J. Cell Sri. 64, 255-264 (1983)
255
Printed in Great Britain © The Company of Biologists Limited 1983
TOROIDAL BANDS IN POLYTENE CHROMOSOMES
OF DROSOPHILA
VEIKKO SORSA
Department of Genetics, University of Helsinki, Finland
SUMMARY
Results obtained from the thin-section electron microscopy of salivary gland chromosomes of
Drosophila melanogaster mainly support the concept of cable-like organization of polytene
chromosomes, with disk-like bands composed of parallel bundles of homologous chromomeres.
Outward orientation of loop fibres may generally cause a toroidal bending in the chromomere
bundles. Both longitudinal and transverse sections of polytene chromosomes indicate that the bands
may contain toroidal subunits. Torus-shaped bands were only found in thin sections of the most
distal and most proximal regions, as well as in certain heavy bands at the late-replicating regions of
polytenized interphase chromosomes. This suggests that an incomplete duplication of chromomeres
may be a reason for torus formation, by preventing the separation of sister chromatids at the earliest
phases of the polytenization process. The appearance of more numerous, but smaller, subunits in
thin-sectioned faint bands is interpreted as a consequence of more complete segregation of sister
chromatids in those bands during polytenization.
INTRODUCTION
Recently, Mortin & Sedat (1982) reported results suggesting that most if not all the
bands of polytene chromosomes of Drosophila melanogaster are torus-shaped. According to Painter's (1934) review of the early history of studies of polytene
chromosomes, the idea of a toroidal organization of the bands was presented by
Leydig two years after Balbiani's first description of these striated ribbons in the
salivary gland cells of Chironomus. Several reports of both disk-like and torus-like
organization of the bands can be found in the early literature of polytene
chromosomes. Some investigators even tried to explain the banding pattern of
polytene chromosomes as a helical type of organization, which was generally found
to be present in dividing chromosomes, both in mitosis and meiosis (cf. Painter, 1934;
and reviews by Bauer, 1935; and Beermann, 1962).
The polyteny concept presented by Koltzoff (1934) also showed a preference for
toroidal organization of bands, while Bridges (1935) depicted the polytene
chromosomes as cable-like structures with disk-like bands. Indisputable evidence for
the disk concept of bands was presented by Bauer (1935). He demonstrated clearly
that in cross-sections of fixed and stained polytene chromosomes, as well as in optical
sections of chromosomes in living salivary gland cells, the bands are evidently filled
with groups of parallel chromomeres. According to Bauer (1935) longitudinal sections
also provided no evidence to prove the existence of a hollow axis in the polytenized
interphase chromosomes (cf. Beermann, 1962).
The hollow centre of the chromosomes, and particularly of the bands, if it does
256
V. Sorsa
exist, should be detectable in longitudinal thin sections cut along the long axis of
the whole polytene chromosome. However, electron micrographs of serial sections
along the central axis of salivary gland chromosomes of D. melanogaster, used for
the electron microscopic (EM) revision of Bridges' maps, have not provided us with
any evidence in favour of the torus concept as a general mode of organization of the
bands (cf. Saura & V. Sorsa, 1979a,b,c,d; Saura, 1980; V. Sorsa & Saura, 1980a,b;
V. Sorsa, Saura & Heino, 1983). Similarly, the earlier electron-microscopic observations on polytene chromosomes, fixed and prepared for the EM by using several
different methods, have also supported the disk-like organization of band chromatin
(cf. e.g. Swift, 1962; Sorsa & Sorsa, 1967, 1968; M. Sorsa, 1969; Berendes, 1970;
Lossinsky & Lefever, 1978; Semeshin, Zhimulev & Belyaeva, 1979; Mott, Burnett
& Hill, 1980; ten Tusscher & Derksen, 1982; Zhimulev, Semeshin, Kulichkov &
Belyaeva, 1982).
In the present report the possible existence of toroidal bands and their distribution
in polytene chromosomes has been studied in a large collection of electron
micrographs taken from thin sections of both unsquashed and squashed salivary gland
cells of D. melanogaster. On the basis of the results obtained some hypotheses are
presented for explaining the appearance and possible organization of toroidal bands
in polytene chromosomes.
