ABSTRACT
Na, K and Cl measurements have been made on frozen sections of chick red blood cells throughout embryonic development, using electron-probe microanalysis. There is an apparent fluctuation in the levels of these elements during maturation, although the Na/K ratio remains fairly constant. The nuclear Na concentration resembles that of the cytoplasm, rather than that of the medium, at all stages. Inhibitor studies indicate that when cytoplasmic Na, K and Cl levels are altered, their corresponding nuclear levels are similarly affected. Additionally, the measurements in nuclei isolated in anhydrous media from lyophilized cells have shown arte-factual accumulation of high Na, K, Ca and Mg.
INTRODUCTION
Chick red blood cells provide a suitable system for the study of terminal differentiation. During development of the chick embryo, 2 populations of red blood cells are produced whose maturation is asynchronous: the first erythrocytes (primitive series) are made exclusively in extra-embryonic tissue, while the second population (definitive series) is produced in the yolk sac after about 4 days of development, and later by the liver and spleen (Romanoff, 1960; Lemez, 1964). By 8 days, although the primitive series consists almost entirely of mature erythrocytes (Small & Davies, 1972), they constitute only 10 to 20% of the erythroid cells in the blood (Dawson, 1936); the remaining erythroblasts of the definitive series now begin maturation, but less synchronously.than-the-primitive-red blood cells (Lemez, 1964).
The maturation of both primitive and definitive erythrocytes is marked by a series of striking morphological changes (Small & Davies, 1972) which are associated with the almost complete genetic inactivation of the cells (Seligy & Neelin, 1970). The proliferating erythroblasts from days 4 and 5 of embryonic development give rise to maturing erythrocytes which are incapable of division; this quiescence is accompanied by the loss of nuclear proteins (Dingman & Sporn, 1964; Gershey & Kleinsmith, 1969), by cessation of DNA synthesis (both replicative and unscheduled) (Cameron & Prescott, 1963; Darzynkiewicz, 1971), by a marked reduction in RNA synthesis (Cameron & Prescott, 1963; Madgwick, Maclean & Baynes, 1972; Zentgraf, Scheer & Franke, 1975) and by a loss of DNA supercoiling (Cook & Brazell, 1976). Despite this, both RNA and DNA polymerases are present (Scheintaub & Fiel, 1973; Longacre & Rutter, 1977) and it has been postulated that changes in the genetic activity of these cells are related to altered patterns of association between regulatory proteins and DNA (e.g. Ruiz-Carrillo, Waugh, Littau & Allfrey, 1974). There is considerable evidence from prokaryotes that interactions between specific DNA sequences and regulatory molecules are determined by the ionic environment (Anderson, Nakashima & Coleman, 1975; Lin & Riggs, 1972). Thus, precise quantification of electrolyte levels within the cell nucleus may demonstrate the nature of the environment which controls the conformation and activity of chromatin (Lezzi & Gilbert, 1970; Rensing & Fisher, 1975). Our aim has been to determine whether the maturation of the chick erythrocyte is associated with changes in the levels of Na and K in the nucleus and whether these levels can be altered by means of inhibitors of the Na/K pumping systems located in the plasma membrane (Hoffman, 1966; Glynn, 1968).
