Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Geophys. J . R. astr. Soc. (1976) 44, 601-623 Crustal Structure at the Continental Margin South of South Africa R. A. Scrutton (Received 1975 September lo)* Summary Four gravity profiles over the continental margin south of South Africa are interpreted in the light of existing seismic, gravity and magnetic results. There is a marked difference in shallow and deep crustal structure between the rifted segment of margin facing the Atlantic Ocean and the margin offset facing the Indian Ocean. Both margin segments have a mass excess beneath the continental rise and a mass deficit beneath the outer continental shelf and continental slope, but this distribution is much more pronounced at the margin offset. It can be explained by a rapid transition from continental to oceanic crust which, in the Indian Ocean, probably results from the mode of formation of the margin (strike slip faulting). In the Atlantic Ocean the cause of the rapid transition is less obvious. The shallow structure at the continent-ocean boundary changes around the margin. At the rifted segment, the sedimentary sequence of the continental rise north of 36" S gives way southwards to a basement high thought to be of volcanic origin. At the margin offset the Agulhas fracture zone appears to contain slivers of both oceanic and continental crust. The latter is in the form of a marginal fracture ridge. Beneath the continental shelf a high density lower crust may compensate for the low density sedimentary rocks in the Outeniqua Basin. I. Introduction One of the programmes of the International Decade of Ocean Exploration (IDOE) is a reconnaissance geophysical and geological study by the Woods Hole Oceanographic Institution of the Eastern Atlantic continental margin (EACM). Survey work began in January 1972, with Cruise 67 of RV Atlantis II working between Cape St Francis, South Africa (Fig. 1) and the Congo River. Continuous recordings of bathymetric, seismic reflection, magnetic and gravity data were obtained together with results from a number of expendable sonobuoy stations (Hoskins, Rogers & Woo 1975). Emery et al. (1975) have presented an overall appraisal and interpretation of these data. On Leg 3 of Cruise 67 four crossings (Fig. 2) were made of the continental margin around the Agulhas Bank south of South Africa. An interpretation of the gravity * Received in original form 1975 June 20. 601 602 R. A. Scrutton 330s 34 35 36 37 38 39OS FIG.1. Detailed bathymetric chart of the continental margin south of South Africa, redrawn from a chart compiled by E. S. W. Simpson at the University of Cape Town. Depths are in metres. The inset shows the position of the study area with respect to Africa. records obtained on the crossings is presented here as one of the more detailed studies of the EACM project. The traverses cross the ' corner ' of southern Africa, where two continental margins of markedly different origin meet (Scrutton et al. 1974). Facing the Atlantic Ocean is a rifted margin, but facing the Indian Ocean is a 1200 km long margin offset. With this study it is hoped to reveal some of the major differences in crustal structure between these two margins and provide evidence for their modes of evolution during and after continental break-up. The morphology of the sea floor around Agulhas Bank is shown in Fig. 1. It has been described by several authors (e.g. Simpson 1970; Scrutton et al. 1974), so only a brief comment on the prominent features is given here. In the Atlantic Ocean, west of 20"E, there is a narrow continental shelf, moderately steep (up to 5"), canyonated continental slope and a well-developed continental rise. These border the Cape Basin which lies at a depth of 4500 to 5000rn. In the southern part of the Cape Basin is the Cape Rise basement arch (Fig. 2) striking east-north-east towards Agulhas Bank. The bathyrnetry in the Indian Ocean is quite different and much more complex. The shelf break is often poorly defined, but if we use the 200-m isobath to define it, the shelf is seen to narrow rapidly eastwards from a width of 250 km south of Cape Agulhas to 50 km south of Cape St Francis. Where the shelf is widest, it terminates in a steep continental slope, but where it narrows, there is a wide, gently inclined upper 603 Crustal structure south of South Africa 34 34 35 35 36 36 37 37 38 38 393 17'E' 18 19 20 21 22 23 24OE 39% FIG.2. Structural elements of the continental margin defined by seismic reflection and seismic refraction data. Sediment thicknesses were obtained from the ' Atlantis I1 ' seismic reflection profiles using an interval velocity of 2-0 km s-'. Seismic refraction stations 4,5 and 6 are by Hales et al. (1970), I50 and 151 by Ludwig et al. (1968), G & H by Green and Hales (1966), S by Spence (1970) and Hales & Nation (1972), and the remainder (between Cape Town and Cape St Francis) by Leyden et al. (1971). The 200-m isobath is shown. slope and a narrow, steep slope (du Plessis & Simpson 1974). The continental rise is poorly developed, being largely made up of the Mallory Seamount chain paralleling the slope. Beyond the rise is the Agulhas Basin at over 5000 m depth and the Agulhas Plateau rising to less than 2500 m depth. The geological history of the region has been fully described by Dingle (1973b), Scnitton (1973a), Dingle & Scrutton (1974) and Scrutton et al. (1974). All the features to be studied here came into existence during Mesozoic and Cenozoic times. Immediately prior to the Mesozoic, the continental parts of the region existed within the Cape fold belt (Fig. 2) within the supercontinent of Gondwanaland. Falkland Plateau bordered Agulhas Bank at that time (Scrutton 1973a; DSDP Scientific Staff 1974) and was also part of the late Palaeozoic-early Mesozoic Cape fold belt (Greenway 1972). During late Triassic-early Jurassic times, a phase of approximately east-west tensional stress, possibly associated with the breakaway of east Gondwanaland from west Gondwanaland, caused subsidence and the initiation of the Outeniqua Basin on Agulhas Bank (Fig. 2). At this time, therefore, the faulted eastern flank of the Agulhas Arch, bordering the Outeniqua Basin, began to evolve. The arch itself consists of Palaeozoic strata folded on an east-south-east-west-north-weststrike underlain by R. A. Scrutton 604 c field 1 1 36 37 38 17'E 18 19 20 21 22 23 24'E FIG.3. Features of the magnetic field south of South Africa. The ' m ' sequence of lineaticns in the Atlantic Ocean is from Larson & Ladd (1973). ' M ' is from Talwani & Eldholm (1973) and is thought to mark the continent-ocean boundary. The remainder of the features are from du Plessis & Simpson (1974). Note the gently iindulating magnetic field with local intense anomalies over the Agulhas Arch and inner shelf and the quiet field over the Outeniqua Basin and marginal fracture ridge. The Cape slope anomaly marks the edge of oceanic crust in the Indian Ocean. The 200-m and 4500-m isobaths are shown. Precambrian granitic and metamorphic rocks. With the separation of South America and Falkland Plateau from Africa in the late Jurassic-early Cretaceous (Larson & Ladd 1973), both the Atlantic and Indian Ocean continental margins of Agulhas Bank were formed. The newly-formed Agulhas fracture zone truncated both the Agulhas Arch and the Outeniqua Basin. It is probable that the Agulhas marginal fracture ridge came into existence at this time. As soon as the South American block had broken away the new rifted margin segment west of Agulhas Arch would have attempted to subside. However, since Falkland Plateau remained in contact with the south-east edge of Agulhas Bank until mid-Cretaceous time, restraints would have been imposed on this subsidence. Also, subsidence of the evolving continental edge in the Indian Ocean would have been delayed. Only since the middle Cretaceous have the margins around Agulhas Bank been free to evolve in a tectonically inactive setting. A local igneous event took place on the Alphard Banks (Fig. 3) in the Palaeogene (Dingle & Gentle 1972), but otherwise crustal subsidence and sediment accumulation have predominated throughout the Upper Cretaceous and Cenozoic. Crustal structure south of South Africa 605 FIG.4. Contour chart of free-air gravity anomalies offshore, b s e d on Afhntis II traverses 10,12, 14 and 15, Graham &Hales (1965) and Talwani & Kahle (in press), and Bouguer anomalies onshore, redrawn from Smit er a[. (1962). Where the Arlanris ZZ traverses are dashed no gravity reccrd was obtained. The contour interval is 20 mgal. The 200-m isobath is shown. 