Sorting Of Alluvial Diamonds

ECONOMIC

GEOLOGY AND

BULLETIN

VOL. 77

OF

THE

SOCIETY

THE

OF

NOVEMBER,

ECONOMIC'GEOLOGISTS

1982

No. 7

The Transportand Sortingof Diamondsby _Fluvialand Marine Processes DONAI•D

G. SUTHERI•AND

PlacerAnalysisLtd., 2 LondonStreet, EdinburghEft3 6NA, Scotland Abstract

This paper describesthe systematicvariationsin size and quality of diamondsthat result from transportby fluvialand marineprocesses. It isshownthat diamondsoriginatingin rather restrictedsourceareas can be spreadacrossmany tens of thousandsof square kilometers. Downstreamor alongshoretransportcan be followedfor severalhundredkilometers,with a systematicreductionin averagediamondsizeaccompanyingsuchtransport.The diamondsize distributionis progressively modified with increasingtravel distance,and the quality of the diamondsimprovesas inferior typesare destroyed.The accompanyingvariationsin crystal form are outlined.Certain implicationsfor explorationprogramsare •nentioned. Introduction

1N the late 1940s, productionof dia•nondsfrom alluvial sourcesrepresentedabout 40 percent of the

to producegeneralstatementsthat are capableof quantifieationand valid for all alluvial diamonddepositswherever they occur. The Extent of Dimnond

Distribution

world total.Thirty yearslater, alluvialdiamondproductionhad more than doubled,and despitethe deThe hardnessof diamond when subjected to a velopmentof majornewkimberlitemines,it stillrep- grindingaction,its stabilityat relativelyhigh temresentedmore than $0 percent of total natural dia- peraturesand pressures, and itschemicalinertness all mond production. The economic importance of suggest that oncediamondis releasedfrom its prialluvial diamondsis thus considerable,and it is fur- mary sourcerockit shouldtend to persistin the sedther emphasizedby the fact that alluvial diamonds imentaryrecord.The brittleness of diamond,which are of consistently higher quality than diamondsre- leadsto shatteringunder a sharpimpact, will only covered from the source kimberlites and also because

act to reduce its size. Moreover, as has happened in

certain countries(e.g., Sierra Leone, Central African

Ghana(Junner,1948),diamondsare ableto undergo Republic)are economically largelydependenton the metamorphiceventsthat alter the surrounding rock productionof alluvial diamonds.Furthermore, while typesand to be releasedto the fluvial systemat a it is generallyconsideredthat kimberlite pipesrep- later date. resentthe more promisinglongtermmining target, Thisability of diamondto withstandmostsurface it isusuallythroughthe presence of alluvialdiamonds and near-surfaceprocesses is reflected in the large that kimberlite source rocks of economic interest are size of many diamond fields. The fields of Sierra first suspected. Leone,for example,coverabout 20,000 kms (Hall, The wide distribution and economicimportance 1968)and thoseastridethe Zaire-Angolaborder,over of alluvial diamondshave given rise to a large lit- 60,000 kms (Bardet, 1974). The wide distributionof erature, but the great majority of the studiespub- alluvialdiamondsin certainregionscan be attributed lishedare descriptiveand tend to concentrateon the in part to the occurrenceof multiple primary sources, particular characteristicsof the local deposits.This as is the casealongthe River Vaal in SouthAfrica concernwith the unique aspectsof the depositshas (Wagner,1914)and in Guinea(Bardet,1974),aswell led to the continuationof the old adage that "dia- as to alluvial dispersion.Consideration,however,of mondsare where you find them," and many mineral the Zaire-Angolaand Sierra Leone examplesestabexplorationstrategiesare basedon the blanketsample lishesthat diamondscan be widely distributedfrom coverageapproachthat this viewpoint implies. It is relativelylimited sourceareas. The diamondfieldsthat straddlethe Zaire-Angola the purposeof this paperto demonstratethat alluvial diamondsare as amenableto systematicanalysisas border are illustratedin Figure 1; the M'buji-Mayi any othergeologicphenomenonand that it is possible deposits in Zaire are not consideredhere. In thisarea

