BookI
　　1
　　WEhavealreadydiscussedthefirstcausesofnature，andallnaturalmotion，alsothestarsorderedinthemotionoftheheavens，andthephysicalelement-enumeratingandspecifyingthemandshowinghowtheychangeintooneanother-andbecomingandperishingingeneral。Thereremainsforconsiderationapartofthisinquirywhichallourpredecessorscalledmeteorology。Itisconcernedwitheventsthatarenatural，thoughtheirorderislessperfectthanthatofthefirstoftheelementsofbodies。Theytakeplaceintheregionnearesttothemotionofthestars。Sucharethemilkyway，andcomets，andthemovementsofmeteors。Itstudiesalsoalltheaffectionswemaycallcommontoairandwater，andthekindsandpartsoftheearthandtheaffectionsofitsparts。Thesethrowlightonthecausesofwindsandearthquakesandalltheconsequencesthemotionsofthesekindsandpartsinvolve。Ofthesethingssomepuzzleus，whileothersadmitofexplanationinsomedegree。Further，theinquiryisconcernedwiththefallingofthunderboltsandwithwhirlwindsandfire-winds，andfurther，therecurrentaffectionsproducedinthesesamebodiesbyconcretion。Whentheinquiryintothesemattersisconcludedletusconsiderwhataccountwecangive，inaccordancewiththemethodwehavefollowed，ofanimalsandplants，bothgenerallyandindetail。Whenthathasbeendonewemaysaythatthewholeofouroriginalundertakingwillhavebeencarriedout。
　　Afterthisintroductionletusbeginbydiscussingourimmediatesubject。
　　2
　　Wehavealreadylaiddownthatthereisonephysicalelementwhichmakesupthesystemofthebodiesthatmoveinacircle，andbesidesthisfourbodiesowingtheirexistencetothefourprinciples，themotionoftheselatterbodiesbeingoftwokinds：eitherfromthecentreortothecentre。Thesefourbodiesarefire，air，water，earth。Fireoccupiesthehighestplaceamongthemall，earththelowest，andtwoelementscorrespondtotheseintheirrelationtooneanother，airbeingnearesttofire，watertoearth。Thewholeworldsurroundingtheearth，then，theaffectionsofwhichareoursubject，ismadeupofthesebodies。Thisworldnecessarilyhasacertaincontinuitywiththeuppermotions：consequentlyallitspowerandorderisderivedfromthem。（Fortheoriginatingprincipleofallmotionisthefirstcause。Besides，thatclementiseternalanditsmotionhasnolimitinspace，butisalwayscomplete;whereasalltheseotherbodieshaveseparateregionswhichlimitoneanother。）
　　Sowemusttreatfireandearthandtheelementslikethemasthematerialcausesoftheeventsinthisworld（meaningbymaterialwhatissubjectandisaffected），butmustassigncausalityinthesenseoftheoriginatingprincipleofmotiontotheinfluenceoftheeternallymovingbodies。
　　3
　　Letusfirstrecallouroriginalprinciplesandthedistinctionsalreadydrawnandthenexplainthe’milkyway’andcometsandtheotherphenomenaakintothese。
　　Fire，air，water，earth，weassert，originatefromoneanother，andeachofthemexistspotentiallyineach，asallthingsdothatcanberesolvedintoacommonandultimatesubstrate。
　　Thefirstdifficultyisraisedbywhatiscalledtheair。Whatarewetotakeitsnaturetobeintheworldsurroundingtheearth？Andwhatisitspositionrelativelytotheotherphysicalelements。（Forthereisnoquestionastotherelationofthebulkoftheearthtothesizeofthebodieswhichexistaroundit，sinceastronomicaldemonstrationshavebythistimeprovedtousthatitisactuallyfarsmallerthansomeindividualstars。Asforthewater，itisnotobservedtoexistcollectivelyandseparately，norcanitdosoapartfromthatvolumeofitwhichhasitsseatabouttheearth：thesea，thatis，andrivers，whichwecansee，andanysubterraneanwaterthatmaybehiddenfromourobservation。）Thequestionisreallyaboutthatwhichliesbetweentheearthandtheneareststars。Arewetoconsiderittobeonekindofbodyormorethanone？Andifmorethanone，howmanyarethereandwhataretheboundsoftheirregions？
　　Wehavealreadydescribedandcharacterizedthefirstelement，andexplainedthatthewholeworldoftheuppermotionsisfullofthatbody。
　　Thisisanopinionwearenotaloneinholding：itappearstobeanoldassumptionandonewhichmenhaveheldinthepast，forthewordetherhaslongbeenusedtodenotethatelement。Anaxagoras，itistrue，seemstometothinkthatthewordmeansthesameasfire。
　　Forhethoughtthattheupperregionswerefulloffire，andthatmenreferredtothoseregionswhentheyspokeofether。Inthelatterpointhewasright，formenseemtohaveassumedthatabodythatwaseternallyinmotionwasalsodivineinnature;and，assuchabodywasdifferentfromanyoftheterrestrialelements，theydeterminedtocallit’ether’。
　　Fortheumopinionsappearincyclesamongmennotoncenortwice，butinfinitelyoften。
　　Nowtherearesomewhomaintainthatnotonlythebodiesinmotionbutthatwhichcontainsthemispurefire，andtheintervalbetweentheearthandthestarsair：butiftheyhadconsideredwhatisnowsatisfactorilyestablishedbymathematics，theymighthavegivenupthispuerileopinion。Foritisaltogetherchildishtosupposethatthemovingbodiesareallofthemofasmallsize，becausetheysotous，lookingatthemfromtheearth。
　　Thisamatterwhichwehavealreadydiscussedinourtreatmentoftheupperregion，butwemayreturntothepointnow。
　　Iftheintervalswerefulloffireandthebodiesconsistedoffireeveryoneoftheotherelementswouldlongagohavevanished。
　　However，theycannotsimplybesaidtobefullofaireither;forevenifthereweretwoelementstofillthespacebetweentheearthandtheheavens，theairwouldfarexceedthequantiturequiredtomaintainitsproperproportiontotheotherelements。Forthebulkoftheearth（whichincludesthewholevolumeofwater）isinfinitesimalincomparisonwiththewholeworldthatsurroundsit。
　　Nowwefindthattheexcessinvolumeisnotproportionatelygreatwherewaterdissolvesintoairorairintofire。Whereastheproportionbetweenanygivensmallquantityofwaterandtheairthatisgeneratedfromitoughttoholdgoodbetweenthetotalamountofairandthetotalamountofwater。Nordoesitmakeanydifferenceifanyonedeniesthattheelementsoriginatefromoneanother，butassertsthattheyareequalinpower。Foronthisviewitiscertainamountsofeachthatareequalinpower，justaswouldbethecaseiftheyactuallyoriginatedfromoneanother。
　　Soitisclearthatneitherairnorfirealonefillstheintermediatespace。
　　Itremainstoexplain，afterapreliminarydiscussionofdifficulties，therelationofthetwoelementsairandfiretothepositionofthefirstelement，andthereasonwhythestarsintheupperregionimpartheattotheearthanditsneighbourhood。Letusfirsttreatoftheair，asweproposed，andthengoontothesequestions。
