Table1.Dataforthedeterminationoftheturquoisebeadsourcearea;ppm,partspermillion.SnaketownHimalayaElementGroupGroupUnclassi-GroupGroupCrescentABfiedABTraceelementdata(ppm)Co1.671.781.951.301.781003.0Cr1.661.441.201.542.3067.3Eu0.260.320.280.270.63*Sb2.102.901.431.762.104.66Sc35.1101.515.136.2106.465.3Ta*****326.0SimilaritycoefficientsSnaketownGroupA1.000GroupB.8251.000Unclassified.544.3371.000HimalayaGroupA.946.808.6131.000GroupB.861.991.565.8401.000Crescent.143.177.092.208.1011.000*Notdetected.limitsfortheHimalaya-Snaketowntur-quoise,butwerepresentinmanyoftheothersamples.ThesebasicdifferencesservedtodistinguishtheCaliforniasourceareafromotherlocalitiesbysimpleexam-inationofthedata.TheSnaketownbeadsfellintotwogroupsoffiveandsevensampleseach,basedonSccontentsof35ppm(groupA)and100ppm(groupB),withIsampleun-classified.ThedataforsamplesfromtheHimalayaminealsofellintothesametwogroups,emphasizingtheneedformultiplesamplesfromeachgeographicsourcearea.Inordertocharacterizethemineareasstatistically,itwasnecessarytocomparetheconcentrationsofalltraceelementssi-multaneously.ThemultivariatestatisticdevisedbyBorchardtetal.(8),inwhichtheratioofeachelementconcentrationinapairofsamplesissummedanddividedbythenumberofelements,wasfoundtobethemostuseful forthesedata(9).Thecoefficientsthusobtainedarerepresenta-tiveofthedegreeofsimilarity,ordis-similarity,betweentwoormoresamples.Thatis,thecoefficientsareameansofquantifyingthedegreeofcorrelationbasedonsampleanalyticaldata.Perfectsim-ilarityresultsinavalueof1.0,andlargedifferencesresultincoefficientsnearzero.Borchardtetal.determinedbyreplicateanalysesthatcoefficientsabove.800wereindicativeofanaccuratecorrelationatthe95percentconfidencelevel.Conversely,valuesbelow.560indicatedthatapairofsamplesprobablywerenotfromthesamesite.TableIshowsthecorrelationcoefficientmatrixforthetwoSnaketowngroups,theunclassifiedbead,thetwoHimalayaminegroups,andfortheCrescentmineinNe-vada.OnthebasisofBorchardt'scriterion,thereisgoodcorrelationbetweensamples460fromSnaketownAandHimalayaA,andbetweenSnaketownBandHimalayaB.TheCrescentsampleisclearlyunrelated.Thesimilaritycoefficients,then,sub-stantiatedtheconclusionthattraceele-mentpatternsfortheSnaketownbeadscorrelatedwiththatofturquoisefromtheHimilayagroupofminesnearHalloronSprings,butnotwiththedatafromthe23otherminesanalyzed.The13turquoisebeads,however,representonlyaportionoftheturquoisefromthesite.SinceturquoisewaswidelytradedintheSouth-west,itispossiblethatothersourcesmaybyrepresentedinotherartifactsfromSnaketown.Culturesandculturaltiesareconstantlychanging,anditisexpectedthattradepat-temswillalsochangewithtimebecauseofpoliticalandeconomicpressures,orthedepletionofaresourceprocurementarea.Chemicalanalysisofacommonlyutilizedmaterialsuchasturquoiseisonemethodofdetecting,ormonitoring,thesechanges.Furtheridentificationofsourceareasuti-lizedbyparticularculturalgroupsshouldprovideadditionalinformationonthena-tureofprehistoricresourceacquisitionandexchangeroutes.ANNECOLBERGSIGLEODepartmentofGeosciences,UniversityofArizona,Tucson85721ReferencesandNotes1.ThechemicalformulaforturquoiseisCuAl6(P04)4(OH),*4H20.2.A.M.C.Sigleo,thesis,UniversityofNewMexico(1970).3.TheseareMineralPark,CanyonCreek,andCourtland,Arizona;Miami,Morenci,andBisbeehavenoreportedprehistoricworkings.4.E.W.Haury,Kiva31,1(1965).5.J.MurdochandR.W.Webb,Calif.Div.MinesGeol.Bull.No.189(1966).6.R.A.Schmitt,T.A.Linn,Jr.,W.Wakita,Radio-chim.Acta13,200(1970);G.E.Gordon,K.Randle,G.G.Goles,J.B.Corliss,M.H.Beeson,S.S.Oxley,Geochim.Cosmochim.Acta32,369(1968).7.F.J.Flanagan,Geochim.Cosmochim.Acda37,1189(1973).8.G.A.Borchardt,P.J.Aruscavage,H.T.Millard,Jr.,J.Sediment.Petrol.42,301(1972).9.Thesimilaritycoefficientndab=ZRi/nwhereR,=X,a/XibifXib>Xia;orXibdXiaifXia>Xib;Xia=thecontentofelementiinsamplea;Xib=contentofelementiinsampleb;andn=thenumberofelements.10.IthankG.W.NelsonandM.E.Wacksforhelpwiththeanalyticalaspectsofthisresearch;theArizonaStateMuseum,Tucson,fortheartifacts;GeodataSystems,Inc.,forpermissiontocollectturquoisesamplesontheirproperty;andG.A.Borchardtfortheuseofhissimilaritycoefficientprogram.7March1975;revised20May1975.ClimaticChange:AreWeontheBrinkofaPronouncedGlobalWarming?Abstract.Ifman-madedustisunimportantasamajorcauseofclimaticchange,thenastrongcasecanbemadethatthepresentcoolingtrendwill,withinadecadeorso,givewaytoapronouncedwarminginducedbycarbondioxide.Byanalogywithsimilareventsinthepast,thenaturalclimaticcoolingwhich,since1940,hasmorethancompensatedforthecarbondioxideeffect,willsoonbottomout.Oncethishappens,theexponentialriseintheatmosphericcarbondioxidecontentwilltendtobecomeasignificantfactorandbyearlyinthenextcenturywillhavedriventhemeanplanetarytemperaturebeyondthelimitsexperiencedduringthelast1000years.Thefactthatthemeanglobaltemper-aturehasbeenfallingoverthepastseveraldecadeshasledobserverstodiscountthewarmingeffectoftheCO2producedbytheburningofchemicalfuels.InthisreportIpresentanargumenttoshowthatthiscomplacencymaynotbewarranted.Itispossiblethatweareonthebrinkofasev-eral-decades-longperiodofrapidwarming.Briefly,theargumentrunsasfollows.The'IOrecordintheGreenlandicecore(1)stronglysuggeststhatthepresentcoolingisoneofalongseriesofsimilarnaturalcli-maticfluctuations.Thiscoolinghas,overthelastthreedecades,morethancom-pensatedforthewarmingeffectproducedbytheCO2releasedintotheatmosphereasaby-productofchemicalfuelcombustion.Byanalogywithsimilareventsinthepast,thepresentnaturalcoolingwill,however,SCIENCE,VOL.189 on October 20, 2010 www.sciencemag.orgDownloaded