Mill’sinsistenceuponthisimaginary’pluralityofcauses’
issignificant。Itindicatestheprecisestageinthedevelopmentoftheideaofcausetowhichhisdoctrinecorresponds。Takingwhatwemaycallthepopularsenseofcausation，the’plurality’
expressesanobvioustruth；andwecanunderstanditsplausibility。Wetake，infact，twoconcreteeventswhichfolloweachother，andcallthemcauseandeffect。Weuseatool——aknifetocutbread，forexample；weareforcedtoattendtothefactthateverydifferenceintheknifewillhaveaneffectontheresult。Theworkisbetterorworse，astheknifeissharperorblunter。Ifwedidnotrecognisethisineverypurposefulaction，allactionwouldbeintrinsicallyuncertain。Weare，therefore，impressedwiththenecessityofadmittingthattheeffectisdeterminedbythecause。But，ontheotherhand，theknifeisthere。Itmayhavebeenmadebyfiftydifferentmethods，andyetbethesame。Thehandlemayhavebeenfirstmadeandthentheblade，orviceversa，andsoforth。thereforewebelieve，andinthissenseofcausebelievecorrectly，thatoneeffectmaybetheproductofanynumberofdifferent’causes。’Inordertoreachthemorescientificsenseofcausation，wehavetotakeintoaccountallthatwehaveneglected。Theknifeisoneproductofanindefinitemultitudeofprocesses，andisthereforenotthetotal’effect’oftheconcreteantecedent，butonlyapartofitarbitrarilysingledout。Wedonotattendtoallthesecollateralresults，becauseforusatthemomenttheyhavenointerest；butwhenwesystematicallycarryoutthe’uniformityofnature’
principle，itisobviousthattheymustbetakenintoaccount。Wethenseethatalthoughpreciselysimilarproductsappearinaninfinitevarietyofconcreteprocesses，theycorrespondonlytoapartofthoseprocesses，andmayalwaysbeanalysedintoidenticalelements。Theeffectcannomorehavetwocausesthanacausetwoeffects，forcauseandeffectaredistinguishedbyobservingthesameprocessinadifferentorder。Itwasjustbecausemenofscienceheldthattheoneeffectmusthaveonecausethattheycouldmakeacoherenttheoryofheat。Mill，howevergoesastepfurther。Bacon’serrorwastheassumptionthattherewasonlyone’form’ofheat。Nowitisspeciallyfutile，saysMill，toseekforthecausesof’sensiblequalitiesofobjects……Inregardtoscarcelyanyofthemhasitbeenfoundpossibletotraceanyunityofcause。’Bacon，therefore，wasseekingfor’whatdidnotexist，’andtothisMilladdsthesurprisingstatementthat’thephenomenonofwhichhesoughtfortheonecausehasoftenestnocauseatall，and，whenithas，dependsasfarashithertoascertainedonanunassignablevarietyofcauses。’74*
ToexplainthisratherstartlingassertionwemusttakeonemoreofMill’stheories。Howfromthedoctrine，whichhefullyadmits，thateveryeventhasacausecanhereachtheconclusionthatsomethingshave’nocauseatall’？Oncemorewehave，I
think，themisapplicationofanundeniabletruth。A’law’ofcausation，takenbyitself，willobviouslynotfullyaccountforasinglefact。Itcannotleadtotheconclusion：’thisfactmustexist，’butonlytotheconclusion：thisfactmustexistifcertainpreviousfactsexisted。Wesomewhereassumeaninitialstage。Howeverfarbackwecango，wemaystillrepeatthequestion。Givenasinglestateoffactsandthe’lawsofcausation，’wecangoindefinitelybackwardsorforwardsintime。
