No Evidence for Significant Age Spreads in Young Massive LMC Clusters

5.3 MB·English

by  Florian Niederhofer

Astronomy&Astrophysicsmanuscriptno.niederhofer (cid:13)c ESO2015
January13,2015
No Evidence for Significant Age Spreads in Young Massive LMC
Clusters(cid:63)
F.Niederhofer1,2,M.Hilker1,N.Bastian3,andE.Silva-Villa4,5
1 EuropeanSouthernObservatory,Karl-Schwarzschild-Straße2,D-85748GarchingbeiMu¨nchen,Germany
e-mail:[email protected]
2 Universita¨ts-SternwarteMu¨nchen,Scheinerstraße1,D-81679Mu¨nchen,Germany
3 AstrophysicsResearchInstitute,LiverpoolJohnMooresUniversity,146BrownlowHill,LiverpoolL35RF,UK
4 Centre de Recherche en Astrophysique du Que´bec (CRAQ) Universite´ Laval, 1045 Avenue de la Me´decine, G1V 0A6 Que´bec,
Canada
5 FACom-InstitutodeFsica-FCEN,UniversidaddeAntioquia,Calle70No.52-21,Medell´ın,Colombia
5
1 Preprintonlineversion:January13,2015
0
2 ABSTRACT

n
Recent discoveries have put the picture of stellar clusters being simple stellar populations into question. In particular, the color-
a
magnitudediagramsofintermediateage(1-2Gyr)massiveclustersintheLargeMagellanicCloud(LMC)showfeaturesthatcould
J
beinterpretedasagespreadsof100-500Myr.Ifmultiplegenerationsofstarsarepresentintheseclustersthen,asaconsequence,
9
young (<1 Gyr) clusters with similar properties should have age spreads of the same order. In this paper we use archival Hubble

Space Telescope (HST) data of eight young massive LMC clusters (NGC 1831, NGC 1847, NGC 1850, NGC 2004, NGC 2100,
]
A NGC2136,NGC2157andNGC2249)totestthishypothesis.Weanalyzedthecolor-magnitudediagramsoftheseclustersandfitted
theirstarformationhistorytoderiveupperlimitsofpotentialagespreads.Wefindthatnoneoftheclustersanalyzedinthiswork
G showsevidenceforanextendedstarformationhistorythatwouldbeconsistentwiththeagespreadsproposedforintermediateage
. LMCclusters.Testswithartificialsingleageclustersshowthatthefittedagedispersionoftheyoungestclustersisconsistentwith
h spreadsthatarepurelyinducedbyphotometricerrors.AsanadditionalresultwedeterminedanewageofNGC1850of∼100Myr,
p
significantlyhigherthanthecommonlyusedvalueofabout30Myr,althoughconsistentwithearlyHSTestimates.
-
o Keywords.galaxies:starclusters:general-galaxies:individual:LMC-Hertzsprung-RusselandC-Mdiagrams-stars:evolution
r
t
s
a
[ 1. Introduction the cluster must have been 10 -100 times more massive in the

pastthanobservedtoday(Conroy2012).Analternativescenario
1 Globular clusters (GCs) have long been thought to be simple
has been proposed by Bastian et al. (2013a) where chemically
v stellar populations (SSP), as their stars formed out of the same
enrichedmaterialfromrotatingstarsandmassiveinteractingbi-
5 molecular cloud at approximately the same time. Therefore all
naries falls on to the accretion disks of low-mass pre-MS stars
7
stars in a GC should have (within a small range) the same age
ofthesamegenerationandisfinallyaccretedonthestillform-
2
and the same chemical composition. However, recent discov-
2 ingstars.Inthisscenarioallstarsarefromthesamegeneration
eries have put this simple picture into question. Features like
0 withoutsignificantagespreadsamongthem.
extended main sequence turn-offs (MSTOs), double main se-
. BesidestheanomalousfeaturesobservedinoldGCswhich
1 quences (MS) and multiple subgiant and giant branches in the
0 pointtowardsamorecomplexscenariothanasinglestellarpop-
color-magnitude diagrams (CMDs) of GCs as well as abun-
5 ulation,intermediateage(1-2Gyr)clustersintheLMCshowex-
dance variations of light elements (e.g. C, N, O, Na, Al) have
1 tendedordoubleMSTOs(e.g.Mackey&BrobyNielsen2007;
beenfound(e.gGratton,Carretta&Bragaglia2012;Piottoetal.
: Mackeyetal.2008;Miloneetal.2009;Goudfrooijetal.2009,
v 2012).Thereareseveralattemptstoexplaintheobservedanoma-
2011a,b). This might be a consequence of an extended period
i lies in GCs. The most common model implies the presence of
X ofstarformation(100-500Myr).Othertheoriesrelatetheex-
multiplegenerationsofstarsinsideGCs.Inthismodel,thesec-
r tendedMSTOstointeractingbinaries(e.g.Yangetal.2011)or
ondgenerationofstarsisthoughttobeformedoutofenriched
a stellarrotation(e.g.Bastian&deMink2009;Yangetal.2013;
materialfromthefirststellargeneration.Proposedsourcesofthe
Lietal.2014).Thecentrifugalforceinrotatingstarsdecreases
ejecta are AGB stars (D’Ercole et al. 2008), fast rotating stars
theeffectivegravitywhichcausesalowereffectivetemperature
(Decressin et al. 2009) and interacting binaries (de Mink et al.
andluminosity(e.g.Meynet&Maeder1997).However,Platais
2009). A large drawback of this model, however, is that in or-
et al. (2012) analyzed the CMD of the galactic open cluster
der to produce the observed amount of second generation stars
Trumpler 20 and did not find evidence that rotation affects the
CMD. Possibly, extended or multiple MSTOs in intermediate
(cid:63) Based on observations made with the NASA/ESA Hubble
ageclustersareassociatedwiththepresenceofmultiplestellar
Space Telescope, and obtained from the Hubble Legacy Archive,
populationsobservedinoldGCs(Conroy&Spergel2011).
which is a collaboration between the Space Telescope Science
Institute (STScI/NASA), the Space Telescope European Coordinating However,recentstudiessuggestthatintermediateageLMC
Facility (ST-ECF/ESA) and the Canadian Astronomy Data Centre clusters, in contrast to galactic GCs, do not have spreads in
(CADC/NRC/CSA). chemicalabundances.Mucciarellietal.(2008)analyzedtheex-
1
F.Niederhoferetal.:LMCClusters
tended MSTO clusters NGC 1651, NGC 1783 and NGC 2173, 0.5
andalsotheolderclusterNGC1978anddidnotfindchemical
anomalies. Furthermore, NGC 1866 (Mucciarelli et al. 2011),
NGC 1806 (Mucciarelli et al. 2014) and NGC 1846 (Mackey 0.4
et al. in prep.) also do not have abundance spreads. This calls 2B
d
into question the idea that the extended MSTO feature is due +
toagespreads,assuchscenarioswouldnaturallypredictabun- 2V0.3
d
(cid:0)
dancespreadsthroughself-pollution.Thus,anunderstandingof =
tshigehptrionptehretieesvoolfutiinotneromfeGdiCaste. ageclusterscangivevaluablein- or(B-V) 0.2
Ifitistruethatintermediateageclustershostmultiplestellar Err
generationsthenyoung(<1Gyr)clusterswithsimilarproperties 0.1
should display signs of age spreads inside the cluster as well.
