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
]OUI•NAI. OX'G•.O?a¾SXCAL RES•-AI•Ca VOL. 69, NO. 6 MARCH 15, 1964 Crustal Warping--A Possible Tectonic Control of Alkaline Magmatism D. K. BAILEY • Geophysical Laboratory,CarnegieInstitution o] Washington,Washington,D.C. Abstract. Alkaline and peralkaline magmatism is peculiar to the epeirogeniccontinental environment; it is characterizedby unusual concentrationof alkalis, volatiles, and rare elements,the large volumesof alkaline rock contrastingwith the small amountsof suchrocks in alkali basalt provinces. Probably the most outstanding region of alkaline and related carbonatite activity is in east and central Africa, where localization of the activity along the rift valleys has long been recognized.Recent discoveriestend to establishthis relationship. The restriction of major alkaline provincesto the stable continents has led to the suggestion that long undisturbedperiods must be necessaryfor slow accretion beneath the crust; the implicationsare that the continentalplate has a passiverole and that any associationof alkaline magmatismwith rifting is fortuitous.It is here argued that the regional tectonicscontrol the generationof alkaline magmasand their localization along the rift zones.The African shield showsa distinct 'basin and swell' pattern, and many of the rifts occupy the axial zones of the long,narrowupwarps,separatingthe basins.A seriesof erosionsurfaces,dating at least as far back as the Jurassic,showsincreasingelevation and separationtoward the crestsof the swells and indicates uplift of these zones in several distinct stages: the continuity of such erosion surfaces shows that the continental plate behaved competently during these uplifts. Upwarpingof a rigid continentalplate wouldrelievethe pressurelowerin the crust,allowing partial melting and the concentrationof fugitive constituentsfrom the underlyingmantle. In east and central Africa, similar alkaline magmas of salic differentiate character,rich in volatiles,and oftenwith associated carbonatites, wereproducedin differentregionsduringdifferent periodsof rifting; the unusualvolume of suchrockscan more reasonablybe explained by partial melting of crystallinesourcerocksand collectionof juvenile fugitive constituents than by the usuallyfavoredprocessof fractionalcrystallization,whichwould requirea vastly greater volume of parent magma. Introduction. Many petrologists havenoted sumablythis tectonicrequirementcouldbe met that the major alkalineand peralkalineigneous in the oceanbasins.Also, if someprocessother complexes appearto be restrictedto the stable than crystal-liquid equilibrium were required continental or shield areas of the crust, an ar- for the formation of alkaline rocks,it is difficult rangementsuccinctlydescribedby Backlund t.o reconcile their concordance with the lowof 'petrogeny'sresidua [1932] as the epeirodiatresis relationship,and temperaturecompositions one which has given rise to the view that t.he system' [Bowen, 1937]. The central questionrespecialtectonicconditionsof these regionsare mains,therefore: Why shouldtheserocksof disnecessaryto the formation of such magmas. tinct salic differentiate character show a marked There is no generalagreementas to the precise developmenton the continents? Probably the most spectaculararray of alkarole of the tectonicsin this magma generation; it has frequentlybeen suggested that the long line igneousrocks and associatedcarbonatitesis periodsof quiescencehave favored the accumu- in Africa, particularly in the easternand central lation of alkali-rich magma by such means as parts of the continent,an excellentsummary of extensivegastransfer in a magma suchas alkali the information on the activity being provided basalt. If lack of disturbancewere the criterion, by Kin• and Sutherland [1960]. That this alkawith however, •he continental associationof these line magmatismshowsa generalcoincidence magmas would still need explanation,for pre- the rift pattern in this par• of Africa has been noted by many writers,includingKing and Sutheftand (p. 299), but there has been considerable • Permanent address: Department of Geology, divergenceof opinionas to the relationshipbeTrinity College, Dublin, Ireland. 