An attempt has been made to clarify the origin of the supernumerary regenerates of triple legs by transplantations between two species of cockroaches, Gromphadorhina portentosa and Leucophaea maderae. The three ways of combining stump and transplant tissues were: heteropleural, dorso-dorsal, antero-posterior (I); heteropleural, dorso-ventral, anteroanterior (II); homopleural, dorso-ventral, antero-posterior (III). The transplantations have been performed at the level of the tibia as well as at the level of the coxa.

(1) The supernumerary regenerates were partly mixed, i.e were composed of a mosaic of the tissues of both species of cockroach, partly homogeneously built up by the tissues of only one of the two species. In combination I nearly all regenerates were mixed; in the other two combinations homogeneous regenerates were relatively numerous, sometimes even the most numerous.

Both supernumerary regenerates of a triple leg might be mixed; or one was homogeneous and the second mixed; or they were both homogeneous. In the latter case one regenerate was built up by Gromphadorhina tissues, the other by Leucophaea tissues, or, more rarely, both were built by up Gromphadorhina tissues.

(2) The intraspecific transplantations simultaneously done in Leucophaea gave results which correspond in general with the results of the interspecific transplantations.

(3) From the composition of the supernumerary regenerates it must be concluded that the four different properties of a cross-section (anterior, posterior, etc.) are not yet determined irreversibly in the tissues of the cockroach legs. Thus, missing properties may be completed by the other properties to give an entire cross-section.

Biologists have always been very interested in naturally occurring multiple branched appendages, especially the three-pronged appendages often observed in arthropods. The symmetric relations of the branches have proved remarkably uniform (Bateson, 1894). It was this arrangement of the different branches that induced Przibram (1921) to propose a hypothesis about their origin.

In spite of a great number of investigations in Arthropoda (Balazuc, 1948; Bart, 1971 a, b; Bodenstein, 1937, 1941; Bohn, 1965; Bullière, 1970; Furukawa, 1937, 1940; Hoarau, 1969; Lheureux, 1970, 1971; Noulin, 1970; Penzlin, 1965) the participation of the host and transplant tissues in the formation of the supernumerary branches is not yet clear. The first experiments in this respect were done by Bodenstein (1937), Bart (1971 a, b), Bullière (1970) and Lheureux (1971). Bodenstein, Bullière and Lheureux on the one hand and Bart on the other come to different conclusions, which can hardly be due to the different objects used by them since in most other regeneration phenomena all arthropods behave remarkably uniformly.

The difficulties of interpretation result from the fact that the leg parts combined have always been taken from one species, although from different segments. But often the structural differences of the appendages from different segments of the body are not great enough to allow identification of every part of the regenerate without doubt. Transplantations between the two cockroaches Leucophaea maderae and Gromphadorhina portentosa (Bohn, 1971), the tissues of which can easily be distinguished by their different pigmentation, offered a method to solve the problem of the origin of the supernumerary regenerates definitively.

Multiple regenerates regularly form when the axial organization of stump and transplant tissues do not harmonize with one another. In the main there are three kinds of disharmonie orientation (Fig. 1, I—III). Combinations I and II are obtained by carrying out the transplantations between the legs of the left and right side of the body (heteropleural transplantation) ; in combination II the transplant is in addition turned round the longitudinal axis through 180°. Combination III is obtained by doing transplantations between the legs of the same body side combined with a 180° rotation of the transplant (homopleural transplantation).

Fig. 1.

The three possible combinations of disharmonie orientation of stump and transplant. The outer circles symbolize the cross-section of the stump, the inner circles that of the transplant (looked at from distal). The letters mark the four symmetry properties: d = dorsal; v = ventral; a = anterior; p = posterior. For further explanations see text.

Fig. 1.

The three possible combinations of disharmonie orientation of stump and transplant. The outer circles symbolize the cross-section of the stump, the inner circles that of the transplant (looked at from distal). The letters mark the four symmetry properties: d = dorsal; v = ventral; a = anterior; p = posterior. For further explanations see text.

The operations have been performed with two different leg segments: the experiments of group 1 with the tibia of the midleg cut at a median or distal level; the experiments of groups 2 and 3 with the coxa of the foreleg and hind leg cut at a median level. In each of the three groups of experiments all three possible combinations (I to III) have been done. Only recently moulted larvae of Gromphadorhina portentosa (first instar) and Leucophaea maderae (second instar in group 2, third instar in groups 1 and 3) have been used for the operations. The transplantations have been performed either interspecifically (groups 1 and 3) or intraspecifically (group 2, transplantations between foreleg and hind leg of Leucophaea; see also Table 1).

Table 1.

Compilation of the results

Compilation of the results
Compilation of the results

The supernumerary regenerates appear as a rule after the first (group 1) or second (groups 2 and 3) postoperative moult. Since the regenerates of groups 2 and 3 are easily lost by autotomy, it was often necessary to wait for further moults till well-developed triple legs were available.

The tissues of Leucophaea and Gromphadorhina may easily be distinguished by their different colour : the legs of Leucophaea have a yellowish brown, those of Gromphadorhina a dark brown or even black pigmentation. The characteristics of the foreleg and hind leg of Leucophaea, important for the experiments of group 2, are specified elsewhere (see p. 194). For characterization of the position of the supernumerary regenerates (anterior, posterior, etc.) I always refer to the organization of the stump tissues.

The results of all experiments are summarized in Table 1. Statements about the frequency of distinct cases, if not mentioned in the text, may be looked up there. An explanation of the half-schematic drawings of Figs. 2, 3, 6, 8 and 11 may be found in the legend of Fig. 2.

Fig. 2.

A–G. Exp. 1/I, right midleg tibia of Gromphadorhina (dark pigmentation) rotated 180° round the longitudinal axis and transplanted on to the left midleg tibia (light pigmentation). Two tarsi develop at the suture, which either are completely separated (A) arising anteriorly (a) and posteriorly (p), or are partly or wholly fused (B) and then have a dorsal position. As a rule the extra tarsi are mixed, i.e. composed of Gromphadorhina and Leucophaea tissues (A,.C, D, F, G); in one case (E) one tarsus consists only of Gromphadorhina tissues, the other only of Leucophaea tissues (the regenerates in this case exceptionally are arranged one behind the other, i.e. proximally (prox) and distally (dist) on the posterior surface of the tibia). In most cases the amounts of both tissues forming one pair of supernumerary regenerates are more or less equal (C, D, E), more rarely the Gromphadorhina tissues predominate (G); only in one case do Leucophaea tissues slightly predominate (F). A, Dorsal view; B, posterior view. In C–G, as in Figs. 3, 6, 8 and 11, the correlated regenerates of triple legs are put together by pairs, a (anterior), p (posterior), d (dorsal), and v (ventral) indicate from which surface of a triple leg the supernumerary regenerate arises. The names are always used with reference to the axial organization of the stump. The regenerates are delineated as if cut up longitudinally on the ventral surface and stretched out in a plain. The positions of the four surfaces of a leg are specified in Ca. Light parts: Leucophaea tissues; dotted: Gromphadorhina tissues; hatched (only in Figs. 3F, 8E and 11 A): segments lacking or borderline not identifiable.