MATERIALS AND METHODS
Collection of the red blood cells
White Leghorn eggs were incubated at 39 °C. After various intervals red blood cells were collected from embryos as follows: (1) the blood vessels of 4-day embryos were teased apart and the fluid collected; (2) the blood vessels, leading from the yolk sac to embryos of 6–12 days of age, were cannulated using microsyringes; (3) blood was withdrawn directly from the heart of 15-to 18-day embryos, using a hypodermic syringe. These methods reduced contamination of the blood cells by allantoic and amniotic fluids. The cells were collected and pooled from 6–20 embryos from each age group, washed twice in Dulbecco’s modification of Eagle’s minimal essential medium (Gibco) and resuspended in medium plus 10–20% (w/v) Dextran (mol. wt 230000; Sigma). Dextran was added to the medium to reduce ice-crystal damage during quench-freezing, to improve section cutting at low temperature, and also to form a suitable matrix so that the sectioned medium could be used as a peripheral standard for the quantification of X-ray data (Gupta, Hall, Maddrell & Moreton, 1976; Gupta, Hall & Moreton, 1977a). Some of the cells were incubated in medium with either 10−4 M ouabain or 10−4 M ouabain plus 10−4 M ethacrynic acid for 6–24 h at 39 °C. As the binding of ouabain to cells depends on the external K concentration (Boardman, Lamb & McCall, 1972; Baker & Willis, 1970), the K in the medium was kept at the same level (8 mM) throughout inhibitor treatment for each age group. Control cells were incubated in Dulbecco’s medium alone. The cells were washed and resuspended in medium plus Dextran as before.
Preparation of sections of whole cells for microanalysis
A small droplet of the cell suspension was placed on the end of a copper rod of 1 · 5 mm diameter. Each specimen was plunged into Freon 13 (monochlorotrifluoromethane) at —181 °C and stored in liquid-nitrogen until required. 1-μm-thick frozen sections were obtained, mounted and stored, as described by Gupta et al. (1977a). The specimens were loaded onto the modified cold stage (—170 °C) (Taylor & Burgess, 1977) of a JEOL JXA-50A electron-probe microanalyser, where they were additionally protected by an anticontamination cap cooled by liquid-nitrogen.
The sections were examined in the microanalyser in the scanning transmission mode at an accelerating voltage of 40–50 kV. A beam of 2–10 nA, focused to a diameter of 100-300 nm, was used either as a static probe or to scan a small rectangular area (raster) localized within cytoplasm or nucleus at magnifications of 5000 to 40000 times. Heterogeneity of electrolyte concentration within subcellular compartments is more likely to be revealed by static probes than by scanning rasters. Each field was analysed for 80 s real time. Although the spatial resolution is expected to be of the order of 300 nm (Hall, 1975; Gupta et al. 1977a), the beam or specimen or both may drift during measurement, thereby producing anomalous results. This is of especial importance if the area to be measured lies at an interface, e.g. nucleus/ cytoplasm or cytoplasm/medium. The centres of nuclear and cytoplasmic fields were therefore generally selected for study. Nevertheless, some data were collected from dense areas which lay on the periphery of nuclei of older chick red blood cells, and which probably corresponded to regions containing condensed chromatin.
Characteristic X-rays for Na, Mg, K and Ca were recorded on diffracting spectrometers (Gupta et al. 1977a) and the complete X-ray energy spectrum was obtained simultaneously from a Kevex Si (Li) energy dispersive spectrometer. The latter provided information about both the characteristic X-rays from K, Cl, P and Fe, and the continuum or white radiation which was used as a measurement of local mass of the specimen under the beam (Gupta et al. 1977a).
Ideally, microanalytical data should be obtained from sections in the hydrated state (Gupta et al. 1977a; Gupta, Berridge, Hall & Moreton, 1978a). Unfortunately, hydrated sections are poor in contrast, and, as the cells under investigation were scattered widely throughout each section, they were difficult to resolve. Therefore, a compromise was reached whereby the majority of sections were examined after dehydration for 1-2 min under high vacuum in the microanalyser column (Gupta et al. 1976), but a few sections were analysed in the hydrated state for comparison. The equations used in the quantification of X-ray data to determine elemental concentrations are given by Gupta et al. (1977a, 1978a).