1103 1104 D.K. BAILEY I000 Miles Scale ]Fig. 1. The East and Central African rift zone. tween the t.wo, if any exists;King and Suther- can Rift zone, has focused attention on the•n land, for instance,wereclearlyperturbedby the and made them a subject of controversy since lack of detailed correlation between the igneous the time of the classicdescriptionby Gregory centers and the rift faults. But the African rift. [1921]. But, as clearly shown by Dixey [1956] zonesare one of the world's major tectonic fea- in his survey of the East African Rift, much of tures and if they are in any way a locusof alka- the Tertiary movement was on the site of lateline magmatism they undoubtedly hold the key or post-Karroo (Jurassic) movements,and as to the relationshipof suchmagmatismwith epei- the Western Rif• is traced from Lake Rukwa rogenicearth movements.The following discus- into the Nyasa trough, and farther south, the sion will be confinedto this one region,but it age of the major movement is predominantly is felt that the basicrelationshipthat is deduced, post-Karroo. New work is showingthat south of namely, partial melting by relief of lit,hostatic LakesTanganyikaand Rukwa a pat.ternof post- load, may have a more general application to the problem of the alkaline rocks. Karroo rifting emergesthat is comparablein ex•tentand scalewith the Tertiary rifting to the The rift zones. The impressivetopogra. phic expressionof the Tertiary rifts in east,and northeast Africa, commonlyknown as the East Afri- north, and only lessstrikingbecauseits topographicexpressionhas been subduedby longer erosion:the mid-Zambezi-Luangwa rift, for in- CRUSTAL WARPING 1105 EXPLANATION TERTIARY-RECENT •..........Volcanic fields o Alkaline centres and volcanoes PRE-TERTIARY + Alkaline centres IOOO Miles Scale Fig. 2. Distribution of alkaline centers and volcanic fields. stance,is merely a p,•r• o.f a much longer rift zoneof post-Karrooage [Bailey, 1961]. Almost certainlythere was pre-Karroo rifting too, and Dixey [1956] has alsostressed the tendencyof the rift zonesto parallel the major structural trendsin the Precambrianrocks; the latter are largely orogenicstructures,however,and, as hintedby Shackleton[1956], this may be essentially accommodation of the rift pattern to a previouslyestablished'grain' in the crust. The presentlyknown pattern of rifting in ea. stern and southernAfrica is shownin Figure1. activity around rift intersections[Dixey, 1955, p. 28; Bailey, 1961] and in the strongcorrelation of Tertiary and post-Karroo activities with those sections of the rift in which movements of these periods respectivelypredominate,makes it. hard to avoid the conclusion that the mag- matism and the rifting are both expressionsof the same major process. The type and characterof the alkaline igneous activity have alreadybeenreviewedby King and Sutherland [1960], with a comprehensivebibliography,so that only a brief survey is in order The alkaline magmatism. When the distri- here. The extensive Tertiary to Recent lava bution of alkalineigneousactivity is superim- fieldsin Ethiopia are generallyrecordedas being posedon the rift pattern, as shownin Figure2, predominantly basaltic, but this may well be there is a remarkablecorrespondence. Certainly a convenientgeneralizationbecauseto the south, the activity is not solely confinedto the actual in Kenya, although basalts (often basanites rift faults; it. would seemnaive to expectthis, according to Kent [1944]) are widespread, but the generalconcurrence of the two patterns, phonolite is very commonand is perhaps the especially noticeable in the concentration of predominantlava throughoutthe Eastern Rift 1106 D.K. BAILEY and in adjacent areas suchas the Uasin Gishu plateau, the Aberdare Mountains and Mount Kenya. Trachytes,nephelinites, and alkali rhyolites are also well representedin these fields. Farther south, in Tanganyika, the lava fields give way to central volcanoesbuilt largely of fragmental materials, mainly nephelinitesand phonolitesalong the