Search for Past Life on Mars: Possible Relic Biogenic Activity in Martian

Home , Martian meteorite

ME RESEARCHGARTICLE am dicates that they are indigenous to the earch for Past Life on Mars: meteorite and were not formed during its 13,000-yearresidence in the Antarctic en- Possible Relic Biogenic Activity vironment (13). The 813C values of the carbonate in ALH84001 range up to 42 per mil for the in Martian Meteorite ALH84001 large carbonate spheroids (12) and are higher than values for carbonatesin other David S. McKay,Everett K. Gibson Jr., SNC meteorites.The sourceof the carbon Kathie L. Thomas-Keprta,Hojatollah Vali, is the martian atmospheric C02, which has ChristopherS. Romanek, Simon J. Clemett, been recycledthrough water into the carbonate (12). The carbonisotopic composi- Xavier D. F. Chillier, Claude R. Maechling, Richard N. Zare tions of ALH84001 are similar to those measuredin CM2 carbonaceouschondrites Fresh fracture surfaces of the martian meteorite ALH84001 contain abundant polycyclic (14). Consequently, the carbonates in aromatic hydrocarbons (PAHs). These fresh fracture surfaces also display carbonate ALH84001 and the CM2 meteorites are globules. Contamination studies suggest that the PAHs are indigenous to the meteorite. believed to have been formedby aqueous High-resolution scanning and transmission electron microscopy study of surface tex- processeson parent bodies. The b83C in tures and internal structures of selected carbonate globules show that the globules martianmeteorite carbonatesranges from contain fine-grained, secondary phases of single-domain magnetite and Fe-sulfides. The -17 to +42 per mil (12, 15, 16). This carbonate globules are similar in texture and size to some terrestrial bacterially induced range of 13Cvalues exceeds the range of 13C carbonate precipitates. Although inorganic formation is possible, formation of the glob- generatedby most terrestrialinorganic pro- ules by biogenic processes could explain many of the observed features, including the cesses(17, 18). Alternatively,biogenic pro- PAHs. The PAHs, the carbonate globules, and their associated secondary mineral phas- cessesare known to producewide rangesin es and textures could thus be fossil remains of a past martian biota. 813C on Earth(19, 20). ALH84001 arrived on Earth 13,000 yearsago (15, 21) and appearsto be essen- tially free of terrestrialweathering (8). A long-standingdebate over the possibility biomarkerson the basis of what we know ALH84001 does not have the carbon iso- of present-daylife on Marswas addressed by about life on Earth.Therefore, if there is a topic compositionstypically associated with the Viking landerexperiments in 1976. Al- martianbiomarker, we may not be able to weatheredmeteorites (12, 15), and detailed though the resultswere generallyinterpret- recognizeit, unlessit is similarto an earthly mineralogical studies (8) show that ed to be negative for life in the tested biomarker.Additionally, no informationis ALH84001 has not been significantlyaf- surfacesoils, the possibilityof life at other available on the geologic context of this fected by terrestrialweathering processes. locations on Mars could not be ruled out rock on Mars. ALH84001 is somewhat friable and (1). The Viking lander'smass spectrometry ALH84001 is an igneousorthopyroxen- breaks relatively easily along preexisting experimentsfailed to confirmthe existence ite consistingof coarse-grainedorthopyrox- fractures.It is these fracturesurfaces that of organicsfor the martiansurface samples ene [(Mg,Fe)SiO3]and minor maskelynite displaythe carbonateglobules. We analyzed analyzed.Furthermore, the Viking results (NaAISi308), olivine [(Mg,Fe)SiO4], chro- freshly broken fracturesurfaces on small contained no informationon possiblefos- mite (FeCr204),pyrite (FeS2), and apatite chips of ALH84001for polycyclicaromatic sils. Another source of informationabout [Ca3(PO4)2j (4-6). It crystallized 4.5 bil- hydrocarbons(PAHs) using a microprobe possible ancient martian life is the Sher- lion yearsago (Ga) (6). It recordsat least two-step laser mass spectrometer (piL2MS) gotty-Nakhla-Chassigny(SNC) class of two shock events separatedby a period of (22, 23). meteorites,which appearto have come to annealing.The age of the first shock event Polycyclic aromatic hydrocarbons. Spa- the Earthby impactevents on Mars(2, 3). has been estimatedto be 4.0 Ga (7). Unlike tial distribution maps of individual PAHs We have examined ALH84001, collected the other SNC meteorites,which contain on interior fracture surfaces of ALH84001 in Antarcticaand recently recognizedas a only trace carbonate phases, ALH84001 demonstrate that both total PAH abun- meteoritefrom Mars (4). Ourobjective was containssecondary carbonate minerals that dance and the relative intensities of indi- to look for signs of past (fossil) life within formglobules from 1 to -250 ,umacross (4, vidual species have a heterogeneous distri- the pore spaceor secondaryminerals of this 6, 8, 9). These carbonate globules have bution at the 50-p,m scale. This distribution martianmeteorite. Our task is difficultbe- been estimatedto have formed3.6 Ga (10). appears to be consistent with partial geo- cause we only have a small piece of rock Petrographicand electron microprobere- chromatographic mobilization of the PAHs fromMars and we are searchingfor martian sults (4, 11) indicate that the carbonates (24). The average PAH concentration in formed at relatively high temperatures the interior fracture surfaces is estimated to D. S. McKay, Mail Code SN, NASA Lyndon B. Johnson (-700?C); however,the stableoxygen iso- be in excess of 1 part per million (25). The Space Center (JSC), Houston, TX 77058, USA. tope data indicate that the carbonates PAHs were found in highest concentration E. K. Gibson Jr., Mail Code SN4, NASA-JSC, Houston, 00 in rich in carbonates. TX 77058, USA. formed between and 80?C (12). The regions K. L. Thomas-Keprta, Lockheed Martin,Mail Code C23, carbonateglobules are found along fractures From averaged spectra we identified two 2400 NASA Road 1, Houston, TX 77058, USA. and in pore spaces.Some of the carbonate groupings of PAHs by mass (Fig. IA). A H. Vali, Department of Earth and Planetary Sciences, globules were shock-faulted(4, 5). This middle-mass envelope of 178 to 276 atomic McGillUniversity, 3450 UniversitySt., Montreal, Quebec, H3A 2A7 Canada. shock event occurredon Marsor in space, mass units (amu) dominates and is com- C. S. Romanek, Savannah River Ecology Laboratory, and thus rules out a terrestrialorigin for posed mostly of simple 3- to 6-ring PAH Drawer E, Universityof Georgia, Aiken, SC 29802, USA. the globules (3, 8, 13). The isotopic com- skeletons with alkylated homologs account- S. J. Clemett, X. D. F. Chillier,C. R. Maechling, R. N. Zare, Department of Chemistry, Stanford University,Stanford, position of the carbon and oxygen associ- ing for less than 10% of the total integrated CA 94305-5080, USA. ated with the carbonate globules also in- signal intensity. Principal peaks at 178, 202,

