An experimental analysis of bivalent interlocking in spermatocytes of the newt Triturus Vulgaris

When interlocked bivalents, in the simplest case 2 ring bivalents, are observed at 1 st meiotic metaphase, it is generally assumed that this abnormal topological relationship came about because 2 pairs of homologous chromosomes trapped one another during synapsis (see Fig. 1). If sufficient chiasmata in the appropriate places are established thereafter, the 2 bivalents remain interlocked until they are released at ist meiotic anaphase.

There can be no doubt that the trapping of chromosomes during synapsis is responsible for many of the recorded cases of interlocking; indeed, at the time of the telosynapsis/parasynapsis controversy in the early 1900s, reviewed by Wilson (1925), the observation by Gelei (1921, 1922) of interlocked bivalents at zygotene in oocytes of the planarian Dendrocoelum was used as a strong argument in favour of parasynapsis. There have been several subsequent observations of a similar kind, by Belar (1928) in the snail Viviparus, by Levan (1933) in Allium, by Buss & Henderson (1971a, b) in heat-shocked locusts, and by John (1976) in the grasshopper Chloealtis. All the above are examples of interlocking observed at zygotene or pachytene.

Interlocks have much more frequently been recorded at diplotene or first meiotic metaphase. Darlington (1937) has provided an extensive list of organisms in which interlocks have been reported, including Mather’s (1933) famous example of complex interlocking at diakinesis in Lilium regale, which was of crucial import at the time of the classical /chiasmatype controversy regarding the relationship between chiasmata and genetical crossing-over. With one exception, all such observations of interlocking have been attributed to chromosome trapping during synapsis. The exception is provided by Taylor (1949) who recorded an increase in the frequency of interlocked i st metaphase bivalents of Tradescantia paludosa brought about by excising and culturing anthers in an artificial medium. Taylor observed a marked increase in the frequency of interlocked 1st metaphase bivalents even in anthers which had been excised after synapsis and part of pachytene had been completed on the plant. Taylor was puzzled as to how this observation could be explained, though he suggested as a possibility that partial desynapsis might follow excision of an anther, and that if it were followed by chromosome ends ‘reuniting’ at metaphase, interlocked bivalents could result. The observations on newt male meiosis to be described in the present paper bear out Taylor’s contention that bivalent interlocking can be generated in meiocytes subjected to a heat shock as late as pachytene, i.e. several days after synapsis has been completed in normal fashion.

Ten years ago one of us (H.G. C.) gave 24-h heat shocks to males of Triturus vulgaris (2n = 24) with the hope that the precise stage at which chiasmata are established could be determined. Information from autoradiography was already available (Callan & Taylor, 1968) concerning the durations of the 5-phase and the 4 stages of meiotic prophase, and the experiment was conducted in the anticipation that a just-sublethal heat shock, if applied during synapsis or at any time up to the stage at which chiasmata are formed, would lead to a significant diminution in chiasma frequency, as Henderson (1966) had shown to be the case in male meiosis of the locust Schistocerca, whereas a heat shock applied after chiasmata had been established would not be expected to affect chiasma frequency. In the event, the heat shock was found to have no very dramatic, if indeed any, effect on chiasma frequency in these newts and the fixed testes from animals which had experienced heat shocks were pooled with other material and used for class teaching. Some years later a few of the stained preparations were examined again, and in certain of these an unusually high frequency of interlocked ist metaphase bivalents was noticed. One such was photographed and used as an illustration for a paper by Callan & Perry (1977). As an outcome of this casual observation it was decided that the heat shock experiment would be worth repeating, but this time paying attention to what stages of meiosis are sensitive to heat shocks in the sense that they generate interlocked bivalents at ist meiotic metaphase.

