Havingnowendeavouredtoshowthattheseveralforegoingassumptionsaretoacertainextentsupportedbyanalogousfacts，andhavingalludedtosomeofthemostdoubtfulpoints，wewillconsiderhowfarthehypothesisbringsunderasinglepointofviewthevariouscasesenumeratedintheFirstPart。Alltheformsofreproductiongraduateintooneanotherandagreeintheirproduct;
foritisimpossibletodistinguishbetweenorganismsproducedfrombuds，fromself—division，orfromfertilisedgerms;suchorganismsareliabletovariationsofthesamenatureandtoreversionsofthesamekind;andas，accordingtoourhypothesis，alltheformsofreproductiondependontheaggregationofgemmulesderivedfromthewholebody，wecanunderstandthisremarkableagreement。Parthenogenesisisnolongerwonderful，andifwedidnotknowthatgreatgoodfollowedfromtheunionofthesexualelementsderivedfromtwodistinctindividuals，thewonderwouldbethatparthenogenesisdidnotoccurmuchoftenerthanitdoes。Onanyordinarytheoryofreproductiontheformationofgraft—hybrids，andtheactionofthemaleelementonthetissuesofthemother—plant，aswellasonthefutureprogenyoffemaleanimals，aregreatanomalies;buttheyareintelligibleonourhypothesis。Thereproductiveorgansdonotactuallycreatethesexualelements;theymerelydeterminetheaggregationandperhapsthemultiplicationofthegemmulesinaspecialmanner。Theseorgans，however，togetherwiththeiraccessoryparts，havehighfunctionstoperform。Theyadaptoneorbothelementsforindependenttemporaryexistence，andformutualunion。Thestigmaticsecretionactsonthepollenofaplantofthesamespeciesinawhollydifferentmannertowhatitdoesonthepollenofonebelongingtoadistinctgenusorfamily。ThespermatophoresoftheCephalopodaarewonderfullycomplexstructures，whichwereformerlymistakenforparasiticworms;andthespermatozoaofsomeanimalspossessattributeswhich，ifobservedinanindependentanimal，wouldbeputdowntoinstinctguidedbysense—organs，——aswhenthespermatozoaofaninsectfindtheirwayintotheminutemicropyleoftheegg。
Theantagonismwhichhaslongbeenobserved（27/55。Mr。HerbertSpencer（’PrinciplesofBiology’volume2page430）hasfullydiscussedthisantagonism。），withcertainexceptions，betweengrowthandthepowerofsexualreproduction（27/56。Themalesalmonisknowntobreedataveryearlyage。
TheTritonandSiredon，whilstretainingtheirlarvalbranchiae，accordingtoFilippiandDumeril（’AnnalsandMag。ofNat。Hist。’3rdseries1866page157）
arecapableofreproduction。ErnstHaeckelhasrecently（’MonatsberichtAkad。
Wiss。Berlin’February2，1865）observedthesurprisingcaseofamedusa，withitsreproductiveorgansactive，whichproducesbybuddingawidelydifferentformofmedusa;andthislatteralsohasthepowerofsexualreproduction。
Krohnhasshown（’AnnalsandMag。ofNat。Hist。’3rdseriesvolume191862
page6）thatcertainothermedusae，whilstsexuallymature，propagatebygemmae。SeealsoKolliker’MorphologieundEntwickelungsgeschichtedesPennatulidenstammes’1872page12。）——betweentherepairofinjuriesandgemmation——andwithplants，betweenrapidincreasebybuds，rhizomes，etc。，andtheproductionofseed，ispartlyexplainedbythegemmulesnotexistinginsufficientnumbersfortheseprocessestobecarriedonsimultaneously。
Hardlyanyfactinphysiologyismorewonderfulthanthepowerofregrowth;
forinstance，thatasnailshouldbeabletoreproduceitshead，orasalamanderitseyes，tail，andlegs，exactlyatthepointswheretheyhavebeencutoff。Suchcasesareexplainedbythepresenceofgemmulesderivedfromeachpart，anddisseminatedthroughoutthebody。Ihaveheardtheprocesscomparedwiththatoftherepairofthebrokenanglesofacrystalbyre—
crystallisation;andthetwoprocesseshavethismuchincommon，thatintheonecasethepolarityofthemoleculesistheefficientcause，andintheothertheaffinityofthegemmulesforparticularnascentcells。Butwehaveheretoencountertwoobjectionswhichapplynotonlytotheregrowthofapart，orofabisectedindividual，buttofissiparousgenerationandbudding。
Thefirstobjectionisthatthepartwhichisreproducedisinthesamestageofdevelopmentasthatofthebeingwhichhasbeenoperatedonorbisected;
andinthecaseofbuds，thatthenewbeingsthusproducedareinthesamestageasthatofthebuddingparent。Thusamaturesalamander，ofwhichthetailhasbeencutoff，doesnotreproducealarvaltail;andacrabdoesnotreproducealarvalleg。Inthecaseofbuddingitwasshowninthefirstpartofthischapterthatthenewbeingthusproduceddoesnotretrogradeindevelopment，——thatis，doesnotpassthroughthoseearlierstages，whichthefertilisedgermhastopassthrough。Nevertheless，theorganismsoperatedonormultiplyingthemselvesbybudsmust，byourhypothesis，includeinnumerablegemmulesderivedfromeverypartorunitoftheearlierstagesofdevelopment;
andwhydonotsuchgemmulesreproducetheamputatedpartorthewholebodyatacorrespondingearlystageofdevelopment？
Thesecondobjection，whichhasbeeninsistedonbyDelpino，isthatthetissues，forinstance，ofamaturesalamanderorcrab，ofwhichalimbhasbeenremoved，arealreadydifferentiatedandhavepassedthroughtheirwholecourseofdevelopment;andhowcansuchtissuesinaccordancewithourhypothesisattractandcombinewiththegemmulesofthepartwhichistobereproduced？Inanswertothesetwoobjectionswemustbearinmindtheevidencewhichhasbeenadvanced，showingthatatleastinalargenumberofcasesthepowerofregrowthisalocalisedfaculty，acquiredforthesakeofrepairingspecialinjuriestowhicheachparticularcreatureisliable;andinthecaseofbudsorfissiparousgeneration，forthesakeofquicklymultiplyingtheorganismataperiodoflifewhenitcanbesupportedinlargenumbers。Theseconsiderationsleadustobelievethatinallsuchcasesastockofnascentcellsorofpartiallydevelopedgemmulesareretainedforthisspecialpurposeeitherlocallyorthroughoutthebody，readytocombinewiththegemmulesderivedfromthecellswhichcomenextinduesuccession。Ifthisbeadmittedwehaveasufficientanswertotheabovetwoobjections。