Preparation of isolated nuclei for microanalysis
A method for the isolation of nuclei by nonaqueous means which yields clean nuclei retaining certain water-soluble constituents and high enzymic activities was selected to minimize extraction and redistribution of free nuclear electrolytes (Kirsch et al. 1970; Gurney & Foster, 1977). Erythrocytes were collected from embryos as described and centrifuged in capillary tubes. The top half of each packed red cell column was discarded and the remaining cells placed in a plastic Petri dish which was immersed in Freon 13 chilled to its freezing point by liquid-nitrogen. The frozen samples were lyophilized at —40 °C over a PSOS trap for 4 days at 10−3torr (0 · 133 N m-1). Desiccated glycerol at approximately 2 °C was added directly to the lyophilized red cells while maintaining vacuum. Homogenization of the red cells and collection of nuclei were carried out essentially as described by Kirsch et al. (1970). Droplets of nuclei suspended in glycerol were smeared at 2 °C over aluminized nylon film supported by Duralium collars (Gupta et al. 1977a, 1978a). Excess glycerol was removed and the nuclei w’ere topcoated with a thin aluminium film prior to mounting into the microanalyser. X-ray data were recorded as described above. Quantification of these data, using external Na, K, Ca and Mg standards, is given in Appendix I (p. 79).
Interferometry
Red blood cells were collected from embryos after 4 days and 18 days of development as indicated above. They were examined with a Baker double refracting interference microscope using the shearing optical system and a half shade eye-piece (Baker, 1957), with monochromatic illumination (λ = 546 nm). The retardation of light produced by nucleus and cytoplasm was measured with respect to the bathing medium. The dry mass per unit volume of nucleus and cytoplasm was calculated according to Barer (1955).
RESULTS
Typical X-ray energy spectra taken from saline, nucleus and cytoplasm in sections are shown in Figs. 1-4. The concentrations of Na, K and Cl from such samples are given in mM kg−1 dry wt in Table 1.
Within the limitations of the technique, the average level of nuclear Na declines in the population from 4 days to 6 and 8 days (P < 0·1 %), returning to its previous level by 12/13 days (P < 01 %) from which time it remains fairly constant. There is a peak in K concentration at 12/13 days (P < 0 ·1 %). Cl concentrations also fluctuate during development; a substantial decrease from 4 days to 6–8 days (P < 0 ·1 %) follows the decline in both Na and K, but after this the level increases and remains relatively constant throughout the rest of development.
The average levels of Na in those areas of nucleus and cytoplasm which were measured are similar within the age groups, even at 6 and 8 days when electrolyte concentrations have dropped. K, on the other hand, is similar in concentration in nucleus and cytoplasm only at 4 and 6 days, but, later in development, the nuclear level increases and is significantly higher than that of the cytoplasm (P < 0 ·1 % for 8 days, 15/17 days and 18 days). There is a decline in nuclear K at 15/17 days (P < 0 ·1 %) after which time the level remains at the value seen at 4 days. The cytoplasmic K concentration does not change significantly from 15–18 days.
The amount of water in embryonic red blood cells taken from chicks of various ages and treatments is presented in Table 2. These data are subject to a number of technical uncertainties (Gupta, 1978; Gupta et al. 1977a, 1978a; Gupta, Hall & Moreton 1977b), but probably provide a good approximation to the amount of water present. In an attempt to check these measurements, red blood cells taken from embryos of 4 days and 18 days of age were studied further using an interferometric method (Baker, 1957). The results are presented in Table 3. Both sets of data indicate that there is an overall increase in nuclear dry mass per unit volume during maturation (Table 3(b)), and hence in the degree of hydration.
The dry mass determinations (and therefore the estimation of wet wt and H2O fractions) are necessarily carried out in the same regions as are used for X-ray quantification of electrolytes; thus it is possible to convert electrolyte concentrations from mM kg-1 dry wt into approximate mM kg−1 wet wt concentrations (Appendix II, p. 81) or into mM l. −1 H2O (Table 4) assuming that all the measured electrolyte elements are free and all water is solvent.
In contrast to the data from sectioned erythrocyte nuclei the Na and K concentrations in nuclei isolated by a nonaqueous method are extremely high, and in nuclei from older embryos, at least, they are very variable (Table 5).
