published based on electron microscopic analysis which appear to deny the existence of the early phase of intimate cell association (Gould, Day & Wolpert, 1972; Searls, Hilfer & Mirow, 1972; Gould, Seiwood, Day & Wolpert, 1974). According to these authors, the mesenchyme cells differentiate into chondroblasts and commence matrix secretion without ever exhibiting the transient condensation stage.

It is important to establish exactly what does happen in the pre-cartilaginous skeletal blastema at the cellular level. Knowledge of the early ‘behaviour’ of the mesenchyme cells is important not only to a fuller understanding of subsequent histogenesis, but further it relates to current hypotheses of pattern formation in the pentadactyl limb (for example Ede, 1971) and of cartilage induction (Holtfreter, 1968).

This communication describes a light- and electron-microscopic analysis of chick limb-bud chondrogenesis. The morphology of the cells and matrix in the femur, tibia/fibula region is described. The results are in line with what might be predicted at the electron-microscope level from the classical light-microscope studies. The mesenchyme cells are more closely associated during the early stages of chondrogenesis before they become spatially isolated by progressive matrix accumulation. The findings are discussed in relation to earlier analyses by light microscopy and in the context of reports of in vivo chondrogenesis in embryonic limbs of other vertebrates. Some attempt is made to resolve the apparent conflict arising from recent publications with regard to the classical story. The significance of the condensation phenomenon is discussed in the light of recent reports concerning intercellular relationships during in vitro chondrogenesis.

A preliminary survey was made of the density of cell packing at standard points in the limb-bud during early development: stages 20, 22, 24 and 26 (Hamburger & Hamilton, 1951) were chosen. The two sampling areas, taken at the mid-point along the proximo-distal axis (Fig. 4, Thorogood, 1972) were (i) central (stages 20, 22) and central-chondrogenic (tibia, stages 24, 26), and (ii) peripheral-undifferentiated and non-chondrogenic : the latter was taken from the anterior margin thus avoiding the myogenic areas (see later, Figs. 2–4). Hind limb-buds from Light Sussex embryos were fixed in Bouin’s fluid and wax sections were cut at 6 μm and stained in Ehrlich’s haematoxylin. Sectioning of the limb-buds was transverse and serial along an antero-posterior axis. Counts were made under oil immersion at × 600 within an area of the section measuring 0·069 by 0·05 mm. Ten counts were made for each category of sampling area at each of the four stages chosen. The means and standard deviation were calculated and the results presented as a histogram.

It was felt that a ‘mapping’ of the mesenchyme cell condensation (both myogenic and chondrogenic) at the light-microscope level was necessary before proceeding further. A series of whole limb-buds were fixed with a modified electron-microscope fixation procedure incorporating Alcian blue (Behnke & Zelander, 1970). The inclusion of the Alcian blue resulted in a subsequent staining of any developing cartilage matrix. Dehydration and embedding followed the procedure described later for electron microscopy (see below). The embedded material was sectioned at 1 μm on an LKB 11800 Pyramitome. The tissue was orientated in such a manner that sectioning was transverse across the limb-bud (from the anterior to the posterior edge) and vertical (from the dorsal to the ventral surface). These transverse sections were serial in a distal proximal sequence from the tip of the limb towards the body wall. For analysis representative sections were collected and retained at 50 μm intervals. In this way, by counting the sections as they were cut, the distance of any section from the limb tip could be ascertained. The sections were stained in 1 % toluidine blue in borax. Camera-lucida tracings of the tissue topography within the limb were made. In Figs. 2–4 representative transverse sections have been selected at 250 μm intervals.