First meiotic metaphases in spermatocytes of Triturus vulgaris are particularly convenient for scoring the interlocked condition, because the chromosomes are large, most of the chiasmata are formed near the ends of the bivalent arm pairs, the centromeres are median or sub-median, and thus the majority of interlocked rings can be identified without any ambiguity. It was assumed that in normal circumstances interlocking is a rare event; one of us (H.G.C.) had never previously noticed interlocked bivalents in this species, or in the related Triturus helveticus, which also has proterminal chiasma localization in male meiosis. It was also assumed that interlocking would be observed only in those spermatocytes which had been subjected to heat shocks up to, but not after, zygotene. Both these assumptions proved to be wrong.

Most of the male specimens of Triturus vulgaris which were used for the heat shock experiments in 1968, and all of those used in 1977, were collected from ponds in Tentsmuir forest, northeast Fife, during the spring months, when the newts were aquatic and in full breeding condition. They were placed in aquaria in a room held at a temperature of about 16 °C (minimum 15 °C, maximum 18 °C) where, as is usual with this species in captivity, the newts soon lost their breeding condition (the first sign of which is a reduction in height of the dorsal crest) and started to climb out of the water. As the newts left the water they were transferred, 5 at a time, to small plastic tanks (25 × 15 × 10 cm) each containing a wetted paper tissue and a piece of undulating asbestos/cement roofing material under which the newts took cover. The tanks were cleaned out once a week, and no attempt was made to feed the newts once they had become terrestrial.

The capacity of Triturus vulgaris to survive a 24-h heat shock was tested in 1968, using a water-jacketed incubator. Plastic tanks each containing 4 newts were placed inside the incubator, the temperature of which was checked at frequent intervals and found to vary no more than ±0·2 °C from the set temperatures. Of 8 newts placed at 33.5 °C, only one was still alive after 24 h and this one died a day later. Four newts placed at 30 °C all survived, so did 4 newts placed at 32.5 °C, but of a further 4 placed at 33.0 °C one was dead 24 h later and the remaining 3 in poor condition. As it was the intention to use a heat shock from which the newts would recover fully (many were to be kept for a month at or about 16 °C after experiencing the shock) the decision was taken to use a temperature of 31’5 °C for 24 h. All of the experimental animals survived this treatment, and were healthy when killed and their testes fixed.

The intention at the outset was to give a 24-h heat shock, then make fixations of 3 newts’ testes immediately thereafter (+ 0 days) and similarly of 3 newts’ testes at daily intervals until +20 days, then at 2-day intervals until +36 days. The 1968 fixations provided some preparations falling within the period + 12 days to +20 days (see Table 1).

It is impossible to determine from a newt’s external features whether its testes will contain the required stage of 1st meiotic metaphase; this becomes apparent only after squash preparations have been made. Consequently some testes were fixed too early, some too late. Batches of newts were given 24-h heat shocks on 31st May, 2nd June, and 6th, 7th and 15th July, 1977, and as far as possible gaps appearing in the earlier fixations, where insufficient or no examples of 1st meiotic metaphase were present in the preparations, were made good in later fixations.

In a study of the time course of male meiosis in Triturus vulgaris at 16 °C, carried out in 1966 (Callan & Taylor, 1968), it was shown inter alia that spermatocytes take 20 to 21 days to pass from the end of premeiotic S to 1st meiotic metaphase. So as to keep a check on the time course operating in 1977, when the temperature of the newt room was overall marginally lower owing to a difference in outside temperature and degree of insolation between the 2 years, the newts just before being given a heat shock were anaesthetized with MS222 and each injected with 10 μCi of [³H]thymidine, 5.0 Ci/mM.

