Somite Abnormalities Caused by Short Heat Shocks to Pre-Neurula Stages of Xenopus Laevis

/. Embryol. exp. Morph. Vol. 45, pp. 283-294, 1978 283 Printed in Great Britain © Company of Biologists Limited 1978

Somite abnormalities caused by short heat shocks to pre-neurula stages of Xenopus laevis

By JONATHAN COOKE1 From the Division of Developmental Biology, National Institute for Medical Research, London

SUMMARY This paper describes the small disturbances, in the regular pattern of the somites and the fissures between them, that are seen following short (around 300 s) heat shocks at 37-5 °C delivered to pre-neurula stages of Xenopus laevis. Affected groups of cells still finally dif- ferentiate as somite muscle, but the normally precise spatio-temporal sequence in which they move beforehand to give rise to the actual pattern of somite blocks, is disrupted. Examination of the position and sizes of patches of disrupted morphogenesis, in relation to the precise embryonic stage at shock, leads to certain conclusions about the nature of the disturbance induced by a brief period at high temperature, in cells due to form somites. The pattern of results is compared with that produced by similar temperature shocks given to tail-bud (later) staged embryos. The discussion includes a brief consideration of how the various results of heat shocks, given at different embryonic stages, might be understood in terms of one particular model (Cooke & Zeeman, 1976) for the spatio-temporal control of the developing somite pattern.

INTRODUCTION Periods of a few minutes exposure to temperatures around mammalian blood heat (37 °C) are proving a useful probe into temporal organization of development in amphibian embryos, even though the biochemical effects of such temperature shocks are currently unknown. Thus, Elsdale, Pearson & White- head (1976) have been able to show that a transient phase of temperature sensitivity in pre-somite cells, sweeping back like a wave through the future body axis, underlies the normally very precise pattern of somites and the fissures between them. That work had used laboratory-laid clutches of Xenopus laevis eggs, but subsequent work has taken advantage of the natural synchrony and homogeneity of wild-collected clutches of Rana temporaria eggs, since parallel phenomena are elicited by heat shocks at all equivalent morphogenetic stages in Rana and in Xenopus. In the latter, synchronous groups of embryos are obtained by manual selection. The present paper describes, for Xenopus, the incidence of small disruptions in the somite pattern that follow short (4-8 min) shocks at 37-5 °C during late 1 Author's address: Division of Developmental Biology, National Institute for Medical Research, The Ridgeway, Mill Hill, London, N.W.7. U.K. 284 J. COOKE blastula and gastrula stages. This incidence has a different pattern than that caused by such shocks given at neurula to post-neurula stages. It is chiefly distinguished in that antero-posterior position of occurrence of disruptions within the body pattern bears no close relationship with precise time in morphogenesis when shock has been administered, and that incidence of abnormality in individuals is probabilistic rather than reliable following a certain number of minutes at high temperature. In both Rana and Xenopus, shock periods as short as 100 sec produce recognizable somite abnormalities. Those resulting from exposure of neurulae and post-neurulae are reliably present, and occur at positions in the somite rows which are systematically related to the precise embryonic stage at shock. Since blastular and gastrular shocks give a probabilistic incidence of abnormalities, however, shocks of a few minutes have been chosen in the present study, giving a statistically useful incidence of readily scorable abnormalities. This is coupled with almost 100% survival, and minimal non-specific cell damage. A report on these early shocks delivered to Rana, presenting a comparable but perhaps slightly different picture, is to be given elsewhere (Elsdale & Pearson, unpublished). A striking feature in both species is that the early developmental period of susceptibility to heat shocks, expressed in the pattern of abnormalities described below, comes to an end appreciably before onset of the later phase of susceptibility that gives body position-specific, reliable defects as earlier described (Elsdale et al. 1976). There is thus a transient period, at the close of gastrulation, when heat shocks which do not kill the embryos leave the subsequent somite pattern quite normal. It is important to realize that all the heat shocks described in this and the previous work produce pattern abnormalities by affecting cells which will not participate in somite formation until anything from 2\ to some 24 h later (in Xenopus - longer in Rand). Only the spatial patterning of restricted groups of cells, not their normal histo-differentiation as muscle, is affected. The bearing of such observations as these on theories of the spatio-temporal organization of the developing system will be briefly discussed.

