A
rainfall
data
analysis
for
the
archeological
drawing
of
the
Augustan
aqueduct
route
Francesco
De
Paola
a,
Maurizio
Giugni
a,
Maria
Emanuela
Cornacchia
b,
Giovanni
Libralato
c,
Giusy
Lofrano
a,d,e,∗aDepartmentofcivil,architecturalandenvironmentalengineering,universityofNaples“FedericoII”,21,viaClaudio,80125Naples,Italy bTBCGénérateurd’innovation,25,boulevardVictor-Hugo,31770Colomiers,France
cDepartmentofenvironmentalsciences,informaticsandstatistics,universityCàFoscariVenice,2737/b,campodellaCelestia,30122Venice,Italy dDepartmentofchemistryandbiology,universityofSalerno,132,viaGiovanniPaoloII,84084Fisciano(Sa),Italy
eInstituteofmethodologiesforenvironmentalanalysis,nationalresearchcouncil(CNRIMAA),C.daS.LojaZ.I.TitoScalo,85050Potenza,Italy
a
r
t
i
c
l
e
i
n
f
o
Articlehistory:Received24September2015 Accepted25January2016 Availableonline12February2016 Keywords: Cumae Naples Romanbath Augustanaqueduct Watersupply
a
b
s
t
r
a
c
t
WasCumaesuppliedbythecollectionofrainwater,widelypracticedforlongtimeintheMediterranean area,orbyabranchoftheAugustanAqueduct?ThemaingoalofAquaAugustawastoprovidewaterto Puteoli(civilian)andMisenum(military)thatweretwoofthemainharborsoftheEmpire.However,the ruinsofthebranchthatwouldhaveflowedthroughCumaehavenotbeenexcavatedyet.Theaqueduct structurehasnotbeenstudiedindetailduetothedifficultyininspectingandthemissingarches.A rainfalldataanalysiswascarriedouttoassesstheflowanduseofthewaterconveyedbytheAqua AugustaaqueducttoCumae.ResultsindicatedthatrainwaterwasnotsufficienttosupplyCumaeandthe AquaAugustashouldhaveplayedagreatroleindeliveringwatertothecity.
©2016ElsevierMassonSAS.Allrightsreserved.
1. Introduction
Traditionally, the archaeology of water has been formerly focusedontheaestheticandarchitectural features ofhydraulic worksratherthanontechnicalandmechanicalaspects[1].Over thecenturies,civilizationsdevelopedstructurestobringdrinking watertothepopulationcenters,especiallythroughtheuseof aque-ducts.Therearemanyhistoricalrecordsonthepurposeanddesign ofaqueducts.Manyoftheseaqueducts,thoughthousandsofyears old,arestillstanding[2].Althoughthecrucialimportanceofwater iswellknown,watersupplyhasnotbeenconsideredasparticularly interesting[3–5].Thegreatbridgesconstituteanexceptionlikethe longarcadesofRomanaqueducts,which continuetoenrichthe landscapeofcountrysideandcitiesallaroundEurope(e.g. Pont-du-Gard,France). Surprisingly, engineering and aestheticissues were alwayseffectively conjugated. Whenever possible Roman engineersfollowed,thesteadydownhillcourseat/orbelowground levelforconstructingaqueducts[6–9].Rome’saqueduct system
∗ Correspondingauthor.Tel.:+393397117349;Fax:+390815938936. E-mailaddresses:[email protected](F.DePaola),[email protected] (M.E. Cornacchia), [email protected] (G. Libralato), [email protected], [email protected](G.Lofrano).
