Encirclingbarrier-reefsareofallsizes，fromthreemilestonolessthanforty-fourmilesindiameter；andthatwhichfrontsoneside，andencirclesbothends，ofNewCaledonia，is400mileslong。Eachreefincludesone，two，orseveralrockyislandsofvariousheights；andinoneinstance，evenasmanyastwelveseparateislands。Thereefrunsatagreaterorlessdistancefromtheincludedland；intheSocietyarchipelagogenerallyfromonetothreeorfourmiles；butatHogoleuthereefis20milesonthesouthernside，and14milesontheoppositeornorthernside，fromtheincludedislands。Thedepthwithinthelagoon-channelalsovariesmuch；
　　from10to30fathomsmaybetakenasanaverage；butatVanikorotherearespacesnolessthan56fathomsor363feetdeep。Internallythereefeitherslopesgentlyintothelagoon-channel，orendsinaperpendicularwallsometimesbetweentwoandthreehundredfeetunderwaterinheight：
　　externallythereefrises，likeanatoll，withextremeabruptnessoutoftheprofounddepthsoftheocean。
　　Whatcanbemoresingularthanthesestructures？Wesee[picture]
　　anisland，whichmaybecomparedtoacastlesituatedonthesummitofaloftysubmarinemountain，protectedbyagreatwallofcoral-rock，alwayssteepexternallyandsometimesinternally，withabroadlevelsummit，hereandtherebreachedbyanarrowgateway，throughwhichthelargestshipscanenterthewideanddeepencirclingmoat。
　　Asfarastheactualreefofcoralisconcerned，thereisnotthesmallestdifference，ingeneralsize，outline，grouping，andeveninquitetriflingdetailsofstructure，betweenabarrierandanatoll。ThegeographerBalbihaswellremarked，thatanencircledislandisanatollwithhighlandrisingoutofitslagoon；removethelandfromwithin，andaperfectatollisleft。
　　Butwhathascausedthesereefstospringupatsuchgreatdistancesfromtheshoresoftheincludedislands？Itcannotbethatthecoralswillnotgrowclosetotheland；fortheshoreswithinthelagoon-channel，whennotsurroundedbyalluvialsoil，areoftenfringedbylivingreefs；andweshallpresentlyseethatthereisawholeclass，whichIhavecalledFringingReefsfromtheircloseattachmenttotheshoresbothofcontinentsandofislands。Again，onwhathavethereef-buildingcorals，whichcannotliveatgreatdepths，basedtheirencirclingstructures？Thisisagreatapparentdifficulty，analogoustothatinthecaseofatolls，whichhasgenerallybeenoverlooked。Itwillbeperceivedmoreclearlybyinspectingthefollowingsectionswhicharerealones，takeninnorthandsouthlines，throughtheislandswiththeirbarrier-reefs，ofVanikoro，Gambier，andMaurua；andtheyarelaiddown，bothverticallyandhorizontally，onthesamescaleofaquarterofaninchtoamile。
　　Itshouldbeobservedthatthesectionsmighthavebeentakeninanydirectionthroughtheseislands，orthrough[picture]
　　manyotherencircledislands，andthegeneralfeatureswouldhavebeenthesame。Now，bearinginmindthatreef-buildingcoralcannotliveatagreaterdepththanfrom20to30fathoms，andthatthescaleissosmallthattheplummetsontherighthandshowadepthof200fathoms，onwhatarethesebarrier-reefsbased？Arewetosupposethateachislandissurroundedbyacollar-likesubmarineledgeofrock，orbyagreatbankofsediment，endingabruptlywherethereefends？
　　Iftheseahadformerlyeatendeeplyintotheislands，beforetheywereprotectedbythereefs，thushavingleftashallowledgeroundthemunderwater，thepresentshoreswouldhavebeeninvariablyboundedbygreatprecipices，butthisismostrarelythecase。Moreover，onthisnotion，itisnotpossibletoexplainwhythecoralsshouldhavesprungup，likeawall，fromtheextremeoutermarginoftheledge，oftenleavingabroadspaceofwaterwithin，toodeepforthegrowthofcorals。Theaccumulationofawidebankofsedimentallroundtheseislands，andgenerallywidestwheretheincludedislandsaresmallest，ishighlyimprobable，consideringtheirexposedpositionsinthecentralanddeepestpartsoftheocean。Inthecaseofthebarrier-reefofNewCaledonia，whichextendsfor150milesbeyondthenorthernpointoftheislands，inthesamestraightlinewithwhichitfrontsthewestcoast，itishardlypossibletobelievethatabankofsedimentcouldthushavebeenstraightlydepositedinfrontofaloftyisland，andsofarbeyonditsterminationintheopensea。Finally，ifwelooktootheroceanicislandsofaboutthesameheightandofsimilargeologicalconstitution，butnotencircledbycoral-reefs，wemayinvainsearchforsotriflingacircumambientdepthas30fathoms，exceptquiteneartotheirshores；forusuallylandthatrisesabruptlyoutofwater，asdomostoftheencircledandnon-encircledoceanicislands，plungesabruptlyunderit。Onwhatthen，Irepeat，arethesebarrierreefsbased？Why，withtheirwideanddeepmoat-likechannels，dotheystandsofarfromtheincludedland？Weshallsoonseehoweasilythesedifficultiesdisappear。
