Paper - The origin of the vertebrate limb (1912)

From Embryology
Embryology - 22 Oct 2020    Facebook link Pinterest link Twitter link  Expand to Translate  
Google Translate - select your language from the list shown below (this will open a new external page)

العربية | català | 中文 | 中國傳統的 | français | Deutsche | עִברִית | हिंदी | bahasa Indonesia | italiano | 日本語 | 한국어 | မြန်မာ | Pilipino | Polskie | português | ਪੰਜਾਬੀ ਦੇ | Română | русский | Español | Swahili | Svensk | ไทย | Türkçe | اردو | ייִדיש | Tiếng Việt    These external translations are automated and may not be accurate. (More? About Translations)

Geddes AC. The origin of the vertebrate limb. (1912) J Anat. Physiol. 46(4): 350-383. PMID 17232933

Online Editor  
Mark Hill.jpg
This historic 1912 paper by Geddes describes development of the limb.

Modern Notes: limb

Musculoskeletal Links: Introduction | mesoderm | somitogenesis | limb | cartilage | bone | bone timeline | bone marrow | shoulder | pelvis | axial skeleton | skull | joint | skeletal muscle | muscle timeline | tendon | diaphragm | Lecture - Musculoskeletal | Lecture Movie | musculoskeletal abnormalities | limb abnormalities | developmental hip dysplasia | cartilage histology | bone histology | Skeletal Muscle Histology | Category:Musculoskeletal
Historic Embryology - Musculoskeletal  
1853 Bone | 1885 Sphenoid | 1902 - Pubo-femoral Region | Spinal Column and Back | Body Segmentation | Cranium | Body Wall, Ribs, and Sternum | Limbs | 1901 - Limbs | 1902 - Arm Development | 1906 Human Embryo Ossification | 1906 Lower limb Nerves and Muscle | 1907 - Muscular System | Skeleton and Limbs | 1908 Vertebra | 1908 Cervical Vertebra | 1909 Mandible | 1910 - Skeleton and Connective Tissues | Muscular System | Coelom and Diaphragm | 1913 Clavicle | 1920 Clavicle | 1921 - External body form | Connective tissues and skeletal | Muscular | Diaphragm | 1929 Rat Somite | 1932 Pelvis | 1940 Synovial Joints | 1943 Human Embryonic, Fetal and Circumnatal Skeleton | 1947 Joints | 1949 Cartilage and Bone | 1957 Chondrification Hands and Feet | 1968 Knee
Historic Disclaimer - information about historic embryology pages 
Mark Hill.jpg
Pages where the terms "Historic" (textbooks, papers, people, recommendations) appear on this site, and sections within pages where this disclaimer appears, indicate that the content and scientific understanding are specific to the time of publication. This means that while some scientific descriptions are still accurate, the terminology and interpretation of the developmental mechanisms reflect the understanding at the time of original publication and those of the preceding periods, these terms, interpretations and recommendations may not reflect our current scientific understanding.     (More? Embryology History | Historic Embryology Papers)

The Origin of the Vertebrate Limb

By A. C. Geddes,

Professor of Anatomy, Royal College of Surgeons in Ireland.

Technical Note

As this paper is of the nature of a preliminary communication, and of necessity has had to be kept within certain limits of length, only a selection from, rather than a full statement of, the facts upon which many of the deductions are based is given in the text.

The laboratory work undertaken has included :—

A. The examination of the following sets, or parts of the following sets, of serial sections of embryos from the R.C.S.I. Anatomical Department Collection :—

Thickness of Probable; Series. | Class. | Nos. | Sections in Stain. Age, Remarks. Micro, mm. Days.

G. Human | 209 15 Hem. and Kosin 70 ” » | 2100 15 » Bis. Br. 60 ” ” 2100 15 ” a9 60 ” > 175 15 ” Eosin 20

” 23 10 ;. . 23 ” ” 28 10 ” ” 28 ” ” 35 10 ” ” 27 ” ” 42 15 ” ” . 35 » , 63a 15 . . 63 ” ” 63b 15 Van Gieson 63 ” ” 64a 15 Hem. and Eosin 64 ” » | 648 15 i. 64 |

. Pig 21 10 ” ” .. | 12mm. ” Chick | 12 15 : ” 6 ”» Frog 10 15 ” ” vee Freeswimming. ” Axolotl 1 15 ” ”» vee Unhatched. ” ” 2 15 ” ” wee Newly hatched. ” » 18 15 » ” see Free swimming. %9 ” 19 15 ” Bis, Br. ees Free swimming (limbs appearing).

” » 279 15 ” ” wee Young larva.

” 280a 15 » ” wee Young larva. 9 ” 2806 15 ” ” see Young larva. R. Trout 9 5 » Iron, Alum 9 ” ” 10 5 : 9 99 ce) 10 ” ” 11 5 Acid F., Gen. V. and Orange G. 11 ” ” 18 10 Hem. and Orange G. 18 ” ” 19 10 ” ” 19 ” ” 20 10 ” ” 20 The Origin of the Vertebrate Limb i 351 “me | Thickness of Probable Series. | Class. | Nos. | Sections in Stain. Age, Remarks, Micro, mm. Days.

R. Trout | 22 10 Osmic acid 22 ” ” 24 10 Gen. V. and Orange G. 24 ” ” 25 10 Osmic acid 25 ” . 28 10 Hem. and Osmic acid 28 ” ” 32 10 Borax Car, and Lyons Blue 32 ” ” 33 10 Hem. and Orange G. 83 » » 34a 10 Borax Car., Iodine Gr. 34 ” ” 346 10 ” ” : 34 ” ” 35 10 ” Lyons Blue 35 ” ” 36 10 Hem. and Orange G. 36 ” ” 38a 10 Borax Car., Lyons Blue 38 ” » 38d 10 ” 38 ” » 38c 10 » Lyons Blue 38 » | 39 10 im ' 39 ” ” 40a 10 ” ” 40 ” ” 400 10 ” ” 40 ”» ’ 40c 10 ” 40 yy ” 44 10 Van Gieson 44 ” » 48a 10 Iron, Alum, and Cochineal 48 ” ” 486 10 ” ” 48 49 10 Borax Car. 49

The Series G. and R. Human, comprise 9831 sections; G. Pig, 856; G. Chick, 522; G. Axolotl, 6361; k. Trout, 11,094 : more than 27,000 sections in all.

B. The preparation of reconstruction models of :—

1. 2. 3. Developing shoulder girdle of Larva, Series G. Axolotl, No. 280a, (see fig. 15).

4, 5 6. 7. The hinder end of Embryo, Series R. Human, No. 23 (see figs. 5 and 6). Anterior half of Embryo, Series G. Axolotl, No. 18 (see fig. 14).

Developing pelvic girdle of Larva, Series G. Axolotl, No. 2806 (see fig. 11).

and 19).1 C. The study of the development of the external body form of the frog, penguin, duck, and chick (complete series). D. The examination of the skeletons of numerous adult vertebrates.

Whole Embryo, Series R. Trout, No. 22 (see fig. 16). Whole Embryo, Series R. Trout, No. 28 (see fig. 17). . Pectoral region of Embryo, Series R. Trout, No. 36 (see figs. 18

1 Models 1-7 and a selection of the microscopic sections were demonstrated to the Anatomical Society of Great Britain and Ireland at Aberdeen, June 1911. The models are preserved at the Royal College of Surgeons in Ireland, where the sets of serial sections

also are.

E. The observation of a large number of dissections of human subjects made in the Anatomical Department.

In addition there were available in the collection of the Anatomical Department, R.C.S.I., the models of developing shoulder girdle and fore limb prepared by Dr Rutherford and demonstrated by him to the Anatomical Society at London, December 1911.

Part I. — Introduction

In 1774, Vieq d’Azyr (1) published the first description of the correspondences existing between the parts of the limbs. Since then many anatomists have tried to establish their genesis and homogenesis. Thus to Oken (2), 1843, the arms and legs were bundles of liberated ribs; to Maclise (8), 1832, they were modified ribs to the elbow and the knee, but more distally corresponded with the interspinous bones and the rays of the azygos fins of fishes; to Owen (4), 1848, the free limbs were divergent appendages attached to the ribs and corresponded with the uncinate processes of the ribs of birds and the branchiostegal rays of fishes, whereas the girdles were compounded of pleurapophyses and hemapophyses; to Goodsir (5), 1857, the limbs were actinapophyses and corresponded with the epipleural spines of fishes; to Humphry (6), 1871, the pelvic bones of the ventral fins and the carpals of the pectoral fins were modified interspinous bones. These hypotheses are now only of historical interest.