Fig. 2.

A–G. Exp. 1/I, right midleg tibia of Gromphadorhina (dark pigmentation) rotated 180° round the longitudinal axis and transplanted on to the left midleg tibia (light pigmentation). Two tarsi develop at the suture, which either are completely separated (A) arising anteriorly (a) and posteriorly (p), or are partly or wholly fused (B) and then have a dorsal position. As a rule the extra tarsi are mixed, i.e. composed of Gromphadorhina and Leucophaea tissues (A,.C, D, F, G); in one case (E) one tarsus consists only of Gromphadorhina tissues, the other only of Leucophaea tissues (the regenerates in this case exceptionally are arranged one behind the other, i.e. proximally (prox) and distally (dist) on the posterior surface of the tibia). In most cases the amounts of both tissues forming one pair of supernumerary regenerates are more or less equal (C, D, E), more rarely the Gromphadorhina tissues predominate (G); only in one case do Leucophaea tissues slightly predominate (F). A, Dorsal view; B, posterior view. In C–G, as in Figs. 3, 6, 8 and 11, the correlated regenerates of triple legs are put together by pairs, a (anterior), p (posterior), d (dorsal), and v (ventral) indicate from which surface of a triple leg the supernumerary regenerate arises. The names are always used with reference to the axial organization of the stump. The regenerates are delineated as if cut up longitudinally on the ventral surface and stretched out in a plain. The positions of the four surfaces of a leg are specified in Ca. Light parts: Leucophaea tissues; dotted: Gromphadorhina tissues; hatched (only in Figs. 3F, 8E and 11 A): segments lacking or borderline not identifiable.

Fig. 3.

A–F. Exp. l/II, right midleg tibia of Gromphadorhina transplanted on to the left midleg tibia of Leucophaea without rotation. The two extra regenerates develop dorsally and ventrally (A, d and v), sometimes they fuse together and then lie posteriorly (B). All regenerates are mixed (A, C, D, F) with the exception of one pair of homogeneously built tarsi, one of which consists of Gromphadorhina, the other of Leucophaea tissues (E). Mostly the Gromphadorhina tissues predominate (C), more rarely the amounts of both tissues are of equal size (D, E) ; only in one case do the Leucophaea tissues predominate (F). A, Posterior view; B, ventral view.

Fig. 3.

A–F. Exp. l/II, right midleg tibia of Gromphadorhina transplanted on to the left midleg tibia of Leucophaea without rotation. The two extra regenerates develop dorsally and ventrally (A, d and v), sometimes they fuse together and then lie posteriorly (B). All regenerates are mixed (A, C, D, F) with the exception of one pair of homogeneously built tarsi, one of which consists of Gromphadorhina, the other of Leucophaea tissues (E). Mostly the Gromphadorhina tissues predominate (C), more rarely the amounts of both tissues are of equal size (D, E) ; only in one case do the Leucophaea tissues predominate (F). A, Posterior view; B, ventral view.

(1) Multiple formations of the tibia (group 7)

(a) General phenomenona

In combinations I and II all operated legs with surviving transplant had two supernumerary regenerates consisting of tarsus and the distal parts of the tibia. The position of these extra regenerates was in accordance with earlier results (Bohn 1965). The regenerates as a rule appeared at that point of the suture, where tissues of different symmetry properties met (Fig. 1), i.e. anterior and posterior in combination I (Fig. 2A), dorsal and ventral in combination II (Fig. 3A). They had the tendency to move together in a dorsal (comb. I) or posterior (comb. II) direction and even to fuse partly or in full length (Figs. 2B, 3B).

In Exp. l/III complete regenerates occurred rather rarely (Fig. 4C). More often there were more or less reduced tarsus-like formations (Fig. 4B); often only a tiny part of the joint region (combining tibia with tarsus) was formed (Fig. 4 A). In one case regenerates were completely absent. The reduced tendency to form regenerates in this combination is due to the fact that the disharmonie orientation may be harmonized by a subsequent movement of the transplant. In most cases the transplants have turned back to their original position by a rotation of 180° round the longitudinal axis (Fig. 4 A, B).

Fig. 4.

A–C. Exp. l/III, left midleg tibia of Gromphadorhina rotated 180° round the longitudinal axis and transplanted on to the left midleg tibia of Leucophaea. The rotated transplant mostly turns back to its normal position (A; only partly in B). Two fully developed regenerates appeared only in two cases (C); in the other cases the tarsi are incomplete and have fused together longitudinally (B) ; often only a socket and condyle (g in A) are present or extra regenerates are missing. The amounts of the two kinds of tissues forming the regenerates are, except for the leg shown in C, approximately of equal size. All legs in posterior view.

Fig. 4.

A–C. Exp. l/III, left midleg tibia of Gromphadorhina rotated 180° round the longitudinal axis and transplanted on to the left midleg tibia of Leucophaea. The rotated transplant mostly turns back to its normal position (A; only partly in B). Two fully developed regenerates appeared only in two cases (C); in the other cases the tarsi are incomplete and have fused together longitudinally (B) ; often only a socket and condyle (g in A) are present or extra regenerates are missing. The amounts of the two kinds of tissues forming the regenerates are, except for the leg shown in C, approximately of equal size. All legs in posterior view.

(b) The composition of the supernumerary regenerates

In each of the three combinations (I—III) one triple leg was found, the one tarsus of which was formed only by Gromphadorhina tissues, the other by Leucophaea tissues (Figs. 2E, 3E). In four cases with partly fused regenerates one of the distally free branches consisted only of Gromphadorhina tissues, the other being mixed (Fig. 4C). The supernumerary regenerates of all other animals were made up of a mixture of the tissues of both species.

Combination I. The borderline between the tissues of both species of cockroach, marked by the different pigmentation, is often approximately in the middle of the dorsal and ventral side of the tarsus (Fig. 2,Ca, Cp, Da, Fa). The anterior half of the anterior tarsus consisted of Leucophaea tissues, the posterior half of Gromphadorhina tissues. The other supernumerary tarsus of the same triple leg was composed inversely. Thus the two accessory tarsi in this respect were symmetric to each other (Fig. 2C). In other cases one of the borders was medio-ventral (Fig. 2,Gp) or medio-dorsal (Fig. 2,Ga), the second border, however, was more anterior or posterior. Rarely both borderlines deviated considerably from the median (Fig. 2 Dp).

