ABSTRACT
Suggestive evidence for the extragonadal origin of germ cells in birds was first presented by Swift (1914), who described primordial germ cells in the chick embryo at as early a stage as the primitive streak. According to Swift, primordial germ cells are originally located extra-embryonically in the anterior part of the blastoderm and occupy a crescent-shaped region (‘germinal crescent’) on the boundary between area opaca and area pellucida. Swift also found that primordial germ cells later enter into the blood vessels, circulate together with the blood throughout the whole blastoderm and finally penetrate into the genital ridges, where they become definitive germ cells. Swift’s views have been confirmed in numerous descriptive and experimental investigations. Among the latter, the publications of Willier (1937), Simon (1960) and Dubois (1964a,b, 1965a,b, 1966) merit special attention. Dubois finally proved that the genital ridges exert a strong chemotactic influence on the primordial germ cells.
Duck embryos have not as yet been so intensively investigated in this respect as those of chicks. The recently published reports by Fargeix (1966, 1967a, b, c) and Fargeix & Theilleux (1966) suggest that the history of primordial germ cells in the duck is similar to that in the chick.
Although the history of primordial germ cells (PCGs) in birds has been thoroughly investigated, the cause of their differentiation at a given time and in a given territory is still unknown. One of the possible ways of attacking this problem is to disturb the original organization of the blastoderm by transection, which results in the formation of twin embryos. Studies on the transection in ovo of bird blastoderms have been made for several years by Lutz and his co-workers (Lutz, 1949, 1953, 1955, 1964, 1965; Fargeix, 1964a,b, 1965, 1967a,b,c) and have shown that in the duck, as well as in the quail, goose and turkey, the embryo can form from any sufficiently large fragment of the blastoderm. The method used by these authors (Wolff & Lutz, 1947; Lutz, 1949) makes it possible to obtain twin embryos, which may survive as long as 10 days of incubation.
Germ cells in duck twin embryos, obtained as a result of transection of the blastoderm, have been investigated mainly in embryos at an advanced stage of development (Lutz, 1949; Lepori, 1961; Paia, 1963; Fargeix, 1964a,b). It was found that both twin embryos contained germ cells, which is quite natural in view of their common circulatory system. The original localization of these cells before migration had not been examined, however, and forms the chief object of this study. During its preparation for publication two papers by Fargeix (1967a,b) appeared, in which the author particularly describes the localization of primordial germ ceils in single embryos obtained after elimination of one half of the blastoderm.
MATERIALS AND METHODS
Duck eggs of the Pekin strain were used for the experiments. Unincubated blastoderms were transected by the method of Wolff & Lutz (1947) and Lutz (1949), either perpendicular or parallel to the long axis of the shell (Rogulska, 1968). The eggs were incubated for 48-98 h at 37·8°C. Embryos developing after operation were always smaller than the control embryos, retarded in development and often abnormal. The operated and control blastoderms were fixed in Bouin or Gendre fluid, embedded in paraffin, sectioned at 6μ and stained with Ehrlich’s haematoxylin and eosin. Some of the blastoderms were photographed before embedding in paraffin wax, having previously been stained in toto with Ehrlich’s haematoxylin and cleared in benzene for this purpose. A total of thirty-one normal and forty-two transected blastoderms were examined.
Primordial germ cells were identified by virtue of their dimensions, spherical shape and the presence of yolk granules or vacuoles. Consecutive sections containing primordial germ cells were drawn by means of a drawing apparatus and the position of cells marked on the drawing. The total number of sections into which the blastoderm was cut, as well as the position of PGCs within each section having been established, it was then possible to determine the approximate position of PGCs within the blastoderm by entering the position of each cell on a large photograph of the whole blastoderm. The data thus assembled were then converted into diagrams of the type shown in Text-fig. 1.
Representative blastoderms, showing localization of PCGs (cf. Plate I). Stippled areas indicate distribution of PGCs (the dots do not represent individual cells).
RESULTS
1 Primordial germ cells in normal duck blastoderms
The control material consisted of thirty-one blastoderms in stages varying from head process to thirty-five somites (Table 1).