Oa61-O128/S2/SS/161a-852.50

1618

1614

DONALD G. SUTHERLAND i

20øE

nshoso

ZAIRE

ANGOLA

I 100km

de Corvoiho

FlC;.1. Thediamond fieldsof theZaire-Angola border. Symbols: 1, alluvialdiamond workings; 2, kimberlite localities; $, limitof occurrence of diamonds in Cretaceous hostrocks; 4, crystalline basement

rocks;5, internationalboundary.Henriquede Carvalhohasbeen renamedSaurimo.Sources: Bardet (1974); Fieremans(1961, 1977); Reis (1972).

almostall the mininghasbeenof deposits relatedto during the Cretaceous.The concentrationof diathe presentdrainagenetwork,but as hasbeenmost mondsin the Cretaceous conglomerates is not suffifully documented by Fieremans(1955,1961),the ini- ciently high to be of economicinterest,but the subtial distribution of diamonds acrossthe area was efsequentreworkingand reconcentrationof the diafectedduring the Cretaceous, when a massiveflu- mondsduringincisionof the presentriver system, viodeltaicdepositwas built out into the seathat was alliedto a certainamountof transportfrom the kimthenslowlyrecedingfrom the Zairebasin.Fieremans berlitesby theserivers,hasresultedin the formation (1961) was able to demonstratethat the diamonds of economically viabledepositsas much as 600 km containedwithin the Cretaceous conglomeratic ho- from the source.

rizonsfocused on an areain northernAngola(Fig.

In Sierra Leone, which has limited areas of known

1) where numerouskimberlite bodieshave been dis-

kimberlites, no intermediate hostrockshaveplayed anyrolein thedistribution of thediamonds alongthe presentriversor their formercourses (Fig. 2). The present Atlantic-directed drainage hasprobably been evolving sincetheinitiationof Atlanticriftingin this areaabout185m.y. ago(Dalrympleet al., 1975)and musthavebeena well-developed riversystem by the timeof kimberliteintrusion about92 m.y. ago(Bar-

covered(Friere de Andrade,1952;Reis, 1972). As discussed later in this paper,the diamondsdiminish systematicallyin size away from thesekimberlites;

thisobservation, in additionto thegeographic focus of the alluvialdiamonddeposits on the kimberlites,

establishes thismajordiamondfield asthe product ofessentially oneperiod oferosion anddeltabuilding

TRANSPORT AND SORTING OF DIAMONDS

1615

g-

GUINEA

,gema

Ton( ooo

8ON

0

ATLANTIC

12

km

11øW

I

FIe;.2. Thediamondfieldsof SierraLeone.Symbols: 1, areasof extensive alluvialdiamondworkings; 2, localizedalluvialdiamondworkings;$, kimberlite;4, BullomGroupcoastalsediments; 5, international

boundary.Sources: Hall (1968)and unpub.repts.,SierraLeoneGeological Survey.

det, 1974). The regionaltectonicsituationis that of a slowly subsidingcontinentalmargin allied to a steadilyuplifting watershedarea, and this has produced an incisingriver network on the resultingregional slope.This incision(perhapsacceleratedby Quaternary climatic and sea-levelfluctuations)has given rise to an apparentlysuperimposed drainage network cutting acrossthe maior rock formations and showsclose correspondencewith more minor structural

features at a local level.

distributionof diamonddepositsin someareasaway from the trunk streams,the Sewa and the Moa. Hall

(1968) hasarguedthat an unknowntype of diamond sourcerock is responsiblefor the wide distribution

of diamonds in SierraLeone.The systematic changes in the sizeand characteristics of diamondsalongthe riversaway from the known kimberlites(seebelow) indicate, however, that while undiscovered sources

may exist,they are subsidiaryto the known sources as suppliersof diamondsto the fluvial system.It is

River capture hasbeen relatively frequent in such therefore clear that since the Late Cretaceous diacircumstances, as the incisingriver network hasad- mondshave beentransportedfrom the known source justedto the variationsin the underlying structural areasas far as the coastalBullomGroup sediments, framework, and this has resulted in a rather wide

a distance of over 200 km.

1616

DONALD

G. SUTHERLAND

ZO

•

Nomibson coast

'" ....

Kasoi

"•

Kwango

'" •

N'Go•r•

ß

0.5 i

650 km

0.0 0

50

,oo

km

zoo

,;o

'

350

FI(;. $. The variationin averagediamondsizewith travel distanceß ct/st = carat per stone.

