ASTROPHYSICALJOURNAL PreprinttypesetusingLATEXstyleemulateapjv.03/07/07 THETRAILSOFSUPERLUMINALJETCOMPONENTSIN3C111 M.KADLER1,2E.ROS2,M.PERUCHO2,Y.Y.KOVALEV2,3,D.C.HOMAN4,I.AGUDO5,2,K.I.KELLERMANN6,M.F.ALLER7,H.D. ALLER7,M.L.LISTER8,ANDJ.A.ZENSUS2 (Received2007September26;Accepted2007December19) ABSTRACT In1996,amajorradioflux-densityoutburstoccuredinthebroad-lineradiogalaxy3C111. Itwasfollowed bya particularlybrightplasmaejection associatedwith a superluminaljetcomponent,whichhasshapedthe parsec-scalestructureof3C111foralmosta decade. Here, wepresentresultsfrom18epochsof VeryLong 8 Baseline Array (VLBA) observationsconductedsince 1995as partof the VLBA 2cm SurveyandMOJAVE 0 monitoringprograms. Thismajoreventallowsustostudyavarietyofprocessesassociatedwithoutburstsof 0 radio-loudAGNinmuchgreaterdetailthanhasbeenpossibleinothercases:theprimaryperturbationgivesrise 2 totheformationofaleadingandafollowingcomponent,whichareinterpretedasaforwardandabackward- n shock. Both componentsevolve in characteristically differentways and allow us to draw conclusionsabout a theworkflowofjet-productionevents;theexpansion,accelerationandrecollimationoftheejectedjetplasma J in an environmentwith steep pressure and density gradientsare revealed; trailing componentsare formedin 7 thewakeoftheprimaryperturbationpossiblyasaresultofcouplingtoKelvin-Helmholtzinstabilitypinching modes from the interaction of the jet with the external medium. The interaction of the jet with its ambient ] mediumisfurtherdescribedbythelinear-polarizationsignatureofjetcomponentstravelingalongthejetand h passingaregionofsteeppressure/densitygradients. p - Subjectheadings:galaxies:individual:3C111–galaxies:active–galaxies:jets–galaxies:nuclei o r t s 1. INTRODUCTION peratures (Homanetal. 2006), or their Lorentz factor dis- a tribution (Kellermannetal. 2004) and luminosity function Direct evidence for the existence of bulk relativistic out- [ (Cara&Lister2007). flows along the jets in blazars and other radio-loud ac- The relativistic-jet model(e.g., Blandford&Konigl1979) 2 tive galactic nuclei (AGN) comes from Very-Long-Baseline v Interferometry (VLBI) observations. The first evidence has become the de-facto paradigm in multiwavelength re- 7 for apparently superluminal structural changes was found search on blazars and other AGN, but VLBI observations 1 from changes in the fringe visibility curves of 3C279 and have demonstrated that the basic concept of ballistically- 6 moving isolated jet knots is clearly oversimplified: jet cur- 3C273(Whitneyetal.1971;Cohenetal.1971). Subsequent 0 vature (e.g., Vermeulen&Cohen 1994), stationary compo- higher-quality VLBI observations (see, e.g., compilation by . nents (e.g., Jorstadetal. 2001), and non-radial and acceler- 1 Vermeulen&Cohen 1994, and references therein) have es- ated motions (e.g., Kellermannetal. 2004), are found to be 0 tablished the “core-jet” type milliarcsecond-scale structure 8 of compact extragalactic jets: the core being a bright and common features of relativistic jets. Within individual jets, 0 unresolved flat-spectrum component at the end of a lin- there are often characteristic velocities suggesting the pres- v: ear structure, and the jet being composed out of individ- ence of an underlying continuous jet flow, but the “com- ponents” themselves most likely represent patterns moving i ual steep-spectrum components or “knots”. The knots fre- X at a different speed than the underlying flow, e.g., as hy- quently move away from the core with apparent veloci- drodynamicallypropagatingshocks(Marscher&Gear1985; r ties exceeding the speed of light. Monitoring observa- a Hughesetal.1985). tions of large source samples (Vermeulen&Cohen 1994; Recentyearshavebroughtmajorimprovementsinnumeri- Jorstadetal.2001;Kellermannetal.2004;Pineretal.2007) calsimulationsofrelativisticjets(see,e.g.,Gómez2005,fora haveprovidedimportantstatisticaltoolsforprobingrelativis- review). Itisnowpossibletosimulatethree-dimensionalrel- tic beaming and the intrinsic properties of extragalactic ra- ativistic jets(e.g.,Aloyetal. 