MATERIALS
AND
METHODS
Salivary glands of third-instar larvae oiDrosophila melanogaster -werefixedwithin a few seconds
of dissection, using procedures that allow exact comparison of the banding patterns with the revised
reference maps of Bridges (cf. Lindsley & Grell, 1968), also at the electron-microscopic level.
Fixative used for this study were the following (v/v): (a) acetic acid/ethanol (1:3), (b) acetic
acid/methanol (1:3), (c) formic acid/methanol (1:3), (d) formaldehyde (4%) in insect Ringer.
Fixation was carried out in a refrigerator at +4°C and fixation time was approx. 1 h for the acidic
fixatives and approx. 15 min for the formaldehyde-Ringer. Some of the acetoethanol-fixed glands
were embedded after dehydration and staining, without squashing. Some material was embedded
after a light squashing. Most of the salivary glands were squashed on silicon-coated slides and
Fig. la-d. Electron micrographs of thin-sectioned salivary gland chromosomes of D.
melanogaster. a. Partially cross-sectioned tip of X chromosome shows toroidal shape of
bands close to the telomere (t). Acetomethanol fixation and uranyl acetate staining, b.
Toroidal group of chromatin fibres (arrow) at the chromocentre area close to the
centromere of chromosome 4. c. An axial section of the distal end of 2L chromosome
showing strongly concave bands close to the tip (t) of chromosome arm. The cup-shaped
bands obviously appear as closed toroids in certain types of cross-sections indicated by the
parallel lines (cs). Nucleosome fibres (indicated by thin arrows) and some cross-sections
and side-views of larger coils (thick arrows) can be recognized. Formaldehyde(4%)Ringer fixation and uranyl acetate staining, d. Cross-section of the heavy doublet 3C 2-3
at the proximal end of the inversion zeste +64b9 (stock obtained from M. M. Green) shows
closed toroids in both homologues (white arrows). The torus-shaped doublets 3C 2-3
include some DNA from the rightmost end of originally triplicated white +R sublocus
attached to the remnants of band 12B 9 at the proximal breakpoint of this inversion. Most
of the chromometric material of ring bands proposedly belongs to the late-replicating
region demonstrated by Arcos-Teran (1972). Acetomethanol (1: 3) fixation, uranyl acetate
staining, cf. Fig. 3a. Bars, 0-1 ^m.
Toroidal bands in polytene chromosomes
257
prepared for the EM as previously described (V. Sorsa, 1983a,b). The EM studies were carried out
in the Department of Electron Microscopy of the University of Helsinki, using a Philips 200 EM
operated at 80keV and the micrographs were taken on 35 mm fine-grain film.
\12B
1A
Fig. 1
258
V. Sorsa
RESULTS AND
DISCUSSION
As has been shown at the light-microscopic level (Bridges, 1935), and also in
electron micrographs (Fig. la; 2d, e), most of the less-prominent bands are split into
parallel groups or bundles of homologous chromomeres (cf. also V. Sorsa & Saura,
1980a,6; V. Sorsa, I982a,b). The cross-sections of heavy bands in unsquashed
polytene chromosomes (Fig. 2a, b) usually show several hollow centres, suggesting
that those bands are formed by several toroidal subunits. Thin sections of distal ends
of chromosome arms show that many bands are strongly concave (Fig. lc), as demonstrated previously by Bauer (1936) (cf. also Beermann, 1962). The cup-shaped bands
may appear as rings in certain cross-sections, as shown in Fig. lc. The actual shape
of the chromatin rings that appear in thin sections of the proximal ends of polytene
chromosomes is difficult to demonstrate (Fig. lb). Occasionally, a torus-like organization of chromatin can be also found in certain heavy bands in the middle of
chromosome arms (Fig. 2c). Clear evidence for the existence of torus-shaped bands
comes from the heavy doublet 3C 2—3 in the X chromosome (Fig. Id). In this frontally
sectioned band at the proximal end of the inversion zeste +64b9 the asynapsed groups
of parallel chromatids, originating from the paired parental strands, seem to form two
closed rings of chromatin. In the less-prominent bands the chromomeres often appear
bent or as toroidal groups (Fig. 2d-i). In general, the results of the present study seem
to suggest that a toroidal organization of chromatin appears in subunits of bands rather
than in the entire bands of polytene chromosomes (cf. Sorsa & Sorsa, 1968).
According to the model proposed by Mortin & Sedat (1982), the axial fibres in both
the interbands and bands form a wide hollow cylinder in the polytene chromosome.