Eastern Rift and trachytes and phonolitesto the east in the Kilimanjaro range. In eastern Uganda a chain of Tertiary volcanoes,largely composedof fragrnental nephelinites, stretches northward from Mount Elgon [Davies, 1952], one of these, Napak [King, 1949], having an exposedijolite-carbona- ment in rare elements.Implicit in nearly all discussions of such alkaline rocks is the view that they are differentiatesof a primary magma,such as alkali basalt: Harkin, however,clearly recognizes that vast quantities of basalt would be requiredto yield the observedvolumesof salic rocks. In addition to the Panda Hill carbonatite in the Rungwevicinity there are other old carbonafire and alkaline complexessuchas the Nkumbwa Hill carbonatite and the Mivula nepheline syenite in Northern Rhodesia and the Ilomba nephelinesyeniteand anothergroup of carbonatires in Nyasaland. A most impressive concenrite core. Farther to the south and west lie the tration of pre-Tertiary carbonatites,nepheline pre-Tertiary carbonatite centers of Tororo, syenites,and syenitesis found in southern Nyasalandand adjacentparts of Mozambique.The Sekulu,and Bukusu.• The potash-rich lavas of Toro-Ankole and majority of theseare of Chilwa age, i.e., immeBirunga in the Western Rift are well known diate post-Karroo,but some, such as those of [Holmes and Harwood, 1932; Combe and Port Herald, are pre-Karroo. All theseare deepHolmes, 1945], but not all the voleanieity of seated complexes, and the large masses of this rift is strongly potassie,as indicated by nepheline syenite, and especially syenite (or Bowen [1938], although it is characteristically perthosite),suchas Zomba and Mlange, which alkaline, basaltsbeing well representedonly in distinguishthis region,are especiallysignificant. South Kivu. Such rocks are older, plutohie equivalentsof Farther south, where the Eastern and Western rifts intersect,is the more isolatedTertiaryRecentvolcanicfield of Rungwe,whichhasbeen studied in detail by Harkin [1960]. Alkali basalt is the predominantlava, and in degreeof alkalirfityandscopeof differentiation the magma seriesis comparedby Harkin to that of Tahiti; the widespreadphonolitesand trachytes of the Tertiary rifts to the north, and they indicate that the sametype of magmatismcharacterized the post-Karroorifts. Theseplutonsalsotend to negate any suggestionthat the Tertiary volcanics are products of differentiation in intracrustal magma chambers. Similarconsiderations apply to the morescatbut the volumeof salielavas,mainly trachytes and phonolites,with somenephelinites, is about tered groups of post-Karroo and pre-Karroo half the total eruptive materiM, and Harkin is complexesto the south and west of Nyasaland forced to conclude that the alkali basalt has (Figure 2). The only exceptionwould seemto been contaminatedby earbonatiteand alkaline be the group of pre-Karroo carbonatitesand material,suchas formsthe neighboring but dis- alkaline complexesfalling in a roughly E-W tinctly oldercomplexof PandaHill. Essentially, zonein the Transvaal,South Africa. These are this problem is the same as that of t,he whole not related to any well-definedrift, but when of the volcanism of the East African Rift--a it is rememberedthat even the post-Karroo preponderanceof alkaline salie rocks,in many rift pattern is blurredby erosionthis is perhaps easesfragrnental,indicatingvolatile-richexplo- not surprising; they may be related to Presive activity, and frequenfiy showingsigns of assoeated carbonatite with its attendant enrich- 2 In a summary of the igneousactivity in Kenya, just published, Wright [1963] has drawn the additional distinction that within the rift valley the younger lavas (post-Miocene) are largely nepheline-free. This may have important implications in the sequence and level of magma generation, but it does not greatly modify the outline given here. cambrian rifting which .is now almost undecipherable. The large-scalecorrelationof the rifting and igneousactivity in time and space,as indicated above,seemsindisputable. Divergenceof opinion seemsto arise when the picture is examinedin detail; e.g., the nephelinitevolcanoesof