The Influence of Peripheral Connexion on the Diameter of Regenerating Nerve Fibres

Nerve fibres subserving different functions have different diameters, not only in a normal nerve but also in one which has been crushed and has undergone degeneration and regeneration (Gutmann & Sanders, 1943). Therefore, there must be factors operating during the regeneration of any particular nerve which tend to restore the fibre size pattern typical of that nerve.

Simpson & Young (1945) have shown by cross union of visceral and somatic nerves containing fibres of different diameters that the central stump exerts an influence in determining the size of fibres regenerated. Thus after union of the central stump of a somatic nerve with the postganglionic trunk of the anterior mesenteric nerve, the latter comes to contain many medullated fibres, whereas normally this nerve contains very few medullated fibres in the rabbit.

However, there is also evidence of a strong peripheral influence upon the size attained by regenerating fibres. When a mixed nerve is severed and its ends united, there is such confusion of fibres at the line of suture that if central influences alone controlled the diameter of regenerated fibres, then all branches of the nerve below the suture would be expected to regenerate fibres of similar sizes. This in fact is not the case. Nageotte & Guyon (1918) and Sanders & Young (1944) have shown that below a suture in a mixed nerve the muscular branches come to contain larger fibres than the cutaneous. After degeneration the cutaneous branches contain smaller Schwann tubes, and it has been suggested that it might be the restrictive influence of these which produced the difference (Gutmann & Sanders, 1943 ; Sanders & Young, 1944). Such a restrictive influence operates to some extent, but further experiments have shown that the most powerful influence lies not in the peripheral stump itself, but in connexion with the end-organ (Sanders & Young, 1945 ; Simpson & Young, 1945). Fibres which become connected with a muscle become larger than those which end blindly. Weiss &.Taylor (1944) found evidence of a similar effect in the course of experiments in which they completed the disproof of the theory of ‘neurotropism’. More recently, Weiss, Edds & Cavanaugh (1945) have extended this work and shown in various nerves of rats that a stretch of nerve regenerating without opportunity for its fibres to reach the periphery becomes filled with many and small nerve fibres.

The present work contains an investigation of the numbers and sizes of fibres found in nerves regenerating with and without normal peripheral connexions. Since fibres with no peripheral connexions are shorter than those regenerating from the same level all the way to the periphery, a control investigation has been made of the effect of length of nerve on the diameter reached during regeneration. Finally, in order to discover whether the influence of peripheral connexion can override that of size of peripheral Schwann tubes, we have also made a study of regeneration through nerve grafts which originally contained fibres of different sizes. This serves also to answer a question of considerable practical importance to the surgeon who wishes to use pieces of cutaneous nerve for grafting into defects in motor nerve trunks (see Sanders & Young, 1944; Hammond & Hinsey, 1945).

The difficulty of this, as of many other investigations of nervous regeneration, has been to obtain similar lesions of the nerves in successive experiments. If two nerves are cut and reunited in such a way that the interval between the ends differs, then the size of the fibres regenerated in the two cases will also differ (Simpson & Young, 1945; Gutmann & Sanders, 1943). We have overcome this difficulty by interrupting the axons by crushing them with narrow smooth-tipped watchmaker’s forceps. If carefully performed, this operation interrupts the axons but leaves intact the connective tissue tubes in which they run, thus leaving optimal and standard opportunities for regeneration.

The peroneal nerves of rabbits were crushed in this way in the thigh, and then severed lower down, the ends being so treated as to provide different conditions for the reception of the regenerating fibres. The details of the various experiments are given below. All unions of the nerves were made by means of concentrated cockerel plasma, for the supply of which we are indebted to Mr P. B. Medawar.