Sorting During Transport

During transportdiamondsare sortedby sizesuch

thisvariationin sizewith distancedownriveror alongshore.

In generalthe transportcurveshavea similarform, erage size of the diamonds.In more detail the size a rather steeplydeclininginitial portionfollowedby distribution of the diamonds may also be expected a more gentlydecliningtail. The final value,toward to changewith transport,while the morphologicchar- which all the curvestend, is due in part to the minacteristicsof the diamondpopulationshouldalsovary imum recoverysize of the mining operations,but in systematically.These three effectsare examined in somedepositsit is claimed that few diamondsoccur at smallersizes(e.g.,Hallam, 1964, p. 716). The rapid turn. initial diminutionin averagediamondsizeistoosteep for the data to be representedby a simpleexponential Size sortingin the directionof transport curve,and insteada goodfit wasfound with the folIn Figure $ the variationsin the mean sizeof dia- lowing modifiedexponentialmodel: mondsalongthe directionof transportfor a number y = a. exp(-b. xU2), of depositshave been plotted. Three of the deposits representfluvial sortingprocesses: Kasai(Fieremans, where y is the average diamond size expressedin 1961), the KwangoRiver (Fieremans,1977), and the carats per stone,x is the distancefrom sourcemeaN'Go•r• in the Central African Republic(Berthoum- suredin kilometers,and a and b are parametersdeieux and Delany, 1956).The otherdepositrepresents terminedby a leastsquaresmethodof curvefitting. alongshoretransportby marine processes in the west In geologicterms,a is equal to the averagediamond coastof Namibia (Stocken,1962). The N'Go•r• de- sizein the sourcearea (i.e., at x -- 0 km) and b is the posithasdata for only about 50 km of transport;the decayconstantthat relatesto the rapidity with which others have all been traced for hundreds of kilomethe averagesizediminisheswith transport.Curvesof ters.The irregularitieson the graphsare likely to be this type have been fitted to the data in Figure g as the result of nonuniformsampling,of local sorting well as to certain other unpublisheddata and give effects, and most probably, of the introductionof goodagreementwith the observations. The theoretdiamonds from secondarysourcessuch as in the icalcurvefor the Namibiandepositisshownin Figure northern part of the Namibian deposits(Hallam, g, and the valuesof the relevant parameters'for all 1964). The term "transport curve" is proposedfor the depositsare given in Table 1. that the farther from the source, the smaller the av-

TRANSPORT AND SORTINGOF DIAMONDS

As shownin Table I there is closeagreementbetweenthevaluesof thedecayconstants, andalthough therearerelativelyfew deposits considered here,this

agreement suggests thatvaluesof b maybe similar in other deposits.The value of the parametera is

clearlyspecific to a givensource area.Themodelcan be appliedto a givenriver systemif the meandiamond size in the sourcearea is known. Thus along the River Sewa in Sierra Leone, if the averagediamond size recoveredfrom the alluvial depositsim-

T^BLE 1.

1617

Parameters for the Theoretical Size Diminution

of Diamondswith Transport

Deposit

a

b

1.93 0.92

0.16 0.15

Kwango

0.83

0.13

N'Go•r•

0.69

0.18

Namibian Kasai

coast

a = averagediamondsizein caratsperstonein the sourcearea b = decay constant

mediatelyadjacentto the kimberlitesat Yengema (Fig. 2) is assumed to be about1 caratper stone(a

notunrealistic figuregiventhe frequentrecoveryof morefrequentlyreportednumberpercent.This has of beingin line with usualsedimenlargediamondsin this area) and if the Sewais as- the advantage sumed to be analogousto the River Kwango in its tologicpracticeand of beingof greaterrelevanceto of the deposit. There is an transportcharacteristics, the predictedaveragesize the economicassessment of diamonds about 170 km downstream is 0.18 carat

artificial lower limit (40.01 carat) to the <0.04-carat

perstone,a figureclosetotheapproximately 0.2 carat classowingto the nonrecovery of very smalldiaper stoneaveragesizerecoveredin prospecting op- mondsduringmining.With the exception of thefarerations in this area. thestdowntransport depositin which there is a sigA finalpointshouldbemadeabouttheimplications nificantpercentage of diamondsin the lowestclass, in this class(whichaccord of thisgeneralmodel:if theaverage diamondsizein the weightpercentages the sourcearea is small,the rapid initial diminution with the smallnumberpercentages for similarclass in sizesuggests thatdiamonds largeenoughto be of sizesreportedfrom otherdeposits, e.g.,Sichel,1975, economicinterestmay only be foundwithin several table$) suggest that there •nay not in fact be many tens of kilometers of their source, rather than the

diamondsbelowthis sizeoccurringin thesedeposits, hundredsof kilometersexemplifiedby the abovede- althoughproblems of recoveryandidentification of posits. verysmalldiamonds makethisstatement difficultto Variation

in size distributions

verify. A numberof syste•naticvariationsare apparentin the size distributions.With increasedtransportit is