eastern Ugandaare not on a recognizablerift line [King and Sutherland,1960], the volcanicphasesof CRUSTAL WARPING 1107 uplifts of some thousandsof feet along zones 300 to 400 miles in width and several thousand milesin length.'Sucherosionsurfaceshave their maximumvertical separationalongthe flanksof the rifts (Figure 3) and convergeand overlap as they are traced into the adjacent basinal areas; they give preciseexpressionto the longknown fact that the rifts are located in the more uplifted portionsof the continent (most clearly seen in the Tertiary East African Rift zone, which lies along the continental divide). This 'basin and swell' structure Fig. 3. Vertical section illustrating the 'rise to the rift' with increasing separation of several erosion surfaces; the upper surface may be taken to represent a Jurassicpeneplane,the lowest an endTertiary peneplane. The heavy broken line marks a Jurassicdatum plane within the crust,at an arbitrary depth. The vertical attitude of the rift fracture is diagrammatic. activity in Kenya appearto be terminatedby rift movements [Kent, 1944], and there is no correlation between the volumes of the rift troughsand the effusivematerials.But theseare real anomaliesonly if a restrictedview is taken of the relationshipbetweenthe magmatismand rifting; they refer to the premise that any relationshipmust be simply oneof causeand effect, whereasit is just as likely that the rifting and magmatism are both expressionsof a more fundamental process.Some such processbecomes of Africa has been well illustrated by Holmes [1944], whosemap is reproducedas Figure 4. Uplift of the flanks of the rifts in many placesis also suggestedby the increasein metamorphic grade of the Precambrian crystalline rocks as the rift is approached,but this evidenceis not always unequivocal.In Rhodesia,for instance,there is an increasein metamorphicgrade in the Lomagundi/Katanga rocks (625 _ 20 m.y.) as the Jurassicmid-Zambezirift is approached[Macgregor, 1947]; pegmatites in this high-grade belt give agesof 430 to 460 m.y. [Nicolaysen, 1962], showingthat this is not simply exposure by uplift of originallydeeperKatanga rocksbut probably also indicates localization of the rift alonga distinctmetamorphicbelt which formed after the main Katanga orogeny. The riseof erosionlevelsof differentagesover the rift zonesshowsthat the continentalplate has beenwarpedin severaldistinctphases,at least sincethe Jurassic,and that the rifts are apparent as soon as the rifts themselves are relatively subordinatefeatures on the crests of lookedat in a regionalcontext. Many people have noted that the uplift of parts of the African continent is far more impressivethan the rifting, this being expressed most forcibly by Willis [1936], who considered that the major problem connected With the rifts was the uplift of the neighboringplateaus.Uplift of the plateaus is a misleadingterm, however, becauseit has becomeincreasinglyclear that 'the rise to the rift,' to which Wayland [1930] called attention, is a generalfeature, the rifts beingessentiallylongitudinalfracture zones along the crestsof broad geanticlines.The geomorphologicalevidencehas been reviewedby the crustal swells.Obviouslythe causeof the warping is as subjectto contentionas the cause of rifting, but it is possibleto clarify the situa- Dixey [1956], and the riseof erosionsurfacesof different ages (Jurassicto Quaternary) toward of uplift couldlead to partial melting by relief of lithostatic load, but compressionwill be adopted as a working hypothesisbecauseit is tion a little. Dixey [1956], for instance,favors the conceptthat the basinshave saggedand that the intervening swellsare largelylag aress, but, otherconsiderations aside,the arrangement of erosionlevels,with the maximum separation on the swellsand the hingelinesin the basins, requirespositiveuplift of the swellswith respect to the base levels. This observation is critical because positiveuplift canbe accomplished only by lateral compressionor direct vertical movement below the uplifted area. Either mechanism the rift shouldersis striking evidence of 'the archingthat accompaniedthe rifting, involving more consistentwith present observations. 