Before fixation the newts were anaesthetized with MS222, their testes removed and fixed in a freshly prepared mixture of 3 parts absolute ethanol and 1 part glacial acetic acid. The fixed testes were stored in a refrigerator until preparations were to be made. A small fragment of testis was placed on a slide cleaned from 95% ethanol. Time was allowed for residual fixative to evaporate, and then 2 small drops of 0.5% orcein (G. T. Gurr’s synthetic) in 45% acetic acid were added. The testis fragment was tapped out to dissociate the cells and examined, uncovered, under the low power of a light microscope. If divisions proved to be present, the testis fragment was tapped out much more extensively, and the slide placed in a moist chamber for 3 min. The material was now spread out lengthwise over the slide and examined under a binocular dissecting microscope against a white background. Fibrous connective tissue strands were removed with forceps, and the material covered with a No. 2 22 × 40 mm siliconized coverslip cleaned from 95% ethanol and gas-blasted to remove fluff. The preparation was now squashed out firmly between the folds of a folded filter paper circle, placed on solid CO₂ in a plastic ‘igloo’ until well frozen (about 10 min), then the coverslip flipped off with a razor blade and the slide placed at once in 95% ethanol. After a minute or two the slide was transferred to absolute ethanol, the preparation then mounted in Euparal under a No. o 22 × 40 mm coverslip, and dried on a hot plate.

These stained preparations were later examined under a Zeiss WL microscope fitted with a green interference filter, and sketches made of well-spread 1st meiotic metaphases, recording the number of chromosome arm pairs associated by chiasmata, the total number of chiasmata, the frequency of pairs of univalent chromosomes, and the number of bivalents interlocked with one another. Both authors scored 20 different 1st meiotic metaphases in this manner; if there was a significant disparity between the 2 sets of observations, which for the most part were carried out on different slides, more metaphases were scored. This was deemed necessary only in the cases of 5 out of a total of 81 newts which provided usable material. Photographs of 1st meiotic metaphases were taken under a Zeiss planapochromat x 100 oilimmersion objective, N.A. 1.32, using Kodak Pan F film.

Autoradiography was used to check the time course of meiosis in the 1977 fixations, the methodology being just as described by Callan & Taylor (1968).

Testes from a few untreated newts were fixed as controls, one testis in ethanol/acetic acid for making squash preparations, and the other in Sanfelice’s fluid for wax embedding, sectioning, and staining in iron haematoxylin. Likewise one testis from each of the +o-day animals, i.e. immediately following heat shock, was fixed in Sanfelice’s fluid for sectioning, to compare with sections of control testes.

The information collected from the squash preparations is shown in Table 1. One male newt, 77 + 18C, proved to be triploid, and will not concern us here. The first point deserving attention is that there is a low, but appreciable, frequency of interlocked bivalents in control animals. We thought that the Tentsmuir newt population, which lives close to the sea, might be abnormal in regard to bivalent interlocking, and so we examined meiosis in preparations from 3 male T. vulgaris collected from a dew pond at Forret Hill, Logie, some 10 km to the southwest of the Tentsmuir ponds and 140 m above sea level. However these also showed interlocking. This makes clear the fact that a cytologist, when studying preparations of meiosis for chiasma frequency and distribution, tends to pick out those divisions in which bivalents are well separated and easily analysable when given the choice!

If we accept from Table 1 that an interlock frequency of up to, say, 10 interlocks in 40 spermatocytes is about the maximum likely to be encountered in normal, control newts (the highest frequency recorded being 7 amongst the 8 controls) then clearly for the first 3 days following a 24-h heat shock the interlock frequency is within the normal range. But from 4 days after the heat shock until 28 days after the heat shock the frequency of interlocks in the majority of the newts (38 out of 56) is much above control level, while from 30 days after the heat shock until + 36 days, which was the latest fixation made, interlock frequency is back to control level. Within the sensitive period, from +4 to +28 days, there is a wide scatter of interlock frequencies, ranging up to 55 in 40 spermatocytes of newt 77 + 24C and down to control frequencies (assumed to be 10 and less) in 18 of the heat-shocked animals. This scatter is portrayed in Fig. 2.