2. Previous geophysical studies These studies have been reviewed for the whole of offshore southern Africa by Scrutton et al. (1974), Siesser, Scrutton & Simpson (1974) and Emery et al. (1975). It is now necessary to consider the seismic and previous gravity results from the Agulhas Bank region where they have a bearing on the following gravity interpretation. Two maps (Figs 2 and 3) summarize the seismic and magnetic work over the continental margin. The magnetic work is not considered in detail here. Seismic refraction The base of the continental crust beneath Agulhas Bank has been shown by Hales & Nation (1972) to be at a depth of about 32 km. An unusually high mean Pwave velocity for the topmost mantle (8.95 km s-') was observed, however, causing Hales and Nation to speculate on the possibility that a more normal P, phase of about 8 . 0 km s-' had been masked in their records. If such a phase were present, the calculated depth of 32 km to the 8.95 km s-l layer would be increased to about 35-5km and the top of the undetected 8.0 km s-' layer, which then becomes the redefined base of the crust, would occur between 30 and 33 km depth. This depth R. A. Scrulton 60 6 40 u) .20 ___ i" 0 Isostatic anomaly --20 ..... .' .., _dJ ..................... . . .... . . ._. . :. __.- Free-air anomaly: -Observed---- - ._ - 5 [-20 Calculated le 5, h [1.03] ................... (2,671 6-33 -___ 12,871 7z4 - * L3.41 , 400 100 200 300 0 Kilometres Profile 14 A -+sostatci anomaly ... . _.. ......... -20 Free - air anomaly: -Observed,--.-Colculated le ._.._ .. [1,031 .................................................. . - - A 7 ... s 1e.h 4 46 4 .2 8 4%+.35*34 &* 4.38 2.3: 5.33 1 12.871 500 400 g .-% 360 200 Kilometres FIG.5 100 0 Crustal structure south of South Africa 607 estimate is consistent with the value of 30 km for the thickness of the crust at the coastline obtained by Smit, Hales & Gough (1962) and Talwani (1962) from gravity data. On the basis of these data, T = 30 km is used in the gravity interpretation below. Within the continental crust five major layers can be recognized seismically. These are the postPalaeozoic sedimentary layer (V, < 4.0 km s-I), the Palaeozoic* layer (V, N 4 . 3 to 5.3 km s-'), the Precambrian* layer (V, N 5 . 8 km s-I), the main upper crustal layer (V, CI 6 . 2 km s-') and the main lower crustal layer ( Vp N 7-2kmsd'). The mean velocity of the sedimentary layer has been calculated by averaging all the observed velocities in the constituent layers (Green & Hales 1966; Ludwig et al. 1968; Spence 1970; Leyden, Ewing and Simpson 1971) weighted in proportion to the thicknesses of these layers. 2.72 km s-l was obtained as the mean. The density corresponding to this velocity on the extended Nafe-Drake curve (Ludwig, Nafe & Drake 1971) is 2 . 2 g cmd3. This is the value used in the following gravity calculations. The Palaeozoic layer varies in thickness between 1 and 6 km, but Hales & Nation (1972) have calculated that the underlying Precambrian layer also varies in thickness such that its base occurs at approximately 8 . 3 km depth throughout the Agulhas * The terms Palaeozoic and Precambrian are used for convenience; the ages of the rocks in these layers only approximate to these periods. 20 1 01 0 100 Kilometres FIG.5 200 608 R. A. Scrutton .g% -20 -0j - 20 ” Isostatic onomoly 4, ....... A,,, ...... . . + Ll+ k-: .. ............................. ’ 24 .:............................... ...... ......... .., , . I I 5.,J ~ .-. ,.-_I-_- I(i(, 0 --’/ 200 ... 300 .. 4 03 500 K ilornetrcj FIG.5. Profiles 15,14,12 and 10 of Atlantis ZI showing, from bottom to top in each case: an isostatic model of the Earth‘s crust including seismic reflection data on the thickness of the sediment layer (dashed lines), seismic refraction data (lu Ludwig ct a/. (1968), le Leyden et a/. (1971), g Green & Hales (1966), s Spence (1970), h Hales & Nation (1972), velocities in Icm s - l ) and densities in g ~ m (in- square ~ brackets);the observed free- air anomaly (dotted where interpolated from contour chart) and the anomaly calculated for the isostatic model; and the isostatic anomaly. Bank region. This horizon is incorporated in the crustal models in this study (Figs 5 and 6) and the Palaeozoic and Precambrian layers are grouped together into a single layer of mean density 2.67 g ~ m - This ~ . density is chosen because it is in common use for the upper part of the crust and because it corresponds to a P-wave velocity of 5.6 km s- which is intermediate between the velocities in the Palaeozoic and Precambrian Iayers. The main upper and lower crustal layers are also grouped together for the initial stages of the gravity interpretation. A density of 2.87 g cm-3 is assigned to these since this is the value commonly chosen for the bulk of the lower continental crust. It corresponds to a P velocity of 6.4 km s-’. Later in the gravity interpretation, however, the 7.2 km s-l layer is introduced as a separate entity. Hales & Nations’ (1972) results suggest that the layer occurs primarily beneath the Outeniqua Basin as a form of compensation for the low-density sedimentary rocks. A lower crustal layer of velocity about 7.4 km s-’ is frequently found below sedimentary basins and at inactive continental margins (Drake & Nafe 1968), providing part, if not all, of the isostatic compensation for the sediments at the top of the crust. There are no seismic refraction data on the deep structure of the oceanic crust west of the Agulhas Arch but there are three stations in the north-east Agulhas Basin (Fig. 2). The weighted mean velocity for the oceanic crust is 5.98 km s-’, which falls midway between the 5.6 km s-’ and 6.4 km s-’ velocities for the Palaeo- ’, 609 Crustal structure south of South Africa Profile 15 Free-air anomaly : 1-20 ___ Observed,----------Colculoted 3 10 [2.87] % c E ._ Y 20 30 2 00 300 400 100 0 Kilometres Profile 14 500 400 160 300 200 Kilometres F I ~6. 100 0 610 R. A. Scrutton 20! Profile 10 200 I00 Kilornetres 0 f - _0 .- __ -. I00 200 ~ .. ' __ 3 00 . 400 .. -~.-.. 