924 SCIENCE * VOL. 273 * 16 AUGUST 1996 228, 252, and 278 amu are assigned to by gL2MS did not show the presence of (31). We also conducted experimentsin phenanthrene (CI4Ho), pyrene (C16H1o), terrestrial PAHs within detection limits, which chips of ALH84001were cultured in chrysene (C18H12), perylene or benzopy- which suggests an upper limit for terrestrial nutrientmedium under aerobic and anero- rene (C20H12), and anthanthracene contamination of ALH84001 of 1%. Mea- bic conditions;we found the chips to be (C22H12) (26). A second weak, diffuse surements of four interior fragments of two sterile. high-mass envelope extends from about 300 Antarctic ordinary chondrites (ALH83013 With the use of the piL2MStechnique, to beyond 450 amu. The peak density is and ALH83101) of petrologic classes H6 PAHs have been found in a wide rangeof high and shows a periodicity at 14 and 2 and L6 showed no evidence of indigenous extraterrestialmaterials, including carbona- amu. This distribution implies that there is PAHs. These represent equivalent desorp- ceous and ordinarychondrites (29), inter- a complex mixture of PAHs whose parent tion matrix blanks; previous studies have planetarydust particles(23, 32), and inter- skeletons have alkylated side chains with shown that no indigenous organic material stellargraphite grains (33). Eachmaterial is varying degrees of dehydrogenation; specif- is present in meteorites of petrologic class characterizedby differing PAH distribu- ic assignments are ambiguous. six (29). tions reflectingthe differentenvironments Contamination checks and control ex- Studies of exterior fragments of in which the PAHs formedand their sub- periments indicate that the observed organ- ALH84001 with intact fusion crust show that sequentevolution (for example,as a result ic material is indigenous to ALH84001. no PAHs are present within the fusion crust of aqueousalteration and thermalmetamor- The accumulation of PAHs on the Green- or a zone extending into the interior of the phism). Comparisonof the mass distribu- land ice sheet over the past 400 years has meteorite to a depth of -500 gm (Fig. 1, B tion of PAHs observedin ALH84001 with been studied in ice cores (27). The total through E). The PAH signal increases with that of PAHs in other extraterrestrialma- concentration of PAHs in the cores varies increasing depth, leveling off at -1200 gm terials indicates that the closest match is from 10 parts per trillion for preindustrial within the interior,well away from the fusion with the CM2 carbonaceouschondrites times to 1 part per billion for recent snow crust. This concentration profile is consistent (34). The PAHs in ALH84001, however, deposition. Because Antarctica is in the less with volatilization and pyrolysisof indigenous differ in several respects from the CM2 industrialized Southern Hemisphere, we PAHs during atmosphericentry of the mete- chondrites:Low-mass PAHs such as naph- may expect that concentrations of PAHs in orite and formation of a fusion crust (30), but thalene (C10H8; 128 amu)and acenaphtha- Antarctic ice lie between these two limits. inconsistent with terrestrial introduction of lene (ClHA; 152 amu) are absent in The primary source of PAHs is anthropo- organic material into the interior of ALH84001; the middle-mass envelope genic emissions, which are characterized by ALH84001 along cracks and pore spaces dur- showsno alkylation;and the relativeinten- extensive alkylation (- 10-fold greater than ing burial in the Antarctic ice sheet. These sity of the 5- and 6-ring PAHs and the that of the parent PAHs) (28) and by the resultsindicate that the PAHs are indigenous relativeintensity and complexityof the ex- presence of abundant aromatic heterocy- to ALH84001. tended high-massdistribution are different. cles, primarily dibenzothiophene (CI2H8S; No evidence can be found for laboratory- On Earth,PAHs are abundantas fossil 184 amu). In contrast, the PAHs in based contamination introduced during pro- molecules in ancient sedimentaryrocks, ALH84001 are present at the part per mil- cessing. Samples for analysis were prepared coal, and petroleum,where they arederived lion level (-103 to 10' times higher con- at the meteorite clean labs at NASA John- from chemical aromatizationof biological centration) and show little alkylation, and son Space Center and sealed in containers precursorssuch as marineplankton and ear- dibenzothiophene was not observed in any before they were transported to Stanford ly plantlife (35). In suchsamples, PAHs are of the samples we studied. University. A contamination study conduct- typicallypresent as thousands,if not hun- Analysis of Antarctic salt deposits on a ed prior to analysis of these samples showed dredsof thousands,of homologousand iso- heavily weathered meteorite (LEW 85320) no evidence for any PAH contamination meric series;in contrast,the PAHs we ob-

I 40r-B - 30 -C co 3Mass 178 "Mass202 20- 20- lO.0 , 10 -

- 0 0 400 800 1200 0 400 800 1200 CD 'ii 15C , , Extended mass 15 C -bD distribution ~ 0Mass 228 Mass 252

C~~~~~~~0 \ 50 - 50 - 0) ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~~~1

0 400 800 1200 0 400 800 1200 Distance from fusion crust exterior (Rum) functionof distance fromthe ALH84001 fusion crust for the four primaryPAHs shown in (A). The fusion crust fragment, which showed no preexisting fractures, was cleaved immediatelyprior to analysis using a stainless steel scapel and introduced in <2 min into the V?L2MS.Each plot represents a section perpendicular to the fusion crust surface, which starts at the exterior and sur-face of ALH84001 . The spectrum represents the average of 1280 individual extends a distance of 1200 1Lminward. The spatial resolution is 100 wm along spectra defining an analyzed surface region of 750 by 750 ,urn mapped at a the section line and is the average of a 2 by 2 arrayof 50 by 50 wLmanalyses, spatial resolution of 50 by 50 ,um. (B through E) PAH Signal intensity as a with each analysis spot being the summed average of 5 time-of-flightspectra.