There are several studies that search for age spreads or ongo-
ingstarformationinyoungmassiveclusters(YMCs).Cabrera- 0.0
15 16 17 18 19 20
Zirietal.(2014)analyzedthespectrumofaYMCinthemerger V
galaxy NGC 34 and concluded that the star formation history √
(SFH) is consistent with a single stellar population. Bastian et Fig.1: Color errors ( dB2+dV2) of NGC 2136 as a function
al.(2013b)presentedacatalogcontainingmorethan100galac- of the V-band magnitude . The filled (red) squares indicate the
ticandextragalacticyoung(<100Myr)massiveclustersanddid dispersionwidthofthemainsequence.
notfindanyevidenceofongoingstarformationintheirsample.
Bastian & Silva-Villa (2013) started a project to constrain
possibleagespreadsinyoungmassiveLMCclusters.Theyana- Parsec1.1isochronesetofthePadovaisochrones(Bressanetal.
lyzedtheCMDsofthetwoyoungclustersNGC1856(281Myr) 2012)andweassumedaSalpeter(1955)IMF.
andNGC1866(177Myr)andfittedSFHstotheclusters’CMDs. The structure of the paper is the following: Section 2 de-
Theyfoundnoagespreadsinthesetwoclustersandconcluded scribes the used data set and the further processing of the data.
thatbothareconsistentwithasingleburstofstarformationthat WeperformtestswithartificialclustersinSection3.Theresults
lasted less than 35 Myr. In this work, we continue the study ofthefittingoftheSFHaregiveninSection4.Wediscussthe
by Bastian & Silva-Villa (2013) by searching for potential age resultsanddrawfinalconclusionsinSection5.
spreadsineightmoreYMCsintheLMC.
We studied the CMDs of the clusters and fitted theoretical
isochronestothemtodeterminetheirage,metallicity,extinction 2. ObservationsandDataProcessing
and distance modulus (DM). With these parameters we created
modelHessdiagramsofdifferentagesandcomparedthemwith ForouranalysiswemadeuseofdatatakenwiththeWideField
theobservationstoputinitialconstraintsonanyagespreadthat andPlanetaryCamera2(WFPC2)onboardoftheHubbleSpace
might be present within the clusters. In order to provide more Telescope (HST). The data set is presented in detail in Brocato
quantitative results, we also fitted the star formation history of et al. (2001) and Fischer et al. (1998). NGC 1850, however, is
eachclusterusingthecodeFITSFH(Silva-Villa&Larsen2010 notpartofthesestudies,weretrievedthedatatablesofthisclus-
andLarsenetal.2011).ThiscodecreatesatheoreticalHessdi- ter from the HST Legacy Archive (HST Proposal 5475, PI: M.
agram taking into account the photometric errors and some as- Shara).NGC2157ispresentedinFischeretal.(1998)whereas
sumptions (e.g. metallicity). It searches for the best match be- the Brocato et al. (2001) data set covers the other six clusters.
tweenthedataandatheoreticalmodelwhichisalinearcombi- TheimagesoftheclustersweretakenintheF450WandF555W
nation of Hess diagrams constructed from different isochrones. filters (Brocato et al. 2001 data set and NGC 1850) and in the
Wedonottakeintoaccountbinariesanddifferentialreddening F555W and F814W filters (Fischer et al. 1998). All the mag-
(exceptforNGC2100whichhasaconsiderablevariationofex- nitudes in the HST filter system were transformed to standard
tinctionacrossthecluster)inthefittingoftheSFH.Theireffects Johnson BVI magnitudes. We use the fully reduced data tables
on the CMDs will only increase any potential age spread that whichconsistofthestars’xandypixelcoordinatesonthedetec-
mightbepresentintheclusters.Additionally,thearchivalpho- torsystem,themagnitudesandtheirstandarddeviationsintwo
tometric data tables that we use for this work contain only the filterseach.
standard deviation of the data reduction as photometric errors. TheHLAcatalogofNGC1850doesnotincludephotomet-
Figure1shows,asanexample,theerrorsinB−V colorofNGC ricerrors.Sowecreatedartificialerrorsthatweremodeledfrom
2136asafunctionoftheV-bandmagnitude.Thefilledsquares theBrocatoetal.(2001)dataset.Wetooktheseclustersfromthe
(red) indicate the width of the MS, given by the standard devi- datasetthathavethesameexposuretime(40s)asNGC1850as
ation. The width of the MS is consistent with the overall trend areferenceandwesimulatedtheerrorssuchthattheyfollowthe
of the errors, especially in the region 17 ≤ V[mag] ≤ 19 that sameexponentialincreasewithfaintermagnitudesastheerrors
is not sensitive to age. The real scatter of the errors around the from the reference clusters. Figure 2 shows the artificial errors
global relation might be larger as the photometric standard de- inB−V colorasafunctionoftheV-bandmagnitude.Thefilled
viations underestimate the real errors. However, underestimat- (red) squares indicate the width of the MS (standard deviation)
ingthephotometricerrorswillalsoleadtolargerestimatedage of NGC 1850. It is comparable with the mean error in the in-
spreads.Therefore,allagespreadspresentedinthisworkareup- terval 17 ≤ V[mag] ≤ 19 where the MS is not affected by age
perlimitsofrealonesintheclusters.Theaimofthisworkisnot effects.Weareawarethatthescatterofthemodelederrorscould
toquantifytheexactextentoftheSFH,butrathertotestifsig- alsobeunderestimatedastheerrorsweremodeledfromthestan-
nificant(tenstohundredsofMyr)agespreadsarepresentwithin darddeviationofthephotometricreductionofotherclusters(see
theclusters.Forallmodelsandfittingsinthispaperweusedthe previousSection).