Gegenbaur (7), 1872-1898, has suggested that the shoulder and pelvic girdles are modified branchial arches and that the free limbs are modified rays. As such a hypothesis practically demands that the archipterygium be a fin of the Ceratodus type and that the body wall be homogenetic with the pharyngeal, and in addition fails even to suggest a reason why there should be two pairs of limbs and never more than two pairs, it is improbable that it will ever be universally accepted as a complete theory of the origin of the vertebrate limb.

The majority of present-day anatomists appears to support the lateral fin hypothesis associated with the names of Mivart (8), Balfour (9), and Thacher (10). Briefly this hypothesis is :—

Primitive vertebrates at one time possessed a continuous median fin, and in addition a pair of continuous lateral fin folds. Traces of these, beginning with a proliferation of the mesoderm, can still be recognised in young elasmobranch embryos and to a less extent in other fish and batrachian embryos. These mesodermic proliferations, though never continuous, are present on the intermediate mytomes. They extend backward along the sides of the body from just behind the head, gradually converging towards the anal region, where they become continuous with the ventral part of the median fin fold and in this respect resemble the persistent lateral or metapleural folds of amphioxus. As is usually the case with the median fin, certain parts of the lateral folds have undergone reduction, only the anterior and posterior portions remaining to form respectively the pectoral and pelvic fins. These must, therefore, be looked upon as the localised remains of a continuous lateral fin fold on either side of the body (11).

Such is the hypothesis. Two logical objections stand in the way of its acceptance as a complete explanation of the origin of the paired limbs.

1. If there were two complete lateral fins, why have they completely disappeared in the ontogeny of all known fishes and in all known larval forms (12), whereas complete median fins are not at all uncommonly present in larval and even in adult forms? It is not an adequate explanation that the interrupted form of fin is the more advantageous. The occurrence of continuous median fins in the lower and of interrupted median fins in the higher fishes, can only be interpreted as showing that the interrupted median fin is a more useful structure than the continuous, and yet the continuous fin, the fin of the less advantageous pattern, is of common occurrence. Further, in torpedo, in which the pectoral fin encroaches on the region of the pelvic, the growth along the line of the hypothesised lateral fin takes place from before backwards, 7.e. from the anlage of the pectoral fin, during the third stage of development (12). Such developmental facts certainly do not suggest that material (cell groups) which can form limbs or part of limbs is available along the whole lateral line.

2. If the limbs are developed from a continuous fin, why are there two pairs of limbs and never more than two pairs, even though one of the pairs be partly modified to form copulatory organs, eg claspers in male elasmobranchs? So far as the hypothesis goes there might be several successive pairs of limbs, just as there often are several successive median fins. It is not an adequate explanation that four limbs are more useful than six or eight. This may be true in the case of most land animals, but when the successful occurrence of many pairs of limbs in elongated invertebrates is considered, it is difficult to believe that it was beyond the powers of the tissues to evolve some non-lethal structural arrangement which would have permitted the use of three or more pairs of limbs by some snake-like vertebrate, if the evolution of more than two pairs of limbs had been possible to the vertebrate type of organisation. Further, in fishes, if the development of more than two pairs of libs were possible, it would, one may say, have occurred. Even were it mechanically unsound for a fish to have more than two pairs of limbs, it can hardly be imagined that the existence of two pairs of limbs and a pair of claspers developed from a third pair would have had a negative selection value. Nor is it a sound argument to urge, that because a structure is temporarily useless it will not occur, for many animals have progressively developed structures not only not useful to them but positively harmful, and yet have persisted for long periods of time (13).

Mivart (14) pointed out this logical objection to the hypothesis that he himself helped to formulate.

Further, in so far as the lateral fin hypothesis presupposes an origin of the lateral fin from within, by a primary outgrowth from the mesodermic somites, it is almost certainly incorrect. This is dealt with later. In the meantime, it is sufficient to note that strong evidence against the prebability of such an origin for the limbs is provided by the paired extremities of man. These are neither unalterable nor inevitably connected with (innervated from) any particular group of body segments. In short, the limbs vary considerably in their relative metameric positions and in their nerve supply (15), (16), (17), (18). Itis hardly conceivable that this could be the case if the framework of the limbs took origin directly from the primitive segmental framework, unless each unit of the segmented axial framework possessed a pair of limb-framework units just as each possesses a pair of costal units. In other words, backward and forward homeeosis is readily understandable where there is a complete chain of serially homologous structures, but is practically inconceivable where there is not a complete chain. There is no such complete chain in the case under consideration, and yet it is hypothesised that an axial segment can in one person initiate an outgrowth to form a limb, and in another individual not only not initiate such an outgrowth but invariably fail to present any indication of possessing a rudiment capable of initiating such an outgrowth.

These objections render the lateral fin hypothesis impossible of acceptance as a complete theory of the origin of the limbs.

The hypothesis which is now advanced does not in reality oppose the lateral fin hypothesis. Rather it expands it in one direction and refines it in another. It appears, too, to link together it and the modified gill arch hypothesis in a manner not a little surprising. The foundations of the hypothesis and the application of certain deductions which flow from it are discussed in Part IT.

Part II. — The Foundations of a Cell-Accumulation Hypothesis of the Origin of the Limbs

1. One of the fundamental conceptions of much of the anatomical teaching of to-day is that the vertebrate body is organised on the same general plan as the body of a segmented Invertebrate, and is essentially an assemblage of similar parts repeated. along a longitudinal axis. This thought was expressed by Cunningham (19) as follows: “In the construction of vertebrates series of similar parts are repeated along a longitudinal axis one after the other. Thus the series of vertebra which build up the backbone, the series of ribs which gird round either side of the chest, the series of intercostal muscles which fill up the intervals between the ribs, the series of nerves which arise from the brain and spinal cord are all examples of this. An animal exhibiting such a condition of parts is said to present the segmental type of organisation, and in the early stages of development this segmentation is much more strongly marked and is to be seen in parts which subsequently lose all traces of such a subdivision.”

Fig. 1. — Diagrain to illustrate the development of Menobranchus lateralis. For description see text, paras, 8-8. The diagram is founded upon a figure of Keibel’s. (Ref. 21.)

2. Is it true that at any stage of development the body of a vertebrate consists of a series of essentially similar parts repeated along a longitudinal axis, and of nothing else ? (20).

3. It is only necessary to study the development of one of the anamniota, eg. Menobranchus lateralis (Necturus maculatus) (21), to recognise (fig. 1) how late a phenomenon is the process of segmentation of the body wall. This vertebrate is developed, as are all the anamniota, as a segmented rod upon the surface of an unsegmented sphere which is elongated and flattened out to form the belly wall. Into this, when development is far advanced and the general body form determined, outgrowing processes of the segments pass to establish a secondary segmentation of the primarily unsegmented belly wall.

4, This subdivision of the forerunners of the definitive body into dorsal segmented and ventral unsegmented groups of structures is demonstrable in the embryos of the higher mammals, even in those of man (figs. 2 and 3). Fig. 2 is a diagrammatised sketch of the embryo Klb from the Normentafel of Keibel and Elze (fig. 3d) (22). Mesially is the medullary region with five or six pairs of primitive somites at its sides; lateral to these (obliquely shaded) are the unsegmented areas from which the somatopleure (parietal layer of the lateral unsegmented mesoderm) derives. From the deep aspect of the same areas the splanchnopleure (visceral layer of the lateral unsegmented mesoderm) is split off at the time of the formation of the embryonic ceelom. From this the permanently unsegmented connective and muscular tissues of the abdominal viscera arise. Fig. 3 is a diagrammatised sketch of the His embryo Lr. (No. 6, His’ Normentafel) (23), and shows the unsegmented area of the belly wall (membrana reuniens) obliquely shaded.

Fig. 2. — Diagram of the human embryo Kb. from the Normentafel of Keibel and Elze. To show the relations of the segmented to the unsegmented parts of the developing embryo. For description see text.