The chromomeric loops are located mainly peripherally in this tube. The diameter of
the axial tube would thus be essentially dependent on both the number and the
thickness of individual axial fibres in the polytene chromosomes. Simple calculations
for estimating the diameter of the axial cylinder, for example in a polytene
chromosome of D. melanogaster, representing the polyteny degree of 512 strands per
homologue, suggest that the 'hollow axis' model may work only if: (a) the axial fibres
are less than 5 nm thick, or (b) the degree of polyteny, at least in the axis, is lower than
512 per homologue. For instance, if the interchromomeric fibres are supposed to be
approx. 30 nm thick helices of nucleosomes (Laird, Ashburner & Wilkinson, 1980),
the diameter of a hollow axis formed by a single layer of 512 such fibres would be
Fig. 2a-i. Toroidal structures in the thin sections of unsquashed (a and b), slightly
squashed (c) and well-squashed (d-i) salivary gland chromosomes of D. melanogaster.
a, b. A general view of a thin section of an unsquashed cell (a) and a higher magnification
of the cross-sectioned heavy band (b) showing four toroidal subunits in a band (white
arrows), c. An open-ended toroid in an unidentified heavy band. Acetoethanol fixation
(1: 3), and uranyl acetate staining. Bars: a, 1 /mi; b, c, 0-2^m. d, e. A general view of an
axial thin section through the region 4D-9F of the X chromosome. The borders of
divisions and subdivisions are marked by thin arrows according to the revised reference
map of X (Bridges, 1938). Circled areas (indicated by arrowheads f-i) are presented as
larger magnifications in f-i showing loop-like organization of chromomeres in many
bands, cf. Fig. 3b—c. Bars: d—e, ljUm; f-i, 0-1 (Jin.
Toroidal bands in polytene chromosomes
Fig. 2
259
260
V. Sorsa
Fig. 3a—c. Schematic representation of proposed structural organization in a torus-shaped
band (a), and in a tightly packed narrow band (b), composed of 32 parallel bundles of eight
chromomeres per homologue. Interchromomeric fibres are depicted only in some of the
bundles, c. Proposed organization in an individual bundle of eight chromomeres each
containing approx. 75 nucleosomes.
Toroidal bands in polytene chromosomes
261
roughly 5/im, even if the axial fibres are placed close together. In the polytene
chromosomes flattened by squashing the axial tube alone should be even wider. An
illustration of highly polytenized salivary gland chromosomes of Chironomus, having
a degree of polyteny of approx. 8000-16000 (cf. Beermann, 1962), reveals quite
unreasonable results if the hollow axis model of Mortin & Sedat (1982) is adopted.
Conversely, if the hollow axis model is used for a rough estimation of the polyteny
degree, e.g. in the torus-shaped doublet 3C 2-3 shown in Fig. Id, it gives the following results. In the cross-section of band 3C 2-3 the inner contour length of chromatin
rings representing the homologues is approx. 500 nm. According to the structural
model proposed by Mortin & Sedat (1982), it corresponds to the axial tube formed
by the layer of axial fibres in the polytene chromosomes. The contour length (500 nm)
is equal to approx. 250 diameters of DNA double helix or to about 16 diameters of
'solenoid-sized' (30 nm) chromatosome helices. According to the thin-section EM of
squashed polytene chromosomes of Drosophila the average thickness of individual
interchromomeric fibres is about 5 nm (cf. V. Sorsa, 1982a,6). By using this average
diameter (5 nm) for the axial fibres the hollow axis model reveals a degree of polyteny
between 64 and 128 for the toroidal bands (3C 2—3) shown in Fig. Id.
The cross-sectional area of the ring bands 3C 2-3 (Fig. Id) is about 0-2/im2 per
homologue. This value can be used for a rough estimation of the degree of polyteny
by comparing the average axial length of the doublet 3 C 2-3, which is approx .0-4 /xm
in similarly prepared X chromosomes, with the axial length of the fraction of the same
band at the left end of the inversion, in (1) zeste +64b9 (stock obtained from M. M.