Initially a number of electron-microscopy fixation procedures were employed but fixation artifacts were encountered. Embryonic tissues often present such problems (Szollosi, 1967), believed to be due to their high water content. In this case fixation resulted in partial disruption of membrane systems, especially of the mitochondria, where cristae often disintegrated and vacuolation was observed. To eliminate this problem a simultaneous double fixation procedure was employed (Trump & Bulger, 1966). This minimizes the effects of lipid extraction and cell shrinkage normally encountered and with the tissue being examined in this investigation gave a reliable standardized fixation.¹ Hind limb-buds were removed from embryos at stages 22–26 and processed by this method. The tissue was block stained in 5 % uranyl acetate, dehydrated through an ethanol series, cleared in propylene oxide and embedded in TAAB resin (Taab Laboratories, Reading, Berkshire, England). Survey sections (1 μm) for light microscopy were cut on an LKB 11800 Pyramitome, and stained with 1 % toluidine blue in borax. Ultrathin sections were cut at 60–90 nm on an LKB 4800A Ultratome I, and double stained with 5 % aqueous uranyl acetate and lead citrate (Reynolds, 1963). The sections were examined in an AEI EM 6B and a Philips EM 300 electron microscope at 60 kV.

The analysis of comparative density of cell packing in peripheral and central (chondrogenic) regions in the normal developing hind limb is represented in Fig. 1. In the peripheral areas the cell density remains approximately constant from stages 20 to 26 inclusive. However, in the central sampling areas along the proximo-distal axis of the limb, which are to become chondrogenic, there is a considerable rise in cell packing. Cell numbers are constant at stages 20 and 22 and of a similar order to those in the peripheral areas, but by stage 24 there is a rise of 62 % over the previous stage-22 figure. It is at this time that the skeletal condensation first becomes distinguishable in wax sections. This increase in cell packing density continues into stage 26 although by this time the rate of increase has perceptibly lowered, the stage-26 figure being only a 10 % rise over the equivalent stage-24 density figure. Thus the condensation phase in development of the skeletal elements involves a considerable increase in the density of the mesenchyme cell packing just prior to stage 24.

Generally there is a gradation of differentiation along the proximo-distal axis of the limb-bud at each stage. Differentiation of cartilage, muscle and other limb tissues is more advanced proximally and this, presumably, is the consequence of the proximo-distal growth of the early limb-bud. This gradient is reflected within the early condensations themselves. The earliest recognizable form of any condensation, myogenic or chondrogenic, is a localized area of increased mesenchyme cell numbers but without any obliteration of intercellular spaces. More proximally the condensations begin progressively to exhibit the cell and tissue morphology characteristic of either early muscle or cartilage.

In the limb at this stage well-defined myogenic condensations of closely packed mesenchyme cells are present at the proximal level (see Fig. 2). These condensations lie peripherally beneath, but separate from, the dorsal and ventral ectodermal surfaces. In transverse section the condensations appear as flattened sheets of cells and in the intact limb they are elongated along the proximo-distal axis of the limb. Distally they merge into less distinct areas of increased cell packing immediately subjacent to the ectoderm. These dorsal and ventral myogenic blastemae are believed to represent the primordial flexor and extensor muscle blocks (Milaire, 1965). The skeletal blastema is not well formed at this stage and exists as a vaguely defined central core of increased cell packing in the proximal part of the limb-bud. (This is not registered in Fig. 1 as central sampling areas at stage 22 were distal to this early pre-chondrogenic condensation.) Intercellular spaces are still readily distinguishable amongst the pre-chondrogenic cells and no extensive close apposition of cell surfaces exists. Within this region is found an extensive area of cell death -the opaque patch. This, like the ANZ seen along the anterior edge of the limb, is composed of moribund and dead cells together with macrophages and is particularly prominent due to the intense staining properties of these cells.

The muscle anlagen are still represented by the simple dorsal and ventral myogenic condensations which have now increased in volume at the proximal level (see Fig. 3). They are composed of very closely packed myoblastic cells and the intercellular spaces are virtually eliminated. Distally the condensations merge into areas of increased mesenchyme cell packing immediately below the ectoderm. The chondrogenic tissue is now better defined than at stage 22 and has assumed the Y shape of the early skeletal blastema. Proximally, there is a single central chondrogenic condensation, the prospective femur, which forms two arms distally. These are destined to form the paired long bones, the tibia and fibula. At this level the two arms are separated by the opaque-patch region of cell death. It has been suggested elsewhere that this has a morphogenetic role in the shaping of the skeletal blastema by dividing distally the simple axial pre-chondrogenic condensation and thus gives rise to the two arms of the Y-shaped blastema (Dawd & Hinchliffe, 1971; Hinchliffe & Thorogood, 1974). More distally the tibia and fibula are represented by areas of increased mesenchyme cell packing.