　　Sincewaterisgeneratedfromair，andairfromwater，whyarecloudsnotformedintheupperair？Theyoughttoformtherethemore，thefurtherfromtheearthandthecolderthatregionis。Foritisneitherappreciablyneartotheheatofthestars，nortotheraysrelectedfromtheearth。Itisthesethatdissolveanyformationbytheirheatandsopreventcloudsfromformingneartheearth。Forcloudsgatheratthepointwherethereflectedraysdisperseintheinfinityofspaceandarelost。Toexplainthiswemustsupposeeitherthatitisnotallairwhichwaterisgenerated，or，ifitisproducedfromallairalike，thatwhatimmediatelysurroundstheearthisnotmereair，butasortofvapour，andthatitsvaporousnatureisthereasonwhyitcondensesbacktowateragain。Butifthewholeofthatvastregionisvapour，theamountofairandofwaterwillbedisproportionatelygreat。Forthespacesleftbytheheavenlybodiesmustbefilledbysomeelement。Thiscannotbefire，forthenalltherestwouldhavebeendriedup。Consequently，whatfillsitmustbeairandthewaterthatsurroundsthewholeearth-vapourbeingwaterdissolved。
　　Afterthisexpositionofthedifficultiesinvolved，letusgoontolaydownthetruth，withaviewatoncetowhatfollowsandtowhathasalreadybeensaid。Theupperregionasfarasthemoonweaffirmtoconsistofabodydistinctbothfromfireandfromair，butvaryingdegreeofpurityandinkind，especiallytowardsitslimitonthesideoftheair，andoftheworldsurroundingtheearth。Nowthecircularmotionofthefirstelementandofthebodiesitcontainsdissolves，andinflamesbyitsmotion，whateverpartofthelowerworldisnearesttoit，andsogeneratesheat。Fromanotherpointofviewwemaylookatthemotionasfollows。Thebodythatliesbelowthecircularmotionoftheheavensis，inasort，matter，andispotentiallyhot，cold，dry，moist，andpossessedofwhateverotherqualitiesarederivedfromthese。Butitactuallyacquiresorretainsoneoftheseinvirtueofmotionorrest，thecauseandprincipleofwhichhasalreadybeenexplained。Soatthecentreandrounditwegetearthandwater，theheaviestandcoldestelements，bythemselves;roundthemandcontiguouswiththem，airandwhatwecommonlycallfire。Itisnotreallyfire，forfireisanexcessofheatandasortofebullition;butinreality，ofwhatwecallair，thepartsurroundingtheearthismoistandwarm，becauseitcontainsbothvapourandadryexhalationfromtheearth。Butthenextpart，abovethat，iswarmanddry。Forvapourisnaturallymoistandcold，buttheexhalationwarmanddry;andvapourispotentiallylikewater，theexhalationpotentiallylikefire。Sowemusttakethereasonwhycloudsarenotformedintheupperregiontobethis：
　　thatitisfillednotwithmereairbutratherwithasortoffire。
　　However，itmaywellbethattheformationofcloudsinthatupperregionisalsopreventedbythecircularmotion。Fortheairroundtheearthisnecessarilyallofitinmotion，exceptthatwhichiscutoffinsidethecircumferencewhichmakestheearthacompletesphere。Inthecaseofwindsitisactuallyobservablethattheyoriginateinmarshydistrictsoftheearth;andtheydonotseemtoblowabovethelevelofthehighestmountains。Itistherevolutionoftheheavenwhichcarriestheairwithitandcausesitscircularmotion，firebeingcontinuouswiththeupperelementandairwithfire。Thusitsmotionisasecondreasonwhythatairisnotcondensedintowater。
　　Butwheneveraparticleofairgrowsheavy，thewarmthinitissqueezedoutintotheupperregionanditsinks，andotherparticlesinturnarecarrieduptogetherwiththefieryexhalation。Thustheoneregionisalwaysfullofairandtheotheroffire，andeachofthemisperpetuallyinastateofchange。
　　Somuchtoexplainwhycloudsarenotformedandwhytheairisnotcondensedintowater，andwhataccountmustbegivenofthespacebetweenthestarsandtheearth，andwhatisthebodythatfillsit。
　　Asfortheheatderivedfromthesun，therightplaceforaspecialandscientificaccountofitisinthetreatiseaboutsense，sinceheatisanaffectionofsense，butwemaynowexplainhowitcanbeproducedbytheheavenlybodieswhicharenotthemselveshot。
　　Weseethatmotionisabletodissolveandinflametheair;
　　indeed，movingbodiesareoftenactuallyfoundtomelt。Nowthesun’smotionaloneissufficienttoaccountfortheoriginofterrestrialwarmthandheat。Foramotionthatistohavethiseffectmustberapidandnear，andthatofthestarsisrapidbutdistant，whilethatofthemoonisnearbutslow，whereasthesun’smotioncombinesbothconditionsinasufficientdegree。Thatmostheatshouldbegeneratedwherethesunispresentiseasytounderstandifweconsidertheanalogyofterrestrialphenomena，forhere，too，itistheairthatisnearesttoathinginrapidmotionwhichisheatedmost。Thisisjustwhatweshouldexpect，asitisthenearestairthatismostdissolvedbythemotionofasolidbody。
　　Thisthenisonereasonwhyheatreachesourworld。Anotheristhatthefiresurroundingtheairisoftenscatteredbythemotionoftheheavensanddrivendownwardsinspiteofitself。
　　Shooting-starsfurthersuffixtoprovethatthecelestialsphereisnothotorfiery：fortheydonotoccurinthatupperregionbutbelow：yetthemoreandthefasterathingmoves，themoreaptitistotakefire。Besides，thesun，whichmostofallthestarsisconsideredtobehot，isreallywhiteandnotfieryincolour。
　　4
　　Havingdeterminedtheseprinciplesletusexplainthecauseoftheappearanceintheskyofburningflamesandofshooting-stars，andof’torches’，and’goats’，assomepeoplecallthem。Allthesephenomenaareoneandthesamething，andareduetothesamecause，thedifferencebetweenthembeingoneofdegree。
　　Theexplanationoftheseandmanyotherphenomenaisthis。Whenthesunwarmstheearththeevaporationwhichtakesplaceisnecessarilyoftwokinds，notofoneonlyassomethink。Onekindisratherofthenatureofvapour，theotherofthenatureofawindyexhalation。Thatwhichrisesfromthemoisturecontainedintheearthandonitssurfaceisvapour，whilethatrisingfromtheearthitself，whichisdry，islikesmoke。Ofthesethewindyexhalation，beingwarm，risesabovethemoistervapour，whichisheavyandsinksbelowtheother。Hencetheworldsurroundingtheearthisorderedasfollows。Firstbelowthecircularmotioncomesthewarmanddryelement，whichwecallfire，forthereisnowordfullyadequatetoeverystateofthefumidevaporation：butwemustusethisterminologysincethiselementisthemostinflammableofallbodies。Belowthiscomesair。Wemustthinkofwhatwejustcalledfireasbeingspreadroundtheterrestrialsphereontheoutsidelikeakindoffuel，sothatalittlemotionoftenmakesitburstintoflamejustassmokedoes：forflameistheebullitionofadryexhalation。Sowheneverthecircularmotionstirsthisstuffupinanyway，itcatchesfireatthepointatwhichitismostinflammable。Theresultdiffersaccordingtothedispositionandquantityofthecombustiblematerial。Ifthisisbroadandlong，weoftenseeaflameburningasinafieldofstubble：
