VOL. 13, NO. 4 REVIEWS OF GEOPHYSICSAND SPACEPHYSICS AUGUST 1975

Classificationand Properties of Iron Meteorites

EI•W^RI• R. D. SCOTT 1 AND JOHN T. WASSON

Departmentsof Chemistry and Planetary and Space Science and Institute of Geophysicsand Planetary Physics Universityof California,Los Angeles,California 90024

Most (86%)iron meteoritescan be assignedto one of 12 geneticgroups on the basisof systematic variationsin their chemical,mineralogical, and structuralproperties; the remaining14% are termed anomalous.The groups are best resolved on Ga-Ni or Ge-Niplots, but they may also be defined using other elements,the distributionand morphologyof characteristicminerals, and very often kamacite bandwidths.The powerof thisclassification to revealcorrelations of numerousand diverseproperties withinthese,groups and systematic variations between groups emphasizes its validity. Its useis essential for understandingthe formationof ironmeteorites. A cornpar!sonof the 12groups suggests that there are two typeswith very different histories: (1) themajor groups IIAB, IIIAB, andIVA (11, 32,and 8% of all irons, lCalJCCuvciy/,- ...... :"•"" probably I•,•, '•-' liD, andIVB, andpossibly CtI•U•'*- IC, IIIE, andIIIF ' withinthese groups, mostproperties are correlatedand chemicaland mineralogicaltrends closely similar and (2) the large groupIAB (19%of all irons),IIICD, andprobably IIE, in whichcorrelations between properties are gen- erallymuch weaker and the observedtrends distinctly different from thosein groupsof the formertype. Eachgroup very probably formed in itsown parent body. Groups of thesecond type seem to havediverse formationalhistories, but we believethat, unlike the first type, they were not oncepart of molten cores. Thisstudy is basedon results from nearly 500 different iron meteorites, which are listed with their classifi- cationtogether with 70 otherpaired irons. We presenta comparativestudy of the propertiesof the 12 groupswithout attempting to fit thesedata to detailedmodels for their formation.

CONTENTS we can apply to the classificationscheme: (1) whether the distributionsof other parametersreinforce this classification, Taxonomicproperties ...... 527 Chemical classification ...... 528 for example,by delineatingthe samegroups when taken in Iron meteoriteproperties ...... 530 combination with each other or with one of the defining Iron meteoritegroups ...... 534 parameters,and (2) whetherthe classificationcan reveal cor- Group characteristics...... 540 relationswithin groupswhich are absentfor all the ironstaken Implications ...... 540 as a whole. Our faith in the geneticusefulness of a scheme INTRODUCTION will dependon its successat interpretingother propertiesin The iron meteorites were earlier considered as a fairly this way. We first describesome of the propertieswhich may homogeneouspopulation which shareda commonorigin. be usedto classifyirons and reviewsome of the schemespro- Detailedinvestigations have revealed this assumption to be er- posed.Our aim is to demonstratethat there is now a highly roneous and showed that there is a need for a classification successfulgenetic classification of iron meteorites. systemthat groupstogether genetically related irons, i.e., those TAXONOMIC PROPERTIES that formed in the same locality in the solar systemand ex- perienceda similarchemical and physical history. In thispaper Structural. The structural classificationsystem has been we describe a classification scheme for the irons which has widely usedsince it was developedby Tscherrnak[1872, 1883] beendeveloped during the last 9 yearsand whichwe believe and Brezina [1885, 1904]. It is based on the structure which is fulfils theserequirements. Although these investigations have visible when a polishedsurface is etchedin acid. The majority providednumerous constraints on the historyof the irons,we of irons show an octahedral array of kamacite (a-Fe, Ni) will not attempt to discussthe interpretationof the data in- bands and are classifiedaccording to the width of thesebands. sofar as it relates to the various models that have been pro- Nearly all the remainder are either hexahedritescomposed posedfor the formation of the irons. Instead,we will describe almost entirely of kamacite, which showsa cubic cleavage,or the classificationscheme and the most Usefultaxonomic pa- ataxites composedlargely of taenite (•-Fe, Ni). For a discus- sion of the octahedral Widmanst•itten structure see Axon rameters and review the general chemical and mineralogical properties of the irons. In the secondhalf of this paper we [1968a] and GoldsteinandAxon [1973]. A recentversion of this give a comparative survey of the groups. classificationscheme by V. F. Buchwald (private communica- A varietyof parametershave been used at different times to tion in Wasson[1970a]) is shown in Table 1 together with the classifyiron meteorites.These includethe more obviousones bandwidth divisions of Brezina [1885]. Slightly different values for the limits of the octahedrite classes have also been such as chemical, structural, or mineralogicalproperties and less obvious ones which were determined indirectly, e.g., proposedby Loveringet al. [1957] and Goldstein[1969]. Each cosmicray exposureages or coolingrates. There is no a priori systemrepresents an attemptto fit the limitsto the minima in a method for guessingwhich parameteror set of parameterswill histogramof bandwidthmeasurements. We preferBuchwald's' be the most useful for establishinga genetically significant schemein which the bandwidths in each classvary by a factor classification.In the absenceof a singleparameter that shows of 2.5, since, as we will see, his bandwidth divisionsgenerally several well-defined hiatuses the best combination is that which fall at the edgeof geneticgroups. Figure 5, which is discussed most clearly resolvesthe irons into groups.There are two tests later, showsa histogramof bandwidthvalues for the iron meteorites. •Now at Department of Mineralogy and Petrology,University of As we demonstratebelow, this single-parameterclassifica- Cambridge, Cambridge, England. tion schemedoes not divide the irons into geneticallyrelated Copyright¸ 1975by the AmericanGeophysical Union. groups. The fact that bandwidth itself is a function of two 527 528 SCOTTAND WASSON:CLASSIFICATION AND PROPFRTIFSOF IRON METFORITFS parameters, bulk Ni content and cooling rate, does not, as Brown [1965], who definedthree Ru-Rh and three Ir-Pt groups Goldstein [1969] suggested,necessarily mean that it could from their analysesof 24 irons. There was however, no clear never be usedto classifyirons into genetically related groups. relationshipbetween these two setsof groupsor with the Ga- In fact, virtually all hexahedritesdo form one geneticgroup. Ge groups.Such conflictingquantizations would be difficult to However, in many other casesthere is overlap of bandwidths comprehendbut might conceivablyreflect earlier fractionation betweenthe groups. Nevertheless,the relatively small overlap events,evidence for which had not been completelyerased by of the major groups and the easeand simplicityof the method the event which fractionatedGa and Ge. However, subsequent have ensured the continued usefulness of this scheme. investigations[Wasson, 1967; Crocket, 1972] showed that' Chemical. Although it was recognizedvery early [e.g., Far- these gaps were filled by other irons and the quantizations rington, 1907] that the bulk Ni contentsgenerally increaseas were entirely spurious. the bandwidths decrease,chemical parameterswere not an es- Otherproperties. Here we briefly summarizeother proper- sential part of the classification scheme described above. ties of the irons which might give useful genetic clues. The Goldberget al. [1951] found a good correlation between the results will all be discussedin greater detail in later sections. bandwidthsand the Ga content. But, more importantly, their Theoretical models for the growth of kamacite have been Ga data on 45 irons were quantized into three fairly well developedby Wood [1964] and Goldsteinand Short [1967a, b]. separatedranges: 45-100, 17-22, and 1.7-2.5 ppm. Lovering et Using the relevant equilibrium phasecompositions and diffu- al. [1957] analyzed 88 irons for Ga and Ge and discoveredthat sioncoefficients, these authors calculated cooling rates for the the first of theseranges could be dividedin two. Becausethe Ge irons (at •500øC) by matchingpredicted and observedNi in- concentrations were also quantized, these clusterswere called homogeneitiesin taenite. Goldstein and Short found that they Ga-Ge groups and labeled I-IV in order of decreasingGa could estimatethe cooling ratesof many irons from a knowl- and Ge contents.Eleven of the irons fell outside thesegroups edgeof their bulk Ni contentand kamacitebandwidth. Gold- and were called anomalous. It was this pioneering work of stein [1969] usedthese parameters to developa cooling rate Brown and co-workers which provided the basis for a classificationscheme which supportedthe chemicalclassi- geneticallysignificant classification scheme. fication and is discussed below. Workingwith greatersensitivity, Wasson [i967] resolvedthe Another parameter available for classificationis the cosmic lowest Ga-Ge group into two clusters which were also cor- ray exposureage, which datesthe time when the meteoritewas related with structure. These were designatedIVA and IVB reducedto a meter-sizedchunk in space. Becauseof the oc- and consisted exclusively of fine octahedrites and ataxites, casional discrepanciesbetween ages obtained from different respectively.Extensive analytical data for many other trace techniques[e.g., Lipschutzet al., 1965], we have chosento use elements were published at this time by Sinales et al. [1967] only the extensive results of Voshage [1967] based on the and Cobb [1967]. Although these authors discovered some 4øK/4•K method. He found extensive support for the cor- other groupings,they did not interprettheir data in termsof a rectnessof the chemicalclassification. Additional j•roperties genetic classificationscheme but used instead the structural which might be usedto classifyirons are the abundancesand classification.In a series of papers, Wasson and co-workers compositionsof other minerals besideskamacite and taenite measuredNi, Ca, Ge, and lr concentrationsin nearly 500 iron and the shock damage which these minerals reveal. Mineral meteorites [Wassonand Kimberlin, 1967; Wasson,1969, 1970a; abundanceslargely reflectthe concentrationof minor elements Wassonand Schaudy, 1971; Schaudyet al., 1972; Scott et al., like C, P, N, Cr, S, and Zn, while the shock data give informa- 1973;Scott and Wasson,1975]. By combining structural obser- tion about transientpressures during the breakup of the parent vations with greater analytical precisionand sensitivity,they body. found that the four groupingsof Loveringet al. [1957] could CHEMICAL CLASSIFICATION be resolvedinto a seriesof 12 geneticgroups. This 'chemical' classificationscheme forms the framework for this paper. It is Wasson and co-workers [Scott and Wasson, 1975, and described in detail in the following sections after other earlier paperslisted therein] usedthis title to distinguishthe parametershave been discussed. classificationfrom that based solely on structure (Table 1). To illustrate the potential pitfalls of searchingfor quantiza- Structure was, however, used as a secondaryparameter for tions in analytical data, we quote the work of Nichiporukand classifyingirons. They found that neither Ga nor Ge alone or

TABLE 1. Structural Classification of Iron Meteorites

Bandwidth, mm

Class Abbreviation Brezina [1885] Buchwald* Comments

Hexahedrite H >50 Taenite absent Octahedrite Coarsest Ogg• >2.5 >3.3 Octahedral orientation of kamacite; taenite usually present' Coarse ¸g 1.5-2.0 1.3-3.3 Og, Om, and Of display prominent Widmanst[ittenpatterns Medium Om 0.5-1 0.5-1.3 Fine Of 0.2-0.4 0.2-0.5 Finest Off <0.2 <0.2 Continuousplate network Plessitic$ Opl <0.2 Kamacite platelets do not form continuous network Ataxite D Fine piessite,generally some isolatedkamacite platelets Anomalous:l: Anom Any other iron

* V. F. Buchwald,personal communication in Wasson[1970a]. • The g of Og and Ogg is the first letter of grob, German for coarse. :l:Categories created by Buchwald. SCOTT AND WASSON: CLASSIFICATION AND PROPERTIES OF IRON M ETFORITFS 529

is slightlyless) but that within a group, variationsseldom ex- IOO ceed +20% from the mean. About 14% of the irons do not [it thesegroups and are calledanomalous. They are discussedin a later section.A few group membersshowing minor chemical or structural deviations are called 'anomalous members' and identified by the suffix -An. ---IO E Theterminology t•sed by Wassonand co-workers is based on thatof Loveringet al. [1957].Instead of renurnberingthe ßgroups whenever a new one was discovered,they assignedlet- ters within each of the four original 'Ga-Ge groups.' Unfor- tunately, the continueduse of the original numberingsystem J - has misled.some authors [e.g., Frick and Hammerbeck, 1973; Vdovykin, 1973] into believingthat the groups establishedby Wasson and co-workers were mere subgroups of the four original Lovering groups. We emphasizethat this is incorrect. The continued use of terms suchas group III [e.g., Jain eta!.,

7 I0 14 20 25 1972]is to be deplored.Group IIlE, for example,is as little Ni•:kel (%) relatedto IIlF asit is to groupiVA. Thereare only four cases Fig. 1. Logarithmicplot of Oa againstNi showingthe 5utlineof where there is good evidencefor geneticrelationships between the ironmeteorite groups. Apart from IAB andIIICD, thegrou•s two groups;IA-IB, IIA-IIB, IIlA-IIIB, and lIIC-IIID. In each showvery limited ranges of Ga contents,less than 4-20% a5out tBE case the first listed has the lower Ni content. Each of these four mean. About 14% of known iron meti•orites are anomalous (i.e., are pairswe considerto be a singlegenetic sequence, and we use notmemberg of thesegroups) and are not shown (see Figure 9). The straight dashed line through groups IA, IC, and IIAB shows the the labels lAB, etc., to refer to the composite group. Ga/Ni ratio for C1 chondrites. Data for C1 ratios in the figures are Nearlyall the groupsshow correlations of Ga, Ge, andNi. taken from Chou et al. [1975]. Note that different vertical scalesare Figure9, whichis discussed later, shows that Ga and Ge are used in Figures 1-3. nearly alwayscorrelated, always positively, within eachgroup. Cori'elations with Ni, however, are positive in some groups togethercan entirely resolvethe groups. In fact, a Ga-Ni or and negativein others (Figures 1 and 2), as discussedlater in Ge-Ni plot (Figures 1 and 2) is the most effectivetwo-element the group descriptions.Gallium and Ge are the only two ele- plot for definingthe groups,but an additionalparameter is mentswhich showboth positiveand negativecorrelations with sometimesneeded. Two pairs of groups,IIE and IIIAB and IC Ni in the various groups. The fractionation of theseelements and IIAB, have overlappingGa valuesbut are resolvedon the within groups is very mild in comparisonwith that observed Ge-Ni diagram, while IIC and IID are better separatedby Ga. for mostother elements. However, the significantinterelement Groups IIC and IB overlap on both Figures 1 and 2, but the correlationswithin groups are further proof of the genetic secondarytaxonornic parameters used by Wasson and co- significanceof this classificationand provide a method of se- workers, Ir contents and bandwidth measurements(Figures 3 quencingthe ironswithin each group (see below). and 4), easilydistinguish them. Notice that the total rangeof

Ge valuesin the groupscovers over 4 ordersof magnitude(Ga JO0' ' '

IOO I

TVA 0.1

o.I TVB

5 7 I0 14 20 25 Nickel (%)

Fig. 2. Logarithmic plot of Ge againstNi; n6te the similarity to Nickel (%) Figure I. Most groupsexcept IAB and IICD show Ge variations whichare very smallin comparisonwith the total rangeof more than 4 Fig. 3. Logarithmicplot of Ir againstNi showingthe outline of the ordersof magnitudeshown by all the groups.Germanium is probably iron meteoritegroups. The major groups,IAB, IIAB, IIIAB, and IVA, the most usefulsingle trace elementfor classifyingirons; only in one are shownin heavieroutline. Groups IAB and IIICD are drawn with case(lIC and lID) is Ga better able to resolvetwo groups. In those dashed lines to contrast them with the other groups, most of which caseswhere groups overlap, other propertiesmay be readily usedto show strong negative correlations and steep slopes. Some 1ow-lr resolve them. Group IB is very sparselypopulated with only 8 members were excluded from the lIE field to reduce confusion at the members and is shown in dotted outline. center of the plot. 530 SCOTT AND WASSON: CLASSIFICATION AND PROPFRTIESOF IRON METEORITFS

TITCD

Nickel(%) Nickel(%) Fig. 4. Logarithmicplot of (a) Au and (b) Ru againstNi. Most groupsare shownonly in outline;this is dashedwhen there are few data points in a group. Although the resolutionof the groupsis poorer than in Figures 1 and 2, the chemical classificationreveals many correlationswithin groupswhich are not apparent for all the irons. Gold is positivelycorrelated with Ni in most groups,the trendsbeing much lesssteep in groupsIAB and IIICD, while Ru is negativelycorrelated with Ni. For data sources see text.