SCIENCE * VOL. 273 * 16 AUGUST 1996 925 servedin ALH84001appear to be relatively tend to be discoidrather than sphericaland in diameter,for analysisby TEM and elec- simple.The in situ chemical aromatization areflattened parallel to the fracturesurface. tron microprobe(37). On the basisof back- of naturallyoccurring biological cyclic com- Intact carbonateglobules appear orange in scatterelectron (BSE) images(Fig. 2), the pounds in early diagenesiscan produce a visible light and have a rounded appear- larger globules (>10 ,um) have Ca-rich restrictednumber of PAHs (36). Hence, we ance; many display altemating black and cores (which also contain the highest Mn wouldexpect that diagenesisof microorgan- white rims. Under high magnificationste- abundances)surrounded by altematingFe- ismson ALH84001could producewhat we reo light microscopyor SEM stereo imag- and Mg-richbands (Fig. 3). Near the edge observed-a few specific PAHs-rather ing, some of the globulesappear to be quite of the globule,several sharp thin bandsare than a complexmixture involving alkylated thin and pancake-like,suggesting that the present.The first band is rich in Fe and S, homologs. carbonatesformed in the restrictedwidth of the second is rich in Mgwith no Fe, and the Chemistry and mineralogyof the car- a thin fracture.This geometrylimited their third is rich in Fe and S again (Fig. 3). bonates.The freshlybroken but preexisting growthperpendicular to, but not parallelto, DetectableS is also presentin patchyareas fracturesurfaces rich in PAHs also typically the fracture. throughoutthe globule. display carbonate globules. The globules We selected a typicalglobule, -50 ,um In situ TEM analysesof the globule in Fig. 4 revealed that Fe- and Mg-richcar- bonates located nearerto the rim range in Fig. 2. False-color compositionfrom ferroan magnesite to pure backscatter electron magnesite.The Fe-richrims are composed (BSE) image of frac- mainlyof fine-grainedmagnetite ranging in tured surface of a chip size from -10 to 100 nm and minor from ALH84001 meteorite showing distri- amountsof pyrrhotite(-5 vol%) (Fig. 3, bution of the carbon- region I, and Fig. 4A). Magnetitecrystals ate globules. Orthopy- arecuboid, teardrop, and irregularin shape. roxene is green and Individual crystals have well-preserved the carbonate glob- structureswith no lattice defects.The mag- ules are orange. Sur- netite and Fe-sulfideare in a fine-grained rounding the Mg-car- carbonatematrix (Fig. 4, A through C). bonate are a black rim Compositionof the fine-grainedcarbonate (magnesite) and a matrixmatches that of coarse-grainedcar- white, Fe-rich rim. bonates located adjacent to the rim (Fig. Scale bar is 0.1 mm. [False color produced 4A). by C. Schwandt] High-resolution transmission electron microscopy(HRTEM) and energy dispersive spectroscopy(EDS) showed that the Fe- sulfide phase associatedwith the Fe-rich rims is pyrrhotite(Fig. 4C). Pyrrhotitepar- Fig. 3. BSE image and A ticles are composedof S and Fe only; no electron microprobe oxygenwas observed in the spectra.Particles maps showingthe con- have atomicFe/S ratiosranging from -0.92 centrationof five ele- to 0.97. The size and shape of the FeS ments in a carbonate OP fromALH84001. The el- particles vary. Single euhedralcrystals of ement maps show that pyrrhotiterange up to --100 nm across; the carbonateis chemi- A polycrystallineparticles have morerounded cally zoned. Colors shapesranging from -20 to 60 nm across range through red, _ . (Fig.4C). HRTEMof theseparticles showed green, light blue, and that their basal spacingis 0.57 nm, which deep blue,reflecting the correspondsto the {1111reflection of the highest to lowest ele- pyrrhotitein a 4C monoclinicsystem. The ment concentrations. magnetite is distributeduniformly in the Scale barsfor all images rim, whereasthe pyrrhotiteseems to be dis- are 20 ,um. (A)BSE im- tributedrandomly in distinctdomains -5 to age showinglocation of F E orthopyroxene (OPX), 10 ,um long (Fig. 3C). Magnetitegrains in clinopyroxene (CPX), ALH84001 did not contain detectable apatite(A), and carbon- amounts of minor elements. In addition, ate (MgC,C). Iron-nch - these magnetite grains are single-domain nms (R) separate the crystalshaving no structuraldefects. center of the carbonate A distinct region, located toward the (C)from a Mg-nchcar- center of the carbonatespheroid but com- bonate (MaC)nm. Re- pletely separate from the magnetite-rich gion in the box is described in Figs. 5 and 6. (B) Ironis most abundantin the parallelrims, -3 ,urmacross, and rim describedabove, also shows accumula- in a region of the carbonate -20 ,um in size. (C) Highest S is associated with an Fe-rich rim; it is not homogeneously distributed,but ratherlocated indiscrete regions or hot spots inthe rim.A lowerS abundance tion of magnetite and an Fe-rich sulfide is present throughoutthe globule in patchy areas. (D) Higherconcentrations of Mg are shown in the Fe-poor (Fig. 3A, region II, and Fig. 5A). This outer region of the carbonate. A Mg-richregion (MgC),-8 ,urmacross, is located between the two Fe-rich region displaystwo types of textures:The rims. (E)Ca-rich regions are associated with the apatite, the Fe-richcore of the carbonate, and the clinopy- first one is more massive and electron-dense roxene. (F)P-rich regions are associated with the apatite.