The borderline was not always exactly parallel to the longitudinal axis of the tarsus. Especially at the posterior surface of the tarsus the border was often rather oblique (Fig. 2 Dp, Fp, Ga). Thus, the composition of the tarsus at the base could be different from that at the top.

Adding the portions of each kind of tissue participating in both supernumerary regenerates of a triple leg gives, as a rule, a relation of 1:1 (Fig. 2C, D, E). More rarely the Gromphadorhina tissues predominate considerably (Fig. 2G). in only one case did the Leucophaea tissues show a weak predominance (Fig. 2F).

Combination II As in comb. I, one of the borderlines was often approximately medio-dorsal or medio-ventral: the other borderline, however, was in most cases at the posterior surface of the tarsus (Fig. 3D). Thus, the amounts of the two different tissues forming a supernumerary regenerate were very unequal. But even if the portions from the two correlated regenerates are taken together the amounts of both tissues are rarely equal (Fig. 3 D, E). In most cases the Gromphadorhina tissues predominate (Fig. 3C); in only one case was there slightly more Leucophaea than Gromphadorhina tissues (Fig. 3F).

Combination III. In this combination only two triple legs occurred with fully developed extra tarsi. In one case one tarsus consisted completely of Gromphadorhina tissues, the second of Leucophaea tissues. In the other case (Fig. 4C) both tarsi fused together in segments one and two. The free distal segments of the one tarsus were built up only by Gromphadorhina tissues, the other tarsus was mixed, 1/3 consisting of Leucophaea tissues. The incomplete tarsi as a rule were composed half and half by tissues of Leucophaea and Gromphadorhina (Fig. 4B).

(2) Multiple formations at the level of the coxa (groups 2 and 3)

(a) General phenomena

The results are very similar to those of group 1, except the supernumerary regenerates which consist of tarsus, tibia, femur, and the distal half of the coxa (Figs. 5, 7, 9). The extra regenerates have an anterior and posterior (comb. I) or dorsal and ventral position (comb. II). In combination III there was often a re-rotation of the rotated transplant. But complete regenerates now appeared more often as compared with group 1.

Fig. 5.

A–E. Exp. 2/I, right foreleg coxa rotated 180° round the longitudinal axis and transplanted on to the left hind-leg coxa (Leucophaea only). A, General view of a triple leg; the mostly mixed extra regenerates lie anteriorly (aReg) and posteriorly (pReg, last tarsal segment missing); B-E, the three branches of the same triple leg separately shown. B, Foreleg transplant (Tra); C, the bend of the tibia and tarsus of the anterior regenerate is an indication for its mixed composition; D, E, posterior regenerate, the foreleg structure of the femur and the tibia (arrows) can clearly be recognized on the anterior surface (E). The posterior surface (D) consists of hind-leg tissues as indicated by the folds on the tibia and the first tarsal segment. D, Posterior view; in all other cases anterior view.

Fig. 5.

A–E. Exp. 2/I, right foreleg coxa rotated 180° round the longitudinal axis and transplanted on to the left hind-leg coxa (Leucophaea only). A, General view of a triple leg; the mostly mixed extra regenerates lie anteriorly (aReg) and posteriorly (pReg, last tarsal segment missing); B-E, the three branches of the same triple leg separately shown. B, Foreleg transplant (Tra); C, the bend of the tibia and tarsus of the anterior regenerate is an indication for its mixed composition; D, E, posterior regenerate, the foreleg structure of the femur and the tibia (arrows) can clearly be recognized on the anterior surface (E). The posterior surface (D) consists of hind-leg tissues as indicated by the folds on the tibia and the first tarsal segment. D, Posterior view; in all other cases anterior view.

Fig. 6.

A–D. Exp. 3/I, right foreleg or hind-leg coxa of Gromphadorhina rotated 180° round the longitudinal axis and transplanted on to the left hind-leg coxa of Leucophaea. As a rule both accessory regenerates are mixed (A, B, C, Da), only rarely is one of the supernumerary legs formed completely by Gromphadorhina tissues (Dp). The amounts of Gromphadorhina and Leucophaea tissues forming the extra regenerates of a triple leg are mostly of equal size (B, C), sometimes the Gromphadorhina tissues predominate (D). A, Dorsal view.

Fig. 6.

A–D. Exp. 3/I, right foreleg or hind-leg coxa of Gromphadorhina rotated 180° round the longitudinal axis and transplanted on to the left hind-leg coxa of Leucophaea. As a rule both accessory regenerates are mixed (A, B, C, Da), only rarely is one of the supernumerary legs formed completely by Gromphadorhina tissues (Dp). The amounts of Gromphadorhina and Leucophaea tissues forming the extra regenerates of a triple leg are mostly of equal size (B, C), sometimes the Gromphadorhina tissues predominate (D). A, Dorsal view.

Fig. 7.

A–E. Exp. 2/II, right foreleg coxa transplanted on to a left hind-leg coxa without rotation (Leucophaea only). A, General view of a triple leg; B-E, the three branches of the same triple leg separately shown. The extra regenerates develop ventrally (vReg) and dorsally (dReg) with respect to the position of the transplant (Tra). The dorsal regenerate in most cases is clearly mixed. The anterior surface (E) shows foreleg structures (arrows), the posterior surface (D) consists of hind-leg tissues as indicated by the folds on the tibia. The ventral regenerate (C) is more or less a pure foreleg. D, Posterior view; in all other cases anterior view.

Fig. 7.

A–E. Exp. 2/II, right foreleg coxa transplanted on to a left hind-leg coxa without rotation (Leucophaea only). A, General view of a triple leg; B-E, the three branches of the same triple leg separately shown. The extra regenerates develop ventrally (vReg) and dorsally (dReg) with respect to the position of the transplant (Tra). The dorsal regenerate in most cases is clearly mixed. The anterior surface (E) shows foreleg structures (arrows), the posterior surface (D) consists of hind-leg tissues as indicated by the folds on the tibia. The ventral regenerate (C) is more or less a pure foreleg. D, Posterior view; in all other cases anterior view.

Fig. 8.

A–F. Exp. 3/II, right foreleg or hind-leg coxa of Gromphadorhina transplanted on to the left hind-leg coxa of Leucophaea without rotation. Some of the supernumerary regenerates are partly mixed (A, B, C, Dd, Ed), some consist completely of Gromphadorhina tissues (Dv, Ev, F). The amounts of Gromphadorhina and Leucophaea tissues forming the regenerates are rarely of equal size (B), mostly the Gromphadorhina tissues predominate (A, C, D, E), in some cases even both regenerates are formed only by Gromphadorhina tissues (F). Av, Ventral view; Ad, dorsal view. The extra regenerates shown in B exceptionally lie anteriorly (a) and posteriorly (p).

