Environmental Research Understanding international crime trends The legacy of preschool - PDF document

EnvironmentalResearch104(2007)315…336Understandinginternationalcrimetrends:ThelegacyofpreschoolleadexposureRickNevinNationalCenterforHealthyHousing,USAReceived12August2006;receivedinrevisedform20February2007;accepted23February2007Availableonline23April2007 E-mailaddress:ricknevin@verizon.net ofblood)cancauseseizuresanddeath,bloodleadoverg/dLisharmfultolearningandbehaviorandthereisnolowerbloodleadthresholdforIQlosses(USCentersforDiseaseControlandPrevention,1991Schwartz,1994Can“eldetal.,2003).Thehalf-lifeofleadinbloodis30days,butpreschoolbloodleadoftenchangesslowlyduetocontinuingexposure,andthatleadburdenaccumulatesinteethandbones(WorldHealthOrganization,1995Needlemanetal.(2003)foundyouthswithhighboneleadaretwiceaslikelytobedelinquent,aftercontrollingforconfounders.Otherstudiesalsolinkpreschoolleadexposuretoaggressiveanddelinquentadolescentbehaviorandlatercriminalviolence(Denno,1990Needlemanetal.,Dietrichetal.,2001StreteskyandLynch(2001)foundUSAcountieswithhigh1990airlead,mostlyfromindustrialemissions,had1989…1991murderratesfourtimeshigherthancountieswithlowairlead,aftercontrollingfornineairpollutantsandsixsociologicalfactors.Thisstudylikelyre”ects1970sadditivepreschoolleadexposure,becauseifmurderweremuchaffectedbycontemporaneousairleadthenthehomicideratewouldhavefallenasgasolineleadandairleadfellover70%from1975…1984(USEnvironmentalProtectionAgency,1986).Most1990lead-emittingfacilitieswereinoperationfordecades,inareaswitholderhousingandsometraf“c,so1989…1991murderrateslikelyre”ectedhigher1970sbloodleadwherechildrenhadadditiveexposuretoleadinpaintandgasolineandindustrialemissions.Nevin(2000)found1941…1975gasolineleaduseexplained90%ofthe1964…1998variationinUSAviolentcrime.Thebeststatistical-“tlagof23-yearsisconsistentwithneuraldamageininfancyandpeakagesofviolentoffending.Nevinshowedabest-“tlagof18yearsforgasolineleadversus1960…1998murders,and21yearsforpercapitapaintleaduseversus1900…1959murders.Thedifferenceinbest-“tmurderlagsisconsistentwithwhenpaintandgasleadmostaffectedpreschoolleadexposure.Gasleadsettledoverafewweeksormonths,andheavilyleadedcirca-1900leadpaintdeterioratedviachalkingafter3years(SchwartzandPitcher,1989vanAlphen,19981.1.LeadexposurepathwaysandpopulationbloodleadtrendsElevatedbloodleadcanbeduetoleadpaintchipingestion,inhaledairlead,andotherpathways,butpaintandgasolinehadespeciallypervasiveeffectsduetoleadcontaminateddustingestedvianormalhand-to-mouthactivityaschildrencrawl.Averagedailyleadingestedby2-year-oldsexposedtodustcontaminatedbyinteriorleadpaintissimilartotheaveragefor2-year-oldsexposedtodustcontaminatedbysettledcityairlead,andaverage2-year-oldleadingestionviadustismanytimesaverageingestionviainhaledairlead,dietarylead(fromairleadsettledoncropsand/orleadsolderinfoodandbeveragecans),orotherpathways(USEnvironmentalProtectionAgency,1986LeadusedinpaintaccountedforalmostathirdoftotalUSAleadoutputfrom1900…1914,whentheUSAproducedover40%ofworldleadoutput(Nevin,2000USGeologicalSurvey,2006).ThehighUSApercapitauseofleadinearly-1900spaintcausedmoresevereUSAleadpainthazardsthroughoutthe20thCentury.TheleadshareofUSApaintpigmentsfellfromnear100%in1900to35%inthemid-1930s(MeyerandMitchell,1943),buttheUSAdidnotbanresidentialleadpaintuntil1978.Pre-1940and1940…1959homeseachaccountedforaboutathirdofUSAhomesintheearly-1980s,andabout80%ofpre-1940and46%of1940…1959homesstillhadsomeinteriorleadpaintin1999(USCensus,1977…2003Jacobsetal.,2002).Sincethe1980sUSAphaseoutofleadingasoline,preschoolbloodleadprevalenceover10g/dLhastrackedUSAtrendsintheprevalenceofhousingwithdusthazardscausedbyinteriorleadpaint(JacobsandNevin,2006Trendsinpreschoolbloodleadprevalenceover10areespeciallyaffectedbywidespreadexposuretoleaddusthazards,butpaintchipingestionisoftenafactorinsevereleadpoisoning.A1989…1990studyfoundthatchildrenwithX-rayevidenceofrecentpaintchipingestionhadaveragebloodleadof63g/dL(McElvaineetal.,1992PercapitauseofleadingasolinesurgedintheUSAafterWorldWarII,androseataslowerrateinnationswithlowerpercapitagasolineconsumption.Leademissionsfromurbantraf“ccausedgreaterleadexposureforcitychildrenbecause10%ofleademissionssettledwithin100moftheroadand55%within20km,howeveratmosphericemissionsalsoaffectedbloodleadinareaswithlittletraf“c(OrganizationforEconomicCo-Opera-tionandDevelopment,1993).Nationaltrendsinaveragebloodleadandtheuseofleadingasolinewerehighlycorrelated,withmedianof0.94inGreece,Spain,SouthAfrica,Venezuela,Belgium,Sweden,Mexico,Finland,Canada,NewZealand,Italy,Switzerland,BritainandtheUSA(Thomasetal.,1999).Childrenexposedtoleadinpaintandgasolinehadagreaterriskofelevatedbloodleadbecauseleadingestionisadditive,butaveragebloodleadcloselytrackedgasolineleaduseduetoslowchangesinleadpaintexposureafterthe1930s.Leadexposurealsospannedawiderangeduetogasleadfalloutrelatedtocitysizeandroadproximity.USAcitieswithpopulationoveramillionhadearly-1960sambientairleadtwicethatincitiesof250,000toamillion,whichhadairlead40%higherthancitiesof100,000…250,000.Airleadbesideaheavilytraf“ckedCincinnatistreet(2150cars/horabout50,000cars/day)was15timesthecitysambientairleadUSPublicHealthService,1966,1965).Severeleadexposurewasanunrecognizedconsequenceoflocatingpublichousingbesidehighways.Forexample,ChicagoslongnarrowRobertTaylorHomesprojectthatopenedin1962wasallwithinabout400mof1963DanRyanexpresswaytraf“cof150,000vehicles/day(AmericanHighwayUsersAlliance,2004Manychildrenhadadditive1950…1970exposuretocityairleadandseverelydeterioratedleadpaintincirca-1900ARTICLEINPRESSR.Nevin/EnvironmentalResearch104(2007)315–336 slumhousing.Inthe1960smanyinnercityhospitalshadlargenumbersofcomatoseandconvulsingchildrenwithleadpoisoning,withfatalityratesof5…28%(Jackson,).Therewasextensiveslumdemolitionasurbanrenewalprojectsinexecutionroseseven-foldfrom1956…1966,butslumclearanceslowedinthelate-1960sUSDepartmentofHousingandUrbanDevelopment,).Publichousingcollocatedwithhighwaysonslumclearancelandalsocausedsevereairleadexposureaspercapitagasleadrose50%from1962to1970.Citybloodleadscreeningin1970showedabout25%ofyoungchildrentestedhadbloodleadover40g/dL.found95%of1970Censustractvariationinchildrenover40g/dLwasexplainedbythetractpopula-tionunderagesevenandprevalenceofdeterioratedordilapidatedhousing,butNewYorkCitychildrenoverg/dLrelativetosubstandardhousingprevalencewas20%higherthansmallercities,consistentwithhigherNewYorkCityairlead.The1976…1980NationalHealthandNutritionExam-inationSurvey(NHANES)revealedaverageUSApre-schoolbloodleadof15g/dL,andthe1988…1991NHANESshowedbloodleadfellsharplywiththeleadedgasphase-out(Pirkleetal.,1994).Percapitauseofleadingasolinepeakedlaterinmostnationsbutpercapitapaintleadusepeakedearlier,andleadpainthazardsraisedUSAelevatedbloodleadrisks.Australia1995averagepreschoolbloodleadwas50%abovethe1990USAaverage,but9%of1990USAchildrenversus7%of1995Australiachildrenwereover10g/dL(Pirkleetal.,1994AustralianInstituteofHealthandWelfare,1996).CanadianandUSAlate-1970saveragebloodleadweresimilarbut4%ofCanadianversus18%ofwhiteand52%ofblackUSAchildrenwereover20g/dL(RoyalSocietyofCanada,1986).In1960,blacksoccupied15%ofcentralcityhouseholdsand56%ofsubstandardcentralcityhousing,andthepercentofallcentralcityblacksinsubstandardhousingwas25%in1960and16%in1966(Kristof,1968Koebel,1996).Percapitagasleadfellfrom1956…1962buthitnewhighsfrom1966…1974,when62%ofblacksunderagesixlivedincentralcities,versus24%ofwhitechildren,withblacksconcentratedinolderhousing(USCensus,1960Averagebloodleadforblack2-year-oldsinChicagoandNewYorkCityfellbyabout30%from1970…1978,butthe1976…1980USAaverageforblackchildrenages6…36monthswasstill50%abovethewhiteaverage,andtheblackprevalenceover40g/dLwas800%higher(forToxicSubstancesandDiseaseRegistry,19881.2.Braingrowth,leadexposure,IQ,andbehaviorCriticalgrowthspurtsingrayandwhitematteroccurbeforeagetwo,whenelevatedmaternalandpreschoolbloodleadcausemanyneurologicaleffectsthatestablishabasisforimpairmentsinIQ,learning,andbehavior(etal.,1997LidskyandSchneider,2003Matsuzawaetal.,).Outcomesarealsoaffectedbyexposureseverity,duration,andtiming,andinteractionswithdietandsocioeconomicstatus(Bellinger,2004).BehaviorproblemscouldbeanindirecteffectofIQorthedirecteffectofbraindamageimpairingimpulsecontrol(Needlemanetal.,Gottfredson(1998)observesthatyouthswithIQof75…90areseventimesmorelikelytobeincarceratedthanthosewithIQof110…125,andstates:noothertraitorcircumstanceyetstudiedissodeeplyimplicatedinthenexusofbadsocialoutcomesaslowIQ.Thisperspective,however,doesnotaddresshowIQthatisstableafterchildhood(Neisseretal.,1996)mightrelatetoanage15…17propertycrimearrestratethataveragedninetimestheover-25ratefrom1970…2003.(BureauofJusticeStatistics,2004).Adifferentperspectiveisprovidedbymagneticresonanceimagingstudiesthatrevealasecondgraymattergrowthsurgejustbeforepuberty,predominat-inginthefrontallobe,theseatofexecutivefunctions„planning,impulsecontrolandreasoning(InstituteofMentalHealth,2001Sowelletal.(1999)reportsscansatages12…16and23…30showingalargefrontallobedifferenceinmyelin,whichprogressivelyinsulatesandthickenswhitematterconnectionsbetweenneuroncellbodies.Bartzokisetal.(2001)reportsfrontallobewhitemattergrowthtoage50,asgraymatterdeclines,andexplains:Whatkeepsgrowingisthemyelin[which]affectsthespeedofthesignalsthattravelfromneurontoneuron[and]allowsyourbraintoworkinconcert;yourenotaspronetoimpulse(Foster,2001Bartzokisetal.