Table Of Content

Mon.Not.R.Astron.Soc.000,1–7(0000) The myth of the molecular ring C. L. Dobbs(cid:63)1,2,3 and A. Burkert1,2† 1 Max-Planck-Institut fu¨r extraterrestrische Physik, Giessenbachstraße, D-85748 Garching, Germany 2 Universitats-Sternwarte Mu¨nchen, Scheinerstraße 1, D-81679 Mu¨nchen, Germany 3 School of Physics, University of Exeter, Stocker Road, Exeter EX4 4QL 2 1 0 2 10January2012 n a ABSTRACT J WeinvestigatethestructureoftheMilkyWaybydetermininghowfeaturesinaspatial 9 map correspond to CO features in a velocity map. We examine structures including logarithmic spiral arms, a ring and a bar. We explore the available parameter space, ] A including the pitch angle of the spiral arms, radius of a ring, and rotation curve. We G showthatsurprisingly,aspiralarmprovidesabetterfittotheobservedmolecularring than a true ring feature. This is because both a spiral arm, and the observed feature . h knownasthemolecularring,arecurvedinvelocitylongitudespace.Wefindthatmuch p of the CO emission in the velocity longitude map can be fitted by a nearly symmetric - 2 armed spiral pattern. One of the arms corresponds to the molecular ring, whilst the o opposite arm naturally reproduces the Perseus arm. Multiple arms also contribute to r t further emission in the vicinity of the molecular ring and match other observed spiral s a arms.WhethertheGalacticstructureconsistsprimarilyoftwo,orseveralspiralarms, [ thepresenceof2symmetriclogarithmicspirals,whichbegininthevicinityoftheends of the bar, suggest a spiral density wave associated with the bar. 1 v 1 INTRODUCTION thanthestellar,withinterarmspurs,andbranchesbetween 5 spiral arms absent in the stellar distribution. 7 Despite decades of observations, determining the spiral COandHimapsoftheGalaxy(e.g.Dameetal.2001) 7 structureofourGalaxyisstillintrinsicallydifficult.Itisnot clearly show the Perseus and Outer arms. However there 1 even clear whether the Galaxy contains 2, 3 or 4 primary are also difficulties with using gas tracers: i) it is difficult . spiral arms (Vallée 2005, 2008; Benjamin 2008; Steiman- 1 to map the opposite side of the Galaxy, ii) the emission in 0 Cameronetal.2010),andwhetherthespiralstructureisdif- the inner part of the Galaxy is dominated by a broad band 2 ferentinthegasandthestars.Traditionally,muchoftheCO in velocity-longitude (hereafter l−v) space, and iii) ambi- 1 emission of the Galaxy has been associated with a feature guities in calculating the distance to gaseous features from : known as the ‘molecular ring’ (Stecker et al. 1975; Cohen v therotationcurveandvelocitycrowding.Thusmappingthe &Thaddeus1977;Roman-Duvaletal.2010),around4kpc i spatial structure from the gas is far from straightforward. X fromtheGalacticCentre.Howeveritisunclearwhetherthis In this paper we take a slightly different approach. r is truly a ring, or simply emission from nearby spiral arms, Rather than using the molecular emission, or stellar distri- a as suggested by simulations of spiral galaxies (Englmaier & bution, to estimate the spiral structure, we instead assume Gerhard 1999; Rodriguez-Fernandez & Combes 2008; Baba the gaseous spiral arms exhibit some pattern and see how et al. 2010). well they fit the observed CO emission. We do not perform Inthepast,thespiralstructurefortheGalaxyhaspre- numerical simulations rather we simply assume the gaseous dominantly been determined from the stellar distribution spiral arms follow a logarithmic spiral pattern, assumed to (Vallée2005andreferencestherein).Measuringdistancesto arise from the gas response to a density wave. This has the stars is difficult for distances larger than a few kpc how- caveat that our results neglect streaming motions. However