2
F.Niederhoferetal.:LMCClusters
0.5 x (pc)
−20 −10 0 10 20
20
0.4
1500
2B
d
+
2V0.3 1000
d 10
(cid:0)
=
B-V) 0.2 500
Error(0.1 y (pixel) 0 0 y (pc)
−500
0.0
15 16 17 18 19 20 −10
V −1000
√
Fig.2: Modeled color errors ( dB2+dV2) of NGC 1850 as a
−1500
functionoftheV-bandmagnitude.Thefilled(red)squaresindi-
−20
catetheobserveddispersionwidthofthemainsequence.
−1500 −1000 −500 0 500 1000 1500
x (pixel)
Additionally to the photometry of NGC 2157, Fischer et
Fig.3: NGC 1847; Positions of the stars on the WFPC2 chip.
al. (1998) provide data of a background field which is located
26(cid:48)(cid:48)east and 110(cid:48)(cid:48)north of the cluster for subtraction of field The inner circle (red) corresponds to two times the core radius
ofthecluster.Theintersectionoftheoutertwocirclesforming
stars.Forouranalysiswerestrictourselvestotheinnerregions
an annulus with the corners of the WFPC2 field of view is the
oftheclusters.
area which is used for field star removal. The asterisk symbol
AstheclustersintheBrocatoetal.(2001)datasetandNGC
marksthecenteroftheclusterasdeterminedbyus.
1850arenotcenteredonanyofthefourWFPC2detectorchips
we had to determine their cluster centers in the x,y CCD co-
ordinate system. We did this by first creating artificial blurred
images of the clusters by convolving the flux weighted spatial tered on the respective cluster that intersect the corners of the
positionsofthestarsineachclusterwithaGaussian.Wechose chipssuchthattheintersectionareaoftheannulusandthechip
Gaussianswithaσrangingbetween40and60pixels(changing isapproximatelythesameastheareawherewewanttosubtract
from cluster to cluster) in order to get a smooth flux distribu- the contaminating stellar population (see Figure 3 as an exam-
tioninthecenterofeachcluster.Afterwardswefittedelliptical ple). After defining the areas we created CMDs for the stars in
isophotes to the created images using the IRAF1 task ellipse bothregionsandsubtractedforeverystarinthefieldCMDthe
andusedthecenteroftheinnermostisophotesastheclustercen- star in the corresponding cluster CMD that has the closest ge-
ter.Onlythosestarswhicharelocatedinsidetwotimesthecore ometricaldistanceincolor-magnitudespace.However,wehave
radiusRcore (radiusatwhichthedensityishalfthecentralden- tobearinmindthatalsointhisouterregionsacertainfractionof
sity)givenbyBrocatoetal.(2001)areusedforfurtheranalysis. clusterstarsisstillpresentthatissubtractedasfieldstars.This
Rcore isbetween2.0and∼4pc(seeTable1),atanassumeddis- becomes clear if we compare the field of view of the WFPC2
tance of 50 kpc (de Grijs et al. 2014). NGC 2157, however, is detector with the tidal radii of the clusters. The tidal radii are
approximatelycenteredonthePCchipoftheWFPC2.Thestel- always larger than 30 pc (McLaughlin & van der Marel 2005,
lardensityishighestonthischipwithasteepfallofftowardsthe assumingaKing(1966)profile)whereasthefieldofviewofthe
otherthreechips.Forthisclusterweusejustthestarslocatedon WFPC2 is ∼36 pc in diameter (assuming a distance of 50 kpc
the PC chip which has 800 × 800 pixels with a pixel scale of totheLMC).However,anover-subtractionisnotaseriousissue
0(cid:48).(cid:48)046perpixel.Assumingadistanceof50kpctotheLMCthe asitaffectsmostlythewellpopulatedregionsintheCMDsand
PCchipcoversanareaof8.9×8.9pc.Thisareaiscomparable keepstheoverallstructureunchanged.
with the area given by the projected half-light radius of NGC
Asalreadymentionedweareprovidedwithanimageoffield
2157(5.4pc,McLaughlin&vanderMarel2005assumingKing
stars near NGC 2157. Therefore, in this case, we made CMDs
1966models).
ofthestarsinthefieldandtheclusterregionofeverychipand
For our purpose we need to further analyze the data sets.
subtractedthefieldstellarpopulationofeverychipseparatelythe
The first step is the subtraction of the field star contamination
samewayaswedidforthepreviousclusters.Figure4showsthe
which is a combination of Galactic foreground stars and LMC
CMDsofallclusters.Theblackdotsindicateallstarsthatwere
fieldstars.Foralltheclusters,exceptforNGC2157,wedonot
used in our analysis and the triangles (cyan) are the subtracted
haveextrafieldexposuressowehavetousetheclusterimages
’field’stars.
themselvesforthefieldstarsubtraction.Tominimizethecontri-
Brocatoetal.(2001)andFischeretal.(1998)performedar-
butionofclusterstarsweconstructedareasthatareasfaraway
tificial star tests to infer the completeness curves of their clus-
fromthecenteroftheclusteraspossible.Wechoseannulicen-
ters.WeadoptlimitingV-bandmagnitudesthatareatorbrighter
1 IRAF is distributed by the National Optical Astronomy than the 90 % completeness limit (cf. their Table 3). Fischer et
Observatories,whichisoperatedbytheAssociationofUniversitiesfor al.(1998)determinedforNGC2157a90%completenesslimit
Research in Astronomy, Inc., under cooperative agreement with the inthe I bandof∼20.5mag(cf.theirFigure3).ForNGC1850,
nationalScienceFoundation. however, we do not have any measure of the completeness. To
3
F.Niederhoferetal.:LMCClusters
beonthesafeside,wechoosealimitof18.5magintheV band tersisnotonlyduetophotometricerrors.Wewilldiscussthisin
forthefurtheranalysisofthiscluster.Thislimitismorethan1 Section5.
magnitude brighter than the completeness limits of the cluster
in the Brocato et al. (2001) data set with comparable exposure
times. 4. Results
The aim of this work is to search for potential age spreads in
3. TestswithArtificialStarClusters a sample of eight young (<1.1 Gyr) massive (> 104M(cid:12)) star
clustersintheLMC.Oursampleofclusterscoversanagerange
BeforewefittheSFHofoursampleofLMCclusterswemade
from20Myrtoabout1Gyr.Theclustersanalyzedinthiswork
tests with artificial coeval star clusters to assess the magnitude
and in Bastian & Silva-Villa (2013) have similar properties as
ofapparentagespreadsthatarepurelyinducedbyphotometric
theintermediateage(1-2Gyr)LMCclustersthatshowextended
errors.Foreachclusterinoursamplewemodeledacorrespond-
ordoubleMSTOs.TheleftpanelofFigure6showsthemasses
ing artificial cluster with the same metallicity and age that we
of the clusters of our sample as a function of the effective ra-
found for the real cluster. We assigned every cluster a random
number of stars that is drawn from a Gaussian distribution that dius Reff along with the clusters presented in Goudfrooij et al.