The elaboration of myogenic and chondrogenic patterns has continued (see Figs. 4, 5). Proximally the muscle condensations subdivide as the complex of antagonistic muscles and associated tendons in the upper limb emerges. The myogenic condensations extend distally to the prospective ankle region.

Amongst the massed myotubes, some multinucleate myotubes are present. These are particularly prominent when sectioned longitudinally. Within the cytoplasm of the myotubes, fibrillar material can be distinguished, and the presence of actomyosin in these cells (Thorogood, 1973) strongly suggests that myofibrillogenesis has commenced within the myogenic condensations. Ultrastructural evidence confirms this interpretation (Hilfer, Searls & Fonte, 1973).

The Y-shaped chondrogenic blastema within the limb-bud shaft is well defined and distally in the digital plate localized areas of increased mesenchyme cell packing presage the ankle elements and the digital condensations. Proximally the femur condensation is composed of differentiated chondrocytes and matrix. The tibia and fibula at this stage contain a gradation of cell types. Proximally the cells are beginning to accumulate a metachromatic matrix between them but distally the tibia and fibula comprise closely packed chondroblasts (Fig. 5) with a negligible amount of matrix. At the periphery of the skeletal condensations mesenchyme cells are densely packed, presumably demonstrating appositional growth of the skeletal blastema by recruitment of local mesenchyme cells (Ede & Agerbak, 1968).

It was mentioned earlier that there is a gradient of differentiation within the skeletal blastema, and therefore it is important to define precisely the position of areas of interest. For the purposes of this communication a description of the ultrastructural appearance of chondrogenic condensation is confined to the stage-26 and stage-28 tibia, where all the early chondrogenic cell phenotypes may be found. (A similar sequence of cellular events takes place in the formation of the other skeletal condensations in the limb-bud.) Proximally within the tibia there are early chondrocytes and a rudimentary matrix, and distally the cell condensation with its characteristic close association of cells exists. The later stages of chondrogenesis have been described comprehensively elsewhere (Goel, 1970; Searls et al. 1972), and it was felt unnecessary to repeat them here.

Undifferentiated mesenchyme from peripheral, marginal areas adjacent to the tibia at stage 26 (and in the limb at earlier stages) is a loosely constructed cellular network with large intercellular spaces (see Figs. 6A, B, 10A). The degree of local packing may vary considerably. The intercellular areas are quite featureless apart from the occasional bundle of fine fibrillar material thought to be small, immature collagen fibrils. Such fibres are sparsely distributed in undifferentiated mesenchyme and are only locally abundant in the mesenchyme adjacent to the basement lamina of the ectoderm. The mesenchyme cells are rather stellate in appearance and have a considerable number of filopodia which are often transected in section. Infrequently a cell may be observed which possesses a single cilium : the kinetosome is embedded some distance into the cytoplasm and the cilium itself may project well away from the cell surface. The cells are generally in contact where their filipodia are juxtaposed. At this stage the majority of intercellular contacts are of the tight junction or zonula occludens type (Farquhar & Palade, 1963); such junctions were usually ‘focal’ although the extended form was occasionally seen. The zonula adhaerens junction with a 100–200 Å gap does occur but is not seen as frequently. A desmosome-like junction, the macula adhaerens diminuta (after Hay, 1968), may be observed but is comparatively rare. The large nucleus contains a prominent nucleolus (or nucleoli). The endoplasmic reticulum is only sparsely distributed and has very few ribosomes attached. A distinct Golgi complex is present which is small but well-defined. The mitochondria are small -usually spherical or oval: the cristae are few in number and the mitochondrial matrix reasonably electron-lucent. There are large numbers of unattached ribosomes, usually in polyribosome clusters.

As previously mentioned, the first indication of a condensation is a localized increase in mesenchymal cell number yet without the formation of extensive areas of cellular contact or any overt indication of differentiation (see Fig. 10B). This is shown by the appearance of the stage-26 digital condensations, which are judged to be pre-chondrogenic condensations by their position rather than any histospecific morphology of the cells.