The above argumentsare basedon the use of a singlemeasureto providean estimateof diamond apparentthat, togetherwith a shifttowardsmaller sizeat a particularlocality.Inevitablysucha measure diamonds, the distributions become much more is rather imprecise, and it is more common in sedi- peakedand the tail of coarsedia•nonds that is very mentologicalwork to considerthe sizedistributionof prominent in thenear-source deposits isprogressively the particlesunder study, as this distributionmay diminished. In conventional statistical terms the dedepartsignificantly from normaland hencediminish positsshowan increasein sortingand kurtosiswith the utility of the mean (or any othersinglestatistic) a decrease in skewnessas the mean size diminishes asa summaryof the populationsize.This matter has with greatertransport. an added importance in the considerationof diaThesesortingeffectshaveobviousimplications for mondsbecausethere is a nonlinear relationshipbetween the size of a diamond and its value; a small

the prospecting of deposits that are locatedat differentpartsof thetransport curveaswellasfor ultimate

percentageof large diamondsmay representa very economicreturns.The greater frequencyof large diamonds near the source areas means not only a significantproportionof the mining income. It is thereforeof both geologicand economicin- higheraveragevalue(dollarsper carat)for the diaterest to note that there are systematic variations mondsbut alsonecessitates that prospectingsamples

alongthe directionof transportin the sizedistribu- be largerandof greaterfrequency thanthosein detions of diamond populations.These variationsare positsthat haveresultedfrom greatertransport(cf. depictedin Figure 4. In Figure 4a two size distri- Applin, 1972). bution curvesare shownfor depositscloselyrelated Variations in other diamond characteristics to their sourcerocks.Figure4b depictstwo deposits A diamondpopulation is notjustcomposed of diathat are from the middle portion of the transport curve, while Figure 4c representsa depositin the mondsof differing sizesbut alsoof varyingcolors, lower portionof the transportcurve. The sizedistri- crystalforms,and diamondtypes.Diamondpopubutionsare basedon the weight percentof diamonds lationsfrom different sourcesthus are composedof of bort,octahedra, dodecahedra, occurringin the givenclassintervalsrather than the varyingpercentages

1618

DONALD G. SUTHERLAND

4O

TABLE2. (o)

Diamond Types along the River Sewa,Sierra Leone

Locality

Clear

Coated

Bort

(wt %)

•0

/'-'x._

-

-

n = 1700 ct

20 I///X•\\- n=/.90Oct 004

,

i

I

I

I

1

2

3

&

5

"l'-• 6

7

8

9

10

11

12

c!

50

lb)

/"orl•\

.... n:3/,0Oct

20

10

2

3

/,

5

6

7

8

9

10

11

12

ct

60

50

40r I

50 58

45 88

5 4

Middle

66

80

4

77

20

Lower

Sewa Sewa

cleavagestones,translucentor coateddiamonds,etc. (Cotty and Wilks, 1971; Harris et al., 1975; 1979). Transportof diamondsapparentlyaffectsthe original populationby preferentiallydestroyingcertaintypes and more easilytransportingothers. Hall (1968)hasgivenfiguresthat relateto the per-

centagesof clear,coated,and bort diamondsat certain localitiesalong the River Sewa in Sierra Leone, including the area around the kimberlites at Yengema. Table 2 summarizesthese statistics.Quite clearlythere is a preferentiallossof bort and coated