(2009, 2011a). Both, young and intermediate age clusters fol-
peaks at the number of stars present in the data table of the re-
spective observed cluster. The masses of the cluster stars were lowthesametrendofincreasingReffwithincreasingmass.NGC
1847 is not included in this plot as it has a large uncertainty in
drawn stochastically from a stellar initial mass function (IMF)
with an index α of −2.35 (Salpeter 1955). The lower limits of itseffectiveradiusduetoitsshallowprofile.Therightpanelof
Figure6showsthecoreradiusR asafunctionoflogarithmic
thestellarmasseswerechosensuchthattheCMDsoftheartifi- core
clustermassforthesameclusters.Itisexpectedthat,duetody-
cialclusterscoverthesamemagnituderangeastherealclusters.
namical evolution, the spread in R increases with the age of
Wecreatethephotometryforeachstarbylinearlyinterpolating core
theclusters(e.g.Kelleretal.2011).WenotethatalltheYMCs
theisochronegrid(Parsec1.1fromthePadovaset)attherespec-
havesystematicallysmallercoreradiithantheintermediateage
tiveagesoftheclustersandaddingphotometricuncertaintiesto
clusters. One reason for this could be that the core radii given
the synthetic photometry that follow the same behavior as the
byGoudfrooijetal.(2009,2011a)areconstructedfromthesur-
observederrors.Forallobservedclusterswefittedanexponen-
face number density profiles, whereas McLaughlin & van der
tial curve to the photometric standard deviations as a function
Marel(2005)usedsurfacebrightnessprofiles.Bothmethodsdo
of the magnitude in each band. We then assigned every star an
notyieldnecessarilythesamevalueforR .Duetodynamical
error that is drawn from a Gaussian distribution with a width core
masssegregationinsideclustersthesurfacebrightnessprofileis
that corresponds to the value of the exponential function at the
more concentrated towards the center and therefore results in a
respective magnitude plus a small random scatter comparable
smallervalueforthecoreradius.
to the scatter of the real errors around the fitted curve. The up-
per panels of Figure 5 show as an example the CMDs of four We fitted theoretical isochrones from the Parsec 1.1
ofthemodeledclusterstogetherwiththetheoreticalisochrones isochrone set (Bressan et al. 2012) to the observed CMDs to
thatwereusedtocreatethephotometry. estimatetheclusters’metallicity,distancemodulusandthered-
When comparing the artificial clusters with the observed deningtowardstheclusters.Thissetofisochronesusesavalue
ones (cf. Figure 4 and 5) we note that the MS of the synthetic of 0.0152 for the solar metallicity Z(cid:12). Table 1 lists the basic
clustersismuchbroaderatfaintermagnitudesthantheMSofthe propertiesoftheLMCclustersthatarethesubjectofthiswork.
realclusters.Thisisduetothefactthatwemodeledtheclusters For the further analysis we dereddened the magnitudes of all
using the photometric errors of the corresponding real clusters. stars in each cluster by the same value of the derived redden-
IfwelookatFigure1weseethattheMSatfaintermagnitudes ing(weassumenodifferentialreddening,exceptforNGC2100).
isnarrowerthanwouldbeexpectedfromthephotometricerrors. ThedataofNGC2157isalreadycorrectedforareddeningvalue
However,thisdoesnotaffectouranalysisaswedonotusethese ofE(B−V)=0.1byFischeretal.(1998).
regionsintheCMDforthefittingoftheSFH(seeSection4). As a second step we fitted the SFH of each cluster using
We carried out 100 Monte Carlo realizations of each clus- the code FITSFH (Silva-Villa & Larsen 2010). For the fitting
terandfittedtheirSFHthesamewayaswedidfortheobserved weusedthepreviouslydeterminedmetallicityanddistanceand
clusters(seeSection4).Theresultsoftheclustersthatareshown assumedaSalpeter(1955)IMF.
asanexamplearepresentedinthelowerpanelsofFigure5.The Weperformedthefittingintworegions(a”blue”anda”red”
dots represent the mean contributed mass fraction that results fittingbox)oftheCMDforeachcluster.Theblueboxcontains
from the 100 Monte Carlo realizations, at individual ages. The theMSoftheclusterwhereastheredboxcoverstheregionsof
errorbars are the standard deviations. The (red) dashed line is theevolvedstars.Thelimitsofalltheboxesandthetotalnumber
thebestGaussianfittothehighestpeakinthedistribution,tak- ofstarsinthoseboxesaresummarizedinTable2.Thefaintlim-
ing into account the errors. The thick vertical line at the x-axis itsoftheboxescontainingtheMSarechosensuchthattheyare
marks the input age of the cluster. All fits reproduce the input atleast0.5magbrighterthanthe90%completenesslimits.We
ageoftheclustersverywellwithintheerrors.Wenotethatthe didseveralfitswithdifferentchoicesoftheboxlimits.Thereby
fitstotheclustersshowsomeadditionallowamplitudesofstar we noticed that the overall result does not depend on the exact
formationathigheragesasitisobservedintherealclusters(see choice of the boxes. To assess the statistical errors of the fit-
Section4).Thisisafirstsignthatthesefeaturesareduetothefit- tingsthatresultfromthestochasticIMFpopulationofstarswe
tingprocessandnotintrinsictotheclusteritself.Thefittedages performed additional bootstrapping tests. We created for each
andstandarddeviationsofallmodeledclustersaresummarized cluster50bootstrapsamplesandrantheSFHforeachofthese
inTable3wherewecomparethemwiththeresultsfromtheob- samples. The figures of the SFH fits presented in the next sub-
servedclusters.Thedispersionofagesislowerthanthespreads sections show the mean values at individual ages (black dots)
wefoundforthehighestpeakofmostoftherealclusterswhich and the one sigmaerrorbars that follow from the bootstrapping
suggest that the spread in the fitted SFH of the observed clus- procedure.