wasundergroundbyapproximately87%[3].Manyofthesesections weredeeptunneledthroughlimestoneandtufarock.Otherswere excavatedintothehillsidewithanengineeredrockand mortar tunnelsetintotheexcavationandthenbackfilledwiththenative soil[10–12].Thus,thestudyoftheirroutesisnotaneasytaskto carryout,asinthecaseofthedouble-branchedAugustan Aque-duct(AquaAugusta)builtinCampaniaRegionbetween20and30 AD[13].TheformerbranchgushedfromthespringofSerinolocated intheTerminio-TuoroMountaintakingwateruptoBenevento.The otherbranchoriginatedat“Acquaro-Pelosi”,stillfromSerinospring, feedingMisenum[11].Theformerandalmostentirelyunderground branchwas103kmlonghavingmanysecondarybranchesbeing 60kmlong.AsectionofAquaAugusta,thelongestaqueductofthe RomanEmpire[12],wasbroughttolightinFebruary 2012bya landslideoccurredafteraheavysnowfall[14].Thefirsthistorical surveyofitsopen-airrouteandthedeep-tunneledsectionswas carriedoutbyAbate[15]inordertoassessthepossibilityofusing theancienttunnelsforanewaqueductsupplyingtheCityofNaples. Nowadays,thestudyoftheoriginalrouteoftheaqueductcannot beconsideredconcluded.Unfortunately,theAquaAugustahasnot beensufficientlyinvestigatedforthefollowingreasons:
• thestructureismostlyundergroundandinspectionsaredifficult; • itisnotconsideredofarchitecturalinterest.
http://dx.doi.org/10.1016/j.culher.2016.01.006
396 F.DePaolaetal./JournalofCulturalHeritage19(2016)395–401
Fig.1.LocationofCumaealongtherouteoftheAugustanaqueductfromSerinotoMiseno. AdoptedfromDeFeoetal.[31].
Aninscription,attestingitsrepairatthetimeofConstantinethe Great(324/326AD),indicatedthattheaqueductservedthecities ofPuteoli,Neapolis,Nola,Atella,Cumae,Acerrae,BaiaMisenumand thecoastalcitiesofVesuvius(PompeiandHercolanum).Theruinsof thisbranchflowingthroughCumaehavenotbeenexcavatedyet, sothereisnorealproofofitsroute[11].
Cumae(Cumaarcheologicalsite, No.40.848356,E14.054078) islocatedonthenorthernedgeofCampiFlegrei(Naples, Campa-niaRegion,Italy),facingsouthandparalleltotheDomitiancoast (Fig.1).
ThesettlementwasestablishedbyGreekcolonistsintheVIII CenturyBCE,sackedbyOscansintheVCenturyBCEand incorpo-ratedintotheRomanEmpireintheIVCenturyBCE[16].Nowadays, themostcelebratedsiteatCumaeistheSybil’sCave,the Archaeo-logicalParkhouses,aseriesofancientruinsandartefacts,including twoRomanbathsandvarioustunnelsandcisterns.Bathswere pop-ularplacesofmassresortand thefirstsanitary facilities,being standardequipmentinmodernhouses,buttheywereoftennot includedinancient romandwellings.ThemainscopeofRoman aqueducts was to fed baths continually [3]. Manderscheid [7] assumedthatsincethethermalbathswerelocatedoutsidecities, accomodationsforhousingandfeedingpatientsmusthavebeen furnished.Accordingtohim,sincetherelativetemperaturesofthe springsvaryconsiderablybetween26◦Cand70◦C,theremusthave beensomeprovisionsforcoolingthewaterdowntoatemperature whichwasbeneficialorperhapscomfortabletothehumanbody. Indeed,inonlyaveryfewcaseswouldthewaterhave automati-callycooledoffsufficientlyonitswayfromthesourcetothebathing pools.Thearcheologyevidencerevealstwopossibilities:
• aninstallationofcoolingintowhichthewaterwouldflowbefore beingintroducedintothepiscinae;
• themixingofhotthermalwaterwithfreshwaterwhichhave mustbeentransportedseparatelyoverarelativelylongdistance bymeansofanaqueduct.
Theuseofcisternswasverycommoninromanbaths[17].For instance,intheBathsofCaracalla(Rome),waterwassuppliedby theAquaNovaAntoninianaandAquaMarciaaqueductsandlocal springsandstoredin18 cisternsuringbathtime,peoplegathered dailytheirsociallifeinaninformalcontextdrinkingwineand hav-ingsexualrelationships.Despitetheeventualmoraldisapproval, thepopularityofbathsenduredforoveramillenniumandspread allovertheRomantime[18,19].