Fig. 8.

A–F. Exp. 3/II, right foreleg or hind-leg coxa of Gromphadorhina transplanted on to the left hind-leg coxa of Leucophaea without rotation. Some of the supernumerary regenerates are partly mixed (A, B, C, Dd, Ed), some consist completely of Gromphadorhina tissues (Dv, Ev, F). The amounts of Gromphadorhina and Leucophaea tissues forming the regenerates are rarely of equal size (B), mostly the Gromphadorhina tissues predominate (A, C, D, E), in some cases even both regenerates are formed only by Gromphadorhina tissues (F). Av, Ventral view; Ad, dorsal view. The extra regenerates shown in B exceptionally lie anteriorly (a) and posteriorly (p).

Fig. 9.

A–D. Exp. 2/III, left foreleg coxa rotated 180° round the longitudinal axis and transplanted on to the left hind-leg coxa (Leucophaea only). A, General view of a triple leg; B–D, the three branches of the same triple leg separately shown. The rotated transplant mostly turns back to its normal position (Tra, A). Mostly one of the extra regenerates possesses hind-leg features (Z1 C), the other foreleg features (Z2, D). The two regenerates have different axial organization: the hind leg is a left leg, the foreleg a right leg. D, Posterior view; in all other cases anterior view.

Fig. 9.

A–D. Exp. 2/III, left foreleg coxa rotated 180° round the longitudinal axis and transplanted on to the left hind-leg coxa (Leucophaea only). A, General view of a triple leg; B–D, the three branches of the same triple leg separately shown. The rotated transplant mostly turns back to its normal position (Tra, A). Mostly one of the extra regenerates possesses hind-leg features (Z1 C), the other foreleg features (Z2, D). The two regenerates have different axial organization: the hind leg is a left leg, the foreleg a right leg. D, Posterior view; in all other cases anterior view.

The more complex structure of the accessory regenerates makes it possible to determine their axial organization. In the combinations I and II both regenerates are in accordance with the stump in this respect, i.e. if one transplants a part of a right leg on to a stump of a left leg two supernumerary left legs are formed. In contrast one left and one right leg is formed in combination III in which a left leg in reverse position is transplanted on to a stump of a leg of the same side of the body.

(b) The composition of the supernumerary regenerates

Combination I. In group 2 parts of a foreleg and hind leg of Leucophaea have been combined. So it is necessary to describe briefly the different structures of these legs. The foreleg differs from the hind leg in having a longitudinal row of relatively robust bristles at the antero-ventral edge of the femur. Besides this there are differences in the number and pattern of the bristles, especially at the anterior surface of the tibia. The tibia and the first tarsal segment in the foreleg are considerably shorter than in the hind leg (Fig. 5B). Naturally, these rather scanty differences do not allow absolutely clear identification of every part of a regenerate, all the more so as these structural characteristics often are only weakly developed in the regenerates. Nevertheless, they are sufficient for elucidating the approximate composition of at least a part of the supernumerary regenerates.

The supernumerary regenerate lying posteriorly is in nearly all cases (26 of 31) clearly mixed (Fig. 5D, E). The anterior surface of the regenerate consisted of tissues of the foreleg, indicated by the bristles on femur and tibia (Fig. 5E). The posterior surface was formed by tissues of the hind leg marked by the folds of the epidermis of the tibia and the first tarsal segment (Fig. 5D) and by the pronounced bending of these segments in a posterior direction (the convex side being posterior). Both phenomena are due to the different length of those segments in the foreleg and hind leg. The different length of both sides of the segments cannot be compensated by the bend alone; therefore the epidermis gets creased.

Likewise, the other extra regenerate lying anteriorly was often considerably bent, but in the opposite direction (the convex side being anterior, Fig. 5C). So it might be suggested that the tissues of the hind leg mainly participate in forming the anterior side of the leg, whereas the posterior side is built up by foreleg tissues. Since characteristic patterns are lacking at the leg’s posterior side, this composition is not completely certain. Therefore the corresponding regenerates are listed separately in Table 1.

The composition of the regenerates is much easier to identify, when the tissues of Gromphadorhina and Leucophaea are combined (group 3). The results confirm the conclusions of the preceding group of experiments. The extra regenerates consist approximately half-and-half of Gromphadorhina and Leucophaea tissues. The anterior half of the posterior leg and the posterior half of the anterior leg is built up by Gromphadorhina tissues, the remaining parts consist of Leucophaea tissues. In most cases the border between the two tissues lies approximately medio-dorsal and/or medio-ventral (Fig. 6A, Ba, Bp, Da; Cp partly) sometimes posterior (Fig. 6,Ca ; Cp partly). The ratio of the two kinds of tissues forming the two correlated regenerates is nearly 1:1 (Fig. 6B, C) ; but in some cases the Gromphadorhina tissues predominate considerably. Examples of the last case are those two legs, the posterior regenerates of which are completely built up by Gromphadorhina tissues (Fig. 6 Dp). Pure Leucophaea regenerates were never observed.

Combination II. Group 2. In contrast to the preceding combination the supernumerary regenerates mostly lie dor sally and ventrally (Fig. 7A). In a great number of cases especially the dorsal regenerates clearly show a mixed composition, the anterior side having foreleg structures, the posterior surface those of the hind leg (Fig. 7D, E). The ventral regenerate in most cases seems to consist uniformly of foreleg material, but minimal participation of hind-leg material cannot be excluded. In some cases the mixed composition is certain, but the stripe with hind-leg structures on the anterior or ventral surface of the leg is always very small. One animal has two uniformly structured regenerates: the ventral regenerate is a foreleg, the dorsal regenerate a hind leg.

Group 3. Most accessory regenerates are mixed, but a certain percentage is built up completely by Gromphadorhina tissues. In five cases the ventral regenerate is homogeneous, whereas the corresponding dorsal regenerate is mixed (Fig. 8D, E); in three cases even both extra regenerates are completely built up by Gromphadorhina tissues (Fig. 8F). The position of the border between the two tissues in the mixed regenerates is not chiefly median as in combination I, but seems to be regularly distributed to all four sides. The amounts of transplant and host tissues rarely are equal (Fig. 8B); in most cases the Gromphadorhina tissues predominate (Fig. 8 A, C, D, E, F).