In short, the vertebrate body, at its commencement, is not completely segmented. Primarily, its dorsal part is segmented, its ventral part unsegmented. Secondarily, the dorsal segmented part through its possession of the central nervous system comes to dominate the unsegmented part and indeed succeeds in moulding it, partly at least, to its own plan of - organisation.

This is an altogether different conception from that which postulates that the vertebrate body is essentially an assemblage of similar parts repeated along a longitudinal axis.

5. It should here be noted that this recognition of two primarily distinct parts of the embryo, a dorsal segmented and a ventral unsegmented, does not involve any special belief as to the origin of the mesoblast. For our present purpose it is sufficient that the mesoblast exists in the axial region, as repeated masses round repeated diverticula of the coelom; in the ventral region, as a continuous sheet. The mesoblast may all have arisen from one source; that, for the moment, is immaterial. What is material is, that, primarily, the mesoblast of the body wall is not segmented, though, secondarily, it is invaded by the mesoblastic segments which grow from behind towards the mid-ventral line.

Fig. 3. — Diagram of the His embryo Lr. To show the later distribution of the areas seen in fig, 2.

6. The most reliable evidence available as to the ultimate fate of the primary segments within the body of the adult, is provided by the arrange-. ment of the segmental nerves. In man, where this arrangement is best known, this evidence shows that though all the primitive segments (except possibly certain of the cephalic) grow laterally and ventrally, two groups of segments fail to pass beyond the lateral line, with the result that there are “faults” in the ventral segmentation. The actual number and identity of the segments involved in these faults is liable to somé variation, but as a general rule the anterior or pectoral fault involves all the segments between the fourth cervical and second thoracic (5th-9th) segments, the posterior or pelvic fault all those between the first lumbar and second sacral (21st-26th) segments.

The recognition of these “faults” is founded upon the distribution of the dorsal and ventral branches of the spinal nerves. Thus all the dorsal branches of the spinal nerves from the first cervical to the coccygeal are distributed either to skin or to muscle, covering or belonging to the axial segmented region of the body. The majority supply both skin and muscle. Of the few that usually do not possess cutaneous branches, the 6th, 7th and 8th cervical and the 4th and 5th lumbar, all at times send branches to the skin (5th lumbar through its loop with Ist sacral). Ventrally, in contradistinction to this there are two great gaps in the series of nerves supplying the skin. Thus the area supplied by the 4th cervical nerve meets, along a line which crosses the upper part of the chest between the roots of the upper limbs, the areas supplied by the Ist and 2nd (2nd and 3rd) thoracic nerves. Similarly the area supplied by the Ist lumbar nerve meets, along a line which crosses the lower part of the trunk (external genitals) between the roots of the lower limbs, the areas supplied by the 2nd and 3rd (8rd and 4th) sacral nerves. The nerves which fail to supply cutaneous areas on the ventral aspect of the trunk fail to supply any muscles belonging to the ventral portion of the body wall. Exceptionally, the marginal (cephalad or caudad) nerves of each group which fails to supply a ventral cutaneous area may give twigs to true ventral segmental musculature. (General references for section, 17, 24, 25, 26, 27.) :

From their position it is easy to assume that these “faults” are connected with the outgrowth of the limbs, but it in no way follows that the limbs are therefore to be regarded as processes of ventrally defaulting segments (29).

7. The primary unsegmented portion of the developing vertebrate shows two secondary extensions—an anterior and a posterior. These are most easily recognised in the development of such a member of the class anamniota as is shown in fig. 1,d. The obliquely shaped area is the primary unsegmented portion, the small stippled areas continuous with it anteriorly and posteriorly are the secondary or derived unsegmented areas. They form the ventral parts of the head and tail projections. (Fig. 1, stages a, b, and c, show the derivation of the anterior unsegmented extension from the surface of the “yolk sac,’? to form the ventral part of the cephalic projection, and stages b and c the derivation of the posterior unsegmented part from the same source to form the small unsegmented ventral part of the caudal projection. The tail soon frees itself from this unsegmented part and posteriorly is segmented from dorsum to venteral

1 The pectoralis major, though topographically ventral, may draw its supply from all the nerves from the 5th cervical to 1st thoracic inclusive. It belongs to the skin sheet (26).

2 The epidermal covering of the “yolk sac” of the anamniota corresponds with the amnion of the amniota.

8. At the stage shown in fig. 1, stage d, the body of the developing animal is composed of five distinct cell territories :—

1. The right axial region (right dorsal unsegmented area).

2. The left axial region (left dorsal segmented area).

3. The central visceral region (primary ventral unsegmented area).

4. The anterior derived visceral region (anterior secondary unsegmented area).

5. The posterior derived visceral region (posterior secondary unsegmented area).

9. In early human embryos comparable areas can be recognised ; see fig. 2. In this the anlagen of the anterior and posterior derived visceral regions are indicated by stippling. The anterior derived area is not yet complete, a considerable part of its ultimate extent being still included in the central visceral region. Similarly, in fig. 3 the stippled areas indicate the extent of the anterior and posterior derived visceral regions: The anterior derived visceral region is from its commencement more important than the posterior. It undergoes a sort of segmentation to form the branchial arches, but this is altogether independent of the axial metamerism.

So far it is legitimate to say that in the development of man and of Menobranchus lateralis five distinct areas or cell territories are recognisable. These territories can equally be recognised in the development of the selachians, ganoids, teleosts, ceratodus, batrachians, reptiles, birds and mammals (30).

10. It has already been pointed out that the unsegmented visceral regions are secondarily invaded by the lateral extensions of the segments of the axial regions. This invasion is most complete in the central visceral region, in the anterior part of which the invasion may be said to be fully successful (formation of thoracic segments).

11. The invasion of the anterior derived visceral region is comparatively slight, being limited to neurogenic and myogenic elements which abandon their own metamerism and adapt themselves to the segmentation (branchial arches) of the region. The posterior derived visceral region is extensively invaded and assumes a segmented form. (Distribution of lower sacral and coccygeal nerves (27).)

It is vmportant to note that the lines of the “faults” in the secondary ventral (derived) segmentation coincide in position with the lines of junction between the central and anterior and the central and posterior visceral regions.

12. Itis onlynecessary to watch the cleavages which mark the early stages of development of the frog’s egg to be convinced that, mechanically at least, a cell acts upon and is acted upon by the cells in its immediate neighbourhood. It is no less true that cell aggregates in contact, function as units in a similar interaction. The results of such mass action and reaction are visible in the mid-dorsal line where the right and left axial regions are in contact. There, where each half vertebral arch meets its fellow of the opposite side, fusion normally takes place but growth does not stop. Each half continues to grow, but, prevented by the opposition of its fellow from crossing the median plane, grows in a new direction, almost at right angles to its former course, backwards in contact with its fellow to form the spinous process. That this is the true method of formation of the spine, that it is essentially a bilateral and not a median structure, is evidenced by the known facts of development and by the not infrequent occurrence of vertebra in which the two half arches have failed to fuse posteriorly. In such cases each half arch carries a demi-spine. See fig. 4.

Fig. 4. — First thoracic vertebra (human). To show complete failure of union between the right and left halves of the vertebral arch. Each half carries a demi-spine.

A precisely comparable heaping up of cells takes place during development along the line of. contact between the cartilaginous vertebral arch and cartilaginous centrum of every vertebra. This cell-accumulation forms a collar round the margins of the area of contact between the parts, persists for a time and then disappears. Such a phenomenon suggests that the two parts of the vertebra function for a time as independent structures, and then when the vertebra finds itself, as it were, and the parts lose their identity in the greater whole, opposing cell proliferation is abandoned and the collar, no longer reinforced by new cells, ceases to be evident as the neighbouring parts increase in size. These “collars” are well shown in the models prepared by Dr Rutherford (s.v. Technical Note).

Even more striking is the fact that there is not infrequently a median vertical projection on the posterior surface of a developing cartilaginous centrum, making evidently a line of frontier between two half centra.