MATERIALS AND METHODS Wild-caught South African clawed toads {Xenopus laevis) were kept in the laboratory at 21 °C on a diet of raw beef heart. During a maximum of 2 years of laboratory life, eggs were obtained by induced matings using injection of Human Chorionic Gonadotropin ('Pregnyl' - Organon Laboratories Limited), 150 i.u. for males and 350 i.u. for females. Pairs of toads were not used at less than 5-week intervals. Eggs were manually de-jellied down to the vitelline membrane and left to develop in 1/10 strength Niu Twitty solution (Rugh, 1962) at 21-22 °C. Groups Somite abnormalities in pre-neurula stages o/Xenopus 285

Fig. 1. Drawings of typical local abnormalities in left-hand somite rows of Xenopus. {a), (b) Abnormalities of the extent referred to as 'disturbances', (c), (d) Abnormal- ities referred to as ' minimal abnormalities'. ev, ear vesicle, marking the beginning of the somite series for purposes of the present experiments. Chevron-shaped lines show the regular positions of normal somite fissures before and after the areas of deranged and/or absent fissures which mark disturbances, while the overall outline is of the dissected out column of somite tissue. of 15-25 embryos, synchronous to within 20 min in morphological stage (Nieuwkoop & Faber, 1956, plus recognizable intermediate stages) between 9 + (late blastula) and 13^-14 (early neurula), were allowed to drop from the sub- merged mouth of a pipette into a 30 ex. boiling tube of 1/10 Niu Twitty kept at 37-5 °C in a circulating water bath. After a specified period of between 4 and 8 min they were retrieved from the bottom of the tube with the wide-mouthed pipette and dropped into a dish with a large volume of 1/10 Niu Twitty at 21-22 °C, resting there for 5 min before storage in shallow dishes of the same solution to develop. Somites were examined in developed larvae at 30's stages, when some 40 somites have formed in the extended axes and the skin is translucent. They were relaxed in a solution of 1:2000 MS 222 (Sandoz) for 5 min, then immersed in 2 % acetic acid in aqueous 0-5 % potassium dichromate for 3 min, whereupon the skin could be stripped cleanly from both sides to reveal somite rows in

E M B 45 Fig. 2. Horizontal histoiogical sections at notochord level showing local somite abnormalities. (— arrows), (a) A 'disturbance', (b) A 'minimal abnormality'. The chaotic rotation of cells and the lack of normal fissuring can be seen in (a), and the uneven sizes of adjacent somites with a partial lack of rotation locally in (b). Examples with some degree of disruption on both sides of the notochord have been chosen. n.c, Notochord; s, somites, ep, epidermis. lateral view. Somite abnormalities were scored as to nature and position (by fissure number counting back from the ear vesicle). Selected rows of somites and their fissures were drawn in camera lucida. A selection of embryos was also fixed overnight in complete Smith's fluid (i.e. the above fixative with addition of 4 % formalin), then washed for 24 h in running tap water, dehydrated with alcohols and propylene dioxide and embedded in Araldite for horizontal sectioning at 1 /*m through notochord and somites.