500 K ilometres FIG.6. Profiles 15, 14, 12 and 10 of Atlantis II showing in each case: below, a crustal structuremodel based on the seismic and gravity data (densities in g ~ m - ~ ) ; above, the observed free-air anomaly and the anomaly calculated for the model. CSA is the position of the Cape slope anomaly. Crustal structure south of South Africa 61 1 zoic-Precambrian and main crustal layers beneath the continent. A two-layer model for the oceanic crust, laterally continuous with the 2.67 g cm-j and 2.87 g cm" layers beneath the continent, is therefore considered acceptable for the initial gravity interpretation. Seismic reflection In a number of areas, the geological basement crops out extensively on the sea floor; these areas are shown in Fig. 2. Those areas holding the greatest implications for the gravity interpretation are the Agulhas Arch antiform of Palaeozoic and Precambrian rocks (Gentle 1970), the Agulhas marginal fracture ridge (Scrutton & du Plessis 1973) and the Mallory Seamount chain. Isopachs of sediment thickness over the Agulhas Arch show the extent of the basement outcrop and the shape of the west and east flanks of the arch. On the west, straight contours suggest a rather simple slope in basement; on the east, uneven contours suggest downfaulting and the presence of minor anticlinal spurs extending into the Outeniqua Basin. There are thick deposits in the Outeniqua Basin, but a relatively thin sediment layer is developed beneath the continental slope on the west and little or no post-Palaeozoic material is present on the steep fault-controlled southern slope of Agulhas Arch (Emery et al. 1975). Beneath the deposits of the Outeniqua Basin a number of small basement ridges strike in a west-north-west to east-south-east direction as extensions of anticlinal arches in the Cape Fold Belt (Dingle 1973a). It is doubtful, however, whether they will have a noticeable effect on the gravity field, not only because of their small size but because the density contrast between Palaeozoic basement and the most deeplyburied sediments is probably low. At the lower continental slope the Outeniqua Basin is effectively closed by the Agulhas marginal fracture ridge (Scrutton & du Plessis 1973). The relationship between the ridge and the Agulhas Arch is at present unknown. The Mallory Seamount chain lies on oceanic crust and parallels the marginal fracture ridge. Du Plessis & Simpson (1974) have suggested that towards the southwest the seamounts pass into a small but continuous ridge that in places crops out on the sea floor. Together, the seamounts and ridge constitute the southern edge of the Agulhas fracture zone. Adjacent to the Agulhas Arch the fracture zone is in the form of a fault-bounded trough containing at least 2 km of low-density sediments (Scrutton et al. 1974). The marginal fracture ridge, seamount ridge, and trough all act as sediment traps confining detritus to the shelf and slope so that the crust in the Agulhas Basin carries only a thin veneer of sediment (Ewing et al. 1969). On the north side of the fracture zone, in the southern Cape Basin, oceanic basement is deeper, with about 2 km of predominantly clastic material on top of it (Emery et al. 1975). Towards the continental slope bordering the Cape Basin, the sediment thickness varies between 0 km and at least 3 km due to the presence of ridges and troughs in basement. Gravity Previous gravity surveys have revealed few details of the crustal structure of this continental margin. Graham & Hales (1965) presented a free-air anomaly contour map for the Agulhas Bank region, but their small amount of interpretation was hampered by a lack of seismic control. Even so, they were able to deduce that a steep gradient occurs in the Moho beneath the continental slope in the Indian Ocean. Hales & Nation (1972) made a further, brief interpretation of Graham and Hales' data to show that their new seismic results were not ' in serious conflict with the gravity data available '. No conclusions were drawn from this, however. Lamont-Doherty Geological Observatory have published their gravity data from the region in two forms: as a free-air anomaly contour chart (Talwani & Kahle, in press) and as isostatic anomaly profiles (Talwani & Eldholm 1973). The latter show 612 R. A. Scrutton that all round the Agulhas Bank a mass deficit exists beneath the continental slope accompanied by a mass excess beneath the continental rise. It is interesting to note that this mass distribution is the opposite of that discovered by early gravity measurements in the north-west Atlantic (Worzel 1965). Apparently, a mass deficit beneath the rise is not an ubiquitous feature of rifted margins. Emery ef al. (1975) have made a preliminary interpretation of the data used in this paper. 3. Gravity data A vibrating string accelerometer (VSA) mounted on a Sperry Mark 19 Mod 3C gyrocompass was used aboard Atlantis I1 to measure variations in the earth‘s gravity field. Data reduction to free-air and Bouguer anomalies (relative to the IGF 1930) took place on board ship. Details of the gravimeter installation are given by Bowin, Aldrich & Folinsbee (1972) and the data reduction system is described by Groman et al. (1972). It is thought that the accuracy of the gravity data is high, although there are no track cross-overs to check the internal accuracy of the survey. A satellite navigation system supplemented by ‘ Omega ’ and dead reckoning sensors provided navigational accuracy of If: 1 km or better, but the inherently high drift rate of the VSA may have introduced errors in the final gravity values. It is assumed that the free-air anomalies are accurate to + 2 mgal. During interpretation no attempt has been made to fit calculated gravity anonlalies to the observed to better than + 5 mgal. No regional gradients caused by variations in mantle structure or composition have been removed from the data during interpretation (See Regional Interpretation below). A contour chart of free-air anomalies has been constructed for the continental margin around Agulhas Bank (Fig. 4). The chart is redrawn from Graham & Hales’ (1965) and Talwani & Kahle’s (in press) charts, using the new Atlantis ZI data as control. There do not appear to be any systematic errors between the three sets of data; errors are randomly positive or negative and average out to approximately zero. In regions of steep gravity gradient, however, individual errors of up to 35 mgal are observed, probably as a result of navigational inaccuracies in the older data. In view of this degree of comparison a contour interval of 20 mgal is considered appropriate for the contour chart. Also shown in Fig. 4 are the Bouguer anomaly contours of the gravity survey of South Africa (Smit et al. 1962). Where these intersect the coast they were used as a further control on contouring the offshore data, knowing that Bouguer and free-air anomalies should be equal at the coast-line. 4. Gravity interpretation This has been carried out in three stages: a preliminary interpretation of the gravity values on a regional scale, a qualitative description and interpretation of the contour chart, and model fitting to the observed Atlantis 11, profiles 10, 12, 14 and 15. Regional interpretation In order to investigate the degree of isostatic balance of the region the mean of the free-air anomalies measured by Atlantis ZI was calculated. Although there is a general trend from positive free-air values over the shelf to negative values over the abyssal areas, reflecting the continental edge effect, the mean obtained was - 1 mgal. This is an indication that the region as a whole is in isostatic balance. Furthermore, this might show that there are no density anomalies within the mantle beneath this region. Evaluations of the regional free-air anomaly from two other sets of data, namely, Crustal structure south of South Africa 613 measurements of satellite orbital characteristics and older marine data, support this conclusion (Talwani 1971). Free-air anornaZy contour chart (Fig. 4) As