everduetoextinction.Alternativelywecanuseagastracer ifwecanfitthespiralpatternevenintheabsenceofstream- suchasCOorHi.Inothergalaxies,e.g.M83,M51,thespiral ing motions, this is a strong indication that the spiral pat- armsalsotendtobemuchnarrowerinthegasthanthestars, ternforourGalaxycanberepresentedbyasimplem=2or indicatingthatgasislikelyabettertracerofspiralstructure. m=4 pattern. There are also two direct advantages of our Foraspiraldensitywave,thegaseousandstellarspiralarms method;thefirstthatwecanreadilyexplorealargeparam- are expected to occupy slightly different patterns, with the eter space, and the second that we do not need to include gaseous arms slightly offset from the stellar arms except at the pattern speed, or spiral potential strength, which are corotation,andwithasmallerpitchangle(Gittins&Clarke unknown parameters. A similar approach has been carried 2004) 1. The gaseous structure is also much more complex thatforakinematicwavedrivenbyatidalinteraction,thestellar 1 although recent simulations of M51 (Dobbs et al. 2010) find andgaseousarmsarenotsystematicallyoffset. (cid:13)c 0000RAS 2 C. L. Dobbs out by Russeil (2003) for star forming complexes, but the 2.1 Fitting technique distribution they use does not display strong spiral struc- We can compare how well our models match the CO ob- ture. Steiman-Cameron et al. (2010) also fit spiral patterns servations by matching features such as the molecular ring, to the intensity of FIR cooling lines at each position in the Perseus Arm, Outer Arm simply by eye. However we also Galaxy. carry out a χ2 fitting between the velocity longitude map of Dame et al. (2001) and our models. The difficulty of the latteristhatwehavetomakenumerousassumptionstocon- vertourmodelsintoemissionmaps.Weassumetheemission 2 METHOD followsaGaussiancentredontheproposedspiralarmswith Toobtainspiralarms,weassumethatthemoleculargaslies avelocitydispersionof7kms−1.Wealsohavetomakesome ina2or4armedlogarithmicspiralpattern.Fromstandard assumptionsabouthowtheemissionscaleswithradius.We densitywavetheory,thegeneralexpressionforalogarithmic suppose the intensity falls with 1/rL2SR where rLSR is the spiral pattern is distancetothelocalstandardofrest(locatedatR=8kpc). We also assume that the amount of molecular gas falls off (cid:18) (cid:19) φ=A(r)cos tannilog(r/r0)−(θ−Ωpt) (1) nasor1m/raGliAseLtwhheeermerisGsAioLnissothtehartadtihuestoofttahleeGmaislasixoyn.Wmaettchheens that of Dame et al. (2001). We calculate the difference be- where A(r) provides the amplitude of the spiral, n is the tweentheobservedandmodelemission,σ2 =(I −I )2, number of spiral arms, i is the pitch angle, r is a constant obs mod 0 and minimise over the spiral arm orientation or molecular which controls the orientation of the arms, and Ω the pat- p ringradius.Whilstdeparturesfromtheseassumptions(e.g. ternspeed.Sinceweonlyrequirethepatternatthepresent adifferentvelocitydispersion,changesinscalingwithr time,wecansett=0.Tofindtheminimaofthepotential, LSR andr )changeσ,howwellthemodelsfitrelativetoeach we then simply have GAL other does not change. n There are still some difficulties with our fitting process log(r/r )=θ, (2) tani 0 when we compare to spiral features. In the Galaxy, local andgivensomevaluesofiandr wecanmapthepositions emissionispresentatalllongitudes,butabsentinourmod- 0 of the spiral arms. els.AwayfromtheGalacticCentre,thisemissiondominates Wethencomputethevelocitylongitudemap,requiring overfeaturessuchastheOuterArm.Thuswecannotreally a given rotation curve. We adopt a flat rotation curve for test for these features without introducing some arbitrary a logarithmic potential (Binney & Tremaine 1987), of the weighting, so we instead restrict our fit