RESULTS Eggs from seven separate layings were used to explore this phenomenon, results from each experiment being essentially the same. Statistical records were derived from two experiments, in which the layings of eggs concerned showed (a) a particularly homogenous developmental rate of progression through the stages, and (b) little mortality and hardly any non-specific cell loss or damage, even though a useful incidence of abnormalities involving two to five successive somites was seen following shocks of 6 and 7 min. Cell loss and damage, if it occurs, is always visible as a cloudy mass of extruded yolky debris within the perivitelline space of embryos during the 34 h after shocking. Embryos without such damage seem to be delayed only by a uniform 2-3 h of development, relative to controls, and to be quite normal in overall behaviour and morphology as larvae. Somite abnormalities in pre-neurula stages of Xenopus 287

Table 1. Probability of abnormality per somite row {i.e. side of an embryo) as a function of stage at temperature shock

Total of all abnormalities (i.e. disturbances + Total of Total somite minimal Probability minimal Probability Embryonic stage rows (i.e. abnormal- per somite abnormal- per somite at shock embryos x 2) ities) row ities row 9 + 62 46 0-74 28 0-45 10 86 61 0-71 32 0-37 101-104 94 61 0-65 28 0-30 10| 96 72 0-75 32 0-34 11 92 57 0-62 32 0-35 11* 80 43 0-54 33 0-41 12 80 16 0-20 13 016 124-13 80 0 0 0 0 134-14 80 75 0-94 6 008 Stage 124-13 is the refractory period. The near 100% incidence for the next stage represents the onset of the second type of reliable position-specific disturbance (here in anterior somites) and is, of course, excluded from the data of Fig. 3a-c.

The continuous range of local abnormalities in somite pattern observed was arbitrarily divided into two categories of severity, referred to as ' disturbances' (more extended disruptions) and 'minimal abnormalities' (involving at most two successive somite fissures). Figs. 1 and 2 show, respectively, camera lucida drawings and horizontal histological sections of examples of the two categories. Hamilton (1969), Cooke (1975a, 1977) and Cooke & Zeeman (1976) have described the process of somite formation in anuran amphibians, with its deviation from the basic vertebrate pattern. Lines of cellular de-adhesion, normally tracing a chevron («) shape as seen laterally in the columns of pre-somite cells, successively separate off groups of these cells prior to the latters' rotation through 90° into their definitive positions to form somites - wedges of longi- tudinally elongated cells. Unpublished observations (Elsdale, Pearson and Cooke) have revealed that such chevron-shaped lines of cellular de-adhesion are themselves formed with a characteristic dorsal-to-ventral time course, and that the time courses of formation of successive fissures are highly regular and coordinated in normal development. Abnormalities of the sort seen after short heat stocks are understandable as local areas (i.e. patches of tissue) in which the timing and positioning of fissure formation has been disorganized, although the cells concerned have still formed somitic tissue and histodirTerentiated as muscle. Posterior to any given patch of disorganized tissue, the remainder of a normal complement of somites is usually found. Minimal abnormalities correspond to small patches of

19-2 288 J. COOKE

1 2 Somites

(c)