1108 D.K. BAILEY IRAN Pladeira DESE EL-JUF CHAD • ß .... 5•::•. .' .:.-.- .. ..'-.?.• ..-.. ..... ::;'-&' ..,.. •;: '•.... :.::" •,,.•: .. •• •;:: ..... :.,, .... :.•',,-- -;¾,•1 ß:• R•n,on Fig. 4. Generalized map of tectonic basins,plateaus, and swells of Africa. (Reproduced from Principleso• Physical Geology by Arthur Holmes, 1944,with the author'spermission.) Compression,uplift, and melting. Lateral compressionon a large scaleis suggestedeven in the simplebasin and swellstructureof Africa, whichhasa tessellatedpattern suchas might be expectedfrom compression of a rigid but anisotropic plate that has tended to yield along 'grains'establishedby earlier metamorphicbelts. In an isotropicplate of uniform thicknessthe compressionrequired for initial warping would be much greater than that necessaryto increase the flexure,and it is to be expected,therefore, that in a continental plate the initial flexures would appear where the plate was already thickened or already uplifted. Such conditions would obtain in old orogenicbelts, especiallyif in a state of over-recovery from previous isostatic imbalance,which would accord with the tendency of swells and rifts to parallel earlier orogenictrends. Present lack of knowledgeof the mechanismof major earth processesmakes it fruitlessto dwell long on the possiblecauseof this compression, but if the southerncontinents were once welded together and have since been driven apart, then Africa, the Gondwananucleus,has presumablybeen under compression from all sides. Upwarping of the rigid continental plate would producerelief of lithostatic load below the crest of the flexure. Where the bulk com- positionand temperaturelower in the crustwere favorable, this would lead to partial melting [Yoder, 1952],3 and the zone of pressurerelief would act alsoas a trap for volatticsand fugitive constituentsfrom the underlying mantle even if melting in the mantle were not possible.It was noted before that the alkaline rocks were a Since the original presentation of this paper (Trans. Am. Geophys. Union, •, 115, 1963) my attention has been directed to a lecture, 'The Teetonic Significance of Alkaline Igneous Activity,' bv R. M. Shackleton, at the first (British) InterUniversity Geological Congress, 1953. In the printed summary he states,'Subcrustalmelting was due to the lifting of part of the load on the layers under the sxvells,and eruption was facilitated by the fractures.'This is an essentialpart of the process elaborated in this paper. CRUSTAL WARPING 1109 generally regardedas derivative magmas,but if it is possiblefor them to form as residualliquids from the crystallizationof a magma such as several times since the Jurassic, and •his implies that they were in isostatic equilibrium at the end of each erosion cycle. To achieve such a alkali basalt, then before any crystallinebasalt condition lighter material would have to be above the eclogite-basalttransition zone can be added at depth to compensateeach succeeding rendered completely molten it should partly uplift. The relevanceof this to the presentsugmelt to yield alkaline liquids. This mechanism gestionof partial melting below the rigid layer produces'derivative' magmaswithout requiring of the crust and withdrawal of light materials the vast quantities of heat necessaryto com- from the underlyingmantle needsno emphasis. pletelymelt all the parent material,and at the But this doesintroduce,too, the alternative that same time it disposesof the volume problem addition of lighter material at the ba•seof the wherebythe volume of 'derivative'magma is crust might, in itself, be responsiblefor the far t,oo large in comparisonwith that of the uplift. The main objectionto sucha mechanism 'parent.' It may be that only belowthe cover is the difficulty in explainingthe rifting in the of the continentsare all the conditionsappro- crestof the upwarp, for this would requirewithpriate for partial melting of basaltic materials drawal of material from belowthe rifts and long at the baseof the crust,hencethe restrictionof sections of these have