　　WecomenowtoourthirdclassofFringing-reefs，whichwillrequireaveryshortnotice。Wherethelandslopesabruptlyunderwater，thesereefsareonlyafewyardsinwidth，formingamereribbonorfringeroundtheshores：wherethelandslopesgentlyunderthewaterthereefextendsfurther，sometimesevenasmuchasamilefromtheland；butinsuchcasesthesoundingsoutsidethereefalwaysshowthatthesubmarineprolongationofthelandisgentlyinclined。Infact，thereefsextendonlytothatdistancefromtheshore，atwhichafoundationwithintherequisitedepthfrom20to30fathomsisfound。Asfarastheactualreefisconcerned，thereisnoessentialdifferencebetweenitandthatformingabarrieroranatoll：itis，however，generallyoflesswidth，andconsequentlyfewisletshavebeenformedonit。Fromthecoralsgrowingmorevigorouslyontheoutside，andfromthenoxiouseffectofthesedimentwashedinwards，theouteredgeofthereefisthehighestpart，andbetweenitandthelandthereisgenerallyashallowsandychannelafewfeetindepth。Wherebanksorsedimentshaveaccumulatedneartothesurface，asinpartsoftheWestIndies，theysometimesbecomefringedwithcorals，andhenceinsomedegreeresemblelagoon-islandsoratolls，inthesamemannerasfringing-reefs，surroundinggentlyslopingislands，insomedegreeresemblebarrier-reefs。
　　Notheoryontheformationofcoral-reefscanbeconsideredsatisfactorywhichdoesnotincludethethreegreat[picture]
　　classes。Wehaveseenthatwearedriventobelieveinthesubsidenceofthosevastareas，interspersedwithlowislands，ofwhichnotonerisesabovetheheighttowhichthewindandwavescanthrowupmatter，andyetareconstructedbyanimalsrequiringafoundation，andthatfoundationtolieatnogreatdepth。Letusthentakeanislandsurroundedbyfringing-reefs，whichoffernodifficultyintheirstructure；andletthisislandwithitsreefs，representedbytheunbrokenlinesinthewoodcut，slowlysubside。
　　Now，astheislandsinksdown，eitherafewfeetatatimeorquiteinsensibly，wemaysafelyinfer，fromwhatisknownoftheconditionsfavourabletothegrowthofcoral，thatthelivingmasses，bathedbythesurfonthemarginofthereef，willsoonregainthesurface。Thewater，however，willencroachlittlebylittleontheshore，theislandbecominglowerandsmaller，andthespacebetweentheinneredgeofthereefandthebeachproportionatelybroader。Asectionofthereefandislandinthisstate，afterasubsidenceofseveralhundredfeet，isgivenbythedottedlines。Coralisletsaresupposedtohavebeenformedonthereef；andashipisanchoredinthelagoon-channel。Thischannelwillbemoreorlessdeep，accordingtotherateofsubsidence，totheamountofsedimentaccumulatedinit，andtothegrowthofthedelicatelybranchedcoralswhichcanlivethere。Thesectioninthisstateresemblesineveryrespectonedrawnthroughanencircledisland：infact，itisarealsectiononthescaleof。517ofaninchtoamilethroughBolabolainthePacific。Wecannowatonceseewhyencirclingbarrier-reefsstandsofarfromtheshoreswhichtheyfront。
　　Wecanalsoperceive，thatalinedrawnperpendicularlydownfromtheouteredgeofthenewreef，tothefoundationofsolidrockbeneaththeoldfringing-reef，willexceedbyasmanyfeetastherehavebeenfeetofsubsidence，thatsmalllimitofdepthatwhichtheeffectivecoralscanlive：——thelittlearchitectshavingbuiltuptheirgreatwall-likemass，asthewholesankdown，uponabasisformedofothercoralsandtheirconsolidatedfragments。
　　Thusthedifficultyonthishead，whichappearedsogreat，disappears。
　　If，insteadofanisland，wehadtakentheshoreofacontinentfringedwithreefs，andhadimaginedittohavesubsided，agreatstraightbarrier，likethatofAustraliaorNewCaledonia，separatedfromthelandbyawideanddeepchannel，wouldevidentlyhavebeentheresult。