Theaimofthispaperistoinvestigatethehypothesisthat rain-waterwasnottheonlywatersourceforCumae.Astatisticalanalysis ofrainfalldatawascarriedoutovera2000-yearperiodestimating thewaterdemandinCumae.Theanalysisconsideredthatbetween
I-IICenturiesADCumaeexperiencedanincreasedwaterdemand due to a great economic, culturaland social growth.Based on thereliefoftheCumaeForumbathoperatedbytheauthors,two fundamentalhypothesisaboutCumaewatersupplysystemwere assessed:
• thecollectionofrainwater,whichhadbeenpracticedforlong timeintheMediterraneanarea[19–21];
• thehistorical recordsof a branch of theAugustan Aqueduct, whichwouldhavereachedthecity[22].
2. Methodologicalapproach
Inordertodeterminewhethertheforumcisternwassupplied withrainwater,theaverageannualrainfallinCumaewasinferred intheI CenturyADusinghydrologicaldatafromthe1916–1999 annalsassumingtemporal recurringrainfall events.De Martino etal.[23]suppliedfurtherdatafortheperiod2000–2006.Thereare nodataavailablebefore1916.Therainfallstationschosenforthe analysiswerelocatedalongtheborderofCumaethatisPozzuoli, NaplesCapodimonte(Observatory),Camaldoli(Hermitage),Ischia (Forio)orIschia(Porto)andLicola.Rainfalleventsrangedbetween 408mm(1957)and1383mm(1979).
Duetothecomplexityofthemonthlyrainfallanalysisandlack ofhistorical data,rainfallestimation wasworkedoutusingthe valueoftheaveragerainfallsinferredbyanexponential smooth-ingstatespace modelstarting fromtheaverageannualrainfall between1916and2006.Thismodelusedathree-characterstring identifyingtheframework terminology of Hyndmanand Khan-dakar[24] throughR package forecast.By usingZZZ asdefault values,themodelautomaticallyselectedthebestmodel(Additive orMultiplicative)fortheerror,trendandseasonalityoftherainfall series.Twoautomaticforecastingalgorithmswereimplemented intheforecastpackage forR[24].Thefirstoneisbasedonthe innovationstatespacemodelsthatunderlieexponential smooth-ingmethods.Thesecondisastep-wiseforecastingalgorithmwith Autoregressiveintegratedmovingaverage(ARIMA)modelthatisa generalizationofanautoregressivemovingaverage(ARMA)model. Thesemodelsarefittedtotimeseriesdataeithertobetter under-standthedataortopredictfuturepointsintheseries(forecasting). Thealgorithmsareapplicabletobothseasonalandnon-seasonal data.Inordertoobtainarobustandwidelyapplicableautomatic forecastingalgorithm,wefollowedthefollowingsteps:
• applicationofallmodelsforeachseriesoptimizingthe parame-ters(bothsmoothingparametersandtheinitialstatevariable)of themodelineachcase;
Fig.2.Topographicsurveyofforumbath.
• selectionofthebestmodelaccordingtotheAkaike’sinformation criterion,ameasureoftherelativequalityofstatisticalmodels foragivensetofdata;
• pointforecastingusingthebestmodel(withoptimized parame-ters)forasmanystepsaheadasrequired;
• obtainthepredictionintervalsforthebestmodeleither simulat-ingfuturesamplepathsusingthenormaldistributionforerrors (parametricbootstrapthatisthepracticeofestimating proper-tiesofanestimator(suchasitsvariance)bymeasuringthose propertieswhensamplingfromanapproximatingdistribution).
3. Resultsanddiscussion
3.1. Routeassessment 3.1.1. Forumbathdescription
Cumae hosted the Forum bath and the smaller Republican Romanbath,known astheCentral Romanbathor, improperly, named the“Tombof theSibyl” [16]. A topographicalsurvey of theForum bath isshown in Fig. 2.Its construction dated back toatime ofintensebuildingactivity,whichoccurredinCumae somedecades afterthe construction of viaDomitiana (95 AD). Atthattime,theCentralRomanbathwasalreadypresentinthe city centre.The construction ofthe Forum bathsuggested that thepopulationincreasedaswellas thevisitorsfor commercial purposesmainlyduetothepresenceoftheportarealinkedtovia
Domitiana,allowingafastconnectiontoRome,withouttheneed toreachviaAppia[21].