to longitudes be- form tween50◦and-50◦.Westillfoundhoweverthatthismethod was biased towards lower pitch angles, simply because the v=v0r2/(Rc2+r2) (3) arms cross the region l = ±50◦ multiple times. Therefore we also carried out a fit just to the part of the main spiral wherev andR areconstants.R determineshowfarfrom 0 c c arm which coincides with the molecular ring. We refer to the centre of the Galaxy the rotation curve becomes flat. the two fits as ‘total’ (i.e. for all the parts of the arms in PastobservationshaveindicatedtherotationoftheGalaxy our models between l =50◦ and -50◦) and ‘arm’ (i.e. just is between 210 and 240 km s−1 over the majority of the betweenthetangentpointsofthearmwhichcoincideswith Galaxy(Clemens1985).ThestandardreferenceisΘ =220 0 the molecular ring). kms−1atR =8kpc.Howevermorerecentmeasurementsof 0 Inthefirstpartoftheresultswecomparea2armedand the distance to masers suggest that Θ may be 250 km s−1 0 4 armed spiral pattern, so we use the ‘total’ fit. In Section or higher (Reid et al. 2009). We tried both v =220 and 0 3.1 we vary the pitch angle and compare results with both 250kms−1,thoughweonlyshowresultsforv =250kms−1. 0 the‘total’and‘arm’fits.Thebestfitorientationofthearm Astheobservationsshowthatthevelocitycurveisstillvery does not depend on which technique is used, but the pitch high close to the centre of the Galaxy, we choose R = 0.1 c angle does. kpc. Then we place the observer a distance of 8 kpc from the centre of the Galaxy. We show results where we adopt a spiral arm pattern, andwhereweassumethatthemolecularringistrulydueto 3 RESULTS a ring. Given that we simply use the computed locations of the spiral arms, or ring, we can investigate a large parame- In Figure 1 we show our best fit to the molecular ring for a ter space. The free parameters in our models are the pitch 2armedspiralpattern(toppanel)adoptingapitchangleof angleofthespiralarms;theradiusofaringfeature;theori- 11◦ (seenextsectionforresultswithdifferentpitchangles). entationofthespiralarms,thelengthandorientationofthe This model includes a bar of radius 3 kpc, which we have bar;theGalocentricradiusandtherotationvelocityv .We simplyplacedacrosstheGalacticcentre45◦ clockwisefrom 0 mainly consider the orientation and pitch angle of the spi- the position of the Sun. In all figures, we simply show the ral arms, but we also briefly mention the other parameters. lines tracing peak emission along the arms, rather than our In principle, this analysis could be carried out without any synthetic emission maps. The lines are overplotted on the priorknowledgeofthestructureoftheMilkyWay,butgiven velocity longitude map of Rodriguez-Fernandez & Combes thelargeparameterspacewehavestartedwiththelocation (2008), which used the data of Dame et al. (2001). of the observer, the rotation curve, and the orientation of From Figure 1, top panel, we see that the Scutum- the bar roughly based on observations. Centaurus (hereafter Sct-Cen) arm provides a good fit to (cid:13)c 0000RAS,MNRAS000,1–7 The myth of the molecular ring 3 300 200 100 y (kpc) 0 -100 -200 -300 80 60 40 20 0 -20 -40 -60 -80 x (kpc) 300 200 100 y (kpc) 0 -100 -200 -300 80 60 40 20 0 -20 -40 -60 -80 x (kpc) 300 10 4kpc 4kpc 5 kpc 200 5 kpc 6 kpc 6 kpc 5 100 y (kpc) 0 y (kpc) 0 -100 -5 -200 -10 -300 -10 -5 0 5 10 80 60 40 20 0 -20 -40 -60 -80 x (kpc) x (kpc) Figure1.OnthelefthandsideweillustratepossibleGalacticfeatures,a2armedspiralpattern(top),a4armedspiralpattern(centre) and3ringsofdifferentradii(lower).Thepositionoftheobserverismarkedbythecrossat8kpc.Ontherighthandpanelsweshowthe location of the spiral arms in a velocity longitude plot, from Rodriguez-Fernandez & Combes (2008) who used the data of Dame et