Inhibitors of Na pump(s) such as ouabain and ethacrynic acid are known to increase the concentration of cellular Na (Whittembury & Fishman, 1969; Lubowitz & Wittain, 1969; Leblanc & Erlij, 1969; Lamb & McCall, 1972; Kennedy & De Weer, 1976; Williams, Withrow & Woodbury, 1971; Hoffman & Kregenow, 1966; Pichon & Treherne, 1974; Glynn & Karlish, 1975; Kregenow, 1977). We have therefore determined whether treatment of erythrocytes with these inhibitors results in changes in the concentration of cytoplasmic and nuclear Na and K. Ouabain treatment alone (10−4 M for 24 h in standard growth medium) had little effect on the concentrations of any of the elements investigated in 6-day cells. Moreover, even in 6-day cells treated with both ouabain and ethacrynic acid (10−4 M), no statistically significant change in nuclear Na was observed, although a decrease in nuclear K was found (P < 5 %). When older red blood cells were treated with ouabain and ethacrynic acid, the cytoplasmic Na concentration increased dramatically (P < 0·2 % at 12/13 days, P < 0·1% at 18 days) with a concomitant decline in K (P < 5 % at 12/13 days, P < 0·1 % at 18 days) (Tables 1 and 4). The nuclear levels of Na and K reflect a change in cytoplasmic concentrations of these elements (for Na: P < 0·1 % at 12/13 days; for K: P < 0 · 1% at 12/13 days and 18 days). Cl levels were unaffected by combined inhibitor treatment at 12–13 days, but in older cells cytoplasmic Cl was greatly affected (P < 1 %) while nuclear Cl increased only slightly (P < 5 %).
The accuracy of the conversion of X-ray analytical results from rπM kg-1 dry wt into rπM kg−1 wet wt depends upon various assumptions, e.g. (1) that the thickness of the section is even, (2) that there is no mass loss during data collection (Gupta et al. 1977a, 1978a). However, the ratios of the elements present are independent of these factors. A histogram is therefore presented from which the ratios of Na, K and Cl in nucleus and cytoplasm throughout erythrocyte maturation can readily be estimated (Fig. 5).
Analysis of the characterisitic X-ray counts produced by phosphorus indicates that its distribution between nucleus and cytoplasm changes during development. As the phosphorus present in cells is not ‘free’ in the same sense that electrolytes are, no attempt has been made to quantify the amount of phosphorus present. Instead, a comparison has been made between the ratios of characteristic X-ray counts for phosphorus to continuum counts in nucleus and cytoplasm as these ratios should indicate the relative concentrations (Table 6). Up to 12–13 days, the nuclear and cytoplasmic phosphorus/continuum X-ray count ratios are not significantly different within the age groups. In older cells, however, the values diverge, presumably reflecting the decreased RNA metabolism in the cytoplasm.
Iron is found in nucleus and cytoplasm throughout erythrocyte maturation, in agreement with the fact that haemoglobin is distributed in both these compartments (Davies, 1961; Small & Davies, 1972).
DISCUSSION
The electrolyte measurements described in this work have been made largely on frozen dehydrated erythrocyte sections, and as such are subject to the limitation that ion translocation has occurred during dehydration. However, whenever parallel measurements have been made on hydrated sections, elemental concentrations were found to be very similar, giving us confidence in the data. Moreover, microprobe studies from this laboratory (Gupta et al. 1976, 1977a, 1978a, b) and other laboratories (Dörge et al. 1975; Somlyo, Shuman & Somlyo, 1977; Appleton & Newell, 1977) on a variety of tissues have shown that if frozen sections are carefully dehydrated at very low temperatures, no translocation of diffusible electrolytes can be detected, except in spaces lacking an organic matrix. Dehydrated sections were generally chosen for microprobe analysis because of superior image resolution (Fig. 6); nevertheless the specific location of the static probe or scanning beam within a dehydrated nuclear section could only be described rarely, and no conclusions can be drawn about the variation in electrolyte concentrations and their possible association with subnuclear structures. Most microprobe measurements were made in the central regions of the nucleus to avoid overlap with cytoplasm. Since the majority of condensed regions in these nuclei are peripheral (Small & Davies, 1972) the measured electrolyte concentrations probably apply to central decondensed regions. These regions appear to have a water content similar to the younger, more active nuclei.