little or no associated major alkalinemagmatismto this environment. volcanism.Expressedin another way, accretion The cause of the rifting in the crest of the arch is, of course,a thorny subject of debate, but there is litt.le reason to supposethat the rifts representsubsidence due to removalof underlyingmagmato the surface;e.g.,there is no volcanismassociatedwith the deepestrift, that of Lake Tanganyika, its floor being 2500 feet below sea level. Strong argumentshave been made that the rifts are compression structures of light material to producethe upwarpsrequiresthat rifting is a consequence of magmatism, whereasthe alternative preferred here is that the accretionof light material,the magmatism, and the rifting are all consequences of lateral compression and upwarpingof the crust. Direct vertical movementbelow the swells, the alternativemechanismt• lateral compression, might be supposedto result from uplift helddownbv the overridingflanks,but the sup- above a rising convectioncurrent. The latter porting evidenceof negativegravity anomalies in the rifts [Bullard, 1936] has beenweakened by drilling evidenceof a considerablethickness of light sedimentsin the best example,the Albert Rift [Dixey, 1956]. The wholesubjectof possibility willundoubtedly findadvocates among rifting, like most geologicproblems,probably suffersfrom our naturaltendencyto fit a simple explanationto a complexphenomenon. If the swellswere producedby a seriesof compres- Atlantic rise with its median 'rift valley'; he t.hose who believe tha•t such a mechanism is re- sponsiblefor the midoceanrises,for Heezen [1962] has shown that the form of the African swells and rifts is similar to that of the midalso shows the East African Rift zone as a con- tinuation of the Indian Oceanrise, but the detailed pattern of the Gulf of Aden and Red Sea sions,as indicatedby the seriesof erosionsur- rifts requiresthat this be an extremelytortuous faces,then in a periodof relief of compressiontransition. If, however,the African rift zones a rift zone might be initiated as an elastic-release fracture zone, which, once established, wouldbe perpetuatedand aggravatedby subsequent compressions and releases.An explanation of this kind might resolvethe conflict between the opposingtensionand compression hypothesesof rifting. Perhaps the most interesting result of Bullard's [1936] gravity survey of the rift zones, albeit overshadowedby the discoveryof the negative anomaliesof the rifts, is that the adjacent plateaus are in isostatic equilibrium. Theseareashave beenuplifted and peneplaned are the continental analogs of the midocean rises, they pose difficulties for a convection origin, becauseit is postulatedthat the rising convectioncurrent causesthe uplift., splits to give the median'rift,' and then spreads,giving ocean-floorspreadingand separationof the continents. There is so far no evidence of similar spreading of the African continent around the rift zones; in fact, it is held by some that the rifts are compression features,and it is difficult to see why a tensionalprocessshouldhave the same expressionthrough the different thicknesses and structures of the oceanic and con- 11 I0 D.K. BAILEY tinental crusts(perhapsthe similar wavelengt. hs are merely a function of the physicalstate of the outer shellsof the earth). Oceanspreading and concomitantcontinental drift are also sup- posedto have started in the Cretaceous, but there is evidenceof rifting in the Jurassic,and probably much earlier, in Africa. Other digculties are that the African rift pattern would require the operationof numeroussmall convection cells (one at least as small as the Lake Victoria basin!) and the situation of rifts along earlier metamorphic belts requires that there are now rising currentswhere previouslythere were descendingones. These remarks are intended chieflyas a warning agains•too facile an application of the convectionhypothesisto all major crustalprocesses, for even if it couldbe demonstrated that the African swells were in- termittently uplifted by rising convectioncurrents the pressurewould be relieved during the intervals and the