The next stage in the formation of a chondrogenic condensation is seen at the distal end of the stage-26 tibia and at the peripheral surface of the tibia, where there has been a considerable increase in the density of cell packing (see Figs. 7A, B, 11A). The cells have become intimately associated with the surfaces of adjacent cells close together, although not in direct contact, and the intercellular spaces considerably reduced. The cells have rounded up and possess fewer filopodia. No obvious ultrastructural cytoplasmic changes appear to have occurred, but rather, a change in the surface relationships of the cells. It is important to note that although the cells are very closely packed, each cell profile is well defined and there are no signs of any cell fusion (see Fig. 11B). This is an important distinguishing feature between the chondrogenic and the myogenic condensations; in the latter, fusion of the condensing cells-that is, the myoblasts -to form multinucleate myotubes has commenced (Hilfer et al. 1973).

The temporal sequence of events in chondrogenesis can be followed by examining the cells in a more proximal position along the axis of the stage-26 tibia, where the close association of the chondrogenic cells becomes progressively lost and the cells separated from each other. This is concomitant with the first definite appearance of cartilage matrix components in the intercellular areas. Abundant collagen fibrils with periodic banding appear in longitudinal and transverse section and are often closely associated with the surface of the cells. Whether this collagen is cartilage-specific type collagen or is the same as that seen in the undifferentiated mesenchyme remains to be established. In addition granular structures, often associated with clumps of amorphous material, appear sparsely scattered in the intercellular spaces: it has been suggested that these represent forms of the mucopolysaccharide component of the early matrix (Searls et al. 1972). Matrix production has commenced and the accumulating matrix is presumably responsible for pushing the chondroblasts apart (see Fig. 8A). In longitudinal section the orientation of these chondro-blastic cells is at right angles to the long axis of the condensation and the cells are approximately parallel to each other (see Fig. 9). In transverse section it is difficult to distinguish any clear pattern of orientation and the distinctive concentric whorling of chondrogenic cells described in the amphibian limb (Anikin, 1929) is not nearly as marked in the chick embryo.

More proximally the chondroblasts exhibit a more pronounced spatial isolation (see Fig. 8B), although some intercellular contact is maintained by an occasional zonula adhaerens junction between the cytoplasmic processes of adjacent cells. A greater concentration of fibrillar and granular matrix components is evident. Within the cells the endoplasmic reticulum is increased in complexity and roughness: large and swollen cisternae are present with granular contents. The Golgi body is more elaborate and vacuolated and there is an increase in the electron density of the mitochondrial matrix and in the number of cristae. The cell profile is no longer rounded but instead is indented in a crenate fashion.

At stage 28 early chondrocytes may be found in the proximal tibia and femur (see Fig. 11C). The cells have maintained the orientation described earlier and each cell has a distinct scalloped profile and is surrounded by a fibrillar and granular matrix. Large areas of parallel rough endoplasmic cisternae are present within which are granular contents -of an electron density approximating to that of the matrix. The endoplasmic reticulum is studded with large numbers of attached ribosomes. The mitochondria are more elongated, with an electron-dense matrix and large numbers of swollen cristae.

Beginning at stage 28 and seen frequently in the later stages are the early events of chondrocyte hypertrophy. Occasionally, cells can be seen that contain extremely large areas of swollen endoplasmic reticulum which are full of granular material. This appearance is interpreted as the phase of increased metabolism heralding chondrocyte hypertrophy. The development of the endoplasmic reticulum is such that most of the other cellular structures are almost obliterated.

