IwillventuretosayanotherwordupontheuniformityofNaturedifficulty。Itiseasy，saysMill，toconceiveofthingshappeningatrandom。Itis，indeed，inonesenseperfectlyeasy。
’Raw’things，orunanalysedconcreteevents，dohappenatrandom，thatis，withoutuniformantecedents。Nothingiseasierthantothinkofthingswithoutthinkingoftheircauses。Theprimitivemind，andeventhecultivatedmind，maysimplywatchtheseriesofeventswithouttryingtofindanyconnectionorindulginginanyreasoning。Butthisisquitedifferentfromthinkingofthingsaspositivelyuncaused。Aphenomenonsuddenlyintrudeswithoutwarning。Imayacceptitwithoutaskingwhenceitcomes，orwhy。Butthereisnoreallypositivemeaninginthestatementthatitiscausedby’nothing。’Itdoesnotimplyacontradiction，suchasoccurswhenIputtogetherthewordscrookedandstraight，roundandsquare；butitrepresentsnointelligiblemeaning。Itcorrespondstoasimpleabsenceofthought。WhenIspeakoftheuniformityofNature，ImeansimplytoindicateaconditionofthinkingaboutNatureatall。Imayceasetoreasonortothink；butifIthink，Imustthinkcoherently，andassumewhathasbeencalledthe’UniversalPostulate。’92*Thephraseseemstometobeinadequate；andatanyrateitisapostulatewiththispeculiarity，thatwecannotmakeanyother。Todenyitistoallowcontradictorystatementsonthemostintimatetissueofourreasoning。Itisasimpossibletodowithoutitastodowithouttheprincipleofcontradictioninpurelogic。Ithelpsustonopositivestatement；butitisawarningthatourstatementsmustbecoherent。Hence，wemustallowthemindtohavethismodestcapacityforworkingupitsexperience。Ifitstartsfromsounprejudicedapointofviewastoadmitcontradictions，orallowofinconsistentstatementsaboutthings，itwillneverbeabletogetanywhere，andwhenMillhasreducedallourknowledgetotherelationsbetweenideasinthemind，itisreallyquiteinconsistenttoallowthemindnopowerofputtingideastogether。Withoutsuchapoweritisdifficulttosaywhatisevenmeantbytheperceptionof’coexistences’and’sequences。’Theprogressofknowledge，then，mustbeunderstoodascorrespondingtotheprocessbywhichthechaosofimpressionsandideasisgraduallyreducedtocosmos；
andasstartingfromapositioninwhichnocausehasbeenyetdiscoveredforgreatmassesoffacts，notfromapositioninwhich’nocause’isanequallyprobablealternativewith’somecause。’Toreasonatallaboutfactsistoarrangetheminorderofcausation，andtosupposethemashavingcertaintime-andspace-relations。Togetbehindthatprimitivegermofreasoningisreallytomakelogicimpossiblefromthestart。Mill’sdreadofaprioriintuitionsandnecessaryresultsthusledhimintoperfectlygratuitousdifficulties。Grantingthe’necessity’ofarithmeticorgeometry，itisstillahypotheticalnecessity。Itcannevertakeusbeyondexperience。Suchtheoremscannottellusoftheexistenceofasinglethingorofitsnature。Theycanonlysaythatifweseethingsinspacetheywillhavecertainrelationswhicharededuciblefromthespecialconfirmation。
Withoutthatpowertheuniversewouldbeundecipherable，butwithitourknowledgestillhasthroughoutacompletelyempiricalbase。Notasinglestatementoffactcanbemadewhichisnotderivedfrom，andjustifiedby，experience；norcanourexperienceevergetbeyondsayingthatanygivensectionofthewholeisdevelopedoutof，anddevelopsinto，precedingandsucceedingsections。
VIII。THEFOURMETHODS
IhavedweltuponthesemisconceptionstoshowwhyMillwasdrivenindefenceofexperiencetoassumetheburthenofprovingparadoxeswhichwouldbedestructivetoourverycapacityforobtainingexperience。Millpridedhimselfwithsomereasononhis’fourmethods。’Althoughtheyhavebeenseverelycriticised，93*
theyhave，Itakeit，agenuinevalue；and，ifweaskhowtheycanbevaluableinspiteofhiserrors，asatisfactoryanswermayperhapsbegiven。Inthefirstplace，hisassumptionsrepresentonegenuine’moment’inallreasoningaboutfacts。Theprimitiveintellectmaybesupposedtoregardfactsassimplyconjoined，andtobeguidedby’associationofideas。’TheearlygeneralisationsofwhichMillspeaks——’fireburns，’’waterdrowns，’andsofortharereallyofthiskind，andareapparentlyformedevenbydogsandmonkeys。Millisquiteright，moreover，inholdingthatapurelyempiricalelementrunsthroughthewholefabricofknowledge。Theerror，Ithink，isinhisfailuretoallowforthewayinwhichitismodifiedinscientificconstruction。Theultimateelementoutofwhichthatconstructionisdevelopedisalwaysanobservationoffact，butthefactmeansadefiniterelationoftimeandspace。Westartfroma’fact，’