In Table 2 we list the number of members,the range of Ni, IRON METEORITE PROPERTIES Ga, and Ge contents,and kamacitebandwidths in eachgroup. In this section we discussthe extent to which the other pa- These statisticsare basedon analysesof 479 differentirons rameters describedearlier support the chemical classification (known paired irons are eliminated) by Wasson[1970a], Was- of irons. It will be seenthat this classificationreveals many re- son and Schaudy [1971], Schaudy et al. [1972], Scott et al. lationships between parameters which are not evident for the [1973], and Scott and Wasson[1975] and four by Reed [1972]. irons taken as a whole. All thesemeteorites plus the paired irons are listed in the ap- Trace and minor elements. Resultsfrom Ir analysesof 479 pendix in alphabetical order with their classifications.It can be irons [Scott and Wasson,1975, and earlier paperslisted there- seen that group IIIAB is the largest with 32 + 3% of all irons in] are shown in Figure 3 on a logarithmic Ir-Ni plot. The con- and the three next largest, IAB, IIAB, and IVA, contain 38 + trast with the Ga-Ni and Ge-Ni plots could hardly be more 3%. The errors are calculated from the square roots of the obvious. At least two groups show a variation of more than 3 numbers of separately analyzed members. The ranges of orders of magnitude, and only one iron has an Ir concentra- chemical and structural parameters found in each group tion outsidethe rangeshown by IIAB. Although most groups (Table 2) are only overall limits; the compositionsare defined on this plot overlapto someextent, it is important to note that much more preciselyin plots like thoseof Figures 1, 2, and 5. within nearly all groups Ir and Ni are negativelycorrelated.

TABLE 2. Structural and Compositional Properties of Genetic Groups of Iron Meteorites

Frequency, Bandwidth, Ni, Ga, Ge, Group N umber % mm Structure % ppm ppm

IAIB 828 17.01.7 0.01-1.01.0-3.1 Om-OggD-Om 6.4-8.78.7-25 55-10011-55 190-52025-190 IIAIC 3910 2.18.1 ><3 50 Anom,H Og 6.1-6.85.3-5.7 49-5557-62 212-247170-185 II B 13 2.7 5-15 Ogg 5.7-6.4 46-59 107-183 IIC 7 1.4 0.06-0.07 Opl 9.3-11.5 37-39 88-114 lid 13 2.7 0.4-0.8 Of-Om 9.6-11.3 70-83 82-98 I1E 12 2.5 0.7-2 Anom* 7.5-9.7 21-28 62-75

_•111AI11B 12036 24.87.5 0.9-1.30.6-1.3 Om 8.4-10.57.1-9.3 16-2117-23 27-4632-47 11IICII D 75 1.41.0 0.01-0.050.2-0.4 Off-OfD-Off 16-2310-13 1.5-5.211-27 1.4-4.08-70 II 1E 8 1.7 1.3-1.6 Og 8.2-9.0 17-19 34-37 I 11F 5 1.0 0.5-1.5 O m- Og•' 6.8-7.8 6.3-7.2 0.7-1.1 IVA 40 8.3 0.25-0.45 Ot' 7.4-9.4 1.6-2.4 0.09-0.14 I V B 11 2.3 0.006-0.03 D 16-26 0.17-0.27 0.03-0.07

* Also Om and Og. I' Also Ogg and Of. SCOTT AND WASSON: CLASSIFICATION AND PROPERTIESOF IRON METEORITES 531

Even in IIAB, where the gradient is very steep,the correlation bandwidth measurements in octahedrites and ataxites; the is highly significant. Significant correlations within groups measurementsare almost entirely by Buchwald, taken from where none exists for the irons as a whole are reassuringevi- papers by Wasson[1970a], Schaudyetal., [1972], Scott etal. dence for the successof the chemicalclassification at isolating [1973], and Scott and Wasson[1975]. Also shown are separate meaningful groups. This plot can often provide valuable histogramsfor the groups that dominate each of the structural evidencefor the position of an iron within a group sequence classes(excluding Off). The four major groups IA, IIAB, and sometimes even its classification. lllAB, and IVA are nearly resolved by bandwidth measure- Publishedanalyses for some20 other trace and minor ele- ments (the single-crystal hexahedrites of llA, for which ments in iron meteorites were examined by Scott [1972]. In bandwidths are not defined, would plot to the right of this many cases,most groups could be resolvedon element-Ni diagram). The structural classification, however, fails to iden- plots. Strong correlationswere generallyobserved within the tify the smaller groups, e.g., lIC, llIE, and those with a large groups, and on most diagrams, regression lines for the spread of bandwidths, i.e., IB and llICD. The boundaries of different groups had similar slopes.He concludedthat there the structuralclasses were adjustedby V. F. Buchwald(private could be few remainingdoubts as to the geneticsignificance of communicationin Wasson[1970a]) to maximize the number of the chemical classification. We will attempt to summarize membersof each major group in a singlestructural classand some of the conclusionsfrom these plots in subsequentsec- are listed in Table 1. tions. Here we show only two of these interelement plots. The structural classificationtends to separate the major Figure 4a is a Au-Ni plot usingAu analysesfrom Bauerand groupsbut failsto isolatethe smallergroups or anon4alous Schaudy [1970], Cobb [1967], Crocker [1972], Fouch• and irons. A bandwidth-Ni plot (Figure 6) however,resolves some Sinales [1966], Goldberg et al. [1951], and E. R. D. Scott additional groups (IIC, IIIF, and IVB) and shows trends in (unpublisheddata, 1975). The total range in Au valuesshown many of the groups. A small number of irons meet the by all irons is barely 2 ordersof magnitude,and within groups chemical but not the structural requirementsfor a group. The the variation is lessthan a factor of 5. Strong positive correla- Ni, Ga, Ge, and Ir contentsof Juromenha[Scott etal., 1973] tions and steepslopes are visible within all groups for which appear to fit IIlA, but any octahedral structure has been there are sufficientdata, with the exceptionof lAB and IIICD. obliterated, presumably by cosmic reheating. Another exam- Scott [1972] noted that the distributionof As is almost identi- ple is Arltunga [Scott and Wasson, 1975], which has a cal to that of Au, while P, Pd, and Sb show similar positive bandwidth 100 times narrower than other lid members and is correlationswithin groups but somewhatdifferent fractiona- classedliD-An. Two groups, IC and liE, are not plotted in tions betweengroups. Cobalt-Ni and Mo-Ni correlation lines Figure6; althoughthe membersof eachare chemicallyvery showlower slopes,the Co and Mo valuesranging over factors similar, they display a bewilderingvariety of structures.Thus of 3 and 10, respectively. bandwidths are a useful but not infallible guide to classifica- In Figure 4b we show a Ru-Ni plot basedon the analysesof tion. Hara and Sandell [1960], Nichiporuk and Brown [1965], and Coolingrates. Following the observationby Goldberget al. Crocket [1972]. Negative correlations are visible within the [1951] of a limited correlation of bandwidth and Ga content, major groups, excluding IA. The limited data suggestmany Yavnel [1961] interpreted the bandwidth-Ni plot in terms of similarities to the Ir-Ni plot (Figure 3), although the total four intersectingzones of decreasingbandwidth with increas- variation of Ru is only a factor of 11Y,instead of 104for Ir. Rhodium and Pt show very similar distributions to Ru, while 1 those for Os and Re are almost identical to that of lr [Herr et al., 1961; Kimberlin et al., 1968; Crocket, 1972; Scott, 1972]. Ioo The last author noted that the Cr-Ni plot also showed steep _ negativegradients within groups but the variation betweenthe --Offi OfTrrA-+-Om-+--Og--t•Ogg-- t groups was rather different from these Pt metals. •O.L- 40"- For some elements(e.g., N, C, In) the measuredconcentra- tions scatter sufficientlythat only a few generalizationsmay be made and no interelement correlations observed [Scott, 30-

1972]. But thoseelements discussed above are nearly all highly ZA correlated within each group; exceptionsare mainly found in 20- group IAB. Thus the members of a given group can be ar- ranged sequentially, so that once a new member is located within the sequence,its completetrace and minor elementcon- I0- tent can be estimatedwith accuracy.This property of groups was predicted by Wassonand Kimberlin [1967] and fully lXIBI[C documented by Scott [1972] and Yavnel [1972], who in- 0.01 0.1 I I0 Komocite Bondwidth (mm) dependentlyobserved these correlations. It is apparentthat the term 'Ga-Ge classification' is no longer appropriate, since Fig. 5. The bold outline shows the distribution of kamacite bandwidthsin 365 irons on a logarithmic scale.The shadedareas show numerous other elements could be used to classify an iron. the bandwidth distributions in groups IVB, IIC, IVA, IIIAB, IA, and However, none show the combination of small intragroup and liB, which are the major groups in each of the structural classes(ex- large intergroup variations exhibited by Ga and Ge. Wasson cluding Off) shown at the top of the figure. Note that shadedareas are [1967] suggested'chemical classification' when a distinctionis plotted independentlyand are occasionallysuperimposed. The largest needed between it and the structural classification, but as we groups (including l lA, which might be plotted to the right of this figure) occur as peaks in the bandwidth distribution, but the smaller will showin following sections,any adjectiveis rather unneces- groups and anomalousirons are not well-definedon this histogram. sary; there is only one genetic classificationof the irons. Measurementsare mostly by Buchwald [Wasson, 1970a; Schaudyet Kamacite bandwidths. Figure 5 is a histogram of al., 1972; Scott etal., 1973; Scott and Wasson, 1975]. 532 Scott ANDWASSON: CLASSIFICATION AND PROPERTIES OFIRON METEORITES ingNi; onefor eachof thefour Ga-Ge groups. Yardley [1966; IO discussedby Axon, 1968a]reinvestigated the data and sug- gestedthat threecurves for coolingrates of about2 ø , 5ø , and 10øKMy -• for groupsI and II, III, and IV, respectively,gave a better fit. However, the variation of the coolingrates in the irons was not well understood until the work of Goldsteinand Short [1967a,b] andShort and Goldstein [1967]. These authors prepareda bandwidth-Niplot showing a numberof curvesfor coolingrates between 0.25 ø and 500øK My -•. Someof these are drawnin Figure6, with the assistanceof Wasson's[1971] equationfitted to the Goldstein-Shortresults. Goldstein and Short [1967b]found small cooling rate rangesof about 1ø-3ø K My -• in groupsIA andIIIB. GroupsIVA andIIIA showed approximatelytenfold increases in coolingrate with increasing bulk Ni content. Figure 6 also showsthat groupsIAB, lid O.Ol IIICD, and IIIE are consistentwith a 2øK My -• curve.The idea of Yardley[1966] that coolingrates are constantwithin groupsapparently holds for groupsIAB, IID, IIIB, IIICD, IIIE, and possiblyIIC and IIIF. Nickel (%) Figure 6 showsthat three groupshave characteristically Fig. 6. Logarithmicplot of kamacitebandwidth against bulk Ni highcooling rates: IIC with ratesof 100ø-500øKMy -•, IVA contentshowing the outline of the groupsexcluding IC and liE, which showlarge ranges,and IIA, which consistsof hexahedrites.Diagonal with 5ø-200øK My -•, and IVB with 5ø-200øK My -•. Errors lines are isocoolingcurves from Goldsteinand Short [1967b]. Groups in the last namedgroup of ataxitesare large,however, because IIC, IVA, and IVB have high cooling rates resultingin characteristic of difficultyin measuringthe narrowkamacite platelets. These structureswhich may be easily recognized.Groups lAB, liD, IIIB, threegroups can thus be easilydistinguished from the others and IIICD have cooledmore slowly and lie within the error limits of a 2øK My-• coolingcurve. Also shownare the bandwidthranges for the which all have ratesof lø-5øK My -•, with the exceptionof various structural classes. IIIF, which appearsto be intermediate.Below bulk Ni con- tents of 7% thesecooling rate curvescannot be usedbecause should have been reducedto meter-sizedchunks by a single impingement,a secondmechanism of bandgrowth, becomes collision,but nearlyall membersof groupsIIIB and IVA seem dominant. to have formedin this way. It is noteworthythat groupIIAB, Goldstein[1969] demonstratedthat Ni and bandwidth whichshows so many chemicalsimilarities to thesetwo groups measurementswere sufficientto resolvemany of the chemical [Scott,1972], was split by multiplecollisions. The clustering of groups.However, Figure 6 showsthat severalgroups are not agesin groupsIIIAB and IVA providesthe bestevidence that resolved;group IAB overlapsliD, IIIB, IIICD, and IIIE. the membersof each group residedin a singleparent body Goldsteindivided the Ni-bandwidthplot into 24 boxeswhich [Voshage,1967]. The smoothvariations of bandwidthwith Ni he called'cooling rate' groupsand notedthat 10 were devoid content within most groups discussedabove provide ad- of members. We do not consider this classification schemevery ditional though weaker evidence. usefulbecause of the incompleteresolution of severalgroups Mineralogy. Therehave been no studiesrelating the com- and our earlier conclusions[Scott et al., 1973;Scott and Was- pletemineralogy of the ironswith the geneticclassification; son, 1975]that bandwidthmeasurements alone should not be Table 3 representsa first attempt to summarizethese used to exclude irons from a group. Some authors [e.g., relationships.In viewof the virtualabsence of publishedquan- Gooleyet al., 1971] have assignedmeteorites to a Goldstein titative estimatesof mineral abundancesin any iron meteorite coolingrate group, but in the absenceof traceelement data it and the absenceof even qualitativedata for many minerals, is better to quote the chemicalgroup or groups that the muchof Table 3 is basedon personaland possiblyrather sub- bandwidthand Ni data suggestthe iron shouldbelong to. Ad- jectiveevaluations. Troilite is an especiallydifficult mineral to ditionalmineralogical and petrologicalevidence will often al- low a unique classificationfor the iron. • IA •6•I!IAB Datofrom Voshage (1967) Cosmicray exposureages. Voshage[1967] noted that his 8 - - data for 62 iron meteorites (Figure 7) showed two groups •' • 'IX7ALr-• A7 B whosemembers had closelysimilar cosmicray exposureages. 6- - All but one of 18 membersof group IIIAB appear consistent with an ageof 650 + 100My, whileall but two of 9 IVA irons clusteraround 400 My. His error limits are large enoughto suggestthat the resolutionof two peaksin the IIIAB cluster o [ , , oooIOOO• I may not be significant.Most other groupsshow a wide range O 02 O.• 0.6 O.8 •.• L• of values:500-900 My for IAB, 100-1200 My for IIAB, and Cosmic m• age 100-800 My for IIICD. Wasson[1969] arguedfor an inverse correlation between Ir and exposure age in IIAB; in fact, a •on m•eo•es by •he •K-•K m•hod [F•g•, •967]. Two more cautiousstatement'is that agesare low (<300 My) in IIA showwe•bde•n•d p•ks, [[[AB • 6•0 •y •nd [VA • 400 •y. The and high (300-1000 My) in IIB. Scott and Bild [1974] note an o•h• •oups [d•nfi•d, lAB •nd [VB, show w[d• mn•s o[ inversecorrelation between Ni and exposureage for five group IIICD members but no very significantcorrelation of ages with compositionalparameters in IAB. There is no a priori reasonwhy all the membersof a group SCOTT AND WASSON: CLASSIFICATIONAND PROPERTIESOF IRON METEORITES 533 estimate becauseof the irregular distribution of nodules 1 cm of small groups where trends are lesswell deftfled,e.g., car- in diameter or larger [e.g., Henderson and Perry, 1958; bides in group IIIE [Scott et al., 1973]. An experienced Buchwald, 1971b]. Phosphide abundancescan be easily es- microscopist,e.g., Buchwald[1975], can classifyperhaps 90% timated from a microscopicsurvey unless P is very abundant, of the irons on the basis of mineralogyand microstructure in which caselarger sectionsare needed[Reed, 1969;Doan and alone. It is interestingto note that Cohen[1905] in his survey Goldstein,1969]. Carbide and graphite abundanceshave been of irons managed to group together on the basis of discussedby Scott [1971a, b] and Scott and Agrell [1971], mineralogy,structure, and Ni content many meteoriteswhich carlsbergiteby Buchwaldand Scott [1971], and phosphatesby we now know to be geneticallyrelated. For example,four out Bild [1974]. Other usefulpapers include those of Bunchet al. of the five membersof his Cape of Good Hope group are [1970] for groups IAB and IIE, El Goresy[1965], plus papers membersof IVB. In the sectionsdescribing the groups, we which are listed in following sectionsby Axon and co-workers discusssome of the mineral occurrences,especially those of and Buchwald on individual meteorites. Wasson and co- silicates, in greater detail. workers [Scott and Wasson, 1975, and earlier papers listed Further clues to the classification of an iron can be extracted therein]have alsosupplied brief noteson somecharacteristic from the morphology and environment of certain minerals. minerals in each group. With the exceptionof severalrare For example,piessite if often pearliticor spheroidized[Perry, silicateand sulfideminerals [Mason, 1972], few other minerals 1944] in the high C groups,IAB and IIICD [Scott and Bild, besides those discussed in the text or Table 3 have been 1974], and silicatesangular in IAB but rounded in liE. Other observed in irons. examplesare the fieldsof cohenire-bearingkamacite bands in Notice that many mineral abundancesvary systematically IA or the large (>•2-cm) troilite nodulesin IIIB which are well- within groups. Thus phosphideand nitride concentrations, armored with schreibersite.These clues are clearly omitted respectively,increase and decreasewith bulk Ni content in from Table 3 but are discussedin the group descriptions group IIIAB. But other mineral abundances,especially in below.Unlike chondrites,iron meteoritescannot generally be groups lAB, IIE, and IIICD, vary randomly or show no classifiedby the compositionof their constituentminerals, significantcorrelation with Ni content. In Table 3, systematic although silicate and chromite compositions[Bunch et al., variations with increasingNi content are indicated by an ar- 1970] in group IAB and lie irons are diagnostic. row. A dash between two numbers indicates either an ap- Shockfeatures. Table 3 also attemptsto estimatethe fre- parently random variation (e.g., silicatesin IAB) or uncer- quency of two shock indicators, shock-melted troilite and tainty, but usuallyit is an attemptto suggestan intermediate shock-hatchedkamacite [Axon et al., 1968;Lipschutz, 1968], in abundance.It should also be noted that only relative abun- the groups. Jain and Lipschutz [1971] review the shock dancesof a given mineral in the groupsare listed. Even when historiesof the irons, thoughgenerally ignoring the additional ubiquitous, carlsbergite,for example, is much less common evidencerecorded by the troilite, which has also been studied than schreibersite. by Buchwald[1975]. In a single meteorite,apart from oc- Because it is the commonest mineral, schreibersite is often casional shocked grains of different orientations, all the the most useful mineral for classifyingirons. However, the kamaciteappears to showa uniform recordof shockhistory. presencebut not the absence of carbides, graphite, and However, within a group there is often inconsistentbehavior, 'silicatesis a veryuseful characteristic of somegroups and is es- which reduces the value of shock-hatched kamacite as a tax- pecially usefulfor identifyinganomalous irons and members onomicparameter. For example,most but not all group IIIAB