The autoradiographic evidence for computing the time course of male meiosis in the newts used in 1977 was less comprehensive than it should have been because, for one reason or another, there was little [³H]thymidine incorporation in many of the newts. There was, however, sufficient information to relate the time course in 1977 to that obtaining in 1967, from which Callan & Taylor (1968) made their detailed analysis. In 1967 the longest interval between [³H]thymidine injection and fixation which showed ‘late-label’ distribution in pachytene nuclei (i.e. deriving from spermatocytes which were labelled at the end of the pre-meiotic 5-phase) was 18 days, whereas in 1977 the corresponding longest interval was 20 days plus the 24 h spent at 31.5 °C (information from newt 77 + 20A). This indicates that the 1977 spermatocytes were taking about 3 days longer to pass from the end of pre-meiotic .S’ to the end of pachytene. This figure is corroborated by information from newt 77 + 26A, in which all 1st meiotic metaphases were labelled, and these labelled uniformly, not late-labelled, a condition encountered at its earliest some 23 to 24 days after [³H]thymidine injection in the 1967 fixations. In short, the 1977 newts took some 15% longer to pass through meiotic prophase, and this factor has been applied to the data on the durations of the meiotic stages known from the 1967 fixations, giving some 11–12 days for premeiotic 5 (which laps over into leptotene for about 1 day), 7 days for leptotene, 10 days for zygotene, 6 days for pachytene, and 2 days for diplotene.

These stage durations have been entered in Fig. 2, so that the stage at which the spermatocytes experienced the heat shock, all of them having been scored for interlocks when they had reached 1st meiotic metaphase, can be appreciated. As regards the generation of an abnormally large number of interlocked bivalents, it is apparent that newt spermatocytes are sensitive to a heat shock from some time towards the end of premeiotic S until the last day or two of pachytene, i.e. some 4 days after synapsis has been completed.

As interlocked bivalents will be recognized at 1st meiotic metaphase only if the appropriate chiasmata are formed after interlocking has occurred, such chiasmata must have formed right at the end of pachytene. This finding is entirely in keeping with classical cytological doctrine, but the finding that interlocks can be generated between bivalents which had already completed synapsis is not.

A ‘ring’ bivalent is one in which both arm pairs are associated by one or more chiasmata, whereas a ‘rod’ bivalent is associated in one arm pair only. Some typical photographs of 1st metaphases including interlocked bivalents are shown in Figs. 3 to 6. Fig. 3 shows a simple ring through ring interlock, Fig. 4 likewise, Fig. 5 shows a rod interlocked with a ring, and Fig. 6 shows 3 rings interlocked with one another.

At the outset we expected to be confronted with the problem of ‘losing’ potentially interlocked bivalents because insufficient chiasmata might be formed in some of the newts, and that to make sense of the data we might need to make some allowance for this. Thus 2 rod bivalents, each of which has formed only a single chiasma, are unlikely to remain interlocked at 1st meiotic metaphase unless the centromeres of both assume non-disjunctional orientations, and strain against one another, in the manner depicted by Buss & Henderson (1971a, b) for locusts; in the newts we encountered no examples of such ‘orientational’ interlocking. However it will be apparent from Table 1 that there is no great variation in mean frequency of associated chromosome arm pairs per spermatocyte throughout the fixations, the overwhelming majority falling between 21.0 and 22.9, with low variance values.

Moreover, and rather to our surprise, plenty of rod bivalents remain recognizably threaded through rings (we scored 94 rods through rings, 8·8%, compared with 979 rings through rings, 91·2%) at 1st meiotic metaphase. The mean of the mean frequencies of chromosome arm pairs associated per spermatocyte is 22.08, and therefore if pairs of univalents are neglected (their numbers are, except for the maverick newt 68+13B, trivial, as can be seen from the right-hand column of Table 1), the average frequency of ring bivalents per spermatocyte is 10.08, and of rod bivalents 1.92. If we were dealing with a large ‘population’ of ring and rod bivalents mixed in the aforesaid proportions, we would expect to find rings associated in pairs at a frequency of 10.08², as compared with single rods associated with single rings at a frequency of 2×1.92× 10.08; however only one half of the latter would lead to interlocks because each rod bivalent has, by definition, only one of its 2 arm pairs available for interlocking with a ring. On a percent basis these expectations are therefore 84 and 16% respectively.