For measurement of the regenerating fibres the nerves were fixed in Flemming’s fluid,^* sectioned transversely at 5μ after paraffin embedding, and stained with a modified Weigert method (Gutmann & Sanders, 1943). This did not always prove satisfactory, since the fixative in some cases failed to penetrate adequately to the centre of the nerve. The fibres were counted and measured on negative photographs made by direct projection on bromide paper at la magnification of 750 ×. In some cases the whole nerve was photographed, while in others sample strips across the nerve were used. After photographing, a number of sample areas were selected for counting by marking out squares 50 × 50 mm. on the finished photograph, usually in a ring round the periphery of the nerve, where fixation, by ordinary histological criteria, was deemed to be adequate. The area of the whole nerve was also determined in each case by planimetry either of the photograph of the whole nerve or of a projection. The fibres in each sample square were then counted and marked off in size-groups with an interval of 2μ. The fibres of various sizes are not scattered quite at random throughout the nerve but sufficient squares were taken to provide an adequate total sample of the nerve. Except in special cases, mentioned elsewhere in the text, 25–30 squares were measured in the case of each nerve, an area equivalent to rather more than a quarter of the total area of the nerve. From the total number of fibres in the sample measured and the total nerve area obtained by planimetry an estimate of the total population could be obtained. The errors involved in such counts are discussed by Gutmann & Sanders (1943) and Simpson & Young (1945). In the present experiments nerves of the opposite sides of individual animals were compared, and since the technique was kept constant, errors of absolute estimation were unimportant.

In quantitative investigations of the present kind it is important to use a measure which shall express adequately the constitution of each individual nerve examined. The average fibre diameter is not very suitable for this purpose. Since the curve of distribution of fibre sizes.departs quite widely from the form of the ‘normal curve’, being in many cases quite strongly bimodal, the average fibre diameter does not typify the diameter of the greatest number of fibres in the nerve. Moreover, there is such a numerical preponderance of small myelinated fibres in most nerves that nerves which differ strikingly in fibre constitution (e.g. muscular and cutaneous bundles) differ only slightly in average fibre diameter. Accordingly, we have in this paper used the root mean square of the whole population of diameters as our estimate of ‘mean’ fibre diameter. This figure (D) was calculated separately for each square of the sample counted, and the value of D given for each nerve was the mean of all the D’s found for separate squares within that nerve. This method of calculating D has the advantage that an estimate of error can be made in each case, and the values of D obtained for different nerves compared statistically. This estimate of the ‘mean’ diameter has an additional advantage in that it is directly related to, the total volume of nerve fibres present in the nerve. If the nerve were wholly composed of fibres of diameter D it would have a volume of fibres per unit length equal to that actually found in the nerve, but there spread over fibres of various diameters.

The object of the experiments described in this section was to discover whether the diameter reached by the fibres at any level in a regenerating nerve is affected by the length of fibre regenerating peripheral to that level. Extension of a nerve beyond its normal length can be achieved either by the use of a graft or by joining it with another nerve. Both methods have considerable disadvantages. The difficulty of obtaining suitable lengths of nerve for autografting and the inevitable variability in the resulting junctions suggested that the method of cross union might be more suitable. The peroneal nerves were therefore crushed as high as possible on both sides of each animal, cut lower down, and joined to the tibial nerves in such a way as to make on one side a nerve of normal length, and on the other one in which fibres have to grow much farther than is usual.

This was done by freeing the peroneal nerve along its whole length, cutting it peripherally, turning it back along itself and joining it to the tibial high in the thigh. The fibres regenerating from the crushed points thus have to pass through a much shorter length of peroneal nerve on the one side before entering the tibial and passing to the end-organs of the latter.

This procedure has the disadvantage that none of the fibres reaches its normal end-organ. However, both the tibial and the peroneal are nerves which have both muscular and cutaneous destinations, the tibial being the larger. It is known that the peroneal motor fibres which enter the gastrocnemius muscles make endings such that stimulation of the nerve causes contraction of these muscles, but it is uncertain to what extent such abnormal connexions function during life. In any case, the final terminations are equally abnormal on both sides of the animal, and the only difference is therefore in the length of nerve to be regenerated.

Two experiments of this kind were made, rabbit ion being killed 107 days, and rabbit 1010, 200 days after operation.^*

In 1011 the distance on the side with a short nerve (AB; see Text-fig. 1) between crushed point and the union with the tibial nerve was 20 mm., whereas A’B’, the distance from the crushed point to the union on the opposite side of the animal, was 75 mm. From the results of Gutmann, Guttmann, Medawar & Young (1942) we may assume that the tips of the regenerating axons on both sides cross the region of crush 5 days after the operation, and advance through the peroneal nerve at 4·4 mm./day. Conditions in the two nerves are the same, except that on the one side the fibres have to travel an extra 55 mm. We may, therefore, expect that the fibres will arrive in the muscles supplied by the long nerve 55/4·4= 12·5 days later than on the short side. Out of a total regeneration time of 107 days, the fibres in the short nerve will have had 12·5 days longer for maturation after the arrival of fibres.