30

0.04 1

Yengema Upper Sewa

(c)

n:t,700ct

stonesduring transport. It seemsprobable that the bort is lost by breakageinto tiny particles;in an experiment with a ball mill, Linari-Linholm (1978) showedthat only six hoursof milling was necessary to reducebort diamondsfrom M'buji-Mayi in Zaire to lessthan 60-mesh size (i.e., < 0.001 carat), while after 950 hoursof milling, gem quality diamonds from the coastaldepositsof Namibia lost only 0.01 percentof their weight. The consistentreductionin the proportion of coated diamondsdown the Sewa may be due to the effectsof surfaceattrition, since the greencoatingson the SierraLeonediamondsare frequentlyrather shallowand are particularlybrittle (Grantham and Allen, 1960). The proportions of cleavageand broken diamonds (in the sizesrecovered) also decreaseswith transport; the Namibian coastaldiamonds,for example,are more than 95 percentcomposed of wholecrystalforms(Hallam, 1964). The dominant crystalforms of diamond in kimberlite

are the octahedron

and the dodecahedron

(Harris et al., 1979). Transportappearsto favor the dodecahedron as the depositsalongthe westcoastof South Africa have disproportionatelyhigh percentagesof this form. This effect is understandable in 20 termsof the mechanicsof initiation of particle movement;the morenearlysphericaldodecahedral crystals would become entrained more easily than the more 10 elongateoctahedra.Dodecahedrawill therefore be subjected,on average,to a greaternumber of transport eventsand will alsobe in motion longerduring OJ3/. 1 2 3 4 5 6 7 8 9 10 11 12 any one suchevent. C! Surfaceabrasionfeaturessuchas percussionmarkingsand roundedcornersmay be expectedto result FIG. 4. Weight percentagediamondsizedistributionsfrom difbut abradeddiamondsarealsowidely ferent partsof the transportcurve.a. near source.b. middle part from transport, of curve. c. lower curve. ct = carat. reported from kimberlites (Grantham and Allen,

*•30

TRANSPORT AND SORTING OF DIAMONDS

1619

1960; Harris et al., 1975). This factor, togetherwith the apparentlyvariableresponse of differentdiamond typesto abrasion,suggests that there may be considerable differencesin the degreeof developmentof thesesurfacefeaturesbetweenseparatedepositional trainsthat are independentof the distancetraveled. In summary,it is apparentthat transporttendsto increasethe quality of the diamonds,preferentially removingbort, mechanicallyweakerdiamonds,and cleavagestones.The greater easewith which more

Thisanomalyof verylargeaverage-sized diamonds nearthemouthof theOrangeRivercanbe explained if (1) the generaltransportmodeldoesnot holdfor the OrangeRiver, (2) the averagediamondsizein the sourcearea is very high, or ($) there is a source (or sources) for a significantproportionof the dia-

rounded

workingsin the BarklyWestarea.The information

forms are entrained

also contributes

to the

mondsmuchnearerthan the Kimberleyarea. There

is no independentevidencefor the first possibility, while the secondis contradictedby the size of the diamonds that were recovered from the River Vaal

pointsto the third possigeneralincreasein quality becausesuchdiamonds on diamondcharacteristics elsewhere(Keyser, are likelyto sufferlessweightlossduringcutting.The bility, whichhasbeensuggested line of reasoning. The absolutevalue (dollarsper carat) of a diamond pop- 1972)from an independent ulation from a particular part of a transportcurve is lack of kimberlite satellite minerals associatedwith therefore a balance between the reduction in size and

the increasein quality that are complementaryto the

thesediamonddeposits arguesagainsta nearbyprimary source,and it may be that the diamondswere hostedin Karroo or older rocks and subsequently

transport process.

freedby erosionassociated with the downcutting of Discussion

the Orange River.

Size diminutionduring transportationcan result from eitherthe preferentialcarryingof smallerparticlesor breakageand attrition of larger particles. Both theseeffectsoperateduring the transportation of diamonds,breakageand attritionbeingindicated by the disappearance of inferior quality diamonds and the roundingof the cornersof more resistant ones,while preferentialcarry is indicated by the The decrease in diamond size northward from the changingproportions of crystalforms.The Namibian are of particularinterestbecause the mouth of the Orange River depicted in Figure $ is coastaldeposits similar to the variations in size northward from the diamondsthere almostentirely exhibitcrystalforms mouthsof the Buffels, Swartlientjes,and Olifants riv- and are not brokenor cleavagestones.Breakagecaners along the west coastof South Africa (Hallam, not, therefore, be important in producingthe size 1964). These relationshipsare most easily explained diminutionalong the coast,which must be domiby the fact that the rivers are the meansof intro- nantly a sortingprocess. In the precedingsectionsthe generalfeaturesof the variationsin diamond populationshave been outlined. The data for thesegeneralizations have been takenfrom a variety of localities,includingexamples of both alluvial and coastaldeposits.An anomaly is revealed,however,by the data on size diminution. This concernsthe coastaldepositsof Namibia.