Anyhow，pangenesisseemstothrowaconsiderableamountoflightonthewonderfulpowerofregrowth。
Itfollows，also，fromtheviewjustgiven，thatthesexualelementsdifferfrombudsinnotincludingnascentcellsorgemmulesinasomewhatadvancedstageofdevelopment，sothatonlythegemmulesbelongingtotheearlieststagesarefirstdeveloped。Asyounganimalsandthosewhichstandlowinthescalegenerallyhaveamuchgreatercapacityforregrowththanolderandhigheranimals，itwouldalsoappearthattheyretaincellsinanascentstate，orpartiallydevelopedgemmules，morereadilythandoanimalswhichhavealreadypassedthroughalongseriesofdevelopmentalchanges。Imayhereaddthatalthoughovulescanbedetectedinmostorallfemaleanimalsatanextremelyearlyage，thereisnoreasontodoubtthatgemmulesderivedfrompartsmodifiedduringmaturitycanpassintotheovules。
Withrespecttohybridism，pangenesisagreeswellwithmostoftheascertainedfacts。Wemustbelieve，aspreviouslyshown，thatseveralgemmulesarerequisiteforthedevelopmentofeachcellorunit。Butfromtheoccurrenceofparthenogenesis，moreespeciallyfromthosecasesinwhichanembryoisonlypartiallyformed，wemayinferthatthefemaleelementgenerallyincludesgemmulesinnearlysufficientnumberforindependentdevelopment，sothatwhenunitedwiththemaleelementthegemmulesaresuperabundant。Now，whentwospeciesorracesarecrossedreciprocally，theoffspringdonotcommonlydiffer，andthisshowsthatthesexualelementsagreeinpower，inaccordancewiththeviewthatbothincludethesamegemmules。Hybridsandmongrelsarealsogenerallyintermediateincharacterbetweenthetwoparent—forms，yetoccasionallytheycloselyresembleoneparentinonepartandtheotherparentinanotherpart，orevenintheirwholestructure:noristhisdifficulttounderstandontheadmissionthatthegemmulesinthefertilisedgermaresuperabundantinnumber，andthatthosederivedfromoneparentmayhavesomeadvantageinnumber，affinity，orvigouroverthosederivedfromtheotherparent。Crossedformssometimesexhibitthecolourorothercharactersofeitherparentinstripesorblotches;andthisoccursinthefirstgeneration，orthroughreversioninsucceedingbudandseminalgenerations，ofwhichfactseveralinstancesweregivenintheeleventhchapter。InthesecaseswemustfollowNaudin（27/57。Seehisexcellentdiscussiononthissubjectin’NouvellesArchivesduMuseum’tome1page151。）andadmitthatthe\"essence\"
or\"element\"ofthetwospecies，——termswhichIshouldtranslateintothegemmules，——haveanaffinityfortheirownkind，andthusseparatethemselvesintodistinctstripesorblotches;andreasonsweregiven，whendiscussinginthefifteenthchaptertheincompatibilityofcertaincharacterstounite，forbelievinginsuchmutualaffinity。Whentwoformsarecrossed，oneisnotrarelyfoundtobeprepotentinthetransmissionofitscharactersovertheother;andthiswecanexplainbyagainassumingthattheoneformhassomeadvantageovertheotherinthenumber，vigour，oraffinityofitsgemmules。
Insomecases，however，certaincharactersarepresentintheoneformandlatentintheother;forinstance，thereisalatenttendencyinallpigeonstobecomeblue，and，whenabluepigeoniscrossedwithoneofanyothercolour，thebluetintisgenerallyprepotent。TheexplanationofthisformofprepotencywillbeobviouswhenwecometotheconsiderationofReversion。
Whentwodistinctspeciesarecrossed，itisnotoriousthattheydonotyieldthefullorpropernumberofoffspring;andwecanonlysayonthisheadthat，asthedevelopmentofeachorganismdependsonsuchnicely—balancedaffinitiesbetweenahostofgemmulesandnascentcells，weneednotfeelatallsurprisedthatthecommixtureofgemmulesderivedfromtwodistinctspeciesshouldleadtopartialorcompletefailureofdevelopment。Withrespecttothesterilityofhybridsproducedfromtheunionoftwodistinctspecies，itwasshowninthenineteenthchapterthatthisdependsexclusivelyonthereproductiveorgansbeingspeciallyaffected;butwhytheseorgansshouldbethusaffectedwedonotknow，anymorethanwhyunnaturalconditionsoflife，thoughcompatiblewithhealth，shouldcausesterility;orwhycontinuedcloseinterbreeding，ortheillegitimateunionsofheterostyledplants，inducethesameresult。Theconclusionthatthereproductiveorgansaloneareaffected，andnotthewholeorganisation，agreesperfectlywiththeunimpairedorevenincreasedcapacityinhybridplantsforpropagationbybuds;forthisimplies，accordingtoourhypothesis，thatthecellsofthehybridsthrowoffhybridisedgemmules，whichbecomeaggregatedintobuds，butfailtobecomeaggregatedwithinthereproductiveorgans，soastoformthesexualelements。
Inasimilarmannermanyplants，whenplacedunderunnaturalconditions，failtoproduceseed，butcanreadilybepropagatedbybuds。Weshallpresentlyseethatpangenesisagreeswellwiththestrongtendencytoreversionexhibitedbyallcrossedanimalsandplants。
Eachorganismreachesmaturitythroughalongerorshortercourseofgrowthanddevelopment:theformertermbeingconfinedtomereincreaseofsize，anddevelopmenttochangedstructure。Thechangesmaybesmallandinsensiblyslow，aswhenachildgrowsintoaman，ormany，abrupt，andslight，asinthemetamorphosesofcertainephemerousinsects，or，again，fewandstrongly—