Although a considerable amount of data exists on cellular electrolyte levels, there has been much disagreement about the nuclear ionic environment and in particular about the amount of Na present. Microprobe analysis of sectioned erythrocytes from various embryonic ages reveals that nuclear levels of Na, K and Cl are significantly different from the external medium. K is elevated while Na and Cl are reduced, essentially in agreement with microprobe studies on other somatic cell nuclei (Gupta et al. 1976, 1977 0, 1978 0; Gupta, Hall & Naftalin, 1978b; Pieri et al. Cameron, Sparkes, Horn & Smith, 1977; Rick, Dörge, von Arnim & Thurau, 1978), with nuclear electrolyte levels in amphibian oocyte and fish neurons (Riemann, Muir & McGregor, 1969; Dick, 1976; Century, Fenichel & Horowitz, 1970; Katzman, Lehrer & Wilson, 1969), and also with measurements of electrochemical activities in a variety of cells (Palmer & Civan, 1977; Civan, 1978). Our studies reveal that in terms of concentration, the dominant nuclear cation, measured in situ, is K rather than Na. There is also an apparent anion deficit which may be compensated in part by the presence of polyanionic nuclear proteins (Bhavanandan & Davidson, 1975; Hunt & Oates, 1977).
The results are consistent with the hypothesis that the dramatic nuclear maturation in the developing erythrocyte is not associated with a major change in the total concentrations of Na and K, despite the loss of considerable nuclear protein mass. K remains the major cation throughout maturation and the nuclear Na/K ratio is similar in primitive proliferating 4-day cells and in mature definitive 18-day erythrocytes (Fig. 5). The apparent decline in nuclear ion levels at 6 days (100% primitive series) and 8 days (20% primitive series) may be artefactual in that only 3 and 7 nuclei respectively were studied. It is also possible that the cells at this stage are sensitive to Dextran. However, the ratios of nuclear Na: K: Cl in each case are similar to those of the cytoplasm, and also to those from other stages of development (Fig. 5). If this decline is real, it suggests that the new definitive series is low in both nuclear and cytoplasmic ions, especially Na, at the start of maturation, and might therefore be under considerable osmotic stress —a problem which, as yet, remains unresolved. Both Na and K levels are restored by 12/13 days (90% definitive series).
The data obtained from sectioned nuclei conflict with those from nuclei isolated from chick erythrocytes albeit by techniques aimed at minimizing electrolyte loss and redistribution (Kirsch et al. 1970). The concentration of Na in the isolated nuclei is variable, but in most cases, is increased up to 20 times the concentration in sectioned cells, although K levels are not raised to the same degree. Nuclei from a variety of sources isolated by either the non-aqueous Behrens technique, or the glycerol procedure, uniformly show extremely high levels of nuclear Na (Siebert & Langendorf, 1970; Kirsch et al. 1970; this paper) which can be increased still further to 600 mM/kg dehydrated material by incubating the isolated nuclei in Na-containing salines (I to h & Schwartz, 1957). We therefore consider these high levels to be artefactual, resulting from exposure of the nucleus during cell fractionation to extracellular fluids containing high Na, and not a reflexion that the nucleus in the intact cell is in direct communication with the medium (see Moore & Morrill, 1976).