TABLE 3. Relative Abundances of the More Common Minerals in the Iron Meteorite Groups

Sulfides Carbides Shock Signs Phosphide: N itride Troilite Daubr6elite SchreibersiteCoheniteHaxonite Graphite Carlsbergite Melted Hatched Group FeS FeCr•.S4 (Fe, Ni)aP FeaC (Fe,Ni)•.aC6 C CrN Silicates Troilite Kamacite Others

IAIB 2, N 01 2, (M)M, rh 3,1 M 21-2 2, (M) 0I 1-3, M* 1-2I 0I (ch),ch,ph, ph,sp sp IC 2, N 1-2 2, (M), rh 1-3, M 0 dc 1-2 0 1-2 0 ch, sp IIA 1 2 2,rh 1-2 1 dc I 0•' 2-3 0 (ch) lib 1, (N) 0-1 2 -, 3, rh I 1 dc 0 0•' 1 I (ch) IIC 1-2, (N) 0-1 2 0 0 0 0 0 1-2 0 (ph) lid 1-2, (N) 0-1 2 -• 3, M 0 1 0-1 0 0-1 2 I (ch) lie 1-2 0 1-2, (rh) 0 0 0 0 0-3, M 1-2 0 ch, ph Ilia 1-2,N 2 1-, 2,rh 1 0 0-1 3-• 0 0 2 3 (ch),(ph), sp I II B 2-3, N 1 2 -, 3, M 0 0 0 0 0 1-2 3 (ch), ph IIIC 1,(N) 0 2,M 0 2 0 0 0 0 0 sp IIID I 0 2, M 1 2 1, (M) 0 0-1 I 0 ph, sp 1IIE 1-2 2 1-2, rh 0 2-3 dc 1-2 0 1-2 0-1 (sp) IIIF 1 2-3 1-2 0 0 0 0 0 I 0 (ch) IrA 2, (N) 3 • 2 0 • 1 0 0 0 0 0-1 2-3 2 (ch) IVB 1 1-2 I -• 2 0 0 0 0 0 1-2 0 (ch)

Key to abundances:0, not observed; 1, sparselydistributed; 2, common; 3, ubiquitous. Key to abbreviations:parentheses, infrequently observed; arrow, changein abundancewith increasingbulk Ni content;N, nodules>1 cm in diameter;M, macroprecipitates•1 cm in length' rh, rhabdites;dc, graphite from decomposingcarbide only; ch, chromite(FeCr2On): ph. phosphates;sp, sphalerite (ZnS). * Bulk silicate compositionclose to chondritic values. •' See text. 534 SCOTT AND WASSON: CLASSIFICATION AND PROPERTIESOF IRON METEORITES

irons show shock-hatchedkamacite [Jain and Lipschutz, 1960], melting and annealing. Although the variations in taenite but a few are apparently unshocked.Group IIA, where the abundanceprobably reflectgradients in Ni content in Canyon effect is absent, is the only large group showing consistent Diablo, the precipitation of cohenite doesproduce grossvaria- behavior. The abundanceof shock-meltedtroilite in this group tions in C content. The enormous ease of diffusion of C appears at first to be inconsistentwith this evidence, since togetherwith the apparent difficulty of nucleatingcarbides can shock pressuresmuch higher than the 130 kbar needed for produce big C concentration changesover distancesof about hatched kamacite are required to melt troilite [Heymann, 10 cm. Canyon Diablo samples with high Ni contents were 1967]. Several IIA irons are, however, severelyrecrystallized. classedas IA-An by Wasson[1970a]. Scott [1972] considered Perhaps the shock occurred at high temperatures, and the these inhomogeneitiesin group IA additional evidencefor a resultant disordering of the kamacite has since annealed out unique (nonigneous)history for this group. With theseexcep- in the others [Axon and Smith, 1970b]. tions the other chemicaland structuralparameters discussed Other signsof deformation in irons (e.g., shearzones) have earlier appear to be constant within a single iron meteorite. been reviewed by Axon [1969]. The examples he describes In the following sectionswe examineeach group in turn, come from diverse groups such as lid (Puquios), Ilia describing its characteristic mineralogical, structural, and (Descubridora), and lie (Kodaikanal), and there is no clear chemicalproperties. In this comparativesurvey we emphasize correlation of theseeffects within the groups.Some of the de- the similaritiesbetween groups IIAB, IIIAB, IVA, and prob- formation, for example in Canyon Diablo and Campo del ably most of the smallergroups and the very differentbehavior Cielo, most probably occurred on impact with the earth, and in groups IAB and IIICD and also liE. it would appearthat deformationduring breakup of the parent IRON METEORITE GROUPS bodies was relatively localized. Meteorite homogeneity. Implicit in this discussionof iron Group IAB. Group IAB, the second largest group, con- meteorite classification has been the assumption of tains the two crater-forming irons, Canyon Diablo and Odes- homogeneityin a singleiron. Chemical homogeneityhas been sa. Becauseso many group IAB propertiesare not sharedby considered by a number of authors (Goldberg et al. [1951], other groups, it seemshighly probable that IAB had a dis- Lovering et al. [1957], and many others)who find that with few tinctly different formational history. exceptions,replicates agree within the limits of experimental Some changesin terminologyhave recentlybeen proposed errors. As noted earlier, some irons have large centimeter-sized for this group.Those irons Wasson[1970a] called group ! plus inclusions of schreibersite,troilite, graphite, cohenite, or that portion of the categoryI-An2 with Ge contentsabove 190 silicates,and most irons have millimeter-sizedgrains of some ppm we now call group IA. The remainder of I-An2 with less of these minerals. Analytical samplesare usuallychosen to than 190 ppm Ge is designatedgroup IB. Although IAB ap- avoid theseinclusions and are generallyabout 1 g in weight. pearsto be a singlefractionation sequenceon Ga-Ni and Ge- This sizeis almostalways large enough to ensurethat thereis a Ni plots (Figures 1 and 2), we find it usefulto distinguishthe representative ratio of kamacite and taenite, to minimize er- sparsely populated high-Ni IB tail from the more rors for elementsthat are enrichedin either of thesephases. homogeneousIA head. In addition, group IB tends to have However, the concentrationof chalcophilicor lithophilic ele- lower concentrationsof refractory elements than IA. The mentswill clearly be upsetby a nonrepresentativesample of category labeled I-Anl by Wasson[1970a] plus Mertzon and sulfide or silicate inclusions. Few of the elements discussed in Morrill from I-An3 are now called IA-An and the remainder the previoussection show strong enough preferences for these of l-An3 either IC (see below) or anomalous.In the appendix phasesto cause significanterrors. Variations in Zn content are listed 82 different IA and 8 IB irons; 27 were classifiedby may be producedin this way [e.g.,Smales et al., 1967].Clearly, Scott and Wasson[1975], two each by Reed [1972] and Scott et analysesfor S [Hendersonand Perry, 1958] and P [Doan and al. [1973], and the remainder by Wasson[1970a]. The appen- Goldstein,1969] must also include large inclusionsin those dix also showsthat group IAB has more anomalousmembers groupswhere they are frequent.Carbon analyses[Moore et al., (10) than any other group. 1969]show very wide variations between replicates, which may Although irons in group IAB have structuresvarying from a often be caused by a very heterogeneouscarbide content. coarsestoctahedrite to an ataxite, they are consistentwith a Cohenite, for example, is often amazinglyvariable within a cooling rate of 2øK My -•, a value not exceededby any other singlemember of group IA, evenover distancesof 10 cm [e.g., group (Figure 6). Mineralogically, group IAB irons are char- Wasson, 1970a, Figure 3, 1974, Figure XIV-9]. Chromium acterizedby troilite-graphite-silicatenodules and abundant car- variations,which are muchsmaller, generally less than a factor bides.The presenceof all four mineralswould be a fairly sure of 2 from the mean, may be producedby statisticalfluctua- identificationfor a group IAB member. Silicateshave not yet tions in daubrfielite content. been confirmed in many IA irons and are quite rare in all the One meteorite in which significantvariations have been dis- other groups except IIE. The presenceof graphite intermixed coveredis in Canyon Diablo. Moore et al. [1967] and Wasson with troilite, or cohenite forming shells on nodules and [1968] found that the Ni contentsof differentsamples varied millimeter-sizedblebs oriented along kamaciteplates provides from at least7 to 8 wt %, through Wasson'sGa and Ge values a fairly good label for a IAB iron. Sometimescohenite has par- were essentially constant. Increases in bulk-Ni content were tially decayedto graphite and low-Ni kamacite along cracks, correlated with taenite content and anticorrelated with and in Dungannon and Wichita County the processhas gone bandwidth.The scaleof theseinhomogeneities is unsurebut to completion,so that graphiteribs outline the extinctcohenite certainly between 10 and 102cm. Variations in Ni content over precipitates. Spheroidized or pearlitic plessite fields [Perry, thesedistances cannot be producedby exsolutionof kamacite 1944] are characteristicof this group and may be partially fil- or cohenite because Ni diffusion distances are limited to dis- led with haxonite [Scott, 1971a]. Schreibersiteis fairly abun- tancesof <1 mm in the time available.They were probably dant, but unlike most other groups,group lAB showsno in- producedearly in the meteorite'shistory, possibly during ag- creasein schreibersitecontent with increasingbulk Ni content. glomeration, and were not erased during any subsequent Instead, it seems to be uncorrelated with Ni content. SCOTTAND WASSON: CLASSIFICATION AND PROPERTIESOF IRON METEORITES 535