Fig. 3. Histograms of the total incidence of involvement in abnormalities, by somites numbered from the ear vesicle. Results from two fully scored experiments are combined, analysis having shown no inhomogeneity between them, (a) The total score for all abnormalities. (6) The subset of all abnormalities incurred by temperature shock up to stage 10^ only, i.e. slightly more than half of all the abnormalities. They are seen to embrace the full range of affected somites (as does the other subset, of abnormalities incurred after stage 10£). The distribution does not deviate significantly from the total distribution (x2 = 115, 19 d.f.). (c) The subset of all 'minimal abnormalities'. Its distribution is somewhat biased towards later- formed somites (further from the ear vesicle) (x2 = 21-4, 16 d.f.) deranged cells, involving the courses of one or two fissures only; disturbances, to a larger patch of tissue involving the courses of several successive fissures. Table 1 shows the incidence of abnormalities, that is, the probability per somite row of occurrence of each of the two categories of abnormality, as a function of embryonic stage at shock. Data are derived from a total of 60 to 100 somite rows per stage. It is apparent that sensitivity to induction of the effect later causing abnormalities remains approximately uniform throughout the earlier part of gastrulation, but diminishes, and ends completely during stage 12-a small yolk-plug stage. Sensitivity thus comes to an end rather abruptly in the latter part of gastrulation. Stage 13|—14 is the stage at which the pattern of reliable damage is first seen, confined at this stage to the anterior-most visible somites. This corresponds well with the estimate of Elsdale et al. (1976) for the beginning of their wavefront of transient temperature sensitivity, in its passage down the body some 2 h ahead of actual somite formation at each level. Thus we are dealing with two Somite abnormalities in pre-neurula stages o/Xenopus 289 phenomena, or developmental periods of sensitivity to induction of some kinds of lasting perturbation in groups of cells. They are separated by a recognizable period of refractoriness to perturbation. Figure 3 a shows the histogram for incidence of all abnormalities, in each somite numbered in series behind the ear vesicle in the two similar experiments combined. Preliminary analysis showed distributions from the two experiments to be indistinguishable. Fig. 3 b is the histogram for that subset, of the same population of abnormalities, which had occurred only in response to shocks delivered up to stage 10| of morphogenesis, i.e. during the first half of the sensitive period. It does not differ, as a distribution, from the total distribution for shocks over the whole sensitive period (^2 = 11-5, 19 degrees of freedom). Position of perturbations within the future body pattern would therefore appear to be unrelated to the timing of heat shocks delivered within the late blastula/ gastrula sensitive period. This would be strongly indicated in any case by the broad shapes of these histograms which are typical of those seen in other experiments of the series and which stand in sharp contrast to the precise, sequential position of perturbations induced by the neurula/post-neurula shocks (Elsdale et al. 1976). Positions of abnormality in the present experiments are random, with the probability distribution shown. Somites between about 4 and 13 behind the ear vesicle are most liable to abnormality (rather than the very first ones) with significant incidence con- tinuing back to about somite 30. In control populations of 30 larvae (60 somite rows) from each of these egg batches, there were no spontaneous abnormalities before somite number 27. Thus the baseline was taken as zero, and all abnormalities in somites of ordinal number 20 or less considered significant and experi- mentally induced. Incidence of abnormality at any one location was probabilistic, with no correlation at first sight between abnormalities on right and left sides of the same embryo. These two features are again in marked contrast with those of the post-neurula sensitive period for shocks of the same order of duration, where perturbation is reliable and almost always bilaterally present at the appropriate level in the body pattern. Fig. 3 c shows the histogram for the subset of all abnormalities, recorded in Fig. 3 a, which were classified as minimal abnormalities. It shows that their incidence is somewhat biased towards somites of higher ordinal number in the series-the later formed ones (x2 = 21-4, 16 d.f. P = 0-05). The possible meaning of such an observation will be discussed below. The incidence of disturbances on right and left sides within embryos, while by no means closely tied as in the case of post-neurula shocks, is not completely unrelated as appears at first sight. Further analysis, taking into account the low probability of abnormality per somite in the population as a whole, shows that the same or adjacent somites are affected on both sides of an individual several times more frequently than would be expected by chance alone. This allows us to treat patches of abnormality involving the same or overlapping somite

19-3 290 J. COOKE

Table 2. Fit of abnormality incidence to a Poisson distribution in whole embryos between stages 9 + and 12

Experiment 1: total 105 Experiment 2: total 90 embryos embryos

Expected Observed Expected Observed Zero class 39 46 33 30 1 abnorm. 44 38 39 42 2abnorm. 17 18 14 12 3 abnorm. 5 3 4 6 positions on each side as single 'events' (see Discussion). When single events are so defined as to include such instances of bilateral abnormality, then the population of whole embryos can be treated as a population of 'targets' and classified into examples with zero, one, two and three 'hits' or perturbations. The fit of the populations in the two scored experiments to a Poisson distribution for perturbations is then found to be strikingly good, with no biases (Table 2). From this we can deduce that, within an egg-laying, embryos are quite homogeneous as to the susceptibility of their cells to perturbation, rather than including in- herently more resistant and more susceptible individuals. This is understandable, since although genetically heterogeneous in the nuclear sense, amphibian embryos from one female at gastrula stages may well be developing largely under the control of a maternally-inherited egg cytoplasmic system, whose character they would share. The fit to a Poisson distribution also allows the assertion that induction of one pertubation in a given embryo by heat shock is without effect upon the probability of further perturbations being induced elsewhere.