is to be expected in a continental margin region the free-air anomaly chart is dominated by the continental edge effect. Nevertheless, important anomalies can be seen superimposed on this, indicating local variations in crustal structure. In the Atlantic Ocean the edge effect is of an amplitude typical of rifted margins, approximately 50 mgal peak-to-peak. The large positive anomalies observed north of Cape Town (Scrutton 1973b),thought to be due to density variations in basement, are not observed in this area. At 36" S over the continental slope the 0 mgal contour is offset approximately 100km in a left-lateral sense. The position of this offset correlates well with the landward continuation of a fracture zone (Fig. 3) invoked by Larson & Ladd (1973) to account for a small displacement of oceanic magnetic lineations in the Cape Basin. Since there normally are gravity anomalies associated with fracture zones and margin offsets (Cochran 1973), there could be a relationship between the gravity and magnetic displacements observed here. A small offset might, therefore, be present where the proposed fracture zone meets the continental edge at 36" S. On bathymetric grounds, the fracture zone itself would appear to be the northern boundary of the Cape Rise upwarp. The negative component of the continental edge effect becomes very deep south of the tip of Agulhas Bank where it is Aanked by steep gravity gradients of up to 8 mgal km-'. It is the net result of a large edge effect and the effect of the sedimentfilled depression within the Agulhas fracture zone. The gravity data support the contention that the depression is fault bounded. On the southeast side of the depression the steep gravity gradient coincides with the northern limb of the Cape slope magnetic anomaly (du Plessis & Simpson 1974) (Fig. 3), marking the faulted edge of oceanic crust at the Agulhas fracture zone. Just within the oceanic crust the Mallory Seamount chain produces a gravity high, peaking at about 150 mgal above Shackleton Seamount (Graham & Hales 1965). It is found that if the shape of this seamount is approximated ~ , seamount's topography can by a vertical cylinder model of density 2-83 g ~ m - the entirely account for the observed gravity anomaly. It seems unlikely, therefore, that there is any crustal thickening or a solidified magma chamber beneath the seamount chain, as has been observed elsewhere (Case et al. 1973). Towards the north-east the above-mentionedgravity anomalies are diminished as a result of the change in continental margin structure seen in Fig. 2 at about 22.5"E. Despite this change, the gravity high associated with the seamounts can be traced until it is again prominent beyond 24" E, where a small peak in the associated basement ridge may occur. The accompanying gravity low, north-west of the high, is also apparent east of 24" E where, rather surprisingly, it occurs in part over the marginal fracture ridge, which is a basement high. It reaches a peak value of -74 mgal over the ridge's south-eastern flank. Since the ridge has no magnetic signature (Scrutton & du Plessis 1973), the possibility arises of it being composed of predominantly sedimentary rock of moderate density, giving a reduced gravity effect. The gravity low would then of necessity be caused by a deeper structure related to the formation of the Agulhas fracture zone. On the continental shelf, over the crest of the Agulhas Arch, the general level of free-air gravity is +20mgal. This might appear low for an area of outcropping basement, but since the exposed rocks are predominantly granitic and sedimentary (Gentle 1970) it is a realistic value. We can see reflected in the gravity contours the relative complexities of the western and eastern sides of the Agulhas Arch, the eastern being the more complex. It is worth noting that no gravity anomaly is associated with the Alphard Banks igneous province situated on the northern side of Agulhas Arch a 614 R. A. Scrutton (Dingle & Gentle 1972) (Fig. 3). The free-air anomalies in this area are near to zero. Evidently, any magma chamber that fed the vents and plugs is small or no longer present in the crust, or the magma was derived directly from the upper mantle. To the east, the Cape fold belt anticlinal structures occurring beneath the narrowing shelf produce a weak west-north-west-east-south-easttrend in the contours near the coast. A small density contrast between the basement extensions and the most deeplyburied sediments of the Outeniqua Basin would account for the weakness of this trend. South-eastwards, near 353"S, 23YE, where the Outeniqua Basin is about 2 km deep and basement has dropped to 4 km below sea level, the gravity effect of the anticlinal extensions is lost altogether and low gravity values occur. An extremely quiet magnetic field is consistent with the greater depth to basement in this area (Fig. 3). High gravity values, up to 34mga1, characterize the shelf region southwest of Cape St Francis, but the reason for them is not apparent. Since there is no thinning of low density sediments here, density variations in basement might be the cause. Profile interpretations Two-dimensional crustal structure models have been constructed for Atlantis ZZ profiles 10, 12, 14, and 15. Profiles 12 and 14 cross the margin near to the right-angled tip of Agulhas Bank where the assumption that structures are two-dimensional is not perfectly valid. It is estimated, however, that because of this the calculated anomalies for traverses 12 and 14 will only be about 5 mgal too high. Fortunately, errors of this magnitude will not bother us, because a fit between calculated and observed profiles of no better than 5 mgal is acceptable. To remove the overshadowing edge effect from the observed free-air anomalies, and thus detect departures of the crust from perfect isostatic equilibrium, an Airy isostatic model of the crust was designed for each traverse. The free-air values due to the models were subtracted from the observed values to give isostatic anomaly profiles. A standard section for the crust, from which the models were derived, was established at the coastline where analyses of gravity and seismic survey data have shown that the crust is most probably 30 km thick and comprised of a layer 8 . 3 km thick of density 2.67 g cm-3 and a layer 21 - 7 km thick of density 2-87 g cm-'. The mass of a column of unit area of this standard crust together with 5 km of mantle of density 3.4 g cm-3 is approximately equivalent to the mass of other crust or crust plus mantle standard sections 3 5 k m in thickness (Heiskanen & Vening Meinesz 1958; Worzel & Shurbet 1955; Woollard 1959; Talwani, Sutton & Worzel 1959), that is lo7 g m for a 1 cm2 column. Fig. 5 shows the crustal models and their associated gravity profiles. For traverses 10 and 12 the calculated and observed profile are matched at the coastline but for traverses 14 and 15 they are matched over the Agulhas Arch where basement comes within 80 m of sea level. Also shown in Fig. 5 is the seismic control discussed earlier in the paper. At the top of the crust sediment bodies are outlined and within the crust refracting interfaces are indicated. In constructing the final crustal structure models (Fig. 6) use was made of this control, firstly by calculating and removing from the isostatic gravity anomalies the effect of the sediment layer, then modifying the structure of the remainder of the crust to compensate for the sediments and satisfy the isostatic anomalies. Throughout this process other constraints on the