Carlsbergiteand sphaleriteare sparselydistributed as micron- 1 1 1 i i sizedprecipitates in kamacite.Very few lAB irons, perhaps IA IAB IB 5%, show shock-hatched kamacite. [] IA In addition to forming minor constituents of troilite- [] lB graphitenodules, the silicatesmay occuras massive,generally angular, inclusions.Olivine (Fa•_8), orthopyroxene(Fs4_9), [] wrc IC and plagioclase(Ab?6_87) are the major minerals[Bunch et al., IC lie wrCD 1970]. The abundanceof silicatesmay vary enormouslyeven betweenneighboring irons in the lAB sequence.Often they may be unobservedunless very large slicesare available, but 1TA • 24 the volume of silicatesseldom exceeds 15% [e.g.,Mason, 1967; lib Bunchet al., 1972]. In silicate-richlAB irons, the Widmanst•it- m 20 ten pattern is seldom continuous for distancesof more than 15 1TC4 mm, probably becausethe high inclusion content inhibited I[ABIIEAB ,F• ,)-l, o ]]IE grain growth in the parent taenite. There is no correlation between bulk Ni content and the Fe content of the fer- romagnesianminerals. Other papers discussingthe silicate 8 mineralogyof IB ironsinclude Rambaldi et al. [ 1974]and Scott '- [] m'E and Bild [ 1974]. 0 -

Bunchet al. [1970] dividedthe silicate-bearingirons of IAB TrD into Odessa-and Copiapo-typeslargely on the basis of the ]]IF Mn, Zn, and Ti contents of troilite. We do not find this distinc- 4 I]ZB tion convincing or useful, especiallyin view of the inter- 8 •IEF•'TrD mediatetroilite compositionsof Landes [Bunchet al., 1972] 5 6 7 8 9 I0 12 14 16 18 20 215 and the differing troilite compositionsin Tacubaya and Nickel (%) Toluca, despitecompositional and geographicalevidence that Fig. 8. Histogramsshowing the distributionof Ni contentsin each the former is a mislabeledspecimen of the latter [Niningerand of the12 groups. Each group is drawn independently, and overlapping groupsare identified by shading.All largegroups show a peakat the Nininger, 1950; Scott and Wasson, 1975]. low-Niend of thegroup, followed by a steadydecrease toward higher Bulk silicate analysesby Jarosewichare available for two Ni contents,but IAB hasa secondpeak around 8% Ni. The additional IAB irons,Woodbine [Mason, i967] and Campodel Cielo peakin IVA is muchweaker and may not be significant.Notice that [Olsen and Jarosewich, 1970]. These authors note the close groupslAB and IIICD havemuch wider Ni rangesthan any other group, even on the logarithmic scaleused. Data taken from Wasson similaritiesto silicatesin H group chondrites.The major and co-workers[Scott and Wasson,1975, and earlierpapers listed differences are the lower Fe contents of olivine and therein]. orthopyroxene in group IAB, intermediate between those in the H group and enstatite chondrites. As discussedbelow, silicatesin group liE, the only other group to containsignifi- the95% confidence level. However, the negativeslope of the IB cant amounts,are easilydistinguished by the virtual absenceof trendis muchless steep than in the othermajor groups. On olivine and higher Cr and Fe contents. other interelementplots (e.g., Figure 4) group lAB fre- A histogramof Ni contentsin group IAB (Figure 8) shows quentlyfails to showthe trendsvisible in othergroups [Scott, that as in other groups most membersare concentratedat the 1972].We have suggestedthat the absenceof a fully molten 1ow-Ni end; about 50% have Ni contents between 6.5 and stage in the history of this group is responsiblefor these 7.5%. But group IAB, unlike the other major groups,appears differencesbetween group IAB and the othermajor groups to havea secondpeak around8% Ni, althoughFigure 8 shows [Wasson,1970a; Scott, 1972]. One principal piece of support- IVA may also have a bimodal distribution. Mislabeled frag- ing evidenceis the presenceof silicateswhich would quickly mentsof large showerslike Canyon Diablo and Toluca might separate from molten metal. We believe that the metal-silicate easily produce a false peak, but we believe that most such fractionation essential to form these irons from chondritic- meteoriteswere excludedfrom this plot. Note that the high-Ni solar mixtures of the elements occurred in the solar nebula and tail in group lB, which stretchesto San Cristobal with 25%, is not in the parent body. vastly more pronouncedthan in the other groups. GroupIC. Group IC is a group of ten irons with 6.1-6.7% Chemically, group IAB appears to have quite different Ni, whichare describedby Scottand Wasson[1975]. Eight propertiesfrom the other major groups.When normalizedto were analyzed by Wasson [1970a]: Arispe, Bendeg6, Ni, most siderophilicelements are presentin abundancesclose ChihuahuaCity, Mount Dooling,St. FrancoisCounty, Santa to thosein CI chondrites[Wasson, 1970a; Scott, 1972]. There Rosa,Union County(all listedas I-^n3), andNocoleche (then is a trend, however, toward lower abundance ratios for the classedas anomalous);the other two IC members,Etosha 4and refractory elementsand Ga and Ge in IB [Scott and Bild, Murnpeowie,were analyzedby Scott and Wasson[1975]. 1974]. Aside from group IIICD, which showsother similarities Table 2 showsthat theseirons have fairly narrowGa (49-55 to IAB, and the Ni-rich side of multiple groups IIAB and ppm) and Ge ranges(212-247 ppm), if the low Ge valuesof IIIAB, group IAB is the only group to show negativeGa-Ni two IC-An membersare ignored.The Ir-Ni plot (Figure 3) and Ge-Ni correlations(Figures 1 and 2). The ratios of the revealsa muchwider range of Ir contents.If two high-Ir IC- highestto the lowestcontents of Ga and Ge in lAB are larger An ironsare neglected,Ni and Ir are negativelycorrelated, as than thosein other groups(except IIICD) by factorsof at least in nearly all other groups.This correlationand that between 5 and 10,respectively. On theIr-Ni plot(Figure 3), groupIAB Ni and Ga, whichis alsonegative, are significantat the 80% also showsunusual behavior; group IA showsno correlation level.On Ga-Ni andGe-Ni plots(Figures 1 and2) IC liesclose with Ni, while lB showsa negativecorrelation significantat to and beneathgroup IA, partlyoverlapping group lib on the 536 SCOTTAND WASSON'CLASSIFICATION AND PROPERTIESOf • IRON METEORITFS

Ga-Ni plot. GroUp IC seemsto be distantly related to IA, gregatesof grains embeddedin daubr6elite.We have also sincenearly all IC membersContain abundant cohenite(which observedsilica in Sierra Gorda (IIA). in Nocoleche has decayedto graphite) but silicatesare absent. GroupIIC. The membershipof this group of plessiticocta- The scarcity of large phosphidesand the trends in IIAB in- hedrites with 9.3-11.5% Ni [Wasson, 1969] was increased to dicate that IC is not related to group liB, Instrumental seven by the addition of Crath6us (1950) [Scott and Wasson, neutronactivation analyses for Au (Figure 4a) and As (E. R. D. 1975]. Becauseof the very high cooling rates, about 300øK Scott, unpublished data, 1975) show lower concentrations My -x (Figure 6), the oriented kamacite spindles have not than are found in group IA. grown large enough to form a continuousnetwork but remain Structurally, group I C shows great diversity, with Arispe largelyisolated. Schreibersite isubiquitous in thematrix, while and Bendeg6being slowly cooled coarse octahedrites, while the highest-Ni member, Wiley, also contains the phosphate Chihuahua City and Santa Rosa are polycrystallineand show sarscopside/graftonite[Buchwald, 1971a; Bild, 1974]. Neither faint tracesof an octahedralstructure reflecting rapid cooling. carbides nor silicates have been observed in this group. Group IIAB. Groups IIA and IIB were definedby Wasson Buchwald [1971a] has describedthe mineralogy of two IIC [1969]. Although we are fairly sure they form a singlefrac- irons,Wiley and Baliinoo. tionation sequence,group IIAB, it is useful to have separate Chemically,group IiC formsa fairly homogeneousgroup. labelsfor the heavily populated IIA group of hexahedritesand Like othergroups (excluding groups IAB and IilCD and the recrystallizedhexahedrites (the erstwhileNi-poor ataxites)and Ni-rich side of double groups IIAB and IIIAI3), Ga and Ge the far lessnumerous coarsest octahedrites comprising group show positive correlations with Ni, significantat the 90 and liB. The appendix lists 39 IIA and 13 IIB members;of these, 99% levels,respe•ctively. Iridium and Ni showa weak negative 21 arid9, respectively,were classified by Wasson[1969] and the correlation, significantat the 80% level. Although there are few remainderby Scott and Wasson[1975]. well-analyzedmembers, it is probablethat fractionatio• trends The hexahedrite-coarsestoctahedrite division is not always are similarto thosein the major groups(excluding IAB). recognizableunless large sectionsare available.Iridium, Group HD. Scott and Wasson [1975] list 13 members o[ however,which has a strongnegative correlation with Ni in this group, doubling its original size [Wasson,1969]. With the both groups(Figure 3), showsa relativelysmall hiatus from exceptionof Arltunga, IID membersare medium or fine octa- 0.5 to 1.8 ppm betweenthe groups.Gallium and Ge exhibit hedrites with kamacite bandwidths of 0.4-0.8 mm and low much narrowerranges in group IIA than in IIB (Table 1) and cooling rates of lø-2øK My -• (Figure 6). Schreibersiteis very are negativelycorrelated with Ni in the latter. In group IIA, common, its abundance increasing with increasing bulk Ni however,Ni and Ge showa weak positivecorrelation, which is content. Troilite generally seemsto be less abundant than in significantat the 80% level.A similarpositive correlation for group IIIAB, and centimeter-sizednodules are rare. (This is Ni and Ga is not found but could easily be masked by ex- the best means for visually distinguishing IIIB and lid perimentalerrors. Such a reversalof correlationsis foundin memberswhen large sectionsare available.) Carbo, which con- group IIIAB (and perhapsalso in IYA). Despiteits small tains abundant haxonite in the piessite fields, is the only rangeof Ni contents,correlations of otherelements with Ni member known to contain carbides, but the presence of are frequentlyobserved [Scott, 1972]; in all casesthey are graphitein Elbogenand the widespreadoccurrence of brown similar to those in IIIAB and IYA, although gradients are martensiticpiessite, occasionally spheroidizing, suggest that C often steeper. is fairly abundant in group IID. Marvin [1962] reported Most membersof groupIIA are classicalsingle-crystal hex- cristobaliteas a devitrificationproduct from a 0.5-mmgreen ahedrites with abundant Neumann bands and euhedral rhab- glass droplet in a Carbo troilite nodule. Magnesium and AI dites, while the remainder are cosmically reheated and were not detectedin the glass(<0.1%). Interestingly,Widman- recrystallizedlike Mejillones(1905) [Wasson,1974, p. 161]. st/itten'sfirst etching experimentsin 1808 were performed on Phosphideabundances increasi• with increasingNi in IIA and two IID irons, Hraschina and Elbogen [Mehl, 1965]. This more noticeablyin IIB, so that centimeter-sizedhieroglyphic groupdisplays long slenderkamacite lamellae, which make at- schreibersitesare observedat the high-Ni end of the IIAB se- tractive macrosections[Wasson, 1969, Figure 6]. quence[Wasson, 1969, Figure 4]. Sulfidenodules are fairly Arltunga was classedIID-An by Scott and Wasson[1975]. rar6 and iaever more than a few millimeters in diameter. In IIA Although its bandwidth of 0.005 mm is a factor of 100 lower nodules,daubr6elite is ubiquitous.The carbidescohenite and than the IID average, its trace element content seemsconsis- haxonite(the former at leasttwice as abundant)form very tent in all respectswith a 1ow-Ni IID member. minor precipitateswhich have often decayedto graphiteand On interelementplots the group IID field is generally near 1ow-Ni kamacite. Carlsbergitegrains are fairly rare. Despite that of group IIC (Figures2-4), but the groupsare resolvedby the frequencyof shock-meltedsulfides in IIA irons,shock- the higher Ga and Co contentsand larger bandwidths(Figure hatchedkamacite has not been observed,although at leasttwo 6) of IID. Like IIC, group IID containssignificant negative occurrencesin groupIIB are known.Detailed accounts of the Ni-Ir and positive Ni-Ge correlations,though a Ni-Ga cor- mineralogyof a few IIA ironshave been given by Axonand relation is not observedand might be maskedby experimental Waine[1972] and Bachwald[1967, 1971a]. errors.Fi'actionation [rends for otherelements in grouplid Silicateinclusions have beenreported in Sikhote-Alin(liB- appearto bevery similfir to thosein themajor groups, exclud- An) by Zaslavskayaand Kvasha[1974]. Orthopyroxene (Fsxs) ing IAB [Scott, 1972' unpublisheddata, 1975]. and olivine(Fax9) are intimatelyassociated with chromiteand GroupHE. Scott et al. [1973] recognizeda clusterof iron troilite. That these inclusionsare found in an iron of slightly meteoriteswith narrow rangesof compositionsfalling above anomalouscomposition (its Ga contentappears to be too low) IIIAB on Ga-Ni and Ge-Ni plots, which they designated suggeststhat a moredetailed investigation of theclassification Weekeroo Station type (following the nomenclatureof Bunch of Sikhote-Alin is needed.Frondel and Klein [1965] discovered et al. [1970] and Wasson [1970b]). In addition to the type ureyite(NaCrSi•O6) in two IIA members,Coahuila and Hex member it comprised four others with 5-10% by volume of River Mountains. Both occurrencescomprised 0.5-mm ag- silicates(Colomera, Elga, Kodaikanal, and Netscha•Svo),the SCOTTAND WASSON:CLASSIFICATION AND PROPERTIFSOF IRON METEORITFS 537