One of the intentions of this investigation was to analyse the early events of chondrogenesis and to decide how justified were the claims of the early light microscopy reports (Fell, 1925; Saunders, 1948; Streeter, 1949). Fell employed the hind-limb bud (as used in the present study) and expressed developmental stage by size rather than by the Hamilton and Hamburger scheme now commonly used. It is necessary to equate the two in order to evaluate the present findings. Fell & Canti (1935) state that ‘the rudiment of the limb skeleton first appears when the leg bud is about 1 mm in length (i.e. Hamburger and Hamilton stage 22/3) and is represented by a diffuse condensation of mesoderm in the proximal part of the bud’. Further, they found that a limb-bud of 1·6 mm in length (i.e. stage 24/5) contains the beginning of a Y-shaped skeletal blastema and that at a length of 2·8 mm (i.e. stage 26/7) the first sign of a cartilage matrix exhibiting metachromasia appears. The timing of events in early chondrogenesis in the hind-limb which we have described seems fully compatible with the observations by Fell and others. The concept, which has arisen from light microscopy, of a mesenchymal condensation of increased cell packing presaging the development of a cartilaginous skeletal blastema seems borne out by the electron-microscopic demonstration of the changes in cell surface relationships at chondrogenic loci.

Although condensation has long been recognized by light microscopists there have been few precise ultrastructural descriptions to substantiate these claims. Goel’s study of chondrogenesis in the chick limb-bud (1970) concentrated on the later stages, particularly the intracellular ‘membranous systems’ and virtually ignored the mesenchyme cells. Olson & Low (1971) examined the development of the cervical regions of the vertebral column where such pre-cartilaginous condensations do not occur; the sclerotome cells move apart, assume a fibroblastic shape and then form a matrix. However, a more recent publication on myotomal chondrogenesis illustrates a pre-chondrogenic area of closely packed mesoderm cells (see fig. 16, Minor, 1973). In an electron-microscopic analysis of the skeletal blastema in the regenerating amphibian limb it is stated that ‘a very intimate and extensive association between groups of adjacent blastema cells first indicates prechondral condensation. The plasma membranes are closely aligned, one against the other, forming a compact mass of cells’ (Schmidt, 1968). This cellular morphology also exists in the emerging skeletal blastema of the embryonic amphibian (Axolotl) limb (Hinchliffe, 1975). In a study of the embryonic rat epiphysis, Godman & Porter (1960) published pictures of closely packed mesenchymal cells which they describe as ‘precartilage’ and subsequently demonstrated that these cells moved apart as matrix accumulation progresses. In an ultrastructural study of the digital plate in the embryonic human forelimb, Kelley (1970a, b) found that ‘digital regions consist of closely associated mesenchyme cells whereas interdigital zones contain fewer cells and greater intercellular spaces’. The electron micrographs in these cited publications show a marked similarity to those presented here and this evidence seems to establish conclusively that condensation is a real event in early chondrogenesis. However, there have been two recent ultrastructural analyses of limb-bud chondrogenesis which do not describe any such condensations of closely packed mesenchyme cells forming the initial skeletal blastemae (Searls et al. 1972; Gould et al. 1972). It is important to reconcile this contradiction with work by others and with the results presented here. It is claimed that the chondrogenic cells differentiate into chondrocytes and produce cartilage matrix without ever passing through the condensation phase. It is noteworthy that these two investigations were concerned with the wing-bud and not the leg. The early development of the forelimb-bud lags approximately 12 h behind that of the hind limb and consequently there is not a strict synchrony of developmental events (Hamburger & Hamilton, 1951). Direct comparison between events in the forelimb and hind limb without regard to this difference may be misleading. In our experience mesenchyme cell condensation is a transient phenomenon occurring before the cells become more spatially isolated from each other by progressive matrix accumulation. Since Searls et al. were examining wing-bud tissues at two stage intervals and given that the timing of condensation may well be different in the forelimb and hind limb, it seems possible that this transitory phase may well have occurred between observed stages and thus was not seen. Gould et al. reported that in the wing-bud stage-24 pre-cartilaginous cells ‘tended to remain spiky in form with many processes and had only occasional long contacts with adjacent cells’. They found an increase from 11 to 15 cells/1000 μm² in cell density in pre-cartilage, but attributed the ‘impression of a condensed region’ to increased pseudopodial processes and to matrix secretion. The present authors found no evidence that appreciable matrix secretion had begun at this stage in the hind limb and are inclined to the view that the ultrastructural characteristics described by Gould et al. may be due to a fixationshrinkage artifact. It was mentioned earlier (see footnote in the Materials and Methods) that the present authors found an absence of chondrogenic condensations if the Karnovsky method of fixation was employed for stage-24 and stage-26 hind limb-buds, that is, at a time when the double fixation technique (Trump & Bulger, 1966) showed well-defined chondrogenic condensations. Is artifactual cell shrinkage leading to the disappearance of the condensations, or conversely, is an artifactual cell expansion producing them? The intact cellular morphology which we observed is not compatible with the latter explanation. Expansion of cell volume due to a fixation artifact would lead to considerable disruption of ultrastructure and this was not found. The former explanation -that of a cell shrinkage destroying the close association of cells -is more likely, particularly as Karnovsky (1965) states that this fixative causes shrinkage of cells in monolayer cultures and of dispersed cells in suspension. The loosely constructed nature of embryonic mesenchyme has some physical similarities to those in vitro cell systems and may produce a similar shrinkage effect. Evidently the cellular morphology of the early chondrogenic condensation is susceptible to profound alteration by variation in fixation procedure. Any susceptibility of cells at this locus to such artifactual change may reflect the particular metabolic state of the chondroblasts at this point in time.