The electrolyte levels in nuclei can be changed profoundly in intact cells by the inhibition of membrane-associated Na-K ATPases. Simultaneous treatment of erythrocytes with the inhibitors ouabain and ethacrynic acid results in massive increases in nuclear Na to a level approaching that of the external medium. Our results suggest an absence of ouabain-sensitive sites in 6-day erythrocyte membranes; ethacrynic acid-sensitive sites appear to be present throughout maturation. The response of erythrocytes to Na/K pump inhibitors varies somewhat during maturation: erythrocytes from older embryos accumulate more Na and show greater reduction in K levels, perhaps indicating that the number of membrane pumps and/or their efficiency declines with increasing maturation/ageing. A comparable result has been obtained by Palmer & Civan (1977) working with Chironomus salivary gland cells. After ‘ageing’ these cells by prolonged incubation in vitro, they found an increase in intracellular ion levels. Pieri et al. (1977) find a similar gradual deterioration in the functions of the Na/K pump of ageing rat liver and brain cells. It is also possible to replace K by Na as the dominant nuclear cation if cells are incubated in medium lacking K (Gupta et al. 1978 b). What effects these changes would have on nuclear metabolism and whether these effects (if any) are reversible remain to be explored, although depletion of K arrests cells in division (Meeker, 1970).
The technical problems associated with electrolyte measurements in the sectioned nuclei of single-cell suspensions do not yet permit precise statements to be made about the intranuclear localization of ions during erythrocyte maturation. There is some variability in electrolyte microanalytical data from these nuclei, which may indicate that areas of condensed chromatin, like metaphase chromosomes (Cameron et al. 1977), have different electrolyte characteristics and dry mass from the rest of the nucleus.
Changes in the electrolyte levels of the medium lead to gross changes in chromatin condensation and deoxyribonucleoprotein (DNP) fibril dimensions (Hughes, 1952; Robbins, Pederson & Klein, 1970; Brasch, Seligy & Setterfield, 1971; Kellermayer & Jobst, 1970; Leake, Trench & Barry, 1972; Coutelle et al. 1974; Neelin, Mazen & Champagne, 1976) which suggests that DNP packing depends largely on modifications of electrostatic interactions along the nucleohistone backbone, observations which are supported by studies on the organization of isolated nucleoprotein in relation to ionic strength (e.g. Bartley & Chalkley, 1973). Maximum contraction of nucleoprotein is observed at a monovalent cation concentration of 0 · 15 to 0· 2 M (Bradbury, Carpenter & Rattle, 1973; Bradbury, Danby, Rattle & Giancotti, 1975; Billet & Barry, 1974), presumably when electrostatic repulsion along the nucleohistone molecule is minimized. At low ionic strengths, electrostatic repulsion would elongate the molecule, while at high ionic strengths histones are dissociated, which would also result in extension of the nucleoprotein.
Moreover, there is clear evidence that histones are involved in chromatin condensation, although the nature of Hi-DNA interactions in relation to histone modification and changes in ionic strength (Renz & Day, 1976) are not fully understood. In avian erythrocytes it is possible that H5 is the most important regulatory histone since it accumulates during maturation, in part replacing Hi (Sotirov & Johns, 1972), prohibits transcription and (disputedly) promotes condensation (Seligy & Neelin, 1970; Lurquin & Seligy, 1972; Bolund & Johns, 1973). However, the involvement of H5 or other histones in any of these processes will depend on the establishment of histone-DNA electrostatic and hydrophobic bonds, and histone-histone interactions whose stability will be related to the electrostatic charge density in the chromatin and therefore to the ionic milieu of the nucleus. Thus the organization of the heterochromatin nuclear bodies of the erythrocyte (possible chromosome equivalents (Anderson & Norris, 1960; Davies, 1961)) and DNP fibre dimensions, may depend primarily on charge interactions which can be modified by variations of nuclear electrolyte levels.
The data presented in this paper are concerned with total monovalent electrolyte concentrations. However, were all these electrolytes to be free in solution, the electrolyte concentration would still not on average exceed 0 · 2 M, i.e. histone-DNA interactions could exist as envisaged in the literature. The abnormally high electrolyte levels observed in the isolated nuclei would result in nucleoprotein dissociation and major changes in chromatin organization. The concentrations of divalent cations in erythrocyte nuclei remain to be determined; but it should be noted that they are far more effective in promoting changes in DNP organization than are monovalent cations (Bradbury et al. 1973, 1975). To gain a greater understanding of the distribution of electrolytes in nuclei, further information is needed concerning the amount and availability of cation-binding sites in the nucleus (e.g. alkali cation-binding sites described by Besenfelder & Siebert, 1975; and the Mg- and Ca-binding sites of metaphase chromosomes (Cantor & Hearst, 1970; Steffensen, 1961)), and also the amount of nuclear solvent water (Beall, Hazlewood & Rao, 1976).