In order to discover what difference this increased length made to the process of maturation, sections were made of the peroneal nerves 10 mm. below the crushed point. Table 1 shows that the estimated total number of fibres is similar in the two nerves. D, the ‘mean’ fibre diameter (see p. 204), is slightly greater in the longer nerve, but since differences of this magnitude can be expected to occur in 60–70% of cases as a result of the operation of chance factors, the difference cannot be termed significant. Therefore, neither the fact that a greater length of nerve was regenerated, nor that contact with the muscle was made later, had operated to affect the diameter of the fibres immediately below the crushed point (Pl. 4, figs.. 1, 2).

Sections were also made farther down both nerves, and showed plentiful medullated fibres below the suture points on both sides (Pl. 4, figs. 3, 4; 1011f,K). No large difference was evident in the fibre diameters but counts were not made. It seems that at this level two equally good maturation has been achieved on the sides in spite of their unequal distances from the lesion.

Still farther distally this is no longer true. Sections of the posterior tibial nerves above the heel showed only very few medullated fibres in the nerve which had been artificially lengthened (Pl. 4, fig. 5 ; 1011 l), whereas they were more abundant and larger on the other side (Pl. 4, fig. 6; 1011 m). On both sides the fibres were much smaller than at more proximal levels. Evidently the greater length to be regenerated had allowed less full maturation.

We were also interested to discover whether the length of new nerve fibres to be formed affected the diameter of fibres central to the lesion. Table 1 shows that D was in fact smaller in the nerve which had the longer distance to regenerate, the difference between the two sides being of a magnitude that would only be found in 5–10% of cases as a result of random sampling alone. 100 days after injury is the time at which the fibres of the central stump show their greatest decrease in volume (Gutmann & Sanders, 1943), and if this decrease is in some way connected with the outpouring of material to make the new fibres, it might be expected to be greater in the nerve which has the greater length to regenerate.

Sections of the points of union of the peroneal and tibial nerves showed good junctions on both sides (Pl. 4, figs. 7, 8; 1011 e,f) without any waist-like constriction. Although, unfortunately, there is no means of making a quantitative assessment of such junctions, we feel safe in concluding that these are similar on the two sides, and therefore that the absence of any difference in fibre diameter is significant.

In rabbit 1010 the nerves were arranged so far as possible exactly as in 1011, and left to regenerate for 200 days. Using the same assumptions as before, we may conclude that the new fibres arrived at the gastrocnemius muscles 10 days later on the ‘long’ side, and therefore had that much longer for maturation in the short nerve.

Table 1 shows that 10 mm. below the point of crushing there were again approximately the same number of fibres on both sides. At this level, however, there were rather larger fibres on the side with the longer nerve. Moreover, this difference is greater than any to be expected as a result of random sampling. It will be noticed that the fibres are larger than after 100 days, though not greatly so (Pl. 5, figs. 9,10).

Below the points of suture of the peroneal to the tibial nerves no actual counts were made, but the fibres appear smaller on the side with the longer nerve. On both sides there is a distinct difference between levels above and below the suture in that, whereas at the former there is usually only one large fibre per tube, at the latter there are several smaller ones (Pl. 5, figs. 9–12; 1010 c,m,g,q). This is a distinction regularly observed between nerves regenerated below crushed and sutured points. In this case the only injury of the axons was by crushing, and it is therefore demonstrated that the larger number of small fibres is due to interruption of the whole nerve and its tubes.

In the posterior tibial nerves at the heel the fibres were distinctly larger on the side with the shorter nerve (Pl. 5, figs. 13, 14; 1010 j, s). On both sides, however, they were larger than after 107 days.

Sections of the crushed points showed similar conditions on the two sides (Pl. 5, figs. 15, 16; 1010 a, k). The unions between the peroneal and tibial nerves had been rather less successful than in rabbit 1011,(Pl. 6, figs. 17, 18; 1010 f,n). In particular, that on the side with the lengthened nerve showed a considerable lateral deviation of fibres. Though the stumps were closely apposed, conditions had evidently not been favourable for the formation of a clot with good longitudinal orientation. The most probable explanation for this is that the union was made without longitudinal tension, and the case is indeed a warning of the bad results which may follow from a union made under apparently ideal surgical conditions (see Weiss, 1943). The difference between the two suture lines may explain the somewhat smaller fibres found immediately below the crush in the shorter nerve.