Giventhesun，theplanets，andgravitation，wecantracethewholepastandfuturehistoryofthesolarsystem；butthefactsatsomeperiodmustbe’given。’Wecannotsaythattheymust，butonlythattheydoexist。Millhimselfputsthis75*withalldesirableclearness。Heexpressesitbysayingthatbesides’causation’thereis’collocation，’aword，hesays，suggestedbyChalmers。76*Toknowthe’collocation，’therefore，isessential。A’law’doesnottellusthatthere’must’beplumsandsuet，butonlythatiftherearesuchthingsincertain’collocations’aplumpudding’must’betheresult。Allstatementsoffacthavethusanempiricalbasis。This，however，takesapeculiarturninhisexposition，andonewhichischaracteristicofaUtilitarianfailing。Hemakesthedistinctionofrelationscorrespondtoadistinctionofthings。Insteadofsayingthatbothcausationandcollocationareimpliedinallphenomena，hespeaksofsome’uniformities’asdependentuponcausationandothersasdependentuponcollocation。Hethereforewritesachapteron’uniformitiesofcoexistencenotdependentoncausation。’77*This，however，iscloselyconnectedwith，andmustbeexplainedby，anotherdoctrinetowhichheattachedthehighestimportance。AftertellingushowhewasstartedafreshbyStewart’saccountofaxioms，headdsthathecameto’inextricabledifficulties’inregardtoinduction。Hehadcometothe’endofhistether，’and’couldmakenothingsatisfactoryofthesubject。’When，afterfiveyears’halt，heagainsettowork，heintroducedhis’theoryofkinds，’which，asheintimates，gotroundthedifficulty。Hefelt，aswemayconjecture，thathehadnowreducedallthefactstosuchpurelyempiricalconjunctionsthathedidnotseehowtogetanytiebetweenthem。Anycause，sofaraswehavegone，mightleadtoanyeffect。andevenwhenwehaveseenacaseofconjunction，wecangivenoreasonforitsrecurrence。Inductionenablesustopredicateattributesofaclass；butalogicalclassisitselfmerelyabundleofattributesarbitrarilyselected，anditremainstoseewhy，fromathing’spossessionofsomeoftheclassattributes，wecaninferthatithastheothers。Whyshouldnotthesamesetofattributesformpartofdifferentbundles？
andifso，whatisthejustificationfortheprimarylogicalprocedure？Fromfeatherlessbipedismweinfermortality。Butwhymaynotsomeclassoffeatherlessbipedsbeimmortal？Ifweadmitthepossibility，allinductionbecomesprecarious。The’theoryofkinds’was，itseems，intendedasananswertotheseobviousdifficulties。
VI。REALKINDS
Mill’saccountof’realkinds’corresponds，ashetellsus，totheoldlogicians’distinctionbetweengenusandspecies。
Thoughourclassificationmaybearbitraryandnothingproperlydeduciblefromit，exceptthemerefactthatwehavechosentogivenamestocertainclustersofattributes，thereisalsoarealdifference。Someofourclassesdonotcorrespondto’realkinds，’andaremistakenforthem。Others，however，correspondtoarealornaturalkind。Thedifferenceisthis：a’realkind’hasan’indeterminatemultitudeofproperties，notderivablefromeachother，’whereasanarbitraryormerelylogical’kind’mayonlydifferinrespectoftheparticularattributeassigned。
Thus，tosaythatNewtonisamanistoattributetohimthe’unknownmultitudeofproperties’connotedby’man。’TosaythatNewtonisaChristianisonlytoattributetohimaparticularbeliefandwhateverconsequencesmayfollowfromhavingthatbelief。78*Oneclassification，ashesays，’answerstoamuchmoreradicaldistinctioninthethingsthemselves，thantheotherdoes’；andamanmaythusfairlysay，ifhechooses，thatoneclassificationismade’byNature’andone’byourselves，’
providedthathemeansnomorethantoexpressthedistinctionjustdrawn。Now，itiseasytounderstandwhyMillfeltthatthisassertionentitledhimtoa’real’bondwhichwouldkeepphenomenatogetherinamoresatisfactoryway。Allthingshadbecomesolooseanddisconnectedthatitwasdifficulttoexplainanyextensionofknowledgeevenbyinduction。Yet，whateverthereason，thingsdosticktogetherincoherentandmany-propertiedclusters。Thebondseemstoberealwhenitisstated’objectively，’not’subjectively’——asapropertyofthethingsobserved，notoftheclassesmadebytheminditself。Itaketheremarktobebothtrueandimportant；and，moreover，thatMilldeservescreditforperceivingsoclearlythisweakjointinhisarmour。Hisapplication，however，suggests，whenhehadhituponanapparentescapefromhis’inextricable’difficulties，hewastoomuchrelievedtoworkoutitsfulleffectuponhisgeneraltheory。The’theoryofkinds’isinsertedratherthanembodiedinhisphilosophy，andmakesrentsintheattempttofillagap。Itplays，however，soimportantapartinthedoctrinethatitrequiressomefurtherconsideration。