Professor R. M. Cormack has pointed out to us that these probability calculations are not strictly applicable to our data, because each spermatocyte contains a limited number of bivalents, 12, ring and rod bivalents occur in different numbers per spermatocyte (12:0, 11:1, 10:2, 9:3, 8:4, 7:5 and 6:6), and that therefore the expected ring /ring and continuing ring /rod associations must be separately assessed for each; these expectations are respectively 11:0, 10:1, 9:2, 8:3, 7:4, 6:5 and 5:6. The known percentage frequency of occurrence of 12 rings: 0 rods is 4-06, of 11 rings: 1 rod is 39.47, of 10 rings: 2 rods is 32.54, of 9 rings: 3 rods is 15.24, of 8 rings: 4 rods is 6.06, of 7 rings: 5 rods is 2-22, and of 6 rings: 6 rods is 0.42. When these percentages are applied as weighting factors to the relevant expectations, the overall expected frequency of ring/ring associations is 82.9% and of ring/rod associations 17.1%. The interlock frequencies actually found, 91.2% and 8.8% respectively, allow us to infer that about one half of potential ring/rod interlocks are retained up to 1st meiotic metaphase, the other half presumably dissociating as the bivalents move on to the division spindle.

Although we scored total chiasma frequencies as well as frequencies of arm pair associations, for the sake of simplicity mean chiasma frequencies and their variances are omitted from Table 1. The biggest difference between mean chiasma frequency and mean frequency of arm pair associations was 4.9 (Logie control 55 /s) but in none of the other newts was the mean excess more than 2, and the variances of mean chiasma frequencies were much the same as the variances of mean arm pair associations. Evidently a 24-h heat shock at 31.5 °C, whenever delivered, has little or no effect on the frequency with which chromosome arm pairs are associated by chiasmata, nor on total chiasma frequency, in T. vulgaris.

When scoring interlocked bivalents it is hard to avoid the preconception that because some spermatocytes contain several interlocks and others, the great majority, none, some spermatocytes may be particularly susceptible to the influence of a heat shock, but others less so or not at all. This preconception is manifestly false, for it is apparent from Table 1 that the variances of mean interlock frequency per spermatocyte follow the means remarkably closely, in other words each interlock is a chance event, which may or may not be accompanied by other interlocks in the same spermatocyte on a purely random basis.

We encountered 17 whole bivalent interlocks, i.e. where one entire ring was threaded through another, out of the total of 1073 interlocks scored. We found no interlocks between ring bivalents where both arm pairs of both bivalents were associated by 2 chiasmata, and where an interlock might have been expected proximally, i.e. within the rings including the centromeres. We found only 2 interlocks where a simple ring was threaded through the proximal ring, including the centromeres, of another bivalent which had formed 2 chiasmata in each of its arm pairs. Finally, we found no interlocks involving the region between 2 neighbouring chiasmata in onearmpairinanybivalentswhere2chiasmatahadformedwithin one or both arm pairs.