TheForumbath,whichremainsareshowninFig.3a,waslocated inthelowerpartofthecity.Itfacedwest,mostlikelytotakethe advantageofabetterinsolationatallhours[16].Thetwopublic entrancesareshowninFig.2:
• theeastentrancewaslocatednearbytheintersectionwiththe roadtowardstheForum,along thenorthernsideofthe Capi-tolium;
• thevestibuleentranceleadtothefrigidariumpassingthrougha sceniccolonnade(Fig.3b).
Thefrigidariumwascharacterisedbythepresenceofcoldpools. Largewindowsallowednaturallighting.Ontherightandleftsides ofthevestibule,twopublicspaceswerepresentservingasdressing roomsandforadditionalactivitiessuchasmassagesand anoint-ingbeforeenteringthebath.Southward,asmallroomservedas acorridorpreventingheatlossaswellashotrooms.Thepoolsfor immersionshowedinFig.4wereonthenorth,southandwestsides oftheserooms.
Thebathcontainedfivepools:
• two poolsinthefrigidariumlabelled D1 (31 m2,water depth
1.4m,volume43.4m3)andD
2(21m2,waterdepth1.4m,volume
29.4m3)accessiblethroughashortstaircase(Fig.4);
Fig.3. (a)Remainsofforumbath;(b)eastentranceofbathandfrigidarium. Picturestakenbytheauthors.
398 F.DePaolaetal./JournalofCulturalHeritage19(2016)395–401
Fig.4. Poolsbelongingtothefrigidarium–D1andD2fromFig.1. Picturestakenbytheauthors.
Fig.5.(a)Tubuli;(b)Suspensurae. Picturestakenbytheauthors.
• threeinsidethecalidariumidentifiedI3 (12.5 m2,waterdepth
1.4m,volume17.5m3),I
4(14.5 m2,waterdepth1.4m,volume
20.3m3),I
5(13.5m2,waterdepth1.4m,volume18.9m3).
TheorderoftheroomsfollowedtheusualsequenceofRoman baths:tepidarium(mediumtemperature,Fig.2[F]),sudatio(fordry heatbath,Fig.1[G]),andcalidarium(hightemperature,Fig.2[H]). Theheatproducedbywoodcombustionwasconveyedfromthe praefurnium(Fig.2[L])throughouttheabove-mentionedseriesof roomsthankstothetubuli,whichwasasystemofcavitiesalong thewallsmadeofterracottaorbrickpipes(Fig.5a).Thedecking, lowerthanthatoftheadjacentrooms,wasatthesamelevelof thepraefurnium,withsomeopeningsatthesamelevelcalledthe suspensurae,whichwerepartofthefloorheatingsystem(Fig.5b). Theywerecoatedwitha15cmthicklayerofwaterproofplaster madeoflimemortarmixedwithsandorpozzolanaandshardsof brick[16].
Acistern,placedonapedestalinthenorthwestcornerofthe bath,suppliedwater.Thepedestalwaslocatedabout2 moverthe baseofthepoolstofacilitatethedistributionofwater.Thecistern, originallydividedintofourtanks,wasorientatedinawayslightly differentfromthatofthemaincomplex.Fromtheruins’ investiga-tion,thecisternwas15mlong,13mwidthand4.5mdepth(Fig.6). Itsperimeterwallhadathicknessgreaterthanthatoftherestof theRomanbath(minimumthicknessof1m),whilethepartition wallsbetweentankswerethinner.Thewallswereprimarily com-posedofopusreticulatum,interruptedathalfheightbyeightrows ofopuslatericium(Fig.6),coveredwithathicklayerofover15cm thickimpermeableplaster.Theplasterformedthepiercapsatthe cornersbetweenthewallandthefloor,accordingtoatechnical solutioncommontoboththeancientcisternsandmoderntanks.
Thecisternof thebathintheForumoperatedasa castellum aquae,sinceitwaspositionedabout2mhigherthanthebottomof thepools.Thenetareaofthecisternwasabout150m2;assuming
aninteriorwaterlevelof3.50m,itcouldhavecontainedabout525 m3ofwater.Theassumptionofconsideringaninteriorwaterdepth
of3.5mfromthetotalheightofthecistern(4.5m)wasadopted onthebaseofestimatedliteraturevalues[25].Thetotalareaand volumeof thepoolswere92.5 m2 and129.5 m3,respectively.