structure were also considered. Limiting depths to compensating and anomalous features were calculated from the isostatic anomalies using Bott & Smith's rules (1958). It was found that, theoretically, all the major anomalies could be satisfied by adjusting the topography of the Moho or varying the density of the lowermost crust. In practice, however, this would have produced a near vertical Moho in places, which is clearly unrealistic. In general, therefore, the bulk of any anomaly was satisfied by adjusting the Moho or density of the lower crust to a realistic extent and the remainder of the anomaly was Crustal structure south of South Africa 615 accounted for by adjusting the mid-crustal structure. No adjustments were made to the structure of the sediment layer since the seismic data define it well enough for the purposes of this interpretation. General features of the profiles and sections. The four isostatic anomaly profiles in Fig. 5 vary in shape over the continental shelf but have much in common over the slope and upper continental rise. A pattern consisting of negative isostatic anomalies over the slope and positive anomalies over the rise is observed, although this is developed to varying degrees on different profiles. Those profiles across the continental margin in the Atlantic Ocean have anomalies of smaller amplitude and gentler gradients than those in the Indian Ocean. An inspection of Talwani & Eldholm’s (1973) profiles reveals, however, that large anomalies and steep gradients occur again north of traverse 15. Between the negative and positive anomalies is a slope in the isostatic gravity profiles that Talwani & Eldholm believe marks the boundary separating continental from oceanic crust. We can see from Fig. 6 that this is in fact the case, and that it is particularly clear in the Indian Ocean, on profiles 10 and 12, where the gradient coincides with the Agulhas fracture zone (Cape slope anomaly). Talwani & Eldholm (op. cit.) also think that the isostatic low on the continental side of the boundary is due to the subsided but uncompensated edge of continental crust, with perhaps, a thick section of sediments beneath the slope. This could also be true, but it is less obvious on our crustal sections. In order to satisfy the isostatic low along the Indian Ocean margin, where the slope is totally or partially free of sediments, it was indeed necessary to place the Moho at a deeper level (move it seawards) beneath the slope. In the Atlantic Ocean, however, where the sediment layer is thicker, any uncompensated deepening of the Moho is masked by the compensation for the sediments. No explanation is given by Talwani & Eldholm for the isostatic gravity high on the oceanic side of the boundary. From the interpretation presented here it would appear that in general a shallower-than-normal Moho beneath the continental rise could account for it. Shallow structure undetected by the seismic data or density variations in the crust or mantle could also be causes, but refraction data in the Indian Ocean support the presence of a thin crust. Thus, whereas the continental edge has subsided to excess, the adjacent oceanic crust may have been restricted in its subsidence. This situation leads to a steep gradient to the Moho at the continental edge and has implications for the evolution of both segments of this inactive margin. These will be discussed later. When the isostatic profiles are corrected for the effect of the sediment layer, high anomalies are obtained over the continental shelf. These correspond in extent to the Outeniqua Basin and are indicative of compensation for the low-density sediments in the basin. Hales & Nation’s (1972) seismic refraction results suggest that the form of the compensation is an increase in density of the lower crust with no crustal thinning. This is shown schematically in Fig. 6 as a near-vertical-sided layer of density 3-15 g ~ r n - ~The . development of this high density phase in the lower crust may have accompanied, and possibly initiated, the subsidence of the Outeniqua Basin. Since Dingle & Scrutton (1974) have pointed out that the basin began as an intracontinental depression in early Mesozoic time, we could envisage the 3 - 15 g cm-3 compensating layer being locally present at the base of the crust during the formation of the continental margin and the Agulhas fracture zone at the end of the Jurassic period. This possibility is important to the evolution of the offset margin. Individual features of the profles and sections. Traverse 15 reveals the simplest crustal structure. To obtain a reasonable fit between the observed and calculated gravity profiles it was found unnecessary to resort to introducing mid-crustal structure. There is, however, scope in the fit to allow for the presence of a thick section of sediments beneath the slope as suggested by Talwani & Eldholm (1973). The increase 616 R. A. Scrutton in amplitude of their isostatic anomalies to the north, mentioned above, suggests that traverse 15 lies near the southern end of such a structure. The fracture zone proposed by Larson & Ladd (1973) rnight terminate the structure at 36" S. It was also found unnecessary to incorporate a layer of high density in the lower continental crust beneath the Outeniqua Basin, probably because the north-east end of the traverse only just enters the north-western tip of the basin. Oceanward of the shelf break a match with the observed gravity was obtained by thinning the crust by up to 4 km, although small density changes could have been involved. From the resulting structure the continent-ocean boundary would appear to be beneath the base of the continental slope. Traverse 14, crossing the margin between the fracture zone of Larson & Ladd (1973) and the Agulhas fracture zone, shows a very different structure, with no scope for a thick sequence of sediments beneath the slope and rise. A region of crust of intermediate thickness occurs beneath the continental rise and appears to be bounded to the north and south by the two fracture zones. At this point the Cape Rise meets CONTINENTAL CRUST ......... ........., SEDIMENTARY B A S I N ON C O N T I N E N T A L CRUST THICK O C E A N I C ml T H I N N E D , UPLIFTED & M E T A M O R P H O S E D C O N T I N E N T A L CRUST _-- CONTINENTAL 000 VOLCANIC MAJOR CRUSTAL CRUST FRACTURES EDGE IN ATLANTIC OCEAN RIDGE FIG.7. Sketch of proposed crustal types at the continental margin south of South Africa. Crustal structure south of South Africa 617 the continental edge. If, as has been suggested, the rise is the trace of the Bouvet Island mantle plume (Johnson, Lowrie & Hey 1973), the crust of intermediate thickness can be interpreted as the landward limit of a swathe of thicker-than-normal oceanic crust produced by excessive volcanic activity. A very good example of this type of feature is the Greenland-Iceland-Faeroes Ridge (Bott, Browitt & Stacey 1971) in the North Atlantic. On Section 12, a narrow zone of intermediate thickness crust is included beneath the sediment-filled depression between the Cape slope magnetic anomaly and the edge of the continental crust. It constitutes the Agulhas fracture zone at this point. There is clearly one major fracture abruptly truncating the thin oceanic crust at the Cape slope anomaly, while there may be another major fracture against true continental crust. A variable mid-crustal structure is incorporated in this profile in order to satisfy the steeper gravity gradients, and this emphasizes the location of the major fractures. The origin of this zone