(2001)attributesimpulsiveteenbehaviortoincompletemyelination,andlinksmyelindisruptiontodevelopmentaldisorders.Developmentaleffectsofleadexposureincludethedestructionofmyelinsheathsanddecreasedactivityofanenzymeintegraltomyelinsynthesis(LidskyandSchneider,).Moregenerally,Silbergeld(1992)observedthatleadexposureduringcriticalperiodsofvulnerabilitycancausepermanentbraindamage,butneurotransmissioneffectscouldbereversibleabsentcontinuousexposure.GraymatterdamagecausingpermanentIQloss,andneuro-transmissiondamagethataffectsbehavior,couldcauseanIQ…crimecorrelationduetoseparateleadeffects.Age-relatedoffendingcouldbelinkedtoincompletemyelina-tionamongteenswithorwithoutpreschoolleadexposure,butcriminalbehaviorcouldbemorecommonandseverewithimpairedand/ordelayedmyelinationorotherneurotransmissiondamage.Whitemattergrowthtoage50suggeststhatlead-inducedneurotransmissiondisruptioncouldalsoaffectbehaviorwellbeyondadolescence,especiallyifmorecontinuousexposurecausesirreversibleInaSupremeCourtbriefopposingjuvenileexecutions,AmericanPsychologicalAssociation(2004)arguedthattheadolescentbrainhasnotreachedadultmaturity,particularlyinthefrontallobes,whichcontrolmaking.Thatbriefincludedagraphshowingviolentoffensesbuildsteeplyto18,beforestartingtodropoffasoffendingisoftenmoderatedoreliminatedbytheARTICLEINPRESSR.Nevin/EnvironmentalResearch104(2007)315–336 individualinadulthood.Thatsamegraphalsorevealsage-speci“carrestrateshiftsthattrackleadexposureandviolentcrimetrends[Fig.1].Youthsages16…22in1994wereallbornbeforetheearly-1980sfallingasolinelead,andtheage-16arrestratewas29%higherthantheage-22ratein1994,consistentwithcriminalbehaviorbeingmoderatedbychangesinfrontallobedevelopmentofadolescentsandyoungadults.The22-year-oldsin2001werealsobornbeforetheearly-1980sdeclineinleadexposure,butthe16-year-oldswereborninthemid-1980s,andthe2001age-16arrestratewas12%thantheage-22arrestrate.Thisstudyexaminesinternationaltrendsinpreschoolbloodlead,crimerates,andage-speci“carrestsrates,totestwhethertherelationshipbetweenleadexposure,arrest,andcrimetrendsinFig.1isevidentacrossmanycrimecategoriesandacrossnationswithdivergentpreschoolbloodleadandcrimeratetrends.International,racial,andcitydifferencesinsevereleadpoisoningprevalencearealsocomparedwithsubsequentcontrastsinmurderratesandjuvenileviolence,withtheexpectationthatseverepre-schoolleadpoisoningcouldbelinkedtomoreviolentoffendingandespeciallytomurderrates.2.Methods2.1.ArrestrateandincarcerationtrendsUSAindexcrimesincludepropertycrimes(burglaryandtheft)andtheviolentcrimesofmurder,rape,robbery,andaggravatedassault(causinginjuryorwithalethalweapon).BritishindictableoffensesincludeUSAindexcrimesplusthreats,simpleassaultswithoutinjury,andpettytheftsbelowaUSAmonetarythresholdforlarceny.ThefollowingUSAandBritishdataareexaminedtohighlightpeakoffendingagesandtemporalshiftsinage-speci“coffending:1.Britishage-speci“ccautionandconvictionrates(similartoUSAarrestrates)forindictableoffensesin1958versus1997(Taylor,1999CaliforniaDepartmentofJustice(2005)1960…2004juvenileandadultindexcrimearrestrates(age10…17arrestsper100,000ages10…17,andadultarrestsper100,000ages18…69).3.USA1970…2003age-speci“cpropertycrimearrestrates(BureauofJusticeStatistics,20044.USA1980…2001age-speci“carrestratesforpropertyandviolentcrimes,and1980…2003juvenilearrestsbyrace(Of“ceofJuvenileJusticeandDelinquencyPrevention,20045.USAincarcerationbyage,andrecidivismtrends(HarrisonandBeck,BeckandHarrison,2001LanganandLevin,2002BureauofJusticeStatistics,1993,20012.2.PreschoolbloodleadandinternationalcrimetrendsRegressionanalysescompareinternationalcrimeratetrendswithestimatedpreschoolbloodleadtrends.Fig.2showsUSA1936…1999preschoolbloodleadestimatesanchoredbyNHANESdata,andestimatesforallagesforcomparisonwithearlierbloodleaddata(Thomasetal.,Pirkleetal.,1994USCentersforDiseaseControlandPrevention,1997,2000OrganizationforEconomicCo-OperationandDevelopment,).Theseestimatesre”ect1980…1988airlead,1946…1976re“neryleaduse,and1936…1946roadgasolineuse(USEnvironmentalProtectionAgency,2001Nevin,2000;USCensus,1975).Early-1940sre“neryleadincludesaviationfuelusedoverseas,sobloodleadestimatesdeclinewithwartimegasrationing(Chevron,2000Fig.3showsbloodleadestimatesforBritain,France,WestGermany,Finland,Italy,Canada,Australia,andNewZealand.Estimatesareanchoredbynation-speci“cbloodleaddatabecausethe55%ofgasleadthatsettledwithin20kilometersofroadscausedahigherratioofbloodleadtopercapitaleademissionsindenselypopulatednations(ARTICLEINPRESS Fig.1.USAviolentCrimeandLeadExposureTrendsandAge-Speci“cViolentCrimeArrestRateShifts.Legend:Nevin(2000)foundthat1941…1975gasolineleaduseexplained90%ofthe1964…1998variationinUSAviolentcrime,wherethe23-yearlagisconsistentwithneurobeha-vioraleffectsofleadexposureininfancyandthetypicalageofviolentoffenders.Age-speci“cviolentcrimearrestrates(Of“ceofJuvenileJusticeandDelinquencyPrevention,2004)alsorevealthatpeakoffendingshiftedtoolderagesby2001,asthe1990sviolentcrimedeclinewasassociatedwithanespeciallysharpdeclineamongjuvenilesbornaftertheearly-1980sdeclineingaslead. Fig.2.USABloodLeadDataandTrendEstimates.Legend:Preschoolbloodleadestimatesre”ect1980…1988airlead,1946…1976re“neryleaduse(seeFig.1),andpre-1946roadgasolineuse,anchoredbyNHANESdata(whichshowthat1976…1980preschoolbloodleadwashigherthanbloodleadforallages).Bloodleadestimatesforallagesarealsoshownforcomparisonwithotheravailable1956…1976surveydataonnational,largeurbanarea,andsmallcityandruralbloodleadlevels.R.Nevin/EnvironmentalResearch104(2007)315–336 etal.,1999RoyalSocietyofCanada,1986OrganizationforEconomicCo-OperationandDevelopment,1993).Forsomenations,bloodleaddataforadults,olderchildren,and/orurbanpopulationswereadjustedbasedonNHANESratiosofpreschoolbloodleadtobloodleadforotherpopulationgroups(Pirkleetal.,1994).GasleademissiontrendswereavailableforCanada(RoyalSocietyofCanada,1986)andBritishtrendsre”ectpost-1969leademissionsandpre-1970petroluse(UKDepartmentofEnvironment,2004UKInstituteofPetroleum,1938…1970).Emissiontrendsforothernationsre”ectleadedgasuseandaverageleadperliterOrganizationforEconomicCo-OperationandDevelopment,1993InternationalEnergyAdministration,1960…1990OctelLtd,1969…1990UnitedNations,1950…1960).Datasupportestimatesfrom1936intheUSA,Canada,andBritain,andfrom1950intheothersixnations.Estimatesaresubjecttobloodleadsurveyrandomerrorandsomeestimationerror,butpotentialerrorissmallrelativetothelargetemporalchangesandinternationaldivergencesinpreschoolbloodlead[Figs.2and3Recordedcrimesper100,000residentsareexaminedforindexcrime,burglary,robbery,assault,rape,andmurder(BarclayandTavares,2003BureauofJusticeStatistics,2006Triggs,1997UKHomeOf“ce,2004StatisticsCanada,2005Mukherjee,etal.,1997AustralianInstituteofCriminology,1999…2006CouncilofEurope,2000,2003NewZealandPolice,2001,2005Interpol,2005).IndexcrimeforWestGermany,France,andItalyappeartobecomparabletoUSArates,whileBritain,Canada,Australia,Finland,andNewZealandincludeabroaderrangeofcrimes,resultinginhigherindexrates.USA,Canada,Australia,NewZealand,WestGermany,andpost-1968Britainburglaryratesarecomparable,asareUSA,Australia,Britain,Canada,NewZealand,andWestGermanyrobberyrates.Britainonlyrecordedpre-1969burglariesoveramonetarythreshold,andlowratesinFrance(notwell-documented)mayalsore”ectathreshold.BritishwoundingiscomparabletoUSAaggravatedassault(LanganandFarrington,1998)andFrenchassaultdataappeartobecomparable.TheUSA,Britain,France,andFinlandreportrapeoffemale,andAustralia,NewZealand,andCanadareportviolentandsexualassaults.Singleandcombinednationregressionswererunwith5…45yearlagstoidentifybest-“tlagsforeachcrime,withthehighestsigni“cance-value)forbloodleadandpercentofcrimeratevariationexplained(Regressionswerealsorunwithunemploymentratesfornationswithcomparabledata(BureauofLaborStatistics,2004).CombinednationARTICLEINPRESS Fig.3.InternationalPreschoolBloodLeadTrendEstimates.Legend:Internationalpreschoolbloodleadestimatesre”ectleademissiontrends,ancbynation-speci“cbloodleaddatabecausethe55%ofgasleadthatsettledwithin20kilometersofroadscausedahigherratioofbloodleadtopercapitaemissionsindenselypopulatednations.R.Nevin/EnvironmentalResearch104(2007)315–336 resultsarereportedforbest-“tlagswithandwithoutcountrydummies.Crimerecording,cultural,andothercountrydifferenceswilllowerthecombinednationindexcrimeregressionwithoutdummies,andtoalesserextentinothercombinednationregressionswithoutdummies.Toillustratebest-“tlags,Fig.6acrosssingle-nationregressiontimelagsfordifferentcrimecategories,andFigs.7…11showcrimeversusbloodleadtrendgraphsforbest-“tlags.Regressionresultsre”ectdatathrough2002,butsomegraphsincludedatathrough2004.Otherpotentialconfounderswereexcludedbecausepreliminaryanalysisshowednoimpactonlong-termcrimetrends.ThepercentofUSAviolentcrimeinvolvinggunswasfairlystablefrom1973…2004(BureauofJusticeStatistics,2005),asviolentcrimeroseandfellsharply.Only6%ofmurdersin1991and4%in2001werelinkedtodrugoffensesorbrawlsin”uencedbynarcotics,andtherewaslittle1990schangeinthepercentofprisonerswhocommittedcrimestogetdrugmoney(Dorseyetal.,2005asmurderandothercrimesfellsharply.Internationalcrimetrendsareinconsistentwiththeoreticaleffectsofpolicepercapita,incarceration,anddemographictrends(BarclayandTavares,2003;USCensus,20042.3.Cross-sectionalregressionanalysisof1985–1994USAcentralcitymurderratesAseparateanalysiscomparesaverage1985…1994murderratesacrossUSAcitieswithdifferencesincirca-1970leadpaintpoisoningandairleadexposure.Childrenundersevenin1970wereinthehighmurderoffenseagebracketin1985…1994.LP%valueswereconstructedtoestimatethepercentofeachcitys1985…1994populationthathadseverechildhoodleadpaintpoisoningin1970.Citysizedummyvariableswereusedasindicatorsof1970airlead.Gilsinn(1972)usedpopulationundersevenanddeterioratedanddilapidatedhousingprevalencetoestimatethenumberofchildrenwithbloodleadover40g/dLineach1970metroarea.Theseestimatesapproximatethenumberofcentralcitychildrenover40g/dLbecausetherewaslittledeterioratedhousingin1970suburbs.Gilsinnsestimatesweredividedbyaverage1985…1994populationforthecorrespondingcitiestocalculateLP%.RegressionanalysiscomparesLP%withtheaverage1985…1994murderratesin124centralcity/cities,including11combinedcitymurderratescalculatedformetroareaswithmorethanonecentralcity.Thisregressiondoesnotre”ectairleadvariationsbecauseGilsinnsestimateswerebasedonhousingdata.Asecondregressioncomparescitymurderrateswithcitysizedummyvariablesasindicatorsof1970airlead,usingaverage1985…1994populationandmurderratesforthesame124city/cities.AthirdregressionwithLP%andcitysizedummiesexaminestheadditiveeffectofsevereleadpainthazardsandairlead.Theanalysisalsotestsfortheeffectoftheblackpercentofcitypopulation.Limitationsofthisanalysisincludecitiesinthesmallsizecategorywithairleadaffectedbylargecitiesinanothermetroarea(e.g.,NewarkandNewYork),childrenthatmovedbetweencitiesfrom1970to1985…94,and1970cityleadpaintpoisoningthatisoverstatedtotheextentthatsuburbsdidhavesomedeterioratedhousing(re”ectedinGilsinnsestimates).3.Results3.1.ArrestrateandincarcerationtrendsAge-14Britishmaleshadthehighestcautionandconvictionrateforindictableoffensesin1958,butpeakoffendingshiftedtoage18by1997.Theage-10offenseratefell70%from1958…1997,asage18…29offendingratesincreasedthreeto“ve-fold.Malesages12…14in1958,bornasgasleadexposureroseafterWorldWarII,hadhigheroffendingratesthanolderteensbornbeforethatriseinleadexposure.By1997,offendingdeclinedrelativeto1958onlyformalesunder14,bornafterthemid-1980sfallinBritishgasleaduse,whileoffendingratesroseforolderteensandadultsbornoveryearsofrisinggasolineleaduse.USApercapitagasolineleadincreased400%from1945…55,andFig.4showstheCaliforniajuvenileindexcrimearrestratesurgedalmost300%from1965to1975.Theadultarrestrateroseatamuchslowerrate,whenmostadultswerebornbeforethe1950ssurgeingasolineleaduse.Thosetrendsreversedinthe1990swhenarrestratesfellfasterforjuveniles,bornafterairleadpeakedintheearly-1970s.In1975,Californiasjuvenileindexcrimearrestratewastwicetheadultrate,but2000…2004indexcrimearrestrateswerehigherforadults.(The1965…1975arrestratetrendspartlyre”ectin”ationincreasingthepercentoftheftsexceedingamonetarythreshold,revisedovertime,butthiswouldnotaffectjuvenilearrestratesrelativetoadults.)Propertycrimeaccountsforover90%ofindexcrimes.TheoverallUSApropertycrimeratewasaboutthesamein1970and2003,butthepropertycrimearrestrateforyouthsunderage15fell45%from1970…2003,theage15…17ratefell27%,theage18…24raterose8%,andthearrestrateforadultsover-24rose58%.The45%dropintheunder-15arrestratecomparesoffendersin1970bornnearthe1956interimpeakingasolineleadversusoffendersin2003bornaftertheearly-1980sfallingasolinelead.The58%increaseintheover-24arrestratecomparesadultsin1970mostlybornbefore1950versustheir2003counter-partsbornbefore1980.Fig.5compares1980…2001age-speci“cUSAarrestratesand1980…2003juvenilearrestratesbyrace.The1980…2001USApropertycrimedeclinewasledbya70%fallintheblackjuvenileburglaryarrestrate,whichfellmuchfasterthanthewhitejuvenilearrestratefrom1980…1988,narrowingtheracialdifference.Juvenileburglaryrateswerelittlechangedfrom1988…1994,butfellfurtherafter1994.The2003blackjuvenileburglaryarrestratewas43%belowthe1980whitejuvenilerate.PeakoffendingforARTICLEINPRESS Fig.4.CaliforniaAdultandJuvenileIndexCrimeArrestRates.Legend:CaliforniaDepartmentofJustice(2005)datashowthejuvenileindexcrimearrestraterosemuchfasterthantheadultarrestratefrom1965…1975,whenmostadultswerebornbeforethe1950sriseingasolineleaduse.Thosetrendsreversedinthe1990swhenarrestratesfellfasterforjuveniles,bornafterairleadpeakedintheearly-1970s.(Arresttrendsfrom1965…1975partlyre”ectin”ationincreasingthepercentoftheftsexceedingamonetarythreshold,revisedovertime,butthiswouldnotaffectjuvenilearrestratesrelativetoadults.)R.Nevin/EnvironmentalResearch104(2007)315–336 robberyisafewyearsolderthanforpropetycrime,andthe42%fallintherobberyratefrom1980…2001wasentirelyduetosharplylowerarrestratesforjuvenilesandyoungadults,astheage35…44arrestraterose.Theblackjuvenilerobberyarrestratefellfrom1980…1988,narrowingtheracialdifference,buttheblackrateandracialdifferencerosefrom1988…1994beforefallingtonewlowsin2001…2003.Aggravatedassaultoffendingpeaksatanolderagethanrobberyandfallsmoreslowlywithage.Aggravatedassaultarrestsroseforallagesfrom1980…1994,buttheage40…44arrestratecontinuedtorisethrough2001.Blackjuvenilesrecordedthelargestrisefrom1985…1994,andthesharpestfallfrom1994…2001.Theunder-21homicidearrestratesoaredfrom1984…1994astheover-25ratedeclined,butthe1990shomicideratedeclinewasmainlyduetoasharpfallintheunder-21rate.Theblackjuvenilemurderarrestratedriftedlowerintheearly-1980sthenrosesharplybeforefallingtomulti-decadelows.Theracialdifferenceinjuvenilemurderarrestratespeakedin1994,butthe2003differencewasonlyaboutone-fourththeaverageracialdifferencefrom1980…1998.USAincarcerationtrendsechoarresttrends,asoffen-dersoverage34accountedforjust27%ofprisoncommitmentsin1993butaccountedfor40%in2001.TheoverallUSAincarcerationratechangedlittlefrom2000…2004,buttheage18…19maleincarcerationratefell30%andtheage20…34ratefell7%,asthemaleincarcerationraterose5%forages35…39,21%forages40…44,26%forages45…54,and41%forthoseover55.Over60%ofprisonersreleasedinboth1983and1994wererearrestedwithin3years,but35%ofthosereleasedin1983wereages18…25versus21%in1994.Combiningprisonerreleasetrendswithrecidivistoffendingratessuggeststhatprisonersreleasedintheprior3yearscommittedjust6%ofARTICLEINPRESS Fig.5.USAAge-Speci“cArrestRatesandJuvenileArrestRatesbyRace.Legend:TheUSAcrimedeclinewasledbyasharpdeclineinoffendingbyjuveniles,andespeciallyblackjuveniles,asarrestrateschangedlittleforthoseoverage35.The1980sracialconvergenceinjuvenileburglaryarrestratescouldre”ecta1960sracialconvergenceinpreschoolbloodleadasslumdemolitionreducedleadpainthazards.Juvenileviolencealsofellfrom1980…butblackjuvenileviolencesurgedinthelate-1980sasblackjuvenileburglaryarrestschangedlittle.Thiscouldre”ectawiderearly-1970sblackpbloodleaddistributionwithmoresevereleadexposureespeciallyaffectingviolence(e.g.,amongyouthsraisedinpublichousingnearcirca-1970hR.Nevin/EnvironmentalResearch104(2007)315–336 propertycrimesand11%ofviolentcrimesin1979versus28%ofpropertycrimesand35%ofviolentcrimesin2002.3.2.PreschoolbloodleadandinternationalcrimetrendregressionsThebest-“ttimelagforindexcrimeversuspreschoolbloodleadis19yearsinaregressionwithcountrydummiescomparing309yearsofdataacrossninenations.Thesamebest-“ttimelagisevidentinsingle-nationregressionswithandwithoutanunemploymentvariable.Table1regressionresultswitha19-yearlag,Fig.6graphsacrosslagsforeachnation,andFigs.7and8preschoolbloodleadtrendsversusindexcrimerateswitha19-yearlag.Bloodleadishighlysigni“cantincombinedandsingle-nationregressionswithandwithoutcountrydummies.Unemploymentissigni“cantinmostnationsbutitsinclusioninthemodelhasnosubstantiveeffectonthebloodleadcoef“cientvalueorsigni“cance(-value),andlittleimpactoncrimeratevariationexplained().Addingunemploymentraisesfrom:80%to81%fortheUSA;87%to90%forCanada;72%to84%forFrance;andARTICLEINPRESS Table1Regressionsforpreschoolbloodleadvs.indexcrimewitha19-yearlagDependentvariableIndependentvariablesCoef“cientStandarderrorIndexCrime„9NationsCombinedIntercept3675.09311.8211.79.00010.165309PreschoolBloodLead176.2622.677.78IndexCrime„9NationsCombinedIntercept2791.49236.9411.78.00010.774309WithCountyDummies(notshown)PreschoolBloodLead287.7114.3720.03AustraliaIntercept773.36910.820.850.40540.62823PreschoolBloodLead287.5848.315.95AustraliawithUnemploymentRateIntercept1690.66857.681.970.0627PreschoolBloodLead340.7846.157.39.00010.73223UnemploymentRate243.1487.222.790.0114BritainIntercept1665.60268.99.00010.95047PreschoolBloodLead740.5525.2229.37BritainwithUnemploymentRateIntercept1384.25243.47PreschoolBloodLead675.4131.4221.50.00010.96344UnemploymentRate75.9241.591.830.0752FinlandIntercept2312.83251.419.20.00010.88033PreschoolBloodLead589.4439.1815.05NewZealandIntercept2123.73580.893.660.00090.93633PreschoolBloodLead1053.8949.4221.32CanadaIntercept1624.66363.204.47.00010.86841PreschoolBloodLead440.0027.5215.99CanadawithUnemploymentRateIntercept863.50397.252.170.036PreschoolBloodLead336.4039.758.46.00010.89741UnemploymentRate280.7684.733.310.002USAIntercept1629.13241.546.74.00010.79843PreschoolBloodLead193.3015.2112.71USAwithUnemploymentRateIntercept1215.54364.513.330.0018PreschoolBloodLead181.8116.8410.80.00010.80843UnemploymentRate98.6165.831.500.142W.GermanyIntercept4012.72209.2319.18.00010.83422PreschoolBloodLead185.2518.4910.02W.GermanywithUnemploymentRateIntercept3750.86156.9823.89.00010.92322PreschoolBloodLead87.9124.573.580.002UnemploymentRate307.2565.894.660.0002ItalyIntercept830.33111.887.42.00010.92833PreschoolBloodLead148.977.4619.96ItalywithUnemploymentRateIntercept737.51153.034.82.00010.93033PreschoolBloodLead135.5816.778.09UnemploymentRate38.5643.210.890.3792FranceIntercept2386.69343.826.94.00010.72234PreschoolBloodLead175.7319.279.12FrancewithUnemploymentRateIntercept1889.03289.496.53PreschoolBloodLead28.1835.290.800.43060.83634UnemploymentRate367.6679.514.62:Regressionscombiningdataforninenationswererunforpreschoolbloodleadversusindexcrimewithtimelagsof5…45years,andthebest-“tlag-valueforbloodlead)was19years.Recordingdifferenceslowerinthecombinedregressionwithoutcountrydummiesbutbloodleadisstillhighlysigni“cant.Bloodleadisalsosigni“cantandishighineachsingle-nationregressionwitha19-yearlag.Unemploymentissigni“cantbuthaslittleincrementaleffectonregressionR.Nevin/EnvironmentalResearch104(2007)315–336 