al. (2001).Theblackcirclesandcrossesindicateterminalvelocitymeasurements.Thecyancrossshowninthetoprightpanelindicatesthe extremityoftheouterHIarmobservedbyDame&Thaddeus(2011).Thepitchangleforthespiralmodelsis11◦.Thespiralpatterns arethebestfitpatterns(withregardstotheorientationofthearms)totheobservedemissionbetweenl=±50◦.TheScutum-Centaurus armprovidesagoodfittothemolecularring.Thebestfitringhasradius6kpc.Howeversincethespiralarmiscurvedinv−lspace, ithasabetterfittothemolecularringcomparedtoaring.Pleaseviewjournalonlineforcolourversionsofthefigures. (cid:13)c 0000RAS,MNRAS000,1–7 4 C. L. Dobbs the molecular ring. The part of this arm in the lower left Model σ2/σ2 quadrant (negative l and v) also agrees very well with the 4arm location of the new arm as measured in HI by Dame & Ring 2.27 Thaddeus (2011). As also noted in their paper, it is very 1armedspiral 2.02 difficult to get the Sct-Cen arm to continue to the Outer 2armedspiral 1.33 4armedspiral 1.0 arm, without providing a very asymmetric spiral pattern. The second spiral arm provides a good fit to the Perseus Table1.Thedifferencebetweentheobservedandanalyticemis- arm.Thebestfitforthesecondarmwasalmostsymmetric with the Sct-Cen arm, only asymmetric by 10◦. It is possi- sion, σ2, where σ2 = (Im−Iobs)2 is shown for the ring model, and spiral arm models with a pitch angle of 11◦. The 1 armed ble to rotate the arms slightly and produce better matches model refers to just the Sct-Cen arm in Figure 1. The results tothetangentpoints(boxes),stillmatchingthespiralarms are normalised to our best fit model, the 4 armed spiral. The 4 reasonablywell,butherewesimplyshowthebestresultsto armed spiral provides the best fit, though all cases are a better our molecular ring fitting technique. The emission from the fitcomparedtoaring. second spiral arm (green) extends a little outside the scope of the observed CO emission. With a lower rotation curve (v = 220 km s−1), this is avoided. Otherwise there is little Pitchangle(◦) σ2/σ2 (‘total’) σ2/σ2 (‘arm’) 4arm 4arm difference for a lower rotation curve. ItisdifficulttoconstrainthestartingradiusoftheSct- 8.5 2.00 2.13 11 2.02 2.08 Cen Arm, given the complex emission towards the Galactic 13.5 2.11 2.06 Centre. We can however see that the arm marked as the 16 2.18 2.01 Perseus arm cannot extend much further inward, else there wouldbeunobservedemission(atl∼15◦,v∼150kms−1) Table 2.Weshowthedifferencebetweentheobservedandana- for this model. lyticemission,σ2,whereσ2=(Im−Iobs)2 fortheSct-Cenarm InthecentrepanelsofFigure1weshowa4armedspi- withdifferentpitchangles.Fortheleftcolumn,weconsideremis- ral model. To the two armed model, we added a third arm sion along the total length of the arm. For the right column, we (cyan)usingourfittingtechnique.Thisarmnaturallyrepro- consideremissiononlybetweenthetangentpointsoftheSct-Cen duces the Outer arm emission. We tried adding a 4th arm, arm.TheresultsarenormalisedasforTable1.Thebestfitpitch however our fitting technique did not show any minimum angleisdependentonthefittingtechnique,howeverinallcases, in the expected vicinity. Since this arm is close to the Sun thefitisbettercomparedtoaring(Table1). (or observer), the emission, and therefore results became dominated by this arm, whereas for the observations, the andaringof6kpcradius.Thespiralmodelsprovidebetter strongestemissioncoincideswiththemolecularring.There- fits statistically compared to the ring model. The 2 and 4 fore we reduced our calculated emission from this arm by armed spirals are also better fits compared to the 1 armed a factor of 10 compared to the other arms to fit the fourth model or ring, which is not surprising because they allow arm.ThisarmthenreproducestheCarinaarm,andalsothe