marked，aswithmostotherinsects。Eachnewlyformedpartmaybemouldedwithinapreviouslyexistingandcorrespondingpart，andinthiscaseitwillappear，falselyasIbelieve，tobedevelopedfromtheoldpart;oritmaybeformedwithinadistinctpartofthebody，asintheextremecasesofmetagenesis。Aneye，forinstance，maybedevelopedataspotwherenoeyepreviouslyexisted。Wehavealsoseenthatalliedorganicbeingsinthecourseoftheirmetamorphosessometimesattainnearlythesamestructureafterpassingthroughwidelydifferentforms;orconversely，afterpassingthroughnearlythesameearlyforms，arriveatwidelydifferentmatureforms。Inthesecasesitisverydifficulttoacceptthecommonviewthatthefirst—formedcellsorunitspossesstheinherentpower，independentlyofanyexternalagency，ofproducingnewstructureswhollydifferentinform，position，andfunction。Butallthesecasesbecomeplainonthehypothesisofpangenesis。
Theunits，duringeachstageofdevelopment，throwoffgemmules，which，multiplying，aretransmittedtotheoffspring。Intheoffspring，assoonasanyparticularcellorunitbecomespartiallydeveloped，ituniteswith（or，tospeakmetaphorically，isfertilisedby）thegemmuleofthenextsucceedingcell，andsoonwards。Butorganismshaveoftenbeensubjectedtochangedconditionsoflifeatacertainstageoftheirdevelopment，andinconsequencehavebeenslightlymodified;andthegemmulescastofffromsuchmodifiedpartswilltendtoreproducepartsmodifiedinthesamemanner。Thisprocessmayberepeateduntilthestructureofthepartbecomesgreatlychangedatoneparticularstageofdevelopment，butthiswillnotnecessarilyaffectotherparts，whetherpreviouslyorsubsequentlyformed。Inthismannerwecanunderstandtheremarkableindependenceofstructureinthesuccessivemetamorphoses，andespeciallyinthesuccessivemetagenesesofmanyanimals。
Inthecase，however，ofdiseaseswhichsuperveneduringoldage，subsequentlytotheordinaryperiodofprocreation，andwhich，nevertheless，aresometimesinherited，asoccurswithbrainandheartcomplaints，wemustsupposethattheorganswereaffectedatanearlyageandthrewoffatthisperiodaffectedgemmules;butthattheaffectionbecamevisibleorinjuriousonlyaftertheprolongedgrowth，inthestrictsenseoftheword，ofthepart。Inallthechangesofstructurewhichregularlysuperveneduringoldage，weprobablyseetheeffectsofdeterioratedgrowth，andnotoftruedevelopment。
Theprincipleoftheindependentformationofeachpart，owingtotheunionofthepropergemmuleswithcertainnascentcells，togetherwiththesuperabundanceofthegemmulesderivedfrombothparents，andthesubsequentself—multiplicationofthegemmules，throwslightonawidelydifferentgroupoffacts，whichonanyordinaryviewofdevelopmentappearsverystrange。I
alludetoorganswhichareabnormallytransposedormultiplied。Forinstance，acuriouscasehasbeenrecordedbyDr。ElliottCoues（27/58。’Proc。BostonSoc。ofNat。Hist。’republishedin’ScientificOpinion’November10，1869page488。）ofamonstrouschickenwithaperfectadditionalRIGHTlegarticulatedtotheLEFTsideofthepelvis。Gold—fishoftenhavesupernumeraryfinsplacedonvariouspartsoftheirbodies。Whenthetailofalizardisbrokenoff，adoubletailissometimesreproduced;andwhenthefootofthesalamanderwasdividedlongitudinallybyBonnet，additionaldigitswereoccasionallyformed。
Valentininjuredthecaudalextremityofanembryo，andthreedaysafterwardsitproducedrudimentsofadoublepelvisandofdoublehind—limbs。（27/59。
Todd’Cyclop。ofAnat。andPhys。’volume41849—52page975。）Whenfrogs，toads，etc。，arebornwiththeirlimbsdoubled，assometimeshappens，thedoubling，asGervaisremarks（27/60。’CompteRendus’November14，1865page800。），cannotbeduetothecompletefusionoftwoembryos，withtheexceptionofthelimbs，forthelarvaearelimbless。Thesameargumentisapplicable（27/61。AspreviouslyremarkedbyQuatrefagesinhis’Metamorphosesdel’Homme’etc。1862page129。）tocertaininsectsproducedwithmultiplelegsorantennae，forthesearemetamorphosedfromapodalorantennae—lesslarvae。
AlphonseMilne—Edwards（27/62。Gunther’ZoologicalRecord’1864page279。）hasdescribedthecuriouscaseofacrustaceaninwhichoneeye—pedunclesupported，insteadofacompleteeye，onlyanimperfectcornea，andoutofthecentreofthisaportionofanantennawasdeveloped。Acasehasbeenrecorded（27/63。Sedgwick’Medico—Chirurg。Review’April1863page454。）ofamanwhohadduringbothdentitionsadoubletoothinplaceoftheleftsecondincisor，andheinheritedthispeculiarityfromhispaternalgrandfather。Severalcasesareknown（27/64。Isid。GeoffroySaint—Hilaire’Hist。desAnomalies’tome1
1832pages435，657;andtome2page560。）ofadditionalteethhavingbeendevelopedintheorbitoftheeye，and，moreespeciallywithhorses，inthepalate。Hairsoccasionallyappearinstrangesituations，as\"withinthesubstanceofthebrain。\"（27/65。Virchow’CellularPathology’1860page66。）
Certainbreedsofsheepbearawholecrowdofhornsontheirforeheads。AsmanyasfivespurshavebeenseenonbothlegsofcertainGame—fowls。InthePolishfowlthemaleisornamentedwithatopknotofhackleslikethoseonhisneck，whilstthefemalehasatop—knotformedofcommonfeathers。Infeather—
footedpigeonsandfowls，featherslikethoseonthewingarisefromtheoutersideofthelegsandtoes。Eventheelementalpartsofthesamefeathermaybetransposed;forintheSebastopolgoose，barbulesaredevelopedonthedividedfilamentsoftheshaft。Imperfectnailssometimesappearonthestumpsoftheamputatedfingersofman（27/66。Muller’Phys。’EnglishTranslationvolume1
1833page407。Acaseofthiskindhaslatelybeencommunicatedtome。）anditisaninterestingfactthatwiththesnake—likeSaurians，whichpresentaserieswithmoreandmoreimperfectlimbs，theterminationsofthephalangesfirstdisappear，\"thenailsbecomingtransferredtotheirproximalremnants，oreventopartswhicharenotphalanges。\"（27/67。Dr。Furbringer’DieKnochenetc。beidenschlangenahnlichenSauriern’asreviewedin’JournalofAnat。andPhys。’May1870page286。）