Arealkind，wesee，hastwocharacteristics；ithasinnumerableproperties，andthosepropertiesarenot’derivable’
fromtheothers。Infact，aderivativepropertywouldbemerelyamodificationofaprimitiveproperty。Ageometrical’kind，’acurveofthesecondorder，forexample，hasinnumerableproperties，buttheyareallderivativefromthesimplepropertiesexpressedintheaxiomsanddefinitions。Theyreciprocallyimplyeachother。Butcanwesaythesameofthepropertiesofathing——ofaplantorofwaterorofanatom？
Herewehavethedistinctionalreadynoticed。Theso-called’thing’maybemerelyacollectionofseparatethings，andwecandiscoverthe’laws’applicabletoallbycombiningthelawsapplicabletoeach。Fromagiven’collocation’wecaninferpastorfuture’collocations，’andonesetofresultscanbeaddedtoorsuperposedupontheother。Butwhenweproceedtochemicalororganiccompounds，wehave’heteropathic’laws。Thecompoundmaybeanalysedintoelements，butwecannotderivethepropertiesofthecompoundsfromthepropertiesoftheelements。Hydrogenandoxygencanbecombinedintotheformofwater；butwecouldnotinferthepropertiesofwaterfromthepropertiesofthehydrogenandoxygentakenapart。Inorganiccompounds，theproblemisstillmoreintricate。Wehavetoconsideraseriesofinter-relatedchangestakingplacewithintheorganism，anddependentpartlyuponthe’environment’andpartlyuponthecomplexconstitutionoftheorganismitself。Itisaunitinsofarasallitspropertiesmanifestanorganiclaworasystemoforganiclaws。Individualsmaydifferfromexternalcausesasplants，forexample，indifferentsoils，andinthatcasewemayregardthedifferencesassimplyderivative。Differenceswhichbelongtotheorganiclawitselfindicatedifferencesofkind；
andtheseareultimateforus，solongaswecannottracethewayinwhichtheyaredependentupondifferencesofconstitution。
These，roughlystated，arethefactswhichMillrecognises。Now，inanycasewhatever，wecanonly’explain’afactbyassumingboth’collocation’and’causation’；or，inotherwords，wemusthaveastatementoffactsandoflaws。Ouranalysisofthephenomenawillinallcasescometoshowinghowagivenstateresultsfrom，orresultsin，apreviousorsucceedingstate。Ifnewpropertiesappearfromthecombinationofsimplerelements，weshouldinferthattheyresult，thoughwemaybequiteunableintheexistingstateofknowledgetoshowhowtheyresult，fromthepropertiesofthoseelements。Thepropertiesdonotmanifestthemselves，andarethereforenotdiscoverable，tillthecombinationisformed；andarethusonlyknowntous’empirically。’Noprocessofreasoning，thatis，canbeadducedtoshowthattheymustresultfromthecombination。But，inthecasesupposed，wedonotdoubtthattheydoresult，andweassumethattheelementshadcertainlatentpropertiesnotpreviouslydiscoverable。This，however，isthepointuponwhichMilldiverges，owing，asIthink，tohisimperfectviewofcausation。
Thedoctrineof’kinds，’infact，givestheanswertoMill’soldproblem，whyasingleinstanceissometimesconclusive，whereasanynumberofinstancesmaysometimesfailtogivecertainty。Itisthisreciprocalconnectionbetweenthepropertiesofa’kind’whichjustifiestheinferencefromonesetofattributestoanotherattribute——theinferenceimpliedinallinduction。ButMill’sinterpretationofthefactseemsstrangelyinconsistent。Hisfavouriteinstanceistheblackcrow。