83%to92%forWestGermany.ItalyandBritainwithjustbloodleadis93%to95%andunemploymentisinsigni“cant.Graphsacrosstimelagsshow(andblood-value)peaksat18…21yearsinsixnations,at14…15yearsinWestGermany(22)andFrance(33)and26yearsinAustralia(Thebest-“tlagforburglaryis18yearsinacombinedregressionforeightnations(229)withcountrydummies,andfor“venations(169)withunemploy-mentdata(excludingburglarydataforFrancesuggestingamonetarythreshold).Table2showsbloodleadishighlysigni“cantincombinedandsingle-nationburglaryregres-Fig.6acrosstimelagsforeachnation,andFig.9graphsburglaryversusbloodleadwithan18-yearlag.Unemploymentissigni“cantbutitsinclusiononlyincreasesfrom65%to73%fortheUSA;78%to86%forCanada;85%to88%forBritain;and82%to92%forWestGermany.Australiais91%withjustbloodlead(unemploymentisinsigni“cant)andNewZealand86%withjustbloodlead.peaksatlagsof16…19yearsinsevennations,and21yearsinAustralia,basedondatathrough2002.Australiasburglaryratefellabout20%from2002…2004.Thebest-“tlagforrobberyacrosssevennations220)is23years,andunemploymentisinsigni“cantacrosssixnations(Table3showsbloodleadwitha23-yearlagishighlysigni“cantincombinedandsingle-nationregressions.Fig.10graphsrobberyversusbloodleadwitha23-yearlag.AddingunemploymentraisessomewhatfortheUSAandCanadabutunemploymentisinsigni“cantorhasanunexpectedsigninothernations.Thebest“tis20…21yearsinfournations,36yearsinFrance(14)and26…28inBritainandAustraliathrough2002.BritainandAustraliarobberyratesfellabout20%from2002…2004.Table4Fig.11showbloodleadwitha23-yearlagishighlysigni“cantinregressionsforaggravatedassault100)andviolentandsexualassault(67),andwitha24-yearlagforrape(113).Unemploymentisinsignif-icant.Thebest-“tforaggravatedassaultis24yearsintheUSA(43),23inBritain(43),and29…33yearsinFrance(14).Thebest“tlagforviolentandsexualassaultis24yearsinCanada(41),22inNewZealand15),and28…33inAustralia(11).Thebest-“tforrapeis23yearsintheUSA(43),30inBritain(43),29inFrance(14),and27…33inFinland(13).Fig.6showsregressionforaggravatedassault,violentandsexualassault,andrapereachabsolutepeaksacrossarangeoflongerlagsinsingle-nationregressions.However,Table4-valuesareveryhighwiththe23…24yeartimelagforallsingle-nationregressionswithover14yearsofdataof79…94%).Table5showsbloodleadwithan18-yearARTICLEINPRESS Fig.6.acrossSingle-NationRegressionTimeLags.Legend:Single-nationregressionswererunwith5…45yearlagsforeachcrimecategoryversuspreschoolbloodlead,foreverynationwithavailabledata.Despitedivergentinternationalcrimeandbloodleadtrends,regression(andbloodlead-value)isnearitspeakineachnationattimelagsconsistentwithpeakoffendingagesforeachcrimecategory.R.Nevin/EnvironmentalResearch104(2007)315–336 lagissigni“cantinthecombinednationmurderregression209)andunemploymentisinsigni“cant(178).Thebest-“ttimelagis18…19yearsfortheUSA,NewZealand,andBritain,butCanadahasashorterbest-“tandaveragepreschoolbloodleadisnotsigni“cantinmurderregressionsforAustraliaorWestGermany. Fig.7.PreschoolBloodLeadvs.NarrowlyDe“nedIndexCrimewitha19-YearLag.Legend:USAindexcrimeincludespropertycrimes(theftandburglary)andtheviolentcrimesofmurder,rape,robbery,andaggravatedassault(causinginjuryorwithalethalweapon).Nationswithcomparablecindexesallshowindexcrimeratestrackingpreschoolbloodleadtrendswitha19yearlag,despitedivergentcrimetrends.TheUSAindexcrimeratewas22%higherthantheFrenchrateand40%higherthanAustraliasratein1980,buttheUSAratewas39%belowtheFrenchrateand45%belowAustraliasratein2001. Fig.8.PreschoolBloodLeadvs.BroadlyDe“nedIndexCrimewitha19-YearLag.Legend:Broadlyde“nedindexcrimeratesincludeUSAindexcrimesplusthreats,assaultswithoutinjury,andtheftsbelowaUSAmonetarythreshold.Despiterecordingdifferencesanddivergentcrimetrends,broadlde“nedindexcrimeratesalsotrackbloodleadtrendswitha19-yearlag.R.Nevin/EnvironmentalResearch104(2007)315–336 3.3.Cross-sectionalregressionanalysisof1985–1994USAcentralcitymurderratesTable6showstheregressionanalysesof1985…1994averagemurderratesacross124centralcity/cities.Theaverage1985…1994murderratewas33(per100,000)incentralcity/citieswithpopulationoveramillion,21incitiesof250,000toonemillion,and15incitiesof100…250thousand,andcitysizedummyvariablesaresigni“cantinasimpleregression,withof11.4%.LP%isalsohighlysigni“cantinaseparateregression,withof14%.WhenLP%andcitysizearebothincluded,regressionexceedsthevariationexplainedbyseparateregressions(11.4%+13.9%25.3%)andLP%andcitysizevariablesaremoresigni“cant(higher-values).Whenavariableisaddedforblackpercentofpopulation,citysizeisstillsigni“cantandLP%isnot(1.27),buttheLP%coef“cientstillhastheexpectedsignandretainingLP%inthemodelincreasesto69%versus61%withjustthecitysizeandblackpercentvariables.ARTICLEINPRESS Table2Regressionsforpreschoolbloodleadvs.burglarywithan18-yearlagDependentvariableIndependentvariableCoef“cientStandarderrorBurglary„8NationsCombinedIntercept1072.11106.5310.06.00010.060229(UK,US,CAN,AUS,NZ,WG,FR,FIN)PreschoolBloodLead27.637.233.820.0002Burglary„8NationsCombinedIntercept626.5073.098.57.00010.776229WithCountyDummies(notshown)PreschoolBloodLead83.334.5618.26Burglary„5NationsCombinedIntercept586.49107.045.48.00010.299169(UK,US,CAN,AUS,WG)PreschoolBloodLead61.767.328.44Burglary„5NationsCombinedIntercept714.8864.3211.11.00010.819169WithCountyDummies(notshown)PreschoolBloodLead75.814.0818.59Burglary„5NationsCombinedIntercept397.6067.855.86.00010.869169WithCountyDummies(notshown)andUnemploymentRatePreschoolBloodLead53.914.4412.15UnemploymentRate78.919.967.92USABurglaryIntercept397.6784.114.73.00010.65343PreschoolBloodLead46.525.298.79USABurglarywithUnemploymentRateIntercept112.06115.390.970.3373PreschoolBloodLead38.855.307.32.00010.72743UnemploymentRate67.4720.593.280.0022BritainBurglaryIntercept51.84129.150.400.69080.84934PreschoolBloodLead141.0410.5213.41BritainBurglarywithUnemploymentRateIntercept147.83119.741.230.22630.88334PreschoolBloodLead121.7411.3810.70UnemploymentRate44.3314.723.010.0051CanadaBurglaryIntercept245.4075.373.260.00230.78141PreschoolBloodLead66.795.6611.79CanadaBurglarywithUnemploymentRateIntercept69.7672.040.970.3390.86041PreschoolBloodLead38.857.615.11UnemploymentRate72.3015.684.61AustraliaBurglaryIntercept544.87133.164.090.00030.91231PreschoolBloodLead132.387.6517.31AustraliaBurglarywithUnemploymentRateIntercept538.67133.184.040.0004PreschoolBloodLead121.9712.709.60.00010.91531UnemploymentRate24.4323.821.030.314W.GermanyBurglaryIntercept1403.1795.8614.64.00010.81920PreschoolBloodLead65.427.269.01W.GermanyBurglarywithUnemploymentRateIntercept1274.6269.7918.26.00010.92320PreschoolBloodLead31.588.593.670.0019UnemploymentRate120.4525.184.780.0002FranceBurglaryIntercept261.69105.512.480.02890.61414PreschoolBloodLead19.314.424.360.0009FranceBurglarywithUnemploymentRateIntercept272.37114.212.380.0362PreschoolBloodLead20.305.473.710.00340.61714UnemploymentRate3.339.960.330.7447FinlandBurglaryIntercept1065.13378.402.810.02020.28611PreschoolBloodLead78.6241.431.900.0902NewZealandBurglaryIntercept1153.40205.29.00010.86235PreschoolBloodLead254.3717.7314.35:Combined-nationregressionsforpreschoolbloodleadversusburglarywithtimelagsof5…45yearsshowedabest-“tlagof18years.Bloodleadisalsosigni“cantandishighwithan18-yearlaginallsingle-nationregressionswith11.Unemploymentisstatisticallysigni“cantbuthaslittleincrementaleffectonregressionR.Nevin/EnvironmentalResearch104(2007)315–336 4.Discussion4.1.ArrestrateandincarcerationtrendsThe1980sracialconvergenceinjuvenileburglaryratescouldre”ecta1960sracialconvergenceinpreschoolbloodleadduetoslumdemolitionanda1956…1962fallinpercapitagasleaduse,evenasurbansprawlspreadmoregasleademissionstopredominantlywhitesuburbs.A1960sbloodleadconvergenceisalsoconsistentwitharacialconvergenceinNationalAssessmentofEducationalProgress(NAEP)scoresreportedforthesamebirthcohort(Neisseretal.,1996).A2003blackjuvenileburglaryarrestratethatwaswellbelowthe1980whitejuvenilearrestrateisalsoconsistentwithlate-1980saverageblackpreschoolbloodleadwellbelowthe1970saverageforwhitechildren(Pirkleetal.,1994Juvenileviolencealsofellfrom1980…1984,butblackjuvenileviolencesurgedinthelate-1980sasblackNAEPscoresandjuvenileburglaryarrestschangedlittle.Thesetrendscouldre”ectawiderearly-1970sblackpreschoolbloodleaddistributionwithmoresevereexposureespe-ciallyaffectingviolence.Averageblackleadexposuremighthavechangedlittlefromthemid-1960stotheearly-1970sasdecliningleadpainthazardsoffsettheriseinambientairlead,butseverepoisoningprevalencelikelyroseamongblackchildrenlivingnearurbanhigh-ways.Astrongerassociationbetweensevereleadpoisoningandviolenceisalsoconsistentwithracialdifferencesinlate-1970sbloodleadandearly-1990sjuvenilearrestrates.Average1976…1980bloodleadforblackchildrenages6…36monthswas50%abovetheaverageforwhitechildren,butblacksweresixtimesmorelikelytohavebloodleadof30…39g/dLandeighttimesmorelikelytobeover40g/dL.Thosechildrenwerejuvenileswhenthe1990…1994blackjuvenileburglaryarrestratewas60%higherthanthewhiterate,buttheblackjuvenileviolentcrimearrestratewas“vetimeshigherandtheblackjuvenilemurderratewaseighttimeshigher.ARTICLEINPRESS Fig.9.PreschoolBloodLeadvs.Burglarywithan18-YearLag.Legend:Thebest-“tlagforpreschoolbloodleadversusburglaryis18yearsinacombined-nationregression.The1974USAburglaryratewas50%and98%higherthanratesinBritainandAustralia,respectively,andthe2002USAratewas56%and63%lowerthanratesinBritainandAustralia,butpreschoolbloodleadtrendspresagedburglaryratetrendswithan18-yearlagineachR.Nevin/EnvironmentalResearch104(2007)315–336 Mof“tt(1993)distinguishesbetweenrelativelycommonAdolescence-Limited(AL)offendersandmoreviolentLife-Course-Persistent(LCP)offenderswhoaccountformostadultoffending.ShiftsinjuvenileindexcrimeandpropertycrimearrestratessuggestthatpreschoolbloodleadhasamajorimpactonALoffending.However,the2003USAage15…17propertycrimearrestratewasstillseventimestherateforadultsoverage24,showingALoffendingismorecommonacrossbirthcohortswithverydifferentpreschoolbloodlead.RisingarrestandincarcerationratesforolderadultssuggestthatLCPoffendingcouldalsoberelatedtopreschoolbloodlead.Braingrowthalsopresentsintriguingparallelswithlifetimeoffendinginonesampleofjuvenilecriminals:Offenseratesrosesharplyafterage10;propertycrimespeakedinadolescenceandfellalmost90%bytheearly-20s;andviolentoffendingpeakedintheearly-20sandfellafterage30withasharpdeclinebyage50evenamonghigh-ratechronicviolentoffenders(SampsonandLaub,2003Thesepatternsparallelbraindevelopmentfromthesurgeinoffendingandgraymattergrowthnearpuberty,throughgraymatterandoffendingpeaksaroundage20,tothepeakinwhitematterandasharpreductioninoffendingbyage50.ARTICLEINPRESS