butitisnotasasimpleunanalysableunit，butassomethingwhichalreadyisthebaseofarelation。Theunitwhichcorrespondstothefinalcelloutofwhichtissueiscomposedisnotproperlyafact，buta’truth。’Wedonotsaysimply’thisis，’butthisissoandso，andhasacertainorderandconfiguration。Thisisgraduallyelaboratedintophysicalsciencebythehelpofthegeometricalandnumericalrelationsalreadyimplied。Thus，causation，ortheconnectionbetweenphenomena，isnotsimplecollocation，butsupposescontinuity。TheunconditionalsequencewhichMillidentifieswith’causation’
doesnot，andcannot，givethe’cause，’thoughitdoesindicate’causalconnection。’Solongastwothingsareentirelyseparateanddistinguishable，wecannotsay，inthefullsense，thatoneisthecauseoftheother；buttheconnection，ifproved，provesthatthereisacausewhichmayormaynotbediscoverable。Brownwasrightinthinkingthatsomethingwasstillwanting，thoughhismodeoffillingthegapbyanintuitionwaserroneous。Mill’sanswerthatthe’intuition’wasneedlessleftthedifficultywhereHumehadputit。Twofactsaresupposedtobeunrelatedandyetalwayscombined。Thatstatesadifficulty，andonlypronouncesittobeinsoluble。Ithas，infact，tobesurmountedbyscientifichypotheses。Thunderandlightning，forexample，arecausallyconnected，butnotsothatlightningcanbeproperlycalledthecauseofthunder。Theyareregardedasduetoacommoncausetotheprocesseswhichwecallelectricdisturbance，andsoforth。Wecannotgivethe’law’orstatethecasualconnectionadequately，butweregardthemasindicatingsomecommonelement，whichiscontinuousandcapableofbeingdescribedintermsofpurenumberandgeometry。Henceanyobservation，assoonaswebegintoreason，mayberegardedasaparticularcaseofsomegenerallaw，orrather，asbeingconceivablyacaseofanindefinitenumberoflaws。Notonlyso，butanylawunderwhichitmaybearrangedis’necessary’ifalltheconditionsberestored。Theprocessbywhichweselectoneofthepossibleformula，therefore，comestoeliminatingalltheformulawhichareincorrectwhenvariousconditionsarealtered。Weallalongassumethatsomecoherentsystemof’laws’ispossible，orthattheruleisthereifonlywecandiscoverit。Iflightninggoesoncewiththunder，weareentitledtosaynotonlythatitmaygowiththunderhereafter，butthatitmustgowiththunderunderthesameconditions。Thereforethesimpleinferencefromanempiricalconjunctionisjustifiedbythe’lawofcausation’orthe’uniformityofnature。’
Now，Mill’s’fourmethods’areapplicabletothemerelyempiricalconjunctions，whichformalargepartofourknowledge，andareimpliedineverystage。Themethodsdo，infact，Itakeit，formanapproximatelyaccuratemodeofdealingwithsuchknowledge。Hiscasesare，forthemostpart，selectedfromthesciences，chemistryinparticular，whereinpointoffactourknowledgeisstillpurelyempirical，andwecanonlyassertacollocation，orsequence，withoutbringingitunderamoregeneralrule。Healsoobserves，andtheremarkmustberemembered，thatheistryingtogiveamethodofproof，ratherthanofdiscovery。94*Ifthescientifictheorybetrue，thesepurelyempiricaltruthswillholdgood，althoughfromthemalonethetheorymightnothavebeendiscoverable。Thephenomenonwhichwecallthefallofastonewillbepresentedwhenanunsupportedstoneisneartheearth，althoughthelawofgravitationrequiresanapplicationofmethodsnotsummedupbysimpleobservationofconjoinedphenomena。Themostunsatisfactorypartofthe’fourmethods’resultsfromthisview。95*TheprocessofdiscoveryisnotsufficientlyrepresentedbythecaseofAoccurringwithorwithoutB。Thescienceswhichhaverisentobequantitativeadvancebyshowinghowavarietyofcasescanbebroughtwithinsomegeneralandpreciseformula，andeveryapproximationto，ordeviationfrom，thelawbeexactlymeasured。Millpaystoolittleattentiontothisessentialcharacteristic，partly，perhaps，becauseheconsidersmathematicsassimplyonepartofthe’inductive’orempiricalsciences。