We may conclude from these two experiments that the length of nerve to be regenerated below an injured point does not greatly affect the rate of increase of diameter of the fibres in the section of nerve immediately below the injury. Maturation does not proceed uniformly along the whole length of a regenerating fibre, but spreads centrifugally, so that at any given time more distant points show a smaller diameter (see also Gutmann & Sanders, 1943).

Two rabbits (1012 and 1013) were operated on in such a way as to reveal the effect of presence or absence of a connexion with the periphery on regeneration in the sensory and motor bundles of the peroneal nerves. On each side the nerve was crushed in the upper third of the thigh, above the point at which it divides into separate bundles destined to supply muscles and skin respectively. The former bundles of course contain somatic motor as well as sensory fibres, whereas the cutaneous branches (anterior tibial nerve) contain sensory and sympathetic fibres only, the latter being almost entirely non-medullated. The nerve was also severed below the knee, where it passes under the lateral head of the gastrocnemius. On one side the cut lower ends were immediately reunited with plasma, whereas on the other side the severed end of the peroneal was turned aside and fixed in such a manner as to allow no possibility of reunion.

The animals were left to regenerate for 172 and 200 days and counts then made of the number of fibres in the sensory and motor bundles below the crush and above the point of severance. Table 2 shows that in both animals the number of fibres was greater and their diameter smaller in the nerves which were not connected with the periphery. In both cases these differences were found to be statistically significant. The table also shows the numbers and diameters of the fibres in the central stumps, above the region of crushing. In rabbit 1012, which had been allowed to regenerate for 172 days, there were 4986 fibres in the central stump on the side with a peripheral union, and 5097 on that with none, but the fibres were distinctly smaller in the latter. This is in itself a very interesting fact, suggesting that the recovery in diameter of central fibres following their decrease during regeneration is influenced in some way by functional connexion with the periphery.

Below the crushed point in the nerve connected with the periphery, there were 5361 fibres in motor and sensory bundles together, that is to say, only a few more than in the central stump. The diameter was distinctly larger in the motor than in the sensory branches (Pl. 6, figs. 19, 20). The regenerated stretch of nerve which had not been allowed to make connexion with the periphery contained a total of 11,213 medullated fibres, more than twice as many as in the central stump. This confirms the finding of Simpson & Young (1945) that when regenerating fibres fail to make contact with the periphery, far more of them are found in the peripheral than in the central stump. Weiss et al. (1945) have recently shown a similar effect in the nerves of rats. Further, the diameter of the regenerated fibres remains very small in this nerve, and D is not greater in the motor than in the sensory bundles (Pl. 6, figs. 21, 22). This last is a result of the utmost importance, since it strongly suggests that the factor which makes somatic motor fibres large is a peripheral, or ‘functional’ one. For since the upper lesion is a simple crush of the nerve, it is to be presumed that the large central fibres proceeded down large peripheral tubes and yet failed to reach a large size. Examination of the sections indeed shows the large tubes of the motor branches most clearly, each filled with numerous small fibres (Pl. 6, figs. 21, 22; 1012 c, sensory and motor). The sensory and motor branches can still be distinguished by the large tubes of the latter, in spite of the fact that the fibre diameters are similar in both. Histograms reinforce the pictures, showing that not only is the fibre diameter similar in the two’nerves, but further that the large motor fibres have not given rise to any large fibres between the point of crushing and the neuroma, the maximum diameter being 12 μ in the motor and 10 μ. in the sensory bundles. However, the motor bundles contain a larger number of fibres 6 –10 μ in diameter than the sensory bundles, giving some indication of the operation of factors other than peripheral connexion.