DISCUSSION The early sensitive period ends at around stage 12 (i.e. at some 13 h of development), some 2 h before the onset of the second period, which takes the form of a wavefront of temperature sensitivity that sweeps through the embryo from stage 13| (early neurula). Younger blastulae than stage 9 are in the early sensitive period, but in addition they suffer too much non-specific damage to make analysis practicable. Thus, during the early period, a hidden perturbation can be induced which may persist to affect formation of somites that are made from between 5 and 10 h later (the first formed somites behind the ear vesicle), up to 19-24 h later (somite 18-20, say). During the second, long sensitive period following stage 13|, perturbations are always located in tissue due to form somites some 2-3 h after the time at which shock is delivered (Elsdale et ah 1976). These position-specific abnormalities, following shocks given in post-neurula development (Elsdale et al. 1976), Somite abnormalities in pre-neurula stage o/Xenopus 291 indicate the progression of a wavefront of pattern-forming activity affecting successive groups of cells. All pre-somite cells might be temperature sensitive, but the perturbation is only registered permanently in that region of the body where cells are currently responding to the wavefront, the period at high temperature rendering them incoherent in their response rather than, as normally, tightly spatio-temporally organized to give the pattern of somites and fissures. Such shocks may be enabling us to follow the progression of the hidden wavefront through the pre-somite cells, preceding their visible alteration in motor/ adhesive behaviour (actual somite formation) by 2 or 3 h at each level, or by the time taken to form some five somites in Xenopus. Cells which are further back down the axis at time of shock can return to normal intercellular communication and hence coherence by the time the wavefront involves them, thus again forming normal somites. Hence we need not assume that the perturbation, induced in pre-somite cells by post-neurula shocks, lasts very long on a developmental time scale. Cells merely become spatially and temporally unco-ordinated in some fashion for an hour or two in their response to any programmed onset of sudden change that may meanwhile occur. The response itself - rotation and the inception of the muscle-cell programme of differentiation - is still made. The implications, for some theories of the control of the somite pattern, were discussed by Elsdale et al. (1976). Abnormalities scored after the earlier temperature shocks, in the work presented here, have a different implication and interest. Yet any complete model of the dynamic intercellular organization of the embryo, enabling the generation of normal somite pattern, must take account of these also. The final effect of the perturbations is indistinguishable from that of the late-induced ones. Had cell death been involved, one would expect debris in the vitelline space, loss of material evident in anterior abnormalities, and regulation to give normal pattern for all later-formed somites (Cooke, 1975a). It appears, not that cells have died, but that a patch of cells has become and remained so disorganized amongst themselves that their formation into somite tissue has occurred in a spatially disturbed and disrupted way. Given that somites after about number 20 in the series are produced from tissue that is a very small cell group in the gastrula, we can say that somite material anywhere in the body pattern may be affected. To account for such abnormalities, we must assume that following early shocks, perturbations persist in cell-groups anywhere from 5 to 24 h, though a slow recovery process might begin to operate over this time (see Fig. 3 c, where later-formed somites tend to be part of smaller abnormalities). Susceptibility of cells to a given temperature shock falls off a little with development during gastrulation, before ending quite suddenly at the small yolk plug stage (see Table 1). Do the disturbed patches of tissue represent clones, derived from one initially perturbed cell and inheriting its perturbed state? This cannot be the case. Cell division in mesoderm at gastrula stages is already far too slow to account, in 292 J. COOKE the time available, for the tissue patch size in the anterior somite disruptions on this hypothesis. Several - to many hundred cells are involved. Furthermore, in posterior regions each formed somite includes fewer cells, yet the extents of abnormalities in terms of somite numbers are if anything less there. A clonal hypothesis demands that they be greater, due to greater elapsed time for cell division between temperature shock and participation by clones in somite formation. The significant association between perturbed somites on the two sides of the body suggests that, significantly often, the random patches of perturbed tissue include cells that are due to form somites on either side of the median notochord. Taken overall, data show a relatively greater incidence of abnormalities in the same or adjacent somites on either side of the midline, for more posterior body positions. Consideration of the mode of generation of the body by increasing growth posteriorly, from the mesodermal cell population existing as gastrula stages, renders this comprehensible. At the time when temperature shocks occur, rendering abnormal groups of cells of a particular average number or extent, both sides of the body pattern are more likely to be embraced in the presumptive fate of the group where this has much growth still to perform, in contributing to posterior regions of the body (see Holtfreter & Hamburger, 1955, also unpublished results on tail-bud