926 SCIENCE * VOL. 273 * 16 AUGUST 1996 ?^- -- RESEARCH ARTICLE underthe TEM.The secondregion is much ganic models,although more complex mod- genic processes,which are known to oper- less electron-denseand is fine-grainedand els could be proposed. ate under extreme disequilibriumcondi- porous.The porousmaterial occurs mainly In contrast,the coexistenceof magnetite tions. Intracellularcoprecipitation of Fe- in crosscuttingbands and rarelyin isolated and Fe-sulfidephases within partiallydis- sulfides and magnetite within individual patches.We interpretthis poroustexture as solvedcarbonate could be explainedby bio- bacteriahas been reported(42). In addi- a regionin which the massivecarbonate has been partiallydissolved. The nanometer- size magnetiteand Fe-sulfidephases are ev- erywhereassociated with the fine-grained, porous Mg-Fe-rich carbonate. In the regions containing high concentrations of magnetite,dissolution of carbonateis evi- dent (Fig. 5A). In contrastto the magnetite-rich rim, the core area contains few magnetite particles.The Fe-sulfidephases in this magnetite-poorregion have chemical compositionssimilar to that of the pyrrhotite. However, unlike pyrrhotitegrains that have a large variety of morphologies, most of these Fe-sulfideparticles have elon- gatedshapes (Fig. 5B). We couldnot obtain a diffractionpattern of these Fe-sulfidepar- Fig. 4. TEMimages of a thinsection obtained ticles because they were unstable in the frompart of the same fragmentshown in Fig. electronbeam. Possible candidates for these 3A (fromthe regionof arrow1, Fig. 3A). (A) Fe-sulfide minerals include mackinawite Imageat lowmagnification showing the Fe-rich (FeS1I), greigite (Fe3S4), and smythite rimcontaining fine-grain magnetite and Fe-sul- (Fe9S11). Becauseof the morphologicalsim- fide phases and theirassociation with the sur- ilarity to terrestrialgreigite (Fig. 5C), we roundingcarbonate (C) and orthopyroxene that minerals are (opx). (B) Highmagnification of a magnetite- suggest these Fe-sulfide richarea in(A) showing the distributionof indi- probablygreigite (38). vidualmagnetite crystals (high contrast) within Formation of the magnetite and iron the fine-graincarbonate (low contrast). (C) High sulfides. The occurrenceof the fine-grained magnificationof a pyrrhotite-richregion show- carbonate,Fe-sulfide, and magnetitephases ing the distributionof individualpyrrhotite par- ar could be explained by either inorganicor ticles (twoblack arrows in the center)together biogenic processes.Single-domain magne- with magnetite(other arrows) within the fine- tite can precipitateinorganically under am- grainedcarbonate (low contrast). bient temperatureand neutral pH conditions by partial oxidation of ferroussolu- tions (39). This syntheticmagnetite ranges in size from about 1 to more than 100 nm and is chemicallyvery pure (39). Simulta- neous inorganicprecipitation of magnetite and pyrrhotite requiresstrongly reducing conditions at high pH (40). However,carbonate is normallystable at high pH, and the observeddissolution of carbonatewould normallyrequire low pH acidic conditions. It is possiblethat the Fe-sulfides,magnetite, and carbonatesall formed under high pH ..~~~~E...... i conditions,and the aciditychanged at some point to low pH, causingthe partialdisso- 50 ~~~~~~~~~~~~~~~~~~10nm lution of the carbonates.But the Fe-sulfide Fig. 5. TEMimages of a thin section showing the morphologyof the Fe-sulfidephases present in and magnetite do not appearto have un- ALH84001and a terrestrialsoil sample. Ironsulfide phase (greigite?-Fe3S4) is locatedin a magnetite- dergoneany corrosionor dissolution,which poor regionseparate and distinctfrom the magnetite-richrims (Fig. 3A, arrow1I). (A) TEM of a thin would have likely occurred under acidic sectionshowing a cross sectionof a singlecarbonate crystal (large black regions; the apparentcleavage conditions (41). Moreover, as previously featuresare due to knifedamage by ultramicrotomy).A vein of fine-grainedcarbonate (light gray) is mentioned, the dissolutionof carbonateis observed withinthe large carbonatecrystal. Possibly greigite and secondary magnetite(fine, dark alwaysintimately associated with the pres- crystals)have been precipitatedin this fine-grainedmatrix. There is a direct relationbetween the ence of Fe-sulfidesand magnetite.Conse- presence of carbonatedissolution and the concentrationof the fine-grainedmagnetite and Fe-sulfide neither simultaneous phases. This regionshows fewer Fe-richparticles, while regions shown in Fig. 4 containabundant quently, precipitation Fe-richparticles. The cleavagesurface of the carbonatecrystal does not show any dissolutionfeatures of Fe-sulfidesand magnetitealong with dis- (arrows);there is no evidence of structuralselective dissolutionof carbonate.(B) A representative solutionof carbonatesnor sequentialdisso- elongatedFe-sulfide particle, located in the dissolutionregion of the carbonatedescribed in (A), is most lution of carbonateat a later time without likelycomposed of greigite.The morphologyand chemicalcomposition of these particlesare similarto concurrent dissolution of Fe-sulfides and the biogenicgreigite described in (C). (C) High magnification of an individualmicroorganism within a root magnetite seems plausible in simple inor- cell of a soil sampleshowing an elongated,multicrystalline core of greigitewithin an organicenvelope.

SCIENCE * VOL. 273 * 16 AUGUST 1996 927 tion, extracellularbiomediated precipita- the surface of calcite concretions grown An altemative explanation is that these tion of Fe-sulfidesand magnetitecan take from Pleistoceneground water in southern textures, as well as the nanosize magnetite place underanaerobic conditions (43, 44). Italy (52), where they are interpretedas and Fe-sulfides, are the products of microbi- Magnetite particles in ALH84001 are nannobacteriathat have assistedthe calcite ological activity. It could be argued that similar (chemically,structurally, and mor- precipitation. these features in ALH84001 formed in Ant- phologically)to terrestrialmagnetite parti- The origin of these textureson the sur- arctica by biogenic processes or inorganic cles known as magnetofossils(45), which face of the ALH84001carbonates (Fig. 6, A weathering. It is unlikely that reduced phas- arefossil remains of bacterialmagnetosomes and B) is unclear.One possibleexplanation es, such as iron sulfides, would form in Ant- (46) found in a variety of sediments and is that the texturesobserved on the carbon- arctica during inorganic weathering