Such piling up of cells along the lines of frontier between cell regions is not confined to the mesoderm but also affects the epiderm. This is well seen in figs. 16 and 22, where the epiderm along the mid-dorsal line is seen to have grown backwards to form a continuous projecting ridge. Peter (31) was the first to suggest a mechanical cause for the formation of such an epidermal ridge, his explanation of the primarily pure epidermal structure of these surface-ridges being that where there are equal rates of cell division a surface layer must increase more rapidly in area than an underlying aggregate can in the surface areas of its mass. The epiderm, therefore, becomes disproportionately large. The epiderm covering of the neighbouring area is equally disproportionately large. If now invasion be impossible, a process of cell-accumulation must occur along the line of frontier. Such an epidermal crest is afterwards invaded and expanded by the mesoderm, which also heaps itself up along the line of frontier (see figs. 7, 8, 9, and 24). Elevations of the epiderm of comparable origin are well known to occur in the human fcetus along the lines of junction between the secondary cell areas, for example, on the lips at the junction of the oral and facial areas, or at the openings of the nostrils at the junction of the nasal and facial areas (32).

Cell-accumulations such as these take place along the lines of frontier between mutually resistant areas. A somewhat different phenomenon occurs where one of two neighbouring cell aggregates has established sufficient dominance to invade the territory of its neighbour. Nowhere is this more easily observed than along the lines of frontier between the axial regions and the central visceral region in the human embryo. There the axial region is the dominant invader, and as it advances it rolls back the undifferentiated but still proliferating mesenchyme towards the mid-ventral line. This undifferentiated proliferating mass forms the lateral projection known as the Wolffian ridge (see fig. 3, marked by crosses). Throughout the whole region of the developing thoracic and abdominal walls, the undifferentiated mesenchyme of the ridge is forced on by the advancing margins of the segments towards the mid-ventral line, ultimately to form the linea alba and to help in the formation of the sternum.

This interpretation of the meaning of the Wolffian ridge is based partly on direct observation, partly on argument from analogy, and partly on the fact that the ridge does not exist in lower vertebrates.

As is shown below (para. 16), there is in axolotl a packing of cells along the lateral line which appears to foretell the coming of the Wolffian ridge. This packing affects the cells of the loose mesenchyme of the unsegmented belly wall for a short distance in advance of the growing edges of the segments, and has all the appearance of being mechanical in origin (see figs. 7, 8, and 9). The accumulation of mesenchyme affected in this way in axolotl is, however, small in comparison with the accumulation in the human embryo, and the reason suggested by transverse sections of embryos and larvee of lower vertebrates is that in them the unsegmented mesenchyme is so diffuse and contains so much fluid that in spite of the pressure to which it is subjected it is still well within its limits of compression. Further, the absence of the Wolffian ridge in lower vertebrates shows it to be a structure of no great phylogenetic age. It, therefore, is not a vestige of a lost structure (lateral fin) and it is not the rudiment of any adult structure. The reasonable assumption, therefore, is that it is a by-product of the developmental process which has not yet been seized upon by selection. These two points help to strengthen the interpretation of the meaning of the ridge adopted, more particularly as the mesenchyme of the human embryo possesses just that density which the corresponding tissue lacks in the lower vertebrates, and all that is required to transform the cell-accumulation of axolotl into the Wolffian ridge of man is a relatively incompressible mesenchyme.

To resume consideration of the process of development of the human embryo. At the anterior and posterior ends of the wall of the thorax and abdomen, that is at the sites of the pectoral and pelvic faults in ventral segmentation, the advance of the lateral edges of the segments towards the mid-ventral line is arrested. The lines of frontier, between the central and anterior derived visceral regions and the central and posterior derived visceral regions, must like all other lines of frontier between cell areas be lines of pressure. Against the resistance offered by the compressed tissue of the contiguous margins of the visceral areas the axial segments fail to make headway and are, as is shown by the ultimate distribution of the segmental nerves, almost entirely arrested in their advance towards the mid-ventral line. Just opposite the tips of the arrested segments the limb buds, which are masses of undifferentiated vascular mesenchyme, are formed upon the Wolffian ridge. The reason for the formation of limb buds, that is for the accumulation of the mesenchyme at four points, is to be sought partly in the greater resistance offered at these points to the advance of the segments and partly in the fact that at each of the points there are three growing edges in contact, each contributing its quota of cells and each unable to advance because of the resistance of the others.

Fig. 5, model of the hind end of Human Embryo, Series R., No. 23, shows the exact relations of the Wolffian ridge to the remainder. of the central visceral region on the one hand, and to the advancing edge of the axial region on the other. It also shows the relation of the limb to the Wolffian ridge, to the central and posterior derived visceral regions, and to the axial region. Fig. 6, view of the left side of the same model, shows the same general relations from a different point of view, and also shows, through a window made by removing a mass of the vascular mesenchyme, the relation of the segments to the limb bud. At this stage they are sharply limited and take no part in the formation of the limb, nor have they up to this stage, so far as can be discovered, contributed anything to the substance of the developing limb; later they contribute much.

Fig. 5. — View of right side of model 1. Hind end of Human Embryo, Series R., No. 23,

Fig. 6. — View of the left side of model 1, To show the relation of the segments to the limb bud.

13. Just as the Wolffian ridge is a thickening of the lateral parts of the visceral regions of the embryo (membrana reuniens), and just as the limb bud is a special thickening upon the Wolffian ridge to which the segments have contributed nothing, so in the next stage of its development is the limb altogether independent of cellular contributions from the segmented axis. Towards the end of the fourth week in the case of the arm, fifth week in the case of the leg, there is a recognisable condensation of the mesenchyme in the centre of the cell-accumulation. This is the rudiment of the skeleton of the limb. From it the membranous skeleton arises; from the membranous the cartilaginous; from the cartilaginous the osseous.

Of the other structures which constitute the adult limb the nerves and some of the muscles are. axial in origin, that is, are derived from the segments which default ventrally ; the rest of the muscles and the fasciz are probably the representatives of the original mesenchyme; the bloodvessels are formed locally, though at the base of the limb there may be some vessels showing a segmented arrangement (33).

The essential point is that the framework of the limb has nothing to do with the axial segments, and therefore the ground plan of the limb is not segmental. The segmental structures that are borrowed by it are broken up and made to adapt themselves in large measure to the non-segmental plan. At their origin from the cord the limb nerves show a segmental arrangement, as they must, but once within the limb they are redistributed as required, though generally retaining their connexion with the parts developed from the tissues that were nearest to them as they entered, 2.e. the cephalad nerve retains its relation to the cephalad border of the limb, ete.

14. As the free limb develops and increases in size and importance the proximal end of its developing framework begins to grow towards the median plane. The growing end, in the case of the lower limb in the human embryo, divides into three processes which separate to avoid the body cavity. One process passes into the dorsal segmented region, one into each of the two visceral regions which contributed mesenchymal cells to the limb anlage. In this way a sort of three-pronged grapple is developed which attaches the limb to the body. The dorsal prong is the ilium (scapula), the ventral prongs the ischium and pubis (precoracoid and coracoid) (34).

15. Summarising the observations recorded in the last paragraphs, we recognise that in the human embryo the body wall (ventral thoracicoabdominal wall) exists first as a simple unsegmented sheet in which there is a relatively very large opening, the umbilicus. Into the unsegmented sheet lateral outgrowths from certain of the axial segments (8th or 9th to 21st or 22nd) pass. As these grow out, there is formed from the unsegmented mesenchyme along the line of frontier, between the segmented and unsegmented regions, a linear accumulation of cells—the Wolffian ridge. This ridge is not confined strictly to the area of formation of the thoracico-abdominal wall, but extends forward along the line of frontier between the axial and anterior derived visceral regions, distinctly to the 4th spinal segment, and with rather less distinctness as far as the first. It also extends backwards along the line of frontier between the axial and posterior derived visceral regions distinctly as far as the 26th segment, and indistinctly into the lower sacral region. The 5th to the 9th spinal segments, as they grow laterally, come into relation with the lateral ends of the line of frontier between the anterior derived and central visceral areas. The advance of these segments is stayed and a special thickening of the Wolffian ridge takes place opposite to them. This thickening of the Wolffian ridge is formed of vascular undifferentiated mesenchyme derived from the visceral areas, and is the basis of the fore limb. In a precisely comparable manner, though at a later date, the advance of spinal segments, 21st-26th, is stayed, and for a precisely comparable reason, a precisely comparable thickening of the Wolffian ridge appears opposite their extremities. This is the basis of the hind limb.

From these mesenchymal non-segmented masses the skeletal bases, the fasciee, the blood-vessels and possibly some of the muscles of the limbs are developed, but the nerves of necessity are derived from (or at least are connected with) the segmental nerves, and some of the muscles probably are derived from the segmental myotomes.

