Chemical, Multispectral, and Textural Constraints on the Composition And

Home , Fontana (lunar crater), Fontana (Martian crater), Normative mineralogy

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 104, NO. E4, PAGES 8679-8715, APRIL 25, 1999

Chemical,multispectral, and textural constraints on the compositionand origin of rocks at the Mars Pathfinder landing site H. Y. McSweenJr., 1 S.L. Murchie,2 J. A. Crisp,3 N. T. Bridges,3 R. C. Anderson,3 J.F. BellII1, 4 D. T. Britt,5 J. Briickner,6 G. Dreibus,6 T. Economou,7 A. Ghosh,1 M.P. Golombek,3 J.P. Greenwood,1,8J. R. Johnson,9 H. J.Moore, 1ø R. V. Morris,11 T. J. Parker,3 R. Rieder,6 R. Singer,12 and H. W•inke6

Abstract. Rocks at the Mars Pathfindersite are probablylocally derived. Textureson rock surfacesmay indicatevolcanic, sedimentary, or impact-generatedrocks, but aeolianabration and dustcoatings prevent unambiguous interpretation. Multispectral imaging has resolved four spectralclasses of rocks: gray and red, which occuron differentsurfaces of the samerocks; pink, which is probablysoil crusts;and maroon,which occursas largeboulders, mostly in the far field. Rocksare assignedto two spectraltrends based on the positionof peakreflectance: the primary spectraltrend contains gray, red, andpink rocks;maroon rocks constitute the secondaryspectral trend. The spatialpattern of spectralvariations observed is orientedalong the prevailingwind direction. The primary spectraltrend arises from thin ferric coatingsof aeoliandust on darker rocks. The secondaryspectral trend is apparentlydue to coatingby a differentmineral, probably maghemiteor ferrihydrite. A chronologybased on rock spectrasuggests that roundedmaroon bouldersconstitute the oldestpetrologic unit (a flood deposit),succeeded by smallercobbles possiblydeposited by impact,and followed by aeolianerosion and deposition.Nearly linear chemicaltrends in alphaproton X-ray spectrometerrock compositions are interpretedas mixing linesbetween rock and adheringdust, a conclusionsupported by a correlationbetween sulfur abundanceand red/bluespectral ratio. Extrapolationsof regressionlines to zero sulfurgive the compositionof a presumedigneous rock. The chemistryand normativemineralogy of the sulfur- free rock resemblecommon terrestrial volcanic rocks, and its classificationcorresponds to andesite. Igneousrocks of this compositionmy occurwith clasticsedimentary rocks or impactmelts and breccias.However, the spectralmottling expected on conglomeratesor brecciasis not observedin any APXS-analyzedrocks. Interpretationof the rocksas andesitesis complicatedby absenceof a "1 gm" pyroxeneabsorption band. Plausibleexplanations include impact glass, band maskingby magnetite,or presenceof calcium-and iron-rich pyroxenes and olivine which push the absorption band minimumpast the imager'sspectral range. The inferredandesitic composition is most sinfilarto terrestrialanorogenic icelandites, formed by fractionationof tholeiiticbasaltic magmas. Early meltingof a relativelyprimitive Martian mantlecould produce an appropriateparent magma,supporting the ancientage of Patlff•nderrocks inferred from their incorporationin Hesperianflood deposits.Although rocks of andesiticcomposition at the Patlff•ndersite may representsamples of ancientMartian crust,inferences drawn about a necessaryrole for water or plate tectonicsin their petrogenesisare probablyunwarranted.

1DepartmentofGeological Sciences, University ofTennessee, 1. Introduction Knoxville. The Mars Pathfinder landing site is strewn with 2AppliedPhysics Laboratory, Johns Hopkins University, Laurel, Maryland. semiroundedpebbles, cobbles, and bouldersthat comprise 3jetPropulsion Laboratory, California Institute ofTechnology, 16% of the surfacearea [Smithet al., 1997a]. The remainderof Pasadena. the site consistsof fines,partly as windblown"drift" forming 4Centerfor Radiophysics andSpace Research, Cornell University, various aeolian bedforms,and partly as depositstypically Ithaca, New York. strewnwith small pebbles. The measuredrock abundanceis 5Lunarand Planetary Laboratory, University ofArizona, Tucson. similar to an estimatefor the entire Pathfinderlanding ellipse 6Max-Planck-InstitutftirChemie, Mainz, Germany. (18-25%) based on thermal emissivityresults [Christensen, 7EnricoFermi Institute, University ofChicago, Chicago, Illinois. 1986]. The characteristicsof this rocky landscapeare 8Nowat Institute ofGeophysics andPlanetary Physics, University of consistentwith depositionby the Ares and Tiu catastrophic California, Los Angeles. floods, possiblyoverprinted by ejecta from a nearby impact 9U.S.Geological Survey, Flagstaff, Arizona. crater [Golombeket al., 1997a;Parker and Rice, 1997]. The 10Deceased[September 21,1998]. Pathfinder location was originally selected,in part, on the 11NASAJohnson Space Center, Houston, Texas. premisethat it might be a "grab bag" site containingrocks 12CatalinaTechnologies, Tucson, Arizona. carried by floods from the ancient southern highlands [Golombeket al., 1997b]. However,the 800-1andistance to Copyright1999 by the AmericanGeophysical Union. the southern highlands boundary, the recognition that Paper number98JE02551. terrestrialfloods do not carry large rocks great distances,and 0148-0227/98/98JE-02551 $09.00 the angularityof many blocks all suggestthat rocks at the

8679 8680 MCSWEEN ET AL.: MARS PATHFINDER ROCKS

•o•o A-7 , Baker's Bench Yogi A-5 •. Lamb TheDice--•'.-t:[_ A-10 .i• Souffi•-•"A-4•!:' ScoobyIA-8 Doo PhotornetryA-3 •1%A'2 Flats Barnacl&- •:•:'...... ':". MintJulep Bill ...... ,-'-:.:.:..•,• ::•;:••,,...:•: :•..,Mini-Matterhorn

Giver ...... •ds + '¾.: A-23 •mbam •uash . BrokenStimpy +Flat lop Has•k

Moe A-18,-19"•' ':' % 'Wedge [' Half•me A-15 A-17Sha MermaM TheGardenRock Charming• Dune

A-27 .,,,,•,,,•North Chimp WindPrevailing Direction t • 2meters

Figure 1. Sketchmap of the Pathfinderlanding site showingthe approximatelocations of rocksand other featuresdescribed or referencedin the text. Eight rockswhich were analyzedby the APXS are coloredblack, and six soil APXS analysissites are indicatedby small solid boxes. APXS chemicalanalysis numbers are also given. Off the map in the directionindicated by the arroware BakersBench and Seal,which are 8 and 37.5 m from the centerof the lander,respectively. The rovertraverse, spanning 105 m, is alsoillustrated; fine lines connectingsome traverse segments indicate repositioning of the rover'sinternal guidance system based on end-of-dayIMP images.

Pathfinder site may be locally derived [Malin et al., 1997]. The nature of the Pathfinder site rocks is somewhat Some rocks might be related to domes, interpreted to be uncertainbecause of conflicting or ambiguousobservations volcanic [Greeley et al., 1977] or sedimentary[Parker and and measurements.Preliminary imaging results [Rover Team, Rice, 1997], locatedapproximately 100 km eastof the site and 1997; Smithet al., 1997a] suggestedthat a diverseassortment associated with a dark mottled unit that extends into the of rocksis present,an inferencethat appearsto be inconsistent westernpart of the Pathfinderlanding ellipse. More likely, with availablerock chemistry[Rieder et al., 1997a]. Here we the rocks are samplesof a ridged plains unit [Britt et al., describe chemical compositions measured by the rover- 1998] or of older crust that may constitutethe local bedrock mounted alpha proton X-ray spectrometer (^PXS), and exposed,streamlined islands. Rocks at the Pathfindersite multispectral imaging by the Imager for Mars Pathfinder (IMP), and observationsof structuresand surfacetextures by itself are Hesperian-age materials interpreted to be the the IMP and by rover cameras. These data allow new lowermostfloors of the outflow channel system[Rotto and inferencesto be drawn concerningthe mineralogy,petrology, Tanaka, 1995; Tanaka, 1997]. It is likely that the bedrock and origin of theserocks. Our interpretationsare guided,to beneaththis unit is lower Hesperianbut it could be partly someextent, by the geologiccontext of the Pathfindersite, by upperNoachian, or both, dependingon local stratigraphy previous observations at Viking sites, and by the The relative locations of rocks described or alluded to in petrogenesisof SNC meteoriteswhich are commonlythought thispaper, as well as the rovertraverse, are shownin Figure1. to be Martian rocks[McSween, 1994, and referencestherein]. This map can be cross-referencedwith a previouslypublished The integrated chemical, multispectral, and textural data panoramicview of the landingsite showingrock namesand suggestthat the Pathfinder site is apparentlydominated by a locations(Mars Pathfinder, 1997, Plate 6). distinctive,perhaps unusual lithology. MCSWEENET AL.: MARS PATHFINDERROCKS 8681

2. Rock Structures and Textures processesprobably have enlarged the originalvesicles and changedtheir shapes[Greeley and Iverson, 1985]. A

2.1. Nature of the Observations possiblealternative to vesiclesis chemicallyetched pits, such as thoseformed in Antarcticaduring brief periodswhen rock surfaceshave thin films of snow xnelt;this mechanismhas been Observationswere made using camerason both the lander proposedas an explanationfor the pits in rocksat Viking and rover that permit interpretationsof rock textures. The landingsites [Allen and Conca,1991]. Otherpossibilities front cameraon the rover providedclose-up images of rocksat includeetching of softerminerals in a matrixof harderminerals a spatialresolution of 0.7-1 mm per pixel at closestrange. For [Carr, 1981]or wind blastingto carveventifact pits. It is rocksfarther than 1.3 m from the IMP landercamera yet within unclearwhether vesicleson the surfaceof Stimpy (Figure 2) 40 cm of rover cameras, the spatial resolution of the mighthave exerted control on windabrasion or whetherall of uncompressedrover camera images is better than that for the pits couldbe dueto weathering.The differencebetween losslessIMP images. However, "super resolution" images thedeep pits covering most of Stimpy'ssurface and the curved preparedby coregisteringand co-addingmultiple single-frame slottedgrooves at thetop of therock may reflect a relationship IMP images[Parker, 1998] can hnprovethe resolutionof IMP between the impact angle of saltating particles and images by factors of 2 to 3. The option to compressrover susceptibilityto abrasion[Greeley et al., 1982]. A surfaceof imagesat 4.9:1 was selectedonly for a few rover frames,used Moe that has sufferedstrong wind abrasionshows only a few primarily for navigation. Many compressionoptions were smallpits, suggesting either that this rock is not vesicularor available for IMP, and most of the imageswere in the range that the pits were filled by dust. Yogi is anotherrock that 1.3:1 lossless to 12:1 lossy. More description of these appearsto haveonly a few pits. cameraswas provided by Smith et al. [1997b] and Rover Onerock, Squash, exhibits an unusual'knobby shape with Team [1997]. The IMP and rover front cameraseach provide lobesand a protrusionroughly 10 cm in size(Figure 3). No stereoscopicpairs; stereocalibration images were available other rocksat the site have similarprotrusions. This texture for the rovercameras but no flat-field images. Imagesfrom the couldindicate that the rock is an autobrecciatedlava, a pillow rover rear color camera did not prove as useful, due to the basalt,a sedimentcontaining rounded cobbles, a volcanic differentfocal lengths of the red, green,and infrared pixels and rock with lithic fragments,or an impact breccia. The the smalldynamic range of the infraredpixels. protrusionshave lower albedo than the restof the rock,much 2.2. Results of whichappears to be coatedwith dustor hasa matrixthat mineralogicallyresembles the dust. The otherside of Squash facingthe IMP doesnot haveas manyprotmsions, and has a A varietyof surfacetextures and structuresare seenon the near-vertical,fairly dust-freeface and a dust-coveredtop. rocks at the landing site, including pitted, knobby, smooth, Linear features,typically appearingas repeatingsubtle lineated,and bumpytextures, and exfoliationstructures. For eachtype, we providea descriptionand discussthe possible light-dark bands spaced 3-5 mm apart, are seen on many geneticimplications, with emphasisplaced on rocksanalyzed Pathfinderrocks. The moststriking examples are Chimp(Plate by APXS (Table 1). Thesetextures may be indicativeof a 1) andZebra. Theboundaries of thebands cannot be resolved numberof processes,but theyare not diagnosticof a particular well enoughto describethe sharpnessof the transition,but type of rock. they do not appearto pinchout or thicken. In someplaces Pits in the 0.1 to 1 cm size range are found on many rock lineationsmay be bandsdefined by thin subtletopographic surfaces,typically occupying areas of 5-10 vol %. The pitted ridges,but mostappear to be flushwith the rock surface. Other rocks with these linear features include Mini- texturesresemble those of some rocks at the Viking Lander 1 site rather than the spongy-lookingrocks common at the Matterhorn,Yogi (lowereast face), Half Doxne,Ender, Squid, Viking Lander2 site. On somerocks, such as Souffle'and the Flat Top,and Booboo. The lineation is not a cameraor image- Dice, the pits could be volcanicvesicles, although aeolian processingartifact (e.g., the samelineation seen on the front face of Flat Top in rover images also appearsin super resolutionIMP images,at an orientationof about40 ø to its top surface).Sometimes the lineationis moreeasily seen in a Table 1, Observed Surface Textures for Rocks superresolution composite image or an anaglyph(requiring Analyzedby APXS three-dimensionalred/blue glasses)than in a single-frame image,but the featuresare alsofound in single-frameimages. RockName Pits Flutes LineationsBumps In somerocks, such as BarnacleBill, multipleorientations of apparentlineations are faint and nonpervasive.In any case, similarlyoriented linear featuresmight reflect any of the Barnacle Bill X x x following: orientationof stressesthe rock has experienced; Yogi ? x x sedimentary or igneous layering; wind-carved rilles; Wedge X x x metamorphicfoliation; or internal alignmentsof bubbles, Shark x x minerals,or zonesof weakness.Alignment of centimeter-size Half Dome X X x x vesiclesin rocksat the Viking landingsites [McCauley et al., Moe ? X 1979;Cart, 1981] impartsa layeredtexture to someof those Stimpy X X rocks. Alignmentof pits of this sizerange is uncommonin Chimp X x Pathfinderrocks, although Chimp (Plate 1) is an exception. However,it is possiblethat lineationson other Pathfinder x, stronglydeveloped; x, weaklydeveloped; rocks are expressionsof smaller (subpixelscale) aligned ? = subtle or inconclusive. vesicles. A terrestrialanalog might be the horizontalzones 8682 MCSWEEN ET AL.: MARS PATHFINDER ROCKS

Plate 1. Stereo image of Chimp (39 cm tall) producedby combiningtwo right-eye mosaics(two frames mosaicedtogether for the red, and two for the cyan), taken before and after commanding the rover to rotate18 ø in place,while 25 cm awayfrom Chimp. The frameshowing more of the left sideof Chimpwas assignedto the red colorplane of the anaglyph,and the otherframe was assignedto the greenand blue colorplanes (cyan), to producea stereoanaglyph. This imagecan be viewedin 3-D by wearingred-blue glasses. The texturein the upperleft resemblesexfoliation. Prominentbanding is roughlyhorizontal around the lower 2/3 of the rock, wrappingaround the roundednose closest to the camera,and continuesaround the left sideof the rock. These image frames were convertedto linear coordinatesbefore combining them into the anaglyph image: RVR EDR-1253231644-S074032-RIGHT, RVR EDR-1253232878-S074033-RIGHT, RVR EDR-1253233245-S074044-RIGHT, and RVR EDR-1253233615-S074045-RIGHT.

rich in vesicles found in inflated basaltic lava flows, although transport. Yet anotherform of vesiclelayering, on the 10-50 the scaleof theselayers is 10-300cm [Thordarsonand Self, cm scale,has been describedin intrusiverocks [Toramaru et 1996; Cashmanand Kauahikaua,1997; Self et al., 1997]. al., 1996]. The physicsof vesiclelayering in igneousrocks is Inflated flows also exhibit bandingon vertical uplift scarps not well understood,so the scale of such layering on Mars and cracksdue to alignmentof vesiclesand changesin could be different. vesicularity,but this bandingdoes not pervaderocks deeply Only one rock at the landing site, Chhnp, has a structure [Honet al., 1994]. If thePathfinder lineations originated in that strongly resemblesexfoliation (top left side of rock in this manner,all surfacesexhibiting lineations would have to Plate 1), where an outer, possibly xnore highly weathered beexpansion cracks formed during lava emplacement. This is portion of the rock appearsto have spalled off. The rock unlikely, consideringtheir probable abrasion during surfacebeneath the spall either has a higher albedoor perhaps MCSWEENET AL.: MARS PATHFINDERROCKS 8683

Figure 2. Rovercamera image mosaic of Stimpy(25 cm tall), with pits 1-3 cm in diameter.Mosaic is composed of the imageframes with thesePlanetary Data SystemPRODUCT IDs: RVR EDR-1252885649-S070104-LEFT, RVR EDR-1252886142-S070105-LEFT, and RVR EDR- 1252886657-S070106-LEFT.

is rougher and has collected dust more efficiently than the broken off the rock. smootherouter surfacesof the rock. A large crack runs from Rounded bumps on a scale of several millimeters are top to bottom(39 cm). Exfoliationresults from rock expansion common on some Pathfinder rocks such as Wedge, Barnacle and fracturing due to temperaturechanges, mineral expansion Bill, and Shark(Figures 4 and 5 andPlate 2). Bumpscould be due to chemical weathering, or release of internal stresses surface expressionsof crystals, pebbles, cobbles, or rock when overburdenis removed. If exfoliationwas a predominant fragmentsmore resistent to erosion,or theymight have formed weatheringstyle at the landing site in the past, somerounded by aeolianabrasion [Whitney and Dietrich, 1973] or chemical boulders may owe their shapes to spheroidal weathering. etching. If the bumpsare weatheringfeatures, Wedge's bumpy Another example of mechanical breakdown of rocks is texture is the most intensely altered surface of this type. provided by Barnacle Bill, which has small rock fragments Larger(2-4 cm) roundedlumps, like thoseseen on the baseof with similar pitted textures in front of it that appear to have BarnacleBill (Plate2) andBambam, could be roundedcobbles 8684 MCSWEEN ET AL.: MARS PATHFINDER ROCKS

Figure 3. Monochromerover camera image of Squash(11 cm tall). The lobeon the left sideof the rockjuts out -12 cm. Mosaicis composedof two imageframes: RVR EDR-1249070145-N027093-RIGHT and RVR EDR-1249070974-N027094-RIGHT.