4
F.Niederhoferetal.:LMCClusters
12 12 12 12
14
14 14 14
16
16 16 16
18
V V V I
18 18 18 20
20 20 20 22
24
22 22 22
−0.5 0.0 0.5 1.0 1.5 2.0 −0.5 0.0 0.5 1.0 1.5 2.0 −0.5 0.0 0.5 1.0 1.5 2.0 26 −0.5 0.0 0.5 1.0 1.5 2.0 2.5
B V B V B V V I
− − − −
(a)NGC2249 (b)NGC1831 (c)NGC2136 (d)NGC2157
12 12 12 12
14 14 14 14
16 16 16 16
V V V V
18 18 18 18
20 20 20 20
22 22 22 22
−0.5 0.0 0.5 1.0 1.5 2.0 −0.5 0.0 0.5 1.0 1.5 2.0 −0.5 0.0 0.5 1.0 1.5 2.0 −0.5 0.0 0.5 1.0 1.5 2.0
B V B V B V B V
− − − −
(e)NGC1850 (f)NGC1847 (g)NGC2004 (h)NGC2100
Fig.4:CMDsoftheclustersinoursample.Theblackdotsarethestarsthatwereusedforthefurtheranalysis,whereasthe(cyan)
trianglesindicatestarsthatweresubtractedasfieldstars.Obviously,alsosomeclusterstarswereexcludedinourSFHfits,butthis
doesnotaffectourresults.AllCMDscontainonlythestarsintheinnerregionsoftheclusters(twotimesthecoreradius).Notethat
theCMDofNGC2157isinthe(V−I)vsIspace.
Table1:ParametersoftheLMCclusters
Cluster Age(Myr) logMass/M(cid:12) Z(Z(cid:12)=0.0152) Rcore(pc) Vesc(km/s) Vesc(km/s) E(B−V)
lit. thiswork lit. thiswork at10Myr lit. thiswork
NGC1831 700a 926 4.59b 0.016a 0.016 4.24c/4.13b 6.6b 9.3 0.01a 0.0
NGC1847 26d 57 4.44b 0.006e 0.006 3.35c/1.73b 4.2b 4.5 0.1f 0.16
NGC1850 30g 93 4.86b/5.15 0.008h 0.006 −−/2.69b 8.77b/12.3 15.5 0.17h 0.1
NGC2004 20i 20 4.36b 0.004j 0.004 2.18c/1.41b 7.0b 7.5 0.08k 0.23
NGC2100 16i 21 4.36b 0.007j 0.007 3.03c/0.99b 7.9b 8.5 0.24k 0.17
NGC2136 100l 124 4.30b 0.004l 0.005 2.91c/1.59b 7.4b 9.3 0.1l 0.13
NGC2157 100m 99 4.31b 0.008m 0.008 −−/2.00b 6.2b 7.8 0.1m 0.1
NGC2249 1000n 1110 4.48n 0.007a 0.008 2.79c/1.75b 9.4n 12.1 0.01a 0.02
NGC1856 281o 4.88b 0.008o −−/1.75b 11.0b 14.7 0.26o
NGC1866 177o 4.91b 0.008o −−/2.79b 9.5b 12.5 0.05o
Notes.Inthecolumnsthatgivetheage,themetallicityZandtheextinctionE(B−V),thefirstvalueisfromtheliteratureandthesecondoneis
determinedinthiswork.Additionally,wealsofoundnewvaluesforthemassandtheescapevelocityofNGC1850(secondvaluesintherespective
columns).Thecoreradiusoftheclusterscanbedeterminedintwoways:Usingthesurfacenumberdensity(firstvalue)orthesurfacebrightness
density(secondvalue).
Thelasttwoclusters(NGC1856andNGC1866)arenotstudiedinthiswork.TheywerealreadyanalyzedbyBastian&Silva-Villa(2013).
References:
(a)Kerberetal.(2007);(b)McLaughlin&vanderMarel(2005)assumingKing(1966)profiles;(c)Brocatoetal.(2001);(d)Elson&Fall(1988);
(e)Mackey&Gilmore(2003);(f)Nelson&Hodge(1983);(g)Baumgardtetal.(2013);(h)Fischeretal.(1993);(i)Elson(1991);(j)Jasniewicz&
Thevenin(1994);(k)Kelleretal.(2000);(l)Dirschetal.(2000);(m)Fischeretal.(1998);(n)Correntietal.(2014);(o)Bastian&Silva-Villa(2013)
The results of the SFH fitting will give us upper limits of 4.1. NGC2249
potentialagespreadsaswedonottakeintoaccountbinariesand
differentialreddening. NGC2249istheoldestclusterinoursample.Theliteratureage
of the cluster is between 660 Myr (Baumgardt et al. 2013) and
about 1 Gyr (e.g. Kerber et al. 2007, Correnti et al. 2014). It is
thereforenotaYMCanymorebutratherbelongstothecategory
of intermediate age LMC clusters. Adopting an age of 1 Gyr,
Correntietal.(2014)foundamassof3.0·104M(cid:12)forthiscluster.
In the following we present the age and SFH fitting results WematchedtheoreticalisochronestoNGC2249todetermineits
ofeachclusterinorderofdecreasingage. basicparametersforthefurtheranalysis.Wefoundareddening
5
F.Niederhoferetal.:LMCClusters
12 12 12 12
14 14 14 14
16 16 16 16
V I V V
18 18 18 18
20 20 20 20
22 22 22 22
−0.5 0.0 0.5 1.0 1.5 2.0 −0.5 0.0 0.5 1.0 1.5 2.0 −0.5 0.0 0.5 1.0 1.5 2.0 −0.5 0.0 0.5 1.0 1.5 2.0
B-V V-I B-V B-V
(a)20Myr(B,Vphotometry) (b)100Myr(V,Iphotometry) (c)125Myr(B,Vphotometry) (d)1.1Gyr(B,Vphotometry)
1.0
0.7 0.7
Mass Fraction000...465 Mass Fraction000...465 Mass Fraction00..68 Mass Fraction00..68
Contributed 00..32 Contributed 00..32 Contributed 00..42 Contributed 00..42
0.1 0.1
0.00 50 100 150 0.00 100 200 300 400 500 0.00 100 200 300 400 500 0.00 1000 2000 3000 4000 5000
Age [Myr] Age [Myr] Age [Myr] Age [Myr]
(e)20Myr(B,Vphotometry) (f)100Myr(V,Iphotometry) (g)125Myr(B,Vphotometry) (h)1.1Gyr(B,Vphotometry)
Fig.5:Upperpanels:CMDsoffouroftheartificialclustersthatwecreated.Wesimulatedtheseclusterstoestimatetheresulting
agespreadofasingle-agepopulationgivenbyourSFHfittingcodecausedonlybyphotometricerrors.Allstarsinoneclusterhave
thesameageandthephotometricerrorsareestimatedfromtheonesoftherealclustersattherespectiveage.Theclustersshown
herehaveagesof20,100Myr,125Myrand1.1Gyr,spanningtherangeofagesoftherealclustersinoursample.The(red)lineis
thetheoreticalParsec1.1Padovaisochroneattherespectiveageofthecluster.