Therefore,thecisternvolumewasaboutfourtimesgreaterthan theamountofwaternecessarytofillthepools.
3.1.2. Rainfalldataanalysis
Resultsfromtherainfallinferenceanalysissuggestedthat: • generally,arainyyearfollowedadrierone;
• annualvaluesfluctuatedinaperiodicwaywithanaverageof 825mmofrainfall(Fig.7).
Fig.6.Cisternfeedingthepools. Picturestakenbytheauthors(NinFig.1).
Fig.7.CumulativeannualrainfallsinCumaeregion;0-1916generatedwithastatisticalanalysis,inthelastperiod(1916–2006),theconfidencelevelsof80and90%are reportedwiththeobserveddata[24].
Consideringthe hypothesisof rainwateruseand takinginto accountanaverageannualrainfallof825mm,thetotal average watervolumethatcouldhavebeenpotentiallyaccumulatedper yearinthecisternwas:
V=Sc•hm=150•0.825=123.75 m3 (1)
whereScisthenetareaofthecisternandhmtheavergeofrainfall.
Themaximumwatervolumewas207.48m3(withanincreaseof
about67%!)consideringtherainiestyear.
Thus, these volumeswere insufficienteven to fill thepools onceperyear.Thestatisticalhydrologicalanalysisofrainfalldata overaround2000 years (Fig.8)showedthereconstructeddata between0and1916withaforecastconfidencelevelof80%and 90%(Fig.6)harvestingrainfalldatawithin1916–2006(Fig.8).The maximumforecastedvalueofthecumulativeannualprecipitation wasapproximately1,800mm;therefore,nomorethanonetankof 270m3/yearcouldhavebeenfilled.Althoughroofrun-offcouldbe
conveyedtothetank,itscontributionwasnegligiblecomparedto thetotalvolume.
Asaconsequence,therainfallanalysisconfutedthe hypothe-sisoftheexclusiveuseofrainwater,suggestingthattheAugustan AqueductsuppliedCumae.
3.2. Flowassessment
The number of Cumae inhabitants during the first and sec-ondcenturiesADisunknowntohistoriansorarchaeologists.The availablerecordsdidnotprovideaccuratedataforthehistorical referenceperiodandmostofthetownwerenotexcavated.Thus, acalculationofthedailyflowbasedontheavailablewatersupply percapitaandthenumberofinhabitantsisnotpossible.
DataderivedfromthecharacteristicsofRomanbathandthe volumeofcisternswasusedtodeterminebothbaths’water con-sumptionand thepotentialdailynumber ofvisitors [30].From thesevalues,aroughestimationofthenumberofpeoplepresentin Cumaeandtheflowdeliveredfromtheaqueducttothecityresulted feasible.
Assumingthatthebathswereopenuntilsunset,adaily func-tioning of 12h is plausible.The total submersion time of each
400 F.DePaolaetal./JournalofCulturalHeritage19(2016)395–401
personintothepoolsisestimatedtobeanhour(frigidariumand calidarium).Ifeachpersonoccupiedanaveragesurfaceareaof0.36 m2(assumingasquarewithsidesof0.60m),increasingthevalueof
30%,eachpersoncouldoccupyaminimumareaof0.47m2inside
thepools.Sincethetotalareaofthepoolsis92.5m2,amaximum
of198peoplecouldbesimultaneouslypresent.
Assumingthat160l(0.16m3)ofwaterwereusedperdayfor
personalcare,thenecessaryexchangeofwater everyhour, cor-respondingto31.62 m3,wasobtainedbymultiplyingthewater
volumenecessaryforeachperson/bath(160 l/d)bythenumber ofusersthatcouldhavebeensimultaneouslypresentinthepools oftheForumofCumae.Multiplyingthenumberofpeoplepresent atthesametimewithin12hofbathoperation,thetotalnumber ofuserswasequalto2372.Thedailywaterflownecessaryforthe Romanbathfunction,inferredfromthesevalues,correspondedto 379.49m3/d,i.e.4.39l/s.ThecisternoftheForumbath,therefore,
withavolumeof525m3ofwater,wassufficienttosupplywater
forabout1 day.Basedontheaforementioneddata,Cumaecould hostmorethan2000inhabitants.Thebathscouldattenda max-imumof2372 usersperday,whichwaspartlycomposedofthe visitorsparticipatingincitytrade.