In this way the segmented axis contributes to the formation of the definitive limb, but the limb is not an outgrowth of the segments, nor has it, within necessarily narrow limits, an unalterable connexion with any particular dorsal segments. It is as if there were a slight element of chance involved in the precise interrelations of the segments and the extremities of the transverse inter-visceral-regional lines of frontier. This is evidenced by the variations which exist in the formation of the nerve plexuses to the limbs! (high and low lumbo-sacral plexus) and in the variability of the level of attachment of the pelvic girdle to the vertebral column (forward and backward sacral homceosis) (35), (36).

16. In Menobranchus lateralis the relations of the limb rudiments to the areas of the developing body are readily evident. In axolotl, in which the development of the limbs is delayed, the relations are less evident though essentially similar.

Fig. 7 shows a transverse section through the body of Axolotl, Series G., No. 18, taken a short distance anterior to the site of development of the

1 This is independent of the variation in the constitution of individual spinal nerves which appears to be due to variations in the relations of the central nervous system to the mesoblastic segments. This must give rise to variations in the secondary segmentation of the central nervous system (36, also 27).

pelvic limb. The section cuts the posterior part of the. mid gut and the posterior part of the central visceral region a short distance anterior to the line of frontier between it and the posterior derived visceral region. The general structure of the unsegmented mesenchyme is shown in the lower ventral part. A little more than the ventral half of the lateral abdominal walls are still unsegmented. In this the cells show a certain degree of packing which contrasts strongly with the arrangement of the cells in the other parts of the mesenchyme.

Fig. 8 shows a transverse section through the anterior part of the frontier territory of the central visceral region (frontier between ‘posterior derived and central visceral regions). The section passes through the terminal part of the mid gut. The lower part of the slightly concave anterior border of the ventral mesentery of the hind gut is cut in the plane of section (projection on ventral wall of coelom).

Fig. 7. — Microphotograph T. S., Axolotl, Fig. 8.—Microphotograph T.S., Axolotl; Series G., No, 18. x20. Series G., No. 18. x20.

The segmented mesoblast (muscle mass) shows darker in the photograph, and is confined to the upper quarter of the lateral wall of the ccelom (¢f. last section, in which the upper half has been invaded by the segments). Comparison with fig. 7 shows that thée'cellularity of the lateral wall of: ‘the coelom is here more marked. (Increased density of packing.)

Fig. 9 shows a transverse, section just posterior to the commencement, of the hind gut; the ventral mesentery is complete in the section which passes practically through (slightly posterior to the centre of) the line of frontier between the central and posterior derived visceral areis. The segmented mesoblast is confined to the upper quarter of the lateral walls of the ccelom, but the unsegmented lower three-quarters of the lateral wall are so cellular that they have stained even more darkly than the segmented part. This is the cell-accumulation which is the beginning of the hind limb.

Fig. 10 shows a section through the cloaca. The unsegmented mesenchyme belongs entirely to the posterior derived visceral region. The almost complete absence of cellularity contrasts strongly with the appearance shown in the last section.

Fig. 9. — Microphoto oh T. S., Axolotl

Fig. 10. Microphotograph T.S., Axolotl, Series G., seh Series G., 18. x 20.

Fig. 11 shows the pelvic girdle and the femur which develop from the mass of cells seen in figs. 7, 8, and 9. The figure represents the model of the hinder end of Axolotl, Series G., No. 280 (model No. 4). The pelvic girdle shows the three usual parts (v. inf.).

Fig. 12 shows a transverse section of Axolotl, Series G., No. 2, through the region at which the fore limb is to appear. Spreading out on either side dorsally is the operculum with the cartilages of the branchial arches - lying in its concavity. Beneath the notochord is the fore gut, posterior to the pharynx. Ventrally is the bulging unsegmented belly wall with its contents. The angle on the right side of the body wall indicates the position of the line of frontier between the segmented and unsegmented parts.

Fig. 13 shows a transverse section of Axolotl, Series G., No. 18. This section is somewhat oblique, and cuts the free branchiw on the right side, the operculum and branchial arches on the left. At a point on the left side corresponding in position to the angle seen in the last section, an oval mass of cartilage—the developing limb girdle—is visible. On the right side the humerus is cut just before it clears the body wall.

Fig. 11. — View of model 4. Developing pelvic girdle of Larva, Series G., Axolotl, No. 280d.

Fig. 12. — Microphotograph T.S., Axolotl. Fie. 13.—Microphotograph T.S., Axolotl, Series G., No 2 x20. , Series @, Nols x20.

Fig. 14 is a view of a reconstruction model (model No. 2) of the head end of this same embryo (Axolotl, Series G., No. 18). A considerable ventral dissection has been effected to expose the pharyngeal roof and the branchial arches. The limbs are not quite symmetrical in position; the left is the more developed. °

The model shows the precise placing of the rudiments of the fore limb girdles ventral to the posterior limit of the head, that is, at the junction of the central visceral and anterior derived visceral regions. It also shows that the pectoral girdles have developed three bosses—scapula, coracoid, and precoracoid.

Fig. 15. represents the model illustrating the development of the pectoral girdle of Axolotl, Series G., No. 280a (model No. 3). The long, spoon-like scapule are seen to have extended back towards the mid-dorsal line and the coracoids to have developed into ventral plates to support the viscera. The branchial arches in the lateral parts of the ventral walls are shown in the background.

Fig. 14. — View of model No. 2. Anterior end of Axolotl, Series G., No. 18. .

Figs. 14 and 15 are especially interesting as showing the complete independence of the framework of the fore limb girdle from all the branchial structures. It is placed in the body wall; they in the pharyngeal. In the section shown in fig. 13 there is a distinct difference in the staining reaction of the cartilage of the branchial arches and of the limb structures. It is difficult to say whether this is due to an intrinsic difference or merely to a difference in the age of the cartilage cells.

In summary the development of the limbs of axolotl is essentially similar to the development of the limbs of man. The first rudiment of the limb is a cell-accumulation at the junction of the two visceral regions with the axial region. This cell-accumulation increases in density, and forms the skeletal basis of the limb, which grows distally to form the skeleton of the free limb and proximally to form the girdle. Each girdle consists of a three-pronged grapple. Each prong grows at first into one of the regions involved in establishing the cell-accumulation which initiates the growth of the limb. The simple grapple-like condition is soon obscured by the specialisation of the girdle to meet the needs of its environment. 17. In some of the teleosts whose eggs are distended with yolk, there is some difficulty in recognising the anterior derived visceral region, as its clear demarcation is delayed until after the pectoral limb bud has formed.

Fig. 15. — View of model No. 8. Developing pectoral girdle of Larva, Series G., Axolotl, No. 280a.

Figs. 16, 17, 18, and 19 represent models 5, 6, and 7 prepared from Trout, Series R., 22, 28, and 36, to show in general view the chief stages in the development of the pectoral limb of the teleost. These are best interpreted by reference to the sections shown in figs. 20, 21, 22, 23, and 24.

Fig. 20 shows a transverse section of Trout, Series R., No. 18, taken immediately posterior to the cardiac area, which, by tracing development backward from the free-swimming stage, can be recognised as placed at the posterior limit of the anterior derived visceral area (cf. figs. 1 and 3). The section, therefore, passes through the region which corresponds with the line of frontier between the anterior derived and central visceral regions. This shows, in the centre of the field, the axial segmented part of the embryo, with spread out on either side of it the tissues which are to form the belly wall (yolk sac wall). The ccelom is represented by the two cavities which underlie the lateral parts of the embryo, and extend for a short distance on either side beyond the lateral margins of the segmented axis.

Beneath these is the yolk in the yolk sac. The dorsal wall of each part of the ccelom shows an open bay which involves both epiderm and mesoderm.

Fig. 16. — View of model 5 from in front. Whole embryo, Series R., Trout, No. 22. The limb buds are shown as dark cones rising from the centre of slightly elevated areas. The lower part of the ‘‘ yolk sac” is merely approximately correct. The greater part of it was freely worked into the model to provide a base for the embryo.

Fig. 18. — View of the right side of model 7. Pectoral region of embryo, Series R., Trout, No. 36. Toshow the attachment of the limb at the line of junction of the axial part and belly wall, a window has been made on the outer side of the fin to show the developing framework of the free limb and of the two ventral prongs. The dorsal prong shows white throngh a second window made at the side of the axial region.