Until active circulation begins, PGCs are scattered in the anterior part of the blastoderm, occupying the germinal crescent, as Fargeix (1966) has also recently reported. Even before circulation begins, however, some changes are observed in the localization of PGCs caused more by the change in the shape of the territory occupied by them than by active migration. In the pre-somite stages the PGCs are mainly located in the anterior part of the area pellucida, which is relatively wide in front of the embryo (Plate 1 and Text-fig. 1, A) and is not yet completely invaded by mesoderm. In stages with more than six somites the area pellucida in front of the embryo is far narrower, and PCGs are situated mainly laterally on both sides of the embryo’s head. Fargeix (1966) observed that the posterior ends of the germinal crescent gradually become wider and he related this with the growth and lateral spreading of the mesoderm, taking some of the PGCs with it. It is possible that both factors—change in shape of area pellucida and growth of the mesoderm—jointly cause a change in the shape of the germinal crescent.
Even in the early stages, before circulation starts, PGCs could sometimes be found in the posterior part of the blastoderm. They usually lie freely above the mesoderm (Plate 2, fig. L), often very far from the germinal crescent, and never occur in any great numbers. The fact that their occurrence is only sporadic suggests that they may be cells which accidentally moved from the germinal crescent towards the posterior part of the blastoderm. Fargeix (1966), however, considers that in some duck blastoderms a considerable proportion of the PGCs is situated in the posterior part of the blastoderm, and that the cells in question originate in this particular territory.
In the 9-to 10-somite stages almost all PGCs are already in the blood vessels.
They were not observed to penetrate actively into the vessels; it seems more likely that they are passively enclosed while the vessels form. After active circulation has started at the stage of about 16 somites, PGCs circulate in the blood throughout the whole blastoderm, and from the 27-somite stage onwards they begin to appear in the genital ridges. Single PGCs may ‘get lost’ during migration and reach non-typical sites, e.g. the head mesenchyme (Plate 2, fig. K). This phenomenon has been described by many investigators in chick embryos (see Meyer, 1964). By the 35-somite stage penetration of PGCs into the ridges is nearly completed and only single PGCs may still remain in the blood vessels. According to van Limborgh (1958, 1961) the distribution of PGCs between the genital ridges of the duck embryo is symmetrical up to the stage of 36 somites. At the 38-somite stage the primordium of the left gonad already contains 60% of the total number of PGCs. In the embryos which I examined this asymmetry must have begun somewhat earlier, since in all six 31-to 35-somite embryos the left ridge already contained slightly more PGCs than its counterpart (34/23, 55/14, 88/85, 105/54, 109/73, 179/149).
Considerable individual variations in the number of PGCs were observed among the blastoderms in all stages examined. The increase in the number of PGCs during the period under investigation is slight (Table 1). The decrease in the mean number of PGCs during the 16-to 24-somite stages is more apparent than real, and is possibly due to failure to observe all the PGCs distributed through the blood vessels of the whole blastoderm during this period. It would appear that the number of PGCs at the moment of their immigration into the genital ridges is determined to a greater extent by the number of cells originally formed than by mitotic divisions of PGCs during their migration through the blood vessels.
The characteristics of the PGCs change slightly during migration. During the period of their formation in the germinal crescent they contain a relatively large amount of yolk (Plate 2, fig I, J), which is gradually used up during migration. PGCs circulating in the blood vessels (Plate 2, fig. M) and especially those which have settled in the genital ridges contain very little yolk but possess large vacuoles in the cytoplasm instead.
2 Primordial germ cells in transected blastoderms
Forty-two blastoderms were classified into four groups according to the arrangement of the embryos (Rogulska, 1968). Examination was made of nineteen blastoderms with embryos lying parallel or almost parallel to each other, five blastoderms with embryos head to head, nine with embryos arranged head to tail (blastoderms with posterior duplication of the anterior embryo, such as shown in Plate 1, fig. G, were also included in this group), and nine blastoderms with embryos arranged at various angles.
The characteristic features of PGCs in transected blastoderms and their history are the same as in the control blastoderms, although the territory they originally occupy is not always similar in shape to the germinal crescent. Analysis of the distribution of PGCs in transected blastoderms often meets with difficulties. Twin embryos often differ as to size and degree of development. Not infrequently they are markedly retarded in development and their development does not always proceed normally. Disturbances in structure and function of the circulatory system have a decisive effect on the distribution of PGCs in the blastoderm, and absence of circulation, leading in the end to the death of the embryos, results in the PGCs remaining in their original position.
(a) Number of primordial germ cells
The number of PGCs in some of the transected blastoderms was very large as compared with control material. Considerable individual fluctuations also occurred (Table 2), but the mean total number of PGCs for all the transected blastoderms is 168·2, as compared with 8·12 for the control blastoderms. Different mean numbers of PGCs were found in different groups, with group I exhibiting the highest mean number. It is, however, difficult to evaluate the significance of these differences, since the sample was small and great differences occur between blastoderms. The general increase in mean number of PGCs appears, however, to be significant.