Green). This fraction of band 3C 2 includes a complete white"1" locus (cf. V. Sorsa,
Green & Beermann, 1973) attached to a small fraction of band 12B 9. Thus the band
contains at least 15 X 103 bases of DNA in each parallel chromomere. The comparison
shows that the chromatin content of doublet 3C 2—3 is roughly 10 times the chromatin
content of the white"1" locus band in the inversion X. Accordingly, each individual
chromomere of the doublet 3C 2-3 should contain at least 150 X 103 bases of DNA,
which presumably is capable of forming approx. 750 nucleosome units. The crosssectional area of such chromomeres is approx. 1700 nm2, if tightly filled with nucleosome fibre, or approx. 2200 nm2, if filled with 30 nm thick chromatosome helices
formed of six nucleosomes per turn. Thus the maximum number of parallel
chromomeres, i.e. the degree of polyteny, in the ring bands shown in Fig. Id is about
118-91 per homologue. However, the densitometry of EM negatives representing
thin-sectioned band chromatin indicates that the solid chromatinfibresconstitute only
about 40 % of the total volume of thin sections (V. Sorsa, 19826). This suggests that
the real chromomere content and, correspondingly, the number of parallel
chromatids, if they are completely duplicated, is only about 47—36 per homologue in
the ring bands 3C 2-3 shown in Fig. Id. Arcos-Terdn (1972) has demonstrated that
the region of the X chromosome including the doublet 3C 2-3 contains late-replicating
DNA, which may explain the estimated low degree of polyteny in these ring bands.
It has been proposed that, due to retarded duplication of DNA the polytenization
cycles and segregation of sister chromatids have not been completed in the
chromomeric loops of this doublet. 'Normal' polytenization of the distal fragment of
262
V. Sorsa
1
3C 2-3 containing the white" " locus at the other end of the inversion indicates that
several potential replication origins may exist in the DNA of the doublet 3C 2-3, but
replication is prevented towards the weak point between 3C 2-3 and 3C 5-6. The
appearance of closed or open-ended toroids in the less highly polytenized regions at
the chromocentre and at the telomeres (Fig. la-c), as well as in the late-replicating
region 3C 2-3 (Fig. Id) suggests that retarded replication of DNA may be a reason
for the formation of torus-shaped bands. In other words, the existence of a single
toroidal group consisting of all the parallel chromomeres in a polytene chromosome
band indicates that the duplicated sister chromatids have not been able to separate
from each other even after the earliest cycles of polytenization. Correspondingly, an
appearance of four toroidal elements in a band (Fig. 2a, b) may be explained as a
consequence of incomplete segregation of duplicated sister chromomeres after the
four-stranded stage of polytenization etc. Evidently, the greater amount and more
repetitive nature of the DNA may also increase the compactness of those bands or of
smaller toroidal groups.
Because cross-sections of polytene chromosomes showing an entire disk of parallel
chromomeres composing a narrower band are extremely rare in electron micrographs
and the identification of cross-sectioned regions is almost impossible until the thinsectioning methods are refined, the number of parallel bundles and the organization
of fainter bands have been studied only in longitudinal sections. Thus it should be
emphasized that only those fixation methods that allow the exact identification of
bands in thin-section EM have been used in this study. This methodological requirement should not make it impossible to detect the hollow centres of bands if they exist
generally in polytenized chromosomes. The thickness of thin sections (=50— 80 nm)
is evidently much lower than the expected diameter of the central holes in bands,
which means that the possible central hole should be detectable in several serial
sections along the chromosome. Analysis of electron micrographs does not support the
existence of a single 'hollow axis' in polytenized chromosomes. Although a centrally
located hole can be found in certain prominent bands (e.g. in the subdivisions 5B, 6F,
7A, 8C and 9EF in Fig. 2d-e), most probably the explanation is a local asynapsis of
homologues. However, in certain tight bands, like 7D 1—2, the location of a distinct
break in longitudinal thin section may be interpreted as the centre of a rather large
toroidal unit, comparable to the structures seen in cross-section in Fig. Id or Fig. 2b.
Excepting the late-replicating regions, the degree of polyteny, i.e. the total number
of parallel chromatids, seems to be roughly constant through long stretches of salivary
gland chromosomes, which is also indicated by the rather uniform diameter of the
chromosome (cf. Laird, Ashburner & Wilkinson, 1980). The remarkable variation
found in the organization of adjacent bands is seemingly caused by the differential
grouping of parallel chromomeres. The highest countable number of parallel groups
is usually approx. 25 per thin section, which apparently corresponds to a total number
of approx. 250-500 parallel bundles per band in the whole chromosome, depending
on the degree of flattening of the chromosome region. Because this estimate is lower
than the expected full polyteny degree (1024) it suggests that the sister chromomeres
derived from the last or from the two latest cycles of polytenization have not separated.