in a conglomerate,rounded lithic fragmentsin a volcanicrock, newly generatedsuper resolution IMP ixnages. or concretions. Prince Charmingand Shark (Figure 5) are examplesof 3. Multispectral Imaging of Rocks rocksthat have bumpytextures accompanied by roundedpits, mostly in the 0.5 to 3 cm size range. Theserocks have been 3.1. Overview and AnalyticalMethods suggestedto be conglomerates,with some pebbles still embeddedand others pluckedout to producesockets [Rover A descriptionof the designand operationof the IMP was Team, 1997]. Suchrocks might be the sourcefor the rounded given by Smith et al. [1997b]. IMP is an electronically pebblesand cobblesseen in the soil, which are similarin size shuttered,multispectral binocular imager, in which light from to the bumps. Analogous featureswere not observedin eacheye is divertedby a nfirror througha spectralfilter and to Viking soils, nor were rocks with similar bumpsand pits. a separatehalf of a CCD. The filters are locatedin a 24- Althoughthese curious surface textures may be suggestiveof positionfilter wheel, designedso that eacheye looksthrough conglomerates,they are not diagnosticand could occur in a filter at oppositepositions on the wheel. Threefilters (445, otherrock typessuch as weatheredvolcanic rocks. 671, and 967 nm) are duplicated at oppositepositions for Stimpyand Hassock each have three rock faces oriented in color stereoimaging. The remainingfilters are intendedfor such a fashion to suggestthat they are polygonalprisms atmosphericopacity measurementsor for discriminationof [Rover Team, 1997]. The implicationis that the rocksmay iron minerals,principally low-calciumpyroxenes and ferric exhibit columnarjointing, a structurecommon in volcanic phasesthat are actually found in Martian meteoritesor whose flows. Stereometricmeasurements of Stimpy indicatethat the existenceon Mars is suggestedon thermodynanficgrounds. anglesbetween the threefaces are 120ø + 15ø andthe angles For the nfineralogicfilters, additionalspectral bandpasses are of the faces on Hassock are 135-140 ø + 15ø. These two rocks viewed either by the right eye (480, 531,600, 752 nm) or the could be columnar lavas, but other interpretations are left eye (802, 858, 898, 931, 1003 nm). Spectralmeasurements possible. For instance,they might be crackedrock faces are reconstructedfrom the imagesby coregistrafionof images developedduring dehydration or freeze-thawcycles. throughdifferent bandpasses; the spectrumis "handedoff" from Becauseof the equivocalnature of the structuresand one eye to the otherbetween 752 and 802 nm. Figure 6 shows textures observed in Pathfinder rocks, these observations a representativeIMP spectrumof a soil near the rock Mint cannot be used to specify uniquely whether the rocks are Julep. IMP spectra of rocks and soils share the basic volcanic, sedimentary,metamorphic, in•pactites, or some characteristicsof a ferric absorptionedge extendingfrom 440 assortmentof rocks with varying origins. However, some nm (the shortestwavelength filter) to a reflectivitymaximum at rock types,such as highly stratifiedsedin•ents, appear to be 750 or 800 nm, and graduallydecreasing reflectivity from the ruled out. The texturalinterpretations are complicatedby the nmxin•um to the longest wavelength filter at 1000 nm. likelihood that multiple processeswere responsible,and the Differencesamong the spectramay be quantifiedusing several originalrock texturesthat carrypetrogenetic information have parameters. almost certainly been overprinted by aeolian abrasion, Thereflectance (designated R*) is measured relative to the chemicalweathering, and coatingsof dust. Theseissues might reflectance of the onboard calibration target which was be resolvedin the future,if diagnostictextures are revealedin characterizedin detail during laboratory studies[Reid et al., MCSWEEN ET AL.' MARS PATHFINDER ROCKS 8685

:.. ,;

...?½,

Figure 4. Superresolution IMP image of Wedge (30 cm tall), producedby combining25 red-filter left-eye lYames.The individualframes were enlargedby 1000%, sharpened,manually co-registered at 1/10 pixel scale, and then co-added,giving equalweight to eachframe. This produces,in effect, a super-resolutionimage that is sharperthan any individualfrmne. Soil or dustappears to be plasteredupon the rock face, more heavily on the left sideof the rock thanthe right. Imageframes used as input: IMP_EDR- 1248443296-REGULAR-0199020004, IMP_EDR- 1248443332-REGULAR-0199020008, IMP_EDR- 1248443380-REGULAR-0199020012, IMP_EDR- 1248443428-REGULAR-0199020016, IMP_EDR-1248443476-REGULAR-0199020020, IMP_EDR-1248443525-REGULAR-0199020024, IMP_EDR- 1248443573-REGULAR-0199020028, IMP_EDR- 1248443621-REGULAR-0199020032, IMP_EDR- 1248443669-REGULAR-0199020036, IMP_EDR- 1248443717-REGULAR-0199020040, IMP_EDR- 1248443765-REGULAR-0199020044, IMP_EDR- 1248443812-REGULAR-0199020048, IMP_EDR- 1248443860-REGULAR-0199020052, IMP_EDR- 1248443908-REGULAR-0199020056, IMP_EDR- 1248443957-REGULAR-0199020060, IMP_EDR- 1248444004-REGULAR-0199020064, IMP_EDR- 1248444052-REGULAR-0199020068, IMP_EDR- 1248444100-REGULAR-0199020072, IMP_EDR- 1248444148-REGULAR-0199020076, IMP_EDR- 1248444196-REGULAR-0199020080, IMP_EDR- 1248444244-REGULAR-0199020084, IMP_EDR- 1248444292-REGULAR-0199020088, IMP_EDR-1248444340-REGULAR-0199020092,IMP_EDR-1248444388-REGULAR-0199020096, and IMP EDR- 1248444436-REGULAR-0199020100.

1997] Reflectanceat a red (670 or 750 nm) wavelengthis [Shermanet al., 1982; Morris et al., 1985;Bishop, 1995]. representativeof the intrinsic brightnessof rock and soil Variations in the red/blue ratio are functions of the materials,plus photometric effects. Reflectanceat a blue (440 concentrationand mineralogy of the ferric-bearingphases, nm) wavelengthis in generalmuch less variablebetween particle size of both compositepolymineralogic aggregates Martian rocksand soils [Soderblom,1992], but certainrocks and the included particles of ferric minerals, and the at the Pathfindersite are characterizedby atypically high or mineralogy and structure of the composite particles low blue reflectance. The red/blue reflectance ratio indicates [Wendlandtand Hecht, 1966;Kortum, 1969;Sherman et al., the magnitudeof the red slopein the regionof strongferric 1982; Morris et al., 1985; Morris and Lauer, 1990]. R.V. electronic absorptionsby many ferric-bearing minerals Morris et al. (unpublishedmanuscript, 1998) showthere is a 8686 MCSWEEN ET AL.' MARS PATHFINDER ROCKS

Figure 5. Monochromerover cameraimage of PrinceChanning (lower right, 20 cm wide), Shark(upper left, 47 cm tall), and Half Dome (upperright, 51 cm tall). The shinyrounded pebble in the soil to the left of Prince Channing is 3 cm in diameter. Prince Channinghas a bumpytexture with 0.5-3 cm size hollows (dark tops and bright bases)and 0.5 cm protmsions(bright tops and dark bases). Shark'ssurface has a similar bumpy texture. Mosaicis composedof six imageframes: RVR EDR- 1252969487-S071070-LEFT, RVR EDR- 1252969661-S071071-LEFT, RVR EDR-1252969833-S071072-LEFT, RVR EDR-1252970002-S071073-LEFT, RVR EDR-1252970175-S071074-LEFT, and RVR EDR-1252970356-S071075-LEFT.

general correlation between the red/blue ratio and iron band is characterized here relative to a continuum between oxidation state in natural compositeparticles (oxidatively 800 and 1000 nm. Most plausibleferric mineralshave a band altered volcanic tephra). The peak reflectanceof typical center at or beyond 900 nm, whereas hematite has a band Martian spectra [Mustard and Bell, 1994], as well as the centernear 860 nm. Some ferrousminerals, most notably low- spectraof' most ferric minerals[Sherman et al., 1982;Morris calciumand high-calciumpyroxenes, also exhbit a diagnostic et al., 1985;Bishop, 1995], occursnear 750 ran. However,in band centerednear 900-930 nm [Cloutisand Gaffey, 1991]. a few minerals(most notably ferrihydrite and maghemite),the Three main techniqueswere used to assessspectral peak occursinstead at or near 800 nm (Figure7). A subsetof variations in rocks at the Pathfinder landing site. First, rocks and soils at the Pathfinderlanding site, includingthe spectral parameter images were calculatedfrom calibrated Mint Julep soil shown in Figure 6, exhibits the longer- mosaics. Those shown in Plate 3 are produced from the wavelengthreflectance peak. Two crystal-fieldabsorption "gallery panorama" (2:1 compression) because of its bandsare observedat the wavelengthsof IMP at the landing continuousspatial coverageand the large amountof spectral site. The strongerone, centered around 530 nm, is manifested variancethat can be representedusing its 440 nm, 530 nm, and as an inflectionor "kink" in the stronglyred-sloped spectrum. 670 nm channels.Second, using the parameterizedimages as a A weak kink is characteristicof many ferric minerals,but guide to spectralheterogeneities at the site, representative amongthose either found in Martianmeteorites or predictedto spectrawere extractedfrom variousrocks and associatedsoils haveformed under plausible conditions at the Martiansurface to characterize spectral properties in greater detail. To [Gooding,1978, 1992; Burns, 1992, 1994],only ferrihydrite, highlight the differences between spectra, they are shown maghemite,and especially hematite have a strongkink (Figure ratioed to an averageof typical bright red drifts at the site. 7). Depth of the kink is characterizedhere relativeto a Spectral variations were explored in detail using these type continuumbetween 440 and 670 nm. The other absorption, spectra, utilizing various methods such as scatterplotsof which is presentonly in a subsetof the observedrocks and spectralparameters and variousmultivariate techniques. We soils, is centeredat 900-930 nm. The depthof this "900 nm" used spectra taken from both "multispectral spots" (no MCSWEEN ET AL.: MARS PATHFINDER ROCKS 8687

Plate 2. Super resolutionstereo image (see captionfor Figure 5) of BarnacleBill (22 cm tall), producedby combiningseven fight-eye frames and eight left-eye frames from the "superpanorama"taken by the IMP camera. The left-eye superresolution image was assignedto the red color plane of the analglyph,and the fight-eye image to the greenand blue planes(cyan), to producea stereoanalglyph. Two setsof imageswere processed in this mannerand mosaicedto coverthe fifil extentof the rock. The pits on BarnacleBill are 0.25 to I cm in diameter. Dark notchesor tiny ledgeson the left side of the rock constituteone subtle set of lineations. Othersets of very faint lineationsare orientedat 45ø clockwisefroin vertical on the top middleand top fight of therock and at 140ø clockwisefrom vertical on thefront and upper left rockfaces. These image frames were used as input: IMP EDR-1248281345-REGULAR-0182010120, IMP EDR-1248281345-REGULAR-0182010121, IMP EDR-1248281367-REGULAR-0182010122, IMP EDR-1248281367-REGULAR-0182010123, IMP EDR-1248281393-REGULAR-0182010124, IMP EDR-1248281393-REGULAR-0182010125, IMP EDR-1248281454-REGULAR-0182010126, IMP EDR-1248281454-REGULAR-0182010127, IMP EDR-1248281504-REGULAR-0182010128, IMP EDR-1248281504-REGULAR-0182010129, IMP EDR-1248281554-REGULAR-0182010130, IMP EDR-1248281554-REGULAR-0182010131, IMP EDR- 1248281607-REGULAR-0182010132, IMP EDR- 1248281607-REGULAR-0182010133, IMP EDR-1248281660-REGULAR-0182010134, IMP EDR-1248281694-REGULAR-0182010136, IMP EDR-1248281694-REGULAR-0182010137, IMP EDR-1248281716-REGULAR-0182010138, IMP EDR-1248281716-REGULAR-0182010139, IMP EDR-1248281734-REGULAR-0182010140, IMP EDR-1248281734-REGULAR-0182010141, IMP EDR-1248281757-REGULAR-0182010142, IMP EDR-1248281757-REGULAR-0182010143, IMP EDR-1248281797-REGULAR-0182010144, IMP EDR-1248281797-REGULAR-0182010145, IMP EDR-1248281848-REGULAR-0182010146, IMP EDR-1248281848-REGULAR-0182010147, IMP EDR-1248281903-REGULAR-0182010148, IMP EDR-1248281903-REGULAR-0182010149, and IMP EDR-1248281962-REGULAR-0182010150.

compression)and the "superpanormna"(2:1 compression) 3.2. Spatial Variations in Rock Spectral Properties [Smithet al., 1997a]. For consistency,only thosefrom the superpanoramaare shownhere. Finally, basedon the results The spectral properties of rock surfaces facing the of analysisof the type spectra,unit classificationmaps were Pathfinder lander show strong variations azimuthally. derived basedon thresholdingof spectralparameter values. Southwestof the lander, facing the prevailing northeastwind Suchmaps are only a tool for understandingvariations at the direction [Smith et al., 1997a], rock surfacesare generally site; their boundariesare abrupt,in contrastto the typically similar spectrally. Comparedto drift they exhibit a similar low gradational spatial variations in spectral properties. blue reflectancebut a lower reflectanceat longer wavelengths, Superpanoramadata were usedfor this purposebecause they a lower red/blue ratio, and a weaker kink. Northeast of the are the only data coveringsubstantial swaths of the landing lander, on rock facesaway from the prevailingwind direction, site simultaneouslyin all of the visible and near-infrared a greaterdegree of heterogeneityis observed(Plate 3). First, image channelswhich we found necessaryto discriminate gray portionsof many cobblesand small boulders,having a rock and soil units. All data usedwere calibratedusing CCD variety of shapes,resemble spectrally rocks on the southwest correctionsand measurementsof the IMP calibrationtargets as side of the lander (blue arrows). Second, red portions of describedby Reid et al. [ 1997]. cobbles and small boulders exhibit elevated red/blue ratios 8688 MCSWEEN ET AL.: MARS PATHFINDER ROCKS

0.35 _

0.30 ....k L ...... -.•- Wavelengthof. 0.20[ / reflectancepeak .....-•'"Depth epthof530-nm band("kink") 0.05

0.00 400 500 600 700 800 900 1000 1100 Wavelength in Nanometers

Figure 6. IMP spectrumof brownsoil nearthe rock Mint Julep,showing the principalparameters used to characterizereflectance spectra of rocksand associatedsoils. and a stongerkink (bright red arrows). Commonly,both whereasits downwindside is redderand similarto large,red kinds of surfacesoccur on the samerocks, with gray portions rocksin thefar field. Thereare also rare examples of rockswith concentratedat edgesor on the verticesbetween facet-like dominantlyred upwind facesto the southof the lander, and faces. Third, a few tabular, pink rock-like masseshave spectrallyanalogous soils (discussedbelow) occur at all red/blueratios nearly as high as in drift, but exceeddrift in azimuths. brightness,especially at shorterwavelengths (pink arrows). Illuminationgeometry does causetemporal variations in Fourth,large, typically rounded boulders in the far field (dark the measured properties of rock surfaces, but on well- red arrows)have as high a red/blueratio andas stronga kink illuminatedrock surfacesthese changes are smallerthan the as drift, but are darker at all wavelengths. These maroon variation between spectral classes. Large degreesof bouldersalso have facets which exhibit the gray reflectance reddeningdue to sky radianceare observedonly on highly propertiestypical of upwindrock faces. Yogi is one of the shadedrock faces. For example,the superpanoramaimage of bestexamples: the upwindside of the rock is relativelygray, Yogi showsa small "bluer," well-illuminated face on the left

8 Drift ß Hematite .... ß.... Schwertmannite .... e .... Goethite = Maghemite [] Ferrihydrite .... ----- Jarosite ...... Pyroxene (pigeonite)

_ _

07 _- •, .... , , -- - , , 06 - ' "o - ,t ,-,--' '--t,, ,o.. _ - oID 05 ,,,ß - .... ,-.•.. ,.,' e--,-e.--•- •,,• _ t-• , ' ,.,'•.-' '•.. _ . _ - 'UID 04 •;,,' ' --

rr 03 ,.,• -_ i•', --• _ 01 : - : O0 ' 400 500 600 700 800 900 1000 1100 Wavelength in Nanometers Figure 7. Laboratoryspectra of typicalferric minerals which are foundin SNC meteoritesor may be expected thermodynamicallyat the Martian surface,resampled into the bandpassesof IMP. Note that hematite, maghemite,and ferrihydrite (all solid lines) exhibit the observedinflection in the three channelsused to calculate the 530 mn "kink." MCSWEEN ET AL.: MARS PATHFINDER ROCKS 8689

(a) Enhanced color (b) Red/blue Low High

Plate 3. Four renditionsof color propertiesof a portion of the quadrantnortheast from the lander. Note the variability of the spectralcharacter of downwindrock faces. (a) Contrast-enhancedvisible color, constructed from 440, 530, and 670 nm image mosaics. Coloredarrows show examplesof t_,'ayrocks (blue arrows),red rocks(bright red arrows),pink rocks(pink arrows),and maroonrocks (dark red arrows). Representativedrift is shown with a white arrow for comparison. (b) Red/blue (670/440 nm) color ratio, shown in false color where redderhues representhigher red/blue ratios, overlain on 670 nm reflectance. (c) Depth of the 530 nm kink, shownin false color where redderhues represent a strongerabsorption, overlain on 670 nm reflectance.

(d)440 nm reflectance inunits of R* ß Notethe blandness ofthe scene, save for atypically dark large boulders (dark red arrows)and atypicallybright, tabular, rock-like masses in the foreground(pink arrows).

side of the rock; this same area appearsredder in the gallery •nostexposed to prevailingwinds. Consequently,much of the panaoramawhen that portion of the rock surfaceis relatively spectral heterogeneityof the rocks is apparently only "skin shaded. deep." This interpretationis consistentwith many similar The azimuthal variation in rock spectral properties is results from imaging studiesof rock surfacesat the Viking important evidence for understandingheterogeneity. The Lander sites [Evans and Adams, 1979; Sharp and Malin, relative homogeneityand gray color of upwind rock faces, the 1984; Guinness et al., 1987], including longer-term reddercolor (approachingthe color of drift) of the downwind observationsof changesin dust coatingson many rock facets faces of large boulders, the presenceof gray portions on [Gu#messet al., 1982]. generally red rocks, and the concentration of such gray Although the upwind/downwind dichotomy in rock surfacesat edgesand verticesall evokepreferential removal or spectral properties is the most obvious source of spectral lack of depositionof thin, oxidized coatingsfrom those faces heterogeneity, at least three other types of spectral 8690 MCSWEEN ET AL' MARS PATHFINDER ROCKS

(a) Flat Top (b) Ginger (c) Boris

Plate 4. Spectralheterogeneity not related to wind direction.(a) Little Flat Top, a tabular boulder in the Rock Garden. Note the brighteningand reddeningof the upper surfacedue to accmnulateddrift, and the redderband along the lowermostseveral centimeters of the front face. (b) Ginger, a 15-cmcobble southeast of the lander. (c) Boris, a 20-cm cobblesoutheast of the lander

heterogeneitiesare observed.Most commonis a brightening Despite surfacetextures suggestive of layering or included and reddenhagof tabularor downwindsurfaces of relatively fragments, describedabove, very few rocks exhibit color grayblocks, such as Flat Top (Plate4a). In somerocks such as variationsthought to correspondto compositionaldifferences BarnacleBill, the brighteningand reddeningis concentrated within unweathered rock, such as parallel banding or in pits on rock surfaces.On BarnacleBill and a numberof mottling. (Becauseof the necessityfor registrationof images other rocks at the north end of the Rock Garden,the brighter, acquiredthrough different filters, color heterogeneitiesin reddertop surfacesthicken in the downwinddirection into rocks are less resolvable spatially than are morphologic wind tails, indicatingthat either airfall drift accumulatedon features.) This is true even of rocks suggestedto have the rock or was preservedas an erosionalremnant of a pre- "sockets"formed by loss of pebbles from a conglomerate exisitingthicker drift layer [Smithel al., 1997a]. Severalgray [Rover Team, 1997]. However,two rocksto the southof the boulders to the southwest of the lander in the Rock Garden lander exhibit mottled spectralproperties absent from other also exhibit a redder horizontal band along the lowermost rocksha the near field. Ginger, a lobatecobble approximately several centimetersof the rock (Plate 4a). This banding is 15 cm in diameter, exhibits rounded, gray lobes a few interpretedas a thin remnantof a preexistingdrift layer that centimetersin size separatedby a dark, very red groundmass has been nearly removed by aeolian erosion [Smith et al., (Plate4b). Boris,a subangularrock-like mass about 20 cm in 1997a]. size, exhibitssubangular patches --5 cm in size of differing The leastcommon type of spectralheterogeneity is mottling spectral classes(brighter and redder through darker and of the northeastfacing, typically scouredsides of cobbles. grayer),separated by a gray groundmass(Plate 4c).