ThefinalpositionmaybeshortlyillustratedbyMill’srelationtohiscontemporaries。Itwillshowbrieflywhatwerethealternativesbetweenwhichhehadtochoose，andthat，ifthatwhichhechoseleadstoerror，therewereatleastequalerrorsontheotherside。MillfranklystatesinhisprefacethatbutforWhewell’sHistoryoftheInductiveSciencesthecorrespondingpartofhisownbook’wouldprobablynothavebeenwritten。’HeremarkswithequalcandourthatSirJohnHerschel，inhisDiscourseontheStudyofNaturalPhilosophy，96*hadrecognisedthefourmethods。Herschel，however，washisonlypredecessor，andamoredistinctandarticulateexhibitionoftheirnaturewasdesirable。HerschelandWhewellhadgraduatedatCambridgein1813and1816。Bothofthemwereablemathematicians，and，withtheircontemporaryBabbage，haddonemuchtointroduceattheiruniversitythemethodsofanalysisdevelopedontheContinent。Theuniversitywasgraduallyroused；
Herschelwonagreatnameinastronomy，andWhewelltookinearlierlifeaveryactivepartinpromotingscientificstudiesinEngland。97*BothofthemhadmuchcloseracquaintancewiththephysicalsciencesthanMill，forwhomtheyprovidedausefulstoreofmaterials。Herschel，thoughafriendofWhewell，approximatestoMill。A’famous’reviewofWhewell’stwobooksintheQuarterlyofJune184198*giveshisposition；butalthoughheseemstoperceivethesourceofWhewell’sweakness，hescarcelycomestoclosequarters。ItisenoughformypurposetospeakbrieflyofthepointsatissuebetweenMillandWhewell。
Whewell，likehismosteminentcontemporariesatCambridge，wasbecomingawarethatGermanspeculationcouldnolongerbeoverlooked。HerschelwassonofaGerman；andWhewell’sfriends，JuliusHare1795-1855andConnopThirlwall1797-1875weretakingupthestudyofGerman。TheirtranslationofNiebuhr’sHistoryofRome1828-1832markedanepochinEnglishscholarship。WhewellmeanwhilehadreadKant，andbeengreatlyimpressed。Especially，ashesays，heacceptedKant’stheoriesofspace，time，and，insomedegree，causation，althoughhedifferedfromKant’sdoctrineastootherso-called’fundamentalideas。’99*He’gladlyacknowledges，’too，hisobligationstotheScottishschool。100*Infact，itmaybesaidthat，likeSirW。Hamilton，hemadeacompromisebetweentwomodesofthoughtwhichveryrapidlydivergefromeachother。WhewellbeginshisPhilosophyoftheInductiveSciencesbyconsideringthefundamentalantithesisofphilosophy，whichcorrespondstothedistinctionbetweenthoughtsandthings，necessityandexperience，objectandsubject，andsoforth。Timeandspaceare，inhisphrase，’fundamentalideas，’uponwhicharefoundedthemathematicalsciences。Butthereareother’fundamentalideas’——
’cause，’’media，’’polarity，’’chemicalaffinityresemblance，’
’excitability，’and’finalcause’——whichinsuccessionbecomethefoundationofvarioussciences。
Thesefundamentalideasare，asheadmits，somethinglike’innateideas，’exceptthattheycanbe’developed。’Theycansomehowbe’superinduceduponfacts，’andarenot’generatedbyexperience。’Ishallnotattempttoexplainatheorywhichseemstoberadicallyincoherent，andwhichmadenoconverts。ItwillbequiteenoughtonoticetwoofthepointsofcollisionwithMill。MillandWhewellagree101*thatthe’firstlawofmotion’
whichassertstheuniformrectilinearmotionofabodynotacteduponbyaforcewasunknowntillthetimeofGalileo。Whewelladmitsfurtherthat，’historicallyspeaking，’itwasmade’bymeansofexperiment。’Wehave，however，attainedapointofviewinwhichweaethatitmighthavebeencertainlyknowntobetrue，independentlyofexperience，Millnaturallyridiculesthisdoctrine，accordingtowhichweburdenourselveswith’truthsindependentofexperience’andyetadmitthattheywereproved’byor’bymeansof’experiment。’Thehistoryisadmittedonbothsides。Ithadbeenobservedthatthemotionofallbodiesceasesunlesstheyreceiveanewimpulse。Thestatementwastrue，thoughvague，forallbodiesupontheearth。Buttheprogressofastronomyandexactsciencesrequiredamoreprecisestatement。