main thesis of this paper would remain applicable. Conclusion. The positionof the East African Tertiary Rift zone along the continental divide, the generaloff-lap of youngerformationsaway from the rifts, and the rise of the seriesof old erosion bevels to the rift shoulders indicate the is to say that vast quantitiesof parent maqma, such as alkaline o!ivine basalt magma, were availablebut only a fraction of this reachedthe surfaceor even the plutonic levelsthat are exposedtoday in the post-Karroo rifts; moreover, such a state of affairs must have prevailed in many different placesand in severalgeological periods.It would seemmore logicalto regard these salle rocks as products of partial melting of parent crystalline materials, such as alkali basalt, in the lower crustal region, with contributions of juvenile material from the upper mantle, and relief of pressurewould be a suitable mechanism for this. Any discussioninvolving the origin of carbonarite must touch on the origin of kimberlite, for rocks of kimberlitic compositionare associated with carbonatites,and vice versa, and somepeople [e.g., yon Eckermann, 1961] hold that carbonatite is derived from a magma of kimberlite composition.But while the distribution of carbonatite centers shows a strong correlation with the rift pattern, the major kimberlite provincesdo not, althoughthey do appear in uplifted areas; the associatedalkaline igneous activity of carbonatites and their unusualsuite of trace elementsalsoset them apart from kimberlite activity, which is typically expressed as breccia pipes and intrusions with closespatial and temporal relationshipbetween little or no associatedmagmatism.Apparently the rifting and alkaline, peralkaline,and car- the two kinds of activity, in their typical debonatiticmagmatism,characterized by abundant velopment,reflect differing tectonic conditions, volatiles,both in the Tertiary rifts and the older as inferredby James[!956], and thereis a vital rifts such as those of the post-Karroo. Essen- need for more information on the regional tially the questionis, can the generationof these tectonicsof the major kimberlite provincesas vast quantitiesof uncommonrocksbe related •o well as gravity and seismic data from both the warping and rifting of the crust? The con- tectonicenvironments.It seemslikely that bot.h tinuity of the variouserosionsurfacesacrossthe kimberlite and carbonatite are derived from the upwarps indicates that the crust has behaved upper mantle, though not necessarilyrelated competently,and present evidencefavors the through somesimple processsuch as fractionaview that it has warped in responseto lateral tion from a commonparent. The differencein compression. Regardlessof the cause,however, the expressionsof the two activities may lie in it is to be expectedthat there would be relief the fact that below the rift zones the PT condiof lithostatic load below the crest of a rigid tions are such as to produce melting in the arch due to support in the limbs, giving rise, lower crust, as indicated by the igneousactivunder favorable conditions,to partial melting ity; therefore, any kimberlitic material from lower in the crust and perhaps in the upper the mantle would generallyhave to be erupted mantle and withdrawal of fugitive constituents through a crustal zone of melting and magma from the uppermantle.The bulk of the magmas generationand would thus be expectedto reach in the rift zonesare of salic differentiate type, the surfaceonly in such form as melilite basalts e.g. phonolites,trachytes, and nephelinites;to and kindredlavassuchas t,hoseof Toro-Ankole, call them productsof fractional crystallization which show affinities with both k'nuberlite position of the rifts as median fracture zonesin the crests of broad crustal arches. There is a CRUSTAL WARPING 1111 [HolmesandHarwo.od, 1932]and carbonatite central and southernAfrica, Trans. Geol. Soc. S. A•rica, 58, Annexure, 1955. [Holmes,1956; von Knottingand Du Bois, Dixey, F., The East African rift system,Colcmi•l 1961].Awayfromthe riftsthe conditions in the Geol. Mineral Resources(Gr. Brit.), Suppl. Set., crust are not suitablefor extensivemelting, and Bull. Suppl. 