The second animal, 1013, shows essentially similar results. Here the total number of fibres below the crush on the side in which connexion with the periphery was allowed was 4768, slightly less than in the central stump (5475). In the nerve with no peripheral connexion the number of regenerated fibres was again much greater (8845) than that in the central stump (5148). The central stump fibres were, as before, smaller in the nerve with no periphery. Their reduction in diameter without, apparently, reduction in thickness of their myelin sheaths, leads to the production of the remarkable fibres seen in Pl. 7, fig. 24 (1013f) [cf. fig. 23 (1013 a)]. The fibres of the motor bundles which had been allowed to make functional connexions showed a typical increase in diameter, and there had been some increase in diameter also in the sensory bundles which had made a peripheral connexion. In the motor bundles with no peripheral connexion D was at most only slightly greater (4 ·9 μ) than in the sensory bundles (4 ·3 μ). The results from this experiment, therefore, fully confirm that of the previous one in showing that all regenerated fibres remain small unless connected with the periphery, such connexion causing some increase of diameter in sensory and still more in motor fibres.

In the experiments of Simpson & Young (1945) it was shown that pieces of splanchnic nerve grafted to post-intercostal nerves came to contain quite large fibres. The present series included two animals in which a test of the same sort was made by grafting into the peroneal nerve in the thigh. On one side of each animal, the graft was a piece of the peroneal from the opposite side. The defect made by taking this graft was then filled by a piece of posterior tibial nerve from the same animal.

The posterior tibial contains smaller medullated fibres than the peroneal (see Sanders & Young, 1944), and the experiment therefore tests the extent to which the large fibres present in the peroneal nerve in the thigh are able to dilate the small tubes of the posterior tibial when they have passed through the latter and made çonnexion with the peroneal muscles. The grafts were 2 cm. in length and were fixed in place with clotted cockerel plasma.

The first animal (1015) was allowed to survive for 98 days, but the experiment was reduced in value by the fact that the upper union of peroneal with posterior tibial was poorly made, the two stumps being separated by about 0 ·5 mm. of irregular connective tissue. Very many nerve fibres had nevertheless penetrated this, and the grafted bundles of the posterior tibial contained many medullated fibres. Those in the corresponding piece of grafted peroneal nerve on the opposite side were more numerous, but perhaps not larger. However, in view of the uncertainties introduced, by the poor union no counts were made. Below the grafts both peroneal nerves contained many medullated fibres, that with the peroneal nerve graft containing more than that with the posterior tibial graft. This case, therefore, shows that in spite of a poor union the many small-fibred bundles of the posterior tibial, when used as a graft, can allow a connexion to be made.

The second animal (1037) was allowed to survive for 200 days after an exactly similar operation. Here the union of the peroneal and posterior tibial graft had been well made (Pl. 7, fig. 25 ; 1037 b), though the presence of many bundles had led to some complications. However, it may reasonably be assumed that the opportunities for connexion were similar to those presented by the union of the peroneal with the peroneal graft (Pl. 7, fig. 26; 1037 h) on the opposite side. The lower unions of both grafts with the peroneal nerves were also well made.

Table 3 shows that the fibres were somewhat larger in the graft of the peroneal than in that of the posterior tibial (Pl. 7,,figs. 27, 28). In the peroneal the difficulties of non-uniformity of various parts of the graft, owing to poor fixation in the centre, make the difference in fibre diameters observed of doubtful significance. Certainly the small tubes of the posterior tibial graft have not grossly restricted the growth of the fibres, but the result suggests that they have done so to some extent.

No strict comparison is. possible between the diameters here recorded and those shown in Tables 1 and 2, which are of fibres regenerating below a point of crushing and not one of suture as in the present case. However, it is interesting that the diameter reached by the fibres is distinctly above that found in those nerves which made no connexion with the periphery. Evidently, after passing through the grafts, the fibres have been able to make connexion with their end-organs and to undergo a subsequent increase in diameter. Below the grafts fibres of large diameter were found on both sides (Pl. 7, figs. 29, 30).

These quantitative results abundantly confirm the previous evidence that fibres become large if they make connexion with the periphery (Weiss & Taylor, 1944; Sanders & Young, 1945; Weiss et al. 1945). Regenerating stretches of nerves which end blindly contain far more and smaller fibres than others comparable in every way except that the fibres can make connexion with muscles and skin. A very important new result which emerges from these counts is that regenerating motor fibres which do not reach their end-organs become very little larger than sensory fibres. This shows the power of the peripheral effect very strikingly, and suggests that it is of paramount importance. It seems not unlikely that during normal development all fibres are of like diameter, and that the spectrum of fibre sizes develops as a result of the peripheral connexions, perhaps by the very act of functioning. Indeed, it may well be that the influence extends back from the fibres to the cellbodies themselves, and that it is the peripheral connexion which determines the size of cells in the spinal cord (see Young, 1945). Such an effect of the periphery upon the centres recalls the theories of Weiss. He has always maintained that such an influence exists, though not quite in the sense now revealed.

