<<

ScientiJic Advances in the Last Decade By Melissa B. Kirkley, John J. Gurney, and Alfred A. Levinson

Cr;ontiGo nArmnrne ;n tho nnct AornAo he A~P~APn( the loanc ,,TAT mqinr ~ATF-~PPCin mrr UUIYIILIIIY CIU YUI.YD0 111 CllY YClLIC UYUUUY TT)lib UCbLIUb VI LIIb L/VVU OLIVV 1114)VL LIUILIIIUbO 111 VUL hove completely altered our understand- understanding of the age, origin, and emplacement of ing of certain concepts relating to the age iLiamonds. Much of the new data and their inter~reta- and origin of . As a generaliza- 1 %are found in highly technical scientific journais and tion, most diamonds formed more than conference proceedings that are rarely encountered by 990 million years ago, deep within the gemologists. Therefore, we have prepared this review earth, from either of two types, peri- dotite and . They were stored be- article to update gemologists on some of the latest facts low the base of for varying periods and concepts with respect to the above topics. of time, some as long as 3,200 million The information contained in this article is applicable years, before being transported to lhe sur- to virtually all natural diamonds, both gem and industrial face. and lainproite, the two (figure I), except possibly for certain rare types of dia- rock types usucrlly associuted with dia- monds, such as those referred to as "fibrous" or "coated," monds, are only the mechanisms that and microdiamonds, as well as for diamonds related to brought diamonds to the surface and are meteorite impacts. In addition to our own experiences, we in no way related to the formation of have drawn from many volumes in the technical literature. most diamonds. Other topics that are For those interested in pursuing these topics further, we somewhat more speculative, for example, recommend the boolzs by Ross (1989), Nixon (1987), the source of for ihe crystolliza- tion of diamonds and the mechanism of Mitchell (1986), Glover and Harris (1984), Kornprobst kimberlite and emplacement, (1984), and Dawson (1980), and the review articles by are also discussed and ihe latest concepts Gurney (1989) and Meyer (1985). presented. AGE OF DIAMONDS Until recently, one of the major unresolved problems in research revolved around the age of diamonds. I Age dating of diamonds assists in understanding their ABOUT THE Atlrm.9 origin, which is a significant factor in diamond explora- Ds Kirk is post-doclmt r~& dficec end 01.f urmy Is press h h Depalnen ' tion. For many , age can be determined directly a1 Gwhemistry, WwsiIy d Cape Town, using a number of well-established geochronological tech- Randebasch, Sooth Africa, DE Lwihscin k prc niques, such as the - (U-Pb)method. However, fsssor h It)@ Deparfment of Cedogy and Geo physics, U~milyd Calgary, Alberfa. because diamond is essentially pure carbon, it does not ha&. contain any of the radiogenic elements on which such Achnowkdgnents: The authors sincere& than methods depend. Even the well-lznown carbon-14 (14C) 0s A. J A. Arise and Mr. F! d Darragh, lbolh method is useless for diamonds because it is restricted to al Wth, Auslralie, lor lhelr careful m&w d fhe nwivscripr. , organic carbon that has been involved in the earth's recent Gem & Gemology, 161.27,k. I, pp 2-25 near-surface carbon cycle. Although diamonds themselves cannot be dated, some O 1991 demological Institute d Amriia of their minute inclusions, such as and ,

2 Origin of Diamonds GEMS & GEMOLOGY Spring 1991 Figrue I. Today, dia- monds are the most pop- ~~largemstone and a valuable industrial mate- rial. As a consequence, there has been consider- able research into the geologic origins of dia- monds to aid in explora- tion and mining. The last decade, in particu- lar, has prodczced some important advances in our rlnderstanding of the complex processes re- quired /or the formation and deposition of dia- monds. The ultimate re- sult of such research is evident in these superb earrings and necklace. The three large diamonds in the necltlace are (from the left) 9.81, 16.18, and 12.73 ct; they are sur- rounded by 254 dia- monds with o total weight of 67 ct; the ear- rings contain 84 dia- monds wit11 a total weight of 11 ci. {ewelry by Van Cleef and Arpels; photo courtesy of Sotheby's, New Yorl<.

can, because these minerals contain measurable octahedron) of the diamond rather than that of quantities of the elements involved in radioactive their species (figure 3). decay systems. Some inclusions (e.g., garnet) were Several attempts were made prior to 1981 to formed at the same time, and in the same place, as date inclusions in diamonds; the study by Kramers their host (e.g.,diamond), so that the age of (1979) is the most significant. Using lead (Pb) the inclusion is also the age of the host. Detailed isotopic compositions of sulphide inclusions in studies of , garnet, pyroxene, , and diamonds, he determined ages on the order of 2,000 other minerals in diamond have shown that these million years (My)for inclusions in diamonds from minerals were growing adjacent to the diamond, the Finsch and Kimberley pipes in ; which then grew around and enclosed them (fig- those from the Premier mine appeared to be about ure 2). This physical relationship between dia- 1,200 My in age. However, Richardson et al. (1984) mond and its cogenic inclusions is sometimes were the first to date successfully a significant reflected by the crystal form of the silicate inclu- number of inclusions in diamonds, specifically sions, which take on the morphology (called cubo- inclusions of garnet in Finsch and Kimberley

Origin of Diamonds GEMS & GEMOLOGY Spring 1991 3 Figure 3. nigons, which are typical of a dia- mond octahedron, can be seen on the face of this flattened inclusion of garnet, with a Figure 2. This garnet inclusion in a diamond colorless (orthopyroxene) or olivine (I mm across) from the Finsch mine, South crystal at the end. These features are taken as Africa, was dated by geochronological methods evidence that the diamond has forced its crys- to be about 3,300 million years old. Because tal habit on the guest mineral during the simul- such inclusions undoubtedly formed at the taneous crystallization of inclusion and host. same time as their hosts, they are the best Photomicrograph by Eduard 1. Giibelin; trans- means of age dating diamonds, which cannot mitted illumination, magnified 50 x . From Gii- be tested directly by standard dating methods. belin aid Koiv~ila(1986, p. 95). Photo courtesy of Dr. S. H. Richardson.

berley pipe are as much as 3,200 My older than the age of lzimberlite emplacement (i.e., when diamonds, by means of the relatively new samar- the pipe reached the surface, about 100 My ium-neodymium (Sm-Nd) geochronological ago]. This example implies that: (a)diamonds method combined with the rubidium-strontium can be stored deep within the earth for an (Rb-Sr)technique. Table 1 summarizes these re- extended period of time before being carried to sults along with more recent data (Richardson, the surface by the lzimberlite; and (b) lzim- 1986; Richardson et al., 1990) on diamonds from berlite is merely the transporting medium for several other pipes in and Austra- bringing diamonds (aswell as other materials) lia. (For a discussion of some earlier studies that to the surface. This process has been pictur- have a bearing on the dating of inclusions in esquely described by the analogy of an elevator diamonds, see Meyer, 1985, and Gurney, 1989.) or a bus (thelzimberlite] picking up passengers The results presented in table 1 may well be (diamonds) in the earth's along its the most striking information about diamonds to route of ascent toward the surface. Note, emerge in the past decade, not only because they though, that there is no resolvable age differ- put accurate ages on diamonds in millions of years ence between the diamonds and lzimberlite (My) but also because of other implications, as emplacement at the Premier mine. This sug- discussed below: gests that, in this example, they may be contemporaneous. Whether the Premier mine 1. Diamonds are old and may have been forming is unusual awaits the determination of dia- continually, certainly intermittently, through- mond ages from additional localities. out most of earth's history. The 2,300 My period between -3,300 and 990 My represents 3. In samples from the Finsch mine, two ages- about half of the earth's 4,500 My existence, -3,300 and 1,580 My-have been obtained, and inclusions in diamond may yet be found the former for peridotitic, and the latter for that extend this range. eclogitic, inclusions. These two main types of inclusions in diamond are discussed in greater 2. Diamonds are usually very much older than detail below. The presence of diamonds of the lzimberlite that brought them to the sur- different ages within one pipe can, for the face. For example, diamonds from the Kim- present, most easily be explained by the fact

4 Origin of Diamonds GEMS & GEMOLOGY Spring 1991 that lzimberlites may obtain their diamonds mond at an extraneous source and its capture from more than one geologic environment by the Premier Izimberlite. (mantle source) during their rise toward the Although the concept that diamonds are surface. xenocrysts in lzimberlite was proposed as early 4. The data reported in table 1 have settled a long- as 1905, it was not until the Third ~nterna- standing debate in which scientists advocated tional Kimberlite Conference, held in France one of two hypotheses with respect to the in 1981, that "there was a realization that origin of diamonds. The " school" diamonds are xenocrysts in lzimberlites" (Mit- maintained that diamonds originally formed chell, 1986, p. 8). The proof, however, has at depth from the crystallization of a lzim- come only since 1984, with the work of Rich- berlite (molten mass) and, hence, are ardson and his associates on the age of dia- genetically related to the magma and are monds and their host roclzs (table l). ihenocrys-ts(def: a relatively large crystal set In summary, recent geochronological studies in a fine-grained groundmass to which it is clearly show that most diamonds: (a) are much genetically related [Gk, pheno: to show + older than the volcanic roclzs (lzimberlite and cryst(a1)j).The "xenocryst school," on the lamproite) that carried them to the surface; (b)are other hand, believed that diamonds were not genetically related to these volcanic roclzs; and formed prior to the intrusion of lzimberlite, are (c)have crystallized, possibly episodically, during not genetically related to it, and are merely a large part of earth's history. Further, it must be accidental inclusions lznown as xenocrysts emphasized that it is the ability to date mineral (def: a crystal or a fragment of a crystal inclusions separated from diamonds, which has included in a magma and not formed by the only existed duri~gthe last decade, that has made magma itself, i.e., foreign to it [Glz. xeno: age dating of diamonds possible. This is further stran&, foreign + cryst(al)]).In view of the evidence that mineral inclusions have great scien- newly' confirmed fact that diamonds are, in tific value in addition to their gemological impor- general, much older than their lzimberlite (or tance. lamprbite) host roclzs, the xenocryst theory is now lznown to be correct. The data from the ORIGIN OF DIAMONDS Premier mine (diamondand lzimberlite are the The study of the origin (genesis]of diamond ideally same age) could be the result of an extremely involves the collection, assimilation, and inter- short interval between the formation of dia- pretation of a vast amount of data from many

TABLE 1. Ages of diamonds and emplacement of associated kimberlitea pipes in My (millions of years), and type of inclusions.