Fig. 17. — View of model 6 from behind. Whole embryo, Series R., Trout, No. 28, To show the growth of the limbs. The ‘*yolk sac,” except in the immediate neighbourhood of the embryo, was freely worked into the drawing to give a general idea of the relations of the parts.

Fic. 19.—View of left side of model 7. To show the attachment of the limb at the junction of the belly wall and axial region. The developing framework of the free limb and the dorsal prong of the girdle are shown.

The interpretation of this, which is part of a rather flat pyramidal or small dome-shaped projection, is that the cell proliferation in the region extending between the lateral margin of the axial segmented part of the embryo and the lateral margin of the ccelom, is proceeding more rapidly than is necessary to keep pace with the extension of the latter. In result, a cell-accumulation occurs which bends the membrane in a dorsal direction, a purely mechanical phenomenon.

Fie. 20. — Microphotograph T.8., Trout, Series R., No. 18. x50.

Fig. 21 represents the next stage in the development of the limb. It shows how the mesenchyme aceumulates in the concavity of the epidermal bay, and forces the mesodermal projection back into the general plane of the body wall. (This is a section through the anterior part of the limb bud seen in relief in fig. 16, model 5.) It also shows well the mechanism of arrest of the lateral part of the segmented outgrowth; its further advance is obviously prevented by the mass of mesenchyme. Fig. 22 shows the relation of the parts posterior to the limb bud. The section passes through the axial and the central visceral regions. There is a slight accumulation of mesenchyme occupying the space ventral to the segments, between them and the mesoderm. Apart from this the mesenchyme forms a thin, clear layer between the epiderm and mesoderm of the central visceral region of the body wall, which here spreads out almost horizontally.

Fig. 21. — Microphotograph T.S., Trout, Series No, 22. x 50 (see also fig. 16).

Fig. 23 is a transverse section through the developing limbs of Trout,

Fig. 22. Microphotograph 1 T.S., Trout, Series R., No. 22, x50.

Series R., No. 28, shown in fig. 17, model 6. It shows the complete inde pendence ‘of the dense conical masses of limb mesenchyme from the axial mesoblast.

Fig. 24 is a transverse section through the developing limb of Trout, Series R., No. 36, shown in figs. 18 and 19, model No. 7. It shows the condensation of the central mesenchyme to form the skeletal base of the limb, and also very precisely the exact placing of the limb at the junction of the axial and central visceral regions. Figs. 18 and 19, representing the right and left sides of model No. 7, show the development of the skeletal base of the free fin (cartilaginous plate) and the production of the three centrally growing processes, one dorsal and two ventral, which form the girdle (37).

Fig. 23. — Microphotograph T.S., Trout, Series R., No. 28. x 50 (see also fig. 17).

Summary of the Development of the Limb in the Teleost

Although the formation of the anterior derived visceral region (freeing of the ventral surface of the head) is delayed, because of the distension of the yolk sac, a region exactly comparable can be recognised partly through a consideration of its ultimate fate, partly on account of the position of the heart. At the posterior part of this region, in the neighbourhood of its angular junction with the axial and central visceral regions of the embryo, a cell-accumulation occurs which leads at first to an outward (upward) folding of the whole thickness of the body wall. This is soon obscured by the rapid proliferation of the mesenchyme in the region which becomes the core of the limb bud. The later stages of development of the limb essentially correspond with those seen in the batrachians and in man. (The exact placing of the root of the pectoral limb at the junction of the ~ three regions is very evident in the newly hatched trout.)

Fig, 24.—Microphotograph T.S., Trout, Series R., No. 89. x 50 (see also figs. 18 and 19).

18. These observations are in favour of the general accuracy of the hypothesis that the archecentre of the limbs is to be found in the existence of four areas, in the body wall, at which cell-accumulation is liable to occur. At each of these areas three of the five body regions meet. From this tendency towards the formation of four cell-accumulations natural selection has established the four vertebrate limbs. There can only be four such areas; there are only four vertebrate limbs.

This hypothesis of extra-segmental origin of the limbs is considerably strengthened by the known facts with regard to the very variable metameric position of the limbs, especially in fishes. In fact, were the metameric position of the limbs constant the hypothesis in its present form could not legitimately have been advanced. As itis, it stands to some extent midway between the gill arch and lateral fin hypothesis. Its relation to the latter is obvious; it merely replaces the suggestion of a primarily continuous lateral projection, by that of primarily discontinuous projections essentially similar in origin to the continuous median fin. It recognises, in short, that though there is a tendency to produce a lateral projection. the factors upon which this depends are only operative when the resistance presented by the visceral areas to the growth of the axial area is sufficient to cause cellaccumulation. In the lower vertebrates the unsegmented mesenchyme is extremely diffuse, and only at the four points where the resistance is somewhat increased by the interaction of three areas can a sufficient cellaccumulation occur to provide a rudiment which natural selection could fashion into a limb. In higher vertebrates resistance, as a result of the greater density of the mesenchyme, is sufficient all along the line to produce cell-accumulation, but this is ignored by natural selection because the conditions of life of the higher vertebrates make no demand for continuous lateral fins or third pairs of limbs.

The hypothesis is also related to that of Gegenbaur, for the fore limb rudiment, more especially its basal girdle, obviously develops to some extent in series with the branchial arches which find their anlagen in the primitively unsegmented mesenchyme of the anterior derived visceral region. This wholly unexpected linking of the gill arch and lateral fin hypothesis is interesting and important, for it brings practically all the evidence in support of the older hypothesis to the service of the new.

19. Once it is recognised that the archecentre of the vertebrate limb is an accumulation of mesenchyme which differs from the rest of the mesenchyme only in its topographical relations, it becomes clear that attempts to establish elaborate homologies between the parts of the highly specialised free fins of fishes and the parts of the no less highly specialised free limbs of tetrapods are practically unprofitable (38). Even in attempting definitely to homologise parts of the limb of a lower tetrapod with those of a higher there are pitfalls.

For in dealing with any adult organ it is necessary clearly to distinguish between those of its characters which are inevitably produced if it develops at all and those which are impressed upon it in ontogeny by its environment. These may be called the accidentia of the organ ; those collectively its proprium. For example (fig. 25), the human tibia is normally a long bone with a strong prismatic shaft. The great mass, length, and prismatic shape of the shaft are, however, accidentia. This is shown by the fact that these characters are not produced in a leg the victim of infantile paralysis. The bone in such a case is short, slender, and cylindrical.

It is, of course, impossible completely to separate the proprium of a structure from its accidentia, for in the case just referred to the tibia of the paralysed limb, though not acted upon by a normal environment, was without doubt subjected to some influences from without itself. It is clear, however, that accidentia being due to the effect of environment in ontogeny can possess no direct morphological significance. The difficulty of distinguishing between proprium and accidentia and even of conceiving the proprium apart from the accidentia is, however, so great that the utmost care has to be exercised in drawing conclusions from the facts of comparative anatomy. As a rough working rule, in cases lacking fuller evidence, the proprium may be regarded as practically identical with the embryonic rudiment.

Fic. 25.—To show the contrast between the tibia from a normal left leg and that from a case of infantile paralysis.

But further, in dealing with homologies it is at times a matter of difficulty to distinguish them from homogeneous and even heterogeneous homoplasies. To obviate this difficulty the working rule with regard to morphological identity must be that the embryonic rudiment of the structures which it is sought to homologise should clearly be recognisable as a development or amplification of the archecentric condition of a larger The Origin of the Vertebrate Limb 377

group which has as branches the groups whose representatives are under consideration. It is clearly not necessary for the embryonic rudiment to reproduce the definitive archecentric condition; for every definitive structure in the higher animals is compounded of proprium and accidentia, or else during ontogeny there could not have been any adjustment to environment, and archecentric accidentia are devoid of greater significance than neocentric.

The ascertainable archecentre of the skeleton of the vertebrate limb is nothing more than a condensed core of mesenchyme which, growing centrally, tends to divide into three processes or prongs. One of these passes towards the mid-dorsal line and two towards the mid-ventral line— one into each of the neighbouring primitive embryonic regions—to form a grapple to fix the limb to the body wall. The other end of the condensed core of mesenchyme growing laterally into the free limb forms a supporting framework of the type required by the limb to enable it to perform its function. Without a complete knowledge of the phylum, which must be established by considerations independent of the limb, it is impossible to say more.