Thepercapitawatersupplywasevaluatedconsideringtwo ref-erencevalues:
• thevalueofquinaria;
• thepercapitawatersupplyofPompeii[11].
AccordingtoDiFenizio[26]andRodgers[27],aquinaria cor-respondedtoaflow rateof 0.48 l/s.Pompeiihad about12,000 inhabitantsandthedaily flowrate(calculatedaccordingtothe new conduits) wasequal to 6460 m3/d.The per capita water
supply of Pompeii correspondingto 0.54 m3/dper person was
obtainedbydividingthedailyflowratebythenumberof inhabi-tants[25].
Thedifferencebetweenthetwovaluesisundoubtedlydueto thefact that the quinaria takes intoaccount only the personal consumptionofwaterwhereasthevalueforthecityofPompeii wascalculatedastheratio betweentheoverallscopes andthe number of inhabitants – including the water used for domes-tic purposes,for publicpurposes and fordecorative ponds and fountains.
Ithasbeenestimatedthat10,000sailorsand10,000 civilians constitutedthepopulationofMisenum.These20,000peoplewould havehadthesamepercapitawatersupplycalculatedforPompeii, i.e.0.54 m3/dperperson.Thedailyflowshouldhavebeen10,767
m3/d,i.e.125l/s.Thisflowofwaterwouldhaverequiredmorethan
adaytoreplenishthePiscinaMirabilis.
Anotherknown datum is that the total populationof Nola, Acerra,Atella,Naples,Pozzuoli,Cumae,BaiaandMisenum,inthefirst centuryAD,exceeded200,000people[22],excludingPompeiiand Herculaneum,which,atthetime,hadrespectively12,000and5000 inhabitants[28].
ThetotalpopulationservedbySerinoAqueductwouldhavethus metorexceeded217,000 people.Althoughtheothercitieswere notallsimilartoPompeii,ourstudyconsideredthatthepercapita watersupplywasofabout0.54m3/(dperperson).
Anidenticalpercapitawatersupplywasprovidedviathe Augus-tanAqueductbyatotalflowrateof1352l/s.Thecurrentdischarge of“Acquaro-Pelosi”sourceis900l/s[29].Thetwovaluesare well-matchingsincethesourcehasbeeninuseforover2000years.
Assumingthat duringthefirstand secondcenturiesAD,the ancientcityofCumaehadmorethan2000inhabitantswithaper capitawatersupplycorrespondingto0.54m3/d,thewaterflowof
thebranchaqueductdirectedtoCumaefromthemainstretchof theSerinoaqueductcouldbeestimatedat1076.7m3/d.
4. Conclusions
Ahydrologicalanalysiswascarriedoutinordertoexactlyassess route,flowandusesofthewaterconveyedbytheAquaAugusta throughthecityofCumae.
Accordingtothis analysis,thehistorical recordsregarding a branchoftheAugustanAqueduct,whichwouldhavereachedthe cityofCumae,wereconfirmed.Thewaterflowofthebranch aque-ductdirectedtoCumaefromthemainstretchoftheSerinoaqueduct wasestimatedcorrespondingto12 l/s.
Acknowledgments
AuthorsacknowledgethePONAquasystemproject(Gestione IntegratadelCiclodelleAcquefinalizzataall’UsoSostenibiledelle Risorse,all’OttimizzazioneEnergetica,alMonitoraggioeControllo dellaQualitàdell’AcquaneiSistemiAcquedottisticienelleretidi DrenaggioUrbano)fundedbyPONR&C2007–2013,Smartcities andcommunitiesandsocialinnovation.
References
[1]G.Lofrano,J.Brown,Wastewatermanagementthroughtheages:ahistoryof humankind,Sci.TotalEnviron.408(2010)5254–5264.
[2]G.Lofrano,G.DeFeo,J.Brown,Waterpathwaysthroughtheages:Fromearly aqueductstonextgenerationofwastewatertreatmentplants.Advancesin watertreatmentandpollutionprevention,Springer,2012,pp.37–54. [3]A.T.Hodge,Romanaqueducts&watersupply,BristolClassicalPress,2002. [4]J.Brown,G.Lofrano,Overviewofwastewatermanagementthroughtheages
Worldenvironmentalandwaterresourcescongress2015:floods,droughts, andecosystems,in:Proceedingsofthe2015WorldEnvironmentalandWater ResourcesCongress,2015,pp.125–135.