growth in this laboratory). Thus bilateral abnormalities could result either because a perturbed tissue patch already lies mid-dorsally to span the presumptive notochord (anterior somites) or because it is situated mid-ventrally near the yolk plug and will subsequently split to migrate up and form somites on either side (posterior somites - see Keller, 1976). After stage 9+ (late blastula) shocks, occasional embryos are seen having massive areas of disrupted somite tissue, even where cell death has not been evident. Also, incidence of the minimal abnormalities is somewhat biased towards shocks delivered near the end of the gastrula sensitive period, and towards later formed somites after any given time of shock delivery. From this, a very tentative picture might be built up which sees these temperature shocks as inducing initially massive tracts of dynamically disorganized tissue, i.e. of cells amongst which a normally existing physiological communication or co- ordination has broken down. Such 'patches' may progressively diminish in size as a prolonged recovery to normal cellular co-ordination sets in. As the wavefront of some change in cellular activity later progresses down the body, causing somite formation, if it should encounter a remaining patch or group of cells the precise timing of whose responses to it cannot be well co-ordinated, then a local area of abnormal fissure formation, etc. will result. The aim of experiments of the type reported here is to place constraints upon theories that are considered for spatial organization of the pattern. Of course many models are open to us at present, postulating different kinds of cellular damage (destruction of membrane protein or integrity; delay of stage-specific Somite abnormalities in pre-neurula stages of Xenopus 293 gene activation, etc.) as incurred by the period of high temperature. But to be plausible they must each allow an explanation of how pattern disruptions of essentially identical nature, but with quite different spatial distributions and temporal relationships to the stage at shock, can be induced by shocks administered during two discrete development periods. Models must also account for the otherwise normal histodifferentiation of the spatially deranged cells. Elsdale & Pearson (unpublished), using Rana, observe an overall correlation between advancing time of gastrular shock and increasing ordinal position of the somites affected, which was not observed in the present Xenopus experiments. This leads them to a model where the initial perturbation is the mal-ordering of the precise sequence in which pre-somite cells progress into the gastrula interior and take up their positions, the patches of abnormality resulting from this. Cooke & Zeeman (1976) have elaborated a particular model for control of the somite pattern, and a very brief outline is finally given here of the way in which, on this model, the total observed results of temperature shocks might be accounted for. This 'clock and wavefront' model (Cooke & Zeeman, 1976; Cooke, 19756, 1977) postulates that all the pre-somite cells become physiologically phase- linked with respect to some oscillator whose rhythm interacts with a wavefront of rapid cell change, causing spatially regular delays and advances in the overt expression of that wave as cell behaviour. This underlies the spatially regular courses of fissures of cellular de-adhesion segregating successive somites. Further details of this model are inappropriate to an experimental paper, but the above characteristics of it allow us to imagine the possible effects of early temperature shocks. The difference in causes of pattern disturbances after late (stage 13| onwards) and early (stages 9+ -12) shocks is conceived of as follows. Late shocks would cause transient phase-shifting, or loss of amplitude and/or synchrony for the hypothetical oscillator in all pre-somite cells (see Winfree, 1970, 1975 for descriptions of the dynamics shared by a number of cellular oscillators whose biochemistry is unknown). Such disturbance would be transient, over a few cycles of the oscillator only (i.e. during the morphogenesis of a few somites), because the nature of such oscillators in their mature form would seem to be self-exciting (Winfree, 1970, 1975) and because, in the Cooke/Zeeman model, all cells are quite strongly linked by intercellular signalling of metabolic variables. Thus, only where the oscillator was in the process of interacting with the wavefront of cell activation during and just after the shock, would the inter-somite boundaries then in process of being set up be perturbed and disrupted in direction and spacing. Hence the defined, reliable zone of abnormal morphogenesis, distinctive in position to each embryonic 'age' at shock, that is seen (Elsdale et al. 1976). Early shocks, by contrast, are seen on this model as causing damage to the apparatus of the oscillator itself (e.g. enzymic proteins, membrane components, etc.) in precursor cells of the somites, leading to immaturity or retardation with 294 J. COOKE respect to development of the oscillator. Intercellular communication and co- ordination of the oscillator among cells of the developing embryo could also be prejudiced by the shock, retarding the build up of phase-linking. The result might be an embryo that is effectively a mosaic, being composed of tissue in a normal state of dynamic organization interspersed with tissue areas of various sizes in various states of slow recovery towards such organization over many hours. The wavefront of somite cell activation in such embryos, encoun- tering a random incidence of 'incompetent' patches of tissue in its passage down the body, might be giving the observed distribution of visible pattern abnormalities.