Age of Age of emplacement Type of Location diamonds of kimberlitea pipe inclusions Reference (mine) (MY) (MY) in diamonds

Kirnberley, -3,300 -100 Peridotitic Richardson et al. (1984) South Africa Finsch, -3,300 -100 Peridotitic Richardson et al. (1984) South Africa Finsch, 1.580 -100 Eclogitic Richardson et al. (1990) South Africa Premier, 1,150 1,100-1,200 Eclogitic Richardson (1986) South Africa Argyle, 1,580 1,100-1,200 Eclogitic Richardson (1986) Australiaa Orapa, 990 -100 Eclogitic Richardson et al. (1990) Botswana

- - -- =In the case of the Argyle sample, the pipe is a lamproite.

Origin of Diamonds GEMS & GEMOLOGY Spring 1991 5 strange, foreign + lith: stone, rock]. are not formed from the magma itself; rather, they become included in the magma as it rises. Xeno- liths may be as small as a single crystal, boulder size (figure 4), and even much larger (figure 5). Xenoliths in and repre- i;.&<*k-2 ;?. .. -. sent fragments of wall roclz adjacent to an intru- .. .>.'+.lk:> p .& ;?&I;* *+ , ..-.: .?--- ?T. .r sion that have brolzen off and been incorporated '. 5..5.5:'. .,:' into the magma as it works its way along fractures r,, . ,,:, ,*.:-+ ., . or cracks to the surface. Thus, xenoliths may Figure 4. Boulder-size xenoliths of eclogite, represent bloclzs of buried crustal formations which frequently contain diamonds, are seen brought closer to the surface, such as metamorphic here at the Roberts Victor mine, near Kim- rocks derived from deep-seated terrains within the berley; South Africa. Study of xenoliths such as earth's lower , or sedimentary rocks in the these, which were brought up from the earth's upper crust and, perhaps most importantly, roclzs mantle (depth at least 150 krn) by kimberlite, believed to be derived from the earth's upper enable scientists to determine pressure- tem- mantle. Kimberlites and lamproites are the only perature (I)-T)conditions within the mantle means of obtaining samples of such deep roclz and, by analogy, P-T conditions for the forma- types. tion of diamonds. Xenoliths are usually rounded, especially if they originate at great depths, probably because of disciplines related to the physical, chemical, and chemical dissolution at the margins of the frag- mineralogic properties of diamonds, the roclzs in ments (e.g., figure 4), but they are likely to be which they crystallize, as well as the roclzs that angular if they originate from shallow wall rocks. brought them to the surface and in which they now Xenoliths that contain diamonds are extremely are found as primary deposits. Clearly, such a important because they permit us to determine the broad survey is beyond the scope of this article. characteristics of the rock types from which dia- Therefore, we will focus on those aspects of monds crystallize. Such characteristics include diamond genesis that are of the greatest interest to chemical composition, pressure-temperature reg- gemologists, particularly those in which major advances have been made in the past decade. Such topics include: (1)the types and properties of roclzs Figure 5. This rounded of sandstone, in which diamonds form; (2)the pressure-tempera- about 2 m in its longest dimension, is an exam- ture conditions under which diamonds crystallize ple of rock broken off from the walls of the con- within the earth; and (3)the source of the carbon of duit along which the kimberlite magma as- which diamond is composed. cended to the surface. As sandstone is charac- teristic of the upper part of the earth's crust Rocks in Which Diamonds Form. Now that it has (generally found within 10 km of the surface), been established that lzimberlite and lamproite are this xenolith will not contain diamonds. Exact only the transporting mechanisms for bringing location unknown (probably South Africa). diamonds to the surface, and are not genetically related to diamond, the question that must be answered is: From what type of magma or pre- existing material do diamonds actually crystal- lize? The two most rewarding areas of investigation in which answers to this question have been found are: (1) the study of diamond-bearing xenoliths, and (2) the study of mineral inclusions within diamonds.

Diamond-bearing Xenoliths. A xenolith can be defined as a rock fragment that is foreign to the igneous mass in which it occurs [Glz, xeno:

6 Origin of Diamonds GEMS & GEMOLOGY Spring 1991 Figure 6. Altered olivine and other - rich minerals (a "kelyphite" rim) surround this 8-cm-long xenocryst of garnet. The alteration resulted from reaction with kimberlite fluids. Figure 7. The polished surface of this eclogite Cl~emicalreactions, along with fragmentation xenolith specimen displays the pyrope-alman- by physical forces, tend to break up xenoliths, dine garnet (red mineral) and clinopyroxene especially those of the type, releasing (green) that are its chief components. From the xenocrysts of diamond. From the Monastery Roberts Victor mine, near Kimberley South mine, South Africa. Africa. ime, the nature of any (e.g., CO,, H,O) garnet (-pyrope) and green pyroxene present, and other parameters pertinent to de- (technically jadeitic clinopyroxene or a solid solu- ciphering the genesis of diamonds. and tion between jaaeite and ), with minor are the predominant xenoliths found to amounts of , lzyanite, , and contain diamonds (see box A for details of the (figure 7). Eclogitc is indicative of a high pressure- mineralogy, , and classification of these high temperature (with emphasis on the former) two roclz'types). environment, consistent with that in which dia- During transportation of both peridotitic and eclogitic xenoliths within kimberlite or lamproite, fragmentation of the xenoliths may take place, Figure 8. Diamonds protrude from the surface of this broken eclogite xenolith obtained from adding smaller xenoliths or even xenocrysts to the the Ardo-Excelsior mine, South Africa. Such transporting magma; this is probably the best way samples, corroborated by experimental studies to explain the occurrence of diamond xenocrysts, in which the pressure-temperature conditions as well as single crystals of other minerals (e.g., required for the formation of garnet and garnet, chromite), in lzimberlite and lamproite clinopyroxene are taken into account, are con- rocks. It is also important to recognize that xeno- sidered evidence that diamond may crystallize liths in kimberlite and lamproite do not always represent with fidelity the mineralogy and pres- sure-temperature conditions of the mantle in which they are supposed to have formed, because extensive modification may take place as they react chemically with the fluids in the transport- ing (figure 6). Inclusions, however, are shielded from extraneous influences by the sur- rounding diamond, so the conditions of formation they indicate are considered better representations of the actual mantle conditions in which diamonds formed. By comparing i~lclusioiland xenolith com- positions, scientists are confident that they can recognize altered vs, unaltered compositions in their study of eclogites and peridotites. Eclogite is a coarse-grained consisting essentially of a granular aggregate of red

Origin of Diamonds GEMS & GEMOLOGY Spring 1991 7 BOX A: CLASSIFICATION AND NOMENCLATURE OF IGNEOUS ROCKS

The subdivisions of igneous roclts named in figure roxene, , and olivine. Typically a roclt of A-1 are those used by most geologists. Of special this type will have 42-52 wt.% SiO, and 16-26 wt.% significance in this review article are the terms nlafic combined FeO and MgO. (extrusive or volca- and ultramafic, because ltimberlite, peridotite, and nic) and (iiltrusive or plutonic) are the most eclogite are ultramafic roclts. The term (from common mafic roclts. The term ultramafic is applied magnesium and ferric [])is used to describe a roclt to a roclt with even lower SiO, than a mafic roclt (we composed chiefly of one or more iron-magnesium use the division at 42 wt.%) and, by analogy, even (ferro-magnesium)dark-colored minerals such as py- higher FeO and MgO. These roclts are composed of

Rhyolite Andesile Basalt Figure A-1. This illustra- 1 I. Dor~te Gabbro 1 tion shows the miner- alogic compositioi~s,se- I I intermediate I maftc I ultramatic I mineral lected chemical changes, percent and changes in color as- sociated with the com- mon extrusive and intru- sive igneous roclts. The center of each vertical column represents the

average- mineralogic- and chemical con~positionsof each rock. For example, the average basalt (see vertical brolt en line) would have about 48 3 9 16 26 40 FeO+MgO (%) vol. % pyroxene, 28 vol. % I I I I amphibole, 24 vol.% cal- cium-rich , K20+Na20 and a trace of olivine. A chemical arlalysis would yield about 47 wt.% SiO, and 20 \avt.%FeO Light Gray Dark Color + MgO. monds form (figure 8).Typically, eclogite occurs in amounts. Peridotite is believed to be the most deep crustal metamorphic regions below conti- common and abundant rock type in the earth's nents; it became eclogite by means of solid-state mantle. Most peridotitic diamonds are formed in (metamorphic)transformation of previously exist- garnet-bearing , with minor amounts ing rock, probably basalt. Eclogite found in the formed in (Gurney, 1989). mantle probably forms in the same way, through Following our practice of illustrating impor- of crustal rocks (discussed below). As tant concepts and features, such as eclogite xeno- an aside, in its own right unaltered eclogite could liths (figures 4, 7, 8), we would like to show a well be considered a gem material because of the photograph of a diamondiferous peridotite xeno- combination of two very attractive gem minerals lith, but such xenoliths are extremely rare and are (again, see figure 7). typically extensively altered. Whereas over 100 Peridotite is a general term for a coarse- diamond-bearing eclogites have been described in grained ultramafic rock consisting chiefly of ol- detail in the literature, there are probably fewer ivine with or without other mafic (high in Fe and than 20 comparable descriptions for diamond- Mg) minerals, such as (again, see box A). iferous peridotite xenoliths. This is most remark- Garnet and frequently occur in small able, since harzburgite is probably the source of