Both from the results presented in this communication and from the observations by others in different embryonic and regenerative systems, it seems that a condensation phase involving close cell association is a necessary prerequisite for chondrogenesis to occur. This conclusion is strengthened by reports of changes in cellular adhesiveness found during in vitro chondrogenesis. Holtfreter (1968) found that differentiating amphibian chondrocytes cease to wander in an amoeboid fashion, move closer together and establish maximal surface contact with each other. A more precise sequence of events has been established for chick chondrocytes in vitro. Abbot and Holtzer (1966) found that a physical interaction mediated through a cell crowding phenomenon was necessary to induce and maintain chondrogenic activity. The clonal cultures of chondrocytes become strongly adherent to each other and to the substrate and begin to synthesize chondroitin sulphate; absence of such an interaction led to DNA synthesis and cell multiplication. More recently, Holtzer and his co-workers have found that treatment with 5-bromodeoxyuridine and thymidine analogue suppressed chondrogenic activity. It inhibits radiosulphate uptake and the cells lose their adhesiveness for each other and the substrate and assume an amoeboid phenotype (Holtzer, Bischoff & Chacko, 1969).

What is the significance of the condensation phenomenon in relationship to the morphogenesis of chondrogenic pattern? Ede has proposed that chondrogenic cell arrangement in the condensation is produced by aggregative movements of cells onto a central cell or small group of cells (Ede & Agerbak, 1968; Ede, 1971). Thus a single cell may initiate chondrogenesis at a particular locus; surrounding cells are stimulated to converge on this cell and the resulting effect is a ‘huddling together of cells’. A similar centripetal aggregation of chondroblastic cells occurs during the in vitro development of amphibian cartilage nodules (Holtfreter, 1968). Ede has suggested an analogy with the cellular slime moulds; aggregation of myxamoebae prior to grex formation is triggered by a single myxamoeba secreting an increased level of chemotactic factor -acrasin. Holtfreter (1968) interprets the in vitro phenomenon as an induction. Once a chondrogenic centre has been formed it then assumes its own secondary ‘homoiogenetic’ inductive powers and grows in size by recruiting cells from the surrounding area. Whether or not a chemical message -be it morphogen or inducer -operates in limb-bud chondrogenesis is not yet established.

Regions of high-density cell packing as seen in mesodermal condensations could arise in several ways. Small-scale movement of cells to a particular locus could occur -that is, by an in vivo aggregation phenomenon. Alternatively a localized increase in mitotic rate would result in an increased number of cells per unit area, and thus a greater density of cell packing would be the consequence. It has not been established how the limb skeletal condensations are formed initially. Which of these factors is responsible and, if both are operating, what are their relative contributions?