2003) andto computethe rela- dio jets (Cohenetal. 2007), their intrinsic brightness tem- tivisticprocesses(e.g.,Gómezetal.1997)thattransferhydro- 1Astrophysics Science Division, NASA’sGoddardSpaceFlightCenter, dynamic results into observed brightness distributions (e.g., GreenbeltRoad,Greenbelt,MD20771,USA;[email protected] relativistic light abberation and light travel time delays). In 2Max-Planck-Institut für Radioastronomie, Auf dem Hügel69, 53121 particular,interactionsbetweenstrongperturbationsorshocks Bonn,Germany;ros,perucho,ykovalev,[email protected] withtheunderlyingjetflowandthejet-ambientmediumcan 3AstroSpaceCenterofLebedevPhysicalInstitute,Profsoyuznaya84/32, besimulated(Agudoetal.2001).Withthesenewtechniques, 117997Moscow,Russia 4AstronomyDepartment, DepartmentofPhysicsandAstronomy, Deni- itisnowpossibletocomparethegeneration,propagationand sonUniversity,Granville,OH43023,U.S.A.;[email protected] evolutionofemissionfeaturesinsimulatedandobservedrel- 5InstitutodeAstrofísicadeAndalucía(CSIC),Apartado3004,E-18080 ativisticjets. Granada,Spain;[email protected] The nearby (z=0.049)9 broad-line radio galaxy 3C111 6National Radio Astronomy Observatory, 520 Edgemont Road, Char- (PKS B0415+379) shows a classical FRII morphology on lott7eAsvsitlrloe,noVmAy2D29e0p3a,rtUm.eSn.At,;[email protected],AnnArbor,MI48109- kiloparsec-scalesspanningmorethan200′′withahighlycol- 1042,U.S.A.;mfa,[email protected] limated jet connecting the central core and the northeastern 8Department of Physics, Purdue University, 525Northwestern Avenue, WestLafayette,IN47907,U.S.A.;[email protected] 9AssumingH0=71kms- 1Mpc- 1,ΩM=0.3,ΩΛ=0.7(1mas=1.0pc). 2 Kadleretal. lobeinpositionangle63◦whilenocounterjetisobservedto- wards the southwestern lobe (Linfield&Perley 1984). This asymmetry is usually explained via relativistic boosting of the jet and de-boosting of the counter-jet. 3C111 exhibits the brightest compact radio core at cm/mm wavelengths of all FRII radio galaxies, a blazar-like spectral energy distri- bution (Sgueraetal. 2005), and it was one of the first (and only) radio galaxies in which superluminal motion was de- tected (Goetzetal. 1987; Preuss,Alef,&Kellermann1988). Moreover,the (sub-)parsecscale jetof3C111isintimately relatedtoitshigh-energyemission: Marscher (2006)reports a disk-jet connection, similar to the well-established one in 3C120(Marscheretal.2002),inthesensethatdipsintheX- ray light curve indicate accretion events which are followed by VLBI jet component ejections. Recently, R.C.Hartman & M. Kadler (in prep.) showed that the gamma-ray source 3EGJ0416+3650 can be decomposed into multiple individ- ual sources inside the EGRET full-band point-spread func- FIG. 1.— University of Michigan Radio Astronomy Observatory light tion,revealinga significantsignalfromthenominalposition curvesof3C111at4.8GHz,8GHz,and14.5GHz.Theshadedareasindicate of 3C111 in the higher-resolution, high-energy band above theejectionepochsoftheindividuallylabeledjetcomponentsasdiscussed inSect.3.2. Thelightestshadingcorrespondstominorejectionsoftherel- 1GeV. This association of 3C111 with 3EGJ0416+3650, ativelyweakcomponentsI,J,K,LwithfluxdensitiesSbelow0.2Jy,medium whichhadoriginallybeensuggestedbyHartmanetal.(1999) shadingcorresponds tocomponents B,C,G,H,M,Nwith0.2Jy<S<0.6Jy and Sgueraetal. (2005), makes 3C111 one of the very andthedarkestshadingtocomponentsEandFwithS>0.6Jy. rare radio galaxiesdetected at gamma-rayenergiesand sup- ports the view that this source may be considered a lower- luminosityversionofpowerfulradio-loudquasars. Here, we report the results from ten years of Very-Long- Baseline Interferometry (VLBI) observations of 3C111 as part of the VLBA 2cm Survey10 (Kellermannetal. 