duction of the diamonds to the coastal zone, after

whichalongshore waveactiontransports and sortsthe diamondsin a northerly direction accordingto the dominant fetch along this coast.Certain diamonds

Conclusions

This paperhasdemonstrated that diamondsfrom relativelyrestrictedsourceareascanbe transported of kilometers anddistributedover ange River have been recognizedas very probably for manyhundreds of squarekilometerseitherby forhavingoriginatedin theKimberleyarea(Wagnerand tensof thousands recovered from the coast near the mouth of the Or-

Merensky,1928, p. 20), which implies a transport mer riversthat depositedintermediatehostrocksor distance for some of the diamonds of at least 1,600

km (seealsoWilliams, 1952). The averagediamond sizenear the mouth of the Orange River is, however, over 1 carat per stone(Stocken,1962; Hallam, 1964), which is in marked contrastto the predicted small diamondsizesthat shouldprevailafter severalhundred kilometersof transport,accordingto the model presentedearlier and supportedby the size variation along the coastnorth of the Orange River mouth. Furthermore, the diamonds recovered from the ter-

by rivercaptures relatedto the evolution of the present drainagesystem.Littoral processes are alsoeffectivein transporting diamonds for hundreds of kilometers.

Duringtransportation a varietyof effectshavebeen indicated.The averagediamondsizediminishesaccordingto a modifiedexponential rule, and thissize diminution is accompaniedby better sorting,an increasein the kurtosis,and a decreasein the skewness of the diamond size distributions.In general, poorer

qualitydiamonds suchasbort,diamonds with inclusions, andcleavage stones arepreferentially destroyed erageover 1 earat per stone(de BeersGeol. Dept., during transport,while the more roundedcrystal 1976). forms(suchasthe dodecahedra) appearto be more racesthat flank the Orange River for over 50 km inland of its mouth also contain diamonds that av-

1620

DONALD G. SUTHERLAND

easily transported.The general quality of the diamondsis thereforeimproved. The above effects are of considerable relevance for

du Toit, A. L., 1951, The diamondiferousgravelsof Lichtenburg: SouthAfrica Geol. Survey Mere 44, 38 p. IVieremans, C., 1955, Etude g•ologiquepr•liminaire desconglom•rats diamantifb•res d'age M•sozoiqueau Kasai:Inst. G•ol. Louvain M•m., v. 19, p. 224-293. -1961, Origine et r•partition de la min•ralisationdiamantifb•re au Kasai occidentale(Congo) et dans le nord-estde la Lunda (Angola):G•ol. Soc.BeigeBull., v. 70, p. 89-95. -1977, Het voorkomenvan diamant langsheende Kwangorivier in Angolaen Zaire:Acad.RoyaleSci.Outre Mer Brussels,

diamond exploration (Lampietti and Sutherland, 1978), both at the regional scalein identifying primary sources or large volumealluvial targetsand at the small scalein decidingon samplingpatternsfor alluvial deposits.It may be presumed,for example, that depositswith a large proportionof inferior diaNat. Sci., v. 20, pt. 1, p. 4-28. mondsand bort (e.g., the Lichtenburgdepositsin Frierede Andrade,C., 1952,On thediscoveryof a kimberlitetype of igneousrock in the diamondfieldsof Lunda: Internat. Geol. SouthAfrica, du Toit, 1951) have resultedfrom relCong., 19th, Algiers1952, ComptesRendus,Sec.21, p. 55-61. atively slighttransportationand are relatively close Grantham,D. R., and Allen, J. B., 1960, Kimberlite in SierraLeone: to their source. The converse, however, need not be

true, as the primary sourcemay be a high-quality producerwith only a small proportionof inferior diamonds.If the qualityand the sizeof the diamonds in a source area are known, the likelihood of there

being extensiveplacer depositsdownstreamof this zone can be assessed. Thus a sourceexhibitinglow averagediamondsize and/or a high proportionof inferior quality diamondsis lesslikely to produce placer depositsof economicinterestfor more than several tens of kilometers

downstream.