last named being much richer in silicates,and two apparently Wasson and Kimberlin gave separatelabels to the two halves lacking silicates(Arlington and Barranca Blanca). Scott and of the IIIAB sequence,so that IIIA showed positive Ni-Ga Wasson [1975] later found no less than five additional irons and Ni-Ge correlations, while in IIIB these trends were which fell in this cluster, all apparently lacking silicates,and reversed(Figures I and 2). Scott et al. [[973] showedthat there labeled it group IIE. was no hiatus between these two parts of the IIIAB sequence Group IIE lacks many of the systematicfeatures of other and arbitrarily made the division at an lr/Ni ratio of 0.02, meteorite groups. Interelement correlations are weak, and correspondingto Ni contentsof about 8.5-9.2%. They noted kamacite bandwidths vary widely from 0.7 to 2 mm and are ,that bandwidths increase with increasing Ni in Ilia and uncorrelatedwith N i content. There are weak negative Ni-Ga decreasein IIIB (Figure 4) but offered no explanation for the and Ni-Ge correlationssignificant at the 80% level and positive apparent coincidenceof Ga, Ge, and bandwidth maxima. A Ga-Ge and negative Ni-Ir correlationssignificant at the 95% histogram of Ni contents in group IIIAB (Figure 8) shows a level. In Figure 3, three lie memberswith 1ow-lr contentswere peak between7.5 and 8.0%, very closeto the 1ow-Ni end of the excludedfrom the outline of this group largely becauseof con- group. With increasingNi the population decreasesfairly uni- gestionin this portion of the plot. Despite the weaknessof cor- formly, so that there is lessthan 20% with Ni contents in the relations, the small ranges of Ga and Ge concentrations (less range 9-10.5%. than 15% about the mean) are comparable to those in other At the 1ow-Niend of IIIA, phosphideabundances are low; groups. schreibersiteis visible only as submillimeter grain boundary Silicates in liE, apart from those in NetschaiSvo,are precipitatesor rhabdites. In hand specimens,only sulfide in- rounded and dr0plike and more Fe-rich than those in group clusionsare visible, centimeter-sizedones being sparselydis- IAB with orthopyroxenecontaining Fs•4_•.5and rare olivine tributed. As the bulk Ni content increases, the abundance of grainsFa•.•_a•. [Bence and Burnett, 1969'Bunch and Olsen,1968; phosphideand sulfidesincreases. Oriented sulfideprecipitates Bunchet al., 1970;Kt;asha et al., 1974; Wasserburget al., 1968]. called Reichenbach lamellae are well-developed in macro- NetschaEvois classedliE-An becauseof the presenceof relict sections with bulk Ni contents above 8%, and centimeter-sized chondrulesand angular silicateswith distinctly different, more nodulesare abundant in group IIIB. Schreibersiteforms large chondritic bulk composition [Buchwald, 1967; Olsen and laths severalcentimeters long, known as Brezina lamellae, in Jarosewich, 1970, 1971]. Other mineralogical differences group IIIB and also coats troilite nodules. On the submil- between lAB and IIE irons include the presenceof glass and limeter scale, phosphides are also very abundant on rutile in liE, while graphite, carbides,and daubr•elite, which kamacite/taenite grain boundariesabove bulk Ni contentsof are all fairly common in lAB, appear to be absent. The 9%. Carbides and graphite are very uncommon in group members of group IIE contain widely varying amounts of IIIAB. Scott [1971a] observedsmall grains of cohenitein pies- troilite; millimeter-sized nodules are abundant in Barranca site fields of a few members at the low-Ni end of IliA. Bianca but appear to be scarce in the other silicate-free Carlsbergite is plentiful in IIIA irons with less than 8% Ni, members. Phosphide concentrations also vary, but large forming micron-sized platelets and grains in kamacite. precipitates(>5 mm in size) seemto be absent.Axon and co- Phosphateshave been observed in several IIIB irons [Olsen workers lAxon, 1968b; Axon and Faulkner, 1970; Axon and and Fredriksson,1966; Olsen and Fuchs, 1967; Bild, 1974] and Smith, 1970a] have describedthe metallic phasesof three lie one Ilia member [Buchwald,1971b], generally associated with members,and a•fourth was studied by Kirot)a and Dyakonot)a troilite nodules. Sphalerite is rare, and silicateshave not been [1972]. Trace elements analyses apart from those listed by reported. A very large proportion of IIIAB irons have shock- Scott and Wasson [1975] are lacking. Analyses of two IIE hatched or recrystallizedkamacite [Jain and Lipschutz, 1969] membersby Sinaleset al. [1967] suggestthat other trace ele- but some(e.g., Thule, Cape York) show only Neumann bands. mentsare also depletedbelow IA levels. Illustrated descriptions of some IIIAB irons are given by The rounded morphology of the IIE silicates,unlike thosein Buchwald [1966, 1971a]. IAB, showsevidence for silicate-ironliquid immiscibility [Was- Chemically, IIIAB shows fractionation trends common to serburget al., 1965; Wasson,1970b], and there is alsoevidence all well-analyzed groups except lAB and IIICD. lridium of mutually immisciblesilicate liquids [Olsenand Jarosewich, (Figure 3), Re, and Os show the widest ranges,over 3 orders of 1970]. There are three IIE irons whose RB-Sr agesand initial magnitude,giving significant negative correlations with Ni. 87Sr/86Srratios have been determined: Colomera [Sanz et al., Ruthenium (Figure 4b), Rh, and Pt have similar negativecor- 1970], WeekerooStation [Wasserburgand Burnett, 1969],and relations with Ni but less steepgradients; P, As, Au (Figure Kodaikanal [Burnett and Wasserburg, 1967]. Only 4a), Sb, Pd, Mo, and Co are positivelycorrelated with Ni, the Kodaikanal, with an age of 3.8 Gy is significantlydifferent gradients decreasingin the order listed [Scott, 1972]. from 4.6 Gy. Most age determinations of lAB silicatesyield GroupIIICD. Groups IIIC and IIID were definedby Was- agesof about 4.6 Gy but only that of Campo del Cielo (4.7 Gy) son and Schaudy [1971] on the basis of data from five has an uncertainty interval as small as +0.1 Gy. meteorites in each group. The former contained fine octa- GroupIll,lB. This is the largestgroup of irons,accounting hedrites with Ni contents between 11 and 13% and the latter for 32 q- 3% of the independentanalyzed irons (Ilia for 25% finest octahedrites and ataxites with 16-23% Ni. Two more and IIIB for 7%). Its membersinclude Cape York, one of the IIIC members, Hassi-Jekna and Magnesia, were later iden- two largestrecovereft' irons, and four crater-formingirons, tified by Scott et al. [1973] and Scott and Wasson[1975]. Was- Boxhole, Henbury, Wabar, and Wolf Creek. After Wassonand son and Schaudy [1971] thought it more likely than not that Kimberlin [1967] defined this group, Scott et al. [1973] IIIC and IIID were related, and there is now somewhat quadrupledthe membershipand, gave a completelist of Ni, strongerevidence from other trace element data [Scott, 1972] Ga, Ge, and Ir data for all 140 members. Scott and Wasson that thesegroups form a singlesequence, IIICD. As in group [1975] have since analyzed another 14 IIIAB members. All IAB, we use the letters separatelyto distinguishthe high- and IIIAB irons are medium octahedrites, and about 75% of all 1ow-Ni parts of the sequence. mediumoctahed•ites are membersof groupIIIAB (Figure5). Scott and Bild [1974] reviewedthe mineralogyand chemistry 538 SCOTT AND WASSON: CLASSIFICATIONAND PROPERTIESOF IRON METEORITES of group IIICD and found its propertiesfairly similar to those plessitealteration. Unlike group IIIAB, in which the great ma- of group lAB but different from the other major groups.Thus jority of irons show shock-hatchedkamacite, only Cooper- on bandwidth-Ni (Figure 6) or interelementplots (excluding town displaysthis effectin IIIE. The eighthgroup memberto Ga, Ge, and refractories)it is not possibleto distinguishtrends be discovered,Paneth's Iron, has been describedin detail by in IB from those in llICD (e.g., Figure 4a). However, in Buchwaldet al. [1974] and analyzed by Scott and Wasson view of the paucity of well-analyzedmembers, additional data [1975]. Our Ni, Ga, Ge, and Ir data do not show significant would be useful.On plots of Ga, Ge, and refractoriesagainst correlationsin group IIIE, perhapsbecause of the smallranges Ni (Figures 1-3) the steepergradients in group IIICD can be observed for the first three elements. extrapolateda short distanceto convergewith the IAB trends GroupIIIF. Schaudyet al. [1972] describedfour irons con- around 10% Ni. It is possiblethat future discoveriesof IIICD taining 6.8-7.8% Ni (Clark County, Nelson County, M oonbi, membersmight extendthe sequenceinto the IA composition andSt. Genevieve County) which show a narrowrange of Ga, range. Elemental abundancesin group IIICD are close to Ge, and Ir contents.A fifth meteorite, Oakley (iron), with a those in Cl chondrites when both are normalized to Ni, with very similar compositionwas later identifiedand brought the the exception of Ga, Ge, and refractory elementslike Ir and membership to the minimum number necessaryfor group Ru, which have very low abundanceratios (Figures3 and 4). status [Scott and Wasson, 1975]. Schaudy et al. noted that Gallium and Ge are negativelycorrelated with Ni, and all Clark County and Nelson County showedbandwidths vary-

three•:lements show large variations comparable only to those ing. from 1 to 10 mm, while Moonbi and St. GenevieveCounty in group lAB. Thus in group IIICD the concentrationsat op- had bandwidthsof 0.5 mm. They suggestedthat bands wider posite ends differ by factors of 20 and 50 for Ga and Ge, than 1 mm grew by impingement and should be neglectedin respectively,compared with 1-2 in most other groups.Iridium coolingrate calculations •[Goldstein and Short, 1967b]; use of and Ni showa negativecorrelation, which is significantat the the smaller bands results in cooling rates of 20øK My -• 95% level, but Figure 3 showsthat in IIICD as in IAB the slope (Figure 6) for thesefour members.Impingement may also ac- is much lessthan in other groups. Data are sparsefor other count for the bandwidth of 1.4 mm in Oakley, leading to an refractory elementsbut seemto follow the Ir pattern. apparentcooling rate of 5øK My -•. Large sectionsof IIIF Group lIICD lacks the abundant troilite-graphite-silicate members [Scott and Wasson, 1975, Figure 5] show few inclusions frequently found in IAB. Dayton is the only macroscopicinclusions; millimeter-sized troilite nodulesand member known to contain silicates; Scott and Bild [1974] schreibersiteinclusions are infrequent. Microscopic dau- measuredorthopyroxene and plagioclasecompositions of Fs•2 br6elite precipitatesare usually abundant and carbides, sili- and Ab96.The pyroxene Fe content lies betweenIAB and lie cates,and graphiteunrecorded. ranges,while the plagioclaseis more albitic than either group. Althoughthis is a smallgroup, with the exceptionof Oakley The sulfides troilite and daubr6elite are much less common its members have been fairly well-analyzed [Bauer and here than in group IAB. However, IIICD shareswith IAB a Schaudy, 1970; Clarke and Jarosewich, 1972; Cobb, 1967; high abundanceof carbidesand graphite. In group IIIC, hax- Crocker, 1972; Goldberget al., 1951; Herr et al., 1961;Lewis onite characteristicallyprecipitates in plessitefields, spreading and Moore, 1971; Lovering et al., 1957; Moore et al., 1969; over the kamacite, enclosing micron-sized metal and Srnaleset al., 1967]. Group IIIF is characterizedby low Co phosphide grains. We have observed them in all IIIC irons contents(it includesthe lowestin any iron), highCr, and fairly apart from Havana, an AmericanIndian artifact of which only low P. Only Ni and Ir show a significantcorrelation, which is 2 cm2 of polishedsurface was available for study.Other signs negativeas in other groups and significantat the 95% con- of high C concentrationsinclude the presenceof spheroidized fidence level. However, the few available data are consistent and pearlitic piessite,which occasionallyforms in placeof the with the trends visible within the major groups (excluding usual black variety. In group IIID, carbidesand/or graphite lAB), viz., Ni correlated negatively with Cr, Os, Pt, and Re have been observedin all but Fi511inge.Schreibersite is fairly and positively with As, Au (Figure 4a), Co, Mo, Pd, and Sb. abundant, but as in group IAB, there is no obviousincrease in GroupIVA. This is the fourth largestgroup of irons after abundancewith increasingbulk Ni content.The slowlycooled lllAB, lAB, and IIAB, and it containsvirtually all fine octa- octahedritesin this group are among the most handsome hedriteswith lassthan 10% Ni, the largestand bestknown be- specimensafter polishingand etchingand frequentlyappear in ing Gibeon. Wasson[1967] originally definedthis group with museumsand books [e.g., Wood, 1968]. 12 members;then Schaudyet al. [1972] presenteddata on 37 Group HIE. This small group of eight meteorites with members,including the original 12, and discussedthe origin of 8-9% Ni sitslike a blisteron the bend in group IIIAB on plots the group. Scott and Wasson [1975] analyzed three other of Ga, Ge, and bandwidth againstNi (Figures 1, 2, and 6). members. Other publisheddata (e.g., Figure 3) do not distinguishgroup Takenas a whole,group IVA shows• •igr•if•9antincrease in lllE members from those of IIIAB. However, Scott et al. bandwidthwith increasingNi content(Figu?e 6). However, [1973] definedthis clusterof irons as a separategroup because Schaudyet al. [1972] interpretedBuchwald's measurements to of the following: (I) Ga and Ge contentsare lower than those indicatetwo plateaus,one around 0.29 ñ •.04 mmfor low-Ni of lllAB irons with similar Ni contents,(2) bandwidthsare membersof the group and anotherat 0.37 4- 0.04 mm for high- uniformly wider than those in IIIAB, and (3) plessitefields Ni members,and suggestedthat the two subgroupsresided in containabundant haxonite (or graphitefrom carbidedecay), different parent bodies.Further examinationof the data does which is absent in llIAB. Schreibersiteabundances seem com- notsupport the evidence for coolingrate plateaus. In fact,the parable to those in lIIAB meteoriteswith similar Ni contents, bandwidth data cluster at 0.30 and 0.35 mm; 42% of and carlsbergite (CrN) is found in a few IIIE members. In Buchwald's measurements fall at these two values. Consider- those file memberswhich contain graphite filamentsin their ing that the 68% error limits are about 4-0.03 mm, an alter- plessitefields from carbidedecay, Kokstad and Willow Creek, native suggestionis that subjectivebias producedthe step there are other signsof cosmicreheating, e.g., shock-melted function observedon a bandwidth versusNi plot and that sulfides,schreibersite haloes, kamacite recrystallization,and there is no needfor more than one parent body. Goldsteinand SCOTTAND WASSON'CLASSIFICATION AND PROPERTIESOF IRON METEORITES 539