　　Letustakeournewencirclingbarrier-reef，ofwhichthesectionisnowrepresentedbyunbrokenlines，andwhich，asIhavesaid，isarealsectionthroughBolabola，andletitgoonsubsiding。Asthebarrier-reefslowlysinksdown，thecoralswillgoonvigorouslygrowingupwards；butastheislandsinks，thewaterwillgaininchbyinchontheshore——theseparatemountainsfirstformingseparateislandswithin[picture]
　　onegreatreef——andfinally，thelastandhighestpinnacledisappearing。
　　Theinstantthistakesplace，aperfectatollisformed：Ihavesaid，removethehighlandfromwithinanencirclingbarrier-reef，andanatollisleft，andthelandhasbeenremoved。Wecannowperceivehowitcomesthatatolls，havingsprungfromencirclingbarrier-reefs，resemblethemingeneralsize，form，inthemannerinwhichtheyaregroupedtogether，andintheirarrangementinsingleordoublelines；fortheymaybecalledrudeoutlinechartsofthesunkenislandsoverwhichtheystand。WecanfurtherseehowitarisesthattheatollsinthePacificandIndianOceansextendinlinesparalleltothegenerallyprevailingstrikeofthehighislandsandgreatcoast-linesofthoseoceans。Iventure，therefore，toaffirm，thatonthetheoryoftheupwardgrowthofthecoralsduringthesinkingoftheland，[13]alltheleadingfeaturesinthosewonderfulstructures，thelagoon-islandsoratolls，whichhavesolongexcitedtheattentionofvoyagers，aswellasinthenolesswonderfulbarrier-reefs，whetherencirclingsmallislandsorstretchingforhundredsofmilesalongtheshoresofacontinent，aresimplyexplained。
　　Itmaybeasked，whetherIcanofferanydirectevidenceofthesubsidenceofbarrier-reefsoratolls；butitmustbeborneinmindhowdifficultitmusteverbetodetectamovement，thetendencyofwhichistohideunderwaterthepartaffected。Nevertheless，atKeelingatollIobservedonallsidesofthelagoonoldcocoa-nuttreesunderminedandfalling；
　　andinoneplacethefoundation-postsofashed，whichtheinhabitantsassertedhadstoodsevenyearsbeforejustabovehigh-watermark，butnowwasdailywashedbyeverytide：oninquiryIfoundthatthreeearthquakes，oneofthemsevere，hadbeenfelthereduringthelasttenyears。AtVanikoro，thelagoon-channelisremarkablydeep，scarcelyanyalluvialsoilhasaccumulatedatthefootoftheloftyincludedmountains，andremarkablyfewisletshavebeenformedbytheheapingoffragmentsandsandonthewall-likebarrierreef；thesefacts，andsomeanalogousones，ledmetobelievethatthisislandmustlatelyhavesubsidedandthereefgrownupwards：hereagainearthquakesarefrequentandverysevere。IntheSocietyarchipelago，ontheotherhand，wherethelagoon-channelsarealmostchokedup，wheremuchlowalluviallandhasaccumulated，andwhereinsomecaseslongisletshavebeenformedonthebarrier-reefs——factsallshowingthattheislandshavenotverylatelysubsided——onlyfeebleshocksaremostrarelyfelt。
　　Inthesecoralformations，wherethelandandwaterseemstrugglingformastery，itmustbeeverdifficulttodecidebetweentheeffectsofachangeinthesetofthetidesandofaslightsubsidence：thatmanyofthesereefsandatollsaresubjecttochangesofsomekindiscertain；onsomeatollstheisletsappeartohaveincreasedgreatlywithinalateperiod；
　　onotherstheyhavebeenpartiallyorwhollywashedaway。TheinhabitantsofpartsoftheMaldivaarchipelagoknowthedateofthefirstformationofsomeislets；inotherparts，thecoralsarenowflourishingonwater-washedreefs，whereholesmadeforgravesattesttheformerexistenceofinhabitedland。Itisdifficulttobelieveinfrequentchangesinthetidalcurrentsofanopenocean；whereas，wehaveintheearthquakesrecordedbythenativesonsomeatolls，andinthegreatfissuresobservedonotheratolls，plainevidenceofchangesanddisturbancesinprogressinthesubterraneanregions。
　　Itisevident，onourtheory，thatcoastsmerelyfringedbyreefscannothavesubsidedtoanyperceptibleamount；andthereforetheymust，sincethegrowthoftheircorals，eitherhaveremainedstationaryorhavebeenupheaved。Now，itisremarkablehowgenerallyitcanbeshown，bythepresenceofupraisedorganicremains，thatthefringedislandshavebeenelevated：