The archecentre of the tetrapod limb (metacentre of the generalised vertebrate limb) can be slightly more accurately defined. Its proximal growth repeats the archecentric condition forming a three-pronged grapple ; its distal growth produces a terminal piece, rod-like or plate-like, which secondarily divides to form digits; this is supported by two rods which in turn are supported by one rod which connects with the three-pronged grapple fixed in the body wall. Further than this it is impossible to go with certainty.

20. In the comparison of the fore and hind limbs and girdles of any one animal it is necessary to recognise that the hypothesis advanced makes it essential to regard the hind limb of either side as a mirror image repetition of the fore limb of its own side, just as it is a mirror image repetition of the hind limb of the opposite side. The need for this arises from the fact that the cell rudiments of the limbs are formed at opposite ends of a continuous unsegmented, uncephalised area where that area is in contact with secondary unsegmented areas. Thus each limb possesses a border deriving from the central visceral region and a border deriving from a derived visceral region. These borders may be distinguished as paromphalic and apomphalic.

Belief in such an underlying mirror image correspondence between the fore limb and hind limb of either side is not new. Humphry in his Treatise on the Human Skeleton (89) wrote: “The key to the exact homology of the upper and lower limbs, I apprehend is furnished by the fact that they are placed at opposite ends of the trunk and that the opposed surfaces of their upper segments have consequently been made to correspond with one another. Thus the scapula is inclined backwards and the ilium forwards; the hinder edge of the scapula corresponds with the anterior edge of the ilium; and the rough projection near the glenoid cavity for the attachment of the long portion of the triceps—the extensor of the forearm—corresponds with the anterior inferior spine of the ilium which gives attachment to the rectus femoris—the long portion of the extensor of the leg; the coracoid process is homologous with the pubis and the clavicle with the ischium. ... The lesser trochanter of the femur which receives the iliacus muscle coming from the inner surface of the ilium looks backwards, and the lesser tubercle of the humerus which receives the subscapularis muscle coming from the inner surface of the scapula looks forwards. The outer and inner surfaces of the two bones respectively correspond with each other; thus, the great trochanter of the femur and the great tubercle of the humerus are both directed outwards; the rough space for the great gluteus, which comes from the tuber ischii and the sacroischiatic ligament, is upon the outer side of the shaft of the femur; and the rough space on the humerus for the deltoid, which comes from the clavicle and the acromion, is upon the outer side of the shaft of humerus ; the rough space for the long adductor muscle, which comes from the spine of the pubis, is on the inner side of the shaft of the femur; and the rough space for the coraco-brachialis, which comes from the tip of the coracoid process, is on the inner side of the shaft of the humerus; the outer condyle of the femur corresponds with the capitulum of the humerus, and the inner condyle of the femur corresponds with the trochlea of the humerus.

“The knee bends backwards and the elbow bends forwards; the flexure in each case bringing the limb more under the middle of the trunk. The patella which receives, or is developed in, the extensor of the leg, is on the anterior surface of the knee; and the olecranon which receives the extensor of the arm is on the posterior surface of the elbow.”

From this extract it is clear that Humphry regarded the limb girdles as morphologically similar caps, applied, as it were, to the opposite ventroterminal aspects of the trunk where it is joined by the neck and tail. These caps and the proximal segments of the limbs, at least, he regarded as possessing a fore and aft mirror image correspondence. This idea has never wholly died out of anatomical thought. Parsons (40), so recently as 1908, published a figure (reproduced in modified form as fig. 26) showing the “looking-glass” symmetry of the muscles of the fore and hind limb. But apart from this the idea may be said to have been practically killed by Huxley when he riveted the words preaxial and postaxial to the borders The Origin of the Vertebrate Limb 379

of the limbs. The assumption underlying these words is that the limbs from their start form parts of a cephalised system. The evidence that Huxley brings forward in support of it is that any other view “would destroy the homology of the pollex and hallux—which is surely out of reach of doubt” (41). Not only is this homology not out of reach of doubt ; it is hardly within the range of possibility (v. inf., nerve supply).

Sabatier (42) supported Huxley’s view, and states, as conclusive proof of its correctness, that anyone who doubts that the preaxial border of the thoracic limb corresponds with the preaxial border of the pelvic limb and the postaxial with the postaxial is in effect homologising the skull with the last tubercle of the coceyx. This would be a just criticism if it were contended that the limbs were definite processes of the axial segments and were from their start segmental, but it is absurd if the possibility of extraaxial origin of the limbs be admitted.

Fig, 26. — To show the mirror image symmetry of the muscles of the fore and hind limb from the dorsal aspect (modified from Parson’s figure, vol. xlii., p. 389, this Journal).

a, teres major and subscapularis; a’, ilio psoas; 8, triceps (long head); 6’, rectus femoris ; y, extensors of wrist and hand ; y’, extensors of ankle and foot; 6, trapezio-deltoid ; 5’, gluteus maximus. 5%, scapula ; I, ilium ; H, humerus; F, femur.

The strongest evidence available in the adult in favour of the mirror image correspondence of the limbs is derived neither from the skeletal nor from the muscular systems, though the evidence from these is strong, but from the nervous system, much the most reliable guide because of its phylogenetic conservation.

21. Herringham (43), working at the problem of nerve supply to the fore limb found, inter alia, that, of two muscles, that which is nearer the surface tends to be supplied by the higher, that which is farther from it by the lower nerve.

Sherrington’s (44) subsequent investigations showed that the innervation of the muscles of the posterior aspect of the thigh and leg does not follow Herringham’s law. In this case the deep layer of muscles is innervated by roots anterior to those which innervate the superficial muscles.

There is here no contradiction in the principle of supply if the idea of the primitive cephalisation of the limb be abandoned. Utilising the phrases, “ paromphalic in origin,” “and apomphalic in origin,” to describe the nerves which arise nearer to or farther from the trans-umbilical plane, Herringham’s and Sherrington’s laws read as one:

“Of two limb muscles that which is the more superficial is usually supplied by a nerve apomphalic in origin to that which supplies the deeper muscle.”

It would be hard to find more perfect evidence in favour of the mirror image correspondence of the limbs.

Sherrington’s experiments, which bring out the mirror image correspondences in the principles of the nerve supply of the limb, also furnish evidence that the hallux has undergone a late change, a confusion in development, it might almost be called.

According to the mirror image hypothesis the thumb and little toe are developmentally corresponding structures; functionally the thumb and great toe correspond. That this is due to a late transformation of the great toe is made practically certain from its nerve supply. The skin of the great toe is supplied by a nerve paromphalic in origin to that which supplies the skin of the little toe, but the nerve supply to its musculature is derived in part at least from the nerve supply proper to the little toe. In other words, the evidence available from the nervous system goes to show that secondarily, that is subsequent to the establishment of the dominance of the segmented cephalised axial part of the body when the limbs are required to function as part of an animal with a definite head and tail, the great toe approximates in type to the thumb, taking over from the thumb’s real homologue, the little toe, a sufficiency of its nerve supply and of its musculature fully to equip it for its new position (cf. apomphalic, fibular, origin of flexor longus hallucis). In other words, the correspondence between the thumb and great toe is an example of heterogeneous homoplasy.

These facts concerning the musculature and nerve supply of the great toe strongly support the hypothesis of limb origin which has been advanced.

The principles of innervation of the limb muscles, the arrangements of the muscles themselves, and the broad outlines of the bones all-point to the existence of a mirror image correspondence between the fore and hind limbs. The confused innervation of the great toe also supports this view, though, at first sight, the apparent comparison between the thumb and great toe seemed to provide the strongest evidence against it.

Such a mirror image correspondence between the parts of the fore and hind limb can be explained on no hypothesis of limb origin which is wholly independent of the suggestions advanced in this paper. Possible modifications of the cell-accumulation hypothesis there are in plenty. No finality is claimed for its details, but in its broad outlines it is believed to be correct. This completes, so far as this paper is concerned, the discussion of the foundations of the hypothesis and of the principal deduction which flows from it. In the next part the hypothesis is given in summary.