Short [1967b] reported cooling rate estimatesfor 26 IVA irons centrations of the more volatile elements Ga, Ge, Cu, Au, Sb, and suggestedthat there might be a hiatus near 25ø-50øK and As, the suggestionbeing that the metal became isolated My -•. But contrary to the suggestionof Goldsteinand Axon from the cooling solar nebula at high temperatures [Scott, [1973] there is not a very good correspondencebetween the 1972;Kelley and Larimer, 1974]. Elemental abundancesrange Goldstein and Short 'cooling rate' subgroupsand the Schaudy from 0.02 to 0.0001 of the concentrations in CI chondrites et al. 'bandwidth' subgroups(four low cooling rate members when both are normalized to Ni. Despite the narrow range of appear in the low-bandwidth subgroup). There is a small Ni contents, there are significant positive correlations of Ni hiatusin Ni concentrationsin group IVA [Schaudyet al., 1972; with Ga, Ge, and Au and a negative Ni-lr correlation. These Scott and Wasson, 1975] between 8.4 and 8.8% with a cor- data and the variation in phosphideabundances suggest that responding gap in Ir contents between 1.2 and 1.5 ppm. fractionations within this group are very similar to those However, all chemical trends on one side of the gap match ex- observedin the major groups (excluding IAB). trapolations from the other. It is therefore more likely that we Anomalousirons. Currently some67 irons comprising14% have an incomplete seriesof samplesfrom a single IV A se- of those analyzed do not fit into any of the above groups. Of quence rather than two independentsubgroups. the ten largest known meteorites, two are anomalous irons, The fast cooling rates in group IVA are also reflectedin the Mbosi and Bacubirito. In Figure 9 the anomalous irons are morphologyof the piessitefields [Goldsteinand Short, 1967b; plotted on graphsof Ge againstGa and Ni and listedin order Schaudy et al., 1972]. Below bulk Ni contents of •8.5%, of decreasingGe content by Wasson[1974, p. 307]. Nearly taenite borders are quite narrow, with micron-sized taenite 30% are genetically related to one or two other anomalous grains distributed throughout plessite fields. A number of irons; five doublets and three triplets were identified by Was- irons in this group showshock-hatched or cosmicallyreheated son and Schaudy [1971], Schaudy et al. [1972], Scott et al. and recrystallizedkamacite [Jain and Lipschutz,1971; Schaudy [1973], and Scott and Wasson[1975]. These probably repre- et al., 1972]. Group IVA membershave low P contentswhich sent incompletesampling from another eight groups.A few of increasewith increasingNi contents.For a given Ni content,a the remaindermight be reprocessedmembers of other groups, IIIAB member will have nearly 4 times the P content of a IVA but it seemsprobable that we have samplesfrom at least 10 iron [Scott, 1972]. Schreibersiteis correspondinglyscarce or more groups. Thus a large number of parent bodies appear absentat the 1ow-Ni end of the group. Carbides,graphite, and necessary to account for the total population of iron carlsbergiteappear to be entirely absent. Daubr6elite is plen- meteorites. Figure 9 shows that the Ge/Ga ratio of the tiful, and in low-Ni membersit is the most common accessory anomalous irons tends to decreasewith decreasingGe, as it mineral after troilite. Silicatescomprise about 50% of Stein- doesfor the groups.The ratio variesfrom about 0.05 in group bach (IVA-An), chiefly bronzite Fsx5plus tridymite [Schaudy IVA and similarly low-Ge anomalousirons, to slightly above et al., 1972;Reid et al., 1974].As noted by Schaudyet al., trace the CI value of 3.2 in group IA, to over 15 in a few anomalous amountsof tridymite are reported in two other IVA irons, Gi- irons. Many of the anomalous irons are also anomalous in beon and BishopCanyon; in the remainingIVA irons, silicates terms of their mineralogy and structure. appear to be absent. Gibeon has been describedin detail by We know of only two anomalous irons which contain Axon and Smith [1970b]. silicates,Tucson [Bunchand Fuchs,1969] and Kendall County There are significantpositive Ga-Ni and Ge-Ni and negative [Bunchet al., 1970]. The anomalousmeteorite Mount Egerton Ir-Ni correlations in group IVA (Figures 1-3). The positive might also be listed, although it contains much more silicate correlationof Ga and Ge with Ni appearsto disappeararound 9% Ni, and a negative correlation like that observed in the high-Ni portions of IIAB and IIIAB is indicatedif the IVA-An IOOO irons Chinautla and Duel Hill (1854) are included. These two irons also produce a flatteningof the bandwidth-Ni trend like that in IliA. Chemical trends within groups IVA and IIIAB ioo are very similar, the largest difference being the smaller negativeslopes in IVA plots of Ir (Figure 3), Os, Pt, Re, and Ru (Figure 4b) against Ni. Clearly, these groups must have had similar intragroup fractionation histories. Group IVA may be distinguishedby its Co, Ga, Ge, P, and Sb contents, which are lower than those in group IIIAB [Scott, 1972]. GroupIVB. The 11 ataxites comprisinggroup IVB, which include Hoba, the largestindividual meteorite, have a narrow range of Ni contents(16-18%) and are composedof kamacite o.I spindlesand grains in a matrix of fine black plessite[Schaudy et al., 1972]. The high-Ni members contain more abundant kamacite and associatedtiny schreibersiteinclusions [see Per- O.Ol ry, 1944;Axon and Smith, 1972], the latter mineral being vir- o.iI I I. I I I IOI I I iod 5 I0 20 40 tually absent at the 1ow-Ni end of the group. In hand speci- Golliurn(ppm) Nickel (%) ments,inclusions of sulfideor phosphidelarger than a few mil- Fig. 9. Logarithmic plots of Ge against Ga and Ni for 67 limeters in size are very scarce. Instead, oriented bands of anomalousirons with the groupsshown only in outline.On both plots sheenare often found, although this feature was also noted by the CI ratio of elementsis shownby the straightdashed line. Although Axon and Smith in several anomalous ataxites. Carbides, Ga and Ge are strongly correlated in both groups and anomalous irons, the Ge/Ga ratio tendsto decreasewith decreasingGe content. graphite, silicates,and carlsbergitehave not been observedin About 30% of the anomalousirons clusterinto doubletsand triplets. this group. At least20 additional groupsseem to be requiredto accountfor all the Chemically, group IVB is distinguishedby very low con- anomalous irons. 540 SCOTTAND WASSON:CLASSIFICATION AND PROPERTIESOF IRON METEORITES than metal [Cleverly, 1968]. All have silicateswith much lower scenario,one can imagine numerous wa•s to produceuncor- Fe contents(Fsmesosiderites.In contrast, studies are not satisfied.Groups IC and lie are casesin point. Our in which the measuredproperties are comparedwith structural policy in such cases has either been to demand a higher classalone, without regardto group membership,seldom yield minimum population (liE) or the measurementof additional meaningfultrends. Most propertiesof irons can only be under- parameters(As, Au in IC) before designatingsuch a clusterto stood by using the genetic classificationdescribed above. We be a group. In these groups the strength of the genetic bond also urge the reader contemplating a modest study of iron betweenany two arbitrary membersof the group is surelyless meteorite propertiesto considerthe epiloguein the paper by than amongmembers of the typicalgroups. It seemslikely that Scott et al. [1973]. We believe that it is more rewarding to membersof a group formed within a relativelysmall region of make a thorough study of a large typical group suchas IIIAB the solar system. If cooling rates are constant or show a than to analyze a random set of irons in which no more than 3 systematictrend within the group, one can further infer that membersof any one group are included.These results may be the group originated in a single parent body. Even with this contrastedwith data on the nonigneousgroup lAB meteorites. SCOTTAND WASSON:CLASSIFICATION AND PROPERTIESOF IRON METEORITES 541

APPENDIX APPENDIX TABLE 1. (continued)

We list in alphabeticalorder the iron meteoriteswhich have Refer- been classified with references for their classification. Italicized Meteorite Group ences* meteoritesare believedto be piecesof other irons, which are lid identified in the footnotes. Bridgewater 10 Briggsdale Ilia 6 Bristol IVA 4 lid APPENDIX TABLE 1. AlphabeticalListing of Iron Meteorite Brownfield (1966) 10 Bruno IIA 7 Classifications Burgavli IA 8 Refer- Burkett IA 8 Meteorite Group ences* Burlington IIIE? 6 Bushman Land IVA 4 Abakan'•(36) Ilia 10 Butler Anom 8 Abancay(25) IIIF 4,10 Cacaria Ilia 6 Adargas(8) IIIB 6 Cachiyuyal (Om) Anom 6 Aggie Creek IIIA 6 Calico Rock IIA 7 Ainsworth• !IB 7 ff"•,.-, I • ,e• ' iA i0 Akpohon(5) IIIA 1 Cambria Anom 9 Algoma Anom 6 Campbellsville IIIB 6 Alikatnima Anom 3 Campo del Cielo IA 8 Altonah IVA 4 Camp Verde(4) IA 8 Amates (34) IA 10 Canton Ilia 6 Angelica IIIA 6 Canyon City Ilia 6 Angra des Reis (iron) IIA 10 Canyon Diablo• IA 8 Annaheim IA-An 8 Canyon Diablo (1936) (4) IA-An 8 Anoka IIIC 9 CanyonDiablo (1949) (4) IA-An 8 Apoala IIIB 6 Cape of Good Hope IVB 4 Apoalapseudo (11) IIIA 6 Caperr Ilia 6 Aprelsky IIIB 10 Cape York$ Ilia 6 Aragon (6) IIA 1 Carbo lid 7 Arispe IC-An 8, 10 Carlton IIIC 9 Arlington IIE 6,10 Carthage Ilia 6 Arltunga liD-An 10 Casas Grandes Ilia 6 Asarco Mexicana IIIA 6 Casey County IA 8 Ashfork (4) IA 8 Casimiro de Abreu Ilia 6 Aswan IIIA 6 Cedartown$ IIA 10 Auburn(35)? Anom 6 Central Missouri ( 1) lib 7 Augusta County IliA-An 6 Chambord Ilia 6 Augustinovka$ IIIB 6 Chaharal(16)? Ilia 6 Avfie IIA 10 Charcas$ Ilia 6 Avoca (West. Australia) Ilia 10 Charlotte IVA 4 Babb's Mill (Blake's) Anom 4 Chebankol Anom 6 Babb's Mill (Troost's) Anom 6 Chesterville IIA 10 Bacubirito Anom 6 Chico Mountains IIA 10 Bagdad Ilia 6 Chihuahua City IC 8,10 Bahjoi IA-An 8 Chilkoot Ilia 6 Bald Eagle IIIB 6 Chinautla IVA-An 4 Balfour Downs IA 8 Chinga Anom 4,5 Ballinger IA-An 8 Chulafinnee Ilia 10 Ballinoo IIC 7 Chupaderos$ IIIB 6 Baquedano IIIB 6 Clark County IIIF 4,10 Barraba (3) IIA 1 Cleveland IIIB 6 Barranca Blanca lie 6, 10 Coahuila$ IIA 7 Bartlett Ilia 6 Colfax lB 8 BasedowRange (15) Ilia 6 Colomera lie 6,10 Bear Creek IIIB 6 Comanche (iron) IA 8 Bella Roca IIIB 6 Cookeville IA 8 Bellsbank Anom 6 Coolac IA 10 Bendeg6 IC 8,10 Coopertown IIIE 6 BennettCounty IIA 7 Copiapo IA 8 Billings Ilia 6 Corowa Anom 7 Bingera$ IIA 7 Corrizatillo IA 10 Bischt•be IA 8 Cosby's Creek IA 8 BishopCanyon IVA 4 Costilia Peak Ilia 6 Bitburg lB 10 Cowell Ilia 3 Black Mountain IA 10 Cowra Anom 9 BloodyBasin (4) IA 8 Coya Norte (33) IIA 7 Bodaibo IVA 4 Cranbourne IA 8 Bogou IA 8 CratheOs(1931) IVA 4 Boguslavka IIA 7 CratheOs(1950) IIC 10 Bohumilitz IA 8 Cruz del Aire Anom 10 Bolivia IA 8 Cuernavaca (8) IIIB 6 Boogaldi IVA 4 Cumpas Ilia 6 Boxhole Ilia 6 Dalton Ilia 6 Braunau IIA 7 Davis Mountains Ilia 6 Breece (14) IIIB 6 Dayton IIID 9 542 SCOTT AND WASSON: CLASSIFICATION AND PROPERTIES OF IRON METEORITFS

APPENDIX TABLE 1. (continued) APPENDIX TABLE 1. (continued)

Refer- Refer- Meteorite G roup ences* M eteorite Group ences*

Deelfontein IA 8 Horse Creek mnom 10 Deep Springs Anom 4 Houck (4) IA 8 De H oek Anom 10 Hraschina lid 10 Delegate IIIB-An 6 Huizopa IVA 4 Del Rio Anom 10 Idaho IA 10 Denton County Ilia 6 Ider IIIA 6 Deport IA 8 Ilimaes (iron):[: IIIA 6 Dermbach Anom 10 Ilinskaya Stanitza IIIA 6 Descubridora(7) Ilia 6 Illinois Gulch Anom 9 Dexter Ilia 6 Indian Valley IIA 7 Dimitrovgrad Ilia 6 Iquique IVB 4 Dorofeevka Anom 10 Iredell IIB 10 Drum Mountains Ilia 6 Iron Creek IIIA 6 Duchesne:[: IVA 4 Iron River IVA 4 Duel Hill (1854) IVA-An 4 Ivanpah Ilia 6 Duketon Ilia 6 Jamestown IVA 4 Dungannon IA 8 Jenkins IA 8 Durango:l: Ilia 6 Jenny'sCreek IA 8 Edmonton (Canada) IIA 7 Joel'sIron (28) ? Ilia 6 Edmonton (Kentucky) IIIC 6, 9 Joe Wright Mountain IIIB 6 Ehrenberg(4) IA 8 Juncal Ilia 6 Elberton(29) IIA 1 Juromenha Ilia 6 Elbogen lid 10 Kalkaska Ilia 6 El Burro lib 7 Karee Kloof IA-An 8 El Capitan IIIB 6 Kayakent Ilia 6 Elga lie 6, 10 Keen Mountain IIA 10 Ellicott IA 10 Kendall County Anom 10 El Qoseir Anom 9 Kenton County$ Ilia 6 Elton Anom 8 Kingston Anom 6 Elyria Ilia 6 Knowles IIIB 6 Emsland Anom 6 Kodaikanal lie 6,10 Etosha IC 10 Kofa Anom 9 Fair Oaks (4) IA 8 Kokomo IVB 4 Filomena (33 ) IIA 7 Kokstad$ IIIE 6 Fi511inge IIID 9 Kopjes Vlei IIA 7 Forsyth County IIA 10 Kouga Mountains IIIB 6 FossilSprings (4) IA 10 Kumerina IIC 7 Four Corners lB 8 Kyancutta Ilia 6 Franceville Ilia 6 La Caille Anom 9 Frankfort (iron) Ilia 6 La Grange IVA 4 Freda IIID 9 Lake Murray lib 7 Galleguillos(32) IVB 4 Landes IA 10 Garden Head Anom 9 Lanton Ilia 6 Garhi Yasin lie 10 La Porte Ilia 6 Gay Gulch Anom 9 La Primitiva:[: Anom 6 Gibeon:l: IV A 4 Las Vegas(4) IA 10 Gladstone (iron):[: IA 8 Laurens Co unty Anom 9 Glasgow Ilia 6 Leeds IA 8 Glenormiston Anom 6 L•nb, rt6 Ilia 6 Goose Lake IA-An 8 LexingtonCounty IA 10 Grand Rapids Anom 9 LexingtonCounty pseudo Ilia 6 Grant:]: IIIB 6 Lime Creek Anom 6 Greenbrier County Ilia 6 Linwood IA 8 Gressk IIA 8 Livingston(Montana) Ilia 6 Guffey Anom 4 Livingston(Tennessee) Anom 10 Gun Creek Anom 6 Locust Grove IIA 10 Gundaring Ilia 6 Lombard IIA 7 Haig Ilia 6 Lonaconing lie 10 Hammond Anom 6 LoonganaStation (21) Anom 2 Haniet-el-Beguel IA 10 Loreto Ilia 6 Harriman (Of) IVA 4 Los Reyes IIIB 6 Harriman (Om) Ilia 6 Lucky Hill Ilia 10 Hassi-Jekna IIIC 6 Luis Lopez IIIB 6 Havana IIIC 9 Madoc Ilia 6 Hayden Creek Ilia 10 Magnesia IIIC 10 Helt Township(4) IA 10 Magura IA 8 Henbury:l: Ilia 6 Maidyak Ilia 10 Hex River Mountains IIA 7 Mantos Blancos IVA 4 Hill City IVA 4 Mapleton Ilia 6 Hoba IVB 4 Maria Elena IVA 4 Hollands Store IIA 10 Marshall County Ilia 6 Hope IA 8 Mart IVA 4 Hopper(30) IIIB 1 Matatiele (18) IIIE 6 SCOTTAND WASSON: CLASSIFICATION AND PROPERTIES ()!• IRONMt-TF()RVIES 543