Table 2. ParameterizedProperties of Rock SpectralClasses

Rock Spectral 0.75 Red/Blue 0.53 gm ResidualDepth Occurrence Class Reflectance,% (0.67/0.44gin) Absorption of 0.9-Fm Ratio ("Kink"), % Absorptiona, %

Gray 10-20 <_2.9 0-7 <0.5 Pebbles,small cobbles,upwind faces of boulders Red 15-30 2.9-4.5 7-18 _<0.5 Pebbles, small cobbles, downwind faces of boulders Pink >35 >_4.5 18-28 <0.5 Tabular rock-hke masses,crust on pebblesand cobblesin soil Maroon 15-30 >_2.9 -15-28 >0.5 Upwind facesof large rounded (not used boulders,rare on cobbles for classification) aDifferencebetween observed absorption depth and that predicted from linear fit betweenred/blue and absorptiondepth in main spectraltrend. MCSWEEN ET AL.: MARS PATHFINDER ROCKS 8691

Gray' Red' Pink: Booboo ß Half Dome Stimpy ß Broken Wall .... ß.... Scooby Doo Shark ß Wallace

''''1''''1''''1''''1''''1''''1'''' _

_ 0.40 _ Primary Spectral .ß___..... •. _ _

Trend ß . '*-*-.,__ß - ß - - _ ß _

ß _ ß _ ß _ 0.30 - , _

ß , iiImmmmmmmmmmlmmmmmlmllmmmm (D - , _ ,

r- _

_ • o• 020. _ _

_ _

O.lO _

_ ...... Mean Bright Red Drift •

_ 0.00

400 500 600 700 800 900 1000 1100 (a) Wavelength in Nanometers

Gray' Red: Pink: i 1 Booboo ß Half Dome Stimpy ß Broken Wall .... ß.... Scooby Doo Shark ß Wallace

,,,•1,,,,1•,,,i•,, •1,•,, i,,,, i,,,•- o Mean Bright - 1.80 o Red Drift

_ > .'.m_ *' ' ' ...... A---A._A_.•_._A -- 1.40 _ g _

-•.? 1.oo _ _

o (1) <• 0.60 -_Primary

- Spectral Trend _- 0.20 400 500 600 700 800 900 1000 1100 (b) Wavelength in Nanometers

Figure 8. Representativespectra of the threemajor rock spectral classes (gray, red, pink) at the Pathfinder landingsite. (a) Reflectance inunits of R*. (b)Ratioed toan average oftypical drifts at the site ("mean bright red drift"), highlightingspectral differences.

3.3. Rock Spectral Classes Garden (Figure 1). Peak reflectanceoccurs at 750 nm; at longer wavelengths, no well-defined absorption band Each of the four major spectralclasses of rock surfacesis minhnumis observed.The low reflectance,low red/blueratio, distinguished by characteristic properties and modes of and weak or absentkink in gray rock surfacesare consistent occurrence, as described below. Parameterized spectral withweakly weathered Fe2+-containing rocks. The lack of a propertiesfor eachclass are summarizedin Table 2. well-defined absorptioncontrasts with most comparably Gray rock surfaces(Figures 8a and 8b) are observedon colored (dark) regions of Mars, which exhibit a "l-gm" rocksof all sizes. Relative to drift, gray rocksare comparable absorptionband severalpercent in depth,centered at 950-980 in reflectanceat short wavelengthsbut are darker at long nm [Mustardet al., 1993, 1997;Bell et al., 1997]. The same wavelengths. The type specimen, Shark, is in the Rock Martian dark regionstypically also exhibit an absorption 8692 MCSWEEN ET AL.' MARS PATHFINDER ROCKS

Maroon: .... * .... Valentine .... e---- Yogi Weathered .... *---- Seal .... e---- Lamb Pink: .... •---- BoobooWeathered .... e,--- Ginger .... ß.... Bakers Bench 0,50 ''''l .... - Secondary Spectral Trend

0.40 - _

;-'-, _ n' _ 0.30 (D _ &' o _ ,' ,•" ..e--..e.... e,-.,.e,.-e-. $..e _ ,• • ,..''..,---'--....,..,. - : ,' ..,...-..,,...•...... ,..•,;.•:• 0 20 - ,' e'.-',-e,-".. e-- -e--- e--•- -•,.,• ' - ,' '?,';'"•"- -•---*-- -*--*--*--Z-•

- ,,' • .;'.."'•".:::.*''

_ 0.00 • i • i I • • i i I • • • • I .... I .... I . , , • I • • i • 400 500 600 7O0 800 900 1000 11oo ( a ) Wavelengthin Nanometers

Maroon' .... * .... Valentine .... •--- Yogi Weathered .... *---- Seal .... e---- Lamb Pink: .... •---- BoobooWeathered .... e,--- Ginger .... ß.... BakersBench 1.80 .... I .... I .... • .... • .... I .... I ....

- Secondary Spectral _ o 1 60 - Trend .& .... ß. _ß_. -&--ß- - (D .• _ . "• ' 'ß - > • - - -,•• 140 _ -•

,,,,1,,,,I,,,,1,,,,I,,,,I,,,,I 0.40 400 500 600 700 800 900 1000 1100 (b) Wavelength in Nanometers

Figure 9. Representativespectra of maroonrocks at the Pathfinderlanding site, compared with pink rock. (a) Reflectancein units of R*. (b) Ratioedto anaverage of typicaldrifts at thesite, highlighting spectral differences. band near 2100 nm [Mustard et al., 1993, 1997]. The reflectancepeak is at 750 nm, and thereis no evidenceof a positions,centers, and relative depthsof the two absorption near-infraredabsorption band. bands in such regions are diagnosticof pyroxeneswith Comparedto grayrocks, the higherreflectance, higher variable calcium and moderate iron contents,such as those in red/blue ratio, and strongerkink in red rock surfacesall basalticshergottites [Singer and McSween,1993; Mustard et suggestthe presenceof greateramounts of ferric minerals. al.,1993, 1997]. However,the lack of a well-definednear-infrared absorption Red rock surfaces(Figures 8a and 8b) alsooccur on rocksof impliesthat suchminerals must be extremelyfine-grained or all sizes,typically on thosethat exhibitgray surfaceson other poorlycrystalline. On the samerocks, the grayproperties of faces. Comparedto drift, red rocksare higherin reflectanceat surfaces more exposed to abrasion suggest that the short wavelengthsbut are comparablein reflectanceat long brightening,reddening ferric component occurs as a verythin wavelengths.The type specimen,Broken Wall, is at the north coatingor rind either of drift or of an oxidizedweathering end of the Rock Garden (Figure 1). As with gray rocks the product. MCSWEEN ET AL.' MARS PATHFINDER ROCKS 8693

Bright Red Drift' o Behind Barnacle Bill Brown Soil: o Left of Phot. Flats .... <>----Left of Mint Julep o On Flat Top .... 4>---- Near Yogi O5O Drifts and Soils -

_ ß ß,..-KY'"-

c- ß :,---4>---- <3..•. _•. (•- _½ - 20

10 Reflectance maximum in bright red drift

O0 400 500 600 700 800 900 1000 11 O0 Wavelength in Nanometers Figure 10. Comparisonof spectraof subtlydistinct drift materials.Bright red drift (solidlines) has its reflectancepeak at 750 nm andlacks an obviousinfrared absorption. Brown soil has a peakat 800 nm and exhibitsa weak infraredabsorption centered near 900 nm.

Pink rock surfaces(Figures 8a and 8b) occur on several ratherthan 750 nm, and thereis a weak ferric absorptionband tabularrock-like massespartially buried by drift, as crustsin at 900-930 nm. Lamb shows the most pronounced the soil eastof Yogi, and on a numberof pebblesbetween the developmentof the absorptionband and the shift in peak lander and Yogi. The type specimen,Scooby Doo, is north- reflectancetoward longer wavelengths. Many rocks with northeast of the lander (Figure 1) nearby to similar maroonfaces, such as Yogi and Booboo,also exhibit gray occurrences. Relative to drift, pink rocks are higher in faceson their upwindstirfaces. reflectanceat all wavelengths,especially short wavelengths, The highred/blue ratio and strongkink of maroonrocks hence the designation"pink." They are readily distinguished andthe presenceof graysurfaces on the rocks'upwind sides from backgroundmaterials in blue (440 nm) images,in which all suggestthat the maroonspectral properties originate in a they are anomalouslybright (Plate 3d). The reflectancepeak ferric-rich coating, as in the case of red rocks. The is at 750-800 ran. Scooby Doo was a site of rover soil developmentof an infraredabsorption in somemaroon rocks mechanicstests [Rover Team, 1997], and no abrasionof the could,by itself,be takento indicatesimply a largerparticle surface was observedwhen the rover's right rear wheel was sizeor greatercrystallinity of the ferricmineral than in the placed on the rock and turned in place. This result is drift or on the surfacesof red and pink rocks. However,the consistentwith a material at least as induratedas a hardpan magnitudein shiftin thewavelength of thereflectance peak is soil. The similarity of pink rock to typical drift, both in its not an expectedeffect of texturalvariations, and it is not an spectralproperties and elementalabundances [Rieder et al., observedeffect of variationsin ferric mineral crystallinityor 1997a], and its mechanicalstrength suggest that the surfaces particlesize [Bell et al., 1990]. Rather,the wavelength shift of rocksof this groupare encrusteddrift. mayindicate a mineralogicdifference between maroon surfaces Maroon rock surfaces(Figures 9a and 9b) occurmostly on and the majorityof otherred materials,which have their large, rounded boulders including Yogi and comparable reflectancepeak at 750 nm (cf. Figure7). bouldersin the far field. The type specimen,Seal, is a large boulder in the far field on the left side of Plate 3. In addition, maroonsurfaces occur on portionsof two smallerrocks in the 3.4. Relationship of Rock Spectral Classesand Drifts near field, Ginger and Lamb. Relative to drift, maroonrocks are darkerat all wavelengths. Spectralshape is very similar to The pervasivenessof drift at the landingsite, the occurrence that of pink rocks, the main difference being in reflectance. of different spectral classes on the same rocks, and the Comparedwith the more cormnonred rocks, maroonrocks are evidencefor rock spectralvariations arising from thin surface darker at short wavelengths but of comparable infrared coatingsor rinds all suggestrelationships of rock spectral reflectance(Figures 9a and 9b). Maroon rocks are readily classeswith each other and with drift. Suchrelationships distinguishedfrom backgroundmaterials in blue (440 nm) were exploredat the Viking Lander sitesby Adams et al. images,in which they appearanomalously dark (Plate 3d). In [1986] and Guinnesset al. [1987, 1997], who found similar the near-infrared,many maroon rocks are distinct from other evidencefor strongspectral effects of surfacecoatings or finds rocksat the Pathfindersite: the reflectancepeak is at 800 nm on rocks,as well as evidencefor certainlocally deriveddark 8694 MCSWEEN ET AL.: MARS PATHFINDER ROCKS

[] Gray Rocks * Pink Rocks o Bright Red Drift ß Red Rocks ß Maroon Rocks O Brown Soil 3O

25 VisibleCorrelationWavelength of SpectralProperties 2O

00 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 (a) 670/440 nm Reflectance Ratio

[] Gray Rocks * Pink Rocks o Bright Red Drift ß Red Rocks ß Maroon Rocks o Brown Soil 55

o ß ß ß :• 50 o • ß o . ¸ • 45 o

-- 40 • • - • - _ _ • _ _ e 35 -

• - ß _- ß _ E - • - c 30 - ß -

• 25 n -

• 20 - n n n Relationship of - Reflectance and Color -

0.•0 0.•5 0.20 0.25 0.30 0.35 0.40 { b) •70 nm Bef[ectance(B •) Figure 11. Comparisonof parameterizedspectral prope•ies of •e fourspectral classes of rocks,bright red drift,and bro• soil. (a) Re•lue (670/440•) ratioand dep• of the 530m absoftion,or k•. •e close co,elationshows that the reddeningphase or phasesexhibits this feature, which is charactehsticof a subset of femc mineralsincluding hematite, maghemite, and fe•ydfite. (b) Red(670 m) reflectanceand re•lue ratio. Two trendsare evidentm the data:"phmau trend"gray, red, andp•k rocks,and a "seconda•trend" consistingof maroonrocks and bro• soils. (c) Strengthsof the 530 and900 mn absoftions.In pr•a• trendrocks, calculated absoftion strengths are inverselyrelated, whereas in secondau•end rocksand in driftthe strength of the 900nm absorptionvaries without a co•espondingch•ge in s•engthof theki•. Seconda••end rocks and mils f•her fromthe intersection of the two •ends e•ibit a reflectancemaximin at 800m. (d) Re•lue ratioand s•ength of the900 mn absoftion. •e prima• andseconda• special trends appe•as • Figure11 c. •e linearfit of the•o variablesfor phm• •endrocks is sho•, andbound•ies of unit classifications in Plate 5 are indicated.

soilsthat were interpretedas abrasionor spallationproducts absorption. However, there are restricted locations at which from adjacentrocks [Sharp and Malin, 1984]. drift has its reflectancepeak at 800 nm and exhibits a weak In characterizingthese types of relationshipsat the infrared absorption at 900-930 nm. These attributes are Pathfinderlanding site, it is critical to recognizesubtle but analogousto thoseof many maroonrocks. Thesebrown soils importantnear-infrared spectral differences within drift (Figure occur in the lee of Yogi, in associationwith Lamb, and north 10). Mostdrift at thesite, which we call "brightred drift," has of the lander surroundingthe partially buried boulder Mint a reflectance maximum at 750 nm and lacks an infrared Julep. An exposure of a soil unit with a similarly strong MCSWEEN ET AL.: MARS PATHFINDER ROCKS 8695

ß Gray Rocks ß Pink Rocks o Bright Red Drift ß Red Rocks ß Maroon Rocks O Brown Soil OO4

Strengths Secondary Trend: - - 003 •- Streng• Se;;•) dnar•Y•!'er;• d: - •002 I-" ßo . %, ooo ß 0% .

-0 01 ß PrimaryTrend:

-0 02 0.00 0.05 0.10 0.15 0.20 0.25 0.30 (C) 530 nm Absorption

ß Gray Rocks ß Maroon Rocks ß Red Rocks o Bright Red Drift ß Pink Rocks O Brown Soil 0.04 Spectral classification ß 0.03 _ Maroon o ß ß • 002 ß Q. ß ß _ o 0 Gray o <1: O.Ol

E _

o 0.00 o ß Red'' ' . o ß

-0.01 ß Primarytrend ' - . ß linearfit / Pi n'k'- ß ß -0.02 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 (d) 670/440 nm Reflectance Ratio

Figure 11. (continued)

visible-wavelengthspectral curvature was detectedat the Figure1 lb showsthe relationshipbetween reflectance and Viking Lander2 site [Guinnesset al., 1987], and was red/blueratio. Most rocks,which belong to the gray,red, and interpretedas evidencefor hematiticallyaltered volcanic pink classes,exhibit an increasein red/blueratio (and kink) tephra. with increasing reflectance, indicating that they are Key spectralparameters indicating the relationships brightenedby a ferric component.Among maroon rocks and betweenrock groupingsand drifts are reflectanceand the both kinds of drift, there is a distinctlydifferent relationship strengthsof absorptions. Figure 1l a showsthat red/blue between reflectanceand spectral shape: both the red/blue ratio and strengthof the 530 nm absorptionare highly ratio and kink are largely independentof reflectance,which correlated,indicating that the ferric phaseresponsible for variesby a factor of >_2with little variationin spectralshape reddeningrocks also accountsfor the 530 nm absorption. (hence,spectra of maroonrocks ratioedto drift are nearly Thisrelationship does not uniquelyindicate the identityof horizontallines as in Figure 9b). the dominantferric mineral(s), but it doesfavor phases with Figure 1l c showsthe relationshipbetween calculated strongkinks (ferrihydrite, maghemite, and especially hematite) strengthsof the crystalfield absorptionsat 530 nm and 900 and precludethose lacking a significantkink (especially nm. Among gray, red, and pink rocks and much bright red goethite). drift, nominalstrengths of the two absorptionsare inversely 8696 MCSWEEN ET AL.' MARS PATHFINDER ROCKS