White (1973) has suggested that interlocking ‘…must be regarded as an extremely rare accident of meiosis whose frequency of occurrence is two or three orders of magnitude lower than would be the case on the hypothesis of random leptotene orientation’. John (1976) on the contrary, has demonstrated that interlocking is frequent in male meiosis of the grasshopper Chloealtis conspersa (this has been confirmed by H.G.C.) and that it occurs, although less frequently, in 5 other species of grasshopper. We now find interlocking to be not infrequent in Triturus vulgaris male meiosis. Nevertheless White is substantially correct in his claim. It is astonishing that interlocking brought about by synapsis is not of common occurrence in all those organisms, probably the great majority, in which synapsis begins at both arm ends of each pair of homologues, and proceeds from these ends towards the central regions. All text-book illustrations of the origin of interlocked bivalents (see for example, White, 1973, p. 154; Rees & Jones, 1977, p. 45) merely show 2 pairs of synapsed chromosomes trapped together, without illustrating the relationships of the ends of these synapsed chromosomes to the nuclear membrane, as is shown in Fig. i. There are organisms whose chromosomes usually synapse from one end only (for example Bombyx mori females studied by Rasmussen, 1977) and in which one can visualize how an intruding chromosome or bivalent might be progressively excluded as synapsis proceeds, but even in such organisms the exchange of pairing partner in triploids, and the successful completion of synapsis around loops in inversion heterozygotes, clearly demonstrates the inadequacy of the zipfastener explanation that is sometimes offered for synapsis, and which therefore cannot be held to account for the rarity of interlocking.

Several cytologists have claimed that interlocking is rare because homologous chromosomes already become pre-aligned with one another during the premeiotic mitoses, and retain their alignment during interphase because homologous telomeres are attached, alongside one another, to the nuclear membrane. Evidence for the attachment of telomeres to the nuclear membrane, or rather of the ends of the lateral elements (axial cores) of the synaptonemal complex to the membrane, is entirely convincing, whereas evidence of pre-alignment of homologues is, to say the least, conflicting. Indeed there cannot be any pre-alignment of homologues in haploid organisms, such as the Ascomycetes, where the homologues do not meet for synapsis until gamete nuclei fuse.

To account for the rarity of interlocking Rees & Jones (1977) propose that alignment of homologues occurs during premeiotic interphase when ‘. the chromosomes are in vigorous movement’; homologues become associated together at or near one pair of telomeres, which then move through the nucleus pulling the homologous pair of chromosomes towards one another and disentangling them from others, after which the telomeres attach to the nuclear membrane and consolidate the disentangled state. We find such a mechanism difficult to conceive in view of the lengths of chromosomes at premeiotic interphase, lengths which are admittedly not known with precision but which are assuredly greater than chromosome lengths at, for example, leptotene or pachytene. Moreover if one examines mid-zygotene in particularly favourable material, such as the spermatocytes of the plethodont salamander Batrachoseps attenuatus, introduced to us by Dr J. Kezer of the University of Oregon, but already well known to Janssens (1909), homologous chromosomes are manifestly not lying close together except where they are synapsed. At any Y-fork marking the limit of the synapsed region, the as yet unsynapsed regions generally form an obtuse angle with respect to one another, i.e. they are approaching one another from widely separate nuclear domains.

In discussion Dr E. M. De Robertis has suggested to us that a possible explanation for the rarity of interlocking is that the telomeres of single chromosomes may be temporarily joined to one another at the end of the premeiotic mitosis, so that eukaryotic chromosomes are for a time circular, like the chromonemes of prokaryotes. The joined telomeres become anchored to the nuclear membrane and associated with the joined telomeres of the homologous chromosome which is anchored nearby. Thereafter the chromosomes decondense, but so long as their telomeres remain united and homologous ends remain in association, they will occupy nuclear domains which exclude the possibility of trapping other chromosomes during the later synapsis. Such an explanation for the rarity of interlocking is by no means ruled out by the knowledge that, once synapsis has begun, bivalent arm ends are indisputably separate from one another (see, in particular, the electron-microscopic study of synapsis in Locusta by Moens, 1973) because a careful examination of leptotene in Batrachoseps neither shows nor fails to show the ends of individual chromosomes. When based on a light-microscopic study of sections or squash preparations of even the most favourable materials, observational difficulties preclude any firm statement on this question. It would be a subject worthy of study by more refined techniques.