8 Origin of Diamonds GEMS CI: GEMOLOGY Spring 1991 Figure A-2. This classification 01 scheme for ultramafic (intrusive) ------A rocks, recommel~dedby a Committee of the IUGS (hlternational Union of Geological Sciences), is b~lsedon the proportions of olivine (Ol), orthopyroxene (Opx), and clinopyroxene (Cpx). The apexes of the triangle represent 100% each of olivine the above-mentioned minerals. Most orthop~roxefite olivlne websterile diamoi~dsfound in rocks broadly PYROXENtTE called peridotite are actclally from rocks more acc~~ratelyclassified as harzbrlrgite (see text for details). Eclogite is actually a specific case of clinopyroxene plus garnet, as is shown at the lower right conler of Cpx + garnet = ECLOGlTE the diagram. -----

olivine and pyroxene, usually together as in peri- ene. That is, peridotite must contain more than 40 dotite, but each may constitute a rock on its own: vol.% olivine (Ol),with the remainder orthopyroxeile and , respectively. (Opx)and/or clinopyroxene (Cpx).Peridotite, in fact, Tho ultramafic rock peridotite is particularly is actually a general name that includes the more import;lnt in gemology both because it is one of the specific rock types dunite, harzburgite, Iherzolite, and , the distinction between them being the two rocks from which diamonds crystallize (eclogite is the ?ther) and because it is the rock from which relative proportions of 01, Opx, and Cpx. For exam- melts of ltimberlite and lamproite (the two rock types ple, dunite must contain at least 90 vol.% olivine, that transport diamonds to the surface) originate. with Opx and Cpx combined providing the remaining volume percentage. Harzburgite, the variety of peri- From figure A-2, however, we see that there are dotite from which many diamonds crystallize, con- two broad categories of ultramafic rocks, peridotites tains 40-90 vol.% olivine, 5-60 vol.% orthopyroxene, and . The classification of a rock as and not more than 5 vol.% clinopyroxene. As the peridotite or pyroxenite depends on the volume diagram indicates, eclogite is actually clinopyroxene percentages of the mrafic minerals olivine and pyrox- plus garnet.

most of the diamonds in lzimberlite (discussed in mineral inclusions in diamonds has been greatly detail in the next section). Clearly, diamond-con- aided by the advent of new analytical techniques, taining peridotite xenoliths must have been disag- which have resulted in major discoveries during gregated by some incredibly efficient process. Sev- the past decade. As every gemologist lznows, even eral mechanisms have been proposed to explain those "large inclusions" that can be seen with the this phenomenon, based mainly on laboratory nalzed eye in diamonds classified as I, or I, are very experiments involving the gases CO, and H,O, small indeed. Yet, such instruments as the elec- which are lilzely to be found in the mantle. It has tron microprobe and the even more recent ion been suggested that these gases are released follow- microprobe now make it possible to analyze chem- ing certain mineral reactions (e.g.,the brealzdown ically minerals and rock fragments as small as one of dolomite, for CO,] within the peridotite area of micrometer (Fm, one millionth of a meter). Other stability in the mantle, resulting in the self- analytical techniques, such as Raman spectro- destruction of diamondiferous peridotite xenoliths scopy and X-ray diffraction, permit the identifica- (Gurney, 1989, p. 957). tion of minerals while they are still within the diamond but cannot provide chemical data. Mineral Incl~zsionsin Diamond. The study of The subject and importance of mineral inclu-

Origin of Diamonds GEMS & GEMOLOGY Spring 1991 9 sions in diamonds have been reviewed comprehen- anite, corundum, and coesite - are minor constit- sively by Meyer (1987)and by Gurney (1989).As was uents characteristic only of the eclogitic assem- explained above in connection with the dating of dia- blage (figure 9).The remaining 11 minerals are so monds, the assumption is made that in order for rare that they need not be considered here. Unfor-. most minerals (e.g., garnet) or aggregates of min- tunately, inclusions in diamonds cannot be used to erals (e.g., eclogite or peridotite) to be included determine geographic origin, as is frequently pos- within another mineral (e.g., diamond), both the sible with colored stones (Giibelin and Koivula, inclusion and host must have beell forming at the 1986, p. 88). same time and place. Therefore, we can conclude Detailed chemical studies of the most impor- that both have a common origin. Thus, if the inclu- tant minerals listed above have made it possible to sions are of the peridotitic assemblage it follows that characterize with confidence 98% of all included the diamond formed within peridotite host rock. diamonds as peridotitic (mainly harzburgite) or Most of the mineral inclusions in diamond are eclogitic (sometimes referred to, respectively, as very small (- 100 pm)and usually are composed of "Ftype" and "E-type" diamonds). The minerals in just one mineral (monomineralic); however, bi- one type of mineral assemblage chemically will mineralic and polymineralic inclusions do occur. not be the same as those in the other type; that is, The multiphase mineral inclusions are partic- they are mutually exclusive. The only exception ularly important not only in determining peridoti- would be the very rare case in which the diamond tic or eclogitic origin, but also because analyses of began to grow in one environment (e.g., eclogitic) the chemical and physical properties of two or and then was moved and later continued to crystal- more coexisting minerals enable us to estimate the lize in a different environment (e.g., peridotitic). pressure and temperature environments in which However, both P- and E- type diamonds may be they and, by analogy, the host diamond formed. found within the same lzimberlite pipe, indicating In total, 22 contemporaneously formed (syn- that the pipe sampled at least two different "dia- genetic) minerals have been found as inclusions in mond source areas" en route to the surface. diamond, including diamond itself (Gurney, 1989). A logical question could be raised concerning Six of these occur in both the peridotitic and the categorization of a diamond as P- or E-type if eclogitic assemblages: olivine, orthopyroxene, the inclusion consisted of a single mineral, say, clinopyroxene, garnet, chromite, and sulphides garnet, that is common to both types of occur- (e.g., pyrrhotite). Four other minerals -rutile, lzy- rences. This matter may be resolved by determin- ing the chemical composition of that inclusion (see figure 10). For example, the ideal formula for a Figure 9. This ruby is the first definitely iden- pyrope-almandite garnet would be a mixture of tified as an inclusion in diamond. It proves both garnet types; that is, both Mg and Fe can that the diamond is of eclogitic origin. Photo- micrograph by Eduard J Giibelin; darkfield illu- substitute in the silicate structure so that the mination, magnified 100~.From Gubelin and formula for any specific garnet in this series could Koivuln (1986, p. 97). be written as (Mg, Fe),A1,Si,Ol,. However, from figure 10 it can be seen that the relative propor- tions of Fe and Mg and minor Ca in garnet differ, characteristically, for both the P- and E-types of diamond associations; specifically, P-type have higher Mg and lower Fe than do E-type garnets. Other inclusions in diamond (e.g., olivine, orthopyroxene) can be classified by similar analyt- ical methods. Returning briefly again to garnet, experienced individuals can accurately distin- guish between the two types on the basis of color. The pyrope type is typically purple-red whereas almandine is usually orange-red. As discussed earlier, diamondiferous xenoliths of the E-type are relatively abundant, whereas diamondiferous P-types are very rare. Yet P-type

10 Origin of Diamonds GEMS & GEMOLOGY Spring 1991 Figure 10. This plot shows the relorive abun- dances of calcirlm (Ca), magnesium (Mg), and iron (Fe) in garnet inclu- sions from diamonds worldwide. Peridotitic gornets are much higher in Mg, and lower in both Fe and Ca, than garnets from eclogites. Although not shown, peridotitic garnets are higher in than eclogitic garnets; the latter also have minor, but charac- teristic, sodium, that is, >0.09% Na,O. (Adapled from Meyer, 1987.) inclusions are much more common than the in diamonds are valuable in this research. In some E-type. It is clear that once mineral inclusions are cases, the analysis of a single inclusion for just one encapsulated within diamond, they are protected chemical element is sufficient. In other cases, from reactions with the surrounding magma, or particularly in geothermometry, several co-exist- from disaggregation, as occurs with the diamond- ing and touching mineral phases ideally should be iferous peridotite xenoliths. Therefore, it seems present in the diamond (figure 11).For a variety of logical tp qccept the relative abundances of P-type reasons, assumptions usually are made and the and E-type assemblages that occur as inclusions in results sometimes lack the desired precision. diamonds as most representative of their relative (Technically, the temperatures and pressures deter- proportiops and, by analogy, the relative impor- mined are those that existed when the mineral tance of each roclz type as the source material from systems were last in equilibrium, which may not which diamonds originate. Although various sci- entists will propose different proportions, we be- lieve that peridotite-type inclusions outnumber the eclogite type by a ratio of 3:1. P-type diamoilds Figure 1 I. A prlrple chrome pyrope garnet and a colorless pyroxene have nnited to form a tro- arc particularly more abundant in smaller sizes. peze in the host diamond. This bimineralic in- Pressure-Temperature Conditions in Which Dia- clusion lies parallel to an octahedron face. Co- monds Form. A lznowledge of the pressure and existing and touching mineral phases such as these are ideal for geothermoinetry studies. temperature (P-T)regime in which diamonds crys- Photomicrogaph by Henry 0.A. Meyer; bright- tallize is essential for determining the geologic field illumination, magnified 40 x . From origin of diamonds. Geobarometry is the discipline Giibelin and Koivula (1986, p. 95). in geology that employs methods, such as the analysis of pressure-indicative minerals, to deter- mine the pressure under which a mineral or roclz formed. For example, the presence of coesite, lznown to be a high pressure form of SiO, (, at low pressure), yields important information regarding pressure conditions. Geothermometry is the discipline concerned with the temperature of formation of similar materials. Both of these disciplines depend heavily on experimental laboratory procedures, such as the synthesis of specific minerals under carefully controlled pressure-temperature conditions, to simulate natural situations. Again, the inclusions