(a) Enh nced color (b) Red/blue Low High

Plate 5. Associationof the primary and secondaryspectral trends with different rock and soil environments in the quadrantnorthwest of the lander, containingYogi and several other large, round boulders. (a) Contrast-enhancedvisible color, constructedfrom 440, 530, and 670 nm imagemosaics. Red arrowsshow the largestboulders partly or completelycovered with maroonsurfaces. (b) Red/blue(670/440 nm) color ratio, shownin false color where redderhues represent higher red/blue ratios, overlainon 670 nm reflectance. (c) False color representationof depth of the "residual"900 nm absorption. For the primary spectraltrend, a linear fit betweenred/blue and 900 nm band depth was determined. Using this relationship,a "predicted" band depthwas calculatedfor the red/blueratio of eachpixel, and subtractedfrom the observed900 nm band depth. Low residualsare shown in bluer hues, and the highestresiduals in redder hues. Materials on the primary spectraltrend appearin shadesof blue and green,and materialson the secondaryspectral trend in shadesof red. The imagebase is 670 mn reflectance. (d) Spectralgroupings of rocks and spectrallysimilar soils, derivedby thresholdingred/blue ratio as in Table 2. Pink rocks and bright red drift were separated from maroon rocks and associatedbrown soils basedon the latter's residual 900 nm band depth of >0.5%. Gray rocksare coloredin blue, red rocksin yellow, pink rocksand bright red drift in gray, and maroonrocks and brown soils in red. MCSWEEN ET AL.: MARS PATHFINDER ROCKS 8697 related. This result indicates greater concavity in the near- High residualsare also found in some brown soils, in small infraredportions of spectraof grayer,less altered rock surfaces, depositsof drift in the lee of Yogi, and in subsurfacesoils as expectedfrom a pyroxene-likeabsorption, even though a exposedby rover wheel tracks. band minimum is not resolved. This curvature was confirmed The unit classificationmap is shownin Plate 5d. Because by fitting the near-infrared(L750 nm) reflectancesof rocks the classification scheme is blind to differences in reflectance using a second-orderpolynomial. All except the reddestof between rocks and soils, bright red drift classifies as pink the gray rocks, Wedge, e,•ibit a consistentupward curvature rock, brown soil classifiesas maroon rock, and there may be and severalapproach zero slopeat 1000 nm. This curvature crossoversof maroon rocks into the pink category. Four couldresult from a pyroxene,olivine, or glassabsorption even important conclusionsarise from this exercise: First, rocks though a band minimum is not resolved, but it could not and soils forming the primary spectral trend (gray, red, and result from spectral continuum effects of a ferric coating pink rocks,some bright red drift) dominatethe scene. They are [Singer and Roush, 1983; Fischer and Pieters, 1993] the primary constituentof the optical surface,and materials becausethe correlationwith red/blue ratio is of the opposite forming the secondarytrend (maroonrocks and brown soils, senseto that predictedfor a ferric coating. Figure 1l d shows somebright red drift) are uncommon. Second,red and pink the relationship between red/blue ratio and 900 nm band surfacesor accumulationsof bright red drift are commonon depth. Due to the high correlationbetween kink and red/blue rocks of all sizes, shapes, and presumably ages. In other ratio, essentiallythe same relationshipsare observedas in words, whatever processesformed the primary spectraltrend Figure 11c. have actedupon all materialsin the scene. Third, maroonrock The correlatedspectral parameters among gray, red, pink surfaces are nearly restricted to large, typically rounded rocks and bright red drift and their predominanceat the site bouldersinterpreted previously [Smithet al., 1997a] to be the leads us to associatethem into a "primary spectral trend." older of two generationsof rocks at the site. In other words, Examinationof the spectrain Figure 8a showsthat rocks and processesforming the secondaryspectral trend seem to have drift in the primary trend also have in commona reflectance had little effect on youngerpebbles and cobblesthroughout peak at or near 750 nm. Among maroonrocks, brown soil, and the scene. Fourth, disturbedsoils exposedin the rover tracks somebright red drift, a secondtrend is evidentin Figures11 c share spectral affinities with maroon rocks. The spectral and 11d, in which kink and red/blueratio are nearly invariant differencebetween the uppermostdrift and the subsurfaceis but 900 nm absorption strength increases to as much as concordantwith soil mechanicsresults [Rover Team, 1997], several percent. These materials we associate into a indicatinga lessercohesion and finer grain size in the "secondary"spectral trend. Fartherfrom the intersectionof the uppermost-1 cm of drift overlying a poorly sorted,cloddy two trends, the maroon rocks and brown soils exhibit a matehal. reflectancepeak at 800 nm ratherthan 750 nm (Figure 9). At The primary and secondary spectral trends have the intersectionof the two trends, maroon rocks grade into fundamentallydifferent attributes: the relationshipof spectral pink rocks, consistentwith the similarity in their spectral propertieswith reflectance;the relationshipbetween strength shapesdespite reflectance differences. of the two observed absorptions;their spatial association The association of the two trends with different local with rocksof differentsize, shape,and probablyage; and the environments was explored in detail in the quadrant wavelengthof the reflectancepeak. Thesedifferences indicate northwestof the lander, containingYogi and several other that the two trendsoriginate from differentphysical processes large boulders(Plate 5a). The superpanoramaof this area was and suggestthat they may involve different ferric mineral used tO map the four major spectralunits, basedon the two compositions. These possibilities are explored in greater spectraltrends shown in Figures 11c and 1l d and the ranges detail below. of parametervalues for eachclass summarized in Table 2. We classified the units based on the relationship between 3.5. Drift Coatings: An Explanation for Spectral red/blue ratio and 900 nm band depth shown in Figure 1l d, Variations becausethese parameters can both be calculatedfrom left-eye channels, avoiding problems with parallax difference and The spatialpattern of spectralvariations on rocksand their calibrationuncertainty. Rocks on the primary spectraltrend relationshipto wind direction stronglysuggest that the major can be separatedbased on thresholdingof the red/blueratio. sourceof spectralheterogeneity is thin ferric coatings. The Separating maroon rocks is less straightforward;although spectraleffects of suchcoatings has been investigatedin the they lie on a distincttrend, their red/blueratios overlap those laboratory by Fischer and Pieters [1993] and Shelfer and of pink rocksand drift and their nominal900 nm banddepths Morris [1998], in the contextof Viking Lander multispectral overlap the range in gray rocks. We therefore defined a imaging by Adams et al. [1986] and Guinnesset al. [1987, "resid•ml900 nm band depth," which characterizeshow far a 1997], and using telescopicobservations by Singer [1980]. given spectnunlies from the averagerelationship in primary Viking analysesindicated that most rock surfacesappeared to trend rocks between the band depth and red/blue ratio. be relatively unweatheredand simply covered or coated to Residualsare shownin Plate 5c. Low residuals(shown in varying thicknesses with somewhat amorphous, oxidized blue to green hues) characterizethe primary spectraltrend. material. The darkestblock surfaceanalyzed by Guinnesset Higher residuals(red hues) characterizethe unusuallystrong al. [1987], for example, was well matched spectrallyby an 900 nm absorptionsin the secondaryspectral trend. On unoxidized Hawaiian basaltic andesitenaturally coated with rocks, high residuals occur on surfaces of large, round about 30 nun of exogenicpalagonite. The laboratorystudies boulders,covering parts of Booboo(right edge of image) and [Fischer and Pieters, 1993; Sheller and Morris, 1998] reveal Valentine (far left of Yogi) but only recessedparts of Yogi. that on a dark, spectrallyneutral substrate,including but not 8698 MCSWEEN ET AL.' MARS PATHFINDER ROCKS

Mars: Laboratory: = Gray (Shark) ..... Basalt slab = Red (Broken Wall) ..... 4-pm coating •, Pink (Scooby Doo) ..... 38-pm coating e Bright Red Drift Ferric powder 1.8

.i .9• 1.6

u. ß 1.4 mtmm'm• • o n-' e• 12

• c 1.0

o• 0.8 • o o,_ 06 Effects on ßß n' o• 0.4 Spectral Shape: ß•mm• Thickness of Thin Coating 0.2 400 500 600 700 800 900 1000 1100 (a) Wavelength in Nanometers

Mars: Laboratory: Gray (Shark) 100% slab 40% sl + 60% ct Red (Broken Wall) -- -- - 80% sl + 20% ct ..... 20% sl + 80% ct Pink (Scooby) -- -- e Bright Red Drift 60% sl + 40% ct ..... 100% coating 1.8

.E• 1.6 ,,e-• 1.4 >e• 1.2 '% '• ... • '., ,. -.. ,,,..• .• .. -: • (• 1.0

• $ 0.6 Effects on • - = --•=,•--= - 0.4 Spectral Shape' -- --' • • - Coverage of Thick Coating 0.2 400 500 600 700 800 900 1000 1100 (b) WavelengthinNanometers Figure 12. Spectralsimilarities of primarytrend rocks with ferric-coateddark substrates.Both plotsshow typelocations of gray,red, and pink rocksratioed to brightred drift,and measured or modeledferric coatings ratioedto uncompactedferric powder. (a) Variationsin thicknessof a ferriccoating on a darksubstrate. (b) Variationsin arealcoverage by an opticallythick ferric coatingon a dark substrate.

limited to basalt, increasingthickness of an opticallythin, and they have lesserred/blue ratios. Thesedifferences can smoothferric coatingcauses an increasein reflectanceat all arisefrom wavelength-dependenttransparency of the coating wavelengths,an increasein red/blueratio, and an increasein or wavelength-dependentscattering effects. The effectsof strengthof ferric absorptions.A coatingonly severalmicrons increasingcoating thickness follow quite closelythe primary thick drasticallyalters the spectralsignature of the substrate, spectraltrend characterizinggray, red, and pink rocks. As and a coatingthickness of only severaltens of micronscauses rocksbecome redder they become as brightas drift or brighter, the spectralproperties of the rock to approachthose of the especially at short wavelengths, but at comparable coatingphase [Fischer and Pieters, 1993]. However,ferric reflectancesthe ferric absorptionsare weakerin primarytrend coatingsexhibit three major differencesfrom uncompacted rocksthan in drift (Figure 1lb). ferricpowder: they are brighter, ferric absorptions are weaker, The primarytrend is unlikelyto be theresult of weathering MCSWEEN ET AL.' MARS PATHFINDER ROCKS 8699 rinds formed by chemical alteration of the local rocks. As continuumin compositionbetween hematite and ferrihydrite; described below, the soils at the Pathfinder site cannot be either constituent is consistent with the presence of the formedby alterationof local rocksvia any singlemechanism strong kink (R. V. Morris et al., 1998, unpublished we have considered,and the reddeningagent on main trend manuscript). Telescopicmeasurements of regionscovered by rocks exhibitsa clear chemicalsignature of soils at the site. bright red drift reveal an infraredabsorption centered near 860 Hence the same evidence argues against ferric coatingson nm of low but variabledepth, consistent with a few percentor primary trend rocks occurringpredominantly as weathering less of crystallinehematite [Bell et al., 1990]. The primary rinds. Instead, the primary trend results from different trend materialsat the Pathfindersite are probablypart of this thicknessesof a continuous,optically thin coatingof drift on globally distributed material, but on the low end of dark rocks,or it resultsfrom differentareal coveragesof dark crystallinehematite abundance. rock by a thicker (tens of microns),smooth coating of drift. At least three explanationsfor the secondaryspectral trend Figure 12 comparestype spectraof main trend rocks with appearplausible. First, secondarytrend rocks and soils may laboratory data, with both sets of spectra ratioed to the contain ferrihydrite, which exhibits a consistentwavelength uncompactedferric material(bright red drift or ferric powder) positionof the 900 nm band as well as an 800 nm reflectance to isolate the spectral effects of coatings. An increasing peak. Ferfihydriteis predictedto have formedunder a variety thickness of a very thin, continuousferric coating closely of plausible Martian liquid water environments[Burns, 1992, mimicsthe shift in spectralproperties from gray to red to pink 1994], and the requisite liquid water environment is rocks (Figure 12a). In reality, a coating may be spatially supported by landforms at the Pathfinder site. Second, discontinuous and concentrated in the recesses between secondarytrend materialsmay containmaghemite, which has surface irregularities on a rock. To simulate such a many of the same spectralattributes. Maghemite could have configuration,different areal coveragesof a discontinuous formedin weatheringrinds on mafic rocksand in soil particles coatingwere modeledas linear (areal) mixturesof dark rock by oxidation of magnetite, without invoking a necessarily and an opticallyrelatively thick (38 }am)coating [e.g., Singer "wet" environment, and its occurrenceat the landing site and McCord, 1979]. Again, increasing coverage by the would be consistentwith results of the magnetic properties coatingmimics the shift in spectralproperties from gray to red experiment[Hviid et al., 1997]. Third, secondarytrend to pink rocks (Figure 12b). Either hypothesismakes two materialsmay containmore than one iron-bearingmineral, predictions which are testable with APXS elemental yieldingcomposite properties similar to thoseof ferrihydrite abundancemeasurements: progressively redder rocks should or maghemite. For instance,certain impact melt rocksfrom exifibit a progressivelymore drift-like composition,and the ManicouaganCrater containingmixtures of hematite and reddestprimary trend rocks shouldmost closely approach the pyroxenehave reflectivity maxima near 800 nm and a weak compositionof drift. Both predictionsare confirmed by absorptionnear 900 nm [Morriset al., 1995]. analysesas describedbelow. Based on this analysis,our working hypothesisfor the 4. Rock Chemistry origin of the primary spectral trend is variations in the thic-ness and/or coverageof coatingsof drift on gray rocks. 4.1. APXS Analyses Red rocks may be partially covered with coatingstens of microns thick, or they may have a continuouscoating only Detaileddescriptions of the APXS instrumentdesign and several microns thick. Pink rocks may simply be gray rocks operationwere providedby Rieder et al. [1997b], and a with a coating of drift as thin as several tens of microns; typical Martian rock X-ray spectrum(Barnacle Bill) was spectrally,there is little differencebetween such a coatingand illustratedby Riederet al. [1997a]. Five preliminaryrock cemented drift. analyses(X-ray modeonly), previouslypublished by Rieder The differentspectral systematics of the secondaryspectral et al. [1997a], are reproducedin Table 3. Although the trend are inconsistentwith it havingthe samephysical origin compositionof one pink rock (ScoobyDoo) is available as the primarytrend. A simpletextural difference in coatings [Rieder et al., 1997a], this analysisis not reportedhere is insufficientto explain the differencesin the near-infrared becauseit is more appropriatelyconsidered in a discussionof spectraof rocks in the two trends. The simplestexplanation Pathfindersoil compositions.Because atmospheric CO 2 for the difference between the two trends is a different ferric betweenthe APXS sensorhead and the samplesignificantly mineral coatingon the secondarytrend rocks, which assumes impactedthe alphamode spectra, the abundancesof light more of the propertiesof primarytrend rocks as an additional elements will not be available until the instrument is coatingof bright red drift is added. To maintainthe drift-like recalibratedunder Martian conditionsof CO2 pressureand redness,the coatingon maroonrocks must be opticallythick temperature.Proton spectra are unaffectedby CO2, but they (50 gm or more)and rough at the scaleof a wavelengthof light canonly be interpretedalong with the alphadata. Alphaand [Fischer and Pieters, 1993]. protonmode analyses should allow determinationof oxygen, The ferricmineralogy of primarytrend materials cannot be i.e., determinationof the approximateredox state of the rocks, determineddefinitively. The presenceof a strongkink does as well as more preciseanalyses of Na20 and other oxides implicate hematite,ferrihydrite, or maghemite,but none of with overlappingspectral lines (MgO, A1203, andSiO2). A these can be continned because of the lack of a well defined second measurement of Half Dome and measurements of three infrared absor9tion. The best spectral analog to bright additionalrocks (Chimp, Moe, and Stimpy)were obtained Martian soils is a mixture of phases,with poorly crystalline, afterthe roverbattery had expired, resulting in measurements nanometer-sizedparticles of ferric oxide dispersedin an with highthermal noise taken during the Martianday; these aluminosilicatematrix [Morris el al., 1993]. Thesenfixtures analyseshave not yet beendeconvolved. occur naturally in certain palagonitictephras from Hawaii Rock analyses have been normalized to 98% oxides to [Singer, 1982;Morris et al., 1990, 1993' Bell et al., 1993]. correctfor unreportedP205, MnO, andCr203, whichhave The nanophaseferric material in palagonitemay lie along a large errors becausetheir peaks are buried within other 8700 MCSWEEN ET AL.: MARS PATHFINDER ROCKS

Table3. PreliminaryAPXS Analyses a, and Calculated Composition and Normafive Mineralogy of Sulfur-Free Rock

Rock Na20 MgO A1203 SiO2 SO3 C1 K20b CaO TiO2 FeO*

Bamacle Bill A-3 3.2+1.3 3.0+_0.5 10.8+1.1 58.5+__2.92.2+__0.4 0.5+0.1 0.7+__0.15.3+__0.8 0.8+__0.2 12.9+1.3 Yogi A-7 1.7+_0.7 5.9+_0.9 9.1+__0.955.5+__2.8 3.9+__0.8 0.6+0.2 0.5+__0.16.6+1.0 0.9+__0.113.1+1.3 Wedge A-16 3.1+1.2 4.9+0.7 10.0+__1.052.2+__2.6 2.8+-0.6 0.5+-0.2 0.7+__0.17.4+-1.1 1.0+__0.115.4+1.5 Shark A-17 2.0+_0.8 3.0+_0.5 9.9+1.0 61.2+3.1 0.7+__0.30.3+__0.2 0.5+0.1 7.8+1.2 0.7+__0.111.9+1.2 Half Dome A-18 2.4+1.0 4.9+0.7 10.6+-1.1 55.3+__2.82.6+_0.5 0.6+__0.2 0.8+__0.1 6.0+_0.9 0.9+__0.1 13.9+__1.4

Calculated sulfur-free rock 2.6+1.5 2.0+_0.7 10.6+_0.7 62.0+_2.7 0 0.2+_0.2 0.7+_0.2 7.3+1.1 0.7+_0.1 12.0+1.3

Provisional CIPW normc for sulfur-free rock Quartz Orthoclase Albite Anorthite Diopside HyperstheneIlmenite Magnetite 21.0 4.1 22.0 15.2 18.2 15.6 1.3 0.5

aValuesare in unitsof wt % oxides,from Rieder et al. [ 1997a].APXS analyses are normalized to 98%oxides, to allow for unreportedP205, MnO, andCr20 3. bpublishedK20 values may be approximately 50%too low, based on recalibration (R.Rieder, pers. comm.). CCalculatedwt % normassumes a molar Fe202/FeO ratio of 0.026(average basaltic shergottite value of McSween and Jarosewich [1983]).

spectrallines. Further calibration for these componentsis 4.2. Rock Compositionsand Mixing Lines ongoing. All iron is reported as FeO (henceforth called FeO*). Analyticaluncertainties reported in Table3 were Table 4 suxnmarizes coefficients for all possible derivedfrom the rangein differencesfound between certified correlationsbetween oxide abundancesin the five analyzed and measuredvalues for eight referencestandards. Rieder et rocks. Rock compositions,when plotted on two-component al. [1997a] also reportedAPXS analysesof two meteorites, diagrams(Figure 13), form roughly linear arrays with soil including the SNC meteorite Zagami, to demonstratethe analysesfalling consistentlyat one end of the array [Riederet accuracyof the instrument. al., 1997a]. The most straighfowardinterpretation of these The performanceof the APXS instrumentwas excellent arraysis that they are mixing lines, with end membersdefined throughoutthe entire period of operation on the Martian by a single rock composition and the average soil surface. Especially good was its electronicsgain stability composition(the latter presumablyin the form of adhering overthe entire temperature range from -87øC to about-20øC. dust). Differences in the compositionsof Pathfinder and The total energy deviation for a particular energy was less Viking soils indicate that the admixed componentmore than 0.2 channels (12 eV). This excellent stability closelyresembles the local soil than the more sulfur-richsoils contributed to a better than expected peak resolution in at •he Viking sites(e.g. Figures 13a and 14b). The APXS only Martian spectra.In orderto testthe overallprecision of the X- determineselemental abundances to depthsof a few tens of ray mode,10 laboratorymeasure•nents of the samesample have microns, so any surficial dust or weathering rind would beenperformed with differentaccumulation times ranging from greatlyaffect the rock analyses.In addition,the percentageof 1 hour to >24 hours. The results of the 10 analyses were dust componentin a bulk APXS analysisof the rock with a statisticallyidentical for all the major elementsin the sample. thin (<100 mm) dustcoating is higherfor lighterelements like This high precision allows the interpretationof relatively sodiumthan for heavierelements like iron becausethe X-rays small differencesin the compositionsof rocks(see below). for elementswith lower atomicweights are absorbedmore by In some of the following figures, analysesof Pathfinder the sample[Crisp, 1998]. and Viking soils are plotted to illustrate their chemical If we assume that the rocks are igneous, the rock relationshipsto the rocks. Like the rocks, the Pathfinder compositioncan be retrievedfrom plots of sulfur (which is soils [Rieder et al., 1997a] are normalized to 98% total enriched in Pathfinder soils) versus other oxides. The oxides. For comparisonpurposes, XRF analysesof Viking solubilityof sulfurin inagmasat reasonableoxidation states is soils[Clark et al., 1982] havebeen normalized to 95.4% total small, normally no more than a few tenths of a percent by oxides(98% minusthe averagePattffmder soil value of Na20 weight [Carroll and Webster, 1994; Jambon, 1994]. + K20 = 2.6%, oxideswhich were belowdetection limits in Acceptingthat the sulfurcontents of typical igneousrocks are the Viking data set). in the 0-0.2 wt % range,we can extrapolatelinear regression MCSWEEN ET AL.' MARS PATHFINDER ROCKS 8701