Part III. Summary and Conclusion

There exist in every developing vertebrate five cell regions, which have different characters and different histories. The frontiers of these regions are lines of potential cell-accumulation. At points where three regions meet the tendency to cell-accumulation is especially marked. There are four such points, and the four cell-accumulations formed around them provided, far back in the phylogeny of the vertebrate stock, material capable of being selected to form the bases of new organs, and in fact it was selected to form organs which would bring the inner animal into closer relations with its environment. At their commencement the limbs were, and still are, altogether independent of the cephalised axial part of the embryo, but later passed, and still pass, completely under its dominance.

The primary correspondence between the fore and hind limb of any one side is a simple mirror image symmetry, and this persists in the essential parts of the limb though it is blurred by a secondary homoplastic convergence made necessary by the fact that the limbs are utilised by, indeed form part of, an animal which normally moves head first.

The existence of just two pairs of limbs is an absolutely fundamental characteristic of the vertebrate stock ; so too is the early predominance of the fore limb over the hind.


In conclusion, I desire to acknowledge the great help I have received from Dr N. C. Rutherford. I have drawn largely upon the series which he has prepared (Series R.). I am also indebted to him for many pages of careful abstracts of the literature of the development of the limbs.

To the Royal Zoological Society of Ireland I am indebted for the axolotl material used in this investigation.

The photographs are the work of Mr Wm. Gill; the drawings of the models the work of Mr Whelan.


This list refers merely to publications directly referred to or quoted in the text. For full literature list see Comparative Anatomy of Vertebrates, Wiedersheim and Parker, London, 1907.

(1) Vice v’Azyr, “ Mémoire sur les rapports qui se trouvent entre les usages et la structure des quatre extrémités dans l’homme et dans les quadrupédes,” Geuvres recueilltes, tom. iv. p. 313, 1805.

(2) Oxen, Lehrbuch der Natur-Philosophie, p. 330, 1843.

(3) Macuise, Todd’s Cyclopedia, vol. iv. p. 70, fig. 490, 1832.

(4) Owen, Archetype and Homologies of the Vertebrate Skeleton, 1848.

(5) Goopsir, Edin. New Phil. Journ., vol. v. p. 178, 1875, and Anatomical Memoirs, ed. Turner, vol. ii., 1868.

(6) Humpury, Journ. of Anat. and Phys., vol. iv. p. 58.

(7) GuGENBAUR: a. Untersuchungen, Heft 3, P. 181, note (reference from

Mivart ; see 8). 6. Grundriss der vergleich. Anat., p. 494. c. Morphologisches Jahrbuch, Bd. xi. Heft iii. p. 417. d. “Das Flossenskelet der Crossopterygier,” Morph. Jahrb., Bd. xxii., 1895. e. Vergleichende Anat. der Wirbeltiere mit Berucksichtigung der Wirbellesen, Bd. i, Leipzig, 1898. (8) Mivart, Trans. Zool. Soc. of London, vol. x. p. 439, 1879. (9) Batrour: a. Comp. Embryology, 2 vols., 1880-81. b. Monograph on the Development of Klusmobranch Fishes, 1878. c. “ Development of the Skeleton of the Paired Fins of Elasmobranchs,” P. Zool. Soc., 1881.

(10) THacuer, ‘“ Median and Paired Fins,” Trans. Connecticut Acad., 1878.

(11) Goopricn, “ Notes on the Development, Structure, and Origin of the Fins,” Q./.M.S., June 1906. See also Wiedersheim, Comparative Anatomy of Vertebrates, 3rd ed. (6th German), London, 1907. (The short statement of the lateral fin hypothesis in text is modified from that given in this work.)

(12) Bravs, Handbuch der vergleich. u. experiment. Entwickelungslehre der Wirbeltiere (Hertwig), Bd. iii. Teil ii. Kap. v.; Entwickelung der Form der Katremitaten u. des Extremitatenskelets, Jena, 1906. See also papers, Jen. Zettsch. fiir Naturw., Bd. xxix., 1895; Morph. Jahrb., Bd. xxvii., 1899 ; Haeckelfestch. Denkschr. Med. Nat. Ges., Jena, 1904.

(13) WALLACE, A. R., The World of Life, Vondon, 1910. (Cf. also Woodward, Presidential Address, Geolog. Sect. Brit. Ass., 1909.)

(14) Mivart, Trans. Zool. Soc. of London, vul. x. p. 481 (footnote), 1879.

(15) Lanerey, “Lumbar Plexus of Cat,” Journ. Phys., vol. xii. p. 349, vol. xv. p. 210, vol. xvii. p. 296.

(16) SHeRRinGtoy, Journ. Phys., vol. xiii., p. 639.

(17) Paterson, Journ. Anat, and Phys., xxviii. p. 84.

(18) Barpgen and Extina, Anat. Anzeiger, xix. pp. 124-135, 209-238.

(19) CunnincHam, Text-Book of Anatomy, 3rd ed., London, 1909.

p (20) Mixon Lab, Text-Book of Embryology, p. 2 (Metamerism), Philadelphia, a, .

(21) Kerpeu (Hertwig’s Handbuch, ref. 12); short well-illustrated account, Bd. i. Teil ii. Kap. vi. p. 62.

(22) Kerpen and Enza, Normentafel (see also Manual of Human Embryology, Keibel and Mall, London, 1910).

(23) His, Anatomie menschlicher Embryonen (Atlas), Leipzig, 1885.

(24) Saerrineton, “ Experiment in Examination of the Peripheral Distribution of the Fibres of the Posterior Roots of some Spinal Nerves,” Proc. Roy. Soc., lii., 1892, and Phil. Trans., 1893.

(25) Herrineuam, “Minute Anatomy of the Brachial Plexus,” Proc. Roy. Soc., xli., 1886.

(26) Grosser u. Fr6aticH, Morph. Jahrbuch, xxx.

(27) Quain, Hlements of Anatomy, vol. iii. part 2, Schafer and Symington, 11th ed., London, 1909.

(28) Waterston and Geppgs, ‘ Report upon the Anatomy and Embryology of the Penguins collected by the Scottish National Antarctic Expedition,” Trans, Roy. Soc. Edin., vol. xlvii, part 2, No. 10.

(29) Cf. references 1, 2, and 3.

(30) Kerpet, Hertwig’s Handbuch (reference 12), Bd. i. Teil ii. Kap. vi. pp. 15-136.

(31) Persr, Arch. mikr. Anat., Bd. 1xi., Bonn, 1902.

(32) Prnkus, ‘‘ Development of Integument,” Manual of Embryology, Keibel and Mall, vol. i., London, 1910.

(33) Miter, “ Beitrige zur Morphologie des Gefasssystems: 1. Die Armarterien des Menschen ; ii. Die Armarterien der Saugetiere,” Anat. Heft, xxii. and xxvii.

(34) BardeEn and Lewis: a. American Journal of Anatomy, vol. i. p. 1, 1901, plate xi., fig. B. 6. Bardeen, “ Develop. of Skeletal System,” Manual of Embryology, Keibel and Mall, vol. i., London, 1910; also literature.

(35) Parerson, ‘On the Human Sacrum,” Scientific Transactions of the Roy. Dublin Soc., 1893.

(36) Batsson, Materials for the Study of Variation, London, 1894.

(37) SwinneErton, “ Pectoral Skeleton of Teleosts,” Q.J.M.S., November 1905.

(38) Mivart, loc. cit., 14.

(39) Humpury, Treatise on the Human Skeleton, London, 1858.

(40) Parsons, “Further Remarks on Traction Epiphyses,” Journ. of Anat. and Phiys., xiii.

(41) Huxtey, “On Ceratodus Forsteri,” Proc. Zool. Soc., London, 1876, (Quoted from p. 119, vol. iv., Scientific Memoirs of T. H. Huzley.)

(42) Sapatier, Comparatson des Ceintures et des Membres Antérieurs et Postérieurs, Montpellier, 1880.

(43) Herrineuam, Proc. Roy. Soc., xli., 1886, p. 437.

(44) Suerrineton, Proc. Roy. Soc., li., 1892, p. 77.

Cite this page: Hill, M.A. (2020, October 22) Embryology Paper - The origin of the vertebrate limb (1912). Retrieved from

What Links Here?
© Dr Mark Hill 2020, UNSW Embryology ISBN: 978 0 7334 2609 4 - UNSW CRICOS Provider Code No. 00098G