　　ifitburnslengthwiseonly，weseewhatarecalled’torches’and’goats’andshooting-stars。Nowwhentheinflammablematerialislongerthanitisbroadsometimesitseemstothrowoffsparksasitburns。（Thishappensbecausemattercatchesfireatthesidesinsmallportionsbutcontinuouslywiththemainbody。）Thenitiscalleda’goat’。Whenthisdoesnothappenitisa’torch’。Butifthewholelengthoftheexhalationisscatteredinsmallpartsandinmanydirectionsandinbreadthanddepthalike，wegetwhatarecalledshooting-stars。
　　Thecauseoftheseshooting-starsissometimesthemotionwhichignitestheexhalation。Atothertimestheairiscondensedbycoldandsqueezesoutandejectsthehotelement;makingtheirmotionlookmorelikethatofathingthrownthanlikearunningfire。Forthequestionmightberaisedwhetherthe’shooting’ofa’star’isthesamethingaswhenyouputanexhalationbelowalampanditlightsthelowerlampfromtheflameabove。Forheretootheflamepasseswonderfullyquicklyandlookslikeathingthrown，andnotasifonethingafteranothercaughtfire。Orisa’star’whenit’shoots’asinglebodythatisthrown？Apparentlybothcasesoccur：sometimesitisliketheflamefromthelampandsometimesbodiesareprojectedbybeingsqueezedout（likefruitstonesfromone’sfingers）
　　andsoareseentofallintotheseaandonthedryland，bothbynightandbydaywhentheskyisclear。Theyarethrowndownwardsbecausethecondensationwhichpropelstheminclinesdownwards。
　　Thunderboltsfalldownwardsforthesamereason：theiroriginisnevercombustionbutejectionunderpressure，sincenaturallyallheattendsupwards。
　　Whenthephenomenonisformedintheupperregionitisduetothecombustionoftheexhalation。Whenittakesplaceatalowerlevelitisduetotheejectionoftheexhalationbythecondensingandcoolingofthemoisterevaporation：forthislatterasitcondensesandinclinesdownwardcontracts，andthrustsoutthehotelementandcausesittobethrowndownwards。Themotionisupwardsordownwardsorsidewaysaccordingtothewayinwhichtheevaporationlies，anditsdispositioninrespectofbreadthanddepth。Inmostcasesthedirectionissidewaysbecausetwomotionsareinvolved，acompulsorymotiondownwardsandanaturalmotionupwards，andunderthesecircumstancesanobjectalwaysmovesobliquely。Hencethemotionof’shooting-stars’isgenerallyoblique。
　　Sothematerialcauseofallthesephenomenaistheexhalation，theefficientcausesometimestheuppermotion，sometimesthecontractionandcondensationoftheair。Further，allthesethingshappenbelowthemoon。Thisisshownbytheirapparentspeed，whichisequaltothatofthingsthrownbyus;foritisbecausetheyareclosetous，thattheselatterseemfartoexceedinspeedthestars，thesun，andthemoon。
　　5
　　Sometimesonafinenightweseeavarietyofappearancesthatforminthesky：’chasms’forinstanceand’trenches’andblood-redcolours。These，too，havethesamecause。Forwehaveseenthattheupperaircondensesintoaninflammableconditionandthatthecombustionsometimestakesontheappearanceofaburningflame，sometimesthatofmovingtorchesandstars。Soitisnotsurprisingthatthissameairwhencondensingshouldassumeavarietyofcolours。
　　Foraweaklightshiningthroughadenseair，andtheairwhenitactsasamirror，willcauseallkindsofcolourstoappear，butespeciallycrimsonandpurple。Forthesecoloursgenerallyappearwhenfire-colourandwhitearecombinedbysuperposition。Thusonahotday，orthroughasmoky，medium，thestarswhentheyriseandsetlookcrimson。Thelightwillalsocreatecoloursbyreflectionwhenthemirrorissuchastoreflectcolouronlyandnotshape。
　　Theseappearancesdonotpersistlong，becausethecondensationoftheairistransient。
　　’Chasms’gettheirappearanceofdepthfromlightbreakingoutofadarkblueorblackmassofair。Whentheprocessofcondensationgoesfurtherinsuchacaseweoftenfind’torches’ejected。Whenthe’chasm’contractsitpresentstheappearanceofa’trench’。
　　Ingeneral，whiteincontrastwithblackcreatesavarietyofcolours;likeflame，forinstance，throughamediumofsmoke。Butbydaythesunobscuresthem，and，withtheexceptionofcrimson，thecoloursarenotseenatnightbecausetheyaredark。
　　Thesethenmustbetakentobethecausesof’shooting-stars’andthephenomenaofcombustionandalsooftheothertransientappearancesofthiskind。
　　6
　　Letusgoontoexplainthenatureofcometsandthe’milkyway’，afterapreliminarydiscussionoftheviewsofothers。
　　AnaxagorasandDemocritusdeclarethatcometsareaconjunctionoftheplanetsapproachingoneanotherandsoappearingtotouchoneanother。
　　SomeoftheItalianscalledPythagoreanssaythatthecometisoneoftheplanets，butthatitappearsatgreatintervalsoftimeandonlyrisesalittleabovethehorizon。ThisisthecasewithMercurytoo;becauseitonlyrisesalittleabovethehorizonitoftenfailstobeseenandconsequentlyappearsatgreatintervalsoftime。
　　AviewliketheirswasalsoexpressedbyHippocratesofChiosandhispupilAeschylus。Onlytheysaythatthetaildoesnotbelongtothecometiself，butisoccasionallyassumedbyitonitscourseincertainsituations，whenoursightisreflectedtothesunfromthemoistureattractedbythecomet。Itappearsatgreaterintervalsthantheotherstarsbecauseitisslowesttogetclearofthesunandhasbeenleftbehindbythesuntotheextentofthewholeofitscirclebeforeitreappearsatthesamepoint。Itgetsclearofthesunbothtowardsthenorthandtowardsthesouth。Inthespacebetweenthetropicsitdoesnotdrawwatertoitselfbecausethatregionisdriedupbythesunonitscourse。Whenitmovestowardsthesouthithasnolackofthenecessarymoisture，butbecausethesegmentofitscirclewhichisabovethehorizonissmall，andthatbelowitmanytimesaslarge，itisimpossibleforthesuntobereflectedtooursight，eitherwhenitapproachesthesoutherntropic，oratthesummersolstice。Henceintheseregionsitdoesnotdevelopatailatall。
　　Butwhenitisvisibleinthenorthitassumesatailbecausethearcabovethehorizonislargeandthatbelowitsmall。Forunderthesecircumstancesthereisnothingtopreventourvisionfrombeingreflectedtothesun。
　　Theseviewsinvolveimpossibilities，someofwhicharecommontoallofthem，whileothersarepeculiartosomeonly。
　　Thisisthecase，first，withthosewhosaythatthecometisoneoftheplanets。Foralltheplanetsappearinthecircleofthezodiac，whereasmanycometshavebeenseenoutsidethatcircle。Againmorecometsthanonehaveoftenappearedsimultaneously。Besides，iftheirtailisduetoreflection，asAeschylusandHippocratessay，thisplanetoughtsometimestobevisiblewithoutatailsince，astheyitdoesnotpossessatailineveryplaceinwhichitappears。