Table 4. Coefficients for Linear Correlations Between Oxide Abundances in Pathfinder Rocks

Na20 MgO A1203 SiO2 SO3 C1 K20 CaO TiO2 FeO*

Na20 1 MgO -0.36 1 A120 3 0.72 -0.58 1 SiO2 -0.28 -0.77 0.15 1 SO3 -0.09 0.87 -0.40 -0.75 1 C1 0.03 0.76 -0.03 -0.69 o.89 1 K20 0.69 0.02 0.79 -0.46 0.12 o.46 1 CaO 0.36 0.05 -0.56 0.01 -o.34 -o.6o -0.54 1 TiO2 0.29 0.78 -0.14 -1.00 0.77 0.72 o.46 -o.o6 1 FeO* 0.52 0.54 0.09 -0.95 0.51 o.5o 0.60 0.03 0.94

lines throughthe data to zero sulfur and therebyestimate the ,rock plots along the Mars mantle-crust fractionation line compositionof the rock end member,as illustratedin Figure (definedby the compositionsof SNC meteoritesand Martian 14. Linear regressioncoefficients for all oxides versusSO3 soils) in a diagram of Mg/Si versusA1/Si [see Rieder et al., are given in Table 4. For a samplesize of five rocks, the 1997a, Figure 4]. The sulfur-freerock's high Fe/Mg probably correlations that are statistically significant at the 90% reflectsthe FeO abundanceof the Martian mantle [Wanke and confidencelevel are those >0.80' SiO2-TiO2,FeO*-TiO2, Dreibus, 1988], but its low A1/Si might be an artifact of MgO-SO3, andC1-SO 3. However,the regressioncoefficients normalizationto a high silicon content(explained more fully below). The sulfur-free rock has significantlyhigher SiO2 for sevenadditional pairs are >0.70. The effectsof closure(the than SNC meteorites, and its relatively high K20 content constant sum problem for compositionaldata [Aitchison, (and apparently high P205 content, as judged from raw 1986]) reducethe significanceof the negativecorrelations in Table 4. spectra)implies that it is more highly differentiatedand richer The sulfur-freerock compositiondetermined in this way is in incompatibleelements than are basalticshergottites. given in Table 3. This compositionexhibits some of the same 4.3. Rock Norm and Classification chemicalpeculiarities that are found in SNC meteorites. The Pathfinderrocks' high Fe/Mg and low A1/Si are characteristic The chemicalcomposition of the sulfur-freerock has been of shergottites,nakhlites, Chassigny, and ALH84001 [Wanke recast into mineralogy using the CIPW norm formulation and Dreibus, 1988; Rieder et al., 1997b], and the sulfur-free (Table 3). This calculationis basedon idealized,anhydrous

lO 20 Pathfinder Soils

D D 18 D DD D D

Normalized r = o.87 Viking So ils

2 Pathfinder Rocks r =0.51

0 1 2 3 4 4.5 0 1 2 3 4.5 S (Weight %) S (Weight %) Figure 13. Mostoxide abundances in Pathfinder rocks fonn approximately linear arrays when plotted against sulfur,with Pathfindersoils plotting consistently at oneend of the array. Linesrepresent linear regressions for therock data only. All Pathfinderdata are normalized to 98%total oxides, and Viking XRF soilanalyses [Clarket al., 1982]have been normalized to 95.4%oxides (see text). 8702 MCSWEEN ET AL.' MARS PATHFINDER ROCKS

70 65 Rock

Pathfinder Rocks

ß 55 r =0.74 Normalized 0 5o Viking Softs

45 Pathfinder Soils

Pathfinder Soils

Pathfinder Rocks

o.5 ulfur-Free Rock Normalized Viking Softs

0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 S (Weight %) Figure 14. Additional plots of oxides versussulfur, illustratinghow extrapolationof linear regressionlines to zero sulfur gives the co•npositionof the sulfur-free(presumably igneous) rock. Viking soil analyses[Clark et al., 1982] are normalizedas describedin the text.

minerals that crystallize under equilibrium conditions in Of course, we do not know if the rocks at the Pathfinder igneousrocks. In the norm calculation,we have assumedthat site are igneous and, as already discussed,certain textural the Fe203/FeO molar ratio is the sameas that measuredin featuresin some rocks may suggesta sedimentaryorigin basaltic shergottites(0.026) [McSween and Jarosewich, [Rover Team, 1997]. The compositionof the sulfur-freerock, 1983]. The norm for the sulfur-free rock is dominatedby however,is similarto thoseof someigneous rocks. Figure 15 feldspars (albite, anorthite, and orthoclase), pyroxenes shows plots of normative feldspar compositions and (hyperstheneand diopside),and quartz. We emphasizethat nonnative"color index" (defined as the sumof hypersthene+ this norm is only provisional, because changes in the diopside+ iron-titanium-chromiumoxides), with the shaded abundancesof alkalis and other light element oxides and/or areas representingthe fields of cmnmon terrestrial volcanic redox state (which may occur when alpha-protondata are rocks [adaptedfrom Irvine and Baragar, 1971]. The sulfur- available) would affect the proportionof feldspar and other free rock plots within the shadedareas in each case;there is phases. Inclusionof P205 would producenonnative apatite no reason a priori why sedimentary rocks should do so. at the expense of diopside. The nonnative calculation is Althoughthe nonnativefeldspar compositions for the average intendedto emphasizedifferences between rocks, so small Pathfindersoil [Riederet al., 1997a] and the sulfur-freerock changesin chemicalcomposition may resultin more dramatic are similar (Figure 15a), their nonnativecolor indicesare not distinctions in the norm. (Figure 15b). This comparisonsuggests that even if the MCSWEEN ET AL.: MARS PATHFINDER ROCKS 8703

Anorthite i.i [• I ...... l...... 1...... Pathfinder Soil ...... !._ .

I I "-••""•%,1••'•••"*"••- '•-' VolcanicRocks

Albite Onhoclase 0 ' 40 ' 20 0 Norm PlagioclaseComposition Figure15. CIPW nomativecomposition of •e Pa•ffmdersulfur-free rock, comp•ed to •ose of co•on te•es•al voltaic rocks,adapted from I•ine andBaragar [1971] (all nomsin •ese plots•e in tool%, ra•er • • % asgiven in Table3). (a) Nomativefeldsp• compositions, and (b) nomativecolor bdex (hypers•ene+ diopside+ magnetite+ ilmenite) versusnormafive plagioclase composition (•o•te/(•o•i• + albite).•e no•ativecompsition of averagePa•ffmder so• [Riedereta&, 1997a] is shownfor comp•son.

Pathfinderrocks are sedimentary,they are likely to be clastic SiO2 - 5.135)[Gill, '1981]. The abundances of TiO 2 (0.7%) rockscomprised mostly of volcanicfragments. and K20 (0.7% < 3.86% for this silica content)in the sulfur- Based on its chemistry and normafive mineralogy, the free rock compositionsupport the classification. Although sulfur-free rock has an andesitic composition. On a silica reported K20 values may have been underestimated(see versusalkalis diagramcommonly used for classificationof footnote b to Table 3), adjusted values do not affect the volcanicrocks [Le Bas et al., 1986], the sulfur-freerock plots andesite classification. within the field of andesite and the individual (dust- Geochemical classification is also sometimes used for contaminated)rock compositionsfall mostlyin the basaltic sedimentaryrocks, although determinationsbased on grain andesite field (Figure 16). Andesites are hypersthene- size andmineralogy are preferred.Using the schemeof Herron nonnativevolcanic rocks with 57-62 wt % SiO2 (calculated [1988], which distinguishes sandstones based on log on an anhydrousbasis), TiO2 < 1.75%, and K2¸ < (0.145 x Na20/K20 versuslog SiO2/A1203, the sulfur-freerock plots

Tephritej•,trachy- / •andesite Rhyolite Trachy- Dacite

oidite Wedge Barnacle Bill

3 Sulfur-free Rock Picro- I Shark Shergo•itesbasalt • Iiandesite I \ ! I 35 45 55 65 75 85 SiO2 (weight%) Figure 16. Chemicalclassification of volcanicrocks, after Le Bas et al. [1986]. The sulfur-freerock plots within the field of andesite.Shergottites (shown as solidsquares), commonly thought to be Martianigneous rocks, fall within the basalt field. 8704 MCSWEEN ET AL' MARS PATHFINDER ROCKS

A-19

A-20 A-23 A-27

Figure 17. IMP images of the APXS deployedon rocks at the Pathfinder landing site. APXS analysis ntunbersare keyedto rock namesin Figure 1.

in thefield of graywackes.On a plotof logFe203*/K20 1997; Smith et al., 1997b]. Chemical evidencefor variable versuslog SiO2/A1203 usedfor classifyingsandstones and soil coatingson the rocks, already discussed,can only be shales, the sulfur-free rock would be termed an iron-sand. comparedwith spectraldata if IMP and APXS measurements are taken from preciselythe same locations. IMP images 5. Comparison of Chemical and documentingeach APXS deploymenton rocks are shownin Multispectral Data Figure 17. The exact position of each APXS spot was determinedthrough careful analyses of rover cameraand IMP 5.1. Spectral and Chemical Analyses of the Same Spots images.In caseswhere the locationwas especially difficult to estimate, Jet Propulsion Laboratory rover engineering It was recognized early in the developmentof the softwarewas usedto measureaccurately the roverand APXS Pathfindermission that correlatingAPXS andIMP datamight positions.The numberof pixelssubtended by the APXS spot provide important.•nfonnation on the nature of the rocks and in IMP imageswas assumedto be a linear functionof distance insightsinto the processesthat formed them [Golombek, from the IMP. For eachspot, spectra were generatedusing MCSWEEN ET AL' MARS PATHFINDER ROCKS 8705

Table5. SpectralReflectance Data Used to CalculateRed degreeof pixelsaturation due to overexposure.Under optimal (750nm)/Blue (440 nm) Ratios for Pathfin&r Rocks lightingconditions at slightlydifferent times of day, ratios variedby no morethan 0.4. for rocksand 0.1 for soils(except MermaidDune, which variedby 0.5). Becauseof the minor APXS Local SolarTime Reflectance ReflectanceRed/Blue differences,these points were averaged. Site of Measurement 440 nm 750 nm Ratio 5.2. Results A-3 1042 0.0458 0.1248 2.725 A-7 1531 0.0743 0.3075 4.139 The interpretationof rockcompositions at the Pathfinder A-16 0930 0.0708 0.2236 3.158 site as mixtures of rock plus adhering dust is strongly A-16 0948 0.0645 0.2016 3.124 supportedby a correlationbetween rock compositionand A-17 1009 0.0329 0.0848 2.568 spectralfeatures [Bridges et al., 1997]. Thered (750 nm)/blue A-17 0932 0.0422 0.0950 2.250 (440 nm) ratio(Table 5) measuresthe spectralsimilarity of A-18 1012 0.0636 0.1711 2.692 individual rocks to soil, with the reddest rocks being most A-18 0936 0.0626 0.1740 2.779 similar. The red/blueratios are plottedagainst SO 3 because Martiansoils are verysulfur-rich and red relative to therocks. Figure 18 demonstratesthat APXS soil sitesare generally redder than rock sites; the only exceptionis the soil at MermaidDune, whichhas a slightlylower red/blueratio than calibrateddata [Reid et al., 1997]from all IMP geologyfilters Yogi. The red/blueratios for the entiredata set and for the in the superpanoramaand multispectral-spot sequences. For rocks alone correlate very well with SO3 (Figure 18). mostAPXS sitestwo or morespectra, each taken at a different Correlationsof red/blue ratio with other chemical components time of day and under varyinglighting conditions, were available(Table 5). SomeAPXS soil spotswere either inthe rock analyses (e.g., C1 and FeO*) are less significant. disturbedby therover or containedcompacted soils. In these Theseobservations imply that a dustor soil coveringis cases,spectra were measuredfrom areas near the spot presenton all the rocks,even in apparentlyclean areas selected for APXS measurements. As noted previously, locations. Except in caseswhere spotswere heavily shadowed,all spectrawere used. The reflectancesat each spectraof rocksat the VikingLander 1 sitehave also been wavelengthwere adjustedslightly to accountfor small interpretedto reflectmixtures of rockand soil [Adams et al., differencesbetween the left andfight eyesof the IMP caused 1986],although no chemicaldata for thoserocks exist. The sulfi•r detected in APXS rock measurements is strongly by calibrationerrors. Redness of eachspot was measured by takingthe ratio of reflectancesat 750 nm (near-infrared) and affectedby superficialcoatings of dustor soiland is therefore 440nm (blue). In mostcases, reflectances from the IMP right notrepresentative of therock chemistry. eyewere used. The Yogi fight-eye 440 nm image had missing Thesmaller range of C1 and FeO* concentrations inrocks packetsin thearea of theAPXS spot, so left-eye data were andsoils relative to SO3 is probablyone reason for theirpoor usedin this one case. The left-eyeimage contained a minor correlations with red/blue ratios. The lack of correlation of

5.5

O 5.0 Pathfinder0 r

4.5 Soil,,=o E Yogi ./• 4.0- Pathfinder Rocks .,•r • 0.89 3.5 W•edg•.,.,e/ - Barnacle Bill 2.5•S__•_har• •"•--••lalfDøme I .,'•1 I I I I I 1 2 3 4 5 6 SO3 (Weight %) Figure 18. Comparisonof the SO3 contentsin APXS rockanalyses with the red (750 nm)/blue(440 nm) ratiosof theirspectra. The long regression line was calculated for all data(rocks and soils), whereas the short regressionline was determinedfor rocksonly. Rockswith highersulfur have high red/blue ratios, supportingthe conclusion that the rock compositions define a mixingline between a rockend member and adheringdust or soil. 8706 MCSWEENET AL.: MARSPATHFINDER ROCKS red/blueratio with FeO* in rockssuggests that the Coarse clastic rocks have chemical compositionsthat mimic abundancesof spectrallyactive ferric componentsin the dust the compositionsof the rocksthat were disaggregatedto form and rock do not affect the spectralproperties as muchas the them. For example, the Poway conglomeratein California thickness and areal distribution of dust on rock surfaces, contains virtually identical minor and trace element consistentwith our previousresults. abundancesto the volcanic rocks from which its gravel fragmentswere derived[Abbott and Smith,1989]. Evenfiner- 6. Interpretations and Discussion grainedsandstones dominated by rock fragments(i.e., wackes) can retain the chemistryof their sourcerocks, as indicatedby 6.1. Igneous,Sedimentary, Impact, or Alteration? the diagnostictrace elementpatterns in wackesfrom different tectonicsettings [Bhatia, 1985]. Consequently,inferences One of the most basic questionsabout rocks at the drawnfrom suchsedimentary rocks about the petrogenesisof Pathfinderlanding site is how theyformed. The andesite-like their igneousprotoliths may still be valid. compositionof the sulfur-freerock sharesgeochemical It is also possiblethat Pathfinderrocks are impactmelts or characteristics with some common terrestrial volcanic rocks impact breccias. Their high nonnative feldspar contents, and with SlqCmeteorites of undisputedigneous origin. This relative to SNC meteorites,might supportthe hypothesisthat stronglysuggests, but doesnot mandate,an igneousorigin they experiencedimpact melting. Shock-loadedfeldspars are for at leastthe subsetof rocksfor which APXS analyseshave more readily melted than most rock-forming minerals, and beenreported. Pits interpreted as vesicles and facets possibly small-scaleshock melted veins in shergottitesare commonly causedby colunmarjointing also imply a volcanicorigin for enriched in a feldspathic component [McSween and somerocks at the landing site. Physicaltheology models darosewich, 1983]. However, melt rocks from large-scale appliedto the morphologyof lava flows on Martianshield impacts on the Moon and Earth typically show little volcanoes are consistent with viscosities appropriate for differentiationfrom the bulk target materials [Warren et al., magmacompositions ranging from ultramaficto andesitic 1996]. Samples of the oldest Martian stratigraphicunits [Hulme, 1976; Moore et al., 1978; Zimbelman, 1985; would be more likely to have experiencedpervasive shock Cattermole, 1987; Lopes and Kilburn, 1990]. However, metamorphism and melting as part of the late heavy landformssuggestive of silicic volcanism,such as resurgent bombardment[Ash et al., 1996], and the only ancientMartian domes and festooned flows in the Martian highlands, are meteorite has been multiply shocked and partly melted uncommon[Hodges and Moore, 1994]. Calculationof the [Treirnan, 1995]. viscosityof an andesiticmagma having the coinpositionof the Anotherpossibility is that the rocksare shock-melteddust sulfur-freerock, usingthe methodof Shaw [1972], gives a deposits[Schultz and Mustard, 1998]. Impact melting of minimumvalue of 4.5 x 103Pa s at 1100øC,similar to the terrestrialloess produces glassy, vesicular rocks of variable lowerend of the viscosityrange estimated for Martianmagmas. compositionwhich have spectrasimilar to Pathfinderrocks. Finally,the observationthat all analyzedrocks at the site However, the compositionsof soils at the Pathfinder site have similarcompositions and spectralproperties mitigates [Rieder et al., 1997a] are distinct from the sulfur-freerock againstnonigneous rocks that haveconsiderable variability, composition,so fractionationduring shockmelting would be such as layeredsediments. We believethat a plausible, required. perhapscompelling case has been made that many Pathfinder Although a significantamount of ejectacould have been rocksare volcanicin origin, but we mustalso considerother depositedat the Pathfinder site from the large crater to the possibilities. south, estimates based on the crater's size and distance from For mostrocks, spatial variations in spectralproperties the site suggestthat most of its ejectawould be significantly providelittle evidencewith which to distinguishigneous, smaller(< 20 cm diameter)than any of the rocksanalyzed by sedimentary,or impact-generatedorigins. Nearly all spectral APXS [Smithet al., 1997a]. The ejectamodel ofMcGetchin et heterogeneityresults from coveringsof drift, or variationsin al. [1973] predicts that the thicknessof a uniform ejecta spectralshape and reflectanceeasily reconciled with ferric deposit from this crater should be 0.3 m, but new crater mineral coatings. Apparentlyscoured gray rock surfaces morphologydata from Mars Global Surveyor suggestthat exhibit little in the way of organizedspatial variations in ejecta blanket thickness based on this paradigm is color. The two significantexceptions are the cobbles overestimated [Garvin and Frawley, 1998]. The limit of discussedearlier, Ginger and Boris. Ginger'sdark gray lobes continuousejecta [Moore et al., 1974] is only about 1.75 km separatedby a maroongroundmass could be interpretedas a from the crater center, well short of its 2.95-?cmdistance from ferric-cementedconglomerate; alternatively, a small,dark red the landing site. Consequently,it seemsunlikely that any patch behind Ginger may indicate that the rock was large Pathfinderrocks, of sizes like thoseanalyzed by APXS, overturnedby airbagretraction, so that the red materialon are ejectafrom this crater. Ginger could be adheringdisturbed soil. Also, the color Finally, we acknowledgethe possibilitythat the rocksmay variationscould originate from remnantmaroon material be coatedwith siliceousweathering rinds, even after removal preservedbetween eroded, dark grayprotuberances. Boris' of adhering dust. Hawaiian basalts exposed to semiarid varicolored,subangular blotches separated by a dark gray weatheringdevelop hydrous silica coatingsthat thickenand groundmasscould similarly be interpretedas a breccia incorporate detrital material with age [Farr and Adams, cementedby a ferric-poormaterial, possibly"dark soil" or 1984; Crisp et al., 1990]. These coatings tend to be tephra. Spectraldata for a few rocksprovide tantalizing but discontinuous and variable in thickness, up to tens of ambiguousevidence for limited lithologicvariations at the microns,and may be underlainby thinnerzones of iron oxides. site,but theydo not supporta particularorigin for mostrocks. Nearly pure silica coatings are colorless, but weathered As previouslynoted, textural observations may suggest basaltic detritus incorporated into older coatings produces that some rocks at the Pathfinder site are conglomerates. colorsranging from brownto red. MeasuredSiO 2 contentsfor MCSWEEN ET AL.: MARS PATHFINDER ROCKS 8707