Combination III. Group 2. The position of the supernumerary regenerates, provided they are formed at all, is not as constant as in the preceding combinations. Very often the stump seems to pass on directly into one of the regenerates (Fig. 9 A, Z1). The coxa of the transplant mostly arises near the ventral face of the coxa of the stump, from which it is separated by a slight constriction. The second regenerate (Z2) arises directly from or at least near the coxa of the transplant. Z1 is a left leg and as a rule only consists of hind-leg tissues (Fig. 9C); Z2 is a homogeneous right foreleg (Fig. 9D). In some cases, however, one of the regenerates clearly shows mixed features; in some other cases the mixed structure seems rather likely, but cannot be proved with complete certainty.

As in all former experiments multiple formations with more than two accessory regenerates were never observed. The multiple formation shown in Fig. 10 A, B seems to be an exception to this rule. It is composed of two branches.

Fig. 10.

A, B. The only case with apparently more than two supernumerary regenerates (Exp. 2/III). Z2 extra regenerate with predominant foreleg features. The second extra regenerate has fused with the transplant (Tra + Z1). The three light stripes on the femur (arrows) suggest that at least the ventral parts of a third extra regenerate had participated in this leg. A, Posterior view; B, anterior view.

Fig. 10.

A, B. The only case with apparently more than two supernumerary regenerates (Exp. 2/III). Z2 extra regenerate with predominant foreleg features. The second extra regenerate has fused with the transplant (Tra + Z1). The three light stripes on the femur (arrows) suggest that at least the ventral parts of a third extra regenerate had participated in this leg. A, Posterior view; B, anterior view.

Fig. 11.

A, B. Exp. 3/III, left foreleg coxa of Gromphadorhina rotated 180° round the longitudinal axis and transplanted on to the left hind-leg coxa of Leucophaea. One of the two supernumerary regenerates is a right leg (r), the other is a left leg (l). The left leg always was mixed (Al, Bl); the right leg was mixed (Ar) or consisted only of Gromphadorhina tissues (Br).

Fig. 11.

A, B. Exp. 3/III, left foreleg coxa of Gromphadorhina rotated 180° round the longitudinal axis and transplanted on to the left hind-leg coxa of Leucophaea. One of the two supernumerary regenerates is a right leg (r), the other is a left leg (l). The left leg always was mixed (Al, Bl); the right leg was mixed (Ar) or consisted only of Gromphadorhina tissues (Br).

The ventral branch is a right leg with mainly foreleg structures (Z2). The second branch is a compound formation, which is the result of the fusion of more than two legs, because there are three ventral parts of a femur (Tra + Z1). The development of this multiple formation may explain its composition. Originally (after the second postoperative moult) it consisted as usual of three completely separated branches, i.e. the transplant and two accessory regenerates. By an accident this leg was damaged rather severely, as a result of which large parts of the transplant and of one extra regenerate were destroyed. In the course of the succeeding regeneration the abnormal formation described above developed out of the wound area. So, the compound formation consists of the transplant, a supernumerary regenerate, and the ventral parts of an additional supernumerary leg. This third supernumerary regenerate is not formed by the original host–transplant combination, but by a secondary wound. In a similar way it is possible to get a supernumerary although incomplete branch by wounding a normal leg (Bohn, 1965).

Group 3. Only seven animals had well-developed extra regenerates, which in most cases are mixed. Some of the regenerates forming right legs are homogeneously built up by Gromphadorhina tissues. Regenerates consisting only of Leuco-phaea tissues could not be observed. The amounts of the two tissues forming a pair of extra regenerates are either nearly equal (1 case, Fig. 11 A) or the Gromphadorhina tissues predominate (3 cases, Fig. 11B). Because of the small number of triple legs with completely developed supernumerary regenerates nothing generally valid can be said about the position of the border between the two tissues.

Summary of the results

(a) The composition of the supernumerary regenerates

In combination I the results of the three groups of experiments are more or less uniform. In nearly all cases the supernumerary regenerates are mixed. Regenerates consisting only of Gromphadorhina or Leucophaea tissues are exceptions. In group 2 homogeneously formed regenerates were not observed ; but this does not mean that such regenerates cannot occur at all in this group, for it cannot be excluded that one or the other of the 41 regenerates classified as not unequivocally identifiable consisted nevertheless of only one sort of tissues.

The borderline between stump and transplant tissues mostly lay mediodorsally and medio-ventrally; so the anterior longitudinal half of a regenerate consisted of one, the posterior half of the other sort of tissue. The amounts of both sorts of tissues forming the two correlated accessory regenerates were approximately equal; the transplant (= Gromphadorhina) tissues predominated rather rarely.

In combination II, too, the mixed regenerates are the most numerous, especially in group 1. In group 3 a considerable part of the extra regenerates was not mixed. The position of the borderline was preferably dorsal and posterior.

In contrast to comb. I a distinct disproportion could be observed between the tissues of Gromphadorhina and Leucophaea, not only in a single regenerate, but also in almost every pair of supernumerary regenerates. The Gromphador-hina tissues predominated in most cases, in some cases even quite considerably.

In combination III the great number of homogeneously formed regenerates of Exp. 2/III is remarkable, whereas the regenerates of Exp. 3/III predominantly are mixed. On the whole, the Gromphadorhina tissues provide the main part of the material for the supernumerary regenerates.

(b) Number, position and symmetry of the supernumerary regenerates

After heteropleural transplantation (combinations I and II) two extra regenerates are formed invariably, if the transplant survived. They arise from those regions of the suture between stump and transplant, where tissues of different symmetry properties meet together. In combination I this situation is anterior and posterior, in combination II dorsal and ventral. Both regenerates have the same axial structure as the stump, i.e. left legs are regenerated on a left side stump. The supernumerary regenerates, especially those arising from the tibia, may move together (in dorsal (I) or posterior (II) direction) and fuse partly or in full length. This fusion is not accompanied by any regression of the fused regenerates.

After homopleural transplantation (comb. Ill) the rotated transplant very often turns back to its normal position. At most two extra regenerates are formed, which often have a tendency to fuse. In contrast to the heteropleural transplantation this fusion is accompanied by a regression of the regenerates. In the extreme, only traces of the distal following joint region are to be found (in group 1, for instance, the socket and condyle of the tibio-tarsal joint, in groups 2 and 3 traces of the coxa-trochanter joint). Even though rudimental the regenerated structures are always developed twice, i.e. symmetrically. Some of the animals had no regenerates at all; in those cases the transplant had always turned back to its normal position. The two extra regenerates have different axial structures: one is in accordance with the stump, the other with a leg of the opposite body side. Thus one left and one right leg is formed in all cases.