1, 1956. kimberliteis allowedto passfrom the mantle Gregory,J. W., Riit Valleys and Geology o• East Seeley, Service and Co., Ltd., London, to the surfacein its typical form of a gas-trans- A•rica, 1921. ported breccia. Harkin, D. A., The Rungwevolcanicsat the north- Bowen,in 1937,usedthe compositions of the ern end of Lake Nyasa, Mem. Geol. Surv. Tanganyika, 2, 1960. rift valleylavasto test 'petrogeny's residuasystem,' showingthat they were almostcertainly Heezen, B.C., The deep-seafloor, in Continental Drift, edited by S. K. Runcorn, Academic Press, the productsof crystal-liquidequilibrium, which New York, 1962. for him, it seems,was most often synonymous Holmes, A., Principles o• Physical Geology, Ron- with fractional crystallization. But the same ald Press,New York, 1944. consideration would apply to the productsof Holmes, A., Vethandel. Ned. Geol. Mijnbouwk. Genoo.t.,16, 1, 1956. partial melting,and, whenthe volumesof such Holmes, A., and H. F. Harwood, Petrology of the volcanic fields east and south-east of Ruwenzori, Uganda, Quart. J. Geol. Soc. London, 88, 370442, 1932. James,T. C., Carbonatites and rift valleys in Tanganyika, Rept. Intern. Geol. Congr., 20th, Mexico, 1956. Kent, P. E., The age and tectonic relationships of the East African volcanic rocks, Geol. Mag., 81, 15-27, 1944. King, B. C., The Napak area of Karamoja, Uganda, Mem. Geol. Surv. Uganda, 5, 1949. King, B.C., and D. S. Sutherland, Alkaline rocks of eastern and southern Africa, Sci. Progr., 48 Acknowledgments. I wish to thank Professor (190, 191, and 192), 1960. C. E. Tilley and Dr. H. S. Yoder for their helpful Macgregor, A.M., An outline of the geological discussionsand review of the manuscript. history of Southern Rhodesia, Bull. Southern Rhodesia,Geol. Surv., 38, 1947. Nicolaysen, L. O., Stratigraphic interpretation of REFERENCES age measurements in southern Africa, in Buddington Volume, pp. 569-598, Geological Society Backlund, H. G., On the mode of intrusion of deepof America, New York, 1962. seated alkaline bodies, Bull. Geol. Inst. Univ. Shackleton,R. M., In discussionof Dixey, Colonial Uppsala, 24, 1-24, 1932. Geol. Mineral Resources(Gr. Brit.), Suppl. Set., Bailey, D. K., The mid-Zambezi-Luangwa rift and Bull. $uppl. 1, 1956. related carbonatite activity, Geol. Mag., 98, 277- yon Eckermann, H., The petrogenesisof the Aln5 284, 1961. alkaline rocks,Bull. Geol. Inst. Univ. Upsala,40, Bowen, N. L., Recent high-temperature research 25-36, 1961. on silicates and its significance in igneous geol- yon Knorring, O., and C. B. G. Du Bois, Carbonatiogy, Am. J. Sci., 33, 1-21, 1937. tic lava from the Fort Portal area in western Bowen, N. L., Lavas of the African rift valleys and Uganda, Nature, 192, 1064-1065,1961. their tectonic setting, Am. Y. Sci., 35A, 19-34, Wayland, E. J., Rift valleys and Lake Victoria, rocksin the rift valleys are taken into account, partial melting seemsmore feasible.Such a thoughtis at the root of the presentsuggestion of tectonic control of the alkaline magmatismin eastern and central Africa. Obviously in other regionsdifferentconditionsmay determinethe generation of alkaline magmas, but wherever suchmagmasare developedin unusual amounts the possibilityof some control mechanismfor relief of pressureshouldbe examined. 1938. Bullard, E. C., Gravity measurements in East Africa, Phil. Trans. Roy. Soc. London, A, 235, 445-531, 1936. Combe,A.D., and A. Holmes,The kalsilite-bearing lavas of Kabirenge and Lyakauli, south-west Uganda, Trans. Roy. Soc. Edinburgh, 61, 359379, 1945. Davies, K. A., The building of Mount Elgon, Mem. Geol. $urv. Uganda, 7, 62 pp., 1952. Dixey, F., Some aspectsof the geomorphologyof Intern. Geol. Congr. (1929), 2, 323-353, 1930. Willis, B., East African plateaus and rift valleys, CarnegieInst. Wash.Publ. 470, 1936. Wright, J. B., A note on possible differentiation trends in Tertiary to Recent lavas of Kenya, Geol. Mag., 100, 164-180, 1963. Yoder, H. S., Jr., Change of the melting point of diopsidewith pressure,J. Geol., 60, 364-374,1952. (Manuscript receivedJuly 1, 1963; revised December 20, 1963.)