REFERENCES COOKE, J. (1975a). The control of somite number during morphogenesis of a vertebrate, Xenopus laevis. Nature, Lond. 254, 196-199. COOKE, J. (1975b). Experimental analysis and a theory of the control of somite number during amphibian morphogenesis. UCLA Symposium on Molecular and Cellular Biology, vol. n. (ed. McMahon and Fox), pp. 205-226. Squaw Valley: W. A. Benhamin Inc. COOKE, J. (1977). Control of somite number during amphibian development: models and theories. In Limb and Somite Morphogenesis, Symp. Soc. Devi Biol. (ed. Ede & Hinch- cliflfe), pp. 433-448. Cambridge University Press. COOKE, j. & ZEEMAN, E. C. (1976). A clock and wavefront model for control of the number of repeated structures during animal morphogenesis. /. theoret. Biol. 58, 455-476. ELSDALE, T., PEARSON, M. & WHITEHEAD, M. (1976). Abnormalities in somite segmentation following heat shock to Xenopus embryos. /. Embryol. exp. Morph. 35 (3), (1977). 625-635. HAMILTON, L. (1969). The formation of somites in Xenopus laevis. J. Embryol. exp. Morph. 22, 253-265. HOLTFRETER, J. & HAMBURGER, V. (1955). Analysis of Development (ed. Willier, Weiss & Hamburger), p. 230-297. Philadelphia: Saunders. KELLER, R. E. (1976). Vital dye mapping of the gastrula and neurula of Xenopus. Devi Biol. 51, 118-137. NIEUWKOOP, P. D. & FABER, J. (1956). Normal Table o/Xenopus laevis (Daudin). Amsterdam: North Holland Publ. Co. RUGH, R. (1962). Experimental Embryology. Minnesota: Burgess Publ. Co. WINFREE, A. T. (1970). An integrated view of the resetting of a biological clock. J. theoret. Biol. 28, 227-374. WINFREE, A. T. (1975). Unclocklike behaviour of the biological clocks. Nature, Lond. 253, 315-318. {Received 5 December 1977, revised 9 February 1978)