In addition to the development of mesodermal organs and tissues, mesodermal cell condensations are frequently involved in the development of some composite organ systems in conjunction with either ectoderm or endoderm -for example, the avian feather primordium which comprises both ectodermal and mesodermal components. The formation of dermal papillae condensations in feather development is basically a system similar to pre-cartilaginous condensations, in that both localized aggregation and differential mitosis could be involved. Wessells (1965) has claimed that differential mitotic activity plays a major part in the establishment of these dermal condensations and that the condensations are formed by a brief phase of increased mitosis which is followed by an increase in the volume of cells. He proposed that cell movement is not a contributing factor as he saw no decrease in cell density between newly formed condensations. However, a more recent study offers circumstantial morphological evidence that cell movement does contribute to feather papillae condensations, although perhaps at the secondary level, by attracting further cells and thus reducing the cell density in the inter-condensation areas (Ede, Hinchliffe & Mees, 1971). This view is supported by Stuart, Garber & Moscona’s experimental evidence (1972) that arrest of mitosis by colchicine eliminates high mitotic activity as the primary mechanism of condensation formation, which they claim is the result of cell aggregation and migration.

Some skeletal blastemae have been analysed -for example, the scleral bone primordium in the chick embryonic eye which Hale (1956) showed to be formed by a localized wave of high mitotic activity, and the mandibular primordium which Jacobson and Fell (1941) concluded to be the result of both cell proliferation and cell movement. With reference to the formation of limb-bud skeletal blastemae, a causative role for differential mitosis has not been established. Autoradiographic experiments following the incorporation of labelled DNA precursors have not revealed any localized waves of accelerated mitotic rate at the loci at which condensations will subsequently appear (Janners & Searls, 1970; Thorogood, 1972). Both pre-myogenic and pre-chondrogenic areas exhibit a high mitotic rate which falls progressively with time. This drop is presumably a result of the involved cells differentiating and dropping out of the mitotic pool.

Wolpert and his colleagues confirmed the findings of Janners & Searls (1970) of a proximo-distal gradient in mitotic index within the limb beginning at stage 23 (Hornbruch & Wolpert, 1970) and they have investigated the relationship between cell density and mitotic index in the limb mesenchyme (Summerbell & Wolpert, 1972). Summerbell and Wolpert conclude that there is an inverse relationship: at stage 25 cell density is high and mitotic index low proximally, and the relationship reversed distally. They suggest that increasing density inhibits mitosis and use this supposition to construct a model for limb morphogenesis. However, it should be remembered that by stage 25 chondrogenic differentiation has begun proximally, as indicated by the high rate of chondroitin sulphate synthesis centrally (Searls, 1965; Hinchliffe & Thorogood, 1974). It has been established in many differentiating tissues that cell division rates drop (see, for example, Abbot & Holtzer, 1966), and in fact in the stage-24 limb mesenchyme Janners and Searls’ continuous labelling experiments (1970) showed that proliferative index of the pre-cartilage regions was only 25 %, i.e. three-quarters of the pre-cartilage cells have dropped out of the cell cycle. Thus increased cell density and lowered mitotic index along the proximo-distal axis of the limb may be more correctly considered as manifestations of chondrogenesis rather than the control mechanisms responsible for limb morphogenesis and outgrowth.

Thus, although we can confirm with ultrastructural evidence the existence of classical mesenchymal condensations composed of closely associated cells there is no unequivocal explanation for the formation of such condensations. In the absence of any firm evidence of differential mitosis being responsible we conclude that a cell aggregation mechanism is responsible. But major progress in understanding the process of formation of skeletal pattern, both normal and deviant, must await firm experimental identification of the mechanism which generates pre-cartilaginous condensations.

The authors wish to thank Mrs Roberta Whetter for typing the manuscript and Mr John Collins for photographic assistance. A part of this work is taken from a thesis submitted by P. V. Thorogood to the University of Wales for the degree of Ph.D. (1972).

With reference to the banded collagen which appears more abundantly in the intercellular areas of chondrogenic tissue at stages 26 and 28, this is presumably cartilage-specific collagen, since Linsenmayer, T. F., Toole, B. P., and Trelstad, R. L. (1973, Devi Biol. 35, 232–239) have now demonstrated a shift in collagen synthesis by the limb chondrogenic tissue at stages 25–26 to what is thought to be cartilage-specific collagen.