Origin of Diamonds GEMS & GEMOLOGY : Spring 1991 Figure 22. This diagram shows how increasing sili-

3.8 ~~I~~I~I~IIIII~~I1111-2.1 con (Si) and decreasing GARNET 1200° aluminum (Al) + chro- 3.7 - - 2.0 mium (CrJ are accommo- dated in the structure of garnet (which crystallizes 3.6 - - 1.9 at 1200°C) with increasing pressure (libars). (Units are 3.5 - - 1.8 in atoms per unit cell.) Studies such as this en- Si 3.4 - - Al able the determination of - + press~lre-temperature con- 33 Cr ditions for inclusions in diamonds at the time of 3.2 - formation. For example, if - a garnet contains 3.0 3.1 1.4 atoms per unit cell of Si, - and 1.95 atoms per unit 3.0 - - 1.3 cell of A1 + Cr, it crystal- - \ - lized at a pressrrre of

2.9,1 I 11 I I 11 8 ,I :,I 11 11.1.2 about 45 libars. Data are 0 20 40 60 80 100 120 140 160 180 200 220 from eclogitic garnet in- clusions in diamonds from Pressure ( Kbars) the Monastery mine, Sorrth Africa (Moore and Gurney, 1985). agree precisely with the temperature and pressure approximate depths within the earth are 100, 150, of diamond formation.) and 200 lzm, respectively. Eclogitic-type inclu- Examples of the types of minerals used, and sions fell within the same temperature range, but elements determined, for geobarometry include: it was not possible to estimate their pressure of (a)aluminum substitution in orthopyroxene (en- formation. Considering the rate at which tempera- statite) co-existing with garnet, (b) tures increase with depth () substitution in clinopyroxene, and (c) sodium under continental areas, as well as the correspond- substitution in garnet. In all of these cases, ele- ing increase in pressure, the estimated depth of vated amounts of Al, K, and Na are indicative of formation of P-type diamonds is in the range high pressures. Examples of geothermometry in- 150-200 km, which is within the . clude methods based on: (a)the partitioning (rela- E-type diamonds appear to have higher tempera- tive proportions) of the Ca and Mg contents of co- tures of crystallization and to form at greater existing orthopyroxene (enstatite)and clinopyrox- depths than do P-types. In fact, based on geo- ene (diopside),(b) the relative abundances of Fe and barometry, Moore and Gurney (1985)have shown Mg in these same pyroxenes, and (c)the relation- that E-type diamonds in at least one South African ship between Fe and Mg in coexisting garnet and mine (Monastery)have origins that may be deeper orthopyroxene. In figure 12, we illustrate how the than 300 lzm, but still within the upper mantle. It pressure at the time of formation of an eclogitic must not be assumed, however, that all E-type garnet inclusion within a diamond can be deter- diamonds originate from such depths. mined. There have been many determinations made Sources of Carbon in Diamond. The source of the in the past decade of the pressures and tempera- carbon from which diamonds form has been a tures at which mineral inclusions in diamond subject of interest and controversy for over a crystallized (see, e.g., Ross, 1989).Meyer (1985)has hundred years; suggestions have ranged from coal evaluated these and concluded that in the peridoti- in the 1800s to and methane today tic type of inclusions the temperature of crystalli- (see Janse, 1984, and Meyer, 1985, for the historic zation ranged from 900" to 1300°C and pressure aspects). It is now generally agreed that there are from 45 to 60 kbar. At 30, 50, and 60 lzbar the two sources of carbon, as determined by stable

12 Origin of Diamonds GEMS & GEMOLOGY Spring 1991 carbon studies, specifically the ratio of environment by subduction to the depths (>I50 carbon-13 to carbon-12. For the sake of conve- lzm] necessary for the formation of diamonds. nience, these ratios are reported as "delta values" Eclogites have a bulk chemical compositioil and specifically, in the case of the carbon , that is virtually identical to that of basalt, but the as 81". (Technically, a delta value is the difference minerals comprising (primarily the between the isotope ratio in a sample and that in a clinopyroxene and Ca-rich plagioclase feld- standard, divided by the ratio in the standard, and spars; see box A] are different from those of expressed in parts per thousand.) eclogites (that is, garnet and the clinopyroxene A plot of the 813C of several hundred dia- omphacite) because of the different pressure and monds, from many geographic locations as well as temperature environment in which the latter crys- from both eclogitic and peridotitic origins, is tallize. Eclogites are formed at higher pressures presented in figure 13. Inspection of this figure will show a shaded area in the narrow range of -2 to - 9 813C, with a peak between - 5 and - 6 613C. Figure 13. The distribution of carbon isotope ra- This is the area in which almost all peridotitic tios, 6'3C, in diamonds from mony geographic diamonds will plot. Although the 613C of many localions and from both eclogitic and peridoti- eclogitic diamonds will also plot in this narrow tic rock origins is illustrated here. The shaded range, others will plot elsewhere on the diagram; region, wilh 613C val~resof -2 to -9, is the they are not confined to the narrow range. These ronge for diamonds from peridotitic assem- important data in figure 13 have been interpreted blages; eclogitic values may range anywhere to imply that there are at least two carbon sources from + 3 to - 34 (including the peridotitic for diamonds: The peridotitic source, with few range). The letter n indicates the number of samples. See text for further details. From exceptions, is characterized by the narrow 613C Gurney (1989). range of ,- 2 to - 9, whereas the eclogitic source may hav

Origin of Diamonds GEMS & GEMOLOGY Spring 1991 C

Figure 14. The theory of plate tectonics explains how the earth's solid outer portion (the ) is di- vided into a number of rigid thin segn~ents(plates) which move in various ways including downward at certain boundaries, particularly where continents collide. This figure illustrates an oceanic basaltic plate being moved downward (subduction) into a region of higher pressures and temperatures below a craton (a part of the earth that has been stable for a long time; inclr~descontinental shields and platforms). Carbon present within the plate in the form of limestone (calcite) or other carbonate-bearing rocks, or as hydrocor- bons (including organic remains), could be the source of this element (carbon) for diamonds forn~ed in eclogite. (Vertical scale is exaggerated.)

relative to basalt during and following certain EMPLACEMENT OF DIAMONDS earth movements and building (tec- To understand the emplacement of diamonds near tonic) events; basalts caught up in such tectonic the surface of the earth, we need to consider the processes may recrystallize into high-pressure following topics: eclogite. But, could eclogite within the earth's What rocks did the emplacement mantle, at depths of 150 lcm and greater, also + represent basalt that has been converted to + Where emplacement occurred eclogi te? + When emplacement occurred When continents collide, basalt, which is the + How emplacement occurred main rock type in the oceanic basins, is pushed down beneath the continents by a process caIIed What Rocks Did the Emplacement. We have re- subduction, into regions of higher pressures and ferred repeatedly to the fact that diamonds were temperatures (figure 14)) where it eventually can carried to the surface (i.e., emplaced) by kimberlite be converted to eclogite. Carbon, in the form of and lamproite. To understand this emplacement limestone (calcite)or other carbonate rocks, or in mechanism, it is important to understand the the form of hydrocarbons (e.g.,organic matter such similarities, differences, and relationships be- as bacteria, algae), may have been included in the tween lzimberlite and lamproite. subducted slab and thus could be the source The most accurate definition available of the material for eventual conversion to diamond. rock called lzimberlite is presented in box B. For the