Analogouscasesareofsuchfrequentoccurrencewithplantsthattheydonotstrikeuswithsufficientsurprise。Supernumerarypetals，stamens，andpistils，areoftenproduced。IhaveseenaleafletlowdowninthecompoundleafofViciasativareplacedbyatendril;andatendrilpossessesmanypeculiarproperties，suchasspontaneousmovementandirritability。Thecalyxsometimesassumes，eitherwhollyorbystripes，thecolourandtextureofthecorolla。Stamensaresofrequentlyconvertedintopetals，moreorlesscompletely，thatsuchcasesarepassedoverasnotdeservingnotice;butaspetalshavespecialfunctionstoperform，namely，toprotecttheincludedorgans，toattractinsects，andinnotafewcasestoguidetheirentrancebywell—adaptedcontrivances，wecanhardlyaccountfortheconversionofstamensintopetalsmerelybyunnaturalorsuperfluousnourishment。Again，theedgeofapetalmayoccasionallybefoundincludingoneofthehighestproductsoftheplant，namely，pollen;forinstance，Ihaveseenthepollen—massofanOphrys，whichisaverycomplexstructure，developedintheedgeofanupperpetal。
Thesegmentsofthecalyxofthecommonpeahavebeenobservedpartiallyconvertedintocarpels，includingovules，andwiththeirtipsconvertedintostigmas。Mr。SalterandDr。MaxwellMastershavefoundpollenwithintheovulesofthepassion—flowerandoftherose。Budsmaybedevelopedinthemostunnaturalpositions，asonthepetalofaflower。Numerousanalogousfactscouldbegiven。（27/68。Moquin—Tandon’TeratologieVeg。’1841pages218，220，353。Forthecaseofthepeasee’Gardener’sChronicle’1866page897。
WithrespecttopollenwithinovulesseeDr。Mastersin’ScienceReview’
October1873page369。TheRev。J。M。BerkeleydescribesabuddevelopedonapetalofaClarkiain’Gardener’sChronicle’April28，1866。）
Idonotknowhowphysiologistslookatsuchfactsastheforegoing。Accordingtothedoctrineofpangenesis，thegemmulesofthetransposedorgansbecomedevelopedinthewrongplace，fromunitingwithwrongcellsoraggregatesofcellsduringtheirnascentstate;andthiswouldfollowfromaslightmodificationintheirelectiveaffinities。Noroughtwetofeelmuchsurpriseattheaffinitiesofcellsandgemmulesvarying，whenwerememberthemanycuriouscasesgivenintheseventeenthchapter，ofplantswhichabsolutelyrefusetobefertilisedbytheirownpollen，thoughabundantlyfertilewiththatofanyotherindividualofthesamespecies，andinsomecasesonlywiththatofadistinctspecies。Itismanifestthatthesexualelectiveaffinitiesofsuchplants——tousethetermemployedbyGartner——havebeenmodified。Asthecellsofadjoiningorhomologouspartswillhavenearlythesamenature，theywillbeparticularlyliabletoacquirebyvariationeachother’selectiveaffinities;andwecanthusunderstandtoacertainextentsuchcasesasacrowdofhornsontheheadsofcertainsheep，ofseveralspursonthelegsoffowls，hackle—likefeathersontheheadsofthemalesofotherfowls，andwiththepigeonwing—likefeathersontheirlegsandmembranebetweentheirtoes，forthelegisthehomologueofthewing。Asalltheorgansofplantsarehomologousandspringfromacommonaxis，itisnaturalthattheyshouldbeeminentlyliabletotransposition。Itoughttobeobservedthatwhenanycompoundpart，suchasanadditionallimboranantenna，springsfromafalseposition，itisonlynecessarythatthefewfirstgemmulesshouldbewronglyattached;forthesewhilstdevelopingwouldattractothergemmulesinduesuccession，asintheregrowthofanamputatedlimb。Whenpartswhicharehomologousandsimilarinstructure，asthevertebraeofsnakesorthestamensofpolyandrousflowers，etc。，arerepeatedmanytimesinthesameorganism，closelyalliedgemmulesmustbeextremelynumerous，aswellasthepointstowhichtheyoughttobecomeunited;and，inaccordancewiththeforegoingviews，wecantoacertainextentunderstandIsid。GeoffroySaint—Hilaire’slaw，thatparts，whicharealreadymultiple，areextremelyliabletovaryinnumber。
Variabilityoftendepends，asIhaveattemptedtoshow，onthereproductiveorgansbeinginjuriouslyaffectedbychangedconditions;andinthiscasethegemmulesderivedfromthevariouspartsofthebodyareprobablyaggregatedinanirregularmanner，somesuperfluousandothersdeficient。Whetherasuperabundanceofgemmuleswouldleadtotheincreasedsizeofanypartcannotbetold;butwecanseethattheirpartialdeficiency，withoutnecessarilyleadingtotheentireabortionofthepart，mightcauseconsiderablemodifications;forinthesamemannerasplants，iftheirownpollenbeexcluded，areeasilyhybridised，so，inthecaseofcells，iftheproperlysucceedinggemmuleswereabsent，theywouldprobablycombineeasilywithotherandalliedgemmules，aswehavejustseenwithtransposedparts。
Invariationscausedbythedirectactionofchangedconditions，ofwhichseveralinstanceshavebeengiven，certainpartsofthebodyaredirectlyaffectedbythenewconditions，andconsequentlythrowoffmodifiedgemmules，whicharetransmittedtotheoffspring。Onanyordinaryviewitisunintelligiblehowchangedconditions，whetheractingontheembryo，theyoungortheadult，cancauseinheritedmodifications。Itisequallyorevenmoreunintelligibleonanyordinaryview，howtheeffectsofthelong—continueduseordisuseofapart，orofchangedhabitsofbodyormind，canbeinherited。A
moreperplexingproblemcanhardlybeproposed;butonourviewwehaveonlytosupposethatcertaincellsbecomeatlaststructurallymodified;andthatthesethrowoffsimilarlymodifiedgemmules。Thismayoccuratanyperiodofdevelopment，andthemodificationwillbeinheritedatacorrespondingperiod;
forthemodifiedgemmuleswilluniteinallordinarycaseswiththeproperprecedingcells，andwillconsequentlybedevelopedatthesameperiodatwhichthemodificationfirstarose。Withrespecttomentalhabitsorinstincts，wearesoprofoundlyignorantoftherelationbetweenthebrainandthepowerofthoughtthatwedonotknowpositivelywhetherafixedhabitinducesanychangeinthenervoussystem，thoughthisseemshighlyprobable;