APPENDIX TABLE 1. (continued) APPENDIX TABLE 1. (continued)

Refer- Refer- Meteorite G ro up ences* M eteorite Group ences*

Mayerthorpe IA 8 Oscuro Mountains IA 10 Mayodan IIA 7 Osseo IA 8 Mazapil IA 8 Otchinjau IVA 4 Mbozi Anom 6 OwensValley IIIB 6 McCamey (23) IA 8 Pan de Azucar IA 10 Mejillones (1905) IIA 7 Paneth's Iron IIIE 10 Merceditas Ilia 6 Para de Minas IVA 4 Mertzon IA-An 8 Parral (20) Ilia 1 Mesa Verde Park lB 10 Perryville IIC 7 Michigan Iron (34) IA 10 Persimmon Creek IB 6 Milly Milly Ilia 6 Picacho Ilia 6 Misteca IA 8 Piedadedo Bagre An o m 6 Moab (4) IA 10 Pima County IIA 10 Mocte7um a !A ! 0 Pine River IA 8 Monahans Anom 7 Pition Anom 9 Monument Rock (4) IA 8 Pitts IB 8 Moonbi IIIF 4,10 Pittsburg IA 10 Mooranoppin(38) IA 10 Plymouth IIIA 6 Moorumbunna Ilia 6 Point of Rocks(iron) Ilia 6 Morden IA 3 Ponca Creek (1) lib 7 Morito:]: Ilia 6 Premier Downs(21 ) Anom 2 Morradal Anom 7 Providence Ilia 6 Morrill IA-An 8 Puente del Zacate IIIA 6 Mount Ayliff IA 8 Pulaski County(4) IA 1 Mount Dooling IC 8, 10 Puquios lid 7 Mount Edith IIIB 6 Purlpica(33) IIA 7 Mount Joy lib 7 PutnamCounty IVA 4 Mount Magnet Anom 9 Quairading(38) IA 1 Mount Ouray lid 10 Quartz Mountain Ilia 10 Mount Sir Charles An om 3 Queensland(13) IA 8 Mount Stirling (38) IA 10 Quillagua(33) IIA 7 Mount Tabby (1O) IVA I Quinn Canyon:l: Ilia 6 Mundrabilla:l: IA-An 10 Rafrtiti Anom 4 Mungindi IIIC 9 Railway (12) IVA 4 Muonionalusta IVA 4 Rancho de la Pila Ilia 6 Murfreesboro Anom 6 Rateldraai Ilia 6 Murnpeowie IC-An 10 Redfields Anom 10 Murphy IIA 10 Red River Ilia 6 Nagy-Vfi.zsony IA 10 Reed City Anom 6 Narraburra:]: IIIB 6 Rembang IVA 10 Navajo lib 10 Repeev Khutor Anom 10 Nazareth (iron) Ilia 10 Rhine Villa IIIE 6 Nedagolla Anom 4 Richa IID 10 Needles lid 7 Richland IIA 10 Negrillos !IA 7 Rifle IA 8 Nejed (37) Ilia 6 Rio Loa (33) IIA 7 Nelson County IIIF 4,10 Rodeo lid 7 Neptune Mountains IA 10 Roebourne Ilia 6 NetschaEvo liE-An 6, 10 Roper River IIIB 6 New Baltimore Anom 6 Rosario IA 10 New Leipzig IA 10 Rowton Ilia 6 New Westville IVA 4 Ruff's Mountain Ilia 6 N'Goureyma Anom 4 Russel Gulch Ilia 6 Nico (12) IVA 1 Sacramento Mountains Ilia 6 Nieder Finow IA 10 St. FrancoisCounty IC 8,10 N'Kandhla lid 7 St. GenevieveCounty:l: IIIF 4,10 Nocoleche IC-An 8, 10 Salt River• IIC 7 Nordheim Anom 4 Samelia Ilia 10 Norfolk IIIA 6 SamsValley I!IB 6 Norfork Ilia 6 San Angelo Ilia 6 Norristown IIIB 6 SanchezEstate (9) IIA 10 North Portugal (27) lib 1 San Cristobal IB 6 Novorybinskoe IVA 10 Sanderson II1B 6 Nuleri Ilia 10 Sandia Mountains lib 7 Nutwood Downs(15) IIIA 6 Sandtown Ilia 6 Oakley(iron) IIIF 10 San Franciscodel Mezquital IIA 10 Obernkirchen IVA 4 San Francisco Mountains IVA 4 Odessa(iron):]: IA 8 San Martin (33) IIA 7 Ogallala IA 8 Santa Apolonia Ilia 6 Okahandja IIA 10 Santa Catherina Anom 9 Okano IIA 7 Santa Luzia lib 7 Orange River (iron) IIIB 6 Santa Rosa IC 8,10 Oroville IIIB 6 Santiagopapasquiero Anom 4 544 SCOTTAND WASSON:CLASSIFICATION AND PROPERTIESOF IRON METFORITFS

APPENDIX TABLE 1. (continued) APPENDIX TABLE 1. (continued)

Refer- Refer- Meteorite G ro up ences* M eteorite Group ences*

S•o JuliS.o de Moreia:• lib 7 Unter-M•issing IIC 7 Sardis IA 10 Uwet IIA 7 Sarepta IA 8 Uwharrie Ilia 6 Savannah Ilia 6 Vaalbult IA 10 Schwetz IIIA 6 Veliko-Nikolaevsky Priisk Ilia 10 Scottsville IIA 7 Ventura Anom 6 Seel•isgen IA 8 Verkhne Dnieprovsk IIE 10 Seligman IA 8 VerkhneDnieprovsk pseudo (2) IIIB 6 Seneca Falls Ilia 6 ¾erkhne Udinsk Ilia 6 SenecaTownship IVA 4 VictoriaWest An6m 6 Serrania de Varas IVA 4 View Hill Ilia 6 Seymchan lie 10 Wabar•: Ilia 6 Seymour IA 8 Waldron Ridge IA 8 ShingleSprings Anom 4 Walker County IIA 10 Shirahagi' IVA 10 Wallapai lid 7 Shrewsbury IA 8 Warburton Range IVB 4 Sierra Gorda IIA 7 Warialda (3) IIA 1 Sierra Sandon:t: Ilia 6 WashingtonCounty Anom 9 SignalMountain IVA 4 Waterville IA-An 8, 10 Sikhote-Alin liB-An 7 Wathena IIA 10 Silver Bell lib 7 Weaver Mountains IVB 4 Silver Crown IA 10 Wedderburn IIID 9 Skookum IVB 4 Weekeroo Station lie 6, 10 Smithland IVA 4 Welland Ilia 6 Smithonia:• IIA 7 Western Arkansas IVA 4 Smith's Mountain:• IIIB 6 Wichita County IA 8 Smithsonian Iron lib 10 Wickenburg(iron) (4) IA 8 Smithville IA 8 Wiley IIC 7 Social Circle IVA 4 Willamette Ilia 6 Soper Anom 9 Williamstown(17) Ilia 6 Soroti Anom 9 Willow Creek IIIE 6 South Byron Anom 6 Wolf Creek IIIB 6 Southern Arizona IA 8 Wonyulgunna IIIB 6 Spearman Ilia 6 Woodbine IB 8 Ssyromolotovo Ilia 10 Wood's Mountain IVA 4 Staunton IIIE 6 Yanhuitlan IVA 4 Steinbach IVA-An 4 Yardea IA 3 Summit lib 10 Yardymly IA 8 SurpriseSprings IA 8 Yarri Ilia 6 Susuman Ilia 6 Yarroweyah IIA 7 Tacubaya(34) IA 10 Yenberrie IA 10 Tamarugal:t: Ilia '6 Youanmi Ilia 6 Tam bo Quemado IIIB 6 Youndegin:!: IA 10 Tamentit Ilia 6 Ysleta Anom 4 Tanakami Mountain IIIE 6 Zacatecas(1792) Anom 5, 8 Tarapaca(H) (19) Anom 6 Zachtecas(1969) IIIB 6 IIIA 1 Zenda IA-An 8 Tarapaca(Om) (31) :. Tawallah Valley IVB 4 Zerhamra IIIA-An 10 Tazewell IIID 9 Temora(22) IIIB 1 * Key to references:1, Buchwald [1975]; 2, de Laeter [1972]; 3, Ternera• IVB 10 Reed [1972]; 4, Schaudyet al. [1972]; 5, E. R. D. Scott, unpublished Thoreau(23) IA 8 data (1975);6, Scott et al. [1973];7, Wasson[1969]; 8, Wasson Thule Ilia 6 [1970a];9, Wassonand Schaudy[1971]; 10, Scott and Wasson[1975]. Thunda Ilia 6 •' Italicized meteoritesare paired with otherslisted below which can Thurlow IIIB 6 be identified from the adjacent number: 1, Ainsworth; 2, Augusti- Tieraco Creek IIIB 6 novka; 3, Bingera;4, Canyon Diablo; 5, Cape York; 6, Cedartown;7, Tlacotepec IVB 4 Charcas; 8, Chupaderos;9, Coahuila; 10, Duchesne;11, Durango; Tobychan lie 10 12, Gibeon; 13, Gladstone; 14, Grant; 15, Henbury; 16, Ilimaes; •Tocavita(26) ? IIC 7 17, Kenton County; 18, Kokstad;19, La Primitiva;20, Morito; 21, Tocopi!la:]: IIA 7 Mundrabilla; 22, Narraburra; 23, Odessa;24, QUinn Canyon; 25, St. Toluca$• IA 8 Genevieve County; 26, Salt River; 27, Silo Julifio; 28, Sierra Sandon; TombigbeeRiver:]: Anom 6 29, Smithonia; 30, Smith's Mountain; 31, Tamarugal;32, Ternera,; Tonopah(24) Ilia 1 33, Tocopilla; 34, Toluca; 35, Tombigbee River; 36, Toubil River; Toubil River•: Ilia 6 37, Wabar; 38, Y oundegin. Trenton Ilia 6 :• The above meteorites, numbered 1-38, are identified with a T•eysa IIIB-An 6 double dagger to warn the reader that samplesor data from these Tucson Anom 4 irons may be listed under other names in the literature. Turtle River IIIB 6 Twin City Anom 9 Acknowledgments.Our studiesof iron meteoriteswere only possi- Udei Station IA 8 ble through the generoussupport of many curators,especially R. S. Uegit Ilia 10 Clarke, M. H. Hey, R. Hutchison, L. G. Kvasha, C. B. Moore, and E. Union (33) IIA 10 Olsen, who supplied large numbers of specimens,often on short Union Co unty IC 8,10 notice.We acknowledgeour great debt to V. F. Buchwald,our source SCOTTAND WASSON:CLASSIFICATION AND PROPERTIESOF IRON METEORITES 545 of adviceand stimulationfor many years.We also thank R. W. Bild, Cleverly,W. H., Further recoveriesof two impact-fragmentedWestern H. J. Chun, and K. L. Robinsonfor their assistance.This work was Australianmeteorites, Haig and Mount Egerton,J. Roy. Soc. West. largelysupported by NASA grant NGR-05-007-329. Aust., 51, 76-88, 1968. Cobb, J. C., A traceelement study of iron meteorites,J. Geophys.Res., REFERENCES 72, 1329-1341, 1967. Cohen, E., Meteoritenkunde,vol. 3, Classificationund N omenclatur; Axon,H. J., Metallurgyof meteorites,Progr. Mater. Sci., 13, 183-228, K6rnige bis dichte Eisen; Hexai•drite;Oktai•drite mit feinen Lamel- 1968a. Ien, 419 pp., Schweizerbart'scheVerlagsbuchhandlung, Stuttgart, Axon, H. J., The metallographicstructure of the K odaikanal 1905. meteorite,Mineral. Mag., 36, 687-690, 1968b. Crocket, J. H., Some aspectsof the geochemistryof Ru, Os, Ir and Pt Axon, H. J., Preterrestrialdeformation effects in iron meteorites,in in iron meteorites, Geochirn.Cosmochirn. Acta, 36, 517-535, 1972. Meteorite Research,edited by P.M. Millman, pp. 796-805, D. de Laeter, J. R., The M undrabilla meteorite shower, Meteoritics, 7, Reidel, Dordrecht, Netherlands, 1969. 285-294, 1972. Axon, H. J., and D. Faulkner, A metallographicand microprobe Doan, A. S., and J. I. Goldstein, The formation of phosphidesin iron study of the BarrancaBianca meteorite,Mineral. Mag., 37, meteorites,in Meteorite Research,edited by P.M. Millman, pp. 898-904, ! 970. 763-779, D. Reidel, Dordrecht, Netherlands, 1969. Axon, H. J., and P. L. Smith,A metallographicand microprobestudy El Goresy, A., Mineralbestand und Strukturen der Graphit- und of the metal phasesin the WeekerooStation meteorite, Mineral. Sulfideinschltissein Eisenmeteoriten, Geochim. Cosmochim.Acta, Mag., 37, 670-673, 1970a. 29, 1i3i-i151• i965. Axon, H. J., and P. L. Smith,A studyof someiron meteoritesof the Farrington,O. C., Analysesof iron meteoritescompiled and classified, Gibeon association,Mineral. Mag., 37, 888-897, 1970b. Field Mus. Publ. 120, Geol. Ser., 3, 5, 59-110, 1907. Axon, H. J., and P. L. Smith, Metallographicstudy of some iron Fouch6, K. F., and A. A. Smales, The distribution of gold and meteoritesof highnickel content, Mineral. Mag., 38, 736-755, 1972. rhenium in iron meteorites, Chem. Geol., 1, 329-339, 1966. Axon, H. J., and C. V. Waine, A metallographicstudy of somehex- Frick, C., and E. C. I. Hammerbeck,Catalogue of South African and ahedrites,Mineral. Mag., 38, 725-735, 1972. South West African meteorites, Repub. SuM Afr. Geol. Opname Axon, H. J., J. Boustead,and E. D. Yardley,The mechanicaland ther- Bull., 57, 1-47, 1973. mal alteration of iron meteoritestructures, in ShockMetamorphism Frondel, C., and C. Klein, Ureyite, NaCrSi,O6: A new meteoritic of NaturalMaterials, edited by B. M. Frenchand N.M. Short,pp. pyroxene,Science, 149, 742-744, 1965. 585-599, Mono, Baltimore, Md., 1968. Goldberg,E., A. Uchiyama,and H. Brown,The distributionof nickel, Bauer,R., and R. Schaudy,Activation analytical determination of ele- cobalt, gallium, palladium and gold in iron meteorites,Geochim. mentsin meteorites,3, Determinationof manganese,sodium, gal- Cosmochirn.Acta, 2, 1-25, 1951. lium, germanium,copper and gold in 21 iron meteoritesand 2 Goldstein, J. l., The classification of iron meteorites, in Meteorite mesosiderites,Chern. Geol., 6, 119-131, 1970. Research, edited by P.M. Millman, pp. 721-737, D. Reidel, Bence,A. E., and D. S. Burnett, Chemistryand mineralogyof the Dordrecht, Netherlands, 1969. silicates and metal of the Kodaikanal meteorite, Geochim. Goldstein, J. I., and H. J. Axon, The Widmanst•itten figure in iron Cosmochim.Acta, 33, 387-407, 1969. meteorites, Naturwissenschaften,60, 313-321, 1973. Bild, R. W., New occurrencesof phosphatesin iron meteorites, Goldstein, J. I., and J. M. Short, Cooling rates of 27 iron and stony- Contrib. Mineral. Petrol., 45, 91-98, 1974. iron meteorites, Geochim. Cosmochim.Acta, 31, 1001-1023, 1967a. Brezina, A., Die Meteoritensammlungdes k. k. mineralogischen Goldstein, J. I., and J. M. Short, The iron meteorites, their thermal Hofkabinettes in Wien, Jahrb. k. k. Geol. Reichsanst.,35, 151-276, history and parent bodies, Geochim.Cosmochim. Acta, 31, 1733- 1885. 1770, 1967b. Brezina,A., The arrangementof collectionsof meteorites,Proc. Amer. Gooley, R., C. F. Lewis, C. B. Moore, and G. I. Huss, Nazareth (b) Phil. Soc., 43, 211-247, 1904. meteorite, Meteoritics, 6, 93-98, 1971. Buchwald,V. F., The iron-nickel-phosphorussystem and the structure Hara, T., and E. B. Sandell, Meteoritic abundance of ruthenium, of iron meteorites,Acta Polytech.Scand., Chem. Incl. Met. Ser., 51, Geochim. Cosmochim. Acta, 21, 145-150, 1960. 1-46, !966. Henderson,E. P., and S. H. Perry, Studiesof sevensiderites, Proc. Buchwald,V. F., Studiesof six iron meteorites,Analecta Geol., 2, U.S. Nat. Mus., 107, 339-403, 1958. 1-75, 1967. Herr, W., W. Hoffmeister, B. Hirt, J. Geiss, and F. G. Houtermans, Buchwald,V. F., Tritium lossresulting from cosmicannealing, com- Versuch zur Datierung von Eisenmeteoritennach der Rhenium- pared with the microstructureand microhardnessof six iron Osmium-Methode,Z. Naturforsch.,Ser. A, 16, 1053-1058, 1961. meteorites, Chem. Erde, 30, 33-57, 1971a. Heymann, D., The origin of hypersthenechondrites: Ages and shock Buchwald, V. F., A new cutting technique for meteoritic irons, effects of black chondrites, Icarus, 6, 189-221, 1967. Meteoritics, 6, 27-31, 197! b. Jain, A. V., and M. E. Lipschutz,Shock histories of hexahedritesand Buchwald,V. F., Handbookoflron Meteorites,in press,University of Ga-Ge group llI octahedrites,in MeteoriteResearch, edited by P. California Press,Berkeley, Calif., 1975. M. Millman,,pp. 826-837, D. Reidel,Dordrecht, Netherlands, 1969. Buchwald,V. F., and E. R. D. Scott, First nitride (CrN) in iron Jain, A. V., and M. E. Lipschutz,Shock history of iron meteoritesand meteorites,Nature Phys. Sci., 233, 113-114, 1971. their parent bodies:A review, 1967-1971, Chem.Erde, 30, 199-215, Buchwald,V. F., R. Hutchison,and J. M. Hall, Paneth'sIron, a new 1971. groupIIIE iron, Meteoritics,9, 307-311, 1974. Jain,A. V., R. B. Gordon,and M. E. Lipschutz,Hardness of kamacite Bunch,T. E., and L. H. Fuchs,A newmineral brezinaite Cr8S4 and the and shock histories of 119 meteorites, J. Geophys.Res., 77, Tucson meteorite, Arner. Mineral., 54, 1509-1518, 1969. 6940-6954, ! 972. Bunch, T. E., and E. Olsen, Potassiumfeldspar in Weekeroo Sta- Kelley,W. R., andJ. W. Larimer,Trace element composition of metal tion, Kodaikanal, and Colomera iron meteorites, Science, 160, during condensation--Implicationsfor iron meteorites(abstract), 1223-1225, 1968. Meteoritics, 9, 360, 1974. Bunch,T. E., K. Keil,and E. Olsen, Mineralogy andpetrology of Kimberlin, J., C. Charoonratana, and J. T. Wasson,Neutron activa- silicate inclusionsin iron meteorites,Contrib. Mineral. Petrol., 25, tion determination of Ir in meteorites, Radiochim. Acta, 10, 69-75, 297-340, 1970. 1968. Bunch, T. E., K. Keil, and G.I. Huss, The Landesmeteorite, Kirova, O. A., and M. I. Dyakonova, Seymchaniron meteorite(in Meteoritics, 7, 31-38, 1972. Russian), Meteoritika, 31, 104-108, 1972. Burnett,D. S., and G. J. Wasserburg,Evidence for the formationof Kvasha, L. G., J. G. Lavrentev,and N. V. Sobolev,On silicateinclu- an iron meteorite at 3.8 X 10• years, Earth Planet. Sci. Lett., 2, sionsand shockmetamorphism features in the Elga octahedrite(in 137-147, 1967. Russian),Meteoritika, 33, 143-147, 1974. Chou, C. L., P. A. Baedecker,and J. T. Wasson,Allende inclusions: Lewis,C. F., and C. B. Moore, Chemicalanalyses of thirty-eightiron Volatileelement distribution and evidencefor incompletevolatilizao meteorites, Meteoritics, 6, 195-205, 1971. tion of presolarsolids, submitted to Geochirn.Cosrnochirn. Acta, Lipschutz,M. E., Shockeffects in meteorites,in ShockMetamorphism 1975. of Natural Materials,edited by B. M. Frenchand N.M. Short,pp. Clarke, R. S., and E. Jarosewich,Iron meteorite compositions, 571-583, Mono, Baltimore, Md., 1968. Smithson. Contrib. Earth Sci., 9, 65-66, 1972. Lipschutz,-M.E., P. Signer,and E. Anders,Cosmic ray exposureages 546 SCOTTAND WASSON:CLASSIFICATION AND PROPERTIESOF IRON METEORITES