(1) The composition of the supernumerary regenerates

The experiments described above attempt to elucidate the origin of the supernumerary regenerates by using leg tissues with different structures as stump and transplant. Regenerate parts having the same structure as stump or transplant tissues were assumed to arise from the corresponding stump and transplant tissues. This is only permissible if the tissues do not lose their special features during regeneration, which is accompanied by dedifferentiation and cell division. In a recent paper (Bohn, 1971) it has been shown that the features used (differences in pigmentation between Gromphadorhina and Leucophaea; structural differences between foreleg and hind-leg tissues) are firmly determined in this respect and yet can manifest themselves autonomously in a foreign environment.

The experiments show that both possibilities of composition of the extra regenerates occur : in each group there are legs with mixed accessory regenerates as well as uniform regenerates. It is remarkable that the Gromphadorhina tissues, i.e. the transplant tissues, predominate in the regenerates of combinations I and II. This raises the question whether the rules found in the combinations of Gromphadorhina and Leucophaea tissues prove correct in intraspecific combinations, too. The results of Exp. 2/II show that there are, at least in this respect, no great differences between interspecific and intraspecific transplantations. For in Exp. 2/II again usually the transplant tissues seem to predominate: the dorsal regenerate was clearly mixed; the ventral regenerate seemed to be uniformly built up by foreleg tissues or contained only small amounts of hind-leg tissues. Thus, the accessory regenerates are composed of transplant and stump tissues in the ratio of about 2 (or more) : 1.

In comparing Exps. 2/III and 3/III there are beyond question certain differences between intraspecific and interspecific transplantations. In the first experiment uniformly composed regenerates predominate, in the latter mixed. But this difference is only quantitative and without importance with respect to our further conclusions.

In the following our results relating to the composition of the extra regenerates are compared with the results of other authors.

The experiments of Bodenstein (1937) with legs of Lepidoptera (JPyrameis cardui) are comparable with combination III. He transplanted parts of the first or second leg on to a stump of the third leg. One of the supernumerary regenerates consisted of host tissues, the other of transplant tissues. His results are partly in accordance with ours in so far as just in this combination (for instance in Exp. 2/III) often triple legs occur having one transplant and one stump regenerate; mixed regenerates formed considerably more rarely. Since Bodenstein had only four cases with well-developed triple legs, the occurrence of mixed regenerates cannot be excluded in Lepidoptera.

Bullière (1970), in his experiments, used differently structured legs of the same species (Blabera craniifer). But the clearly identifiable structures are limited to a few parts of the leg only, and are not sufficient to allow identification of mixed regenerates in every case (see p. 194). The mere presence of the specific foreleg structures, which are restricted only to the anterior surface of the leg, does not suffice to take such a leg for a pure foreleg; conversely, the absence of these structures does not mean that the leg is a hind leg. Bullière concludes from his experiments that one regenerate consists of host and the other of transplant tissues. This conclusion, in my opinion, is only correct to the extent that triple legs may occur in Blabera in the same way as in Leucophaea, which are built up in this way. But the figures presented by Bullière argue against his own opinion. In Fig. 3B, C, D (Bullière, 1970, p. 344) both supernumerary regenerates show the same bend as our mixed regenerates and thus make it likely that they are composed of both, foreleg and hind-leg tissues.

The same objections may be raised to the experiments of Lheureux (1971). The legs used as stump and transplant (pedipalps respectively hind legs of the spider Tegeneria saeva) are distinguished more or less only by their distal ends having two or three claws respectively. This is not enough to allow clear identification of mixed structures like those presented in Fig. 8 Ad and 8Cv. So it remains undecided, whether or not mixed extra regenerates may be found in spiders.

The results of Bart (1971 b) are in good agreement with ours. Bart did not transplant whole legs, but only sectors thereof, and in most cases got only one accessory leg. This regenerate was clearly mixed at least in those cases or parts that allowed unequivocal identification. Naturally, Bart could not show an exact borderline, since he combined the parts of legs of the same species (Carausius morosus). But he supposes that the adjacent tissues of the stump and transplant extend into the supernumerary regenerate; thus, each forms one longitudinal half of the regenerate. Our experiments have shown that this supposition is correct as far as concerns comb. I, but in the other combinations (II and III) this situation is only exceptionally realized. Besides such clearly mixed regenerates Bart also gets regenerates which seem to consist only of host tissues. This assumption could be confirmed by interspecific transplantation experiments (A. Bart, personal communication). Most curiously, there are never supernumerary regenerates, which consist only of transplant tissues. In phasmids, contrary to the case in cockroaches, the host tissues seem to predominate.

(2) The number, position and axial structure of the supernumerary regenerates

Most authors (Bodenstein, 1937, 1941 ; Bohn, 1965; Bullière, 1970; Lheureux, 1970, 1971) agree with respect to number, position and orientation of the accessory regenerates. The re-rotation of rotated transplants in comb. Ill, previously reported in Leucophaea (Bohn, 1965) has also been established for Blabera (Bullière, 1970). In Carausius on the contrary there seems to be no re-rotation (Bart, 1971 a). Bart is the only one who gets three supernumerary regenerates in several cases after homopleural transplantation. In our experiments there was only one similar case (Fig. 10 A, B). But the development of this multiple leg clearly indicates that originally there had been only two supernumerary regenerates. The third incomplete branch has developed owing to an untypical wounding. Likewise Furukawa (1940) finds a leg with three accessory regenerates as an exception among his multiple legs. In the only figure Bart presents of such a quadruple leg, two of the three supernumerary regenerates have fused together at their bases and are only separated at their distal ends. It is possible that secondary effects have led to these formations, and in the legend of this figure Bart does say that there has been a partial necrosis of the axial regenerate. Nevertheless, it cannot be excluded that Carausius behaves quite differently in this respect from all other arthropods, especially since Bart did not observe any re-rotation of the rotated transplant.

(3) The causes for the development of supernumerary regenerates

The extra regenerates always appear after disharmonie orientation of transplants. Remarkably they always develop at that site where different symmetry properties of stump and transplant (anterior-posterior, dorsal-ventral) meet. Bart (1971 a) therefore concludes that in those places ‘morphogenetic centres’ are established which give rise to the development of regeneration blastemas. This hypothesis seems plausible as far as combinations I and II are concerned. According to Bart’s hypothesis, however, one should get four extra regenerates in the experiments of combination III, since there are disharmonie situations at all four sides of the leg. But Bart never gets more than three accessory regenerates; mostly (75%) he only gets two. Other insects never show more than two accessory regenerates.