of crust of intermediate thickness is discussed in the next section. Traverse 10 crosses the Outeniqua Basin just west of its closure against the Recife Arch (Fig. 7), a feature similar to the Agulhas Arch, and provides a section across the Agulhas marginal fracture ridge. Beneath the ridge the crust is 23 km thick, giving the impression that it is an integral part of the continent. It is possible a major fracture occurs between the ridge and true continental crust, but it is not clear from the gravity data. In front of the ridge we see a structure similar to that on traverse 12: a major fracture abruptly truncates an upturned, thin oceanic crust. This fracture lies beneath the Cape slope anomaly and within the Agulhas fracture zone. 5. Discussion of margin structure Structure at rifted margin segment The gross nature of the continent-ocean transition west of the Agulhas Arch is revealed by traverse 15. A relatively thin sediment layer on the continental shelf passes seawards into a thicker layer on the continental slope and rise. Overcompensation for this sediment thickening is apparent at the base of the crust where there is a rapid seaward rise in the Moho to a depth less than the Airy isostatic depth beneath the continental rise. Comparison of this transition with others at rifted margins shows that it is rather different. Off eastern North America (Emery et al. 1970) and western southern Africa (Rabinowitz 1972; Scrutton 1973b; Emery et al. 1975),for example, there are thicker sediment wedges across the margins and relatively high isostatic anomalies occur over the outer shelf and slope. The anomalies can be totally or partially accounted for by the presence of a basement ridge beneath the shelf break or an increase in basement density, neither of which are found off the Agulhas Bank. Any anomalies remaining after this shallow structure and its compensation has been taken into consideration tend to show a mass deficit beneath the continental rise, the opposite to that found here. The observed configuration of the Moho west of Agulhas Bank, together with the presence of a mass excess beneath the rise, is therefore somewhat unusual and has implications for some of the mechanisms put forward by various authors for the evolution of rifted continental margins. These mechanisms are thermal evolution, as proposed by Sleep (1971), which leads to a relative elevation of the Moho as both continental and adjacent oceanic crust cool and contract; sediment loading and consequent crustal flexure, as advanced by Walcott (1972), leading to a relative depression of the Moho beneath the continental shelf, slope and rise; transfer of material by crustal creep from beneath the shelf and s!ope to beneath the rise, as propounded by Bott (1971), leading to a relative elevation of the continental Moho and depression of the oceanic Moho; migration of the Moho in response to sea-floor 618 R. A. SCrnttoa spreading rate changes at the mid-ocean ridge, causing an overall elevation or depression; and interpretation of the Moho as the basalt-eclogite phase change whose position changes in response to sediment loading at the top of the crust or temperature variations at the base of the crust, which might be applicable to continental regions but not oceanic (Wyllie 1971). It would appear that none of these mechanisms operating alone can explain restricted subsidence of the Moho beneath the continental rise together with its excessive depression beneath the shelf and slope as observed here. Perhaps two, or more of them, operating together, for instance thermal contraction and migration of the basalt-eclogite phase change, can explain it. If so, the structure discovered here would support the hypothesis that a number of mechanisms operate simultaneously as an inactive continental margin evolves (Rona 1974), some outweighing others in their effect. Deep structure at ofSset margin segment The isostatic gravity anomalies and their associated structures seen on the rifted segment of margin are even more pronounced on this offset segment. Particularly apparent is the abrupt change across the Agulhas fracture zone from crust of continental thickness to crust of oceanic thickness. Moho gradients are steep and they could have been made steeper by omitting the mid-crustal structure. However, it is primarily the lack of thickening of the oceanic crust as the continental edge is approached that produces the abrupt transition. In fact, the faulted edge of oceanic crust appears to be upturned at the Agulhas fracture zone, causing further shallowing of the Moho. An explanation of this accentuated structure might lie in the fact that a margin offset perpendicularly links two tension-formed rift zones. In this situation it is not subjected to all the margin-forming processes that affect rifted margin segments. Firstly, the offset segment does not undergo crustal stretching but, instead, experiences shearing stresses. Secondly, in the central part of the offset, away from the rift zones, it is not subjected to the high temperatures that characterize rifts for a prolonged period of time (c. 50 My) prior to crustal splitting. The crust at the central part of the offset therefore does not experience significant thermal expansion and contraction. Thirdly, an offset probably does not receive the overall heavy sedimentation that produces the thick sediment wedges seen at rifted margins. This is because sediment traps, such as the marginal fracture ridge, commonly develop at the continental slope of margin offsets (Francheteau & Le Pichon 1972), thus preventing sedimentation on the continental rise. It is likely, therefore, that crustal thinning and subsidence mechanisms dependent upon any of these three factors-rustal stretching, high temperatures, heavy sedimentation-will be reduced in effect at margin offsets. This ultimately leads to a steep Moho gradient and the development of a large mass excess beneath the continental rise. These arguments are not applicable to the rifted margin segment discussed above. We must still appeal to a complex interaction of fully operative margin-forming processes in order to explain the mass excess there. Shallow structure at ofSset margin segment The geological evolution of this segment of continental margin has been outlined early in the paper. Although it is now clear that the Agulhas fracture zone controls the trend of the margin, the origin of the structures closely associated with it is still puzzling. Prior to the separation of South America from Africa, the Agulhas Arch probably continued as a basement high on to Falkland Plateau. An examination of the seismic refraction data of Ewing et al. (1971) shows that the arch could have been continuous with the Falkland Islands high. If this is so, the 40 km wide strip of crust of intermediate thickness seen beneath the trough on Section 12 must have come into existence since the Palaeozoic and probably since movement between the continents Crustal structure south of South Africa 619 along the Agulhas fracture zone began. Magnetic profiles 24-27 of du Plessis & Simpson (1974) clearly show over the narrow strip a magnetic anomaly similar in size to those associated with the nearby oceanic crust. From this observation, it can be concluded that at least part of the basement beneath the sediment infilling the trough is magnetic and probably basaltic in character. Basement here, therefore, differs from that of the Agulhas Arch and the Agulhas marginal fracture ridge, both of which are nonmagnetic. There is bathymetric