tangentpointatl∼50◦,v∼50kms−1.Thefactorof10is morecomplexityandcoveralargerareawheretheobserved emission lies. However distinguishing between the 2 and 4 somewhatarbitrarybutdoesseemtoindicatethatanyarm armedmodels(andallowingforthedifferentdegreesoffree- between theSunandtheSct-Cen armissomewhatweaker. dom) is probably beyond the scope of our approach. Overallthough,theadditionofextraarmssuperimposedon We also performed a simpler test to compare between the Sct-Cen arm contribute further to the emission of the ourmodelsandtheobservedCOemissiononlybetweenthe molecular ring. tangent points of the Sct-Cen arm (the ‘arm’ fit). This test In the lower panels of Figure 1 we show the results for corrects for any bias due to spiral arms crossing the region rings. We show 3 examples with radii of 4, 5 and 6 kpc. l=±50◦multipletimes.WeshowtheresultsinTable2,and It can be seen from Figure 1 that a ring does not cover againthespiralarmmodelsstillprovidebetterfitscompared as much of the emission of the observed ‘molecular ring’ as to the ring. a spiral arm. The reason for this is because the observed molecular ring is actually curved in v −l space, whereas the emission from an actual ring is a straight line. So we 3.1 Parameter study see that whilst a 4 or 5 kpc radius ring can fit emission for positive longitudes, it misses the emission at negative Inthissection,weinvestigatehowalteringtheparametersof longitudes.Conversely,a6kpcringreproducestheobserved ourmodelsaffectshowwellthespiralarmsfittheobserved emissionatnegativelongitudes,butdoesnotagreewiththe emission. brightestobservedemissionatpositivelongitudes.Fromour The main parameter which we can vary is the pitch fitting technique, the 6 kpc ring gave the best fit, although angle of the spiral arms. In Figure 2 we show models with byeye5kpcappearsbest(thedifferenceisprobablybecause pitch angles of 8.5◦, 13.5◦ and 16◦. From Figure 2 we see theobservedemissionextendsfurtheratnegativelongitudes thatinallcasesthemolecularringiswellreproduced.Again, compared to positive longitudes). these are our best fit models, where we have fitted for the In Table 1 we show the difference between the model orientation of the arms. Thus to a large extent there is a estimated CO emission and that of the observations, σ2 = degeneracy between the pitch angle and the orientation of (I − I )2 for 1 (corresponding to the Sct-Cen arm in thearms,thoughitisnotalwayspossibletoreproduceother m obs Figure 1), 2 (upper panels, Figure 1) and 4 (middle pan- Galactic features. els,Figure1)armedspiralmodelswithapitchangleof11◦ Forthepitchangleof8.5◦,weobtainaveryasymmetric (cid:13)c 0000RAS,MNRAS000,1–7 The myth of the molecular ring 5 300 200 100 y (kpc) 0 -100 -200 -300 80 60 40 20 0 -20 -40 -60 -80 x (kpc) 300 200 100 y (kpc) 0 -100 -200 -300 80 60 40 20 0 -20 -40 -60 -80 x (kpc) 300 200 100 y (kpc) 0 -100 -200 -300 80 60 40 20 0 -20 -40 -60 -80 x (kpc) Figure 2.Thespiralpatternisshownformodelswith2spiralarmsandpitchanglesof8.5◦ (top),13.5◦ (centre)and16◦ (lower).The positionoftheobserverismarkedbythecrossat8kpc.Thearmshavebeenrotatedtoprovideabestfittothedata.Inallcasesthe Sct-Cen arm easily fits the molecular ring. For a pitch angle of 8.5◦, there is a good agreement with the observed CO emission, but the spiral arms are highly asymmetric. For higher pitch angles, we do not reproduce the tangent point at l = 30◦, v = −100 km s−1 as the arms do not extend to the inner part of the Galaxy, and also it is difficult to reproduce this tangent point and the Perseus arm simultaneously.Pleaseviewjournalonlineforcolourversionsofthefigures. (cid:13)c 0000RAS,MNRAS000,1–7 6 C. L. Dobbs pattern. Our fitting technique does not take into