14 Origin of Diamonds GEMS W GEMOLOGY Spring 1991 BOX B: WHAT IS KIMBERLITE?

One would think that the meaning of the term mafic derived from deep in the earth kimberlite would be well understood and that it 1> 150 km below the surface) which occurs near the would have a universally accepted definition. How- surface as small volcanic piies, dikes, and sills. It is ever, although kimberlite was first introduced over a composed principally of olivine (both as phe- hundred years ago, based on descriptions of the nocrysts and in groundmass), with lesser amounts of diamond-bearing pipes of Kimberley, South Africa, , diopside, serpentine, calcite, garnet, il- there is still no unanimity on its definition (see, e.g., menite, spinel, and/or other minerals; diamond is Glover and Harris, 1984; Clement et al., 1984; and only a rare constituent. Mitchell, 19891. Use of the adjectives "volatile-rich, potassic, ultra- The problem stems, in part, from the fact that a mafic" to describe kimberlite indicates that there is satisfactory definition of kimberlite must take into an important and characteristic chemical signature account mineralogic composition, chemical compo- for this roclz. Volatile-rich (readily vaporizable, gas- sition, texture, and origin. This is a difficult task, eous) refers to the high contents of CO, (8.6% indeed. average, mostly in calcite) and H,O (7.2% average, in Most kimberlite professionals now accept the serpentine and phlogopite), in lzimberlites. The aver- definition of Clement et al. (1984, pp. 223-2241, but it age potassium content (K,O = 0.6%-2.0%) is high for is very complex. For a gemological audience, the an ultramafic roclz, whereas the average SiOz content definition of Dawson ( 1984, pp. 104-105) is instruc- (25%-35%) is extremely low for an igneous rock. In tive and, in its imaginative presentation, illustrates the case of kimberlite, Fe203 averages 12.7% and the complexity of defining kimberlite satisfactorily: MgO, 23.8%. (All of the preceding values are from "In short, KIMBERLITE IS A HYBRID ROCK, Mitchell, 1989, p. 35 and analysis 10, table 1, p. 36.) comprising: Although not mehtioned, or even implied, in this definition, there are also unusually high concentra- ~(AGMENTS OF HIGH-T [temperature] tions, for ultranlafic rocks, of certain nonessential PERIDOTITE AND ECLOGITE elements found in small quantities (i.e., "trace ele- plus ments"]. Examples are niobium (Nb],zirconium (Zr), MEGACRYSTS strontium (Sr], barium (Ba), rubidium (Rb), and which have reacted with cerium (Ce).These all occur in amounts significantly RELATIVELY LOW-T, VOLATILE-RICH less than 0.1 %; nevertheless, they are geochemically and, during and after intrusion, with significant and are also important indicators of the HIGH-LEVEL GROUNDWATER presence of kimberlite for exploration purposes. and in many diatreme- kin~berlites Exactly how lzimberlites form is a subiect of there is variable input of intense st;dy. Although several hypothese; have WALL ROCK MATERIAL 1e.g. basalt, gneiss, ]" been presented in the past, Eggler (1989, p. 496) [The following terms mentioned above and not previ- reports that " of carbonated peridotite ously explained are briefly defined. Megacryst: coarse is preferred." That is, when a peridotite containing a single crystal; nongeneric term for a phenocryst or small amount of a such as dol- xenocryst. Matrix: groundmass; finer grained than omite (source of CO,) as well as phlogopite (source of megacrysts. Diatreme: a general term for a volcanic potassium) becomes hotter, a small portion (less than pipe that is emplaced in rocks by a gaseous explosion 10%1 of the roclz will melt initially. The fluid and is filled with angular brolzen fragments called (magma]resulting from this partial mel;ing will have .] the chemical characteristics of kimberlite mentioned Dawson uses the term hybrid in this explanation above (volatile-rich, potassic, ultramafic). This pro- because kimberlite contains a mixture of (foreign] cess probably talzes place at pressures of 50-65 lzbar xenoliths (peridotite, eclogite, and other roclz types] and temperatures of about 1200"-1500°C. Thus, lzim- and xenocrysts (diamonds and others) in addition to berlite magmas, like diamonds, originate from peri- the normal crystallization products from the lzim- dotite in the upper mantle at depths of perhaps berlite magma. 150-200 km. It is important to remember, however, We offer the following simpler, though less pre- that the consensus is that diamonds will only crystal- cise, definition which augments that of Dawson lize from peridotite (in addition to eclogite] and not (19841: from kimberlite (except possibly in special cases) for reasons that are not known, and that diamonds will Kimberlite is a hybrid, volatile-rich, potassic, ultra- occur in lzimberlite only as xenocrysts.

Origin of Diamonds GEMS & GEMOLOGY Spring 1991 15 purposes of this discussion, we need only consider magnesium (Mg),iron (Fe),and calcium (Ca),but the highlights presented in box B and recognize the higher in silicon (Si) and aluminum (Al), than following characteristics of lziinberlite. lzimberlites. Kimberlite is a dark-colored (referred to as "blue Lamproites, like liimberlites, occur as pipes, ground" when fresh) hybrid rock; that is, it is a dikes, and sills, but lamproite pipes resemble mixture of the crystallization products of the champagne glasses, rather than cones, in shape. kimberlite magma itself (e.g., olivine, phlo- A comparison of the minerals present in the two gopite) plus xenocrysts and xenoliths of peri- rock types is particularly informative (table 2).The dotite and eclogite derived from the upper fact that several major, as well as some minor, mantle. minerals are common to both rocli types suggests Chemically, lzimberlite is an ultramafic, po- that their respective magmas have similar chemi- tassic, volatile-rich (CO,,H,O) roclz that formed cal characteristics. Further, the xenocrystic min- deep within the earth (at least at the depth of erals are identical, which suggests that liim- diamond formation) at high pressures and tem- berlites and lainproites were, at some time in their peratures. histories, in the same high pressure-high tempera- Kimberlite is intruded from the mantle into the ture environments characterized by eclogite and earth's crust; near the surface, it talies the form peridotite (in the Argyle olivine lamproite of of a cone-shaped pipe characterized by a volca- Australia, there are far more eclogitic than peri- nic explosion and the formation of breccia dotitic xenocrysts in diamonds). From table 2 we (angular broken fragments) within the pipe (see see that lamproite has a number of additional below). minerals that distinguish it from liimberlite (al- though not all are present in all lamproites).These Lamproite was a relatively obscure rocli type until 1979, when it was found to host primary deposits of diamonds in Australia. To date, it is only the second rock type to have gained this distinction. Although it has been the subject of many studies TABLE 2. Minerals found in kimberlite and lamproite. over the last decade, like lzimberlite it is not easily Minerals Kimberlite Larnproite defined. Currently, lamproite refers to a group of Minerals that crystallize directly from kimberlite and roclis closely related in chemical composition (a lamproite magmas clan) rather than to a specific rock variety. Nev- Major ertheless, certain chemical, textural, and miner- Olivine r/ r/ alogic characteristics are recognized: Diopside r/ r/ Phlogopite r/ r/ Lamproite has a characteristic gray to greenish Calcite r/ gray mottled appearance and, like liimberlite, is Serpentine r/ a hybrid rocli. The primary magmatic crystal- Monticellile r/ lization products, most notably olivine, occur Leucite r/ Arnphi bole r/ both as and as groundmass constit- Enstatite r/ uents. Upper-mantle xenoliths and xenocrysts r/ are the same as those found in lzimberlites. Minor Chemically, lamproite is an ultrapotassic (po- r/ tassium values are typically 6%-8% K,O, com- r/ llmenite r/ pared to 0.6%-2.0% K,O for liimberlites), mag- Spinel r/ nesium-rich (mafic) igneous rocli. Significant Priderite r/ trace elements include zirconium (Zr),niobium r/ (Nb),strontium (Sr),barium (Ba),and rubidium Wadeite r/ (Rb);these same elements also are enriched in Xenocryst minerals derived from the upper mantle kimberlites (see box B).On the other hand, CO,, Olivine r/ r/ which is enriched in liimberlite (average 8.6%), Garnet r/ r/ generally is low (

16 Origin of Diamonds GEMS & GEMOLOGY Spring 1991 include, for example, the minerals leucite (KA1- Kimberlites tend to occur in clusters, with 6 to Si,06), sanidine (KAlSi,O,), wadeite (K,ZrSi,O,) 40 (excludingdilzes) in any one. Certain areas, such and priderite [(K,Ba)(Ti,Fe),016],from which the as South Africa, , and northwestern Tan- potassium- and zirconium-enriched character of zania, contain many clusters. The five main pipes lamproites is derived. at Kimberley, South Africa, cover a circular area In summary, based on their chemical, textural, with a diameter of 10 lzm, but the entire cluster, mineralogic (including those of the xenoliths and incl~ldingdilzes, occupies an area with a diameter xenocrysts), and emplacement (e.g., as pipes) char- of 40 lzm. Janse (1984) suggests that the distance acteristics, lzimberlites and lamproites have much between several major clusters in both South in common. Certainly, the similarities are greater Africa and Siberia is generally about 400 lzm, than the differences. Both rock types probably although the use of such estimates has been formed by partial melting (seeboxB) of similar, yet questioned by Gurney (1989). The ratio of eco- distinctive, peridotitic material at greater depths nomic to noneconomic pipes in clusters varies than any other lznown volcanic roclzs. Pressures of considerably: Janse (1984) gives figures ranging origin can be related to the upper mantle at dcpths from 5 out of 15 at Icimberley and 3 out of 29 at of at least 150 lzm, and temperatures probably were Orapa (Botswana),to 1 out of 30 at Alakit (Siberia); in the range of 1100"-1500°C; lzimberlites appear some clusters, of course, have no economic value. to have crystallized at the middle to upper part of Unfortunately, there are no similar statistics the temperature range and lamproites at the lower for lamproites. Not only is the discovery of the end. Both rock types have transported mantle Argyle deposit in relatively xenoliths and xenocrysts, including diamonds, to recent, but many of the roclzs previously called the surface, although not all lzimberlites and lamproites, as w~llas related types, are presently lamproites contain diamonds. Finally, it should be being reevaluated and reclassified. Such studies, noted tht only recently has lamproite been recog- for example, have resulted in the reclassification of nized as 4 host for diamond deposits and that there the diamond-bearing Prairie Creek (Murfreesboro), is no lznown fundamental reason why other man- Arkansas, pipe from a lzimberlite to an olivine tle-derivid roclzs with sufficiently deep origins lamproite. could not also transport diamonds from the same However, studies of kimberlites, and to a lesser peridotitic and eclogitic sources. extent lamproites, have provided valuable infor- mation about the geologic distribution of the Where Emplacement Occurred. There are two diamond-bearing roclcs in the earth's crust. Figure aspects to the topic of where emplacement of 15 shows that the geographic distribution of pri- kiinberlite and lamproite occurred: geographic mary diamond deposits, predominantly kim- and geologic. Both of these topics were reviewed berlites, is not random but is confined to regions of thoroughly by Clifford (1970), Janse (1984)) and the that are old cratons (defined Dawson (1989).Most emphasis will be placed on below). This is an observed fact that was well lzimberlites because there is far more information formalized by Clifford in several publications, available on these rocks than on lamproites. culminating in Clifford (1970).It illustrates, too, Kimberlites are widespread around the earth, the fact that lzimberlites are never found in oceanic and over 3,000 are lznown in southern Africa alone. environments or young mouiltain belts. A craton This figure includes both dilzes and pipes (sills are (again, see figure 14)is part of the earth's crust that rare), of which the former are more abundant. has attained stability and has been little deformed There are about 100 kimberlites in North Amer- for a very prolonged period of time (generally more ica, with perhaps 60 located in what is called the than 1,500 My). Effectively, the term applies to Colorado-Wyoming State Line District. Of the extensive, stable contineiltal areas and consists of total number of lzimberlites worldwide, fewer than two parts: (a)a shield (the exposed core of a craton, 1,000 contain any diamonds, only 50-60 have ever e.g., the ), and (b)a platform (the been economic, and only about 12 major pipes are part of the craton, covered by generally flat-lying being mined today. Most of the well-lznown dia- sediments and sometimes associated volcanics, mond-producing pipes have a surface area of be- e.g., basalts, that is adjacent to, and an extension of, tween 5 and 30 hectares (about 12-75 acres); in the shield). Cratons are the nuclei of all continents, South Africa, they typically contain about one carat and all present-day continents, except Europe, of rough diamond per five tons of lzimberlite . have more than one craton (again, see figure 15))