The number of PGCs in operated blastoderms does not depend on the general condition of the blastoderm. Blastoderms very retarded in development, with abnormal embryos and poor circulation, did not contain fewer PGCs than those which had developed more normally. Not infrequently the absolute age of such blastoderms (in hours of incubation) was very high in comparison with their stage of development as indicated by the number of somites. If PGCs had undergone intensive mitotic divisions, then their number should systematically increase with the increasing age of the embryo. As Table 3 shows, such an increase does not occur. In addition, examination of the mitotic index of PGCs in several normal and transected blastoderms did not point to significant differences between them.
Numbers of primordial germ cells in transected blastoderms according to the time of incubation

The number of PGCs therefore mainly depends on how many of them were produced initially. In transected blastoderms this number is doubled on the average, because the territory where they are formed is much larger or because two territories are present.
No connexion could be observed between the degree of development or the position of twins and the number of PGCs found around them at the time of primary localization.
(b) Primary localization of primordial germ cells
The basic difference between transected and control blastoderms consists in the different primary localization of PGCs. It is self-evident that this can only be observed in those stages in which the original localization of PGCs has not yet been affected by the activity of the circulatory system.
Primary localization of PGCs was observed in twelve blastoderms belonging to group I. If the twin embryos lie in a common area pellucida, the highest number of PGCs is located in front of the embryos, and a smaller number between them. If two areae pellucidae have formed, then relatively more PGCs occur between the embryos; with parallel or nearly parallel embryos the PGCs are distributed mainly along the line of transection of the blastoderm (Plate 1 and Text-fig. 1, B, D, E). This phenomenon does not occur with other arrangements of the embryos.
In four blastoderms in group II (embryos arranged head to head) the PGCs are encountered chiefly in the boundary region between area pellucida and area opaca, or in the area pellucida at the level of the anterior parts of the embryos (Plate 1 and Text-fig. 1, C). In group II the line of transection and the localization of the PGCs are not clearly associated.
In group III (embryos arranged head to tail) six blastoderms were found showing primary localization of PGCs. The line of transection of the blastoderm did not affect the position of the PGCs (Plate 1 and Text-fig. 1, F, G). No connexion was observed between the number or distribution of PGCs and the position of either embryo. An embryo formed from the posterior part of the blastoderm is usually more advanced, but is not always found to have a higher number of PGCs in its vicinity.
The mixed group IV contains embryos arranged at various angles, usually at right angles. The primary localization of the PGCs was observed in seven blastoderms. In four of them a distinct connexion could be observed between the line of transection and the position of the PGCs: they occur particularly in the ‘central’ part, along the line of transection (Plate 1 and Text-fig. 1, H).
(c) Comparison of behaviour and fate of primordial germ cells in transected and normal blastoderms
Although in general outline the fate of PGCs in transected blastoderms is the same as in controls, certain deviations from the normal pattern were observed. The first PGCs in blood vessels are encountered as early as the 3-somite stage, earlier than in control blastoderms. This suggests that the formation of somites was inhibited, while the formation of blood vessels was not. The penetration of PGCs into the blood vessels normally ends at the 6-to 10-somite stage; PGCs are found outside the blood vessels as late as the 10-somite stage in transected blastoderms. No connexion was observed in transected blastoderms between the width of the area pellucida in front of the embryo and the occurrence of PGCs in it; blastoderms were often encountered which had a large area pellucida in front of the embryos with only a small number of PCGs. In normal development the 16-somite stage marks the beginning of active circulation. From this time PGCs begin to move away from the germinal crescent and to circulate with the blood. In transected blastoderms circulation does not always begin at the 16-somite stage; this may take place at stages with several somites less or more. In addition, in some cases circulation was so poor that PGCs remained in the vicinity of the former germinal crescent.
It has been assumed that it is not until circulation has been established that PGCs can be found in intraembryonic vessels (Meyer, 1964). In transected blastoderms, however, PGCs were repeatedly found in the heart or dorsal aortae of embryos at such an early stage that it would be difficult to assert that they could have reached there in the normal way (Plate 1 and Text-fig. 1, H). This could probably have taken place passively, by the enclosing of some PGCs in the aortae or heart as they formed. From the 31-somite stage onwards normally almost all PGCs should have settled in the genital ridges. In transected blastoderms the invasion of the genital ridges by PGCs may take place at slightly later stages. In general, however, PGCs in transected blastoderms pass through the same phases of migration as in control blastoderms, although the consecutive phases may be reached at different stages of development on account of the retarded development of twin embryos and the asynchrony of developmental processes. The only two significant features by which transected blastoderms differ from normal ones appear to be the general increase in the number of PGCs and the effect of the line of transection, which often takes over the functions carried out in normal development by the germinal crescent only.