because soils (41, 47, 48) and classifiedas single- ate surfaceare a resultof the partialdisso- reported authigenic sulfur-bearing phases domain(-20 to 100 nm) or superparamag- lution of the carbonate-that is, they are from Antarctic soils and meteorites are sul- netic (<20 nm) magnetite (49). Single- erosional remnantsof the carbonatethat fates or hydrated sulfates. In general, authi- domainmagnetite has been reportedin an- happen to be in the shape of ovoids and genic secondary minerals in Antarctica are cient limestonesand interpretedas biogenic elongateforms, perhaps because the carbon- oxidized or hydrated (53). The lack of PAHs (48). Some of the magnetitecrystals in the ate has preferentiallyeroded along grain in the other analyzed Antarctic meteorites, ALH84001carbonates resemble extracellu- boundariesor dislocations. Shock effects the sterility of the sample, and the nearly lar precipitatedsuperparamagnetic magne- may have enhancedsuch textures.Howev- unweathered nature of ALH84001 argue tite particles produced by the growth of er, becausewe know of no similarexample against an Antarctic biogenic origin. As a anaerobicbacterium strain GS-15 (43). from the terrestrialgeologic recordor from control we examined three Antarctic ordi- Surface features and origin of the car- laboratoryexperiments, we cannot fully nary chondrites (ALH78119, ALH76004, bonates. We examined carbonatesurfaces evaluate this possible explanationfor the and ALH81024, all of which do not have on a numberof small chips of ALH84001 textures.A second possibilityis that arti- indigenous PAHs) from the same ice field with the use of high-resolutionSEM (50). facts can be createdduring sample prepara- where ALH84001 was collected, as well as a The Fe-richrim of globules typicallycon- tion or may resultfrom laboratorycontam- heavily weathered ordinary chondrite that sists of an aggregateof tiny ovoids inter- ination. For example, the applicationof a gave negative results for PAHs (LEW mixed with small irregularto angularob- thick Au-conductivecoating can produce 85320). These meteorites were chosen to jects (Fig. 6A). Ovoids in the exampleare texturesresembling mud cracks,and even cover the different degrees of weathering about100 nm in longestdimension, and the dropletsor blobsof Au. Laboratorycontam- observed on Antarctic meteorites. Examina- irregularobjects range from 20 to 80 nm ination can include dust grains, residue tion of grain surfacesat all magnifications in across.These featuresare typical of those on fromsample cleaning, and organiccontam- weathered and unweathered regions of these the Fe-rich rims of many carbonateglob- ination from epoxy. For comparison,we meteorites showed no sign of the ovoid and ules. These objects are similarin size and examined several control samples treated elongate forms seen in ALH84001. Howev- shape to featuresin the Fe-richrims iden- identicallyto the meteoritechips. We con- er, none of these control meteorites con- tified as magnetiteand pyrrhotite(Fig. 4, B clude that the complextextures (Fig. 6) did tained detectable carbonate. andC). These objectsare too smallto obtain not resultfrom procedures used in our lab- Ovoid features in Fig. 6 are similar in size compositionalanalysis under the SEM. oratory.Only interioror freshlybroken sur- and shape to nannobacteria in travertine In the center of some of the globules faces of chips were used (50). We did ob- and limestone (54). The elongate forms (Fig. 2), the surfaceof the carbonateshows serve an artifact texture from our Au-Pd (Fig. 6B) resemble some forms of fossilized an irregular,grainy texture. This surface conductive coating that consists of a mud filamentous bacteria in the terrestrial fossil texturedoes not resembleeither cleavage or crack-like texture visible only at 50,OOOX record. In general, the terrestrial bacteria a growthsurface of syntheticand diagenetic magnificationor greater.None of the con- microfossils (55) are more than an order of carbonates(51). These surfacesalso display trols display concentrations or blobs of magnitude larger than the forms seen in the small regularlyshaped ovoid and elongated coating material.A lunarrock chip carried ALH84001 carbonates. formsranging from about 20 to 100 nm in throughthe sameprocedures and examined The carbonate globules in ALH84001 longest dimension (Fig. 6B). Similar tex- at high magnificationshowed none of the are clearly a key element in the interpreta- turescontaining ovoids have been foundon featuresseen in Fig. 6. tion of this martian meteorite. The origin of these globules is controversial; Harvey and McSween ( 11) and Mittlefehldt (4) argued, on the basis of microprobe chemistry and equilibrium phase relationships, that the globules were formed by high-temperature metamorphic or hydrothermal reactions. Altematively, Romanek et al. (12) argued on the basis of isotopic relationships that the carbonates were formed under low-tem- peraturehydrothermal conditions (56). The nanophase magnetite and Fe-sulfides present in these globules would likely not be detected in microprobe analyses, which normally have a spatial resolution of about I gim. TEM and our S Fig. 6. High-resolution SEM images showing ovoid and elongate features associated with ALH84001 Our observations carbonate globules. (A) Surface of Fe-rich rim area. Numerous ovoids, about 100 nm in diameter, are maps suggest that nanophase magnetite and present (arrows).Tubular-shaped bodies are also apparent (arrows).Smaller angular grains may be the Fe-sulfides, while concentrated in some magnetite and pyrrhotitefound by TEM. (B) Close view of central region of carbonate (away from rim zones, are present in discrete regions areas) showing textured surface and nanometer ovoids and elongated forms (arrows). throughout the globules. The effect of these