　　But，asamatteroffact，noplanethasbeenobservedbesidesthefive。Andallofthemareoftenvisibleabovethehorizontogetheratthesametime。Further，cometsareoftenfoundtoappear，aswellwhenalltheplanetsarevisibleaswhensomearenot，butareobscuredbytheneighbourhoodofthesun。Moreoverthestatementthatacometonlyappearsinthenorth，withthesunatthesummersolstice，isnottrueeither。ThegreatcometwhichappearedatthetimeoftheearthquakeinAchaeaandthetidalwaveroseduewest;andmanyhavebeenknowntoappearinthesouth。AgaininthearchonshipofEuclees，sonofMolon，atAthensthereappearedacometinthenorthinthemonthGamelion，thesunbeingaboutthewintersolstice。Yettheythemselvesadmitthatreflectionoversogreataspaceisanimpossibility。
　　Anobjectionthattellsequallyagainstthosewhoholdthistheoryandthosewhosaythatcometsareacoalescenceoftheplanetsis，first，thefactthatsomeofthefixedstarstoogetatail。ForthiswemustnotonlyaccepttheauthorityoftheEgyptianswhoassertit，butwehaveourselvesobservedthefact。ForastarinthethighoftheDoghadatail，thoughafaintone。Ifyoufixedyoursightonititslightwasdim，butifyoujustglancedatit，itappearedbrighter。Besides，allthecometsthathavebeenseeninourdayhavevanishedwithoutsetting，graduallyfadingawayabovethehorizon;andtheyhavenotleftbehindthemeitheroneormorestars。Forinstancethegreatcometwementionedbeforeappearedtothewestinwinterinfrostyweatherwhentheskywasclear，inthearchonshipofAsteius。Onthefirstdayitsetbeforethesunandwasthennotseen。Onthenextdayitwasseen，beingeversolittlebehindthesunandimmediatelysetting。Butitslightextendedoverathirdpartoftheskylikealeap，sothatpeoplecalledita’path’。ThiscometrecededasfarasOrion’sbeltandtheredissolved。
　　Democritushowever，insistsuponthetruthofhisviewandaffirmsthatcertainstarshavebeenseenwhencometsdissolve。Butonhistheorythisoughtnottooccuroccasionallybutalways。Besides，theEgyptiansaffirmthatconjunctionsoftheplanetswithoneanother，andwiththefixedstars，takeplace，andwehaveourselvesobservedJupitercoincidingwithoneofthestarsintheTwinsandhidingit，andyetnocometwasformed。Further，wecanalsogivearationalproofofourpoint。Itistruethatsomestarsseemtobebiggerthanothers，yeteachonebyitselflooksindivisible。Consequently，justas，iftheyreallyhadbeenindivisible，theirconjunctioncouldnothavecreatedanygreatermagnitude，sonowthattheyarenotinfactindivisiblebutlookasiftheywere，theirconjunctionwillnotmakethemlookanybigger。
　　Enoughhasbeensaid，withoutfurtherargument，toshowthatthecausesbroughtforwardtoexplaincometsarefalse。
　　7
　　Weconsiderasatisfactoryexplanationofphenomenainaccessibletoobservationtohavebeengivenwhenouraccountofthemisfreefromimpossibilities。Theobservationsbeforeussuggestthefollowingaccountofthephenomenawearenowconsidering。Weknowthatthedryandwarmexhalationistheoutermostpartoftheterrestrialworldwhichfallsbelowthecircularmotion。It，andagreatpartoftheairthatiscontinuouswithitbelow，iscarriedroundtheearthbythemotionofthecircularrevolution。Inthecourseofthismotionitoftenigniteswhereveritmayhappentobeoftherightconsistency，andthiswemaintaintobethecauseofthe’shooting’ofscattered’stars’。Wemaysay，then，thatacometisformedwhentheuppermotionintroducesintoagatheringofthiskindafieryprinciplenotofsuchexcessivestrengthastoburnupmuchofthematerialquickly，norsoweakassoontobeextinguished，butstrongerandcapableofburningupmuchmaterial，andwhenexhalationoftherightconsistencyrisesfrombelowandmeetsit。Thekindofcometvariesaccordingtotheshapewhichtheexhalationhappenstotake。Ifitisdiffusedequallyoneverysidethestarissaidtobefringed，ifitstretchesoutinonedirectionitiscalledbearded。Wehaveseenthatwhenafieryprincipleofthiskindmovesweseemtohaveashooting-star：similarlywhenitstandsstillweseemtohaveastarstandingstill。Wemaycomparethesephenomenatoaheapormassofchaffintowhichatorchisthrust，orasparkthrown。Thatiswhatashooting-starislike。Thefuelissoinflammablethatthefirerunsthroughitquicklyinaline。Nowifthisfireweretopersistinsteadofrunningthroughthefuelandperishingaway，itscoursethroughthefuelwouldstopatthepointwherethelatterwasdensest，andthenthewholemightbegintomove。Suchisacomet-likeashooting-starthatcontainsitsbeginningandendinitself。
　　Whenthematterbeginstogatherinthelowerregionindependentlythecometappearsbyitself。Butwhentheexhalationisconstitutedbyoneofthefixedstarsortheplanets，owingtotheirmotion，oneofthembecomesacomet。Thefringeisnotclosetothestarsthemselves。
　　Justashaloesappeartofollowthesunandthemoonastheymove，andencirclethem，whentheairisdenseenoughforthemtoformalongunderthesun’scourse，sotoothefringe。Itstandsintherelationofahalotothestars，exceptthatthecolourofthehaloisduetoreflection，whereasinthecaseofcometsthecolourissomethingthatappearsactuallyonthem。
　　Nowwhenthismattergathersinrelationtoastarthecometnecessarilyappearstofollowthesamecourseasthestar。Butwhenthecometisformedindependentlyitfallsbehindthemotionoftheuniverse，liketherestoftheterrestrialworld。Itisthisfact，thatacometoftenformsindependently，indeedoftenerthanroundoneoftheregularstars，thatmakesitimpossibletomaintainthatacometisasortofreflection，notindeed，asHippocratesandhisschoolsay，tothesun，buttotheverystaritisallegedtoaccompany-infact，akindofhalointhepurefueloffire。
　　Asforthehaloweshallexplainitscauselater。
　　Thefactthatcometswhenfrequentforeshadowwindanddroughtmustbetakenasanindicationoftheirfieryconstitution。Fortheiroriginisplainlyduetotheplentifulsupplyofthatsecretion。
　　Hencetheairisnecessarilydrierandthemoistevaporationissodissolvedanddissipatedbythequantityofthehotexhalationasnotreadilytocondenseintowater-Butthisphenomenontooshallbeexplainedmoreclearlylaterwhenthetimecomestospeakofthewinds-Sowhentherearemanycometsandtheyaredense，itisaswesay，andtheyearsareclearlydryandwindy。Whentheyarefewerandfainterthiseffectdoesnotappearinthesamedegree，thoughasaruletheisfoundtobeexcessiveeitherindurationorstrength。
　　ForinstancewhenthestoneatAegospotamifelloutoftheair-ithadbeencarriedupbyawindandfelldowninthedaytime-thentooacomethappenedtohaveappearedinthewest。Andatthetimeofthegreatcometthewinterwasdryandnorthwindsprevailed，andthewavewasduetoanoppositionofwinds。Forinthegulfanorthwindblewandoutsideitaviolentsouthwind。AgaininthearchonshipofNicomachusacometappearedforafewdaysabouttheequinoctialcircle（thisonehadnotriseninthewest），andsimultaneouslywithittherehappenedthestormatCorinth。