Lower panels: Results of the SFH fits of the four artificial star clusters shown in the upper panel. The dots represent the mean
contributed mass fraction at individual ages of the 100 realizations whereas the errorbars are the standard deviation. The dashed
(red)lineisthebest-fitGaussiantothehighestpeakofthepointstakingintoaccounttheerrorbars.Indicatedasaverticalthinkline
atthex-axisistheactualageoftherespectivecluster.Thefittothe20Myroldclustergivesapeakat20.1Myrwithadispersionof
1.5Myr.Wegotanageof102Myrforthe100Myroldclusterwithanstandarddeviationof8.4Myrandanageof130Myrwitha
dispersionof9.6Myrforthe125Myroldcluster.The1.1Gyroldclusterhasafittedageof1.2Gyr.Thestandarddeviationis111
Myr.
E(B−V)of0.02,ametallicityofZ=0.008andaDM(m−M) theCMD.ThelimitsofthechosenboxescanbefoundinTable
0
of18.3mag.Thesevaluesareingoodagreementwithprevious 2.ThebluefittingboxcontainstheMSdowntoamagnitudeof
studies. Kerber et al. (2007), who also used the Brocato et al. 2.0 which is about half a magnitude above the 90% complete-
(2001) data set, modeled the CMDs of various clusters using ness limit given by Brocato et al. (2001). The red fitting box
Padova isochrones from Girardi et al. (2002). For NGC 2249 coverstheredclumpstars.Figure8showstheresultsoftheSFH
they find a value of 0.01 for the reddening, a metallicity Z of fit. The points are the mean contributed mass fraction of the fit
0.007andaDMof18.27mag.TherecentstudybyCorrentiet atindividualageswithonesigmaerrorbarsthatresultfromthe
al. (2014) that used deep HST photometry finds similar values bootstrapping.ThesolidlineisthebestfitGaussiantakinginto
fittingMarigoetal.(2008)isochronestotheCMDs(E(B−V)= accounttheerrorbars.Thecurvepeaksatanageof∼1.11Gyr.
0.02,(m−M) =18.2mag,Z=0.006). The dispersion of 139 Myr is a measure of the maximum age
0
Figure 7 shows the CMD of NGC 2249. The black dots spreadthatispresentinthecluster.
indicate individual observed stars. The CMD shows two evi-
dent features: The MS of the cluster which extends up to a B
4.2. NGC1831
band magnitude of ∼0.8 and the red clump at B ∼1.3 mag and
B−V ∼0.8.OverplottedaremodelHessdiagramsatthreediffer- NGC1831isthesecondoldestclusterinoursample.Kerberet
entagesfrom830Myrto1.4Gyr(filledcontours).Theposition al.(2007)foundanageof700Myrforthiscluster,ametallicity
oftheredclumpintheCMDandtheMSTOisbestreproduced Zof0.016,anextinctionE(B−V)of0.01andaDMof18.23.Li
by an age of about 1 Gyr, in agreement with the findings by etal.(2014)analyzedtheCMDofNGC1831andfoundthatthe
Kerberetal.(2007)andCorrentietal.(2014).Atanageof830 MSTOisbroaderthanexpected.Theyconcludethatthisfeature
Myr the MS extends to brighter magnitudes than it is observed isbestexplainedwithstarsofdifferentrotationvelocitiesanda
and at 1.4 Gyr the models predict a MSTO that is fainter than dispersionofagesbetween550and650Myr.Fortheiranalysis
observed. they used the following parameters: Z=0.012, E(B− V)=0.03
Foramorequantitativeanalysisoftheclusteragewefitthe and(m−M) =18.4mag.WedeterminedanextinctionE(B−V)
0
SFHoftheclusterusingtheFITSFHcode(Silva-Villa&Larsen of 0.0, a metallicity Z of 0.016 and a distance modulus of 18.1
2010)providingtheobservedmagnitudesandphotometricerrors magthatweusedfortheanalysisoftheSFHofNGC1831.The
ofthestarsasinput.WedidthefitoftheSFHintworegionsin CMDoftheclusterwithsuperimposedmodelHessdiagramsis
6
NGC 1847
F.Niederhoferetal.:LMCClusters
20 10
NGC 1783
8
15 NGC 1783
R (pc)eff10 LW 431 NGNCG C2 11098N8G7C 1831 NNGGCC 1 18N85G66C6 18N0NG6GCC 1 1885406 R (pc)core 46 LW 431 NGNCG 1C9 827108NGC 1831NGC 175N1GC 1806NGC 1846
NGC 2157 NGC 2004 NGC 1751 NGC 1856
NGC 1850
5
NGC 2100 2 NGC 2157 NGC N1G8C47 2249 NGC 1866
NGC 2249
NGC 2136 NGC 2136 NGC 2004
NGC 2100
0 0
4.0 4.2 4.4 4.6 4.8 5.0 5.2 5.4 4.0 4.2 4.4 4.6 4.8 5.0 5.2 5.4
log (Mass/M ) log (Mass/M )
⊙ ⊙
Fig.6:Leftpanel:PlotofthemassversustheeffectiveradiusforoursampleofYMCs(blackcircles)andsomeadditionalclusters
withsimilarproperties(blackdots),takenfromthecatalogofMcLaughlin&vanderMarel(2005).Thetwoclustersthataremarked
withablackdotsurroundedbyablackcirclearethetwoclustersanalyzedbyBastian&Silva-Villa(2013).The(red)trianglesare
clusters of intermediate ages that display extended MSTOs, from Goudfrooij et al. (2009, 2011a). In this plot NGC 1847 is not
shownasitseffectiveradiushasalargeuncertaintyduetoitsshallowprofile.WeincreasedthemassofNGC1850byafactorof2
withrespecttotheliteraturevaluetoaccountforthenewagethatwefound(seeSection4.5).