Table3Regressionsforpreschoolbloodleadvs.robberywitha23-yearlagDependentvariableIndependentvariableCoef“cientStandarderrorRobbery„7NationsCombinedIntercept11.968.061.480.1391(UK,US,CAN,AUS,NZ,WG,FR)PreschoolBloodLead7.880.5813.64.00010.461220Robbery„7NationsCombinedIntercept9.415.591.680.09370.848220WithCountyDummies(notshown)PreschoolBloodLead7.420.3918.89Robbery„6NationsCombinedIntercept1.568.420.190.8530.472190(UK,US,CAN,AUS,WG,FR)PreschoolBloodLead7.640.5912.97Robbery„6NationsCombinedIntercept9.946.051.640.10210.830190WithCountyDummies(notshown)PreschoolBloodLead7.470.4317.41Robbery„6NationsCombinedIntercept9.547.771.230.22110.830190WithCountyDummies(notshown)andUnemploymentRatePreschoolBloodLead7.490.5015.10UnemploymentRate0.091.140.080.9349USARobberyIntercept56.4513.234.270.00010.71543PreschoolBloodLead8.550.8410.14USARobberywithUnemploymentRateIntercept6.9920.430.340.7341PreschoolBloodLead7.630.839.20.00010.76743UnemploymentRate10.593.533.000.0046BritainRobberyIntercept65.959.30.00010.84343PreschoolBloodLead13.400.9014.83BritainRobberywithUnemploymentRateIntercept59.049.70PreschoolBloodLead14.371.0114.20.00010.85643UnemploymentRate2.551.331.910.0627CanadaRobberyIntercept26.205.814.51.00010.71241PreschoolBloodLead4.500.469.82CanadaRobberywithUnemploymentRateIntercept3.666.730.540.58990.81741PreschoolBloodLead2.840.515.55UnemploymentRate5.741.234.68AustraliaRobberyIntercept56.198.05.00010.89030PreschoolBloodLead7.470.5015.03AustraliaRobberywithUnemploymentRateIntercept41.118.20PreschoolBloodLead8.490.5216.27.00010.92230UnemploymentRate4.341.293.370.0023W.GermanyRobberyIntercept30.361.7717.19.00010.80219PreschoolBloodLead1.600.198.29W.GermanyRobberywithUnemploymentRateIntercept27.802.3811.69.00010.82819PreschoolBloodLead1.330.265.21UnemploymentRate1.020.661.540.1421FranceRobberyIntercept23.6282.890.280.78060.17614PreschoolBloodLead5.693.551.600.1351FranceRobberywithUnemploymentRateIntercept225.8969.683.240.0078PreschoolBloodLead8.402.323.620.0040.70014UnemploymentRate25.775.884.390.0011NewZealandRobberyIntercept36.364.12.00010.90830PreschoolBloodLead5.970.3616.66:Thebest-“tlagforpreschoolbloodleadversusrobberyis23yearsinregressionscombiningdataforsevennations.Unemploymentisinsigni“cant.Bloodleadishighlysigni“cantandishighwitha23-yearlaginallsingle-nationregressionswithR.Nevin/EnvironmentalResearch104(2007)315–336 4.2.PreschoolbloodleadandinternationalcrimetrendsItisstrikingthatpreschoolbloodleadishighlysigni“cantatbest-“tlagsconsistentwithpeakoffendingagesforeachcrimecategory.Burglaryandotherpropertycrimearrestspeakatages15…20,andthebest-“tforburglaryis18yearsincombinednationregressionsand16…19yearsinseparateregressionsfortheUSA,Canada,Britain,France,Finland,WestGermany,andNewZealand.Aggravatedassaultpeaksfromage18tothelate-20s,andthebest-“tis22…24yearsforaggravatedassaultintheUSAandBritainandforviolentandsexualassaultinCanadaandNewZealand.Robberyarrestspeakfromage15tothemid-20s,andthebest-“tlagis23yearsinacombinedregressionand20…21yearsfortheUSA,Canada,WestGermany,andNewZealand.Thebest-“tlagforindexcrimeis18…21yearsintheUSA,Britain,Canada,Italy,Finland,andNewZealand.Somenationsshowlongerbest-“tsforsomecrimes,butbloodleadisgenerallystillhighlysigni“cantattheinternationalbest-“tforthatcategory.Althoughtimeseriescomparisonscanresultincoin-cidentalcorrelations,nonationshowsanycorrelationbetweenburglaryandbloodleadatlagsoflessthan10orover38years„thebloodleadcoef“cientinsuchregres-sionsisinsigni“cant.Nonationshowsanysigni“cantrelationshipbetweenrobberyorviolentandsexualassaultversusbloodleadwithalagoflessthan11years,betweenaggravatedassaultandbloodleadwithalagoflessthan14years,orbetweenrapeandbloodleadwithalagoflessthan13years.Changesinwhenunemploymentisaddedarealsoconsistentwithotherevidencethatunemploymenthasasubstantivelysmalleffectonpropertycrime(burglaryandmostindexcrime)andnoclearrelationshipwithviolence.ARTICLEINPRESS Fig.10.PreschoolBloodLeadvs.Robberywitha23-YearLag.Legend:Thebest-“tlagforpreschoolbloodleadversusrobberyis23yearsinacombined-nationregression.TheCanadianrobberyratewas“vetimestherateinBritainin1962,butthe2002CanadianratewaslessthanhalftherateiR.Nevin/EnvironmentalResearch104(2007)315–336 Theveryhighsigni“canceofbloodleadatlagsconsistentwithpeakoffendingagesisespeciallystrikinginlightofdivergentcrimeratetrends.Canadasindexcrimeratewas60%higherthantherateinBritainintheearly-1970s,but20%lowerin2001.TheUSAindexratewas22%higherthantheFrenchrateand40%higherthanAustraliasratein1980,buttheUSAratewas39%belowtheFrenchrateand45%belowAustraliasratein2001.The1974USAburglaryratewas50%and98%higherthanratesinBritainandAustralia,respectively,butthe2002USAratewas56%and63%lowerthanratesinBritainandAustralia.TheCanadianrobberyratewas“vetimestherateinBritainin1962,butthe2002CanadianratewaslessthanhalftherateinBritain.The1960USAaggravatedassaultratewasalmostthreetimestherateinBritain,butthe2002USAratewashalftherateinBritain.The1960USAraperatewaseighttimestheBritishrate,butthe2002USAraperatewasjust50%higherthantheBritishrate.Indexcrimerecordingdifferencesresultinlower(16.5%)inthecombined-nationindexcrimeregressionwithoutcountrydummies,butthesedifferencesalsomakethesigni“canceofbloodleadinthisregressionmoreremarkable.Thehigh(63…93%)ineachsingle-nationindexcrimeregressionwitha19-yearlagalsosuggeststhatbloodleadaffectsmanytypesofcriminalbehaviorincludingsimpleassaultsandpettythefts.Moreuniformrecordingofburglaryandrobberyresultinofalmost30%inthe5-nationburglaryregressionwithoutcountrydummies,andof46%inthe7-nationrobberyregressionwithoutcountrydummies.4.3.Cross-sectionalanalysisof1985–1994USAcentralcitymurderratesItiswellknownthat1980…1994USAmurderratesmainlyre”ectedtrendsinlargecities,butairleadandgasolineleadtrendscanexplainwhythelargestUSAcitieshadsuchhighmurderrates.Citieswithpopulationoveramillionhad1960sairleadabouttwicethelevelincitiesof250,000toamillion,whichhadairlead40%higherthancitiesof100…250thousand.Average1985…1994murderratesincity/citiesoveramillionwerethen57%higherthanincity/citiesof250,000toamillion,whichhadaverageARTICLEINPRESS Table4Regressionsforpreschoolbloodleadvs.aggravatedassaultandviolent&sexualassaultwitha23-yearlag,andvs.rapewitha24-yearlagDependentvariableIndependentvariableCoef“cientStandarderrorAgAssault„3NationsCombinedIntercept99.0525.793.840.0002(UK,US,FR)PreschoolBloodLead9.991.715.84.00010.258100AgAssault„3NationsCombinedIntercept18.9214.571.300.1974WithCountyDummies(notshown)PreschoolBloodLead22.251.2018.54.00010.803100USAAgAssaultIntercept20.6313.201.560.1257PreschoolBloodLead17.190.8420.44.00010.91143BritainAgAssaultIntercept116.8115.14PreschoolBloodLead36.621.4724.89.00010.93843FranceAgAssaultIntercept43.4376.270.570.5796PreschoolBloodLead8.013.272.450.03040.33414V&SAssault„3NationsCombinedIntercept257.0757.844.44(CAN,AUS,NZ)PreschoolBloodLead32.283.998.08.00010.50167V&SAssault„3NationsCombinedIntercept73.3137.241.970.0534WithCountyDummies(notshown)PreschoolBloodLead46.662.9315.92.00010.84467CanadaV&SAssaultIntercept96.1926.133.680.0007PreschoolBloodLead44.692.0621.69.00010.92341NewZealandV&SAssaultIntercept1510.00341.194.430.0007PreschoolBloodLead178.8425.407.04.00010.79215AustraliaV&SAssaultIntercept842.56480.611.750.1135PreschoolBloodLead75.2523.103.260.00990.54111Rape„4NationsCombinedIntercept0.142.040.070.9461(UK,USA,FR,FIN)PreschoolBloodLead1.230.148.63.00010.402113Rape„4NationsCombinedIntercept7.510.68WithCountyDummies(notshown)PreschoolBloodLead1.390.0624.72.00010.951113USARapeIntercept6.301.284.94PreschoolBloodLead1.570.0819.11.00010.89943BritainRapeIntercept5.490.82PreschoolBloodLead1.170.0814.49.00010.83743FranceRapeIntercept3.352.861.170.2648PreschoolBloodLead0.710.125.71.00010.73114FinlandRapeIntercept3.391.831.850.0915PreschoolBloodLead0.620.212.880.0150.42913:Incombinednationregressions,thebest-“tlagforaggravatedassaultandforviolentandsexualassaultis23years,andthebest“tforrapeis24years.Bloodleadisalsosigni“cantwitha23/24-yearlaginallsingle-nationregressions.Unemploymentisnotsigni“cantinassaultandraperegressions.R.Nevin/EnvironmentalResearch104(2007)315–336 1985…1994murderrates40%higherthancitiesof100…250thousand.LP%,re”ecting1970leadpaintpoisoning,isalsohighlysigni“cantinasimpleregression.TheregressionforcitysizeandLP%yieldshigherstatisticalsigni“cance-values)andexplanatorypower()thanseparatesimpleregressions,consistentwitheffectsofgasolinelead(citysize)andleadpainthazards(LP%),plustheadditiveeffectofpaintandgasolinelead(notcapturedbyseparateregressionsforcitysizeandLP%).AnassociationbetweenmurderandmoresevereleadexposurecouldexplainwhyWestGermanandAustralianbloodleadtrendsshownostatisticalrelationshipwithrecentmurdertrends.WestGermanylikelyhadalowprevalenceofseverelyelevatedbloodleadduetodestruc-tionofoldhousing(withleadpaint)duringWorldWarII.Australiadataalsoshowarelativelylow1990sprevalenceofelevatedbloodleadevenwhenaveragepreschoolbloodleadwasrelativelyhigh.Australianmurderrates(andincarceration)didfallfrom1900throughthe1940sfollowedbyalongslowrisesincethe1940s(Graycar,),consistentwithadeclineinpaintleadexposurefollowedbyrisinggasolineleadexposure.USAleadpaintpoisoningmusthavedeclinedasseverelydeterioratedslumsweredemolishedfromthemid-1950sthroughthe1960s,buttheUSAmurderrate”uctuatedfrom8to10per100,000from1971…1994.Therefore,thehypothesisthatmurderisespeciallyassociatedwithsevereexposureimpliesthatseveregasolineexposureincreasedasseverepainthazardsdeclined.Ruralandcitysizemurdertrendsareconsistentwiththatshift.Theruralshareofthepopulationwas26%in1980andin1990buttheruralshareofUSAmurdersfellfrom14%in1976to7%in1994,andtotalruralmurdersfell44%from1980…1994.Thatmurderdeclineisconsistentwithafallinruralpaintleadexposurefrom1940…1970,whentheaveragefarmhomewasabout35yearsold(USCensus,1975),sohalfof1940farmhomeswerebuiltbefore1905withhighlyleadedinteriorpaint,whereashalfof1970farmhomeswerebuiltafter1935wheninteriorleadpaintwasfarlesscommon.Urbanairleadroseasleadpaintexposurefell,anda1980smurderdeclineoutsideofcitiesover100,000wasoffsetbyasharpriseinlargecitieswiththeworst1960sairlead.From1981…1991,USAmurderratesrose3%incitiesof100…500thousand,9%incitiesof500,000to1million,and26%incitiesoveramillion.The1980sphase-outofgasleadleftlittleairleaddifferencebycitysize,andaverage2000…2002murderrateswere14.7(per100,000)incitiesoveramillion,14.6incitiesof500,000toamillion,15.0incitiesof250…500thousand,and9.5incitiesof100…250thousand(FoxandZawitz,2004ARTICLEINPRESS