continuity between the trough and the Cape Rise of thickened oceanic crust, so there may also be a genetic connection. Assuming this, the trough is envisaged as being floored by a sliver of thickened oceanic crust, probably much faulted and dynamically metamorphosed within the Agulhas fracture zone. The trough appears to maintain its character to 22"E, then passes via a bathymetric step into the marginal fracture ridge. With the data now available it is possible to put forward a hypothesis for the origin of this ridge. It appears to have a close relationship with the adjacent continental crust, since it exists only where the Outeniqua Basin and the associated high-density lower crust occur beneath the shelf. It also appears to have a relationship with the formation of the continental edge and the Agulhas fracture zone because of their common parallelism. The lack of magnetic signature and the crustal thickness of 23 km beneath the ridge both suggest affinities with continental crust. Geophysical evidence suggests that the Ivory Coast-Ghana Ridge is an analogous feature, hence the dredging from it of a suite of rocks including coarse-grained and ferruginous micaceous sandstones and schists (Deltiel et ab. 1974) is further evidence for a continental origin. Clearly, the weight of evidence is in favour of a continental origin for the marginal fracture ridge. If we attempt to explain the emplacement of the marginal fracture ridge in terms of Le Pichon & Hayes' (1971) theory for the remobilization of continental crust in a continent-confined shear zone, it is necessary to account for its restriction to the margin of the Outeniqua Basin. Scrutton & du Plessis (1973) have already suggested that the degree of development of such a ridge depends on the gross composition of the crust through which the shear passes. If this is true, then the presence of the embryonic Outeniqua Basin together with, perhaps, the high density lower crust since early Mesozoic times may have facilitated remobilization to produce a significant ridge at the time of continental break-up. Unfortunately, modem examples of major continent-confined shear zones, the San Andreas fault and the Dead Sea fault system, give no indication of the formation of ridge-like features. They do show, however, lens faulting that can isolate a sliver of continental crust. From the evidence presented above it would appear that the best explanation of the marginal fracture ridge is that it is a sliver of continental crust, fault-bounded and possibly dynamically metamorphosed. It would have been produced in the continentconfined shear zone that became the Agulhas fracture zone. The local geology of the crust facilitated the metamorphism, thus restricting the ridge to a limited sector of the offset margin segment. 7. Conclusions Four major conclusions can be drawn from the preceding interpretation and discussion of the Atlantis If gravity data. They concern both shallow and deep crustal structure, a summary of which is given in Fig. 7, where an attempt is made to delineate crust types around the southern tip of Africa. The conclusions are: (1) South of South Africa the free-air gravity field over the continental shelf and upper continental slope is dominated by high values over the Agulhas Arch antiform and low values over the Outeniqua Basin of post-Palaeozoic sediments. This pattern is modified by smaller structures and density variations in the pre- 620 R. A. Scrutton Mesozoic basement. In deeper water areas, fracture zones and their associated structures control the gravity anomalies. (2) Isostatic gravity anomalies reveal a mass deficjt beneath the outer continental shelf and continental slope and a mass excess beneath the continental rise. This distribution of mass is somewhat unusual at inactive continental margins, but it can be explained by interaction of the various processes of continental margin evolution. In the Indian Ocean, the observed mass distribution is more pronounced than that in the Atlantic Ocean. This is attributed to the reduced effect of some of the processes of margin evolution at this margin segment, which is an offset margin segment as opposed to a rifted one. (3) In the Agulhas fracture zone, along the Indian Ocean margin, there is a basement trough and a basement ridge. The former is thought to be underlain by a sliver of thickened, dynamically metamorphosed oceanic crust, the latter by a sliver of upfaulted, dynamically metamorphosed continental crust. These structures may be typical of margin offsets. (4) Two features of excessive volcanic activity are apparent in the oceanic regions south of South Africa. One is the Cape Rise basement high, underlain by thickened oceanic crust and forming the southern limit of the continental rise and slope sediment prism in the Atlantic Ocean. The other is the Mallory Seamount chain, which apparently is not associated with any anomalous deep structure. Acknowledgments I thank K. 0. Emery and E. S. W. Simpson for providing me with the opportunity to sail on Atlantis 11. K. 0. Emery and C. 0. Bowin kindly allowed me to work on the Woods Hole gravity data. I have had fruitful discussions with R. V. Dingle, and A. P. Stacey has given valuable criticism of the paper. Diana Guthrie and Kay Stavert have helped with the preparation of the manuscript. The Eastern Atlantic Continental Margin Programme was funded by National Science Foundation Grant GX-28193 to Woods Hole Oceanographic Institution. Appreciation is expressed to all those who contributed to the collection of the gravity data, especially E. Uchupi, Senior Scientist. Grant Institute of Geology, Uniuersity of Edinburgh, Edinburgh EN9 3JW. References Bott, M. H. P., 1971. Evolution of young continental margins and formation of shelf basins, Tectonophysics, 11, 319-327. Bott, M. H. P. & Smith, R. A., 1958. The estimation of the limiting depth of gravitating bodies, Geophys. Prospect., 6, 1-10. Bott, M. H. P., Browitt, C. W. A. & Stacey, A. P., 1971. The deep structure of the Iceland-Faeroe Ridge, Marine Geophys. Res., 1, 328-351. Bowin, C. O., Aldrich, T. C. & Folinsbee, R. A., 1972. VSA gravity meter system: Tests and recent developments, J. geophys. Res., 77, 2018-2033. Case, J. E., Ryland, S . L., Simkin, T. &Howard, K. A., 1973. Gravitational evidencc for a low-density mass beneath the Galapagos Islands, Science, 181, 1040-1042. Cochran, J. R., 1973. Gravity and magnetic investigations in the Guinea Basin, western equatorial Atlantic, Bull. geol. SOC.Am., 84, 3249-3268. Crustal structure south of South Africa 621 Deltiel, J-R., Valery, P., Montadert, L., Fondeur, C., Patriat, P. & Mascle, J., 1974. Continental margin in the northern part of the Gulf of Guinea, The geology of Continental margins, 297-311. Eds C. A. Burk and C. L. Drake, SpringerVerlag, New York. Dingle, R. V., 1973a. Post-Palaeozoic stratigraphy of the eastern Agulhas Bank, South African continental margin, Mar. Geol., 15, 1-23. Dingle, R. V., 1973b. Mesozoic palaeogeography of the southern Cape, South Africa, Palaeogeog. Palaeoclim. Palaeoecol., 13, 203-21 3. Dingle, R. V. & Gentle, R. I., 1972. Early Tertiary volcanic rocks on the Agulhas Bank, Geol. Mag., 109,127-136. Dingle, R. V., Gentle, R. I., Gerrard, I. & Simpson, E. S. W., 1970. The continental shelf between Cape Town and Cape Agulhas, in: Rep. No 70116, Inst. Geol. Sci., Geology of the East Atlantic Continental Margin, 199-209. Ed. F. M. Delany SCOR Symp., Cambridge, 1970. Dingle, R. V. & Scrutton, R. A., 1974. Continental breakup and the development of post-palaeozoic sedimentary basins around southern Africa, Bull. geol. SOC. Am., 85, 1467-1474. Drake, C. L. & Nafe, J. E., 1968. The transition from continent to ocean from seismic refraction data, in: The Crust and Upper Mantle of the Pacific Area, eds L. Knopoff, C. L. Drake and P. J. Hart, AGU Geophys. Monog. 