account start at the bar, whereas Vallée (2005) assumes that the tangentpoints,orthePerseusarmwhicharealsoconstraints Sagittarius-Carina and Cygnus (Outer) Arms start at the onthesecondspiralarm(green).Howeverevenifweadjust bar(thesearetheequivalenttothearmscolouredcyanand the second spiral to fit the tangent point at l ∼ −30◦, v ∼ yellow on Figure 1). We constrained the models such that −100 km s−1, the structure is still highly asymmetric. the Scutum-Centaurus and Perseus Arm begin at the bar, For pitch angles of 13.5◦ and 16◦ we do not produce becausethesearmsareseeninthestarsandthegas,whereas the tangent point at l ∼ −30◦, v ∼ −100 km s−1 simply the other arms may not be associated with stellar enhance- because the emission does not extend that far inwards. We ments (Drimmel 2000; Benjamin 2008). could continue the arms for another half rotation, however Churchwell et al. (2009) also proposed a schematic of this would lead to a rather short ((cid:46) 2 kpc) bar, with the the Galaxy based on the GLIMPSE infrared survey (see assumption that the arms begin at the bar. In any case also Benjamin 2008). They propose that the Galaxy is a it is difficult to match the tangent point and the Perseus 2-armedspiral,themainarmsbeingtheScutum-Centaurus arm simultaneously, especially when the pitch angle is 16◦. andPerseusArms,withseveralsecondaryspiralarms.They Another constraint is the observed distance to the Perseus adopt a long bar, but do not state the pitch angle of the arm, which is around 2 kpc towards l = 134◦ (Xu et al. arms. This is similar to our 11◦ pitch angle model if we 2006). We get good agreement with this for the models we choose a longer bar, and neglect the first 180◦ rotation of present. However if we constrain the Perseus spiral arm to the spiral arms, or the models we show for larger pitch an- matchthethetangentpointatl∼−30◦,v∼−100kms−1, gles. We did note though in Section 3.1 that starting the the distance to the Perseus arm becomes too large for the arms further out in the disc would likely miss regions of higherpitchangles.Finallywenotethatwedonotgetgood emission in the CO l−v diagram. agreement with the newly observed outer HI arm (Dame & Finally,Steiman-Cameronetal.(2010)proposeamodel Thaddeus 2011) with these pitch angles, in particular 8.5◦ based on [CII] and [NII] cooling lines. This is very similar and 16◦. to our 4 armed model in Figure 1. The main difference is We show in Table 2 how well the models with differ- that they use slightly larger pitch angles (13−16◦), and ent pitch angles fit the observed data, using the ‘total’ and whilst we obtained a reasonable fit with one pitch angle, ‘arm’fits.The‘total’fitfavourslowerpitchangles,because theyuseddifferentpitchanglesforeacharm.Withthepitch thearmscovermoreoftheregionofobservedemission.The angle for the Scutum-Centaurus arm they used (15.5◦), it ‘arm’ fit favours larger pitch angles, and provides a best is difficult to reproduce the HI feature seen by Dame & fit pitch angle of 16.5◦. Both techniques however neglect Thaddeus (2011) as the arm barely extends to the third features such as the Perseus arm, outer HI arm, and sup- Galactic quadrant. posed tangent points, which appear necessary to constrain the pitch angle. With these extra constraints, the 11◦ pitch angle reproduces more of the CO features. 