DISCUSSION
As shown by the present experiments and those of Fargeix (1967a,b), any half of the unincubated blastoderm can give rise not only to an embryo, but also to primordial germ cells. These results become more understandable if it is assumed that in the unincubated blastoderm PGCs have not yet been individualized, and that the embryo must reach a certain stage before their formation can begin. This suggestion is not in agreement with observations by Simon (1960), who states that in the unincubated chick blastoderm these cells are already present, and scattered over the whole blastoderm. New data on this subject were recently presented by Dubois (1967a,b) on the basis of in vitro experiments. According to his hypothesis the primordial germ cells originate in the lower posterior part of the unincubated chick blastoderm and are then transported together with the lower layer towards the germinal crescent by the pregastrulation movements. One of the pieces of evidence for this interesting hypothesis is the fact that after culturing anterior and posterior parts of unincubated blastoderms separately during 48 h the posterior part is found to contain on average more PGCs than the anterior one. However, when a similar experiment was performed in ovo, Fargeix (1967b) came to the conclusion that the posterior part contains on average fewer PGCs. It will be remembered that the duck blastoderm is less advanced at egg-laying than that of the chick; consequently, according to Dubois’ hypothesis more PGCs would be expected in the posterior part. As far as the present experiments are concerned no significant differences between anterior and posterior parts have been observed, although it must be stressed that the number of embryos showing head-to-head and head-to-tail alignment was rather small.
Primordial germ cells arise normally from that part of the boundary region between the area opaca and the area pellucida which is farthest away from the growth centre. The germinal crescent is the region where the completion of the lower layer and the appearance of the mesoderm take place last. In transected blastoderms, however, the primary localization of PGCs cannot always be explained solely by the formation of a germinal crescent in front of each embryo.
In blastoderms showing parallel or nearly parallel arrangement of the embryos an association can be observed between the position of the PGCs and the former line of transection of the blastoderm. The PGCs are located mainly along this line and can be encountered even in the vicinity of the posterior parts of the embryos (Plate 1 and Text-fig. 1, D). Their concentration along the former line of transection, between the embryos, cannot be explained only by an overlapping of the two germinal crescents, since PGCs are often situated far backwards and rarely occur at all on the ‘external’ sides. Thus the line of transection may apparently sometimes play a role similar to that of the germinal crescent in normal development. A similar phenomenon was observed by Fargeix (1967d) even when one of the halves of the blastoderm was eliminated after transection; the developing embryo exhibited a distinct asymmetry of the crescent caused by the more numerous occurrence of PGCs on the transected side. One of the possible explanations of this phenomenon is that disturbances caused by the cut in some way create favourable conditions for the formation of PGCs. It must be emphasized, however, that the phenomenon described occurs only with this particular alignment of the embryos. When they are aligned head to head or head to tail no connexion is observed between the primary localization of the PGCs and the line of transection. This was also noted by Fargeix (1967b) when embryos deriving from one half of a blastoderm were examined.
The numbers of PGCs in both control and transected blastoderms exhibit considerable individual variations, but the transected blastoderms contained on an average twice as many PGCs as the normal blastoderms. In my control material the increase in the number of PGCs before their settlement in the genital ridges was very slight. Fargeix & Theilleux (1966), on the other hand, when examining the number of PGCs in normal duck blastoderms in stages ranging from the primitive streak to fifteen somites, observed a distinct increase in the average number of PGCs, from twenty for the primitive streak stage to as many as 350-400 for stages with over ten somites. Fargeix & Theilleux consider that the formation of new PGCs in the germinal crescent lasts only up to the 4-somite stage, and that the further increase in their number is caused by mitotic divisions of pre-existing cells. They explain the relatively low mitotic index of PGCs by difficulties in identifying the dividing cells. Fargeix (1967b), on the basis of his experiments involving simple transection of the blastoderm or destruction of one of the halves, states that the embryos formed from one half of the blastoderm contain as many PGCs as normal ones. Therefore, after transection the whole blastoderm would have contained twice as many PGCs, which is in agreement with my results.
It seems to me that the number of primordial germ cells in the blastoderm, whether normal or transected, depends to a greater extent on the number of PGCs originally formed than on mitotic divisions of pre-existing PGCs. Transection itself may in certain cases stimulate formation of PGCs and thus evoke an increase in their number from the very beginning of development.