　　Thattherearefewcometsandthattheyappearrarelyandoutsidethetropiccirclesmorethanwithinthemisduetothemotionofthesunandthestars。Forthismotiondoesnotonlycausethehotprincipletobesecretedbutalsodissolvesitwhenitisgathering。
　　Butthechiefreasonisthatmostofthisstuffcollectsintheregionofthemilkyway。
　　8
　　Letusnowexplaintheorigin，cause，andnatureofthemilkyway。
　　Andheretooletusbeginbydiscussingthestatementsofothersonthesubject。
　　（1）Oftheso-calledPythagoreanssomesaythatthisisthepathofoneofthestarsthatfellfromheavenatthetimeofPhaethon’sdownfall。Otherssaythatthesunusedoncetomoveinthiscircleandthatthisregionwasscorchedormetwithsomeotheraffectionofthiskind，becauseofthesunanditsmotion。
　　ButitisabsurdnottoseethatifthiswerethereasonthecircleoftheZodiacoughttobeaffectedinthesameway，andindeedmoresothanthatofthemilkyway，sincenotthesunonlybutalltheplanetsmoveinit。Wecanseethewholeofthiscircle（halfofitbeingvisibleatanytimeofthenight），butitshowsnosignsofanysuchaffectionexceptwhereapartofittouchesthecircleofthemilkyway。
　　（2）Anaxagoras，Democritus，andtheirschoolssaythatthemilkywayisthelightofcertainstars。For，theysay，whenthesunpassesbelowtheearthsomeofthestarsarehiddenfromit。Nowthelightofthoseonwhichthesunshinesisinvisible，beingobscuredbytheofthesun。Butthemilkywayisthepeculiarlightofthosestarswhichareshadedbytheearthfromthesun’srays。
　　This，too，isobviouslyimpossible。Themilkywayisalwaysunchangedandamongthesameconstellations（foritisclearlyagreatestcircle），whereas，sincethesundoesnotremaininthesameplace，whatishiddenfromitdiffersatdifferenttimes。Consequentlywiththechangeofthesun’spositionthemilkywayoughttochangeitspositiontoo：butwefindthatthisdoesnothappen。Besides，ifastronomicaldemonstrationsarecorrectandthesizeofthesunisgreaterthanthatoftheearthandthedistanceofthestarsfromtheearthmanytimesgreaterthanthatofthesun（justasthesunisfurtherfromtheearththanthemoon），thentheconemadebytheraysofthesunwouldterminateatnogreatdistancefromtheearth，andtheshadowoftheearth（whatwecallnight）wouldnotreachthestars。Onthecontrary，thesunshinesonallthestarsandtheearthscreensnoneofthem。
　　（3）Thereisathirdtheoryaboutthemilkyway。Somesaythatitisareflectionofoursighttothesun，justastheysaythatthecometis。
　　Butthistooisimpossible。Foriftheeyeandthemirrorandthewholeoftheobjectwereseverallyatrest，thenthesamepartoftheimagewouldappearatthesamepointinthemirror。Butifthemirrorandtheobjectmove，keepingthesamedistancefromtheeyewhichisatrest，butatdifferentratesofspeedandsonotalwaysatthesameintervalfromoneanother，thenitisimpossibleforthesameimagealwaystoappearinthesamepartofthemirror。Nowtheconstellationsincludedinthecircleofthemilkywaymove;andsodoesthesun，theobjecttowhichoursightisreflected;butwestandstill。Andthedistanceofthosetwofromusisconstantanduniform，buttheirdistancefromoneanothervaries。FortheDolphinsometimesrisesatmidnight，sometimesinthemorning。Butineachcasethesamepartsofthemilkywayarefoundnearit。Butifitwereareflectionandnotagenuineaffectionofthesethisoughtnottobethecase。
　　Again，wecanseethemilkywayreflectedatnightinwaterandsimilarmirrors。Butunderthesecircumstancesitisimpossibleforoursighttobereflectedtothesun。
　　Theseconsiderationsshowthatthemilkywayisnotthepathofoneoftheplanets，northelightofimperceptiblestars，norareflection。Andthosearethechieftheorieshandeddownbyothershitherto。
　　Letusrecallourfundamentalprincipleandthenexplainourviews。Wehavealreadylaiddownthattheoutermostpartofwhatiscalledtheairispotentiallyfireandthatthereforewhentheairisdissolvedbymotion，thereisseparatedoffakindofmatter-andofthismatterweassertthatcometsconsist。Wemustsupposethatwhathappensisthesameasinthecaseofthecometswhenthematterdoesnotformindependentlybutisformedbyoneofthefixedstarsortheplanets。Thenthesestarsappeartobefringed，becausematterofthiskindfollowstheircourse。Inthesameway，acertainkindofmatterfollowsthesun，andweexplainthehaloasareflectionfromitwhentheairisoftherightconstitution。Nowwemustassumethatwhathappensinthecaseofthestarsseverallyhappensinthecaseofthewholeoftheheavensandalltheuppermotion。Foritisnaturaltosupposethat，ifthemotionofasinglestarexcitesaflame，thatofallthestarsshouldhaveasimilarresult，andespeciallyinthatregioninwhichthestarsarebiggestandmostnumerousandnearesttooneanother。Nowthecircleofthezodiacdissolvesthiskindofmatterbecauseofthemotionofthesunandtheplanets，andforthisreasonmostcometsarefoundoutsidethetropiccircles。Again，nofringeappearsroundthesunormoon：fortheydissolvesuchmattertooquicklytoadmitofitsformation。Butthiscircleinwhichthemilkywayappearstooursightisthegreatestcircle，anditspositionissuchthatitextendsfaroutsidethetropiccircles。Besidestheregionisfullofthebiggestandbrightestconstellationsandalsoofwhatcalled’scattered’stars（youhaveonlytolooktoseethisclearly）。Soforthesereasonsallthismatteriscontinuallyandceaselesslycollectingthere。Aproofofthetheoryisthis：Inthecircleitselfthelightisstrongerinthathalfwherethemilkywayisdivided，andinittheconstellationsaremorenumerousandclosertooneanotherthanintheotherhalf;whichshowsthatthecauseofthelightisthemotionoftheconstellationsandnothingelse。Forifitisfoundinthecircleinwhichtherearemostconstellationsandatthatpointinthecircleatwhichtheyaredensestandcontainthebiggestandthemoststars，itisnaturaltosupposethattheyarethetruecauseoftheaffectioninquestion。Thecircleandtheconstellationsinitmaybeseeninthediagram。Theso-called’scattered’starsitisnotpossibletosetdowninthesamewayonthespherebecausenoneofthemhaveanevidentpermanentposition;