Origin of Diamonds GEMS & GEMOLOGY Spring 1991 17 1 - , . -3 F q F. . . 1' .,- I- - . J C . i $. . -'.. 7 ' w 'I L C 7,.,, : -I- '- p, I,,! f i* i,, - .T -- : 1 .r . bF- ' u..:a -+-d 4- I +I- .I '! .3 .'1 p. . . y .f F, - -., i .- PI - &&.r,-.. I . . ' .. , , 8 Y .' - .- -++ 1, 'IC -

Figure 15. This world map of primary diamond deposits (i.e., those in pipes) shows cratonic areas (dashed lines) with major economic diamond deposits (large solid diamonds), minor economic de- posits (small solid diamonds), and subeconomic deposits (small open diamonds). The economic de- posits are in kimherlites except for the Australia11 depositfs), which are in lamproite. Note that sev- eral cratons have no known diamoiid deposits. From Gzrrney (1989) and based on data in Ianse (1984). which are usually of different ages. This suggests lzimberlites is "on-craton" as opposed to "off- that the continents of today may be composites of craton," that is, "on" the part of the craton the remnants of ancient continents, each of which (including the sedimentary platform roclzs above) had its own craton. Kimberlites have been found as opposed to "off" it. The latter location includes within most cratons on all continents. Some any younger part of the craton and adjoining cratons have far more diamond production and mobile belts (i.e., linear regions adjacent to cratons potential than others. For example, the Kaapvaal that were subjected to folding due to cratonic (Kalahari) craton of southern Africa has seven of collisions and later became mountain belts, such the world's 11 established diamond-producing as the present Alps). Mobile belts may eventually clusters. become "fused" to cratons. However, generaliza- Studies of the locations of lzimberlites within tions such as this can be dangerous. For example, cratons (thisdetail is not shown in figure 15)reveal the Argyle lamproite, the largest producer of dia- that the occurrence of lzimberlite pipes is most monds in the world today, is in a mobile belt that common within the younger, generally flat-lying became part of a craton 1,800 My ago. The signifi- sedimentary platform roclzs that rest on the Ar- cance of cratons from the point of view of the chean (>2,500 My old) part of a craton. Very few "storage" of diamonds between the time of their important lzimberlite pipes are lznown in the formation and their being brought to the surface exposed cores (shield areas) of cratons, the princi- will be discussed later. pal examples being the West African and Tanza- nian cratons, probably because they have been When Emplacement Occurred. Diamond-bearing extensively eroded. This has resulted in the gener- lzimberlites and probably lamproites have in- alization among kimberlite specialists that the truded into the earth's crust for a very long period, most favorable location for diamond-containing as evidenced by the occurrence of diamonds in the

18 Origin of Diamonds GEMS & GEMOLOGY Spring 1991 2,600-My-old conglomerate in (b) most lzimberlites and lamproites were em- South Africa. The presence of diamonds in this placed in the last 200 My, although there were paleoplacer (alluvial) deposit requires that a still major intrusions at least as early as 1,600 My ago older kimberlite (or lamproite) existed. and possibly prior to 2,600 My ago. The oldest lznown lzimberlites still preserved are the 1,600-My-old intrusions near Kuruman, How Emplacement Occurred. It is generally ac- northern Cape Province, South Africa, which are cepted that lzimberlite and lamproite magmas situated on the but contain no result from the partial melting of similar, yet diamonds. Subsequently, there was a period of distinctive, peridotitic material 150-350 lzm be- extensive lzimberlite emplacement about 1,200 low the earth's surface, that they intruded into My ago, again in South Africa (e.g., Premier mine; cratons, and that the process of emplacement has see table 1) but also in India and Mali. Other been operative, during some periods more than important episodes of lzimberlite intrusion are others, at least since 2,600 My ago. Beyond this presented in table 3. point matters become more speculative, partic- Although information on lamproites is less ularly on the details of how emplacement occur- readily available, they are known to cover a range red. Topics that must be considered are: (1)the rate from the Argyle pipe, which was intruded -1,200 of ascent of diamond-bearing lzimberlites and My ago, to the Ellendale lamproites (about 50 lamproites; and (2) the configuration and forma- bodies are known and several are potentially tion of diatremes. Again, most emphasis will be economic), approximately 400 km from the Argyle placed on lzimberlite. deposit, which were intruded in early Miocene time (-20 My ago) into platform sediments of Rate of Ascent. The rate of ascent of lzimberlites is Devonian and Permian age (-400-250 My). Some most frequently determined on the basis of the lamproites in Wyoming, , and a few following observations: (a)diamonds are preserved other locblities may have been emplaced within during their ascent to the surface rather than the last one inillion years. reverting to graphite, being converted to carbon We may deduce from the large number of new dioxide (CO,), or dissolving in the kimberlite age dates obtained in the past decade that: (a) magma (figures 16 and 17); and (b)the diamond- kimberlite and/or lamproite intrusions can occur bearing kimberlites also transport large xenoliths at several different times in the same vicinity; and from as deep as 200 Izm below the surface (again, see figure 4). Both of these observations require that the ascent to the surface be reasonably rapid. During slow ascent, or ascent with many inter- TABLE 3. Times of intrusion of selected kimberlite mediate stops, diamond would likely revert to provinces (modified from Dawson, 1989). graphite (which in the pressure-temperature con- - - - Geologic age Time (My ago) Locality ditions in the earth's crust is thermodynamically more stable than diamond)and the heavy xenoliths Eocene Namibia, Tanzania would tend to settle back through the magma. At Upper Cretaceous Southern Cape (South Beni Bouchera, Morocco, for example, we lznow Africa) Middle Cretaceous Kimberley (South that diamonds were transformed to graphite be- Africa), Lesotho, cause they were not transported rapidly to the Botswana, Brazil surface by lzimberlite or lamproite (Slodlzevich, Lower Cretaceous Angola, West Africa, 1983). The fall in temperature and pressure was Siberia sufficiently slow to permit the conversion; only Upper Eastern North America, Siberia with a rapid decrease in temperature and pressure Devonian Colorado-Wyoming, will the carbon atoms "freeze" in the metastable Siberia diamond structure (figure 18). Ordovician Siberia Although we lznow that the ascent is rapid, Upper Northwest Australia estimates of the exact velocity depend on various Middle Proterozoic 1 Premier (South Africa), India, Mali assumptions. For our purposes, the ascent rates Lower Proterozoic Kuruman (South Africa) proposed by Eggler (1989)of 10-30 Izm per hour are realistic. In other words, diamonds are brought to

Origin of Diamonds GEMS & GEMOLOGY Spring 1991 19 Configzzration and Formation of Diatremes. The term diatreme is synonymous in this article with the terms breccia pipe and simply pipe: It is a general term for a that is emplaced in roclzs by a gaseous explosion and is filled with angular brolzen fragments called breccia [Glz. dia: through + trem(a): perforation = to pierce, drive through]. As a general term, diatreme can be used for pipes of many types in numerous geologic situations (e.g., diatremes of basalt erupting along fractures); however, the combination of features described below is possibly uniq~leto lcimberlite diatremes (and related rocks, e.g., lamproites) pri- marily because of the great depths from which they originate as well as the amounts and types of

Figure 16. Exlre~nelywell-shaped diamonds with sharp oclahedral edges, as seen illst below Figure 17. T11e inegzllar rounded shapes fre- the table facet of this host mineral, are un- quently found on diamonds result when the usual. Special conditions were required for their transporling kimberlite (or lamproite) magmas preservation; for example, they may have been react with, and dissolve, the diamond, starting protected within another mineral (diamond, in wit11 the octahedral edges. Isolation of dia- this case) or in an eclogite xenolith. Otllerwise, monds from transporting magmas (see figure it is likely that they would have been resorbed 16), or very rapid ascent to the surface, can (dissolved) in a kimberlite or lamproite magma. minimize this effect. Plloto 0 GIA and Pl~otomicrograpl~by Eduard 1. Giibelin; dark- Tino Hammid. field illzrmit~atioi~,magnified 20 x . From Giibelin and lZoivula (1 986, p. 97). the surface from their storage areas at depths of at least 110 km (at the base of cratons) in 4-15 hours! Further, as the surface is approached, within the last 2-3 lzm, the velocity increases dramatically to perhaps several hundred lzilometers per hour, for reasons that are explained later. An obvious requirement for kimberlites to reach the surface is the availability of fractures that extend from below the base of the craton, through solid roclz, for a distance in the vicinity of 150 lzm; these deep fractures are only possible in geologically stable areas. Exactly how these deep fractures are generated, and are even repeated from time to time in identical localities to account for lzimberlite intrusions of widely different ages within the same pipe, is a matter that is not well understood. For our purposes it is only necessary to recognize the existence of a problem concerning the origin of the deep fractures and to observe that several hypotheses have been proposed, for exam- ple, craclz propagation by magmatic fracturing (Eggler, 1989) and crustal thinning linked with major plate movements (Dawson, 1989). Mitchell (1986)has reviewed older theories.