SUMMARY
In early stages of development of the duck primordial germ cells (PGCs) initially occupy a crescent-shaped area, the germinal crescent. The characteristic features of PGCs and their fate from the time of formation up to the time of settlement in the genital ridges are the same as in the chick embryo. The number of PGCs exhibits considerable individual variations and does not markedly increase until after the PGCs settle in the ridges.
The primary localization of PGCs in transected blastoderms may be different from that in normal blastoderms and is affected by the arrangement of the twin embryos. With parallel or nearly parallel twins the majority of PGCs are located between the embryos, in the area through which the line of transection ran. This connexion between the position of the PGCs and the line of transection is not found when the embryos are arranged head to head or head to tail.
The number of PGCs in transected blastoderms is on an average twice as high as the number of PGCs in normal blastoderms.
Each half of an unincubated blastoderm is equally capable of forming an embryo and of forming PGCs.
Neither the characteristic features of PGCs in transected blastoderms nor the course taken by their migration deviate from those seen in normal development.
RÉSUMÉ
Les cellules germinales primordiales chez les blastodermes normaux ou fissurés du Canard
Pendant les stades précoces du développement du Canard, les cellules germinales primordiales (CGP) occupent initialement une région en forme de croissant, le croissant germinal. Les caractéristiques des CGP et leur destinée depuis leur formation jusqu’à ce qu’elles s’établissent dans les crêtes génitales sont les mêmes que chez l’embryon de Poulet. Le nombre de CGP par individu est très variable, et ne montre pas d’augmentation importante avant l’établissement des CGP dans les crêtes.
La localisation primaire des CGP dans les blastodermes fissurés peut être différente de celle chez les blastodermes normaux selon la disposition relative des embryons jumeaux. Les embryons placés parallèles ou presque parallèles, la plupart des CGP se trouve entre les embryons dans la région où se fit la ligne de fissuration. Cette relation entre les CGP et la ligne de ne se reproduit pas lorsque les embryons sont placés tête à tête ou tête à queue.
Le nombre de CGP dans les blastodermes fissurés est en moyenne deux fois les nombre de CGP chez le blastodermes normaux.
Chaque moitié d’un blastoderme non-jncubé est capable également de la formation d’un embryon et de la formation des CGP.
Ni les caractéristiques des CGP dans les blastodermes fissurés ni le cours de leur migration ne diffère de ce que l’on voit pendant le développement normal.
Acknowlegments
I wish to express my sincere thanks to Dr Andrzej K. Tarkowski for his advice and encouragement throughout every stage of this work. I am also grateful to Professor P. D. Nieuwkoop and Dr J. Faber (Hubrecht Laboratory, Utrecht) for helpful comments and assistance in the preparation of the manuscript.
REFERENCES
EXPLANATION OF PLATES
PLATE 1
Fig. A. Control embryo. One somite, x 12
Fig. B. No. 7 (group I). Eight and ten somites, x 12
Fig. C. No. 21 (group II). Three and eight somites, x 12
Fig. D. No. 11 (group I). Thirteen and fourteen somites, x 2·5
Fig. E. No. 15 (group I). Fourteen and eighteen somites, x 2·5
Fig. F. No. 26 (group III). Three and six somites, x 2·5
Fig. G. No. 33 (group III). Twenty, twenty-two, twenty-six somites, x 1·5
Fig. H. No. 40 (group IV). Eight and twelve somites, x 2·5
PLATE 2
Fig. I. Primordial germ cell in entoderm of area pellucida of normal blastoderm (head process stage). x!200.
Fig. J. Primordial germ cells above entoderm of area pellucida of normal blastoderm (8-somite stage), x 800.
Fig. K. Primordial germ cells (arrows) in head mesenchyme of normal embryo (20-somite stage). x400.
Fig. L. Primordial germ cell above lateral mesoderm of normal embryo (7-somite stage). x400.
Fig. M. Primordial germ cell in blood vessel. A large vacuole can be seen. Transected blastoderm no. 38 (group IV), six and eleven somites, x 1200.
Fig. N. Primordial germ cell in karyokinesis (arrow). Transected blastoderm no. 38 (group IV), six and eleven somites, x 1200.
Fig. O. Group of primordial germ cells above mesoderm (arrow). Transected blastoderm no. 38 (group IV), six and eleven somites, x 300.
Fig. P. Numerous primordial germ cells in blood vessels. Transected blastoderm no. 3 (group I), degenerate somites, x 400.