928 SCIENCE * VOL. 273 * 16 AUGUST 1996 1--x.RESEARCH ARTICLE undetectedoxide and sulfide mineralson iron sulfideparticles that could have result- ment of all PAHs present on a sample surface to a reductionreactions spatial resolution of 40 p.m, and detection limits are the carbonate microprobe analyses may ed from oxidation and in the sub-attomole (> 107 molecules) range (1 make the interpretationof the microprobe known to be importantin terrestrialmicro- amol = 10-18 mol). data (4, 11) uncertain.Alternatively, if the bial systems;and (v) the presenceof PAHs 23. S. J. Clemett, C. R. Maechling, R. N. Zare, P. D. globulesare productsof biologic activity, a associatedwith surfacesrich in carbonate Swan, R. M. Walker, Science 262, 721 (1993). 24. M. R. Wing and J. L. Bada, Geochim. Cosmochim. low-temperatureformation would be indi- globules.None of these observationsis in Acta 55, 2937 (1991). cated. The texturesof the carbonateglob- itself conclusive for the existence of past 25. PAH concentration is estimated from comparison of ules are similarto bacteriallyinduced car- life. Although there are alternativeexpla- the averaged spectra of interiorfracture surfaces of ALH84001 with known terrestrialstandards and the bonate crystalbundle precipitates produced nations for each of these phenomenataken Murchison (CM2) meteorite matrix. In the case of in the laboratoryand in a freshwaterpond individually,when they are consideredcol- Murchison (CM2), the total concentration of PAHs (57). Moreover,the observedsequence in lectively,particularly in view of theirspatial has been independently measured to be in the range of 18 to 28 parts per million [K. L. Pering and C. the martiancarbonate globules-Mn-con- association,we conclude that they are evi- Ponnamperuma, Science 173, 237 (1971)]. The av- taining carbonateproduction early (in the dence for primitivelife on early Mars. erage PAH spectrum of ALH84001 used in this esti- core) followed by Fe carbonateand finish- mate was generated from the average of -4000 REFERENCESAND NOTES single-shot spectra acquired from three separate ing with the abundantproduction of re- fracturesurfaces encompassing an analyzed surface duced Fe-sulfides-is a sequence that is area of -2 mm2, representing regions both rich and 1 .P. Mazur et al., Space Sci. Rev. 22, 3 (1978); H. P. poor in PAHs. common in terrestrialsettings, because Mn Klein,Eos 76, 334 (1995). 26. At a single laser ionizationwavelength, is un- is first reducedby biogenicaction, followed 2. H. J. Melosh, Icarus 59, 234 (1984); B. J. Gladman, [.L2MS able to distinguish between structuralisomers; how- J. A. Burns, M. Duncan, P. Lee, H. F. Levison, Sci- by ferric iron and sulfate (57). Pure Mg- ever, because differentisomers have differentphoto- ence 271, 1387 (1996). ionization cross sections, mass assignments are carbonate(magnesite) can also be produced 3. H. P. McSween Jr., Meteoritics 29, 757 (1994). Anal- based on the most probable isomer. In the case of by biomineralizationunder alkalinecondi- yses of the gases in glassy inclusions in the SNC masses 178 and 202, the possible isomer combina- meteorites EET79001 by Bogard and Johnson [Sci- tions (59). On the basis of these observations are phenanthrene or anthracene and pyrene or ence 221, 651 (1983)], Becker and Pepin [Earth fluoranthene.At the photoionizationwavelength used tions, we interpretthat the carbonateglob- Planet. Sci. Lett. 69, 225 (1983)], and Marti et al. in this study (266 nm), the ,uL2MSinstrument is -19 ules have a biogenicorigin and were likely [Science 267, 1981 (1995)] have shown that the times as sensitive to phenanthrene as to anthracene abundance and isotopic compositions of the formedat low temperatures. and -23 times as sensitive to pyrene as to fluoran- trapped gases in the SNC meteorites and the mea- It is possible that all of the described thene [R. Zenobi and R. N. Zare, in Advances in Mul- sured atmospheric compositions on Mars, mea- tiphoton Processes and Spectroscopy, S. H. Lin,Ed. featuresin ALH84001can be explainedby sured in situ by the Viking landers, have a direct (World Scientific, Singapore, 1991), vol. 7, pp. one-to-one correlation (more than nine orders of inorganicprocesses, but these explanations 1-167]. Hence, masses 178 and 202 are assigned to magnitude in concentrations). This remarkable appearto requirerestricted conditions-for phenanthrene and pyrene. In the case of higher agreement is one of the strongest arguments that the masses, more structural isomers exist, and assign- example, sulfate-reducingconditions in SNC meteorites represent samples from Mars. ments are based on those PAHs known to have high 4. D. W. Mittlefehidt,Meteoritics 29, 214 (1994). Antarctic ice sheets, which are not known cross sections from comparisons with standards. H. Treiman, ibid. 30, 294 (1995). to occur. Formationof the describedfea- 5. A. 27. K. Kawamura and 1. Suzuki, Naturwissenschaften A. J. D. H. ibid. 6. E. Jagoutz, Sorowka, Vogel, Wanke, 81, 502 (1994). tures by organic activity in Antarctica is L. B. M. H. Wies- 29,478 (1994); E. Nyquist, Bansal, 28. W. W. Youngblood and M. Blumer, Geochim. Cos- also possible,but suchactivity is only poorly C. Y. LunarPlanet. Sci. mann, Shih, 26,1065 (1995). mochim. Acta 39,1303 (1975); S. T. Wakeham, C. F. G. Nature 380, 57 understoodat present. However, many of 7. R. D. Ash, S. Knott, Turner, Schaffner, W. Giger, ibid. 44, 403 (1980); R. E. the describedfeatures are closely associated (1996). LaFlammeand R. A. Hites, ibid. 42, 289 (1978); T. E. 8. S. J. Wentworth and J. L. Gooding, Lunar Planet. with the carbonateglobules which, based Jensen and R. A. Hites, Anal. Chem. 55, 594 (1983). Sci. 26,1489 (1995). 29. S. J. Clemett, C. R. Maechling, R. N. Zare, C. M. 0. on textural and isotopic evidence, were 9. K. L. Thomas et al., ibid., p. 1409. D. Alexander, LunarPlanet. Sci. 23, 233 (1992); L. J. likely formedon Marsbefore the meteorite 10. S. K. Knott, R. D. Ash, G. Turner,ibid., p. 765. Kovalenko et al., Anal. Chem. 64, 682 (1992). came to Antarctica.Consequently, the for- 11. R. Harvey and H. P. McSween Jr., Nature 382, 49 30. The interiors of stony meteorites are not heated (1996). above 1000 to 1200C during passage through the mation of possibleorganic products (mag- 12. C. S. Romanek et al., ibid. 372, 655 (1994). Earth's atmosphere. For example, in the CM carbo- netite and Fe-sulfides)within the globules 13. J. L. Gooding, Icarus 99, 28 (1992). naceous chondrite Murchison, amino acids ha've is difficult to understandif the carbonates 14. M. M. Grady, I. P. Wright,P. K. Swart, C. T. Pillinger, been found by Kvenvolden et al. [Nature 228, 923 formed on Mars the and Geochim. Cosmochim. Acta 52, 2855 (1988). (1970)]. Iftemperatures had been above 120?C, ami- and magnetite 15. A. J. T. Jull, C. J. Eastoe, S. Xue, G. F. Herzog, no acids would have degraded. Fe-sulfidesformed in Antarctica.Addition- Meteoritics 30, 311 (1995). 