12, Washington, 174-1 86. DSDP Scientific Staff Leg 36, 1974. Glomar Challenger in the south-western Atlantic, Geotimes, 19, 11, 16-18. de Plessis, A. & Simpson, E. S. W., 1974. Magnetic anomalies associated with the southeastern continental margin of South Africa, Marine geophjw. Res., 2,99-110. Emery, K. O., Uchupi, E., Phillips, J. D., Bowin, C. O., Bunce, E. T. & Knott, S. T., 1970. Continental rise off eastern North America, Bull. Am. Assoc. Petrol. Geol., 54,44-108. Emery, K. O., Uchupi, E., Bowin, C . O., Phillips, J. D. & Simpson, E. S. W., 1975. Continental margin off western Africa: Cape St. Francis (South Africa) to Walvis Ridge (South-West Africa), Bull. Am. Assoc. Petrol. Geol., 59, 3-59. Ewing, M., Eittreim, S . , Truchan, M. & Ewing, J. I., 1969. Sediment distribution in the Indian Ocean, Deep Sea Res., 16,231-248. Ewing, J. I., Ludwig, W. J., Ewing, M. & Eittreim, S., 1971. Structure of the Scotia Sea and Falkland Plateau, J. geophys. Res., 76, 7118-7137. Francheteau, J. & Le Pichon, X., 1972. Marginal fracture zones as structural framework of continental margins in the South Atlantic Ocean, Bull. Am. Assoc. Petrol. Geol., 56, 991-1007. Gentle, R. I., 1970. Pre-Quaternary geology of the continental shelf between Cape Infanta and Cape Town, South African Natl. Comm. Oceano. Res., Mar. geol. Programme, Tech. Rept., 3, 13-27. Graham, K. W. T. & Hales, A. L., 1965. Surface ship gravity measurements in the Agulhas Bank area south of South Africa, J. geophys. Res., 70, 4005-4011. Green, R. W. E. & Hales, A. L., 1966. Seismic refraction measurements in the southwestern Indian Ocean, J. geophys. Res., 71, 1637-1647. Greenway, M. E., 1972. The geology of the Falkland Islands, Brit. Antarctic Suro. ScientiJic Rept. No. 76,29 pp. Groman, R. C., Luyendyk, B. P., Richards, D. A. & Wooding, C. M., 1972. Data processing of geophysical time series data, in: Seismic reflection, magnetic and gravity profiles of the eastern Atlantic continental margin and adjacent deep-sea floor I. Cape Francis (South Africa), to Congo Canyon (Republic of Zaire), eds E. Uchupi and K. 0. Emery, Woods Hole Oceano. Znst. Rept. 72-95, 9 pp. 622 R. A. Scrutton Hales, A. L. & Nation, J. B., 1972. A crustal structure profile on the Agulhas Bank, Bull. seism. SOC.Am., 62, 1029-1051. Hales, A. L. & Nation, J. B., 1973. A seismic refraction study in the southern Indian Ocean, Bull. seism. SOC.Am., 63, 1951-1966. Hales, A. L., Barrett, D. & Spence, D. L., 1970. The Indian Ocean seismic programme: a review, in Oceanography in South Africa, South African CSIR Symp, Durban, 1970. Heiskanen, W. A. & Vening Meinesz, F. A,, 1958. The Earth and its gravityJield, McGraw-Hill, New York, 470 pp. Hoskins, H., Rogers, C. U. & Woo, A. O., 1974. Data report of oblique reflectionrefraction radio-sonobuoy profiles on the African Atlantic continental margin (R/V Atlantis 11 Cruises 67 and 79, Woods Hole Oceano. Inst, Rept. 74-34,40 pp. Johnson, G. L., Lowrie, A. & Hey, R. N., 1973. Marine geology in the environs of Bouvet Island and the South Atlantic triple junction, Marine geophys. Res., 2,23-36. Larson, R. L. & Ladd, J. W., 1973. Evidence for the opening of the South Atlantic in the Early Cretaceous, Nature, 246,209-212. Le Pichon, X. & Hayes, D. E., 1971. Marginal offsets, fracture zones and the early opening of the South Atlantic, J. geophys. Res., 76,6283-6293. Leyden, R., Ewing, M. & Simpson, E. S. W., 1971. Geophysical reconnaissance on the African shelf: Cape Town to East London, Bull. Am. Assoc. Petrol. Geol., 55,651-657. Ludwig, W. J., Nafe, J. E., Simpson, E. S. W. & Sacks, S., 1968. Seismic refraction measurements on the southeast African continental margin, J. geophys. Res., 73,3707-3719. Ludwig, W. J., Nafe, J. E. & Drake, C. L., 1971. Seismic refraction, in The Sea, Vol. 4, Part 1 , 53-84, Wiley-Interscience, New York. Rabinowitz, P., 1972. Gravity anomalies on the continental margin of Angola, Africa, J. geophys. Res., 77,6327-6347. Rona, P. A., 1974. Subsidence of Atlantic continental margins, Tectonophysics, 22, 283-299. Scrutton, R. A., 1973a. Structure and evolution of the sea floor south of South Africa, Earth Planet. Sci. Letters, 19,250-256. Scrutton, R. A., 1973b. Gravity results from the continental margin of southwestern Africa, Marine geophys. Res., 2, 11-21. Scrutton, R. A. & du Plessis, A., 1973. Possible marginal fracture ridge south of South Africa, Nature, 242, 180-182. Scrutton, R. A., du Plessis, A., Barnaby, A. M. & Simpson, E. S. W., 1974. Contrasting structures and origins of the western and southeastern continental margins of southern Africa. Proc. 3rd Internatl. Gondwana Symp., Canberra, 1973. Siesser, W. G., Scrutton, R. A. & Simpson, E. S. W., 1974. Atlantic and Indian Ocean margins of southern Africa, in: The geology of Continental margins, 641654. Eds C. A. Burk and C. L. Drake, Springer-Verlag, New York. Simpson, E. S. W., 1970. The geology of the southwest African continental margin: a review, in: Rep. No. 70/16, Inst. Geol. Sci., Geology of the East Atlantic Continental Margin (ed. F. M. Delany) SCOR Symp., Cambridge, 1970,153-170. Simpson, E. S. W. & du Plessis, A., 1968. Bathymetric, magnetic and gravity data from the continental margin of south-western Africa, Can. J. Earth Sci., 5, 11 19-1 123. Sleep, N. H., 1971. Thermal effects of the formation of Atlantic continental margins by continental break-up, Geophjs. J. R . astr. SOC.,24,325-350. Crustal structure south of South Africa 623 Spence, D. L., 1970. A seismic refraction study of sedimentary structure on the Agulhas Bank, south of Cape Infanta, unpublished MSc Thesis, University of Witwatersrand. Smit, P. J., Hales, A. L. & Gough, D. I., 1962. The gravity survey o f t h e Republic of South Africa, Geological Survey publ., Pretoria, 486 pp. Talwani, M.,1962. Gravity measurements on H M S Acheron in South Atlantic and Indian Oceans, Bull. geol. SOC.Am., 73, 1171-1 182. Talwani, M., 1971. Gravity, in The Sea, Vol. 4, Part 1, 251-297. Wiley-Interscience, New York. Talwani, M. & Eldholm, O., 1973. Boundary between continental and oceanic crust at the margin of rifted continents, Nature, 241, 325-330. Talwani, M. & Kahle, H-G., in press. Free-air gravity charts of the Indian Ocean, in The atlas of geology andgeophysics of the International Indian Ocean Expedition, ed. G. Udintsev. Talwani, M., Sutton, G. H. & Worzel, J. L., 1959. Crustal section across the Puerto Rico Trench, J. geophys. Res., 64,1545-1555. Vening Meinesz, F.A.,1948. Gravity expeditions at sea, 1923-1939,Vol.4,Waltman, Delft, 233 pp. Walcott, R. I. 1972. Gravity, flexure, and the growth of sedimentary basins at a continental edge, Bull. geol. SOC.Am., 83, 1845-1848. Woollard, G. P., 1959. Crustal structure from gravity and seismic measurements, J.geophys. Res., 64,1521-1544. Worzel, J. L., 1965. Deep structure of coastal margins and mid-ocean ridges, in Submarine geology and geophysics, 335-361, eds W. F. Whittard and R. Bradshaw, Colston Papers No. 17, Buttenvorths, London. Worzel, J. L. & Shurbet, G. L., 1955. Gravity interpretations from standard oceanic and continental crustal sections, in Crust of the Earth, ed. A. Poldervaart Geol. SOC.Am. Sp. Pap. 62,87-100. Wyllie, P. J., 1971. The dynamic Earth, Wiley, New York, 416 pp.