Toroidal bands in polytene chromosomes
263
In many bands the chromomeres form even larger groups. In a large number of
Bridges' bands the chromatin seems to be so tightly packed that counting of parallel
subunits is impossible.
The essential results of the present study may be summarized as follows. In thinsection electron microscopy of salivary gland chromosomes of D. melanogaster the
toroidal organization of whole bands was found only in the less-polytenized regions.
The appearance of ring-shaped bands is proposed to be a consequence of incomplete
or retarded duplication of chromomeric loops. This, together with a larger amount
and greater degree of repetitiveness of DNA may prevent the segregation of sister
chromomeres and cause the grouping of chromatids even at the early phases of
polytenization. In fainter bands composed of smaller chromomere units the
chromatin consists of a number of separate groups, which is interpreted as a
consequence of easier segregation of duplicated chromatids during the early cycles of
the polytenization process. Peripheral location of chromomeric loops may cause a
toroidal bending also in smaller groups of parallel chromomeres (cf. Fig. 3).
The study was supportedfinanciallyby the National Research Council of Sciences of Finland.
I thank Dr Virpi Virrankoski-Castrodeza, Ms Anja O. Saura, Ph. Lie. and Ms Riikka Santalahti for
skilful thin-sectioning of the chromosome material.
REFERENCES
ARCOS-TERAN, L. (1972). DNS-Replikation und die Natur der spat replizierenden Orte in
X-Chromosom von Drosophila melanogaster. Chromosoma 37, 233-296.
BAUER, H. (1935). Der Aufbau der Chromosomen aus den Speicheldriisen von Chironomus Thummi Kiefer (Untersuchungen an der Riesenchromosomen der Dipteren. I.). Z. Zellforsch. nrikrosk.
Anat. 23, 280-313.
BAUER, H. (1936). Structure and arrangement of salivary gland chromosomes in Drosophila species.
Proc. natn. Acad. Set. U.SA. 22, 216-222.
BEERMANN, W. (1962). Riesenchromosomen. Protoplasmatologia, 6/D. Wien: Springer Verlag.
BERENDES, H. D. (1970). Polytene chromosome structure at the submicroscopic level. I. A map of
region X, 1-4E of Drosophila melanogaster. Chromosoma 29, 118—130.
BRIDGES, C. B. (1935). Salivary chromosome maps with a key to the banding of the chromosomes
of Drosophila melanogaster. J. Hered. 26, 60-64.
BRIDGES, C. B. (1938). A revised map of the salivary gland X-chromosome of Drosophila
melanogaster. J. Hered. 29, 11-13.
KOLTZOFF, N. K. (1934). The structure of the chromosomes in the salivary glands of Drosophila.
Science, N.Y. 80, 312-313.
LAIRD, C. D., ASHBURNER, M. & WILKINSON, L. (1980). Relationship between relative dry mass
and the average band width in regions of polytene chromosomes of Drosophila. Chromosoma 76,
175-189.
LINDSLEY, D. L. & GRELL, E. H. (1968). Genetic variations of Drosophila melanogaster, publ. 627.
Washington, D.C.: Carnegie Instn of Washington.
LOSSINSKY, A. S. & LEFEVER, H. M. (1978). Ultrastructural banding observations in region
1A-10F of the salivary gland X-chromosome of Drosophila melanogaster. Drosoph. Inf. Serv. 53,
126-131.
MORTIN, L. I. & SEDAT, J. W. (1982). Structure of Drosophila polytene chromosomes, evidence
for a toroidal organization of the bands. J. Cell Sci. 57, 73-113.
MOTT, M. R., BURNETT, E. J. & HILL, R. J. (1980). Ultrastructure of polytene chromosomes of
Drosophila isolated by microdissection. J. Cell Sci. 45, 15-30.
PAINTER, T . S. (1934). Salivary gland chromosomes and the attack on the gene. J. Hered. 25,
465-476.