31. Sources of laboratory contamination fall into three ally, these productsmight requireanerobic 16. R. H. Carr, M. M. Grady, I. P. Wright,C. T. Pillinger, categories: sample handling, laboratoryair, and vir- bacteria,and the Antarctic ice sheet envi- Nature 314, 248 (1985); I. P. Wright,C. P. Hartmetz, tual leaks (that is, sources of PAHs inside the ,L2MS C. T. Pillinger,J. Geophys. Res. 98, 3477 (1993). vacuum chamber). Possible contamination during ronment appearsto be oxygen-rich;ferric 17. A. J. T. Jull, S. Cloudt, C. J. Eastoe, Lunar Planet. sample preparation was minimized by performing oxide formedfrom metallic Fe is a common Inst. Tech. Rep. 96-01, part 1 (Lunarand Planetary nearly all sample preparation at the NASA-JSC me- weatheringproduct in Antarcticmeteorites. Institute,Houston, TX, 1996), p. 22; J. D. Hudson, J. teorite curation facility. In cases where subsequent Geol. Soc. London 133, 637 (1977). sample handling was requiredat Stanford, all manip- In examining the martian meteorite 18. I. D. Clark and B. Lauriol, Chem. Geol. 102, 217 ulationwas performed in less than 15 minutes, using ALH84001we have found that the follow- (1992); N. Nakai, H. Wada, Y. K. lyosu, M. Takimoto, only stainless steel tools previously rinsed and ultra- ing evidence is compatiblewith the exis- Geochem. J. 9, 7 (1975). sonicated in methanol and acetone. Dust-free gloves 19. K. J. Murata, I. Friedman, B. M. Madsen, U.S. Geol. were worn at all times and work was performed on a tence of past life on Mars:(i) an igneous Surv. Prof. Pap. 614B, 1 (1969); A. M. Martini,L. M. clean aluminum foil surface. To quantify airborne Mars rock (of unknown geologic context) Walter, J. M. Budai, T. C. W. Ku, Geol. Soc. Am. contamination from exposure to laboratory air, two that was penetratedby a fluid along frac- Abstr. Prog. 27, A292 (1995); G. E. Claypool and 1. clean quartz discs were exposed to ambient labora- R. Kaplan, in Natural Gases in Marine Sediments, I. tory environments both at NASA-JSC and Stanford. tures and pore spaces,which then became R. Kaplan, Ed. (Plenum, New York, 1974), pp. 99- Each disc received an exposure typical of that expe- the sites of secondarymineral formation 139. rienced subsequently by samples of ALH84001 dur- and possiblebiogenic activity;(ii) a forma- 20. T. 0. Stevens and J. P. McKinley,Science 270, 450 ing sample preparation. No PAHs were observed on tion for the (1995). either quartz disk at or above detection limits. Be- age carbonateglobules younger 21. Priorto landing in the Antarctic ice field, this meteor- cause contamination can depend on the physical than the age of the igneousrock; (iii) SEM ite was in space for about 16 millionyears, based on characteristics of the individualsample (forexample, and TEM imagesof carbonateglobules and cosmic ray exposure data [D. D. Bogard, LunarPlan- a porous materialwill likelygive a larger contamina- featuresresembling terrestrial microorgan- et. Sci. 26,143 (1995)]. tion signal than a nonporous one), additional con- 22. ,uL2MSwas used to analyze fresh fractured samples tamination studies have been previously conducted isms, terrestrial biogenic carbonate struc- of ALH84001 for the presence of PAHs. The ,uL2MS at Stanford[see (29)].Briefly, samples of the mete- tures, or microfossils; (iv) magnetite and instrument is capable of the simultaneous measure- oriticadid residues of Barwell(L6) and Bishunpur

SCIENCE * VOL. 273 * 16 AUGUST 1996 929 (L3.1) were exposed to laboratory air for 1 and 4 625-645); R. B. Frankeland R. B. Blakemore, Eds., Claridge, Antarctica: Soils, Weathering Processes days, respectively. Barwell (L6) is known to contain IronBiomineralization (Plenum, New York, 1991). and Environment, Developments in Soil Science no indigenous PAHs and none were observed on the 43. D. R. Lovely, J. F. Stolz, G. L. Nord Jr., E. J. P. (Elsevier,Amsterdam, 1987), vol. 16, exposed sample. Bishunpur (L3.1), in contrast, con- Phillips,Nature 330, 252 (1987). 54. R. L. Folk, J. Sediment Petrol. 63, 990 (1993). tains a rich suite of PAHs, but no discernible differ- 44. D. Fortin, B. Davis, G. Southam, T. J. Beveridge, 55. J. W. Schopf and C. Klein, Eds., The Proterozoici- ences in signal intensities were observed between J. Indust. Microbiol. 14,178 (1995). ment Biosphere (Cambridge Univ. Press, New York, exposed and unexposed samples. To test for con- 45. J. L. Kirschvinkand S. B. R, Chang, Geology 12, 559 1992). tamination from virtualleaks, the [uL2MSinstrument (1984). 56. Contraryto the interpretations presented by Harvey was periodicallychecked using samples of Murchi- 46. R. P. Blakemore, Annu. Rev. Microbiol. 36, 217 and McSween (11) concerning the equilibriumparti- son (CM2) meteorite matrixwhose PAH distribution (1982). tioning of stable carbon isotopes on Mars, 813C val- has been previouslywell characterized. No variations 47. N. Petersen, T. von Dobeneck, H. Vali, Nature 320, ues for carbonate and atmospheric C02 are incom- in either signal intensity or distributionof PAHs were 611 (1986); J. W. E. Fassbinder, H. Stanjek, H. Vali, patible with a high-temperature origin. For the observed for p.L2MSinstrument exposure times in ibid. 343,161 (1990). CaCO3-CO2 system, solid carbonate is only en- excess of 3 days. No sample of ALH84001 was in 48. S. R. Chang, J. F. S. Tolz, J. L. Kirschvink,S. M. riched in 13C below about 2000C [T. Chacko, T. K. the instrument for longer than 6 hours. The ,uL2MS Awramik,Precambrian Res. 42, 305 (1989). Mayeda, R. N. Clayton, J. R. Goldsmith, Geochim. vacuum chamber is pumped by an oil-free system: 49. R. F. Butlerand S. K. Banerjee, J. Geophys. Res. 80, Cosmochim. Acta 55, 2867 (1991); C. S. Romanek, two turbomolecular pumps and a liquid nitrogen- 4049 (1975). E. L. Grossman, J. W. Morse, ibid. 56, 419 (1992)]. cooled cyropump. 50. We handpicked small chips in a clean bench from Assuming the 813C of ambient CO2 is 27 per mil (the 32. K. L. Thomas et al., Geochim. Cosmochim. Acta 59, our curatorial allocation of ALH84001. Most chips value for trapped gases in the martian meteorite 2797 (1995). were from a region near the central part of the me- EETA79001) [C. P. Harzmetz, I. P. Wright, C. T. 33. S. J. Clemett, S. Messenger, X. D. F. Chillier,X. Gao, teorite and away from the fusion crust, although for Pillinger,in Workshop on the Mars Surface and At- R. M. Walker, R. N. Zare, LunarPlanet. Sci. 26, 229 comparison we also looked at chips containing mosphere Through Time, R. M. Haberle et al., Eds. (1996). some fusion crust. For high-resolution work we used (Tech. Rep. 92-02, Lunar and Planetary Institute, 34. S. J. Clemett, thesis, Stanford University(1996). an Au-Pd coating estimated to be -2 nm thick in Houston, TX, 1992), pp. 67-68], a 613C of 42 per mil 35. B. P. Tissot and D. H. Welte, Petroleum