20 Origin of Diamonds GEMS h GEMOLOGY Spring 1991 Figure IS. The diamond-graphite equilibrium line is plotted here against the geothermal gradients for the continental shield (craton) and oceanic areas. (A geothermal gradient is the rate of increase of temperature with increasing depth in the earth and averages about 25°C per kilometer in the earth's crust.) {Jnder the oceans, temperature rises much more rapidly with depth than it does under the shield areas. The geothermal gradient line for the shield areas intersects the diamond- graphite equilibrium line at 53 ltbars (and about 1325"C), which corresponds to a depth of about 160 km. Under oceanic areas, it intersects at depths greater than 200 Temperature (OC) ltm, which is unsrlitable for the prodrlction of diamonds. Modified from the GIA Diamonds course.

gases they contain. The characteristic features of shows the three different zones within an idealized kimberlite diatremes are: (a] their general shape, lzimberlite. and (b) iheir three distinct depth zones (root, The root zone is the deepest part of the pipe. It diatreme, crater) which, in combination, are the is characterized by an irregular outline and by configurfiion of the pipe (figure 19). numerous distinct intrusive phases of lzimberlite The classic, early 20th-century studies of and igneous features, and it extends about 0.5 lzm lzimberlite pipes in the Kimberley area, South vertically about 2-3 km below the surface. It is Africa, demonstrated that they occur as carrot- composed of crystallized lzimberlite magma with shaped, vertical intrusioils that pinch out at depth. the megascopic appearance of a typical intrusive These observations have stood the test of time. igneous roclz, along with xenoliths and xenocrysts With increasing depth of the mines over the years, and frequently with some small breccia fragments. new observations were made and older ones re- With depth, the root zone grades into individual fined. These include: (a) the fragmental (brecci- feeder dilzes, also of lzimberlite (but without cer- ated) nature of lzimberlite, particularly in the tain characteristics such as breccia), which extend diatreme zone; (b) the fact that lzimberlite is a downward indefinitely, but probably not continu- "cold" roclz, inasmuch as there are very few indica- ously, as the fractures along which the magmas tions of thermal effects (e.g., coiltact metamor- moved probably opened and then closed after phism) on either the wall roclzs or xenoliths of the magma passed through. Root zones and feeder diatreme zone as would be expected for a rising dilzes may contain diamonds among the xeno- molten magma; and (c)the observation that the crysts, but they have been mined only on a small pipes become narrower with depth, eventually scale because of their limited volume. Economic thinning into feeder dikes that are rarely thicker mining is generally limited by the width of the than one meter. In other parts of Africa, important dilzes, which are usually only about 60 cm wide surface features of lzimberlite diatremes, e.g., (although they may, rarely, be 10 m wide, known as and rings (see below), were also recognized. a "blow"). Hawthorne (1975) combined all the above The diatreme zone (not to be confused with features and facts relating to lzimberlite pipes and diatreme = pipe) is much greater in vertical extent developed an idealized model, which is illustrated than the root zone and is the most important in modified form in figure 19. This figure, which is source of diamonds because of its volume. It ranges a basis for modern concepts of emplacement, from 1 to 2 lzm in height and extends to within 300

Origin of Diamonds GEMS & GEMOLOGY Spring 1991 21 Tuff ring DEPTH

EROSION LEVEL Figure 19. This idealized model of o -0rapa ki~nberlitepipe includes the root, diatreme, and crater zones. Also shown are the present erosion levels at the Orapa pipe in Botswana and the jogersfontein, Kirnberley, and Bellsbanlz pipes in South Africo. Adapted jrom Howthorne (1975). -Jagersfontein

-Kirnberley

- Bellsbank

m of the surface, in the idealized situation in the pipe takes on its conical shape and becomes which it has not been eroded (again, see figure 19). wider. At the same time, those areas of lzimberlite The diatreme zone contains xenoliths and xeno- that had already crystallized as roclz undergo crysts from the mantle, as well as roclz fragments fragmentation (brecciation).Because of the expan- derived from the crustal roclzs through which the sion of the gases, the lzimberlite magma cools lzimberlite passed, including wall rock (e.g., basalt, down rapidly so that there are few thermal reac- gneiss, shale) in the vicinity of the pipe. The main tions with the wall roclzs or crustal xenoliths. With characteristics of this zone, however, are the lzim- the temperature sufficiently low in relation to the berlite and other fragmental roclzs (e.g., lower pressure, the diamonds resist conversion to tuffs) associated with explosive magmas. graphite and survive intact (figure 18). These features develop because within the The crater zone occupies the upper 300 m or so lzimberlite magma there are large amounts of of a typical lzimberlite diatreme which, at its dissolved gases, specifically carbon dioxide and formative stage, is a . Whereas most volca- (see box B], under great pressure. At about noes erupt molten , as typified by Mauna Loa 2-3 lzm below the surface, explosions occur in the in Hawaii, such probably is not the case with a ascending kimberlite magma as the gases expand lzimberlite volcano. This is because, by the time enormously at the lower near-surface pressures. the lzimberlite has passed through the diatreme Because of these explosions, the rate of ascent of zone, it is no longer molten and does not flow out. the kimberlite accelerates rapidly, to perhaps sev- Rather, it erupts as brolzen solid fragments of rock eral hundred kilometers per hour. As the lzim- called pyroclastics [Glz. pyro: fire + clastos: bro- berlite breaks through the overhead crustal rocks, ken into pieces], tuffs (a general term for all

22 Origin of Diamonds GEMS & GEMOLOGY Spring 1991 consolidated pyroclastic roclzs), or lapilli (loose It is premature to make more than passing pyroclastic material in thc size range of 2-64 mm), comments on the emplacement aspects of diamon- anlong other terms. Another factor to be consid- diferous lamproites, because only the Argyle mine ered is that the rising lzimberlite magma and/or is in production and only since late 1985. There are pyroclastic material, while not molten, is still hot several scientific reports describing this mine, as and will eventually encounter cooler groundwater well as the Ellendale lamproite pipes about 400 lzm at depths that vary with locality. As the groundwa- distant, in the volumes edited by Ross (1989). ter turns to steam, the eruption becomes even Suffice it to say that there are many similarities more volatile and, in some special situations, very between these pipes and those of lzimberlite. As violent explosions occur. Returning briefly to Daw- mentioned earlier, one interesting difference is in son's (1984) explanation of the term kimberlite shape. Unlike kimberlites, which are carrot-like, (box B), we can now understand the reason for his the root and lower diatreme zones of lamproites iilcludi~lgthe concept of "high-level groundwater." are thin and stem-like, but toward the top the pipes Kimberlite volcanoes have never been ob- flare out in a curved manner that gives a "cham- served to explode in historic times. In Tanzania, pagne glass" shape in cross-section. Mali, and Botswana, however, there are rare exam- ples of the surface expression of this phenoincnon Diamond Sampling in the Mantle. Discussion of that have not yet been eroded away. These include the sampling of diamonds in the mantle and their znaars (low-relief volcanic craters formcd by shal- transportation to the surface requires the integra- low explosive eruptions, which may be filled with tion of many of the topics considered earlier. This water if they intersect the water table, and which is accomplished by reference to figure 20, a sim- are surrounded by crater rings) and tzzffrings (wide, plified version of a diagram first published by low-rimmed accumulations of pyroclastic debris Haggerty (1986).This figure is based on an ideal- of tuff &; lapilli, slightly larger in size than an ized cross-section of the earth through a craton and associated inaar).~t is likely that the explosion lifts its subcontinental regions (thelithosphere), which debris no more than several hundred meters into are characterized by very long-term tectonic sta- the air and that the tuff ring is typically about 50 m bility, such as no mountain building or plate high and is quickly eroded. movements. (Kimberlite magmas, which originate Inspection of depths below the surface in at depths below the base of cratons, do penetrate figure 19 will show that the typical distance from the cratons through narrow fractures, but this the top of the crater zone to the base of the minor volcanic activity does not negate the con- diatreme zone is about 2,300 m. Depending on the cept of a tectonically stable craton.) In all such geomorphologic, topographic, and other charac- areas, because of the tectonic stability, there are teristics of the emplacement site, and assuming an low geothermal gradients (rates of increase in erosion rate of 1 m every 30,000 years (which is the temperature with depth) by comparison with the average for the earth), it could be predicted that the oceanic parts of the world. vast majority (except the root zone) of a typical Figure 20 also reflects, as noted earlier, the fact lzimberlite diatreme would be completely eroded that eclogitic diamonds appear to form at greater away in 69 My. During this period, it would be depths than do peridotitic diamonds. As indicated co~ltinuallyreleasing diamonds into secondary by arrows, the eclogites formed from oceanic deposits, such as alluvials or beach sands. In figure basalts that traveled down from the ocean basins 19, we see that four southern African mines- by means of plate tectonic movements. The dia- Orapa, Jagersfontein, Kimberley, and Bellsbank- gram also shows the relative position of the mobile all of Cretaceous age (-100 My; see table l),are belts 011 either side of the craton. eroded to distinctly different levels. From this it Of great significance is the shaded zone at follows that the Bellsbank mine has limited re- depths of 110-150 lzm and temperatures of maining economic potential because the entire 900"-1200°C. This zone, like the rest of the craton diamond-containing diatreme and crater zones and its subcontinent, has not been actively in- have been eroded away, whereas the Orapa mine volved in plate movements or other major tectonic can loolz forward to potential economic produc- activities for at least 1,500 My. Of possibly even tion to a depth of about 2,000 m which, at the greater significance, however, is the fact that present rate of production, will take several hun- diamonds are stable within this area. Thus, dia- dred years. monds that formed in the deeper parts of this keel-