theserinds are higher(>66 wt %, and commonly>80%) than 6.3. Absence of a Pyroxene Absorption Band that estimatedfor the Pathfindersulfur-free rock, and analyzed oxide sums are lower (<90%, as appropriatefor opal or Basalts similar in mineralogy to some SNC meteorites hydratedsilica glass)than somePathfinder rocks (low sums (specificallybasaltic shergottites) are commonlythought to in someAPXS analysesare thoughtto result from variations representlarge portionsof the Martian surface. The marked in the measurementgeometry [Rieder et al., 1997a]). compositional similarity between these meteorites and However, the discontinuous and thin character of siliceous Martian soils [Baird and Clark, 1981] suggestsa pervasive rinds might allow lower measuredbulk silica contentsand basalticsource for weatheredmaterials [McSween, 1994]. Near-infraredspectral similarities between shergottites and highertotals on APXS-sizedspots. Under ordinaryterrestrial weatheringconditions, silica is leached from basalt, but acid dark regions of the Martian crust, interpretedas relatively conditionsor cyclic wettingand drying can concentratesilica rocky regions, buttress this conclusion [Soderblom, 1992; [Farr and Adams, 1984], or a silica glaze can form as a Mustard et al., 1993; Singer and McSween, 1993; Mustard reactionproduct on the rock surfacein the presenceof dew or and Sunshine, 1995]. The Martian remote-sensingspectra light rain [Casey et al., 1993; Dom and Meek, 1995]. contain two absorptionbands that indicate the presenceof pyroxenes with variable calcium and iron contents. Formation of siliceouscoatings on Martian rocks by an analogous process, however, might require intermittent Deconvolutionof the spectrareveal overlappingabsorption rainfall on Mars, which seemsunlikely in the past billion bands for both pigeonite and augite [Mustard and Sunshine, yearsor so. If coatingsformed in the Noachian,rainfall might 1995; Mustard et al., 1997], with iron-rich compositions havebeen an agent,but thentransportation of rocksby floods similar to those in shergottites. Although these spectral would likely have removedthe coatings. The observation observations are commonly cited as evidence for a that Pathfinderrocks with the highestsilica contentsare least predominantlybasaltic Martian surface,it might be impossible spectrallyreddened is oppositethe trendexpected if siliceous to distinguish pyroxene-bearing andesites. The calculated coatingsincorporated detritus (as do Hawaiianbasalt rinds), norm for the sulfur-free rock at the Pathfinder landing site but couldbe rationalizedif detritus-freecoatings are covered contains both hyperstheneand diopside (Table 3), which by a veneer of less siliceousdust. would presumablybe representedin a volcanic rock by pigeoniteand augite, respectively. Adamset al. [1986] found 6.2. Stratigraphic Implicationsof Rock Spectral Features that the best spectralmatch for rocks at the Viking 1 Lander site was a mixture of gray andesitewith palagonitedust and a A key implicationof the spatialassociations of primary coarserock-like soil. However, they cautionedthat the rather and secondarytrend spectralattributes with rocksof different featureless,three-band spectra did not warrant assignmentof a morphologicclasses is a stratigraphyof rock spectralfeatures, specificrock type. providedthat the large, roundedboulders predate smaller The inability to discerna pyroxeneabsorption band near rocksas interpretedby Smithet al. [1997a]. Figure19 shows 1000 nm in IMP spectra of any Pathfinder rocks is a three-stagesequence of eventsthat can explain these spatial disconcerting, given that this band commonly appears in associations.In stage 1, large roundedboulders and their spectraof dark regionsobserved telescopically and from orbit. marooncoatings were eraplaced. The same material forming the Analysis of the continuumeffects of atmosphericaerosols on coatingsmay have been eraplacedelsewhere, perhaps as the remote data showsthat, at the surface,the typical "1 •m" brown soils exposedin the subsurfaceby rover tracks. In band in dark regions should exhibit a minimum within the stage2, the small cobbleswere eraplaced.Some small rocks wavelength range covered by IMP [Erard et al., 1994]. mayhave retained a partialmaroon coating due to an originby Although adhering dust is very effective in hiding the fragmentationof large,old boulders.Concurrently or later,a spectralsignatures of underlyingrocks [Fischer and Pieters, layer of bright red drift at least severalcentimeters thick was 1993], it is still surprisingthat no pyroxeneband is observed emplaced.In stage3, emplacementof rockshas largely ceased. in gray rockswith the leastdust coating. Aeolianerosion has partially removed the marooncoatings from old rocks,and it hasredistributed the brightred drift A numberof hypotheseshave beenconceived to explainthe absence of the band minimum. Lack of a band minirotan due to leavingthe red bandat the baseof manyrocks. Brownsoil maybe exposedin thesubsurface or by spallationfrom maroon grosscalibration error is refuted by the resolutionof a band minimum in some ferric materials at the site. More rocks. Eraplacementof very smallamounts of brightred drift on grayrocks has produced coated red andpink rocks. Some significantly, absorptionbands are accurately resolved in pinkmaterials may also originate from exposure of preexisting, IMP spectra of different ferric minerals embeddedin color cemented brown soil or drift. chips on the IMP calibration targets [Reid et al., 1997]. A In terms of geologic events, stage 1 may include band could be presentbut extremely weak and not resolved depositionof floodmaterials, stage 2 a secondflood event of above noise in the data, due to a very fine-grained rock lessermagnitude or emplacementof impactcrater ejecta and texture. However, available image resolution(1-5 mm per airfall drift, and stage 3 subsequentaeolian erosionand pixel) couldresolve mineral grainsonly as coarseas thosein redeposition.If this sequenceis correct,then the different medium- to coarse-grainedigneous rock, so the failure of ferric phasesproposed to occur on primary trend and imagesto resolvemineral grainsmay not necessarilyindicate secondarytrend rocks originated in two differentperiods, and a very frae-grainedtexture, e.g., <50 [tin [Mustardand Hays, perhapstwo differentpaleoenvironements. If the ferricphase 1997]. Alternatively,relatively "clean" gray rock surfacesmay occuringin reddertrend rocks is maghemiteor ferrihydriteas contain a silica coatingwhich masks spectralproperties of proposedabove, it may be a mineralogicsignature of the pyroxenein the rock interior. However, the existenceof such wetter flood environmentin which large, roundedboulders coatings is inconsistentwith the lesser abundanceor absence were emplaced. of ferric-rich coatingson the same rock faces, unlesssilica 8708 MCSWEEN ET AL.: MARS PATHFINDER ROCKS

Stage I

Old maroon / rockswith ferric coatingX .•<,,,,,•.•,•. .,,.,,&%?;,<•,•.<,•,,•,•x,x,x,•:•¾•:•:•:•:: ...... x

Stage 2

Bright red drift Fragmentsof Freshly • old maroon exposed rocks cobbles

"'%"%•:.•? ...... •:•.•.•:•.•....

Stage 3 New coatings of

•/ brightred drift Erosion and ....' '""'""::'"'• Abrasionofold redeposition of

...... •.... • coatings bright red drift • ...:.::::ii .... :.::i:ii::

x:.,.::.?:.....x.,.:¾,...%.• .,..,..,.,,•,• ,, ..,.,. >.>..?. •:.,..,.•,,., .,..,,,.,.• ::::• :: > ::>.:::.:.:.:. :.:.:. :., :. :. :. :. :. :. :. > :.: .:. ::.:: ::: .,..+.,.x,,.,.,: • :.:. :.:. :.:. :.:.:. :.:. :.:. x.., > >> >.,::..,.,., .,;.,.,,:., ...,:., .,, ,:, ,...,.,:.,;;:.:. ,, ..-.....-.-.....-.-.-...•-...... ,.

Figure 19. Proposedstratigraphy of spectralheterogeneities on rocksat the Pathfinderlanding site.

coatings are significantly harder to remove than ferric push the centerof the band to or beyond 1000 nm so that it coatings. would not be resolvable by IMP, but would produce the We favor three hypothesesfor the apparentabsence of a observedinfrared spectral curvature. Figure 20 illustrates pyroxeneband. First, the rocks may be volcanicor impact zoned pyroxene compositionsin the Nakhla and Shergotty glasses;the"1 I•m" band in Fe2+-containing glassesis meteorites, with superimposedcontours of the "1 gm,' band centerednear 1100 nm. The IMP spectralrange is insufficient minima for pyroxenesof variouscompositions [Cloutis and to resolve a band minimum in glassesbut could detect the Gaffey, 1991]. Augite compositionsin Nakhla are rich in infraredcurvature like that observedin gray rocks. However, calcium(Wo component),and the absorptionband minimum glasses in very old rocks might be expected to have (illustratedas a shadedbox) lies beyond1000 nm. Basaltic devitrified, especially in the presenceof water. Second, a shergottitescontain both high-calcium (augite) and low- maskingphase dispersed throughout the rocks may obscure calcium (pigeonite) pyroxenes. The absorptionbands for an absorption. Magnetite is common in terrestrial volcanic pigeonite and augite overlap, so that the band positionsof rocks and some SNC meteorites [McSween, 1994], and bulk basalticshergottites (illustrated as a secondshaded box) laboratory spectra show that its presence can drastically fall between the actual pyroxene compositions at band weaken the absorptiondue to pyroxene 'Hunt et al., 1974; positionsless than 1000 nm. However, the compositionsof Cloutis et al., 1990]. Third, the pyroxenecompositions in nonnative pyroxenes in the Pathfinder sulfur-free rock are these rocks could be significantly higher in iron and/or significantlyricher in iron (Fs component)than are Shergotty calciumthan are typical for Mars. Such a compositioncould pyroxenes. Very iron-richpigeonite is metastableand tends MCSWEEN ET AL.: MARS PATHFINDER ROCKS 8709

Nakhla Contoursof Pyroxenes "11zm"Band Minima • Normative 0.5 Wo Diopside

Band Position Pathfinder Shergotty Sulfur-free Pyroxenes.. 0.96 ß Rock 0.94 Band PosEions

En Normative Es Hypersthene Figure 20. Pyroxenequadrilateral (En = MgSiO3, Fs = FeSiO3,Wo = CaSiO3), illustratingthe analyzed compositionsof zonedaugite (open squares) in theNakhla meteorite [Harvey and McSween, 1992], pigeonite (solidcircles) and augite(solid squares)in the Shergottymeteorite [Stoffier et al., 1986], and calculated compositionsof nonnativediopside and hypersthene in the Pathfindersulfur-free rock (Table 3). Contoursof the "1 gin" bandminima for pyroxenesof variouscompositions are taken from Cloutisand Gaffey [ 1991].. Pyroxeneband positionsfor bulk Nakhla and basaltic shergottites(Shergotty and EET^79001B) are illustratedby shadedboxes [McFadden, 1987; Sunshine et al., 1993]. Thedownward slope of thecontours toward the Fs side of the quadfilaterialsuggests that the absorptionband for the iron-rich pyroxenesin the sulfur-freerock might occur at a wavelenth>1000 nm, possiblyaccounting for its absencein IMP spectra. Alternatively, a high proportionof calcium-richpyroxene, as in Nakhla, and breakdownof roetastableiron- rich pigeoniteinto fayalite plus silica, would also push the absorptionband minimum past the IMP spectral range.

to breakdown into fayalite plus silica [Lindsley and and partial melting [e.g., DePaolo, 1981; Hildreth and Anderson, 1983], so the actual assemblagein the rock may be Moorbath,1988; Foden and Green,1992]. ferroaugiteplus fayalitic olivine. The positionsof the band Magmas eruptedat subductionzones (andesitesin this minima for calcium-and iron-richaugite [Cloutis and Gaffey, settingare sometimescalled "orogenic andesites") follow one 1991] and iron-rich olivine [King and Ridley, 1987] lie of two differentiationpaths, depending on whetherthey beyond 1000 nm and thus would not be seenin IMP spectra. exhibit significant iron-enrichment with increasing The tendencyfor the pyroxeneband minima contoursto bend fractionation.Figures 21 and 22 comparethe tholeiiticand downward toward the iron-rich side of the quadrilateral calc-alkalinedifferentiation trends, which are controlledby (Figure 20) suggeststhat even a compositepyroxene band fractionationat low pressuresunder anhydrous conditions or minimum for augite and pigeonitemight occur beyond 1000 at higher(crustal) pressures with waterpresent, respectively [Groveand Baker, 1984; Sisson and Grove, 1993]. In Figure 21 (adaptedfrom Gill [1981]),the liquidlines of descentare 6.4. Andesite Petrogenesis reflectedin increasingSiO2, andorogenic suites fall on both On Earth, andesites are the second most abundant lava sidesof the boundaryseparating tholeiitic and calc-alkaline type,occurring mostly at destructiveplate margins. The direct rocks. Also shownin this figureis a liquid line of descent productionof andesiticmagma has sometimesbeen attributed leadingto anorogenic"icelandites," which are andesiticrocks to melting of mantle peridotiteunder hydrousconditions, formedby fractionalcrystallization of basaltic magmas in other fluxed by metamorphicdehydration of the subductedslab. tectonicsettings such as spreadingcenters and intraplate hot However, high-pressure experiments [Nicholls and spots [e.g., Carmichael, 1964; Ashley et al., 1995]; Ringwood, 1973; Mysen et al., 1974] have demonstratedthe icelandites invariably exhibit tholeiitic trends. The difficulty (or, in some cases,the impossibility) of forming Pathfindersulfur-free rock composition,as well as all the primary andesitic melts under such conditions. The individualrock analyses (which are adulteratedby adhering consensusnow is that andesitesform primarily by fractional dust), clearly fall within the tholeiite field, as do the crystallizationof basaltic magmas,a conclusionreached by compositionsof melts for SNCmeteorites. Figure 22 (adapted Gill [1981] after an extensive review of the published from Grove and Kinzler [1986]) illustratestholeiitic iron evidence. Lesser amountsof andesiticmagmas result from enrichmentin terms of MgO,FeO*, and alkalis; again, the other processes,such as magma mixing, crustalassimilation, Pathfinderrocks and SNC meltsare clearly tholeiitic. 8710 MCSWEEN ET AL.: MARS PATHFINDER ROCKS

Sulfur-free Rock Intercumulus Melts Pathfinder Rocks --• p AnorogenicIcelandites o Nakhlite

Orogenic Suites

5O 55 6O 65 SiO2 (Weight %)

Figure 21. Comparisonof tholeiitic and calc-alkaline differentiationtrends that lead to andesites,adapted from Gill [1981]. The calculatedsulfur-free rock and the compositionsof Pathfinderrocks fall within the tholeiitefield. Orogenicsuites are from volcanoesin Tonga,New Britain, and E1 Salvador(tholeiitic), and in New Britain, New Zealand, and Chile (calc-alkaline); anorogenic icelandites are from Iceland. Also illustrated are the compositionsof SNC meteorite melts: intercumulusliquids in basaltic shergottites (EETA79001, Longhi and Pan [1989]; Zagami, McCoy et al. [1992]; Shergotty,Hale et al. [1997]; Chassigny,Longhi and Pan [1989]; andNakhla, Longhi and Pan [1989] and Treiman [1993]), as well as the bulk compositionfor the QUE94201 shergottite[Warren and Kallemeyn, 1997], which, unlike other SNC meteorites,is thoughtto representa liquid composition[McSween et al., 1997].

FeO* ntercumulus \ Melts Nakhlite Shergoffite Sulfur-f•e• l• )Chessignite Rock •_.-:0,• ,•\ Pathfind•r•thfinder-•-•,•'I•:• t Shprgottite Rocks /•,'•• 'Anoroge nicIcelandire s

Figure 22. AFM diagramcomparing tholeiitic and calc-alkalinedifferentiation trends, adapted from Grove and Kinzler [1986]. The sulfur-free rock composition,Pathfinder rock analyses,and SNC melts (sources given in captionfor Figure 21) plot within the tholeiitic field. Orogenicsuites are from volcanoesin the LesserAntilles and California,and icelanditesuites are from Icelandand the GalapagosSpreading Center. MCSWEEN ET AL.: MARS PATHFINDER ROCKS 8711

Themost obvious chemical differences between orogenic Centerat 85øWand the experimentally determined Galapagos andesiresand anorogenicicelandires is that icelandireshave liquidline of descent[duster et al., 1989]. Onlythe TiO 2 higherFeO* and lower A1203 abundances ata given SiO 2 content differs appreciablyfrom the icelanditesin this content[Gill, 1981]. Theseare the samegeochemical differentiationsequence. propertiesthat make the Pathœmder sulfur-free rock (and SNC Becausethe compositionsof basalticshergottite liquids meteorites)distinctive. In fact, anorogenicicelandires seemto plotat plausiblepositions for a parentalmagma that providethe closestterrestrial analogy for the sulfur-freerock mightgenerate the sulfur-freerock (Figures 21 and22), we composition.Figure 23 comparesthe abundances of SiO2, havealso plotted the QUE94201 bulk composition (the only A1203, TiO2, FeO*,and MgO in thesulfur-free rock with shergottitethought to representa liquid [McSweenet al., thoseof fractionatedlavas from the GalapagosSpreading 1997])and the Shergottyintercumulus melt compositionin Figure23. Like the shergottitesand the norm of the sulfur-free rock, icelanditescommonly contain both high-calciumand 7O low-calciumpyroxenes [e.g., Carmichael, 1964]. Nonnative Fs contentsof diopside (Fs31 ) and hypersthene(Fs61 ) in 65 Sulfur-freeRock icelandites [Carmichael, 1964] are only slightly less iron- rich than normative pyroxenesin the sulfur-free rock (Fs37 Od 6O and Fs75, respectively). Given the youngcrystallization ages Shergotty Intercumulus Melt of shergottites(summarized by McSween [1994]) and old ages (/) 55 of Pathfinder rocks as inferred from their mobilization by ancient floods, it is unrealistic to expect that shergottite 5O magmascould actuallyhave given rise to the sulfur-freerock ,•,,,QUE94201 composition. However, some geochemicalcharacteristics of 45 theMartian mantle, such as A1203 depletion and FeO* 16 enrichment[Wtinke and Dreibus, 1988], might be imprinted on magmasof any age. The shergottiteliquid lines of descent, _ GalapagosLavas : calculated using the MELTS program [Ghiorso and Sack, 1995] at low pressure and appropriate oxygen fugacities 14 ; ..'. - ß I ß (QFM for Shergotty, QFM-3 for QUE94201), deviate . I '. significantlyfrom that of Galapagosmagmas (Figure 23) and ß do not passthrough or near the compositionof the sulfur-free '..." "½'!';' rock. Thus it is difficult to relate the Pathfinderrocks directly to shergottitesby fractionation processesat shallow depth, .--•'--•. Lineof Descent(Ca/c) although they clearly share some distinctive geochemical 10 properties[Rieder et al., 1997a]. We note, however,that the 1 MELTS program does not always correctly predict the crystallizationbehavior of highly evolved magmas [Yang et al., 1996]. The chemical similarities between the sulfur-free rock and terrestrial anorogenicicelandires imply that any attemptsto •• ' ' , ' G;lapa;os'Llqul.• interpret the andesitic compositions of these rocks as Lineof Descent(Exp) requiringa significantrole for water in their petrogenesis,or 4 even the existenceof Martian plate tectonics,are probably unwarranted. The modest amountsof water that may have existedin the lithosphereof Mars were probablytoo small to generate much silica-rich magma [Campbell and Taylor, 1983]. A more plausibleexplanation for these rocks is that they formedby fractionalcrystallization of basalticmagmas, possiblyrepresented by some rocks of the Hesperianridged