Formation most cases. On some samples we used a thin (<1 for carbonate in ALH84001 can only be explained by and Occurrence (Springer, New York, 1978). nm) coating, about 10 s with our sputter coater; a precipitation temperature around 0?C. This tem- 36. R. E. LaFlamme and R. A. Hites, Geochim. Cosmo- these samples usually showed charging effects and perature estimate is compatible with an independent chim. Acta 42, 289 (1978); S. G. Wakeham, C. could not be used for highest resolution imaging. We range of precipitation temperatures (0? to 80?C) Schaffner, W. Giger, ibid. 4.4, 415 (1980). more typically used 20 to 30 s. For backscatter and based on the 8180 of the same carbonates (12). If 37. We removed the spheroid from a thin section with a chemical mapping we used a carbon coat of about 5 carbon isotopes are distributed heterogeneously on micro-coring device, embedded it in epoxy, thin sec- to 10 nm thick. We monitored the possible artifacts Mars in isolated reservoirs (for example, the crust tioned it using an ultramicrotome, and analyzed it from the coating using other reference samples or by and atmosphere), precipitation temperature esti- using a TEOL2000FX TEM [technique described in looking at fresh mineral surfaces on ALH84001. We mates based on stable isotopes are relativelyuncon- (31)]. We made about 50 thin sections from the glob- could also estimate relative coating thicknesses by strained. This is probably not the case, though, as ule, and the remainingcarbonate globule was coat- the size of the Au and Pd peaks in the energy dis- other researchers [(3);L. L. Watson, I. D. Hutcheon, ed with evaporated carbon for conductivity (- 10 nm) persive x-ray spectrum. For our heaviest coating, a S. Epstein, E. M. Stolper, Science 265, 86 (1994); B. and mapped with wavelength dispersive spectros- slight crazing texture from the coating is barelyvisible M. Jakosky, Geophys. Res. Lett. 20, 1591 (1993)] copy for majorand minor elements, using a Cameca at a magnification of 50,OOOx on the cleanest fresh have documented a link between crustal and atmo- SX 100 microprobe (Fig. 3). grain surfaces. The complex textures shown in most spheric isotopic reservoirs on the planet. 38. H. Stanjek, J. W. E. Fassbinder, H. Vali, H. Wegele, of the SEM photographs are not artifacts of the coat- 57. C. Buczynsik and H..S. Chafetz, J. Sediment Petrol. W. Graf, Eur. J. Soil Sci. 45, 97 (1994). ing process but are the real texture of the sample. 61, 226 (1991); also see Figs. 1 D and 3, A and B. 39. B. A. Maher, in IronBiominerals, R. B. Frankeland R. We used a JEOL35 CF and a PhilipsSEM with a field 58. K. H. Nealson and D. Saffarini,Annu. Rev. Microbiol. P. Blakemore, Eds. (Plenum, New York, 1991), emission gun at the JSC. The JEOLand Philips p. (FE0) 48, 311 (1994); J. P. Hendry, Sedimentology 40, 87 179; R. M. Taylor, B. A. Maher, P. G. Self, Clay Miner. SEMs are equipped, respectively, with a PGT and (1993). 22, 411 (1987). Link EDS system. We achieved about 2.0 nm reso- 59. J. B. Thompson and F. G. Ferris, Geology 18, 995 40. H. G. Machel, in Palaeomagnetic Applications in Hy- lution at 30kV. Some images were also taken at drocarbon Exploration and Production, P. Turner lower kV, ranging from 1 to 10 kV. Inmost cases, the (1990). and A. Turner,Eds. (Geol. Soc. Spec. Publ. 98, Geo- chips were coated after handpicking with no further 60. We thank L. P. Keller,V. Yang, C. Le, R. A. Socki, M. logical Society, London, 1995), pp. 9-29; R. M. Gar- treatment. For comparison, several chips were ultra- F. McKay, M. S. Gibson, and S. R. Keprtafor assist- rels and C. L. Christ, Solutions, Minerals and Equi- sonically cleaned for a few seconds in liquid Freon ance; and R. Score and M. Lindstromfor their help in libria(Freeman, Cooper, San Francisco, 1969). before coating to remove adhering dust. Several sample selection and preparation. The guidance of 41. H. Valiand J. L. Kirschvink,Nature 339, 203 (1989); samples were examined uncoated at low kV. We J. W. Schopf at an early stage of this study is appre- H. Vali, 0. Forster, G. Amarantidis, N. Petersen, carbon-coated and examined all of the surfaces an- ciated. We also thank G. Horuchi, L. Hulse, L. L. EarthPlanet. Sci. Lett. 86, 389 (1987). alyzed for PAHs only after the PAH analyses had Darrow, D. Pierson, and personnel from Building37. 42. H. Valiand J. L. Kirschvink,in IronBiominerals, R. B. been completed. The work was supported in part by NASA's Plane- Frankel and R. B. Blakemore, Eds. (Plenum, New 51. J. Paquette, H. Vail, E. W. Mountjoy,Am. Mineral.,in tary Materials Program (D.S.M. and R.N.Z.) and the York, 1990), pp. 97-115; S. Mann, N. H. C. Sparks, press; J. Paquette, H. Vali,A. Mucci, Geochim. Cos- Exobiology Program (D.S.M.). Additional support R. B. Frankel, D. A. Bazylinski, H. W. Jannasch, mochim. Acta, in press. from NASA-JSC Center Director's Discretionary Nature 343,258 (1990); M. Farina,D. M. Esquivel, H. 52. E. F. McBride, M. Dane Picard, R. L. Folk, J. Sed. Program (E.K.G.)is recognized. C.S.R. and H.V. ac- G. P. L. de Barros, ibid., p. 256; D. A. Bazylinski, B. Res. A64, 535 (1994); see fig. 9. knowledge support from the National Research R. Heywood, S. Mann, R. B. Frankel,ibid. 366, 218 53. E. K. Gibson, S. J. Wentworth, D. S. McKay, Proc. Council. X.D.F.C. acknowledges support from the (1993); A. Demitrac, in Magnetite Biomineralization 13th Lunar Planet. Scd. Conf., Part 2, J. Geophys. Swiss National Science Foundation. and Magnetoreception, J. Kirschvink,D. S. Jones, Res. 88, suppl. A912 (1983); M. A. Velbel, Meteorit- B. J. McFadden, Eds. (Plenum, New York, 1985), pp. ics 23, 151 (1988); I. B. Campbell and G. G. C. 5 April1996; accepted 16 July 1996

Associate Editor An editorialposition at SCiENCEis available for a Ph.D. scientistwith a broadrange of interestsand research experience in one or more of the following areas:molecular biology, biochemistry,neuroscience, and/or developmental biology. We are seeking someone with at least 2 or 3 years of postdoctoralexperience in the life sciences and publicationsin peer-reviewedjournals. Editors have varied responsibilitiesincluding solicitation, selection, and editing of manu- scripts. Previous editorial experience is not required.This is a full-time position in either our Washington, DC, or Cambridge,UK, office. Please send resume and list of publicationswith cover letter and salaryrequirements to: Mr. GregoryStokes, AAAS, 1200 New York Avenue,N.W., Washington,DC, 20005

930 SCIENCE * VOL. 273 * 16 AUGUST 1996