　　butifyoulookuptotheskythepointisclear。Forinthiscirclealonearetheintervalsfullofthesestars：intheothercirclesthereareobviousgaps。Henceifweacceptthecauseassignedfortheappearanceofcometsasplausiblewemustassumethatthesamekindofthingholdsgoodofthemilkyway。Forthefringewhichintheformercaseisanaffectionofasinglestarhereformsinthesamewayinrelationtoawholecircle。Soifwearetodefinethemilkywaywemaycallit’afringeattachingtothegreatestcircle，andduetothemattersecreted’。This，aswesaidbefore，explainswhytherearefewcometsandwhytheyappearrarely;itisbecauseateachrevolutionoftheheavensthismatterhasalwaysbeenandisalwaysbeingseparatedoffandgatheredintothisregion。
　　Wehavenowexplainedthephenomenathatoccurinthatpartoftheterrestrialworldwhichiscontinuouswiththemotionsoftheheavens，namely，shooting-starsandtheburningflame，cometsandthemilkyway，thesebeingthechiefaffectionsthatappearinthatregion。
　　9
　　Letusgoontotreatoftheregionwhichfollowsnextinorderafterthisandwhichimmediatelysurroundstheearth。Itistheregioncommontowaterandair，andtheprocessesattendingtheformationofwaterabovetakeplaceinit。Wemustconsidertheprinciplesandcausesofallthesephenomenatooasbefore。Theefficientandchiefandfirstcauseisthecircleinwhichthesunmoves。Forthesunasitapproachesorrecedes，obviouslycausesdissipationandcondensationandsogivesrisetogenerationanddestruction。Nowtheearthremainsbutthemoisturesurroundingitismadetoevaporatebythesun’sraysandtheotherheatfromabove，andrises。Butwhentheheatwhichwasraisingitleavesit，inpartdispersingtothehigherregion，inpartquenchedthroughrisingsofarintotheupperair，thenthevapourcoolsbecauseitsheatisgoneandbecausetheplaceiscold，andcondensesagainandturnsfromairintowater。
　　Andafterthewaterhasformeditfallsdownagaintotheearth。
　　Theexhalationofwaterisvapour：aircondensingintowateriscloud。Mistiswhatisleftoverwhenacloudcondensesintowater，andisthereforeratherasignoffineweatherthanofrain;formistmightbecalledabarrencloud。Sowegetacircularprocessthatfollowsthecourseofthesun。Foraccordingasthesunmovestothissideorthat，themoistureinthisprocessrisesorfalls。Wemustthinkofitasariverflowingupanddowninacircleandmadeuppartlyofair，partlyofwater。Whenthesunisnear，thestreamofvapourflowsupwards;whenitrecedes，thestreamofwaterflowsdown：
　　andtheorderofsequence，atallevents，inthisprocessalwaysremainsthesame。Soif’Oceanus’hadsomesecretmeaninginearlywriters，perhapstheymayhavemeantthisriverthatflowsinacircleabouttheearth。
　　Sothemoistureisalwaysraisedbytheheatanddescendstotheearthagainwhenitgetscold。Theseprocessesand，insomecases，theirvarietiesaredistinguishedbyspecialnames。Whenthewaterfallsinsmalldropsitiscalledadrizzle;whenthedropsarelargeritisrain。
　　10
　　Someofthevapourthatisformedbydaydoesnotrisehighbecausetheratioofthefirethatisraisingittothewaterthatisbeingraisedissmall。Whenthiscoolsanddescendsatnightitiscalleddewandhoar-frost。Whenthevapourisfrozenbeforeithascondensedtowateragainitishoar-frost;andthisappearsinwinterandiscommonerincoldplaces。Itisdewwhenthevapourhascondensedintowaterandtheheatisnotsogreatastodryupthemoisturethathasbeenraisednorthecoldsufficient（owingtothewarmthoftheclimateorseason）forthevapouritselftofreeze。
　　Fordewismorecommonlyfoundwhentheseasonortheplaceiswarm，whereastheopposite，ashasbeensaid，isthecasewithhoar-frost。
　　Forobviouslyvapouriswarmerthanwater，havingstillthefirethatraisedit：consequentlymorecoldisneededtofreezeit。
　　Bothdewandhoar-frostarefoundwhentheskyisclearandthereisnowind。Forthevapourcouldnotberaisedunlesstheskywereclear，andifawindwereblowingitcouldnotcondense。
　　Thefactthathoar-frostisnotfoundonmountainscontributestoprovethatthesephenomenaoccurbecausethevapourdoesnotrisehigh。Onereasonforthisisthatitrisesfromhollowandwateryplaces，sothattheheatthatisraisingit，bearingasitweretooheavyaburdencannotliftittoagreatheightbutsoonletsitfallagain。Asecondreasonisthatthemotionoftheairismorepronouncedataheight，andthisdissolvesagatheringofthiskind。
　　Everywhere，exceptinPontus，dewisfoundwithsouthwindsandnotwithnorthwinds。Theretheoppositeisthecaseanditisfoundwithnorthwindsandnotwithsouth。Thereasonisthesameasthatwhichexplainswhydewisfoundinwarmweatherandnotincold。Forthesouthwindbringswarm，andthenorth，wintryweather。Forthenorthwindiscoldandsoquenchestheheatoftheevaporation。ButinPontusthesouthwinddoesnotbringwarmthenoughtocauseevaporation，whereasthecoldnessofthenorthwindconcentratestheheatbyasortofrecoil，sothatthereismoreevaporationandnotless。Thisisathingwhichwecanoftenobserveinotherplacestoo。Wells，forinstance，giveoffmorevapourinanorththaninasouthwind。Onlythenorthwindsquenchtheheatbeforeanyconsiderablequantityofvapourhasgathered，whileinasouthwindtheevaporationisallowedtoaccumulate。
　　Water，onceformed，doesnotfreezeonthesurfaceoftheearth，inthewaythatitdoesintheregionoftheclouds。
　　11
　　Fromthelattertherefallthreebodiescondensedbycold，namelyrain，snow，hail。Twoofthesecorrespondtothephenomenaonthelowerlevelandareduetothesamecauses，differingfromthemonlyindegreeandquantity。
　　Snowandhoar-frostareoneandthesamething，andsoarerainanddew：onlythereisagreatdealoftheformerandlittleofthelatter。Forrainisduetothecoolingofagreatamountofvapour，fortheregionfromwhichandthetimeduringwhichthevapouriscollectedareconsiderable。Butofdewthereislittle：forthevapourcollectsforitinasingledayandfromasmallarea，asitsquickformationandscantyquantityshow。
　　Therelationofhoar-frostandsnowisthesame：whencloudfreezesthereissnow，whenvapourfreezesthereishoar-frost。
　　Hencesnowisasignofacoldseasonorcountry。Foragreatdealofheatisstillpresentandunlessthecoldwereoverpoweringitthecloudwouldnotfreeze。Fortherestillsurvivesinitagreatdealoftheheatwhichcausedthemoisturetoriseasvapourfromtheearth。
　　Hailontheotherhandisfoundintheupperregion，butthecorrespondingphenomenoninthevaporousregionneartheearthislacking。For，aswesaid，tosnowintheupperregioncorrespondshoar-frostinthelower，andtorainintheupperregion，dewinthelower。Butthereisnothingheretocorrespondtohailintheupperregion。Whythisissowillbeclearwhenwehaveexplainedthenatureofhail。
　　12
　　Butwemustgoontocollectthefactsbearingontheoriginofit，boththosewhichraisenodifficultiesandthosewhichseemparadoxical。
　　Hailisice，andwaterfreezesinwinter;yethailstormsoccurchieflyinspringandautumnandlessofteninthelatesummer，butrarelyinwinterandthenonlywhenthecoldislessintense。Andingeneralhailstormsoccurinwarmer，andsnowincolderplaces。