butwhensuchhabitorothermentalattribute，orinsanity，isinherited，wemustbelievethatsomeactualmodificationistransmitted（27/69。SeesomeremarkstothiseffectbySirH。Hollandinhis’MedicalNotes’1839page32。）;andthisimplies，accordingtoourhypothesis，thatgemmulesderivedfrommodifiednerve—cellsaretransmittedtotheoffspring。
Itisgenerallynecessarythatanorganismshouldbeexposedduringseveralgenerationstochangedconditionsorhabits，inorderthatanymodificationthusacquiredshouldappearintheoffspring。Thismaybepartlyduetothechangesnotbeingatfirstmarkedenoughtocatchattention，butthisexplanationisinsufficient;andIcanaccountforthefactonlybytheassumption，whichweshallseeundertheheadofreversionisstronglysupported，thatgemmulesderivedfromeachunmodifiedunitorpartaretransmittedinlargenumberstosuccessivegenerations，andthatthegemmulesderivedfromthesameunitafterithasbeenmodifiedgoonmultiplyingunderthesamefavourableconditionswhichfirstcausedthemodification，untilatlasttheybecomesufficientlynumeroustooverpowerandsupplanttheoldgemmules。
Adifficultymaybeherenoticed;wehaveseenthatthereisanimportantdifferenceinthefrequency，thoughnotinthenature，ofthevariationsinplantspropagatedbysexualandasexualgeneration。Asfarasvariabilitydependsontheimperfectactionofthereproductiveorgansunderchangedconditions，wecanatonceseewhyplantspropagatedasexuallyshouldbefarlessvariablethanthosepropagatedsexually。Withrespecttothedirectactionofchangedconditions，weknowthatorganismsproducedfrombudsdonotpassthroughtheearlierphasesofdevelopment;theywillthereforenotbeexposed，atthatperiodoflifewhenstructureismostreadilymodified，tothevariouscausesinducingvariabilityinthesamemannerasareembryosandyounglarvalforms;butwhetherthisisasufficientexplanationIknownot。
Withrespecttovariationsduetoreversion，thereisasimilardifferencebetweenplantspropagatedfrombudsandseeds。Manyvarietiescanbepropagatedsecurelybybuds，butgenerallyorinvariablyreverttotheirparent—formsbyseed。So，also，hybridisedplantscanbemultipliedtoanyextentbybuds，butarecontinuallyliabletoreversionbyseed，——thatis，tothelossoftheirhybridorintermediatecharacter。Icanoffernosatisfactoryexplanationofthesefacts。Plantswithvariegatedleaves，phloxeswithstripedflowers，barberrieswithseedlessfruit，canallbesecurelypropagatedbybudstakenfromthestemorbranches;butbudsfromtherootsoftheseplantsalmostinvariablylosetheircharacterandreverttotheirformercondition。Thislatterfactisalsoinexplicable，unlessbudsdevelopedfromtherootsareasdistinctfromthoseonthestem，asisonebudonthestemfromanother，andweknowthattheselatterbehavelikeindependentorganisms。
Finally，weseethatonthehypothesisofpangenesisvariabilitydependsonatleasttwodistinctgroupsofcauses。Firstly，thedeficiency，superabundance，andtranspositionofgemmules，andtheredevelopmentofthosewhichhavelongbeendormant;thegemmulesthemselvesnothavingundergoneanymodification;
andsuchchangeswillamplyaccountformuchfluctuatingvariability。
Secondly，thedirectactionofchangedconditionsontheorganisation，andoftheincreaseduseordisuseofparts;andinthiscasethegemmulesfromthemodifiedunitswillbethemselvesmodified，and，whensufficientlymultiplied，willsupplanttheoldgemmulesandbedevelopedintonewstructures。
TurningnowtothelawsofInheritance。Ifwesupposeahomogeneousgelatinousprotozoontovaryandassumeareddishcolour，aminuteseparatedparticlewouldnaturally，asitgrewtofullsize，retainthesamecolour;andweshouldhavethesimplestformofinheritance。（27/70。ThisistheviewtakenbyProf。Hackelinhis’GenerelleMorphologie’b。2s。171，whosays:
\"LediglichdiepartielleIdentitatderspecifischconstituirtenMaterieimelterlichenundimkindlichenOrganismus，dieTheilungdieserMateriebeiderFortpflanzung，istdieUrsachederErblichkeit。\"）Preciselythesameviewmaybeextendedtotheinfinitelynumerousanddiversifiedunitsofwhichthewholebodyofoneofthehigheranimalsiscomposed;theseparatedparticlesbeingourgemmules。Wehavealreadysufficientlydiscussedbyimplication，theimportantprincipleofinheritanceatcorrespondingages。Inheritanceaslimitedbysexandbytheseasonoftheyear（forinstancewithanimalsbecomingwhiteinwinter）isintelligibleifwemaybelievethattheelectiveaffinitiesoftheunitsofthebodyareslightlydifferentinthetwosexes，especiallyatmaturity，andinoneorbothsexesatdifferentseasons，sothattheyunitewithdifferentgemmules。Itshouldberememberedthat，inthediscussionontheabnormaltranspositionoforgans，wehaveseenreasontobelievethatsuchelectiveaffinitiesarereadilymodified。ButIshallsoonhavetorecurtosexualandseasonalinheritance。Theseseverallawsarethereforeexplicabletoalargeextentthroughpangenesis，andonnootherhypothesiswhichhasasyetbeenadvanced。
Butitappearsatfirstsightafatalobjectiontoourhypothesisthatapartororganmayberemovedduringseveralsuccessivegenerations，andiftheoperationbenotfollowedbydisease，thelostpartreappearsintheoffspring。Dogsandhorsesformerlyhadtheirtailsdockedduringmanygenerationswithoutanyinheritedeffect;although，aswehaveseen，thereissomereasontobelievethatthetaillessconditionofcertainsheep—dogsisduetosuchinheritance。CircumcisionhasbeenpractisedbytheJewsfromaremoteperiod，andinmostcasestheeffectsoftheoperationarenotvisibleintheoffspring;thoughsomemaintainthataninheritedeffectdoesoccasionallyappear。Ifinheritancedependsonthepresenceofdisseminatedgemmulesderivedfromalltheunitsofthebody，whydoesnottheamputationormutilationofapart，especiallyifeffectedonbothsexes，invariablyaffecttheoffspring？Theanswerinaccordancewithourhypothesisprobablyisthatgemmulesmultiplyandaretransmittedduringalongseriesofgenerations——asweseeinthereappearanceofzebrinestripesonthehorse——inthereappearanceofmusclesandotherstructuresinmanwhicharepropertohislowlyorganisedprogenitors，andinmanyothersuchcases。Thereforethelong—continuedinheritanceofapartwhichhasbeenremovedduringmanygenerationsisnorealanomaly，forgemmulesformerlyderivedfromthepartaremultipliedandtransmittedfromgenerationtogeneration。