[5]K. Kollyropoulos, G. Antoniou, I. Kalavrouziotis, J. Krasilnikoff, D. Kout-soyiannis, A.Angelakis,Hydrauliccharacteristics ofthedrainagesystems ofancienthellenictheatres:casestudyofthetheatreofdionysusandits implications,J.Irrig.DrainEng.(2015),10.1061/(ASCE)IR.1943-4774.0000906, 04015018.
[6]A.O.Koloski-Ostrow,N.DeHaan,G.DeKleijn,S.Piras,WaterintheRoman town:newresearchfromCuraAquarumandtheFrontinusSociety,J.Roman Archaeol.10(1997)181–191.
[7]H.Manderscheid,ThewatermanagementofGreekandRomanbaths,in:O. Wikander(Ed.),Handbookofancientwatertechnology,Leiden,Boston,and Köln,2000,pp.467–535.
[8]R.D.Hansen,WaterandwastewatersystemsinImperialRome,WaterRes.Bull. Am.WaterRes.Assoc.19(2)(1983).
[9]Y.Gorokhovich,A.Alexopoulos,V.Gikas,P.Gikas,A.N.Angelakis,Water sup-plyanduseintheRomancityofAptera,Hellas:themysteryoftheancient watersystem,in:Proc.ofthe3rdIWAInternationalsymposiumonwaterand wastewatertechnologiesinancientcivilizations,March22–25,2012,Istanbul, Turkey,2012,pp.380–386.
[10]R.Coates-Stephens,ThewallsandaqueductsofRomeintheEarlyMiddleAges, AD500-1000,J.RomanStud.88(1998)166–178.
[11]G.DeFeo,R.M.Napoli,HistoricaldevelopmentoftheAugustanAqueductin SouthernItaly:20centuriesofworksfromSerinotoNaples,WaterSci.Technol. WaterSupply7(1)(2007)131–138.
[12]G.Libertini,B.Miccio,N.Leone,G.DeFeo,TheAugustanaqueductinthe con-textofroadsystemandurbanizationoftheservedterritoryinSouthernItaly, in:ProceedingsoftheIWAregionalsymposiumonwater,wastewaterand environment:traditionandculture,22–24March2014,Patrass(Greece),2014. [13]F.DePaola,M.Giugni,M.E.Cornacchia,G.Lofrano,Thewatersupplyofancient Cumae:thecasestudyoftheForumbath,in:ProceedingsoftheIWAregional symposiumonwater,wastewaterandenvironment:traditionandculture, 22-24March,Patrass(Greece),2014.
[14]W.F.Lorenz,G.DeFeo,E.Baros,S.DeGisi,AncientPompeiiwatersupply: sources,routes,hydraulicsoftheaqueducts,WrightPaleohydrological Insti-tute,2012.
[15]B.Miccio,U.Potenza,GliacquedottidiNapoli,AziendaMunicipalizzata Acque-dottodiNapoli,Milan,Italy,1994.
[16]G.Bodon,I.Riera,P.Zanoviello,Utilitasnecessaria.Sistemiidraulicinell’Italia romana,Ed.ProgettoQuartaDimensione,1994.
[17]L.Mays,G.P.Antoniou,A.N.Angelakis,Historyofwatercisterns:legaciesand lessons,Water5(2013)1916–1940.
[18]F.Abbate,Primistudisull’acquedottoClaudio:rapportoalsignorsindacodi Napoli,Stamperiadell’Iride,Naples,Italy,1862.
[19]G.G.Fagan,HygienicconditionsinRomanpublicbaths,in:G.C.M.Jansen(Ed.), CuraAquaruminSicilia,Leiden,2000,pp.281–288.
[20]P.Caputo,R.Morichi,R.Paone,P.Rispoli,Cumaeilsuoparcoarcheologico.Un territorioelesuetestimonianze,BardiEditore,Rome(Italy),1996.
[21]G.G.Fagan,BathinginpublicintheRomanworld,UniversityofMichiganPress, 2002.