　　Again，thereisadifficultyaboutwaterfreezingintheupperregion。
　　Itcannothavefrozenbeforebecomingwater：andwatercannotremainsuspendedintheairforanyspaceoftime。Norcanwesaythatthecaseislikethatofparticlesofmoisturewhicharecarriedupowingtotheirsmallsizeandrestontheiar（thewaterswimmingontheairjustassmallparticlesofearthandgoldoftenswimonwater）。Inthatcaselargedropsareformedbytheunionofmanysmall，andsofalldown。Thiscannottakeplaceinthecaseofhail，sincesolidbodiescannotcoalescelikeliquidones。Clearlythendropsofthatsizeweresuspendedintheairorelsetheycouldnothavebeensolargewhenfrozen。
　　Somethinkthatthecauseandoriginofhailisthis。Thecloudisthrustupintotheupperatmosphere，whichiscolderbecausethereflectionofthesun’sraysfromtheearthceasesthere，anduponitsarrivaltherethewaterfreezes。Theythinkthatthisexplainswhyhailstormsarecommonerinsummerandinwarmcountries;theheatisgreateranditthruststhecloudsfurtherupfromtheearth。Butthefactisthathaildoesnotoccuratallatagreatheight：yetitoughttodoso，ontheirtheory，justasweseethatsnowfallsmostonhighmountains。Againcloudshaveoftenbeenobservedmovingwithagreatnoiseclosetotheearth，terrifyingthosewhoheardandsawthemasportentsofsomecatastrophe。Sometimes，too，whensuchcloudshavebeenseen，withoutanynoise，therefollowsaviolenthailstorm，andthestonesareofincrediblesize，andangularinshape。Thisshowsthattheyhavenotbeenfallingforlongandthattheywerefrozenneartotheearth，andnotasthattheorywouldhaveit。Moreover，wherethehailstonesarelarge，thecauseoftheirfreezingmustbepresentinthehighestdegree：forhailisiceaseveryonecansee。Nowthosehailstonesarelargewhichareangularinshape。Andthisshowsthattheyfrozeclosetotheearth，forthosethatfallfararewornawaybythelengthoftheirfallandbecomeroundandsmallerinsize。
　　Itclearlyfollowsthatthecongelationdoesnottakeplacebecausethecloudisthrustupintothecoldupperregion。
　　Nowweseethatwarmandcoldreactupononeanotherbyrecoil。
　　Henceinwarmweatherthelowerpartsoftheeartharecoldandinafrosttheyarewarm。Thesamething，wemustsuppose，happensintheair，sothatinthewarmerseasonsthecoldisconcentratedbythesurroundingheatandcausesthecloudtogooverintowatersuddenly。（Forthisreasonrain-dropsaremuchlargeronwarmdaysthaninwinter，andshowersmoreviolent。Ashowerissaidtobemoreviolentinproportionasthewatercomesdowninabody，andthishappenswhenthecondensationtakesplacequickly，-thoughthisisjusttheoppositeofwhatAnaxagorassays。Hesaysthatthishappenswhenthecloudhasrisenintothecoldair;whereaswesaythatithappenswhenthecloudhasdescendedintothewarmair，andthatthemorethefurtherthecloudhasdescended）。Butwhenthecoldhasbeenconcentratedwithinstillmorebytheouterheat，itfreezesthewaterithasformedandthereishail。Wegethailwhentheprocessoffreezingisquickerthanthedescentofthewater。Forifthewaterfallsinacertaintimeandthecoldissufficienttofreezeitinless，thereisnodifficultyaboutitshavingfrozenintheair，providedthatthefreezingtakesplaceinashortertimethanitsfall。Thenearertotheearth，andthemoresuddenly，thisprocesstakesplace，themoreviolentistherainthatresultsandthelargertheraindropsandthehailstonesbecauseoftheshortnessoftheirfall。Forthesamereasonlargeraindropsdonotfallthickly。
　　Hailisrarerinsummerthaninspringandautumn，thoughcommonerthaninwinter，becausetheairisdrierinsummer，whereasinspringitisstillmoist，andinautumnitisbeginningtogrowmoist。
　　Itisforthesamereasonthathailstormssometimesoccurinthelatesummeraswehavesaid。
　　Thefactthatthewaterhaspreviouslybeenwarmedcontributestoitsfreezingquickly：forsoitcoolssooner。Hencemanypeople，whentheywanttocoolhotwaterquickly，beginbyputtingitinthesun。SotheinhabitantsofPontuswhentheyencampontheicetofish（theycutaholeintheiceandthenfish）pourwarmwaterroundtheirreedsthatitmayfreezethequicker，fortheyusetheicelikeleadtofixthereeds。Nowitisinhotcountriesandseasonsthatthewaterwhichformssoongrowswarm。
　　ItisforthesamereasonthatrainfallsinsummerandnotinwinterinArabiaandEthiopiatoo，andthatintorrentsandrepeatedlyonthesameday。Fortheconcentrationorrecoilduetotheextremeheatofthecountrycoolsthecloudsquickly。
　　Somuchforanaccountofthenatureandcausesofrain，dew，snow，hoar-frost，andhail。
　　13
　　Letusexplainthenatureofwinds，andallwindyvapours，alsoofriversandofthesea。Buthere，too，wemustfirstdiscussthedifficultiesinvolved：for，asinothermatters，sointhisnotheoryhasbeenhandeddowntousthatthemostordinarymancouldnothavethoughtof。
　　Somesaythatwhatiscalledair，whenitisinmotionandflows，iswind，andthatthissameairwhenitcondensesagainbecomescloudandwater，implyingthatthenatureofwindandwateristhesame。Sotheydefinewindasamotionoftheair。Hencesome，wishingtosayacleverthing，assertthatallthewindsareonewind，becausetheairthatmovesisinfactallofitoneandthesame;theymaintainthatthewindsappeartodifferowingtotheregionfromwhichtheairmayhappentoflowoneachoccasion，butreallydonotdifferatall。Thisisjustlikethinkingthatallriversareoneandthesameriver，andtheordinaryunscientificviewisbetterthanascientifictheorylikethis。Ifallriversflowfromonesource，andthesameistrueinthecaseofthewinds，theremightbesometruthinthistheory;butifitisnomoretrueintheonecasethanintheother，thisingeniousideaisplainlyfalse。Whatrequiresinvestigationisthis：thenatureofwindandhowitoriginates，itsefficientcauseandwhencetheyderivetheirsource;whetheroneoughttothinkofthewindasissuingfromasortofvesselandflowinguntilthevesselisempty，asifletoutofawineskin，or，aspaintersrepresentthewinds，asdrawingtheirsourcefromthemselves。
　　Wefindanalogousviewsabouttheoriginofrivers。Itisthoughtthatthewaterisraisedbythesunanddescendsinrainandgathersbelowtheearthandsoflowsfromagreatreservoir，alltheriversfromone，oreachfromadifferentone。Nowateratallisgenerated，butthevolumeoftheriversconsistsofthewaterthatisgatheredintosuchreservoirsinwinter。Henceriversarealwaysfullerinwinterthaninsummer，andsomeareperennial，othersnot。
　　Riversareperennialwherethereservoirislargeandsoenoughwaterhascollectedinittolastoutandnotbeusedupbeforethewinterrainreturns。Wherethereservoirsaresmallerthereislesswaterintherivers，andtheyaredriedupandtheirvesselemptybeforethefreshraincomeson。