of iron meteoritesby the Ne•/A1:8 method,J. Geophys.Res., 70, Scott, E. R. D., J. T. Wasson, and V. F. Buchwald, The chemical clas- 1473-1489, 1965. sificationof iron meteorites,7, A reinvestigationof irons with Ge Lovering,J. F., W. Nichiporuk, A. Chodos,and H. Brown, The dis- concentrationsbetween 25 and 80 ppm, Geochim.Cosmochim. Acta, tribution of gallium,germanium, cobalt, chromium,and copperin 37, 1957-1983, 1973. iron and stony-iron meteoritesin relation to nickel content and Short, J. M., and J. I. Goldstein,Rapid methodsof determiningcool- structure, Geochim. Cosmochim. Acta, II, 263-278, 1957. ing rates of iron and stony iron meteorites,Science, 156, 59-61, Marvin, U. B., Cristobalite in the Carbo iron meteorite, Nature, 196, 1967. 634-636, 1962. Smales,A. A., D. Mapper, and K. F. Fouch6, The distributionof Mason, B., The Woodbine meteorite, with notes on silicates in iron trace elementsin iron meteorites,as determinedby neutronactiva- meteorites, Mineral. Mag., 36, 120-126, 1967. tion, Geochim. Cosmochim.Acta, 31, 673-720, 1967. Mason, B., The mineralogy of meteorites,Meteoritics, 7, 309-326, Tschermak,G., Die Meteoritendes k. k. mineralogischenMuseum am 1972. I October 1872, Mineral. Petrogr. Mitt., 165-172, 1872. Mehl, R. F., On the WidmanstSttenstructure, in Sorby Centennial Tschermak, G., Beitrag zur Classificationder Meteoriten, Sitzber. Symposiumon History of Metallurgy, edited by C. S. Smith, pp. Akad. Wiss. Wien, Math.-Naturwiss. Kl., Abt. I, 88, 347-371, 1883. 245-269, Gordon and Breach, New York, 1965. Vdovykin, G. P., The Canyon Diablo meteorite,Space Sci. Reo., 14, Moore, C. B., P. J. Birrell, and C. F. Lewis, Variations in the chemical 758-831, 1973. and mineralogicalcomposition of rim and plainsspecimens of the Voshage,H., Bestrahlungsalterund Herkunft der Eisenmeteorite,Z. Ganyon Diablo meteorite, Geochim. Cosmochim. Acta, 31, Naturforsch.,Ser. A, 22, 477-506, 1967. 1885-1892, 1967. Wasserburg,G. J., and D. S. Burnett, Status of isotopicage deter- Moore, C. B., C. F. Lewis, and D. Nava, Superioranalyses of iron minations on irøn and stone meteorites,in Meteorite Research, meteorites,in Meteorite Research,edited by P.M. Millman, pp. edited by P.M. Millman, pp. 467-479, D. Reidel, Dordrecht, 738-748, D. Reidel, Dordrecht, Netherlands, 1969. Netherlands, 1969. Nichiporuk, W., and H. Brown, The distributionof platinum and pal- Wasserburg, G. J., D. S. Burnett, and C. Frondel, Strontium- ladium metalsin iron meteoritesand in the metal phaseof ordinary rubidiumage of an iron meteorite,Science, 150, 1814-1818,1965. chondrites,J. Geophys.Res., 70, 459-470, 1965. Wasserburg,G. J., H. G. Sanz, and A. E. Bence,Potassium-feldspar Nininger, H. H., and A.D. Nininger, The Nininger Collectionof phenocrystsin the surfaceof Colomera,an iron meteorite,Science, Meteorites, p. 139, American Meteorite Museum, Winslow, Ariz., 161, 684-687, 1968. 1950. Wasson, J. T., The chemical classification of iron meteorites, 1, A Olsen, E., and K. Fredriksson, Phosphatesin iron and pallasite studyof iron meteoriteswith low concentrationsof galliumand ger- meteorites, Geochim. Cosmochim.Acta, 30, 459-470, 1966. manium, Geochim. Cosmochim.Acta, 31, 161-180, 1967. Olsen, E., and L. H. Fuchs, The state of oxidation of some iron Wasson, J. T., Concentrationsof nickel, gallium, germanium, and meteorites, Icarus, 6, 242-253, 1967. iridium in Canyon Diablo and other Arizona octahedrites,J. Olsen, E., and E. Jarosewich,The chemicalcomposition of the silicate Geophys.Res., 73, 3207-3211, 1968. inclusions in the Weekeroo Station iron meteorite, Earth Planet. Wasson, J. T., The chemical classificationof iron meteorites,2, Hex- Sci. Lett., 8, 261-266, 1970. ahedritesand other irons with germanium concentrationsbetween Olsen, E., and E. Jarosewich,Chondrules: First appearancein an iron 80 and 200 ppm, Geochim.Cosmochim. Acta, 33, 859-876, 1969. meteorite, Science, 174, 583-585, 1971. Wasson, J. T., The chemical classificationof iron meteorites,4, Irons Perry, S. H., The metallographyof meteoric iron, U.S. Nat. Mus. with Ge concentrationsgreater than 190 ppm and other meteorites Bull., 184, 1-206, 1944. associatedwith group I, Icarus, 12, 407-423, 1970a. Rambaldi, E., E. Jagoutz, and J. T. Wasson,BitburgmA group IB Wasson, J. T., Ni, Ga, Ge and lr in the metal of iron-meteorites-with- iron meteorite with silicateinclusions, Mineral. Mag., 39, 595-600, silicate-inclusions,Geochim. Cosmochim. Acta, 34, 957-964, 1970b. 1974. Wasson, J. T., An equation for the determinationof iron-meteorite Reed, S. J. B., Phosphorusin meteoritic nickel-iron, in Meteorite cooling rates, Meteoritics, 6, 139-147, 1971. Research,edited by P.M. Millman, pp. 749-762, D. Reidel, Wasson, J. T., MeteoritesmClassificationand Properties,316 pp., Dordrecht, Netherlands, 1969. Springer, New York, 1974. Reed, S. J. B., Determinationof Ni, Ga, and Ge in iron meteoritesby Wasson, J. T., and J. Kimberlin, The chemical classificationof iron X-ray fluorescenceanalysis, Meteoritics, 7, 257-262, 1972. meteorites,2, Irons and pallasiteswith germaniumconcentrations Reid, A.M., R. J. Williams, and H. Takeda, Coexistingbronzite and between8 and 100ppm, Geochim.Cosmochim. Acta, 31, 2065-2093, clinobronzite and the thermal evolution of the Steinbach meteorite, 1967. Earth Planet. Sci. Lett., 22, 67-74, 1974. Wasson, J. T., and R. Schaudy, The chemical classificationof iron Sanz, H. G., D. S. Burnett, and G. J. Wasserburg,A preciseRb87/Sr s7 meteorites,5, Groups IIIC and IIID and other irons with ger- age and initial Sr•7/Sr• for the Colomera iron meteorite,Geochim. manium concentrationsbetween I and 25 ppm, Icarus, 14, 59-70, Cosmochim. Acta, 34, 1227-1240, 1970. 1971. Schaudy,R., J. T. Wasson,and V. F. Buchwald,The chemicalclassifi- Wood, J. A., The cooling rates and parent planets of severaliron cation of iron meteorites,6, A reinvestigationof ironswith Ge con- meteorites, Icarus, 3, 429-459, 1964. centrationslower than I ppm, Icarus, 17, 174-192, 1972. Wood, J. A., Meteoritesand the Origin of Planets,p. 3 l, McGraw-Hi!l, Scott, E. R. D., Studiesof the structureand compositionof iron New York, 1968. meteorites,Ph.D. thesis, Cambridge Univ., Cambridge, England, Yardley, E. D., Ph.D. thesis,University of Manchester,Manchester, 1971a. England, 1966. Scott, E. R. D., New carbide (Fe, Ni)28C•, found in iron meteorites, Yavnel, A. A., The dependenceof the structureof iron meteoriteson Nature Phys. Sci., 229, 61-62, 1971b. their chemicalcomposition and crystallizationconditions (in Rus- Scott, E. R. D., Chemical fractionation in iron meteoritesand its in- sian), Meteoritika, 20, 114-120, 1961. terpretation, Geochim.Cosmochim. Acta, 36, 1205-1236, 1972. Y avnel, A. A., The chemical groups of iron meteoritesand their Scott, E. R. D., and S. O. Agrell, The occurrenceof carbidesin iron peculiarities(in Russian), Meteoritika, 31, 141-148, 1972. meteorites(abstract), Meteoritics,6, 312-313, 1971. Zaslavskaya,N. I., and L. G. Kvasha, On silicatesand chromitefrom Scott, E. R. D., and R. W. Bild, Structure and formation of the San polymineralic inclusionsin the Sikhote-Alin octahedrite(in Rus- Cristobal meteorite, other IB irons and group IIICD, Geochim. sian), Meteoritika, 33, 117-121 1974. Cosmochim.Acta, 38, 1379-1391, 1974. Scott, E. R. D., and J. T. Wasson, Chemical classificationof iron meteorites,8, Groups IC, IIE and IIIF and data on 97 other irons, (Received December 19, 1974; submitted to Geochim. Cosmochim. Acta, 1975. acceptedMarch 19, 1975.)