Buss & Henderson (1971a, b) found that exposure of male Locusta migratoria to 40 °C for several days induced bivalent interlocking that was detectable both at pachytene and at ist meiotic metaphase. Their analysis was complicated by the fact that exposure to 40 °C also reduces chiasma frequency in Locusta, thereby diminishing the frequency of interlocks detectable at ist metaphase; and to complicate matters still further, after an initial and severe drop in chiasma frequency lasting for several days, when interlocks would be expected to be most often ‘lost’, chiasma frequencies rise back to near normality and it is only when they do that interlocks become detectable. Despite this complication Buss & Henderson interpret their observations by claiming that heat treatment applied to spermatocytes in leptotene, zygotene or pachytene does not induce interlocking, but that the sensitive stage occurs earlier, ‘…at some period between the telophase of the last spermatogonial mitosis and the main premeiotic S phase’. They go on to argue that chromosome pairing must therefore be a 2-step process (see also Maguire, 1977), the first of these being an initial alignment of homologues at premeiotic interphase or the final mitotic telophase, and that it is this initial alignment which is sensitive to heat and which, if it goes awry, will inevitably lead to interlocking at the later stage of zygotene synapsis. Buss & Henderson, like Rees & Jones, consider that the attachment of telomeres to the nuclear membrane stabilizes the mutual alignment of homologues that is said to be achieved prior to synapsis and during the first step of the pairing process, but they discount an alternative possible explanation for the generation of interlocks, that the heat treatment may cause telomeres to detach from the nuclear membrane, allowing non-homologous chromonemata to intertwine and thus become interlocked.

We on the contrary contend that our observations on Triturus support the opposite conclusion. We have shown that interlocks can be generated by a 24-h heat shock applied to spermatocytes at any stage from late premeiotic S until mid-pachytene. Our information regarding spermatocytes exposed to heat shock earlier in the premeiotic S phase is too scanty to warrant a statement that these are insensitive, though what information we have suggests that this may be so. However our significant finding is that interlocks can be generated in spermatocytes which had already completed synapsis, under normal conditions, before they experienced a heat shock.

At first sight this finding seems hard to explain. Like Taylor (1949) we thought of the possibility that the heat shock might cause desynapsis, followed by recovery and a second synapsis, preceding which non-homologous chromosomes might have become intertwined, but the appearance of pachytene stages in sectioned control testes, and in testes fixed immediately after the removal of newts from a heated incubator, proved to be identical. Heat shock does not appear to induce desynapsis in Triturus.

The likeliest explanation for our findings is that a heat shock causes telomeres to detach from the nuclear membrane, or from the lateral elements of the synaptonemal complex where these are attached to the nuclear membrane, allowing non-homologous chromonemata to become intertwined. Thereafter the interlocked condition will be maintained provided recombination between homologues occurs distal to the region of intertwining, in the ‘cloud’ of chromonemata outside the synaptonemal complex, not inside the complex between its 2 lateral elements. Despite the fact that unambiguous evidence for the presence of DNA between the lateral elements of the synaptonemal complex is lacking, some students of meiosis assert that recombination occurs within the complex (Moens, 1968; von Wettstein, 1971; Stern, Westergaard & von Wett-stein, 1975) whereas others, and notably the original discoverer of the synaptonemal complex, M. J. Moses, are less dogmatic. Moses (1977) writes: ‘There is presently no direct evidence from which to conclude whether crossing over takes place in chromatin immediately associated with the synaptonemal complex or in that more peripheral to it’. We think we may have provided indirect evidence in favour of the periphery.

Whether all interlocked bivalents in Triturus vulgaris originate in similar fashion is debatable. We imagine that whole bivalent interlocks, of which we found only 17 in a total of 1073, must have arisen by trapping during synapsis, much as depicted by Rasmussen (1977) in Bombyx mori oocytes. However telomere detachment and intertwining of chromonemata could account for all the rest, whether detachment occurs prior to synapsis, in which case synaptic trapping, i.e. foreign chromonemata lying within synaptonemal complexes, might result, or after synapsis, and involving only the chromonemata peripheral to the complexes.