Wehaveasyetspokenonlyoftheremovalofparts，whennotfollowedbymorbidaction:butwhentheoperationisthusfollowed，itiscertainthatthedeficiencyissometimesinherited。Inaformerchapterinstancesweregiven，asofacow，thelossofwhosehornwasfollowedbysuppuration，andhercalvesweredestituteofahornonthesamesideoftheirheads。ButtheevidencewhichadmitsofnodoubtisthatgivenbyBrown—Sequardwithrespecttoguinea—pigs，whichaftertheirsciaticnerveshadbeendivided，gnawedofftheirowngangrenoustoes，andthetoesoftheiroffspringweredeficientinatleastthirteeninstancesonthecorrespondingfeet。Theinheritanceofthelostpartinseveralofthesecasesisallthemoreremarkableasonlyoneparentwasaffected;butweknowthatacongenitaldeficiencyisoftentransmittedfromoneparentalone——forinstance，theoffspringofhornlesscattleofeithersex，whencrossedwithperfectanimals，areoftenhornless。
How，then，inaccordancewithourhypothesiscanweaccountformutilationsbeingsometimesstronglyinherited，iftheyarefollowedbydiseasedaction？
Theanswerprobablyisthatallthegemmulesofthemutilatedoramputatedpartaregraduallyattractedtothediseasedsurfaceduringthereparativeprocess，andaretheredestroyedbythemorbidaction。
Afewwordsmustbeaddedonthecompleteabortionoforgans。Whenapartbecomesdiminishedbydisuseprolongedduringmanygenerations，theprincipleofeconomyofgrowth，togetherwithintercrossing，willtendtoreduceitstillfurtheraspreviouslyexplained，butthiswillnotaccountforthecompleteoralmostcompleteobliterationof，forinstance，aminutepapillaofcellulartissuerepresentingapistil，orofamicroscopicallyminutenoduleofbonerepresentingatooth。Incertaincasesofsuppressionnotyetcompleted，inwhicharudimentoccasionallyreappearsthroughreversion，dispersedgemmulesderivedfromthispartmust，accordingtoourview，stillexist;wemustthereforesupposethatthecells，inunionwithwhichtherudimentwasformerlydeveloped，failintheiraffinityforsuchgemmules，exceptintheoccasionalcasesofreversion。Butwhentheabortioniscompleteandfinal，thegemmulesthemselvesnodoubtperish;noristhisinanywayimprobable，for，thoughavastnumberofactiveandlong—dormantgemmulesarenourishedineachlivingcreature，yettheremustbesomelimittotheirnumber;anditappearsnaturalthatgemmulesderivedfromreducedanduselesspartswouldbemoreliabletoperishthanthosefreshlyderivedfromotherpartswhicharestillinfullfunctionalactivity。
Thelastsubjectthatneedbediscussed，namely，Reversion，restsontheprinciplethattransmissionanddevelopment，thoughgenerallyactinginconjunction，aredistinctpowers;andthetransmissionofgemmuleswiththeirsubsequentdevelopmentshowsushowthisispossible。Weplainlyseethedistinctioninthemanycasesinwhichagrandfathertransmitstohisgrandson，throughhisdaughter，characterswhichshedoesnot，orcannot，possess。Butbeforeproceeding，itwillbeadvisabletosayafewwordsaboutlatentordormantcharacters。Most，orperhapsall，ofthesecondarycharacters，whichappertaintoonesex，liedormantintheothersex;thatis，gemmulescapableofdevelopmentintothesecondarymalesexualcharactersareincludedwithinthefemale;andconverselyfemalecharactersinthemale:wehaveevidenceofthisincertainmasculinecharacters，bothcorporealandmental，appearinginthefemale，whenherovariaarediseasedorwhentheyfailtoactfromoldage。Inlikemannerfemalecharactersappearincastratedmales，asintheshapeofthehornsoftheox，andintheabsenceofhornsincastratedstags。Evenaslightchangeintheconditionsoflifeduetoconfinementsometimessufficestopreventthedevelopmentofmasculinecharactersinmaleanimals，althoughtheirreproductiveorgansarenotpermanentlyinjured。Inthemanycasesinwhichmasculinecharactersareperiodicallyrenewed，thesearelatentatotherseasons;inheritanceaslimitedbysexandseasonbeingherecombined。Again，masculinecharactersgenerallyliedormantinmaleanimalsuntiltheyarriveattheproperageforreproduction。ThecuriouscaseformerlygivenofaHenwhichassumedthemasculinecharacters，notofherownbreedbutofaremoteprogenitor，illustratesthecloseconnectionbetweenlatentsexualcharactersandordinaryreversion。
Withthoseanimalsandplantswhichhabituallyproduceseveralforms，aswithcertainbutterfliesdescribedbyMr。Wallace，inwhichthreefemaleformsandonemaleformco—exist，or，aswiththetrimorphicspeciesofLythrumandOxalis，gemmulescapableofreproducingthesedifferentformsmustbelatentineachindividual。
Insectsareoccasionallyproducedwithonesideoronequarteroftheirbodieslikethatofthemale，withtheotherhalforthree—quarterslikethatofthefemale。Insuchcasesthetwosidesaresometimeswonderfullydifferentinstructure，andareseparatedfromeachotherbyasharpline。Asgemmulesderivedfromeverypartarepresentineachindividualofbothsexes，itmustbetheelectiveaffinitiesofthenascentcellswhichinthesecasesdifferabnormallyonthetwosidesofthebody。Almostthesameprinciplecomesintoplaywiththoseanimals，forinstance，certaingasteropodsandVerrucaamongstcirripedes，whichnormallyhavethetwosidesofthebodyconstructedonaverydifferentplan;andyetanearlyequalnumberofindividualshaveeithersidemodifiedinthesameremarkablemanner。