Fig.11.PreschoolBloodLeadvs.Assaultwitha23-YearLag(AggravatedAssaultorViolent&SexualAssault).Legend:Incombinednationregressionsthebest-“tlagversusbloodleadis23yearsforaggravatedassaultandforviolentandsexualassault.The1960USAaggravatedassaultratewasalmostthreetimestheBritishrate,andthe2002USAratewashalftherateinBritain,yetthebest-“tlagis22…24yearsforaggravatedassaultinsingle-natiregressionsforboththeUSAandBritain.R.Nevin/EnvironmentalResearch104(2007)315–336 Chicagomurdertrendsalsoprovideanecdotalevidenceofarisingpercentofmurdersrelatedtoseveregasolineleadexposure.In1980,18yearsafterits1962openingbesidetheDanRyanexpressway,RobertTaylorHomesaccountedfor0.5%ofChicagospopulationand11%ofChicagomurders(ONeill,1997Hagedorn(2004)expresswaysandhousingprojectsconcentratedChicagohomicidesinBlackareas,andillustratesthispointbymappinghighwaysagainst1965murderrates,presentedbesideapictureofRobertTaylorHomesandtheDanRyan.Butleadpaintpoisoninginlate-1940sslumsisalsoconsistentwithmurdersnearhighwaysin1965,whenchildrenfromthoseslumswereyouthslivingnearhigh-waysbuiltonslumclearanceland.HighwayairleadthenpeakedabouttwodecadesbeforeChicagos1992murderratepeak.Hagedornnotes:MurderinChicagoisnowmorecommoninthefarwesternandsouthernareasofthecity.Why?Hisspatialanalysisappearstoshow1992murderstrackingexpresswaystothewestandtheDanRyansouth,wherethe50%riseinUSApercapitagasolineleadusefrom1962…1970spreadleadpoisoningwellbeyondtheinnercity.4.4.Temporaltrends,cross-sectionalconfounders,andothercrimetheoriesrevisitedNeedlemanetal.(2003)foundthatsocialfactors,includingraceandsingle-parents,raiseddelinquencyriskforyouthswithlowerbonelead.Preschoolleadexposureishighlycorrelatedwithsocialfactorsbecausepoorchildrenaremorelikelytoliveinolderhousingwithdeterioratedpaint,andblackchildrenwereconcentratedincitieswithhigherairlead.Socialfactorscouldconstituteindependentoffendingrisksforthosewithnopreschoolleadexposure,and/orinteractwithleadexposuretoincreaseoffendingrisk,buttemporaltrendssuggestanyindependentsocialfactoreffectissmallrelativetotheleadeffect.Thejuvenilearrestratesoaredinthe1960s,trackingthesurgeingasleadafterWorldWarII,despitealarge1960sdeclineinthepercentofchildreninpoverty.Thatriseinjuvenileoffendingcoincidedwitha1960sriseintheunwedteenbirthrate,andthe1990sdeclineinjuvenilearrestscoincidedwithafallingunwedteenbirthrate.HigheroffendingduetosingleparentswouldbeconsistentwithjuvenileoffendingthatlaggedtheunwedbirthtrendbyARTICLEINPRESS Table5Regressionsformurdervs.preschoolbloodleadwithan18-yearlagDependentvariableIndependentvariableCoef“cientStandarderrorMurder„6NationsCombinedIntercept0.5230.4681.120.265(UK,US,CAN,AUS,NZ,WG)PreschoolBloodLead0.1920.0345.73.00010.137209Rape„6NationsCombinedIntercept0.0520.1600.320.7471WithCountyDummies(notshown)PreschoolBloodLead0.1140.01110.22.00010.925209USAMurderIntercept3.8270.4019.54PreschoolBloodLead0.2610.02510.34.00010.72343USAMurderwithUnemploymentRateIntercept2.9540.5924.99PreschoolBloodLead0.2380.0278.73.00010.74743UnemploymentRate0.2060.1061.950.0582BritainMurderIntercept0.4580.0647.21PreschoolBloodLead0.0650.00611.47.00010.76343BritainMurderwithUnemploymentRateIntercept0.4610.0647.17PreschoolBloodLead0.0690.0088.16.00010.76443UnemploymentRate0.0060.0110.560.5779CanadaMurderIntercept1.2800.1866.90PreschoolBloodLead0.0560.0144.030.00030.29441CanadaMurderwithUnemploymentRateIntercept1.0100.2054.92PreschoolBloodLead0.0130.0220.610.54250.39241UnemploymentRate0.1110.0452.480.0177AustraliaMurderIntercept2.2900.15914.44PreschoolBloodLead0.0200.0092.210.0350.14431AustraliaMurderwithUnemploymentRateIntercept2.3020.15414.98PreschoolBloodLead0.0400.0152.740.01050.22631UnemploymentRate0.0470.0271.720.0973WestGermanMurderIntercept1.3610.05226.27PreschoolBloodLead0.0020.0040.630.53660.02220WestGermanMurderwithUnemploymentRateIntercept1.3070.04727.93PreschoolBloodLead0.0120.0062.030.05880.35820UnemploymentRate0.0500.0172.980.0084NewZealandMurderIntercept0.5830.5051.150.2575PreschoolBloodLead0.2790.0426.63.00010.60331:Thebest-“tlagformurderis18yearsinregressionscombiningdataacrosssixnations.Unemploymentisinsigni“cant.Bloodleadisnotsigni“cantorhasanunexpectedsigninmurderregressionsforAustraliaandWestGermany,butbloodleadishighlysigni“cantinothersingle-nationregressionsformurderwithan18-yearlag.R.Nevin/EnvironmentalResearch104(2007)315–336 12…17years,aschildrenraisedbysinglemothersbecameteenagers.Thecoincidentriseandfallofunwedbirthratesandjuvenileoffendingisinconsistentwiththetime-precedenceindicatorofcausation.Nevin(2000)showedage-speci“cunwedpregnancyratestrackUSAgasleadtrendswithtimelagsconsistentwithmothersageandleadexposureinthe“rstyearoflife.Cross-sectionalstudiesthatlinkcriminaloffendingtosingleparentscouldre”ectseparateeffectsofpreschoolleadexposureondifferenttypesofimpulsivebehavior,acrossfamilygenerations.Socialtrendscannotexplainwhythe1990shomicidedeclinewassopronouncedamongjuvenileoffenders,andespeciallyblackjuveniles,butbloodleadtrendscan.Bloodleadprevalenceover30g/dLamongwhiteUSAchildrenfellfrom2%in1976…1980tolessthan0.5%in1988…1991,asprevalenceover30g/dLamongblackchildrenplum-metedfrom12%tobelow1%.Thewhitejuvenilemurderarrestratethenfellfrom6.4to2.1from1993…2003,astheblackjuvenileratefellfrom58.6to9.7.That83%fallintheblackjuvenilemurderarrestrateoccurredwithjust36%ofblackchildrenlivingintwo-parentfamiliesin1993,andin2003.Age-speci“carrestratesrelatedtopreschoolbloodleadcanexplainwhycrimepredictionsbasedon1990sdemographictrendsprovedinaccurate,andwhyincarcera-tionandcrimebothrosepriorto1990asincreasedoffendingbyjuvenilesandyoungadultsmorethanoffsettheincapacitationofolderoffenders.Gunuseofferslittleinsightintooverallcrimetrends,butgunhomicidesdidaccountformostofthe1973…2002USAmurderratevariation.Thesetrendsarenotinconsistentwiththehypothesisthatmurderisespeciallyaffectedbysevereleadpoisoning,butsuggestanespeciallylethalinteractionbetweengunaccessandsevereneurobehavioraldamage.Thiscouldexplainwhyruralmurdersfellafter1980despiteeasyruralaccesstoguns.The1990sfallinblackjuvenilemurderarrestscoincidedwithafallinblackyouthscarryingguns,butbloodleadtrendscouldexplainwhysomanyblackyouthsstoppedcarryinggunsatthesametime.Theblackpercentofcitypopulationappearstoexplainmuchofthecross-sectionalvariationinaverage1985…1994USAcentralcitymurderrates,butthe1990smurderratedeclinewasalsoledbyasharpdeclineinoffendingbyblacksincentralcities.NewYorkCitysracialcomposition(28%black)alsoprovidesnoinsightintotheespeciallystrikingdeclineinthatcitysmurderratefrom31(per100,000population)in1990to7in2004.The2004murderratewas20inDallas(27%black),14inPhoenix(6%black),13inHouston(26%black),13inLosAngeles(12%black),and8inSanAntonio(7%black).NewYorkCityhadextensiveslumdemolitionandreducedincin-eratorleademissionsinthe1960s,andbannedleadpaintin1960,resultinginalargereductioninleadpoisoning.NewYorkCitychildrenover60g/dLfellfrom2694in1970to494in1974,andchildrenover40g/dLfellfrom1595in1975to976in1980.NewYorkCityandSt.Louisbothreportedabout1200childrenperyearwithbloodleadARTICLEINPRESS Table6USAcentralcitymurderrateregressionsDependentvariableIndependentvariableCoef“cientStandarderrorCentralCityMurderRateIntercept13.64011.45419.38.00010.1141241million14.75604.11273.590.0005250,000…1million5.31372.16242.460.0154CentralCityMurderRateIntercept8.69472.11644.11.00010.139124LP%0.00780.00184.44CentralCityMurderRateIntercept2.04442.26290.900.36810.3291241million19.49113.67545.30250,000…1million7.07111.91133.700.0003LP%0.01000.00166.19CentralCityMurderRateIntercept6.36351.13845.59.00010.6081241million16.71512.75386.07250,000…1million4.95101.44583.420.0008Black%ofPopulation0.03090.002512.28CentralCityMurderRateIntercept1.40881.55350.910.36630.6881241million12.85522.58504.97250,000…1million4.16941.33473.120.0022LP%0.00170.00131.270.207Black%Population60.37295.177911.66:Theseregressionscompareaverage1985…1994murderratesacrossUSAcitieswithdifferencesincirca-1970childhoodleadpaintpoisoningandairleadexposure.Preschoolchildrenin1970wereinthehighmurderoffenseagebracketin1985…1994.TheLP%variablemeasuresthepercentofeacitys1985…1994populationthathadseverechildhoodleadpaintpoisoningin1970.Citysizedummieswereusedasindicatorsof1970airlead.CitysiandLP%arebothsigni“cantinsimpleregressions.WhenLP%andcitysizearebothincluded,regression(32.9%)exceedsthevariationexplainedbyseparateregressions(11.4%+13.9%25.3%)andLP%andcitysizearemoresigni“cant(higher-values),re”ectingtheadditiveeffectofpaintandgasolineleadnotcapturedbyseparateregressionsforcitysizeandLP%.Whenavariableisaddedforblackpercentofpopulation,citysizeisstillsigni“cantandLP%isnot(1.27),buttheLP%coef“cientstillhastheexpectedsignandretainingLP%inthemodelincreasesto69%.R.Nevin/EnvironmentalResearch104(2007)315–336 