Right panel: Same as the left panel but for the core radius. It is shown, e.g. in Keller et al. (2011), that the core radii of clusters
havelargerspreadstheoldertheclustersare.WenoteherethatthecoreradiiofallclustersfromtheGoudfrooijetal.(2009,2011a)
samplearesystematicallyhigherthantheonesfromoursample.OnereasonforthiscouldbethatthecoreradiigiveninGoudfrooij
etal.(2009,2011a)areobtainedfromthesurfacenumberdensity,whereasMcLaughlin&vanderMarel(2005)usedthesurface
brightnessprofiles.Bothmethodsdonotyieldthesamevalue(seetext).
Fig.7:CMDofNGC2249withoverlaidmodelHessdiagramsatthreedifferentages.Theblackdotsaretheindividualobserved
starsandthefilled(colored)contoursaretheoreticalHessdiagrams.Theirrespectiveagesareindicatedintheupperleftcornerof
eachpanel.
showninFigure9.ItisverysimilartotheCMDofNGC2249 4.3. NGC2136
suggestingthatbothclustershavecomparableages.TheMSof
NGC1831extendsupto M ∼0.5mag.Thepositionofthered NGC 2136 is younger than the previous two clusters. Its age
B
clumpisataBmagnitudeofabout1.2andaB−Vcolorof∼0.9. fromtheliteratureisabout100Myr(e.gDirschetal.2000).By
LookingattheHessdiagramsweseethatthemodelat870Myr overplotting isochrones onto the cluster’s CMD we find a red-
fitsbestthedata. dening E(B−V)of0.13,whichisinagreementwiththevalue
of0.1±0.03foundbyDirschetal.(2000).Furthermore,wede-
terminedthemetallicityZtobe0.005whichissomewhathigher
WefittedtheSFHoftheclusterintworegionsintheCMD. than the value of 0.004 by Dirsch et al. (2000) and a distance
The limits of the fitting boxes are given in Table 2. The boxes modulus of 18.5 mag that is consistent with the mean distance
containtheMSaswellastheHe-burningstars.Theresultsofthe to the LMC (de Grijs et al. 2014). Figure 11 shows the CMD
fitting are shown in Figure 10. A Gaussian fit to the individual of NGC 2136 overlaid with Hess diagrams at ages of 100, 125
pointsthatrepresentthecontributedmassfractionatsingleages and 177 Myr. The top of the MS in this cluster is at B ∼ −2
gives a peak at an age of 924 Myr. The standard deviation of followed by an almost continuous sequence of already evolved
the Gaussian (126 Myr) is the upper limit of the age spread as stars which reaches a top brightness of M ∼ −4 and extends
B
we do not consider other effects like binaries, rotating stars or to a B−V color of ∼1.5. We note that there are stars in the so
differentialextinctionthatmighteffecttheCMD. called ”Blue-Hertzsprung-Gap” between the MS and the sub-
7
F.Niederhoferetal.:LMCClusters
Table2:RegionsintheCMDsoftheclusterswheretheSFHwasfitted
Cluster Colorlimitsofthe Magnitudelimitsofthe Colorlimitsofthe Magnitudelimitsofthe Numberofstars
bluefittingbox bluefittingbox redfittingbox redfittingbox insidetheboxes
NGC1831 −0.1≤(B−V)≤0.5 −6.0≤V ≤2.0 0.5≤(B−V)≤1.4 −6.0≤V ≤1.5 1015
NGC1847 −0.4≤(B−V)≤0.4 −6.0≤V ≤0.5 0.4≤(B−V)≤1.5 −6.0≤V ≤−2.5 280
NGC1850 −0.3≤(B−V)≤0.1 −6.0≤V ≤0.0 0.1≤(B−V)≤1.8 −6.0≤V ≤−2.5 995
NGC2004 −0.4≤(B−V)≤0.3 −6.0≤V ≤0.5 0.3≤(B−V)≤1.7 −6.0≤V ≤−1.5 410
NGC2100 −0.4≤(B−V)≤0.2 −6.0≤V ≤0.0 0.2≤(B−V)≤1.9 −6.0≤V ≤−1.0 375
NGC2136 −0.3≤(B−V)≤0.2 −6.0≤V ≤0.5 0.2≤(B−V)≤1.7 −6.0≤V ≤−0.5 360
NGC2157 −0.3≤(V−I)≤0.2 −6.0≤I≤1.5 0.2≤(V−I)≤1.9 −6.0≤I≤−1.0 1000
NGC2249 0.0≤(B−V)≤0.5 −6.0≤V ≤2.0 0.5≤(B−V)≤1.7 −6.0≤V ≤1.5 390
Fig.9:CMDofNGC1831withoverlaidmodelHessdiagramsatthreedifferentagesfrom750Myrto1.04Gyr.
0.5
0.7
n n
o0.6 o0.4
cti cti
a a
Fr0.5 Fr
ss ss 0.3
a a
M0.4 M
d d
e e
ut0.3 ut0.2
b b
ri ri
nt0.2 nt
o o
C C0.1
0.1
0.0 0.0
0 500 1000 1500 2000 0 500 1000 1500 2000
Age [Myr] Age [Myr]
Fig.8:ResultsoftheSFHfitofNGC2249.Thedotsrepresent Fig.10:ResultsforthefittingoftheSFHofNGC1831.Thedots
thecontributedmassfractionatindividualagesandthesolidline representtheresultsatindividualagesandthesolidlineshows
showsthebestGaussianfittothepointswithapeakat1.11Gyr thebestGaussianfittothepointswithapeakat924Myranda
andastandarddeviationof139Myr. standarddeviationof126Myr.
giantbranchwherenostarsarepredictedbythemodels.These beabittoobright(Figure11left).Ontheotherhand,anageof
starscouldbestarsinthedenseclustercenterthatareaffectedby ∼180MyristoooldforNGC2136(Figure11).Thetheoretical
crowding,fastrotatingstars,interactivebinariesorhighermass Hessdiagramfailstofittheblueloopstarsaswellastheposition
starsthatformedoutofthemergingorcollisionoftwolowmass oftheMSturnoff.Fromthiswecanalreadyruleoutapossible
stars(”bluestragglers”).TheblueloopregionintheCMDthat agespreadofmorethan±50Myr.
ispopulatedbyHe-burningstarsiswellsampledinthiscluster To put our initial estimate on a more quantitative basis we
whichmakesiteasiertoconstrainapossibleagespreadjustby fittedtheSFHusingFITSFH,aswedidinthepreviousclusters.
lookingatthetheoreticalHessdiagrams.Weseethatanageof We adopted two fitting boxes that were chosen such that they
125Myrfitsbestthepositionoftheevolvedstars,especiallythe include the main features of the cluster CMD but exclude the
redsideoftheblueloop(Figure11middle).Alsoanageof100 leftovercontaminationoffaintredfieldstars(compareFigure4).