Reversion，intheordinarysenseoftheword，actssoincessantly，thatitevidentlyformsanessentialpartofthegenerallawofinheritance。Itoccurswithbeings，howeverpropagated，whetherbybudsorseminalgeneration，andsometimesmaybeobservedwithadvancingageeveninthesameindividual。Thetendencytoreversionisofteninducedbyachangeofconditions，andintheplainestmannerbycrossing。Crossedformsofthefirstgenerationaregenerallynearlyintermediateincharacterbetweentheirtwoparents;butinthenextgenerationtheoffspringcommonlyreverttooneorbothoftheirgrandparents，andoccasionallytomoreremoteancestors。Howcanweaccountforthesefacts？Eachunitinahybridmustthrowoff，accordingtothedoctrineofpangenesis，anabundanceofhybridisedgemmules，forcrossedplantscanbereadilyandlargelypropagatedbybuds;butbythesamehypothesisdormantgemmulesderivedfrombothpureparent—formsarelikewisepresent;andasthesegemmulesretaintheirnormalcondition，theywould，itisprobable，beenabledtomultiplylargelyduringthelifetimeofeachhybrid。Consequentlythesexualelementsofahybridwillincludebothpureandhybridisedgemmules;andwhentwohybridspair，thecombinationofpuregemmulesderivedfromtheonehybridwiththepuregemmulesofthesamepartsderivedfromtheother，wouldnecessarilyleadtocompletereversionofcharacter;anditis，perhaps，nottooboldasuppositionthatunmodifiedandundeterioratedgemmulesofthesamenaturewouldbeespeciallyapttocombine。
Puregemmulesincombinationwithhybridisedgemmuleswouldleadtopartialreversion。Andlastly，hybridisedgemmulesderivedfrombothparent—hybridswouldsimplyreproducetheoriginalhybridform。（27/71。IntheseremarksI，infact，followNaudin，whospeaksoftheelementsoressencesofthetwospecieswhicharecrossed。Seehisexcellentmemoirinthe’NouvellesArchivesduMuseum’tome1page151。）Allthesecasesanddegreesofreversionincessantlyoccur。
Itwasshowninthefifteenthchapterthatcertaincharactersareantagonistictoeachotherordonotreadilyblend;hence，whentwoanimalswithantagonisticcharactersarecrossed，itmightwellhappenthatasufficiencyofgemmulesinthemalealoneforthereproductionofhispeculiarcharacters，andinthefemalealoneforthereproductionofherpeculiarcharacters，wouldnotbepresent;andinthiscasedormantgemmulesderivedfromthesamepartinsomeremoteprogenitormighteasilygaintheascendancy，andcausethereappearanceofthelong—lostcharacter。Forinstance，whenblackandwhitepigeons，orblackandwhitefowls，arecrossed，——colourswhichdonotreadilyblend，——blueplumageintheonecase，evidentlyderivedfromtherock—pigeon，andredplumageintheothercase，derivedfromthewildjungle—cock，occasionallyreappear。Withuncrossedbreedsthesameresultfollows，underconditionswhichfavourthemultiplicationanddevelopmentofcertaindormantgemmules，aswhenanimalsbecomeferalandreverttotheirpristinecharacter。
Acertainnumberofgemmulesbeingrequisiteforthedevelopmentofeachcharacter，asisknowntobethecasefromseveralspermatozoaorpollen—
grainsbeingnecessaryforfertilisation，andtimefavouringtheirmultiplication，willperhapsaccountforthecuriouscases，insistedonbyMr。
Sedgwick，ofcertaindiseaseswhichregularlyappearinalternategenerations。
Thislikewiseholdsgood，moreorlessstrictly，withotherweaklyinheritedmodifications。Hence，asIhavehearditremarked，certaindiseasesappeartogainstrengthbytheintermissionofageneration。Thetransmissionofdormantgemmulesduringmanysuccessivegenerationsishardlyinitselfmoreimprobable，aspreviouslyremarked，thantheretentionduringmanyagesofrudimentaryorgans，orevenonlyofatendencytotheproductionofarudiment;butthereisnoreasontosupposethatdormantgemmulescanbetransmittedandpropagatedforever。Excessivelyminuteandnumerousastheyarebelievedtobe，aninfinitenumberderived，duringalongcourseofmodificationanddescent，fromeachunitofeachprogenitor，couldnotbesupportedornourishedbytheorganism。Butitdoesnotseemimprobablethatcertaingemmules，underfavourableconditions，shouldberetainedandgoonmultiplyingforamuchlongerperiodthanothers。Finally，ontheviewheregiven，wecertainlygainsomeinsightintothewonderfulfactthatthechildmaydepartfromthetypeofbothitsparents，andresembleitsgrandparents，orancestorsremovedbymanyhundredsofgenerations。
CONCLUSION。
ThehypothesisofPangenesis，asappliedtotheseveralgreatclassesoffactsjustdiscussed，nodoubtisextremelycomplex，butsoarethefacts。Thechiefassumptionisthatalltheunitsofthebody，besideshavingtheuniversallyadmittedpowerofgrowingbyself—division，throwoffminutegemmuleswhicharedispersedthroughthesystem。Norcanthisassumptionbeconsideredastoobold，forweknowfromthecasesofgraft—hybridisationthatformativematterofsomekindispresentinthetissuesofplants，whichiscapableofcombiningwiththatincludedinanotherindividual，andofreproducingeveryunitofthewholeorganism。Butwehavefurthertoassumethatthegemmulesgrow，multiply，andaggregatethemselvesintobudsandthesexualelements;
theirdevelopmentdependingontheirunionwithothernascentcellsorunits。
Theyarealsobelievedtobecapableoftransmissioninadormantstate，likeseedsintheground，tosuccessivegenerations。