4 DISCUSSION We also tested the orientation of the bar, the rotation curve and the position of the observer in the Galaxy. As It is relatively easy to find a spiral arm configuration such noted before, decreasing the rotation curve to 220 km s−1 that the region corresponding to the molecular ring is re- slightly reduces the scope of the emission in velocity space. produced by a nearby spiral arm. If we rotated the nearest Moving the radius of the observer closer to the Galactic spiral arm (with respect to the Galactic Centre), we would Centre increases the span of the nearer spiral arm in the obtainasimilarfeaturewithadifferentgradient.Increasing velocity plot. Thus to still achieve a similar pattern to Fig- ordecreasingthepitchangleofthearmschangestheextent ure 1, a lower pitch angle would be required, and likewise ofthemolecularring.Otherspiralarmsenhancethemolec- if the observer is further out in the Galaxy, a higher pitch ular ring, as all overlap at least at l = 0. Finally, as the angle would be needed. So again there is a degeneracy be- moleculargassurfacedensitydecreaseswithradius,outside tweentheGalocentricradiusandthepitchangle,butagain, the vicinity of the bar, the maximum emission will be from with significant departures from the observed values, it is the near spiral arm close to the bar. more difficult to reproduce all the features in the CO map Whilst a ring feature can also represent CO emission simultaneously. similar to the observed molecular ring, a ring does not fit Changing the orientation of the bar does not make a theobservationsaswellasaspiralarm.Thisisbecausethe verynoticeabledifferencetothelocationofemissionassoci- spiral arm appears curved in v−l space, similar to the ob- atedwiththebar.Itisdifficulttoreproducethefullextent served molecular ring, but dissimilar to a true ring. Thus ofCOemissioninvelocityspaceassociatedwiththebar,i.e. whilst we cannot rule out that the molecular ring corre- for velocities in excess of +200 km s−1. This could be due sponds to a true ring, we found that a spiral arm produced toahigherrotationcurve,motionsalongthebar,orsimply abetterfitcomparedtoaringoverallourrangeofpitchan- features near the Galactic Centre that we are missing. gles,andthisfindingwasrobusttothedetailsofourfitting technique. Binney et al. (1991) proposed that the molecular ring 3.2 Comparison to other models of the Milky Way couldbeduetotheouterLindbladresonanceofthebar.We There have been several models suggested for the structure alsoperformedsimulationswithabarredpotentialtoexam- oftheMilkyWayinrecentyears.Vallée(2005)proposesa4 ine whether a gaseous ring would form. However generally armedspiralmodelwithapitchangleof12◦.Themaindif- gasfeaturesproducedattheendofthebararehighlyellip- ference between his model, and our 4-armed model is that tical,asseenalreadyinnumericalsimulations(e.g.Wada& we suppose that the Perseus and Scutum-Centaurus Arm Norman (2001); Lin et al. (2008)). The elongated features (cid:13)c 0000RAS,MNRAS000,1–7 The myth of the molecular ring 7 duetothebarinourGalaxymaywellcorrespondtothefar Conference Series, The Spiral Structure of the Galaxy: andnear3kpcarms,seeninthemoleculargasdata.Inour Something Old, Something New.... pp 375–+ simulationsofbars,anyfeaturecorrespondingtoaringwas Binney J., Gerhard O. E., Stark A. A., Bally J., Uchida againsimplythespiralarmsclosetotheGalacticcentre.In K. I., 1991, MNRAS, 252, 210 fact,fewgalaxiesshowobviousringsinthegasattheendof Binney J., Tremaine S., 1987, Galactic dynamics. Prince- thebar–typicallyringsarenuclearringsmuchnearertothe ton, NJ, Princeton University Press, 1987, 747 p. centre, or features caused by large collisions. Our Galaxy is Chakrabarti S., Laughlin G., Shu F. H., 2003, ApJ, 596, probably not unusual in this respect. 