We are especially grateful to Dr S. L. Schor, Dr A. M. Mullinger and Mr R. G. W. Northfield for help in early stages of this investigation, which is supported by grants from the Medical Research Council and the Cancer Research Campaign. R. T. Johnson is a Research Fellow of the Cancer Research Campaign. The Biological Microprobe Laboratory was supported by a grant from the Science Research Council to Drs B. L. Gupta, T. A. Hall and R. B. Moreton. We thank Mrs Kate Barber and Messrs. A. J. Burgess, N. G. B. Cooper, M. J. Day and P. G. Taylor for technical assistance at various stages. Our gratitude is also due to Professor M. M. Ci van for critically reading the manuscript.
APPENDIX I. QUANTIFICATION OF THE X-RAY DATA FROM ISOLATED NUCLEI
[utabl]
For the sectioned material in this report, the coefficients kx were obtained by using the peripheral medium around each section as the standard. In the case of the isolated nuclei there was no suitable peripheral medium and the coefficients were obtained as follows:
Potassium (both energy-dispersive and diffracting spectrometers): Standards were frozen-dried 1-μm sections of quench-frozen Dextran/saline solutions, containing 1000 mM of K per kg of Dextran.
Sodium: Spectrometer sensitivity was calibrated in proportion to potassium, by comparison of Na and K counts from frozen-dried 1-μm sections of quench-frozen Dextran/saline solutions containing equimolar NaCl and KC1.
Calcium: Spectrometer sensitivity was calibrated in proportion to potassium, by comparison of Ca and K counts from dried droplets of solutions of equimolar CaCl2 and KCL.
Magnesium: Spectrometer sensitivity was calibrated in proportion to sodium. Dried solutions of equimolar salts of Mg and Na proved to be unsuitable because of inhomogeneity and segregation. Consequently the efficiency for Mg relative to Na was determined from comparison of Mg and Na counts from bulk standards of pure Mg and NaCl respectively, via a set of estimated correction factors.
Observations on the bulk standard were done with a 20-W electron beam. The observed value for (SM/SN, was 8·45. At 20 W, the ratio nN/nM according to the equation of Green & Cosslett (1961, p. 1211), uncorrected for electron backscatter, is 1 · 41: the ratio nN-c/nN (equations of Hall, 1971, p. 228) is 0·412; the correction factor for differences in backscattering (Hall, 1971, p. 215) is 0·99; and the product nN/nMnN-c/nN is hence 0 · 576. Finally the ratio of absorption corrections fx-c/fM (from the equation of Philibert, 1963, modified by Duncumb & Shields, 1966, as quoted by Hall, 1971, p. 216) is 0·538. Substitution of these values into equation (4) gives the value 2·04 for the relative sensitivity (SM/SN)t. Application of this factor to the sodium coefficient fNa in the Table gives the coefficient fMg = 8600/2 · 04 = 4220.
APPENDIX II. ESTIMATION] OF LOCAL DRY-WEIGHT FRACTIONS, AND CONVERSION OF MEASUREMENTS FROM mM kg−1 DRY WEIGHT TO mM kg−1 WET WEIGHT
Our preferred method for the measurement of local dry-weight fractions (Gupta, 1978; Hall, 1978; Gupta et al. 1978a) is based on the comparison of X-ray signals from the selected region of the specimen before and after dehydration. Because measurements on hydrated specimens were not needed for the main objectives of the present study, almost all of the sections were already dehydrated prior to X-ray analysis. Therefore a different procedure was used to obtain the values listed inTable2.
Equation (9) is the basic equation used by Dörge et al. (1977).