Inahighly—organisedanimal，thegemmulesthrownofffromeachdifferentunitthroughoutthebodymustbeinconceivablynumerousandminute。Eachunitofeachpart，asitchangesduringdevelopment，andweknowthatsomeinsectsundergoatleasttwentymetamorphoses，mustthrowoffitsgemmules。Butthesamecellsmaylongcontinuetoincreasebyself—division，andevenbecomemodifiedbyabsorbingpeculiarnutriment，withoutnecessarilythrowingoffmodifiedgemmules。Allorganicbeings，moreover，includemanydormantgemmulesderivedfromtheirgrandparentsandmoreremoteprogenitors，butnotfromalltheirprogenitors。Thesealmostinfinitelynumerousandminutegemmulesarecontainedwithineachbud，ovule，spermatozoon，andpollen—grain。Suchanadmissionwillbedeclaredimpossible;butnumberandsizeareonlyrelativedifficulties。Independentorganismsexistwhicharebarelyvisibleunderthehighestpowersofthemicroscope，andtheirgermsmustbeexcessivelyminute。
Particlesofinfectiousmatter，sosmallastobewaftedbythewindortoadheretosmoothpaper，willmultiplysorapidlyastoinfectwithinashorttimethewholebodyofalargeanimal。Weshouldalsoreflectontheadmittednumberandminutenessofthemoleculescomposingaparticleofordinarymatter。Thedifficulty，therefore，whichatfirstappearsinsurmountable，ofbelievingintheexistenceofgemmulessonumerousandsmallastheymustbeaccordingtoourhypothesis，hasnogreatweight。
Theunitsofthebodyaregenerallyadmittedbyphysiologiststobeautonomous。Igoonestepfurtherandassumethattheythrowoffreproductivegemmules。Thusanorganismdoesnotgenerateitskindasawhole，buteachseparateunitgeneratesitskind。Ithasoftenbeensaidbynaturaliststhateachcellofaplanthasthepotentialcapacityofreproducingthewholeplant;butithasthispoweronlyinvirtueofcontaininggemmulesderivedfromeverypart。Whenacellorunitisfromsomecausemodified，thegemmulesderivedfromitwillbeinlikemannermodified。Ifourhypothesisbeprovisionallyaccepted，wemustlookatalltheformsofasexualreproduction，whetheroccurringatmaturityorduringyouth，asfundamentallythesame，anddependentonthemutualaggregationandmultiplicationofthegemmules。Theregrowthofanamputatedlimbandthehealingofawoundisthesameprocesspartiallycarriedout。Budsapparentlyincludenascentcells，belongingtothatstageofdevelopmentatwhichthebuddingoccurs，andthesecellsarereadytounitewiththegemmulesderivedfromthenextsucceedingcells。Thesexualelements，ontheotherhand，donotincludesuchnascentcells;andthemaleandfemaleelementstakenseparatelydonotcontainasufficientnumberofgemmulesforindependentdevelopment，exceptinthecasesofparthenogenesis。Thedevelopmentofeachbeing，includingalltheformsofmetamorphosisandmetagenesis，dependsonthepresenceofgemmulesthrownoffateachperiodoflife，andontheirdevelopment，atacorrespondingperiod，inunionwithprecedingcells。Suchcellsmaybesaidtobefertilisedbythegemmuleswhichcomenextindueorderofdevelopment。Thustheactofordinaryimpregnationandthedevelopmentofeachpartineachbeingarecloselyanalogousprocesses。Thechild，strictlyspeaking，doesnotgrowintotheman，butincludesgermswhichslowlyandsuccessivelybecomedevelopedandformtheman。Inthechild，aswellasintheadult，eachpartgeneratesthesamepart。
Inheritancemustbelookedatasmerelyaformofgrowth，liketheself—
divisionofalowly—organisedunicellularorganism。Reversiondependsonthetransmissionfromtheforefathertohisdescendantsofdormantgemmules，whichoccasionallybecomedevelopedundercertainknownorunknownconditions。Eachanimalandplantmaybecomparedwithabedofsoilfullofseeds，someofwhichsoongerminate，someliedormantforaperiod，whilstothersperish。
Whenwehearitsaidthatamancarriesinhisconstitutiontheseedsofaninheriteddisease，thereismuchtruthintheexpression。Nootherattempt，asfarasIamaware，hasbeenmade，imperfectasthisconfessedlyis，toconnectunderonepointofviewtheseseveralgrandclassesoffacts。Anorganicbeingisamicrocosm——alittleuniverse，formedofahostofself—propagatingorganisms，inconceivablyminuteandnumerousasthestarsinheaven。
CHAPTER2。XXVIII。
CONCLUDINGREMARKS。
DOMESTICATION。
NATUREANDCAUSESOFVARIABILITY。
SELECTION。
DIVERGENCEANDDISTINCTNESSOFCHARACTER。
EXTINCTIONOFRACES。
CIRCUMSTANCESFAVOURABLETOSELECTIONBYMAN。
ANTIQUITYOFCERTAINRACES。
THEQUESTIONWHETHEREACHPARTICULARVARIATIONHASBEENSPECIALLYPREORDAINED。
Assummarieshavebeenaddedtonearlyallthechapters，andas，inthechapteronpangenesis，varioussubjects，suchastheformsofreproduction，inheritance，reversion，thecausesandlawsofvariability，etc。，havebeenrecentlydiscussed，Iwillhereonlymakeafewgeneralremarksonthemoreimportantconclusionswhichmaybededucedfromthemultifariousdetailsgiventhroughoutthiswork。
Savagesinallpartsoftheworldeasilysucceedintamingwildanimals;andthoseinhabitinganycountryorisland，whenfirstvisitedbyman，wouldprobablyhavebeenstillmoreeasilytamed。Completesubjugationgenerallydependsonananimalbeingsocialinitshabits，andonreceivingmanasthechiefoftheherdorfamily。Inorderthatananimalshouldbedomesticateditmustbefertileunderchangedconditionsoflife，andthisisfarfrombeingalwaysthecase。Ananimalwouldnothavebeenworththelabourofdomestication，atleastduringearlytimes，unlessofservicetoman。Fromthesecircumstancesthenumberofdomesticatedanimalshasneverbeenlarge。
Withrespecttoplants，Ihaveshownintheninthchapterhowtheirvarieduseswereprobablyfirstdiscovered，andtheearlystepsintheircultivation。
Mancouldnothaveknown，whenhefirstdomesticatedananimalorplant，whetheritwouldflourishandmultiplywhentransportedtoothercountries，thereforehecouldnothavebeenthusinfluencedinhischoice。Weseethatthecloseadaptationofthereindeerandcameltoextremelycoldandhotcountrieshasnotpreventedtheirdomestication。Stilllesscouldmanhaveforeseenwhetherhisanimalsandplantswouldvaryinsucceedinggenerationsandthusgivebirthtonewraces;andthesmallcapacityofvariabilityinthegoosehasnotpreventeditsdomesticationfromaremoteepoch。