220 Ourfiducialmodeladoptsapitchangleof11◦,assuming ChurchwellE.,BablerB.L.,MeadeM.R.,WhitneyB.A., aGalocentricdistanceof8kpc.Allthepitchangleswetried Benjamin R., Indebetouw R., Cyganowski C., Robitaille couldreproduceafeaturesimilartothemolecularring,and T.P.,PovichM.,WatsonC.,BrackerS.,2009,PASP,121, fromfittingthemolecularringalonewefindabestfitpitch 213 angleof16.5◦.Howeverdeviationsfromthe11◦ modeltend Clemens D. P., 1985, ApJ, 295, 422 toreproducefeweroftheotherobservedfeaturesintheCO Cohen R. S., Thaddeus P., 1977, ApJL, 217, L155 l−v diagram, or produce highly asymmetric arms. Larger Dame T. M., Hartmann D., Thaddeus P., 2001, ApJ, 547, pitchanglesseemtopointtowardsalongerbar,whichmeans 792 CO in the lower right quadrant of the v−l plot is absent. Dame T. M., Thaddeus P., 2011, ApJL, 734, L24+ Moreover the supposed tangent point of the inner part of Dobbs C. L., Bonnell I. A., 2006, MNRAS, 367, 873 the Perseus arm, and the distance to the Perseus arm, can- Dobbs C. L., Theis C., Pringle J. E., Bate M. R., 2010, notbematchedsimultaneouslywithlargepitchangles.Our MNRAS, 403, 625 2armedspiralmodelisfairlysymmetric,andisthusconsis- Drimmel R., 2000, A&A, 358, L13 tent with a density wave originating at the bar. Though we Englmaier P., Gerhard O., 1999, MNRAS, 304, 512 cannot rule out that the Galaxy simply consists of several, Gittins D. M., Clarke C. J., 2004, MNRAS, 349, 909 asymmetric, spiral arms, it seems less likely that by coinci- Kim W., Ostriker E. C., 2002, ApJ, 570, 132 dence they match both the inner and outer spiral structure Lin L.-H., Yuan C., Buta R., 2008, ApJ, 684, 1048 simultaneously.ReproducingallthefeaturesintheCO,e.g. MartosM.,HernandezX.,Yań˜ezM.,MorenoE.,Pichardo Outer Arm, Carina Arm, requires 4 spiral arms. B., 2004, MNRAS, 350, L47 It is thought that the Galaxy may exhibit two spiral PatsisP.A.,GrosbolP.,HiotelisN.,1997, A&A,323,762 arms which are evident in both stars and gas, whilst other Reid M. J., Menten K. M., Zheng X. W., Brunthaler A., features, e.g. the Sagittarius Arm, are only seen in the gas MoscadelliL.,XuY.,ZhangB.,SatoM.,HonmaM.,Hi- (Drimmel 2000; Benjamin 2008). The next step would be rota T., Hachisuka K., Choi Y. K., Moellenbrock G. A., to try and produce hydrodynamical models with potentials Bartkiewicz A., 2009, ApJ, 700, 137 based on the 2 armed spiral pattern shown here for exam- Rodriguez-Fernandez N. J., Combes F., 2008, A&A, 489, ple, and see whether there are gaseous spurs or arms which 115 do not correspond to stellar features and whether they cor- Roman-Duval J., Jackson J. M., Heyer M., Rathborne J., respond to observed features in CO. Such features could Simon R., 2010, ApJ, 723, 492 arise from the shearing of clouds in the spiral arms (Kim Russeil D., 2003, A&A, 397, 133 & Ostriker 2002; Dobbs & Bonnell 2006) or resonances in SteckerF.W.,SolomonP.M.,ScovilleN.Z.,RyterC.E., thedisc(Patsisetal.1997;Chakrabartietal.2003;Martos 1975, ApJ, 201, 90 et al. 2004). Steiman-Cameron T. Y., Wolfire M., Hollenbach D., 2010, ApJ, 722, 1460 Vallée J. P., 2005, AJ, 130, 569 Vallée J. P., 2008, ApJ, 681, 303 5 ACKNOWLEDGMENTS Wada K., Norman C. A., 2001, ApJ, 547, 172 Xu Y., Reid M. J., Zheng X. W., Menten K. M., 2006, WethanktherefereeTylerFosterforsuggestionsthathave Science, 311, 54 substantially improved the paper. We are very grateful to NemesioRodriguezforprovidingtheCOvelocitylongitude plot from Rodriguez-Fernandez & Combes (2008), and to ThispaperhasbeentypesetfromaTEX/LATEXfileprepared by the author. Tom Dame for providing the CO velocity-longitude data used for our comparison tests. We also thank Panos Pat- sisforvaluablediscussions.CLDacknowledgesfundingfrom the European Research Council for the FP7 ERC starting grant project LOCALSTAR. REFERENCES Baba J., Saitoh T. R., Wada K., 2010, PASJ, 62, 1413 Benjamin R. A., 2008, in H. Beuther, H. Linz, & T. Hen- ninged.,MassiveStarFormation:ObservationsConfront Theory Vol. 387 of Astronomical Society of the Pacific (cid:13)c 0000RAS,MNRAS000,1–7