Undoubtedly the disharmonie orientation is of great importance for the induction of regenerative processes (Bodenstein, 1937; Furukawa, 1940; limbs of Amphibia see Mangold, 1929); but it seems questionable to reduce the problem of the origin of the extra regenerates to the simple formula: regenerates always develop in accordance with position and number of sites with opposite orientation. Some of Bart’s own experiments argue against such an interpretation. For instance, he gets supernumerary regenerates, when he combines a dorsal part of the scape with a posterior part of the coxa.

In my opinion Bart uses the different qualities anterior, posterior, etc. much too formally. In fact, these names only represent a topographic characterization. Beyond this they may also imply that the surface of an appendage is asymmetrically organized. But instead of distinguishing only four different qualities one could arrange the cross-section into a two-dimensional coordinate system, the axes of which coincide with the antero-posterior and dorso-ventral axis. In this way it becomes much more evident that in reality every cell of a cut surface is different from any other (according to the position within the system), even though lying within the same sector of the section, i.e. anteriorly, posteriorly, etc. The sites bearing the extra regenerates in combinations I and II are thus sites in which these differences are greatest. In the combination III the degree of disharmonie orientation is equal at every point. Thus, contrary to Bart’s opinion there are more than four regions with disharmonie orientation.

The correspondence of the number of accessory regenerates with the number of existing cut surfaces is striking. This fact is in good agreement with the hypothesis of Przibram (1921) about the formation of the ‘Bruchdreifachbildungen’. According to this hypothesis each cut surface forms one regenerate: a normal distal regenerate sprouts from the amputation surface of the stump and a so-called proximal regenerate form the proximal wound surface of the translant. The axial organization of the extra regenerates is in good agreement with this assumption. Taking a left leg as host the supernumerary regenerates formed in combinations I and II are two left legs, but one left and one right leg is formed in combination III. Naturally, the conclusions of Przibram’s hypothesis are not valid with respect to the material used for the regenerates because it may happen that both supernumerary regenerates are built up only by material of the transplant. In modifying the hypothesis of the ‘Bruchdreifachbildung’ one could say: by combining two cut surfaces with divergent orientations each cut surface is stimulated to induce or organize one regenerate, the material for which may come partly or as a whole from the other amputation surface.

Nevertheless, this conception is not completely satisfying because, firstly, the phenomenon of the regular arrangement of the regenerates still rests unsolved unless it is assumed that the points of greatest disharmony are favoured above others in developing blastemas; secondly, it seems possible to get complete regenerates even if only a part of a cross-section had been exposed by a wound (Bart, 1971 a, b). Further experiments must be done before these seemingly contradictory phenomena are resolved under a single hypothesis which accounts for them all.

(4) The determination of symmetry properties

An important problem already discussed by Bart (1971 b) is that of the determination of the symmetry properties. Are cells of a distinct side (for instance anterior) able to transform into or to proliferate cells of another side (for example posterior) ? The experiments with partly severed legs done so far (Bohn, 1965; Bart, 1970) argue against such a transformation, for the distal regenerates arising from such wounds only possess those properties that had been exposed by the wound.

Such a transformation cannot yet be concluded, as Bart (1971 b) does, by the fact that there are triple legs with one extra branch consisting entirely of stump tissues and the other branches built up entirely by transplant tissues. Naturally, a supernumerary regenerate develops within a restricted area at one surface of the wound. But this does not mean that only cells of this restricted region had participated in forming the regenerate. Cells from all regions of a cross-section could have assembled at that position. Thus, it is not necessary to assume transformation of symmetry properties in order to allow the development of complete regenerates. Some of Bart’s own experiments are far more more convincing. He implanted a piece of tissue taken from one surface of a leg into the opposite surface and got a complete regenerate. If the wound areas of each, the transplant and the implantation zone, really included only a quarter of a complete cut surface it must be concluded that it is possible to get a complete regenerate with all symmetry properties from a wound containing only two of four existing sides. But, nevertheless, some doubt remains. It cannot be excluded that there had been some necrosis following as well as deeper wounds during operation. By this cells of the other symmetry properties could have been exposed and could have participated in regeneration.

The experiments with Leucophaea and Gromphadorhina show that transformations of symmetry properties in fact occur much more frequently than has been supposed. What would the supernumerary regenerates look like if such transformations were not possible? The amounts of the Leucophaea tissues of both extra regenerates of one triple leg taken together should make up just one complete cross-section. Apart from numerous cases in which the Gromphadorhina tissues clearly predominate (Figs. 2G, 3C, 6D, 8E) or even occasionally yield all of the material for both regenerates (Fig. 8F), this cannot even be found exactly in those cases where the relation of both tissues is approximately 1:1. In Fig. 2D, for instance, the postero-ventral part of the first tarsal segment of the accessory tarsi consists of Gromphadorhina tissues, in Fig. 2F the anterior claw and anterior part of the arolium is represented twice by Leucophaea tissues, etc. Thus, it must be concluded that regenerates including more than one crosssection, even two in the extreme, may develop from a single cross-section.

There is a further complication as far as the supernumerary regenerates of combination III is concerned. One left and one right leg is formed, the former corresponding to the organization of the stump’s amputation surface, the latter by regeneration from the transplant’s amputation surface in a proximal direction. There are several cases with mixed extra regenerates. This means that tissues originally representing parts of a right leg have been fitted into a left leg, and the reverse. This insertion apparently happened ‘ortsrichtig’ since such mixed legs do not show any anomalies in their patterns. So, in addition to or in connexion with a change of the symmetry properties there is a second change: the axial organization apparently is not yet fixed irreversibly and may be fitted to the conditions obtaining at the time.

Change or transformation of the symmetry properties or of the axial organization does not mean, for example, that a greater part of dorsal tibia tissues may be transformed to ventral tissues as a whole. A great number of experiments, for example the formation of triple legs after disharmonie orientation, argues against such a possibility. It is much more likely that the missing parts will be added to the edges of an incomplete cross-section by proliferation, i.e. may be a kind of lateral regeneration. In this way there could be a doubling of crosssections without any cell changing its determination. Therefore, I would prefer to speak of completion of missing symmetry properties instead of using the term transformation. Thus it remains open whether this completion is attained by an extensive transformation of symmetry properties or not.

According to the results of our experiments it seems to be fairly certain that completion of symmetry properties does occur. Nevertheless, some doubt remains. For simplicity, consider one extreme where two complete regenerates have been formed by a single cross-section. This result could have been achieved by a splitting of the blastema perpendicular to the longitudinal axis, by a kind of cross-section. Such a process does not seem to be very likely, but cannot be excluded. These doubts could be settled by transplanting parts of a cross-section interspecifically. The formation of a complete regenerate by such a sector including only a part of a whole cross-section irrevocably would demonstrate the ability of the leg tissues to complete missing symmetry properties.