Acknowledgments

Thanksare due to M. Dale, Placer AnalysisLtd., and D. Hodgson,Departmentof Geography,University of Edinburgh, for commentson the initial draft of this paper. REVERENCES

Applin, K. E. S., 1972, Sampling of alluvial diamond depositsin WestAfrica:Inst.Mining MetallurgyTrans.,v. 81, sec.A, p. 6277.

OverseasGeol. Mineral Resources, v. 8, p. 5-25. Hall, P. K., 1968, The diamond fields of Sierra Leone. Sierra Leone Geol. Surv. [lull., v. 5, 133 p.

Hallam, C. D., 1964,The geologyof the coastaldiamonddeposits of southernAfrica (1959),in Haughton,S. H., ed., The geology of someoredeposits in southernAfrica:Johannesburg, Geol.Soc. SouthAfrica, v. 2, p. 671-728. Harris,J. W., Hawthorne,J. B., Oosterveld,M. M., and Wehmeyer, E., 1975,A classification schemefor diamondand a comparative study of South African diamondcharacteristics: Physicsand Chemistryof the Earth, v. 9, p. 765-784. Harris, J. W., Hawthorne, J. B., and Oosterveld,M. M., 1979, Regional and local variations in the characteristicsof diamonds from somesouthernAfrican kimberlites,in Boyd, IV. R., and Meyer, H. O. A., eds., Kimberlites,diatremes,and diamonds: Their geology,petrology,and geochemistry: Washington,Am. Geophys.Union, p. 27-41. Junner,N. R., 1943, The diamond depositsof the Gold Coastwith noteson otherdiamonddepositsin West Africa: Gold CoastGeol. SurveyBull. 12, 54 p. Keyset,U., 1972, The occurrenceof diamondsalongthe coastbetween the Orange River estuary and the Port Nolloth reserve: SouthAfrica Geol.SurveyBull., v. 54, 23 p. Lampietti,IV.J., and Sutherland,D. G., 1978, Prospecting for diamonds-somecurrentaspects:Mining Mag., v. 139, p. 117-123. Linari-Linholm,A A., 1973,The occurrence, mining,and recovery of diamonds:London,de BeersConsolidatedMines Ltd., 44 p. Reis, E., 1972, Preliminary note on the distributionand tectonic controlof kimberlitesin Angola:Int. Geol.Cong.,24th, Montreal

Bardet,M. G., 1974, G(•ologiedu diamant:Bur. RecherchesGeol. 1972, Sec. Rept. 4, p. 276-281. Min. [Paris]M•m. 83, v. 2, 223 p. deposits: Berthoumieux,13., and Delany IV., 1956, Rapport sur la mission Sichel,H. S., 1973,Statisticalvaluationof diamondiferous SouthAfrican Inst. Mining Metall. Jour.,v. 73, pt. 7, p. 2.35-243. diamant Ouest Oubangui:Unpub. Rept., Central African ReStocken,C. G., 1962, The diamond depositsof the Sperrgebiet, public, Dir. Mines G•ol. SouthWestAfrica:GeolSoc.SouthAfrica Ann. Cong.,5th, 1962, Cotty, W. F., and Wilks, E. M., 1971,DifferencesbetweendiaIVieldexcursionguide, p. 1-16. mondsfrom different sources:Diamond Research1971, p. 8-11. Dalrymple, G. B., Gromm•, C. S., and White, R. W., 1975, Po- Wagner, P. A., 1914, The diamondfieldsof southernAfrica (1971 facsimileed.): Cape Town, Struik, 355 p. tassium-argon age and paleomagnetism of diabasedykesin Liberia: Initiation of central Atlantic rising: Geol. Soc. America Wagner, P. A., and Merensky,H., 1928, The diamonddepositson the coastof Little Namaqualand:Geol. Soc.SouthAfrica Trans., Bull., v. 86, p. 399-411. v. 31, p. 1-41. de BeersGeol. Dept., 1976, Diamonds,in Coetzee, C. B., ed., Mineral resources of the Republicof SouthAfrica, 5th ed.: Pre- Williams,A. F., 1932,The genesisof the diamond:London,Ernest Benn, 2 vols.,636 p. toria, SouthAfrica Geol. Survey, p. 17-30.