over30g/dLfrom1981…1985,whenNewYorkspopula-tionwas16timeslarger.Chicago,Detroit,Baltimore,Philadelphia,andSt.Louisreport3…4%ofchildrentestedin1998…1999hadbloodleadover20g/dL,butNewYorkCityprevalenceover20g/dLwasjust0.4%(NewYorkCityDepartmentofHealth,2002MissouriDepartmentofHealthandSeniorServices,2004Meyeretal.,2003CentersforDiseaseControlandPrevention,2005landDepartmentoftheEnvironment,2004DonohueandLevitts(2001)theorylinkingearly-1970sabortionlegalizationtothe1990sUSAcrimedeclinehighlightsthelagbetweenbirthandpeakoffendingages.USApreschoolbloodleadpeakedintheearly-1970s,andbloodleadtrendscanexplainearlierUSAandinterna-tionalcrimetrends.DonohueandLevittcreditearlyabortionlegalizationwithearlycrimedeclinesinNewYorkandCalifornia,butstate-wideabortionlegalizationdidnotpresageanearlystate-wideNewYorkcrimedecline.ThatearlycrimedeclinewasevidentonlyinNewYorkCitywheretherewasapronounceddeclineinleadpoisoning(NewYorkState,1999,2004).Californiaalsolimitedgasleadperliterto0.26gin1977and0.18gin1978,beforeanationallimitof0.29gin1983(OctelLtd,1969…1990).Californiapercapita(leaded)gasolineusewasalso30%higherthantherestoftheUSAin1950,20%higheroverthe1950s,and10%higherinthe1960s.Californiasviolentcrimeratewasthen40%higherthantherestoftheUSAfrom1960…1990,anditsburglaryratewas75%higherinthe1960s,55%higherinthe1970s,and27%higherinthe1980s(FederalHighwayAdministration,2003;USCensus,1975BureauofJusticeStatistics,20065.ConclusionsThisanalysisaddstomountingevidencethatpreschoolleadexposureaffectstheriskofcriminalbehaviorlaterinlife.Arrestrateshiftsandinternationaltrendssuggestthatpreschoolbloodleadespeciallyaffectsjuvenileoffendingandrelatedtrendsinindexcrime(mainlypropertycrime)andburglary.Violentcrimetrendsandshiftstohigheradultarrestratessuggestbloodleadalsoaffectsviolentandrepeatoffending.Itislikelythatpolicerecordedcrimeandarrestratetrendsexaminedherealsounderstatetheeffectthatleadhadonthe1990sUSAcrimedecline,becausecrimesurveydatashowanevensteeper1990sviolentcrimedecline,asalargershareofcrimeswerereportedtoandrecordedbypolice,andviolentcrimearrestsfelllessthanpolicerecordedviolentcrimes,asalargerpercentofreportedcrimeswereclearedbyarrest(BureauofJusticeStatistics,2007Thehypothesisthatmurderratesareespeciallyaffectedbysevereleadpoisoningisconsistentwithinternationalandracialcontrastsandacross-sectionalanalysisofaverage1985…1994USAcitymurderrates.Whetherotheroffendingrisksareespeciallyrelatedtoabloodleadthresholdisnotknown.Nothresholdissuggestedbyrisingcrimethattracesbacktoaveragepreschoolbloodleadof5g/dLorlessinnationswhereseverepoisoningwasrare.USAjuvenilearrestratesfallingthrough2003alsoshownoevidenceofathresholdrelatedtohistoricallylowlate-1980sbloodlead.OtherresearchlinkspreschoolleadexposuretoawidevarietyofadverseneurodevelopmentaleffectsincludingADHD,otherbehavioralproblems,andIQlosses.Age-speci“carrestratessuggestIQmayhaveonlyalimitedindirectrelationshipwithcrime,butevidenceofnolowerthresholdforlead-inducedIQeffectswarnsthatthereisnolowerthresholdforneuraldamage,andnoreasontoassumethatlowerbloodleadaffectsIQandnotothermanifestationsofneuraldamage,includingcriminalbehavior.Furtherresearchisneededtospecifythemechanismsbywhichbloodleadaffectsbehavior,andhowbloodleadinteractswithotherriskfactors,butpolicyimplicationsofthisstudyandrelatedresearchareclear:Theassociationbetweencrimeandpreschoolbloodleadshouldlendurgencytoglobaleffortstoeliminatepreschoolleadexposure.Thisanalysishasfocusedongasolineandpaintleadasdeterminantsoftrendsinaveragepreschoolbloodleadandsevereleadpoisoningprevalence,butchildrenandpregnantwomenarealsoexposedtooccupationalandsecondaryleadexposure(leaddustbroughthomeonworkclothes),leadindrinkingwaterfromoldwatermainsandservicelineconnectors,industrialleademissions,lead-contaminatedtoxicwastesites,lead-glazedceramics,andhomeremediesandcosmeticsinsomenations(RapuanoandFlorini,1994Actionisneededtoaddressalltheserisks.Over30nationsstilluseleadedgasoline,andaplannedglobalphase-outby2008mustbeaccelerated(UnitedNations,2005LeadpainthazardsarebyfarthegreatestremainingUSAleadexposurerisk.Theactionsneededtoeliminatesuchhazardsarewellknown,andhazardreductioncostsaremorethanoffsetbyhigheraveragelifetimeearningsresultingfromavoidedcognitivelosses(PresidentsTaskForceonEnvironmentalHealthRisksandSafetyRiskstoChildren,2000USDepartmentofHousingandUrbanDevelopment,1999).Asimplewindowreplacementstrat-egycanalsoyieldlong-termleadpainthazardreductionplusenergysavingsfromhigh-ef“ciencywindowsthatexceedwindowreplacementcosts(JacobsandNevin,2006NevinandJacobs,2006Nevinetal.,1999).Avoidedcrimebene“tsfurtherincreasenetbene“ts,andthisanalysissuggeststhatfurtherreductionsinpreschoolleadexposurewillyieldfurtherreductionsincrime.AcknowledgmentsTheauthoracknowledgesthevaluableregressionanaly-siscontributionsofArtemGonopolskiy.ReferencesAgencyforToxicSubstancesandDiseaseRegistry,1988.TheNatureandExtentofLeadPoisoninginChildrenintheUnitedStates.USDepartmentofHealthandHumanServices.ARTICLEINPRESSR.Nevin/EnvironmentalResearch104(2007)315–336 AgencyforToxicSubstancesandDiseaseRegistry,2000.CaseStudiesinEnvironmentalMedicine:LeadToxicity.USDepartmentofHealthandHumanServices.AmericanHighwayUsersAlliance,2004.UncloggingAmericasArteries:EffectiveReliefforHighwayBottlenecks1999…2004.RetrievedMarch11,2006,fromhttp://www.highways.org/pdfs/bottleneck2004.pdfAmericanPsychologicalAssociation,2004.AmericanPsychologicalAssociation,andtheMissouriPsychologicalAssociationasAmiciCuriaeSupportingRespondent.RetrievedMarch9,2006,fromhttp://www.apa.org/psyclaw/roper-v-simmons.pdfAustralianInstituteofCriminology,1999…2006.Australiancrime:factsand“gures.RetrievedMarch9,2006,from:http://www.aic.gov.au/publications/facts/AustralianInstituteofHealth&Welfare,1996.LeadinAustralianChildren:SummaryoftheNationalSurveyofLeadinChildren.Banks,E.,Ferretti,L.,Shucar,D.,1997.Effectsoflowlevelleadexposureoncognitivefunctioninchildren:areviewofbehavioral,neuropsychologicalandbiologicalevidence.Neurotoxicology18,237…281.Barclay,G.,Tavares,C.,2003.InternationalComparisonsofCriminalJusticeStatistics2001.UKHomeOf“ce,Issue12/03Bartzokis,G.,Beckson,M.,Lu,P.H.,Nuechterlein,K.H.,Edwards,N.,Mintz,J.,2001.Age-relatedchangesinfrontalandtemporallobevolumesinmen:amagneticresonanceimagingstudy.Arch.Gen.Psychiatry58,461…465.Beck,A.,Harrison,P.,2001.Prisonersin2000,BureauofJusticeStatistics,NCJ188207.RetrievedMarch9,2006,from:www.ojp.usdoj.gov/bjs/pub/pdf/p00.pdfBellinger,D.,2004.Lead.Pediatrics113,1016…1022.BureauofJusticeStatistics,1993;2001.MostseriousoffenseofStateprisoners,byoffense,admissiontype,age,gender,race,andHispanicorigin.RetrievedMarch11,2006,fromhttp://www.ojp.usdoj.gov/bjs/dtdata.htm#ncrpBureauofJusticeStatistics,2004.Arrestsbyagegroup,numberandratesfortotaloffenses,violentoffenses,andpropertyoffenses,1970…2003,12/04.RetrievedMarch9,2006,fromhttp://www.ojp.usdoj.gov/bjs/dtdata.htm#crimeBureauofJusticeStatistics,2005.Murders,robberies,andaggravatedassaultsinwhich“rearmswereused,numbersofoffensesandratesper100,000population,1973to2004.RetrievedMarch9,2006,fromhttp://www.ojp.usdoj.gov/bjs/glance/tables/guncrimetab.htmBureauofJusticeStatistics,2006.Criminaloffenses(numberofcrimesreportedtopoliceandcrimerates)since1960.RetrievedMarch9,2006,fromhttp://bjsdata.ojp.usdoj.gov/dataonline/Search/Crime/State/StateCrime.cfmBureauofJusticeStatistics,2007.Fourmeasuresofseriousviolentcrime.RetrievedFebruary12,2007,fromhttp://www.ojp.usdoj.gov/bjs/glance/tables/4meastab.htmBureauofLaborStatistics,2004.ComparativeCivilianLaborForceStatistics:TenCountries1959…2003.CaliforniaDepartmentofJustice,2005.CrimeinCalifornia2004.Can“eld,R.,Henderson,C.,Cory-Slechta,D.,Cox,C.,Jusko,T.,Lanphear,B.,2003.Intellectualimpairmentinchildrenwithbloodleadlevelsbelow10g/dL.NewEngl.J.Med.348,1517…1522.Chevron,2000.AviationFuelsTechnicalReview:AviationGasolineIntroduction.CouncilofEurope,2000.EuropeanSourcebookofCrimeandCriminalJusticeStatistics.RetrievedMarch9,2006,fromeuropeansourcebook.org/index.htmlCouncilofEurope,2003.EuropeanSourcebookofCrimeandCriminalJusticeStatistics„2003,seconded.,RetrievedMarch9,2006,fromhttp://www.ministerievanjustitie.nl:8080/b_organ/wodc/publications/ob212_01.pdfDenno,D.W.,1990.BiologyandViolence.CambridgeUniversityPress,NewYork.Dietrich,K.,Ris,M.,Succop,P.,Berger,O.,Bornschein,R.,2001.Earlyexposuretoleadandjuveniledelinquency.Neurotoxicol.Teratol.23,511…518.DiIulio,J.,1996.FillChurches,NotJails:YouthCrimeandSuper-predators,SenateJudiciaryCommittee,SubcommitteeonYouthViolence.RetrievedMarch9,2006,from:http://www.brookings.edu/views/testimony/diiulio/19960228.htmDonohue,J.,Levitt,S.,2001.Legalizedabortionandcrime.Quart.J.Econom.116,379…420.Dorsey,T.,Zawitz,M.,Middleton,P.,2005.DrugsandCrimeFacts,NCJ165148.RetrievedMarch9,2006,fromhttp://www.ojp.usdoj.gov/bjs/pub/pdf/dcf.pdfFederalHighwayAdministration,2003.NetMotor-FuelVolumeTaxed,byState,1950…1995.RetrievedMarch9,2006,fromwww.fhwa.dot.gov/ohim/summary95/section1.htmlFerrington,D.,Langan,P.,Tonry,M.,2004.Cross-NationalStudiesinCrimeandJustice.USDepartmentofJustice,Of“ceofJusticePrograms,BureauofJusticeStatistics.Foster,R.,2001.Yourbrainsnerveskeepdevelopingeventhroughyour40s.HealthScout.RetrievedMarch9,2006,from:http://www.loni.ucla.edu/media/News/HS_05162001.htmlFox,J.,Zawitz,M.,2004.HomicideTrendsintheUnitedStates,USDepartmentofJustice,Of“ceofJusticePrograms,BureauofJusticeStatistics.RetrievedMarch9,2006,from:http://www.ojp.usdoj.gov/bjs/homicide/homtrnd.htm#contentsFox,J.,1996.TrendsinJuvenileViolence,USBureauofJusticeStatistics.Gilsinn,J.,1972.EstimatesoftheNatureandExtentofLeadPaintPoisoningintheUnitedStates.USDepartmentofCommerce,NationalBureauofStandards.Gottfredson,L.,1998.Thegeneralintelligencefactor.Scient.Am.9,24…29.Graycar,A.,2001.CentenaryArticle„CrimeintwentiethcenturyAustralia,2001YearBookAustralia,ABSCatalogueNumber1301.0.Hagedorn,J.,2004.Gangresearch.net.RetrievedDecember14,2004,from:http://gangresearch.net/Globalization/guggenheim/chihomicide/pages/homwalls65.htmlhttp://gangresearch.net/Globalization/guggenheim/chihomicide/pages/homchang.htmlHarrison,P.,Beck,A.,2005.Prisonersin2004,BureauofJusticeStatistics,NCJ210677.RetrievedMarch9,2006,from:www.ojp.usdoj.gov/bjs/pub/pdf/p04.pdfHarrison,P.,2000.IncarcerationratesforprisonersunderStateorFederaljurisdiction,per100,000residents.BureauofJusticeStatistics,RetrievedMarch9,2006,from:http://www.ojp.usdoj.gov/bjs/dtdata.htm#correctionsInternationalEnergyAdministration,1960…1990.OECDProductSupplyandConsumption(1000tonnes):MotorGasoline,IEAOilInforma-tionDataService.Interpol,2005.InternationalCrimeStatisticsJackson,R.,1998.TheExposureofchildrentolead,byJ.JulianChisholmandHaroldE.Harrison.Pediatrics,1956:18;943…948.Pediatrics102,Jacobs,D.,Nevin,R.,2006.Validationofa20-yearforecastofUSchildhoodleadpoisoning:updatedprospectsfor2010.Environ.Res.102,352…364.Jacobs,D.,Clickner,R.,Zhou,J.,Viet,S.,Marker,D.,Rogers,J.,Zeldin,D.,Broene,P.,Friedman,W.,2002.Theprevalenceoflead-basedpainthazardsinUShousing.Environ.HealthPerspect.110,599…606.Koebel,C.,1996.UrbanRedevelopment,DisplacementandtheFutureoftheAmericanCity.CommunityAffairsOf“ce,FederalReserveBankofRichmond.Kristof,F.S.,1968.UrbanHousingNeedsthroughthe1980s:anAnalysisandProjection.TheNationalCommissiononUrbanProblems,ResearchReportNo.10.Langan,P.,Farrington,D.,1998.CrimeandJusticeintheUnitedStatesandinEnglandandWales,1981…96,BureauofJusticeStatistics.NCJ169284.RetrievedMarch9,2006,from:http://www.ojp.usdoj.gov/bjs/abstract/cjusew96.htmLangan,P.,Levin,D.,2002.RecidivismofPrisonersReleasedin1994,BureauofJusticeStatistics,NCJ193427.Levitt,S.D.,2004.Understandingwhycrimefellinthe1990s:Fourfactorsthatexplainthedeclineandsixthatdonot.J.Econom.Perspect.18,ARTICLEINPRESSR.Nevin/EnvironmentalResearch104(2007)315–336 