264
V. Sorsa
SAURA, A. 0 . (1980). Electron microscopic analysis of the banding pattern in the salivary gland
chromosomes of Drosophila melanogaster. Divisions 23 through 26 of 2L. Hereditas 93, 295-309.
SAURA, A. 0 . & SORSA, V. (1979a). Electron microscopic analysis of the banding pattern in the
salivary gland chromosomes of Drosophila melanogaster. Divisions 21 and 22 of 2L. Hereditas
90, 39-49.
SAURA, A. O. & SORSA, V. (19796). Electron microscopic analysis of the banding pattern in the
salivary gland chromosomes of Drosophila melanogaster. Divisions 30 and 31 of 2L. Hereditas
90, 257-267.
SAURA, A. O. & SORSA, V. (1979c). Electron microscopic analysis of the banding pattern in the
salivary gland chromosomes oiDmsophila melanogaster. Divisions 37, 38 and 39 of 2L. Hereditas
91, 5-18.
SAURA, A. O. & SORSA, V. (1979<i). Electron microscopic analysis of the banding pattern in the
salivary gland chromosomes oiDmsophila melanogaster. Divisions 27, 28 and 29 of 2L. Hereditas
91, 219-230.
SEMESHIN, V. F., ZHIMULEV, I. F. & BELYAEVA, E. S. (1979). Electron microscope autoradio-
graphic study on transcriptional activity of Drosophila melanogaster polytene chromosomes.
Chromosoma 73, 163-177.
SORSA, M. (1969). Infrastructure of the polytene chromosome in Drosophila melanogaster with
special reference to electron microscopic mapping of chromosome 3R. Ann. Acad. Set. Fenn. ser.
A, IVBiol. 151, 1-18.
SORSA, M. & SORSA, V. (1967). Electron microscopic observations on interband fibrils in
Dmsophila salivary chromosomes. Chromosoma 22, 32-41.
SORSA, M. & SORSA, V. (1968). Electron microscopic studies on band regions mDmsophila salivary
chromosomes. Ann. Acad. Sci. Fenn. ser. A, IVBiol. 127, 1-8.
SORSA, V. (1982a). An attempt to estimate DNA content and distribution in the zeste-white region
of the X chromosome of Drosophila melanogaster. Biol. Zbl. 101, 8 1 - % .
SORSA, V. (19826). Volume of chromatin fibers in interbands and bands of polytene chromosomes.
Hereditas 97', 103-113.
SORSA, V. (1983a). Interband fibrils of polytene chromosomes. Cell Differ. 12, 137-142.
SORSA, V. (19836). Electron microscopic mapping and ultrastructure of Dmsophila polytene
chromosomes. In Insect Ultrastructure 2 (ed. H. Akai & R. C. King). New York: Plenum.
SORSA, V., GREEN, M. M. & BEERMANN, W. (1973). Cytogenetic fine structure and chromosomal
localization of the white gene in Dmsophila melanogaster. Nature, new Biol. 245, 34—37.
SORSA, V. & SAURA, A. O. (1980a). Electron microscopic analysis of the banding pattern in the
salivary gland chromosomes of Dmsophila melanogaster. Divisions 1 and 2 of X. Hereditas 92,
73-83.
SORSA, V. & SAURA, A. O. (19806). Electron microscopic analysis of the banding pattern in the
salivary gland chromosomes oiDmsophila melanogaster. Divisions 3, 4 and 5 of X. Hereditas 92,
341-351.
SORSA, V., SAURA, A. O. & HEINO, T . I. (1983). Electron microscopic analysis of the banding
pattern in the salivary gland chromosomes of Dmsophila melanogaster. Divisions 6 through 10
of X. Hereditas 99, 181-200.
SWIFT, H. (1962). Nucleic acids and cell morphology in Dipteran salivary glands. In The Molecular
Contml of Cellular Activity (ed. J. M. Allen), pp. 73-125. New York: McGraw-Hill.
TEN TUSSCHER, B. & DERKSEN, J. (1982). The fourth chromosome of Chimnomus Malpighian
tubules. Chromosoma 85, 643—658.
ZHIMULEV, I. F., SEMESHIN, V. F., KULICHKOV, V. A. & BELYAEVA, E. S. (1982). Intercalary
heterochromatin in Drosophila. I. Localization and general characteristics. Chromosoma 87,
197-228.
{Received 26 April 1983-Accepted 8 June 1983)