The figures of Table 2 show for the first time that sensory as well as motor fibres are affected by their connexions with the periphery. The anterior tibial nerve of the rabbit is a pure skin nerve (see Sanders & Young, 1944), its only motor fibres being sympathetic, presumably non-medullated and therefore not considered here. Yet during regeneration its fibres become larger if they are allowed to reach the periphery. It should perhaps be noted, however, that since the suture at S of Text-fig. 2 is of the whole peroneal nerve, many of these sensory fibres will enter the peroneal muscles. However, Gutmann (1945) has confirmed the previous evidence that sensory fibres cannot form motor end plates in mtiscle. Further evidence about the factors which control the maturation of sensory fibres is needed : these experiments show that the peripheral connexions exercise a large influence.

The present investigations still leave it quite uncertain how connexion with the periphery comes to exert its effect on maturation of the fibres. The experiments in which the nerves were artificially increased in length show that the diameter reached by fibres immediately below a crushed point is not greatly affected by the length of nerve to be regenerated more distally.

Hand in hand with the increase in size of the fibres which have become connected with the periphery goes the reduction and disappearance of those remaining without connexion, but it is impossible at present to say whether the one process controls the other. The growth of one fibre in a tube may have the effect of constricting the others and hence leading to their degeneration. Until such degeneration has been actually observed it is impossible to say more. It may be that the material used to make the large fibres is in some way derived from the others arising from the same parent fibre, thus producing their degeneration. Certainly the whole increase produced by peripheral connexion is not obtained by draining material from other fibres. For the total volume occupied by the fibres in the peripheral stump of an unconnected nerve is much less than that of the fewer, but larger, fibres which develop when a peripheral connexion is allowed. Weiss et al. (1945) also found this in the nerves of rats. The last column of Table 2 contains estimates of the volumes occupied by fibres in the central and peripheral stumps of the nerves. Assuming that the distance from the point of crushing to the neuroma was 45 mm., to the peroneal muscles 65 mm., and to the skin 195 mm., the total volume of the fibres in the peripheral Stump is seen to be very much greater ir the nerves connected with the periphery. The tota increase of some of the fibres cannot therefore have been only at the expense of others. Moreover, it must be remembered that the fibres central to the crush alsc remain small in an unconnected nerve. Assuming that the diameters found immediately above the lesion are maintained to the cord, the volume in the central stumps would be 28 and 31 cu.mm, in the nerves connected with the end-organ, but only 22 and 19 cu.mm, in those ending blindly as neuromas assuming that all crushes are made at a distance of 100 mm. from the cord. Weiss et al. (1945) have already noted that the fibres remain small in the central stumps of a regenerating nerve withoul peripheral connexion. As they observe, the small size of such fibres cannot be due entirely to drainage of axoplasm from the central fibres. The volume oi the many small fibres in an unconnected regenerating nerve is smaller than in a similar nerve having connexion with the periphery (see Table 2), whereas the central stump of the latter contains the larger fibres Hammond & Hinsey (1945) believed that the small decrease which they observed in the central stumps of sutured nerves might be due to the presence oi many small fibres turned back at the junction A similar explanation cannot apply in the present case, since the number of fibres in the central stump of the unconnected nerves was not significantly greater than in those with a peripheral connexion However, it seems likely that some part of the large number of fibres seen above the neuromas was due to ‘doubling back’ towarels the centre as suggested by Weiss et al. (1945).

It is impossible to resist the conclusion that the making of a connexion with the periphery produces an actual increase in the size of the fibres, whereas without such connexion not only do newly regenerated fibres not increase, but also their parent fibres remain small. Functional hypsertrophy is, of course, common enough in muscles, glands, and other tissues, and it is most interesting to find it also in a tissue whose function is conduction. It is much to be hoped that it will be possible to throw further light on the processes by which an effect of such importance for the functioning of the organism is produced.

Our thanks are due to the Rockefeller Foundation for their support which made this work possible.

The photographs of Weigert preparations of transverse sections are direct bromide prints at a magnification of 430 diameters.

The photographs of the junctions are of preparations stained by Bodian’s protargol method at a magnification of 21 diameters.