Myr is compatible with the data, although the model seems to Thefaintlimitofthebluefittingboxisonemagnitudeabovethe
8
F.Niederhoferetal.:LMCClusters
0.5
0.4
n n
o o
acti acti0.4
Fr0.3 Fr
s s
s s
Ma Ma0.3
d d
e0.2 e
ut ut
b b0.2
ri ri
nt nt
Co0.1 Co
0.1
0.0 0.0
0 50 100 150 200 250 300 350 400 0 50 100 150 200
Age [Myr] Age [Myr]
Fig.12:ResultsoftheSFHfitofNGC2136.Thedotsrepresent Fig.15:ResultsoftheSFHfitofNGC2157.Thedotsrepresent
the results at individual ages and the solid line shows the best the results at individual ages and the solid line shows the best
Gaussianfittothepointswithapeakat123.3Myrandastandard Gaussianfittothepointswithapeakat98.3Myrandastandard
deviationof22.6Myr. deviationof13.2Myr.
and a distance modulus of 18.5 mag for the cluster, consistent
withthevaluesassumedfortheclusterbyFischeretal.(1998).
Figure14showstheCMDoftheclustertogetherwiththeoretical
Hess diagrams overplotted. As NGC 2136, which has a similar
age,theCMDofNGC2157showsaclearMSturnoffatabout
M = −3 and a well populated blue loop with a slight over-
V
density of stars at V − I ∼ 1.2 that marks the red envelope of
theloop.InagreementwithFischeretal.(1998)wenotethatan
age of 100 Myr fits best the observed stellar distribution in the
CMD (Figure 14 middle). From the CMD we can constrain an
agespreadtobelessthanabout±30Myr.
We chose the regions in the CMD that we used to fit the
Fig.13:CMDofNGC2136withanoverplottedtheoreticalHess
SFHoftheclustersuchthattheycontainthemainfeaturesofthe
diagramat200Myr.
clusterplussomeadditionalspacebelowandabovethesequence
ofevolvedstarstobetterconstrainyoungerandolderages(see
Table2fortheexactlimits).Theresultsofthefittingprocedure
90%completionlimitgivenbyBrocatoetal.(2001).Thefitting are displayed in Figure 15. As expected, the star formation has
oftheSFHofNGC2136givesaperiodofstarformationwhich aclearmaximumatabout100Myrwithaupperlimitofanage
peaksatanageof∼123Myrwithanupperlimitof±23Myrfor spread of ±13.2 Myr. Additionally, the fit yields a high value
itsduration(seeFigure12),comparablewithourfirstestimate. of star formation at 50 Myr with a 1.5σ significance. This age
Wealsonote,thatthereisanadditionalsignificantpeakat200 correspondstothestarsthatareatthetopoftheMSatM =−3.
V
Myr.Figure13showsaHessdiagramat200Myrsuperimposed However,thepositionoftheevolvedstarsat50Myrdonotagree
over the CMD of NGC 2136. The fitted star formation at 200 withtheobservations.Thereforethisagecanbeexcluded.
Myr is due to the faintest stars at the red end of the blue loop
at B−V ∼1.0. But at this age we would also expect a higher
4.5. NGC1850
densityofblueloopstarsata B−V colorofabout0.4anda B
magnitudeofabout−2.5wherenostarsareobserved.Therefore, NGC1850isabinaryclustersystemconsistingofamainclus-
wecanruleoutthepeakat200Myr. ter(NGC1850A)andasmallanddensesecondarycluster(NGC
1850B) composed mainly of young O and B stars. This cluster
is the only one for which we obtained the data from the HST
4.4. NGC2157
Legacy Archive. The data catalog consists of the spatial posi-
NGC2157isnotpartoftheBrocatoetal.(2001)datasetandits tions of the stars on the detector chip, the RA and Dec coordi-
dataistakenfromFischeretal.(1998)whoprovidesphotometry nates and the AB magnitudes in the F450W and F555W filter,
intheVandIfilters.Thephotometryisalreadycorrectedforex- however without photometric errors. We converted the magni-
tinctionbyFischeretal.(1998)whoadoptedareddeningvalue tudestoVegamagnitudesandafterwardstotheJohnsonBVsys-
E(B−V) of 0.1. We derived the same value and therefore did temandcreatedartificialphotometricerrorsthatfollowthesame
notchangeitforouranalysis.Byplottingtheoreticalisochrones behavior as the observed ones in the Brocato et al. (2001) data
overtheobservedCMDoftheclustertheydeterminedanageof setoftheotherclusters.Inourstudyweconsideronlythecen-
100 Myr for the cluster. We estimated a metallicity Z of 0.008 tral parts of the main cluster NGC 1850A (all stars within two
9
F.Niederhoferetal.:LMCClusters
Fig.11:CMDofNGC2136withoverplottedtheoreticalHessdiagramsat100,126and178Myr.
Fig.14:CMDofNGC2157withoverlaidHessdiagramsattheages80Myr,100Myrand140Myr.NotethatthisCMD,incontrast
totheotherclusterCMDs,isinthe(V−I)vs.M space.
V
Fig.16:CMDofNGC1850withsuperimposedmodelHessdiagramsatagesfrom60to400Myr.
timesthecoreradius).Theyoungstarsofthesecondarycluster was determined by Sebo & Wood (1995) by measurements of
areoutsidethisareaandthereforetheywillnotaffectouranaly- Cepheids.TheresultingCMDofNGC1850AisshowninFigure
sis.WedeterminedareddeningE(B−V)of0.1whichissmaller 16.ItshowsaMSthatextendsuptoaBmagnitudeofabout-3.5
than the value of 0.17 that is assumed by Fischer et al. (1993). andatrackofevolvedblueloopstarsthatgoestoaB−V color
Additionally, a metallicity of Z=0.006 and a distance modulus ofabout1.3.ComparingtheobservedCMDwiththemodelwe
of18.5magreproducesbestthepositionandextentoftheblue seethatforanageofabout90Myrthebestaccordanceisfound.
loop. The distance modulus is a bit smaller than 18.6 mag that Themagnitudeandthemorphologyoftheevolvedstarsmatches
10