Lidsky,T.,Schneider,J.,2003.Leadneurotoxicityinchildren:basicmechanismsandclinicalcorrelates.Brain126,5…19.MarylandDepartmentoftheEnvironment,2004.ChildhoodBloodLeadSurveillanceinMaryland2003AnnualReport,LeadPoisoningPreventionProgram.Matsuzawa,J.,Matsui,M.,Konishi,T.,Noguchi,K.,Gur,R.,Bilker,W.,Miyawaki,T.,2001.Age-relatedvolumetricchangesofbraingrayandwhitematterinhealthyinfantsandchildren.CerebralCortex11,McElvaine,M.,DeUngria,E.,Matte,T.,Copley,C.,Binder,S.,1992.Prevalenceofradiographicevidenceofpaintchipingestionamongchildrenwithmoderatetosevereleadpoisoning,St.Louis,Missouri,1989through1990.Pediatrics89,740…742.Meyer,H.,Mitchell,A.,1943.Leadandzincpigmentsandzincsalts.In:MineralsYearbook1941,USGeologicalSurvey,Washington,pp.165…178.Meyer,P.,Pivetz,T.,Dignam,T.,Homa,D.,Schoonover,J.,Brody,D.,2003.Surveillanceforelevatedbloodleadlevelsamongchildren1997…2001.MorbidityMortalityWeek.Rep.52(SS-10),1…22.MissouriDepartmentofHealthandSeniorServices,2004.FY2003BloodLeadTestingData.Mof“tt,T.,1993.Adolescence-limitedandlife-course-persistentantisocialbehavior:adevelopmentaltaxonomy.Psychol.Rev.100,674…701.Mukherjee,S.,Carcach,C.,Higgins,K.,1997.AStatisticalPro“leofCrimeinAustralia.AustralianInstituteofCriminology,ResearchandPublicPolicySeriesPreparedfortheSecondNationalOutlookSymposium,March1997.NationalInstituteofMentalHealth,2001.TeenageBrain:AWorkinProgress.NIHPublicationNo.01-4929Needleman,H.,Riess,J.,Tobin,M.,Biesecker,G.E.,Greenhouse,J.B.l.,1996.Boneleadlevelsanddelinquentbehavior.J.Am.Med.Assoc.257,363…369.Needleman,H.,McFarland,C.,Ness,R.,Fienberg,S.,Tobin,M.,2003.Boneleadlevelsinadjudicateddelinquents.Acasecontrolstudy.Neurotoxicol.Teratol.24,711…717.Neisser,U.,Boodoo,G.,Bouchard,T.,Boykin,A.,Brody,N.,Ceci,S.,Halpern,D.,Loehlin,J.,Perloff,R.,Sternberg,J.,Urbina,S.,1996.Intelligence:knownsandunknowns.Am.Psychol.51,Nevin,R.,Jacobs,D.,2006.Windowsofopportunity:leadpoisoningprevention,housingaffordability,andenergyconservation.HousingPolicyDebate17,185…207.Nevin,R.,Gazan,H.,Bender,C.,1999.Moreevidenceofrationalmarketvaluesforhomeenergyef“ciency.AppraisalJ.LXVII,454…460Nevin,R.,2000.HowleadexposurerelatestotemporalchangesinIQ,violentcrime,andunwedpregnancy.Environ.Res.83,1…22.NewYorkCityDepartmentofHealth,2002.SurveillanceofChildhoodBloodLeadLevelsinNewYorkCity.LeadPoisoningPreventionProgram.RetrievedMarch9,2006,from:http://www.nyc.gov/html/doh/html/lead/l2002.shtmlNewYorkState,1999.1997CrimeandJusticeAnnualReport.DivisionofCriminalJusticeServices.RetrievedMarch9,2006,http://www.criminaljustice.state.ny.us/crimnet/ojsa/cja_97/contents.htmNewYorkState,2004.2000…2001CrimeandJusticeAnnualReport.DivisionofCriminalJusticeServices.RetrievedMarch9,2006,http://www.criminaljustice.state.ny.us/crimnet/ojsa/cja_00_01/contents.htmNewZealandPolice,2001.NZPoliceStatisticstrendsforyearending30June2001.RetrievedMarch9,2006,from:http://www.police.govt.nz/service/statistics/NewZealandPolice,2005.NewZealandCrimeStatistics2004.RetrievedMarch9,2006,from:http://www.police.govt.nz/service/statistics/2004/calendar/stats-national-20041231.pdfONeill,R.,1997.CHADemolitions:OhYes,TheyCanTakeThatAwayFromMe.TheChicagoReporter.OctelLtd,1969…1990.WorldWideSurveyofMotorGasolineQuality.Octel,London.Of“ceofJuvenileJusticeandDelinquencyPrevention,2004.OJJDPStatisticalBrie“ngBookOnline.RetrievedMarch9,2006,from:http://ojjdp.ncjrs.org/ojstatbb/crime/faqs.asp?ID=T38OrganizationforEconomicCo-OperationandDevelopment,1993.RiskReductionMonographNo.1:Lead.Pirkle,J.,Brody,D.,Gunter,E.,Kramer,R.A.,Paschal,D.C.,Flegal,K.M.,Matte,T.D.,1994.ThedeclineinbloodleadlevelsintheUnitedStates.J.Am.Med.Assoc.272,284…291.PresidentsTaskForceonEnvironmentalHealthRisksandSafetyRiskstoChildren,2000.EliminatingChildhoodLeadPoisoning:AFederalStrategyTargetingLeadPaintHazards.USDepartmentofHousingandUrbanDevelopmentandUSEnvironmentalProtectionAgency,WashingtonDC.Raine,A.,Phil,D.,Stoddard,J.,Bihrle,S.,Buchsbaum,M.,1998.Prefrontalglucosede“citsinmurdererslackingpsychosocialdepriva-tion.Neuropsychiatr.Neuropsychol.Behav.Neurol.II,1…7.Rapuano,M.,Florini,K.,1994.TheGlobalDimensionsofLeadPoisoning:AnInitialAnalysis.AllianceToEndChildhoodLeadPoisoningandtheEnvironmentalDefenseFund.Reaves,B.,2003.Numberoffull-timeswornlawenforcementof“cers,rateper100,000residents,andpercentchangeintheUnitedStates,1975…1998,01/03.BureauofJusticeStatistics.RetrievedMarch9,2006,fromhttp://www.ojp.usdoj.gov/bjs/dtdata.htm#justiceRoyalSocietyofCanada,1986.LeadintheCanadianEnvironment:ScienceandRegulation.Sampson,R.,Laub,J.,2003.Life-CourseDesisters?Trajectoriesofcrimeamongdelinquentboysfollowedtoage70.Criminology41,555…592.Schwartz,J.,Pitcher,H.,1989.TherelationshipbetweengasolineleadandbloodleadintheUnitedStates.J.Off.Statist.5,421…431.Schwartz,J.,1994.Low-levelleadexposureandchildrensIQ:ameta-analysisandsearchforathreshold.Environ.Res.65,42…55.Silbergeld,E.,1992.Mechanismsofleadneurotoxicity,orlookingbeyondthelamppost.FASEBJ.6,3201…3206.Sowell,E.R.,Thompson,P.M.,Holmes,C.J.,Jernigan,T.,Toga,A.,1999.Invivoevidenceforpost-adolescentbrainmaturationinfrontalandstriatalregions.Nat.Neurosci.2,859…861.StatisticsCanada,2005.Crimesbytypeofoffence,CANSIM,table252-0013.RetrievedMarch9,2006,from:http://www.statcan.ca/english/Pgdb/legal02.htmStretesky,P.,Lynch,M.,2001.Therelationshipbetweenleadexposureandhomicide.Arch.Pediatr.Adol.Med.155,579…582.Taylor,R.,1999.FortyYearsofCrimeandCriminalJusticeStatistics,1958to1997.UKHomeOf“ce,ResearchDevelopmentandStatistics.Thomas,V.,Socolow,R.,Fanelli,J.,Spiro,T.,1999.Effectsofreducingleadingasoline:ananalysisoftheinternationalexperience.Environ.Sci.Technol.33,3942…3947.Triggs,S.,1997.InterpretingTrendsinRecordedCrimeinNewZealand.NewZealandMinistryofJusticeUSCensus,1960…1990.StatisticalAbstractoftheUnitedStates.USCensus,1975.HistoricalStatisticsoftheUnitedStates:ColonialTimesto1970.USGovernmentPrintingOf“ce.USCensus,1977…2003.Annual/AmericanHousingSurveys,CurrentHousingReportsSeriesH-150.RetrievedMarch9,2006,fromhttp://www.census.gov/prod/www/abs/h150.htmlUSCensus,2004.InternationalDataBase.RetrievedMarch9,2006,fromhttp://www.census.gov/ipc/www/idbnew.htmlUKDepartmentofEnvironment,2004.Emissionsoflead(Pb)totheatmospherefromallUKsources,1970to2002.RetrievedMarch9,2006,fromhttp://www.environment-agency.gov.uk/commondata/103608/f2_lead_dt_1161608.txtUKHomeOf“ce,2004.Summaryofrecordedcrimedatafrom1898to2002/03.RetrievedMarch9,2006,fromhttp://www.homeof“ce.gov.uk/rds/recordedcrime1.htmlUKInstituteofPetroleum,1938…1970.UKPetrolConsumption.UnitedNations,1950…1960.EnergyDatabase.U.N.StatisticalDivisionUnitedNations,2005.Sub-SaharanAfricaCelebratesLeadedPetrolPhase-Out:TargetSetforaGlobalPhase-Outby2008,U.N.EnvironmentProgramme.RetrievedMarch9,2006,fromwww.unep.org/pcfv/documents/NR2005_66.pdfARTICLEINPRESSR.Nevin/EnvironmentalResearch104(2007)315–336 USCentersforDiseaseControlandPrevention,1991.PreventingLeadPoisoninginYoungChildren,ReportNo.99-2230.USDepartmentofHealthandHumanServices,AtlantaGA.USCentersforDiseaseControlandPrevention,1997.Update:bloodleadlevels„UnitedStates,1991…1994.MorbidityMortalityWeek.Rep.46,141…146.USCentersforDiseaseControlandPrevention,2000.Bloodleadlevelsinyoungchildren„UnitedStatesandselectedstates,1996…1999.MorbidityMortalityWeek.Rep.49,1133…1137.USCentersforDiseaseControlandPrevention,2005.CDCSurveillanceData,1997…2004.RetrievedMarch9,2006,fromwww.cdc.gov/nceh/lead/surv/stats.htmUSDepartmentofHousingandUrbanDevelopment,1971.1970HUDStatisticalYearbook.USGovernmentPrintingOf“ceUSDepartmentofHousingandUrbanDevelopment,1999.EconomicAssessmentoftheFinalRuleonLead-BasedPaint:RequirementsforNoti“cation,EvaluationandReductionofLead-BasedPaintHazardsinFederallyOwnedResidentialPropertyandHousingReceivingFederalAssistance.Of“ceofLeadHazardControl,WashingtonDC.USEnvironmentalProtectionAgency,1986.AirQualityCriteriaforLead:VolumeIofIV,EnvironmentalCriteriaandAssessmentOf“ce,EPA600/8-83-028a-d.USEnvironmentalProtectionAgency,2001.NationalAirQualityandEmissionsTrendsReport,1999.USGeologicalSurvey,2006.LeadStatistics,RetrievedFebruary9,2007,http://minerals.usgs.gov/minerals/pubs/of01-006/lead.pdfUSPublicHealthService,1965.SurveyofLeadintheAtmosphereofThreeUrbanCommunities.EnvironmentalHealthSeries,PublicationNo.999-AP-12.USPublicHealthService,1966.AirQualityDatafromtheNationalAirSamplingNetworksandContributingStateandLocalNetworks1964…1965,CincinnatiOH.vanAlphen,M.,1998.PaintFilmComponents.NationalEnvironmentalHealthForum,GeneralSeriesNumber2.SouthAustralianHealthCommission,Adelaide.WorldHealthOrganization,1995.Inorganiclead.InternationalPro-grammeonChemicalSafety,Environmentalhealthcriteria;ISBN:92-4-157165-9,accessedat:http://www.inchem.org/documents/ehc/ehc/ehc165.htm#SectionNumber:3.3ARTICLEINPRESSR.Nevin/EnvironmentalResearch104(2007)315–336