(5) Cell lineage during regeneration

The organization of the mixed regenerates may help to clarify the problem of cell lineage and growth in the regenerating leg. As indicated above, the borderline between the two different tissues almost always runs in a longitudinal direction. This means that, after the establishment of the blastema, either there had been only longitudinal divisions or that if divisions in the transversal direction had occurred at all they must have occurred to the same extent at every level of the longitudinal axis. But the same holds true even for the establishment of the blastema itself. The completion of missing parts of a crosssection as referred to above is only imaginable if there are divisions perpendicular to the longitudinal axis of the leg, unless, as seems unlikely, rather large areas of tissues may change their side qualities as a whole. So, divisions must have occurred in such a way that sectors with strongly radial borders had formed in a presumably dome-like blastema.

On the other hand, the borderline is not always exactly parallel to the longitudinal axis. It has been mentioned above (see p. 194) that there are legs with slightly but distinctly oblique border lines (Figs. 2Dp, Fp, Ga; 3Cv; 8 Cv); narrow stripes of foreign tissues may even taper away (Figs. 3Cv, Fd; 8 Ad, Cv). These cases are a further indication of the fact that cells are not yet determined irreversibly with respect to a distinct position within the cross-section. A cell or at least the descendants of a cell may alter this position.

Studies on cell lineage and related problems have been made recently by Bryant (Bryant & Schneiderman, 1969; Bryant, 1970) in imaginai wing and leg discs of Drosophila. Their results, especially in the leg discs, are quite similar to ours. The clonally related cells produced by X-ray-induced somatic crossing over become arranged in longitudinal stripes. But the stripes are not always strictly parallel to the longitudinal axis nor of equal width at every level; moreover there are partly overlapping patches. This means that the cells of an imaginai leg disc of Drosophila, like those in larval legs of Leucophaea and Gromphadorhina, are not yet determined, at least until the end of the second instar, with respect to their position within the cross-section.

I thank Dr D. A. Ede for polishing the style of the English translation and Dr P. A. Lawrence for helpful criticism.

Balazuc
,
J.
(
1948
).
La tératologie des coléoptères et expériences de transplantation sur Tenebrio molitor L
.
Mém. Mus. natn. Hist, nat
.,
Paris
(Sér.)
25
,
1
193
.
Bart
,
A.
(
1970
).
Blessures et transformations trochanteriénnes chez Carausius morosas Br
.
Annls Embryol. Morph
.
3
,
379
398
.
Bart
,
A.
(
1971a
).
Morphogenèse surnuméraire au niveau de la patte du phasme Carausius morosus Br
.
Wilhelm Roux Arch. EntwMech
.
166
,
331
364
.
Bart
,
A.
(
1971b
).
Modalités de formation et de développement d’un centre morphogénétique surnuméraire chez Carausius morosus Br
.
Wilhelm Roux Arch. EntwMech. Org
.
168
,
97
124
.
Bateson
,
W.
(
1894
).
Materials for the Study of Variation
.
London
.
Bodenstein
,
D.
(
1937
).
Beintransplantationen an Lepidopterenraupen. IV. Zur Analyse experimenten erzeugter Bein-Mehrfachbildungen
.
Wilhelm Roux Arch. EntwMech. Org
.
136
,
745
785
.
Bodenstein
,
D.
(
1941
).
Investigations on the problem of metamorphosis. VIII. Studies on leg determination in insects
.
J. exp. Zool
.
87
,
31
53
.
Bohn
,
H.
(
1965
).
Analyse der Regenerationsfähigkeit der Insektenextremität durch Amputations-und Transplantationsversuche an Larven der Afrikanischen Schabe Leucophaea maderae Fabr. (Blattaria). II. Achsendetermination
.
Wilhelm Roux Arch. EntwMech. Org
.
156
,
449
503
.
Bohn
,
H.
(
1971
).
Interkalare Regeneration und segmentale Gradienten bei den Extremitaten von Leucophaea-Larven (Blattaria). III. Die Herkunft des interkalaren Regenerates
.
Wilhelm Roux Arch. EntwMech. Org
.
167
,
209
221
.
Bryant
,
P. J.
(
1970
).
Cell lineage relationships in the imaginai wing disc of Drosophila melanogaster
.
Devi Biol
.
22
,
389
411
.
Bryant
,
P. J.
&
Schneiderman
,
H. A.
(
1969
).
Cell lineage, growth, and determination in the imaginai leg discs of Drosophila melanogaster
.
Devi Biol
.
20
,
263
290
.
Bullière
,
D.
(
1970
).
Interpretation des régénérais multiples chez les Insectes
.
J. Embryol. exp. Morph
.
23
,
337
357
.
Furukawa
,
H.
(
1937
).
Reduplication experimentally produced in the earwig, Anisolabis maritima
.
Kontyû
11
,
126
129
.
Furukawa
,
H.
(
1940
).
Transplantation-experiments on appendages of Anisolabis maritima (Dermaptera). III. Transplantation of forceps
.
Jap. J. Zool
.
8
,
510
535
.
Hoarau
,
F.
(
1969
).
La régulation dans les territoires de régénération chez ITsopode terrestre Helleria brevicornis Ebner
.
Annis Embryol. Morph
.
2
,
87
103
.
Lheureux
,
E.
(
1970
).
Etude de schizomélies expérimentales de l’appendice locomoteur chez l’araignée Tegenenaria saeva Bl
.
C. r. hebd. Séanc. Acad. Soi
.,
Paris
, Sér. D,
270
,
709
711
.
Lheureux
,
E.
(
1971
).
Recherches expérimentales sur la participation du greffon et du portegreffe dans la morphogenèse d’appendices surruméraires chez l’Araignée Tegenaria saeva Bl
.
C. r. hebd. Séanc. Acad. Sci
.,
Paris
, Sér. D,
272
,
3299
3302
.
Mangold
,
O.
(
1929
).
Das Determinationsproblem. IL Die paarigen Extremitaten der Wirbeltiere in der Entwicklung
.
Ergebn. Biol
.
5
,
290
404
.
Noulin
,
G.
(
1970
).
Conditions de la différenciation de péréiopodes surnuméraires uniques ou multiples chez l’Oniscoïde Por celtio dilatatus Brandt
.
C. r. hebd. Séanc. Acad. Sci
.,
Paris, Sér. D
,
270
,
2305
2307
.
Penzlin
,
H.
(
1965
).
Experimentelle Erzeugung von Dreifachbildungen bei einem heterometabolen Insekt
.
Naturwissenschaften
52
,
217
.
Przibram
,
H.
(
1921
).
Die Bruchdreifachbildung im Tierreiche
.
Wilhelm Roux Arch. EntwMech. Org
.
48
,
205
444
.