I I I I 2O Figure 23. Chemical variationsin fractionatedlavas from the ß _•-•;•_ .-. GalapagosSpreading Center at 85øWand the experimentally determinedGalapagos liquid line of descent[duster et al., •, 16 ß t'•.• '-•" .•,.,- . • '•. ß 1989] are compared with the Pathfinder sulfur-free rock e• / ß ß ee-• .... composition.The Pathfinder rock is a closematch to highly evolvedicelandite members of this mid-oceanridge basalt -',':.:, (MORB)differentiation sequence, except for its lowerTiO2 8 abundance.Also illustratedare the intercumulusliquid . Sulfur-fr•Rock ß ß compositionfor Shergotty[Hale et al., 1997]and the bulk compositionof the QUE94201 shergottite[Warren and 4 - • I I I I I I I Kallerneyn,1997], as well as theircalculated liquid lines of 0 2 4 6 8 descent. Fractionationof basalticshergottite magma at shallowdepths apparently cannot generate a melthaving the MgO Pathfinderandesite composition. 8712 MCSWEEN ET AL.: MARS PATHFINDER ROCKS

plains unit that cropsout near the Pathfindersite [Britt et al., spectral trends: rocks of the prilnary spectral trend, which 1998]. Rutherford and Hess [ 1981] discussedthe likelihood share a reflectance peak at 750 nm, include most materials that melting and fractionation processesmight generate (both rocks and soils) in the optical surface. Only maroon siliceous magmas on Mars, concluding that it might be rocks and spectrally similar brown soils are assignedto the appropriate to think of Mars as intermediate between the secondaryspectral trend, defined by reflectance peaks at Moon (where silica-rich lithologiesare very uncommon)and longer wavelengths. Iceland (which has -10 vol % of exposedicelandite). 3. The spatial pattern of spectralvariations in rocks and Comparisonof the aluminacontents of fractionatingSNC their relationshipto wind directionsuggest that the sourceof and mid-oceanridge basalt (MORB) melts (Figure 23) with spectral heterogeneity(i.e., the primary spectral trend) is the A1203 abundance of the Pathfinder sulfur-free rock primarily thin ferric coatingsof red aeoliandust on dark rocks. suggeststhat its parent basaltic magma was more aluminous The ferric phasein primarytrend coatings is ambiguousdue to than SNC parent magmas. Such a magma compositionmay lack of a diagnosticspectral signature. The secondaryspectral signalmelting of a relatively primitive Martian mantle source, trend apparentlyrequires coating by a different ferric nfineral one that had not yet been depleted in A1203 by partial with distinct spectral properties (possibly maghemite or melting [Longhiet al., 1992]. Becausedifferentiation of Mars ferrihydrite). occurredwithin the first 100 Myr of solar systemhistory, as 4. Preliminary APXS analysesof most elementsin rocks inferredfrom Sm-Nd and Hf-W isotopicsystematics in Martian plot as nearly linear arrays, interpreted as mixing lines meteorites[Harper et al., 1995;Lee and Halliday, 1997], this between a single rock compositionand adheringdust. This may imply that the rocks at the Pathfinderlanding site are interpretationis supportedby a strong correlationbetween very old. Heavily crateredNoachian crust outcropsnear and rock sulfur contentsand red/blue spectralratios measuredat possiblybeneath channel deposits at the Pathfinderlanding the same spots. Extrapolationsof regressionlines to zero site, and theserocks could have been sampledby floodsor by sulfur give the approximate composition of a presumed impact. igneousrock. Its chemistrycorresponds to that of andesitic The andesiticcomposition of the Pathfindersulfur-free volcanic rock. rock resemblesthe average compositionof Earth's crust. Further possible interpretationsand inferencesderived Riederet al. [1997a] suggestedthat Pathfinderrocks might from the above data are as follows. providea morerepresentative sampling of the ancientMartian 5. The nonnative mineralogy of the sulfur-free rock is crust than does the ALH84001 meteorite, a 4.5 Gyr-old consistentwith norms of common terrestrial volcanic rocks, orthopyroxenite.The widespreadoccurrence of rockswith suggestingthat the relatively large rocks analyzedby APXS andesiticcompositions on Mars was originally suggested are igneous.The calculated(probably minimmn) viscosity of a based on Mariner 9 thermal emissionspectra of suspended magma having the compositionof the Pathfinder sulfur-free dust[Hanel et al., 1972]. Formationof an andesiticcrust by rock overlapsthe range of viscosityestimates for Martian lava simple fractional crystallizationwould require a huge flows based on models that relate rheology to flow quantityof basalticmagma, however, so the petrogenetic morphology. However, the spectralmottling seenin several mechanisminferred to producethe Pathœmdersulfur-free rock rocks (not analyzed by APXS) and the suggestionof rounded would probablybe more consistentwith a mafic average clastsand socketson the surfacesof a few rocksmight imply crustalbulk composition.Production of an andesiticcrust on that coarseclastic sedimentaryrocks or impact brecciasare Mars might have beenaccomplished by repeatedperiods of also present. It is also possiblethat the analyzedrocks are partialmelting, probably over an extendedperiod of time, as coated with siliceous weathering rinds, although the on Earth. Althoughsuch a complexscenario cannot be ruled requirementfor intermittentrainfall or dew, if the processis out, local fractionationof basalticmagma provides a simpler analogousto that on Earth, makesthis less likely. explanationfor the Pathfinderrocks. 6. Large, rounded bouldersof the maroon spectralclass may predate smaller rocks at the site, and the ferric mineral 7. Summary and Conclusions coating that defines the maroon classmay be a mineralogic signatureof the flood environmentin which the boulderswere Chemical,multispectral, and texturalstudies of rocksat the emplaced. Smaller, more angular rocks may represent Mars Pathfinder site allow the following conclusionsand depositionfrom a secondflood event or ejectafroln a nearby interpretationsto be drawn. ilnpact crater. The latter event was followed by aeolian 1. Rock surfacesexhibit combinationsof pitted, knobby, erosion and deposition. smooth,lineated, and bumpytextures. Original textures have 7. The absenceof a "1 gm" pyroxeneabsorption band in the been.overprinted by aeolianabrasion, weathering, and dust spectraof any Pathfinderrocks is perplexing. The presenceof coatings,complicating their interpretations.One rock has a somemafic phaseis supportedby upwardcurvature in infrared structureresembling exfoliation, and several othersresemble portionsof the spectraof gray rock surfaceswith less dust polygonalprisms which may suggestcolumnar jointing. The cover. One possibleexplanation for the lack of a resolved equivocalnature of most observedsurface textures means that band center is that the rocks are volcanic or impact glasses, they cannotbe usedto determineuniquely the origin of the which have absorptionbands centered near 1100 nm (beyond rocks. the IMP spectral range). Alternatively, the band may be 2. Four spectral classesof rock surfacesare identified. maskedby magnetitein the rock, or the presenceof iron-rich Gray andred surfacescommonly occur on the samerocks, with augiteand fayalitic olivine (the latter formedby breakdownof gray portionson the upwindsides or at erodededges and red metastable iron-rich pigeonite) may have pushed the portionson downwindsurfaces. Several pink rocksoccur as absorptionband beyond 1000 nm. tabularrock-like masses,probably soil crusts. Maroon rocks 8. The chemistry of sulfur-free rock, if igneous, is are mostly large, roundedboulders in the far field. Correlated consistentwith fractionation along a tholeiitic trend. Its spectralparameters allow the rocks to be assignedto two composition is most similar to terrestrial icelandites MCSWEEN ET AL.: MARS PATHFINDER ROCKS 8713

(andesites formed by fractional crystallization of basaltic Carr, M. H., The Surface of Mars, 226 pp., Yale Univ. Press,New magma in anorogenic settings), rather than like the more Haven, Conn., 1981. Carroll, M. R., and J. D. Webster, Solubilities of sulfur, noble gases, commonorogenic andesites. The rock compositiondoes not halogens,in Volatilestn Magmas,Rev. Mineral., vol. 30, editedby lie along calculatedliquid lines of descentfor shergottite M. R. Carroll and J. R. Holloway, pp. 232-279, Mineral. Soc. Am., melts at low pressure, but instead may require a more Washington,D.C., 1994. aluminousbasaltic parent. Casey, W. H., H. R. Westrich, J. F. Banfield, G. Ferruzzi, and G. W. 9. Speculationsabout possibleroles for water and plate Arnold,Leaching and reconstructionat the surfacesof dissolving tectonicsin Martian petrogenesis,based on andesiticrocks at chain-silicateminerals, Nature, 366, 253-256, 1993. Cashman, K. V., and J.P. Kauahikaua, Reevaluation of vesicle the Pathfinder site, are not justified. Although these rocks distributionsin basalticlava flows, Geology, 25, 419-422, 1997. may representancient crust, as inferredfrom higheralumina Cattermole,P., Sequence,rheological properties, and effusionrates of concentrationsthat mightcharacterize partial melts from a volcanicflows at Alba Patera,Mars, Proc. Lunar Planet.Sci. Conf. primitivemantle, a complexorigin involving repeated crustal 17th, Part 2, .1.Geophys. Res., 92, suppl.,E553-E560, 1987. Christensen,P. R., The spatialdistribution of rockson Mars, Icarus, 68, meltingseems less likely than their derivation by fractional 217-238, 1986. crystallizationof basalticmagmas. Clark, B.C., A. K. Baird, R. J. Weldon, D. M. Tsusaki, L. Schnabel,and M.P. Candelaria, Chemical composition of Martian fines, J. Acknowledgments.We appreciatethoughtful reviews by Tim Geophys.Res., 87, 10,059-10,067, 1982. Grove,Vicky Hamilton, and Mac Rutherford. Tom Daley assisted with Cloutis, E. A., and M. J. Gaffey, Pyroxene spectroscopyrevisited: imageprocessing and spectral analysis. Support for thisresearch was Spectral-compositional correlations and relationship to providedby NASA through contracts to the individual investigators or geothermometry,J. Geophys.Res., 96, 22,809-22,826, 199!. wascarried out at the JetPropulsion Laboratory under contract with NASA. Cloutis, E. A., M. J. Galley, D. G. W. Smith, and R. St. J. Lambert, Reflectancespectra of mafic silicate-opaqueassemblages with applicationsto meteoritespectra, Icarus, 84, 315-333, 1990. References Crisp,J. A., The effect of thin coatingsof dustor soil on the bulk APXS compositionof the underlyingrocks at the Pathfinderlanding site, Abbott, P. L., and T. E. Smith, Sonora,Mexico, sourcefor the Eocene Proc.Lunar Planet. Sci.Conf., 29, 1962, 1998. PowayConglomerate of southernCalifornia, Geology, 17, 329-332, 1989. Crisp, J., A. B. Kahle, and E. A. Abbott, Thermalinfrared spectral characterof Hawaiian basaltic glasses,J. Geophys.Res., 95, Adams,J. B., M. O. Smith,and P. E. Johnson,Spectral mixture modeling: 21,657-21,669, 1990, A new analysisof rock and soil typesat the Viking Lander1 site,J. DePaolo,D. J., Trace elementand isotopiceffects of combinedwall Geophys.Res., 91, 8098-8112, 1986. rock assimilationand fractionalcrystallization, Earth Planet. Sci. Aitchison,J., The StatisticalAnalysis of CompositionalData, 416 pp., Lett., 53, 189-202, 1981. Chapmanand Hall, New York, 1986. Dom, R. I., and N. Meek, Rapid formation of rock varnish and other Allen, C. C., and J. L. Conca, Weatheringof basalticrocks under cold, rockcoatings on slagdeposits near Fontana, California, Earth SurJq. arid conditions: Antarctica and Mars, Proc. Lunar Planet. Sci. ProcessesLandforms, 20, 547-560, 1995. Conf., 21, 711-717, 1991. Ash,R. D., S. F. Knott,and G. Turner,A 4-Gyr shockage for a Martian Erard, S., J. Mustard, S. Murchie, M.-P. Bibring, P. Cerroni, and A. meteoriteand implicationsfor the crateringhistory of Mars,Nature, Corodine,Martian aerosols:Near-infrared spectral properties and 380, 57-59, 1996. effects on the observation of the surface, Icarus, I I I, 317-337, Ashley, P.M., R. A. Duncan, and C. A. Freebrey,Ebor Volcano and 1994. Crescent Complex, NortheasternNew South Wales - Age and Evans, D. L., and J. B. Adams, Comparison of Viking Lander geologicaldevelopment, Aust. .1. Earth Sci., 42, 471-480, 1995. multispectral images and laboratory reflectance spectra of Baird, A. K., and B.C. Clark, On the originaligneous source of Martian terrestrial samples,Proc. Lunar Planet. Sci. Conf., 10, 1829-1834, fines, Icarus, 45, 113-123, 1981. 1979. Bell, J. F., III, T. McCord, and P. Owensby,Observational evidence of F .art,T. G., and J. B. Adams, Rock coatingsin Hawaii, Geol. Soc. Am. crystalline iron oxides on Mars, .1. Geophys.Res., 95, 14,447- Bull., 95, 1077-1083, 1984. 14,461, 1990. Fischer,E., andC. Pieters,The continuumslope of Mars:Bi-directional Bell, J. F., III, R. V. Morris, and J. B. Adams, Thermally altered reflectance investigations and applicationsto Olympus Mons, palagonitictephra: A spectraland processanalog to the soil and Icarus, 102, 185-202, 1993. dustof Mars, J. Geophys.Res., 98, 3373-3385, 1993. Foden,J. D., and D. H. Green, Possiblerole of amphibolein the origin Bell, J. F., III, M. Wolff, P. James,T. Clancy, S. Lee, and L. Martin, of andesitc - Some experimental and natural evidence, Contrib. Mars surfacemineralogy from Hubble SpaceTelescope imaging Mineral. PetroL, 109, 479-493, 1992. during 1994-1995: Observations,calibration, and initial results,J. Garvin, J. B., and J. J. Frawley, Martian impact cratermorphometry: Geophys.Res., 102, 9109-9123, 1997. Preliminaryresults from the Mars Orbiter LaserAltimeter (MOLA) Bhatia,M. R., Rare earthelement geochemistry of AustralianPaleozoic investigation(abstract), Lunar Planet. Sci., 29, 1495, 1998. graywackesand mudrocks,Sediment. Geol., 45, 97-113, 1985. Ghiorso,M. S., and R. O. Sack, Chemicalmass transfer in magmatic Bishop, J., Spectroscopic analyses of chemically altered proceses,IV., A revisedand internallyconsistent thermodynamic montmorillonitesand applicationsto the soilsof Mars, Ph.D. thesis, model for the interpolationand extrapolationof liquid-solid Brown Univ., Providence,R.I., 1995. equilibria in magmatic systems at elevated temperatures and Bridges,N., R. A. Anderson,J. A. Crisp, T. Economou,and R. Reid, pressures,Contrib. Mineral. PetroL, I 19, 197-212, 1995. Separating dust and rock APXS measurements based on Gill, J., Orogenic Andesitesand Plate Tectonics,390 pp., Springer- multispectraldata at the Pathfinderlanding site, Eos Trans.AGU, 78, Verlag New York, 1981. Fall Meet. Suppl.,F402, 1997. Golombek, M.P., The Mars Pathfindermission, J. Geophys.Res., 102, Britt, D. T., et al., The mineralogyof the Mars Pathfinderlanding site 3953-3965, 1997. (abstract),Lunar Planet. Sci., 29, 1776, 1998. Golombek, M.P., et al., Overview of the Mars Pathfinder mission and Bums, R., The fate of iron on Mars: Mechanism of oxidation of basaltic assessmentof landing site predictions,Science, 278, 1743-1748, mineralsto ferric-bearingassemblages, in Workshopon Chemical 1997a. Weatheringon Mars, LPI Tech.Rep., 92-04, Part 1, 8-9, 1992. Golombek, M.P., R. A. Cook, H. J. Moore, and T. J. Parker, Selectionof Burns, R., Schwertmannite on Mars: Deposition of this ferric the Mars Pathfinderlanding site, J. Geophys.Res., 102, 3967-3988, oxyhydroxysulfatemineral in acidic saline meltwaters(abstract), 1997b. Lunar Planet. Sci., 25, 203-204, 1994. Gooding, J. L., Chemical weathering on Mars: Thermodynamic Campbell,I. H., and S. R. Taylor, No water,no granites- no oceans,no stabilitiesof primary minerals(and their alterationproducts) from continents,Geophys. Res. Lett., IO, 1061-1064,1983. mafic igneousrocks, Icarus, 33, 483-513, 1978. Carmichael,I. S. E., The petrologyof Thingmuli,a Tertiaryvolcano in Gooding,J. L., Soil mineralogyand chemistryon Mars: Possibleclues easternIceland, ,I. Petrol., 5, 435-460, 1964. from saltsand claysin SNC meteorites,Icarus, 99, 28-41, 1992. 8714 MCSWEEN ET AL.: MARS PATHFINDER ROCKS