Origin of Diamonds GEMS & GEMOLOGY Spring 1991 23 Figure 20. This model for the genesis of diamond is sin~plifiedlrom Haggerty (1986). The slable craton and subcratonic areas today are as much as 200 I

shaped area, as well as those that formed elsewhere E-type diamonds. To account for this, Haggerty and somehow were moved into the same area, have (1986) suggests a "complex plumbing system" (a been preserved and stored for as long as 3,200 My system of interconnecting fractures) that is able to (see table 1 and accompanying discussion). One sample appropriate preservation and storage areas. reason for the suggestion that diamonds in the storage zone may be of multiple origins is that in diamondiferous lzimberlites (and lamproites) both P-type and E-type diamonds, in any proportions, CONCLUSIONS may be found. Age-dating techniques applied to the mineral in- A lzimberlite pipe such as K1 (figure 201, clusions (e.g., garnet) within diamond show that ideally situated over the keel of the craton, would the inclusions are very old, ranging from 990 to be likely to contain diamonds, primarily of the 3,300 My, but this range may be extended as more P-type, provided other factors, such as a rapid age determinations are made. Inasmuch as the ascent rate, are favorable. Pipe K2 would lilzely inclusions formed at the same time as their dia- sample an eclogitic enclave hosting E-type dia- mond hosts, it follows that diamonds are equally as monds, whereas lzimberlite diatreme K3, which is old. On the other hand, lzimberlites and "off craton," would probably be barren. The lamproites, the two rock types in which primary lamproite diatreme L1, as exemplified by the diamonds are found, are generally much younger Argyle and Ellendale pipes in Australia, is also "off (the most important diamond pipes range in age craton" in that it is intruded into the mobile belt; from about 100 to 1,200 My) and are not the roclzs yet it is diamondiferous with both P-type and in which diamonds crystallize. Hence, lzimberlite

24 Origin of Diamonds GEMS & GEMOLOGY Spring 1991 and lamproite are merely the transporting media homogeneous source within the earth's mantle, that bring diamonds to the surface by mechanisms whereas the carbon for eclogitic diamonds proba- that are not completely understood. Diamonds are bly originates from sources on the earth's crust formed at depths of 150-200 lzm below the surface; that have been subducted below cratons by plate in the interval between their formation and their movements. transport to the surface, they are stored under Knowledge of the age, origin, and emplace- cratons at least 110 lzm below the surface, where ment of diamonds is extremely valuable in the the high pressure and relatively cool temperatures search for primary deposits. For example, explora- preserved them. tion geologists now know that those areas with the Study of both diamond-bearing xenoliths and greatest potential are ancient cratons that contain mineral inclusions within diamonds shows that ultramafic rocks that originated at great depths. diamonds form within two different rock types, For the gemologist, such lznowledge results in a peridotite and eclogite. Carbon isotope studies on better appreciation of the complex geologic pro- diamonds show that the carbon from which peri- cesses that are required to bring the world's most dotitic diamonds crystallize originates from a important gem to the surface.

REFERENCES Clement C.R., Slcinner E.M.W, Scott Smith B.H. (1984) Kim- A Basis for Concept~~alModels in Exploration, Depart- berlite redefined. Iournal of Geology. Vol. 92, pp. 223-228. ment of Geology and University Extension, University of CliffordTN. (1970)ThestructuraIframeworkof Africa. InT N. Western ~ustrafia,Publication No. 8, pp. 19-61. Cliffoyd,and I. G. Gass, Eds., African ~Magr~~otismand Kranlers J.D. (1979)Lead, uranium, strontium, potassium and Tectonics, Oliver and Boyd, Edinburgh, pp. 1-26. rubidium in inclusion-bearing diamonds and mantle- Dawson J.B.'(1980! Kimberlites and Their Xenoliths. Springer- derived xenoliths from southern Africa. Earth and Plane- Verlag, Berlin. tary Science Letters, Vol. 42, pp. 58-70. Dawson J.B..(1984) Petrogenesis of kimberlite. In J. E. Glover Kornprobst J., Ed. (1984)Kimberlites I: Kilnberlites and Related and I? G. Harris, Eds., Kimberlite Occurrence and Origin: Roclts. Proceedings of the Third International Kimberlite A Basis /or Conceptual Models in Exploration, Geology Conference. Elsevier, Amsterdam. Department and University Extension, University of Meyer H.O.A. (1985) Genesis of diamond: A mantle saga. Western Australia, Publication No. 8, pp. 103-1 11. American Mineralogist, Vol. 70, pp. 344355. Dawson J.B. (1989) Geographic and time distribution of kim- Meyer H.O.A. (1987)Inclusions in diamond. In l? H. Nixon, Ed., berlites and lamproites: Relationships to tectonic pro- Mantle Xenoliths, John Wiley, New York, pp. 501-522. cesses. In J. Ross, Ed., Kinlberlites and Related Rocks, Vol. Mitchell R.H. (1986) Kimberlites: Mineralogy. I, Proceedings of the Fourth International Kimberlite and Plenum Press, New York. Conference, Perth, 1986, Geological Society of Australia Mitchell R.H. (1989) Aspects of the petrology of lcimberlites Special Publication No. 14, Blackwell Scientific Publica- and lamproites: Some definitions and distinctions. In J. tions, Oxford, pp. 323-342. Ross, Ed., Kilnberlites and Related Rocks, Vol. I, Proceed- Eggler D.H. (1989)Kimberlites: How do they form? In J. Ross, ings of the Fourth International Kimberlite Conference, Ed., Kimberlites and Related Roclts, Vol. 1, Proceedings of Perth, 1986, Geological Society of Australia Special Publi- the Fourth International Kimberlite Conference, Perth, cation No. 14, Blaclcwell Scientific Publications, Oxford, 1986, Geological Society of Australia Special Publication pp. 7-45. No. 14, Blacltwell Scientific Publications, Oxford, pp. Moore R.O., Gurney J.J. 1985)Pyroxene solid solution in garnets 489-504. included in diamond. Nature, Vol. 318, pp. 553-555. Glover J.E.,Harris PG., Ecls. (1984)Icilnberlite Occ~rrrenceand Nixon PH. (Ed.)(1987) Mantle Xenoliths. John Wiley, New Yorlt. Origin: A Basis for Conceptunl Models in Exploration. Richardson S.H. (1986)Latter-day origin of diamonds of eclogi- Geology Department and University Extension, Univer- tic paragenesis. Nature, Vol. 322, pp. 623-626. sity of Western Australia, Publication No. 8. Richardson S.H., Gurney J.J., Erlank A.J., Harris J.W. (1984) Giibelin E.J., Koivula J.I. (1986) Photocltlas o/ Inclusio~isill Origin of diamonds in old enriched mantle. Nature, Vol. . ABC Edition, Zurich. 310; pp. 198-202. Gurney J.J. (1989) Diamonds. In J. Ross, Ed., Kirnberlites and Richardson- S.H..- Erlank A.1.., , Harris ,1.W. . Hart S.R. 119901, , Related Roclts, Vol. 2, Proceedings of the Fourth Interna- Eclogitic diamonds of Proterozoic age from Cretaceous tional Kimberlite Conference, Perth, 1986, Geological kimgerlites. Nature, Vol. 346, pp. 54156. Society of Australia Special Publication No. 14, Blacltwell Ross J., Ed. (1989)Icimberlites and Related Rocks (Vols. 1 and2). Scientific Publications, Oxford, pp. 935-965. Proceedings of the Fourth International Kimberlite Con- Hagerty S.E. (1986)Diamondgenesis in a multiple-constrained ference, Perth, 1986, Geological Society of Australia model. Nat~lre,Vol. 320, pp. 34-38. Special Publication No. 14, Blacltwell Scientific Publica- Hawthorne J.B. (1975)Model of a ltimberlite pipe. Physics nnd tions, Oxford. Chen~istryof the Earth, Vol. 9, pp. 1-15. Slodltevich !V (1983) Graphite paramorphs after diamond. JanseA.J.A. (1984)Kimberlites- where and when. In J. E. Glover International Geology Review; Vol. 25, pp. 497-514. and I? G. Harris, Eds., Kirnberlite Occurrence nnd Origin:

Origin of Diamonds GEMS & GEMOLOGY Spring 1991 25