　　Butifanyonewillpicturetohimselfareservoiradequatetothewaterthatiscontinuouslyflowingdaybyday，andconsidertheamountofthewater，itisobviousthatareceptaclethatistocontainallthewaterthatflowsintheyearwouldbelargerthantheearth，or，atanyrate，notmuchsmaller。
　　Thoughitisevidentthatmanyreservoirsofthiskinddoexistinmanypartsoftheearth，yetitisunreasonableforanyonetorefusetoadmitthatairbecomeswaterintheearthforthesamereasonasitdoesaboveit。Ifthecoldcausesthevaporousairtocondenseintowaterabovetheearthwemustsupposethecoldintheearthtoproducethissameeffect，andrecognizethattherenotonlyexistsinitandflowsoutofitactuallyformedwater，butthatwateriscontinuallyforminginittoo。
　　Again，eveninthecaseofthewaterthatisnotbeingformedfromdaytodaybutexistsassuch，wemustnotsupposeassomedothatrivershavetheirsourceindefinitesubterraneanlakes。Onthecontrary，justasabovetheearthsmalldropsformandthesejoinothers，tillfinallythewaterdescendsinabodyasrain，sotoowemustsupposethatintheearththewateratfirsttricklestogetherlittlebylittle，andthatthesourcesoftheriversdrip，asitwere，outoftheearthandthenunite。Thisisprovedbyfacts。Whenmenconstructanaqueducttheycollectthewaterinpipesandtrenches，asiftheearthinthehighergroundweresweatingthewaterout。
　　Hence，too，thehead-watersofriversarefoundtoflowfrommountains，andfromthegreatestmountainsthereflowthemostnumerousandgreatestrivers。Again，mostspringsareintheneighbourhoodofmountainsandofhighground，whereasifweexceptrivers，waterrarelyappearsintheplains。Formountainsandhighground，suspendedoverthecountrylikeasaturatedsponge，makethewateroozeoutandtrickletogetherinminutequantitiesbutinmanyplaces。Theyreceiveagreatdealofwaterfallingasrain（foritmakesnodifferencewhetheraspongyreceptacleisconcaveandturneduporconvexandturneddown：ineithercaseitwillcontainthesamevolumeofmatter）and，theyalsocoolthevapourthatrisesandcondenseitbackintowater。
　　Hence，aswesaid，wefindthatthegreatestriversflowfromthegreatestmountains。Thiscanbeseenbylookingatitineraries：whatisrecordedinthemconsistseitherofthingswhichthewriterhasseenhimselforofsuchashehascompiledafterinquiryfromthosewhohaveseenthem。
　　InAsiawefindthatthemostnumerousandgreatestriversflowfromthemountaincalledParnassus，admittedlythegreatestofallmountainstowardsthesouth-east。Whenyouhavecrossedityouseetheouterocean，thefurtherlimitofwhichisunknowntothedwellersinourworld。BesidesotherriversthereflowfromittheBactrus，theChoaspes，theAraxes：fromthelastabranchseparatesoffandflowsintolakeMaeotisastheTanais。Fromit，too，flowstheIndus，thevolumeofwhosestreamisgreatestofallrivers。FromtheCaucasusflowsthePhasis，andverymanyothergreatriversbesides。NowtheCaucasusisthegreatestofthemountainsthatlietothenortheast，bothasregardsitsextentanditsheight。Aproofofitsheightisthefactthatitcanbeseenfromtheso-called’deeps’andfromtheentrancetothelake。Again，thesunshinesonitspeaksforathirdpartofthenightbeforesunriseandagainaftersunset。Itsextentisprovedbythefactthatthoughtcontainsmanyinhabitableregionswhichareoccupiedbymanynationsandinwhichtherearesaidtobegreatlakes，yettheysaythatalltheseregionsarevisibleuptothelastpeak。FromPyrene（thisisamountaintowardsthewestinCeltice）thereflowtheIstrusandtheTartessus。Thelatterflowsoutsidethepillars，whiletheIstrusflowsthroughallEuropeintotheEuxine。MostoftheremainingriversflownorthwardsfromtheHercynianmountains，whicharethegreatestinheightandextentaboutthatregion。Intheextremenorth，beyondfurthestScythia，arethemountainscalledRhipae。Thestoriesabouttheirsizearealtogethertoofabulous：however，theysaythatthemostand（aftertheIstrus）
　　thegreatestriversflowfromthem。So，too，inLibyathereflowfromtheAethiopianmountainstheAegonandtheNyses;andfromtheso-calledSilverMountainthetwogreatestofnamedrivers，therivercalledChremetesthatflowsintotheouterocean，andthemainsourceoftheNile。OftheriversintheGreekworld，theAchelousflowsfromPindus，theInachusfromthesamemountain;theStrymon，theNestus，andtheHebrusallthreefromScombrus;manyrivers，too，flowfromRhodope。
　　Allotherriverswouldbefoundtoflowinthesameway，butwehavementionedtheseasexamples。Evenwhereriversflowfrommarshes，themarshesinalmosteverycasearefoundtoliebelowmountainsorgraduallyrisingground。
　　Itisclearthenthatwemustnotsupposeriverstooriginatefromdefinitereservoirs：forthewholeearth，wemightalmostsay，wouldnotbesufficient（anymorethantheregionofthecloudswouldbe）ifweweretosupposethattheywerefedbyactuallyexistingwateronlyanditwerenotthecasethatassomewaterpassedoutofexistencesomemorecameintoexistence，butriversalwaysdrewtheirstreamfromanexistingstore。Secondly，thefactthatriversriseatthefootofmountainsprovesthataplacetransmitsthewateritcontainsbygradualpercolationofmanydrops，littlebylittle，andthatthisishowthesourcesofriversoriginate。However，thereisnothingimpossibleabouttheexistenceofsuchplacescontainingaquantityofwaterlikelakes：onlytheycannotbebigenoughtoproducethesupposedeffect。Tothinkthattheyareisjustasabsurdasifoneweretosupposethatriversdrewalltheirwaterfromthesourceswesee（formostriversdoflowfromsprings）。Soitisnomorereasonabletosupposethoselakestocontainthewholevolumeofwaterthanthesesprings。
　　Thatthereexistsuchchasmsandcavitiesintheearthwearetaughtbytheriversthatareswallowedup。Theyarefoundinmanypartsoftheearth：inthePeloponnesus，forinstance，therearemanysuchriversinArcadia。ThereasonisthatArcadiaismountainousandtherearenochannelsfromitsvalleystothesea。Sotheseplacesgetfullofwater，andthis，havingnooutlet，underthepressureofthewaterthatisaddedabove，findsawayoutforitselfunderground。
　　InGreecethiskindofthinghappensonquiteasmallscale，butthelakeatthefootoftheCaucasus，whichtheinhabitantsofthesepartscallasea，isconsiderable。ManygreatriversfallintoitandithasnovisibleoutletbutissuesbelowtheearthoffthelandoftheCoraxiabouttheso-called’deepsofPontus’。Thisisaplaceofunfathomabledepthinthesea：atanyratenoonehasyetbeenabletofindbottomtherebysounding。Atthisspot，aboutthreehundredstadiafromland，therecomesupsweetwateroveralargearea，notallofittogetherbutinthreeplaces。AndinLiguriaariverequalinsizetotheRhodanusisswallowedupandappearsagainelsewhere：
　　theRhodanusbeinganavigableriver。