Greeley,R., andJ. D. Iverson,Wind as a GeologicalProcess, 333 pp., Lindsley,D. H., and D. J. Andersen,A two-pyroxenethermometer, CambridgeUniv. Press,New York, 1985. Proc. Lunar Planet. Sci. Conf. 14th, Part 2, J. Geophys.Res., 88, Greeley,R., E. Theilig,J. E. Guest,M. H. Carr,H. Masursky,and J. A. suppl.,A887-A906, 1983. Cutts,Geology of ChrysePlanitia, or. Geophys.Res., 82, 4093-4109, Longhi,J., andV. Pan,The parentmagmas of the SNC meteorites,Proc. 1977. Lunar Planet. Sci. Conf., 19, 695-709, 1989. Greeley,R., R. N. Leach,S. H. Williams,B. R. White, J. B. Pollack,D. Longhi, J., E. Knittie, J. R. Holloway, and H. W'•ke, The bulk H. Krinsley,and J. R. Marshall,Rate of wind abrasionon Mars,or. composition,mineralogy and internal structureof Mars, in Mars, Geophys.Res., 87, 10,009-10,024, 1982. editedby H. H. Kieffer, B. M. Jakosky,C. W, Snyder,and M. S. Grove, T. L., and M. B. Baker, Phaseequilibrium controls on the Matthews,pp. 184-208, Univ. of Ariz. Press,Tucson, 1992. tholeiiticversus calc-alkaline differentiation trends, or. Geophys. Lopes,R. M. C., andC. R. J. Kilburn, Emplacementof lava flow fields: Res., 89, 3253-3274, 1984. Applicationof terrestrialstudies to Alba Patera,Mars, or. Geophys. Grove, T. L., and R. J. Kinzler, Petrogenesisof andesites,Annu. Rev. Res., 95, 14,383-14,397, 1990. Earth Planet. Sci., 14, 4!%454, 1986. Malin,M., et al., Geologyand geomorphology of the Pathfinder landing Guinness,E. A., C. E. Left, andR. E. Arvidson,Two Marsyears of site,Eos Trans.AGU,, 78(46),Fall Meet.Suppl., F395, !997. surfacechanges seen at the Viking landingsites, or. Geophys. Res., MarsPathfinder, Foldout plates, Science, 278, 1734-1742,1997. 87, 10,051-10,058, 1982. McCauley,J. F., C. S. Breed,F. El-Baz, M. I. Whitney,and M. J. Guinness,E. A., R. E. Arvidson,M. A. Dale-Bannister,R. B. Singer,and Grolier,Pitted and fluted rocks in the westerndesert of Egypt: E. A. Bruckenthal,On the spectralreflectance properties of Viking comparisons,or.Geophys. Res., 84, 8222-8232,1979. matedhalsexposed at the Viking landingsites, Proc. Lunar Planet. McCoy,T. J.,G. J. Taylor,and K. Keil,Zagami: Product of a two-stage Sci. Conf. 17th,Part 2, or.Geophys. Res., 92, suppl.,E575-E587, 1987. magmatichistory, Geochim. Cosmochim. Acta, 56, 3571-3582,1992. McFadden, L. A., Spectral reflectance of SNC meteorites: Guinness,E. A., R. E. Arvidson,I. H. D. Clark,and M. K. Shepard, Relationshipsto Martiansurface composition, LPI Tech.Rep., 88- Optical scattering properties of terrestrial varished basalts 05, 88-90, 1987. comparedwith rocks and soils at the Viking Lander sites, or. McGetchin,T. R., M. Settle,and J. W. Head,Radial thickness variation Geophys.Res., 102, 28,68%28,703, 1997. in impactcrater ejecta: Implications for lunarbasin deposits, Earth Hale, V. P.S., H. Y. McSweenJr., and G. A. McKay, Cumulus Planet. Sci.Lett., 20, 226-236, 1973. pyroxenein Shergotty:The discrepancybetween experimental and McSween, H. Y., Jr., What we have learned about Mars from SNC observationalstudies (abstract), Lunar Planet. Sci., 28, 495-496, 1997. meteorites,Meteorittcs, 29, 757-779, 1994. Hanel,R., et al., Investigationof the martianenvironment by infrared McSween,H. Y., Jr.,and E. Jarosewich,Petrogenesis of the Elephant spectroscopyon Mariner 9, Icarus, 17, 423-442, 1972. MoraineA79001 meteorite: Multiplemagma pulses on the Harper,C. L., L. A. Nyquist,B. Bansal,H. Weismann,and C.-Y. Shih, shergottiteparent body, Geochim. Cosmochim. Acta, 47, 1501- 15113, 1983. Rapid accretionand early differentiationof Mars indicatedby McSween,H. Y., Jr.,D. D. Eisenhour,L. A. Taylor,M. Wadhwa,and 142Nd/144NdinSNC meteorites, Nature, 267, 213-217, 1995. G. Crozaz,QUE94201 shergottite: Crystallization of a Martian Harvey, R. P., and H. Y. McSweenJr., The petrogenesisof the basalticmagma, Geochim. Cosmochim. Acta, 60, 4563-4569, 1997. nakhlites: Evidenceform cumulatemineral zoning, Geochim. Moore,H. J., C. A. Hodges,and D. H. Scott,Multiringed basins - Cosmochim.Acta, 56, 1655-1663, 1992. illustratedby Orientaleand associated features, Proc. Lunar Sc•. Herron,M. M., Geochemicalclassification of terrigenoussands and Conf. 5th, 71-!00, 1974. shalesfrom core or log data, or. Sediment.Petrol., 58, 820-829, 1988. Moore,H. J.,D. W. G. Arthur,and G. G. Schaber,Yield strengths of Hildreth,W., andS. Moorbath,Crustal contributions to arc magmatism flowson the Earth, Mars, and Moon, Proc. Lunar Planet. Sci. Conf. 9th, 335 !-3378, !978. in theAndes of centralChile, Contrib. Mineral. Petrol., 98, 455-489, 1988. Morris,R. V., andH. V. LauerJr., Matrixeffects for reflectivity Hodges,C. A., and H. J. Moore,Atlas of volcaniclandforms on Mars, spectraof dispersednanophase (superparamagnetic) hematite with U.S.Geol. Surv. Prof. Pap., 1534,194 pp., !994. applicationto martianspectral data, or. Geophys. Res., 95, 5101- Hon, K., J. Kauahikaua,R. Denlinger,and K. Mackay,Emplacement 5109, 1990. and inflation of pahoehoe sheet flows: Observations and Morris,R., H. Lauer,C. Lawson,E. Gibson,G. Nace,and C. Stewart, measurementsof active lava flows on Kilauea Volcano, Hawaii, Spectraland other physiochemical properties of submicronpowders Geol.Soc. Am. Bull., 106, 351-370, 1994. of hematite(a-Fe203) , maghemite (g-Fe203) , magnetite (Fe304) , Hulme, G., The interpretationof rheologicalproperties and effusion goethite(a-FeOOH), and lepidocroicite (g-FeOOH), J. Geophys. rate of an OlympusMons lava,Icarus, 27, 20%213, 1976. Res., 90, 3126-3144, 1985. Hunt, G. R., J. W. Salisbury,and C. J. Lenhoff, Visible and near Morris,R. V., J. L. Gooding,H. V. LauerJr., and R. B. Singer,Origins infrared spectraof mineralsand rocks, IX, Basic and ultrabasic of Marslike spectraland magneticproperties of a Hawaiian igneousrocks, Mod. Geol., 5, 15-22, 1974. palagoniticsoil, J. Geophys.Res., 95, 14,427-14,434,1990. Hviid, S. F., et al., Magneticproperties experiments on the Mars Morris, R. V., D.C. Golden,J. F. Bell III, H. V. Lauer Jr., and J. B. Pathfinderlander: Preliminaryresults, Science, 2 78, 1768-1770, Adams, Pigmenting agents in martian soils: Inferencesfrom 1997. spectral,M6ssbauer, and magneticproperties of nanophaseand Irvine, T. N., and W. R. A. Baragar, A guide to the chemical other iron oxides in Hawaiian palagoniticsoil PN-9, Geochim. classificationof the commonvolcanic rocks, Can. J. Earth Sci.,8, Cosmochim.Acta, 57, 4597-4609, 1993. 523-548, 1971. Morris,R. V., D.C. Golden,J. F. Bell III, andH. V. LauerJr., Hematite, Jambon,A., Earthdegassing and large-scale geochemical cycling of pyroxene,and phyllosilcateson Mars: Implicationsfrom oxidized volatileelements, in Volatilesin Magmas,Rev. Mineral., 30, edited impactmelt rocksfrom ManicouaganCrater, Quebec, Canada, or. by M. R. Carroll and J. R. Holloway,pp. 479-517, Mineral. Soc. Geophys.Res., 100, 5319-5328, 1995. Am., Washington,D.C., 1994. Mustard,J. F., andJ. F. Bell III, New compositereflectance spectra of Juster,T. C., T. L. Grove,and M. R. Perfit,Experimental constraints on Marsfrom 0.4 to 3.14 gm, Geophys.Res. Lett., 21,353-356, 1994. the generationof FeTi basalts,an&sites, and rhyodacitesat the Mustard,J. F., and J. E. Hays, Effectsof hyperfineparticles on GalapagosSpreading Center, 85øW and 95øW, or. Geophys. Res., reflectancespectra from 0.3 to 25 micrometers,Icarus, 125, 145- 94, 9251-9274, 1989. 163, 1997. King,T. V. V., andI. Ridley,Relation of thespectrscopic reflectance of Mustard,J. F., andJ. M. Sunshine,Seeing through the dust: Martian olivineto mineralchemistry and some remote sensing applications, crustalheterogeneity and links to the SNC meteorites,Science, 267, or.Geophys. Res., 92, 11,457, 1987. 1623-1626, 1995. Kortum,G., ReflectanceSpectroscopy, 366 pp., Springer-Verlag,New Mustard,J. F., et al., The surfaceof SyrtisMajor: Composition of the York, 1969. volcanicsubstrate and mixing with altered dust and soil, or. Geophys. Le Bas, M. J., R. W. Le Maitre, A. Streckeisen,and B. Zanettin, A Res., 98, 3387-3400, 1993. chemical classification of volcanic rocks based on the total alkali- Mustard, J. F., S. Murchie, S. Erard, and J. M. Sunshine,In situ silicadiagram, o r. Petrol., 27, 745-750, 1986. compositionsof Martianvolcanics: Implications for the mantle,J. Lee, D.-C., and A. N. Halliday, Core formation on Mars and Geophys.Res., 102, 25,605-25,615, 1997. differentiatedasteroids, Nature, 388, 854-857, 1997. Mysen,B. O., I. Kushiro,I. A. Nicholls,and A. E. Ringwood,A possible MCSWEEN ET AL.: MARS PATHFINDER ROCKS 8715

mantle origin for andesitemagmas: Discussionand replies,Earth Sunshine,J. M., L. A. McFadden,and C. M. Pieters,Reflectance spectra Planet. Sct. Lett., 21, 221-229, 1974. of the ElephantMoraine A79001 meteorites: Implicationsfor Nicholls, I. A., and A. E. Ringwood,Effects of water on olivine stability remotesensing of planetarybodies, Icarus, 105, 79-91, 1993. in tholeiite and the production of silica-saturatedmagas in the Tanaka,K. L., Sedimentaryhistory and mass flow structuresof Chryse island-arc environment, d. Geol., 81,285-300, 1973. and Acidalia Planitiae, Mars, J. Geophys.Res., 102, 4131-4149, Parker, T. J., "Super resolution"of the Mars Pathfinderlanding site, 1997. using manual techniques(abstract), Lunar Planet. Sci., 29, 1817, Thordarson,T., and S. Self, Sulfur,chlorine, and fluorine degassing and 1998. atmosphericloading by the Rozaeruption, Columbia River Basalt Parker, T. J., and J. W. Rice Jr., Sedimentarygeomorphology of the Group,Washington, USA, J. Volcanol.Geotherm. Res., 74, 49-73, Mars Pathfinderlanding site, d. Geophys.Res., 102, 25,641-25,656, 1996. 1997. Toramam,A., A. Ishiwatari,M. Matsuzawa,N. Nakamura,and S. Arai, Reid, R. J., A. Dummel, R. B. Singer,J. R. Johnson,J. F. Bell III, and T. Vesiclelayering in solidifiedintrusive magma bodies: A newly Daley, Imager for Mars Pathfindercalibration, Eos Trans.AGU, 78, recognizedtype of igneousstructure, Bull. Volcanol.,58, 393-400, Fall Meet. Suppl., F402, 1997. 1996. Rieder, R., et al., The chemical compositionof Martian soil and rocks Treiman,A. H., The parentmagma of the Nakhla (SNC) meteorite, returned by the mobile alpha proton X-ray spectrometer: inferredfrom magmaticinclusions, Geochim. Cosmochtm. Acta, 57, Preliminary resultsfrom the X-ray mode, Science, 278, 1771-1774, 4753-4767, 1993. 1997a. Treiman, A. H., A petrographichistory of Martian meteorite Rieder, R., H. W/inke, T. Economou, and A. Turkevich, Determination ALH84001: Two shocksand an ancient age, Meteoritics, 30, 294- of the chemical compositionof Martian soil and rocks: The alpha 302, 1995. proton X-ray spectrometer,d. Geophys. Res., 102, 4027-4044, W•inke, H., and G. Dreibus, Chemical compositionand accretional 1997b. historyof terrestrialplanets, Philos. Trans. R. Soc.London, Series A, Rotto, S., and K. L. Tanaka, Geologic/geomorphicmap of the Chryse 235, 545-557, 1988. Planitia region of Mars, U.S. Geol. Surv. Misc. Invest. Ser. Map, 1- Warren,P. H., andG. W. Kallemeyn,Yamato-793605, EET79001, and 2441, 1995. otherpresumed Martian meteorites: Compositional clues to their Rover Team, Characterizationof the Martian surface depositsby the origins,Antarct. Meteorite Res., 10, 61-81, 1997. Mars PathfinderRover, Sojourner,Science, 278, 1765-1768, 1997. Warren, P. H., P. Claeys, and E. Cedillo-Pardo,Mega-impact melt Rutherford, M. J., and P. C. Hess, Granite genesis: Processesand petrology(Chicxulub, Sudbury, and the Moon): Effectsof scaleand variablesfor Mars (abstract),Lunar Planet. Sci., 12, 915-917, 1981. other factors on potential for tractional crystallization and Schultz, P. H., and J. F. Mustard, Martain impact glass:Generation and developmentof cumulates,Spec. Pap. Geol. Soc.Ant., 307, 105- evidence(abstract), Lunar Planet. Sci., 29, 1847, 1998. 124, 1996. Self, S., T. Thordarson,and L. Keszthelyi,Emplacement of continental Wendlandt, W. W., and H. G. Hecht, Reflectance Spectroscopy,298 flood basalt lava flows, in Large Igneous Provinces: Continental, pp., Wiley-Interscience,New York, 1966. Oceamc, and Planetary Flood Volcanism,Geophys. Monog. Ser., Whitney,M. I., andR. V. Dietrich,Ventifact sculpture by windblown vol. 100, edited by J J. Mahoney and M. F. Coffin, pp. 381-410, dust, Geol. Soc.Ant. Bull., 84, 2561-2582, 1973. AGU, Washington,D.C., 1997. Yang,H.-J., R. J. Kinzler,and T. L. Grove,Experiments and models of Sharp,R. P., and M. C. Malin, Surfacegeology from Viking landerson anhydrous,basaltic olivine-plagioclase-augite saturated melts from Mars: A secondlook, Geol. Soc. Am. Bull., 95, 1398-1412, 1984. 0.001 to 10 kbar, Contrib.Mineral. Petrol., 124, 1-18, 1996. Shaw, H. R., Viscosities of magmatic silicate liquids - an empirical Zimbelman,J. R., Estimatesof rheologicalproperties for flows on the methodof prediction,Am. d. Sct., 272, 870-893, 1972. Martian volcano AscraeusMons, Proc. Lunar Planet. Sci. Conf Sheller, T. D., and R. V. Morris, Eft•ct of a ferric weatheringrind on 16th,Part 1, J. Geophys.Res., 90, suppl.,D157-D162, 1985. the optical and M6ssbauer spectra of a basaltic rock (abstract), Lunar Planet. Sci., 29, 1327, 1998. R. C. Anderson,N. T. Bridges,J. A. Crisp,M.P. Golombek,and T. J. Sherman,D., R. Burns,and V. Bums, Spectralcharacteristics of the iron Parker, Jet PropulsionLaboratory, 4800 Oak Grove Drive, California oxideswith applicationto the Martian bright regionmineralogy, d. Institute of Technology, Pasadena, CA 91109-8099. (robert.c. Geophys.Res., 87, 10,169-10,180, 1982. [email protected]; nathan.bridges•jpl.nasa.gov; [email protected]. Singer, R. B., The dark materialson Mars: I. New informationfrom nasa.gov; [email protected]. gov; [email protected]. gov) reflectance spectroscopyon the extent and mode of oxidation, J. F. Bell III, Centerfor Radiophysicsand SpaceResearch, 424 Space Lunar Planet. Sct. Conf., IIth, 1045-1047, 1980. Sciences Bldg., Cornell University, Ithaca, NY 14853-6801. Singer,R. B., Spectralevidence for the mineralogyof high-albedosoils ([email protected]) and duston Mars, d. Geophys.Res., 87, 10,159-10,168, 1982. D. T. Britt, Lunar and PlanetaryLaboratory, University of Arizona, Singer,R. B., and T. B. McCord, Large scalemixing of bright and dark Tucson,AZ 85721.([email protected]) materialsand implicationsfor analysisof spectralreflectance, Proc. J. Brtickner, G. Dreibus, R. Rieder, and H. Wltnke, Max-Planck-Institut Lunar Planet. Sci. Conf. 20th, 1835-1848, 1979. far Chemie,Saarstrasse 23, Mainz 55122, Germany.(bmeckner@mpch- Singer, R. B., and H. Y. McSween Jr., The igneouscrust of Mars: mainz.mpg.de; [email protected];[email protected] Compositional evidence from remote sensing and the SNC mainz.mpg.de;[email protected]. de) meteorites,in Resourcesof Near-Earth Space,edited by J. Lewis, T. Economou,Enrico Fermi Institute,University of Chicago,933 East M. S. Matthews,and M. L. Guerrieri,pp. 709-736, Univ. of Ariz. 56• St.,Chicago, IL 60637.([email protected]) Press, Tucson, 1993. A. Ghoshand H. Y. McSweenJr., Departmentof GeologicalSciences, Singer,R. B., and T. Roush,Spectral reflectance properties of University of Tennessee, Knoxville, TN 37996-1410. particulateweathered coatings on rocks:Laboratory modeling and ([email protected];[email protected]) applicabilityto Mars(abstract), Lunar Planet. Sci., 14, 708-709, J.P. Greenwood, Institute for Geophysicsand Planetary Physics, 1983. Universityof California at Los Angeles,Los Angeles,CA 90024-1567. Sisson,T. W., and T. L. Grove,Experimental investigations of the role ([email protected]) of H20 in calc-alkalinedifferentiation and subductionzone J. R. Johnson,U.S. GeologicalSurvey, 2255 North Gemini Drive, magmatism,Contrib. Mineral. Petrol., I 13, 143-166,1993. Flagstaff,AZ 86001.([email protected]) Smith,P. H., et al., Resultsfrom the Mars Pathfindercamera, Science, R. V. Morris, NASA JohnsonSpace Center, MC SN3, Houston,TX 278, 1758-1765, 1997a. 77058. Smith, P. H., et al., The imagerfor Mars Pathfinderexperiment, d. S. L. Murchie, Applied Physics Laboratory, The Johns Hopkins Geophys.Res., 102, 4003-4025,1997b. University, Johns Hopkins Road, Laurel, MD 20723. Soderblom,L. A., The compositionand mineralogyof the Martian ([email protected]) surfacefrom spectroscopic observations: 0.3 mmto 50 gm,inMars, R. Singer, Catalina Technologies,4740 East SunriseDrive, #312, editedby H. H. Kieffer,B. M. Jakosky,C. W. Snyder,and M. S. Tucson, AZ 85718. Matthews,pp. 557-597,Univ. of Ariz. Press,Tucson, 1992. St6ffler, D., et al., Shock metamorphismand petrographyof the Shergottyachondrite, Geochim. Cosmochim. Acta, 50, 889-913, (ReceivedFebruary 6, 1998;revised June 17, 1998; 1986. acceptedJuly 1, 1998.)