1998; Zensusetal. 2002; Kellermannetal. 2004; Kovalevetal. 2005) and its follow-up program MO- JAVE11 (Lister&Homan 2005; Homan&Lister 2006). We investigate the parsec-scale source structure during a major flux-density outburst and during its aftermath. We find that this outburst was associated with the formation of an exceptionallybrightfeatureinthejetof3C111. Avarietyof processes (beyondthe predictionsof simple ballistic motion models) are observed and discussed in view of modern relativistic-jet simulations. In Sect. 2, our observations and the data reduction are described. A detailed report of the observationalresultsisgiveninSect.3. InSect.4,wediscuss thevariousprocessesobservedinthejetof3C111asaresult FIG. 2.—Spectral-indexcurvesof3C111between14.5GHzand8GHz of the outburst and during the propagation of the new jet (top),and8GHzand4.8GHz(bottom)fromtheUMRAOmonitoringpro- featurealongthejet. InSect.5,weputtheseresultsintothe gram.TheshadedareasarethesameasinFig.1. contextof future simulations and observationswith the goal ofunderstandingtheproductionmechanismsofAGNjets. atthevariousepochsaregiveninTable2. Themodelswere alignedbyassumingthewesternmostcomponent(namely,the 2. OBSERVATIONSANDDATAANALYSIS “core”)tobestationarysothatthepositionofjetcomponents 3C111hasbeenmonitoredaspartoftheVLBA2cmSur- can be measured relative to it. Because of the coupling of vey program since April 1995. The observational details thefluxdensitiesofnearbymodelcomponents,theuncertain- are given by Kellermannetal. (1998). Following the meth- tiesinthecomponentfluxdensitiesarelargerthantheformal ods described there, the data from 17 epochs of VLBA ob- (statistical) errors unless the given model component is far servations of 3C111 between 1995 and 2005 (see Table 1) enoughfromitsclosestneighbor. Throughoutthispaper,er- were phase and amplitude self calibrated and the brightness rorsof 15% are assumed for the flux densities of individual distribution was determined via hybrid mapping. An addi- model-fitcomponents. In mostcases, this shouldbeconsid- tional epoch from June 2000 was made available to us by ered a conservative estimate that accounts for absolute cal- G.Taylor. Thepolarizationcalibrationwasperformedasde- ibration uncertainties and formal model-fitting uncertainties scribed in Lister&Homan (2005). Two-dimensional Gaus- (see,e.g.,Homanetal.2002). Positionuncertaintieswerede- sian componentswere fitted in the (u,v)-domainto the fully terminedinternallyfromthedeviationsofthedatafromlinear calibrated visibility data of each epoch using the program motion. DIFMAP (Shepherd1997). Theparametersofeachmodelfit 3. RESULTS 10http://www.cv.nrao.edu/2cmsurvey/ 3.1. The1996RadioOutburstof3C111 11http://www.physics.purdue.edu/MOJAVE TheTrailsofSuperluminalJetComponentsin3C111 3 FIG. 3.— Naturallyweightedimagesoftheparsec-scalejetof3C111fromthe2cmVLBAmonitoring. Acommonrestoringbeamof(0.5 1.0)masat P.A.0◦wasused.Thetotalrecoveredfluxdensityineachimage,thermsnoise,andthelowestcontoursforeachimagearegiveninTable1.Cont×oursincrease logarithmicallybyafactorof2.OnlycomponentsE,G,Handtheircorrespondingtrailingcomponentsareindicatedbycirclesenclosingacross. Astrongfluxdensityoutburstof3C111occurredin1996, the two lower frequencies, the flux-density maximum was which was first visible in the mm band and some months reachedatsubsequentlatertimes, inmid1997at8GHz and later at lower radio frequencies. This outburst was first de- in late 1997 at 4.8GHz. The profile of the outburst in the tected at 90GHz with the IRAM interferometer at Plateau flux-densityvs. timedomainshowsanarrow,high-amplitude de Bure in January 1996 with flux densities greater than peakbetweenearly1996andlate1997,whichisalmostsym- 10Jy (Alefetal. 1998), at 37GHz in March 1996, and at metric.Afterlate1997,aslower-decreasingcomponentdom- 22GHzinAugust1996withtheMetsähoviradioobservatory inatesthelightcurves,mostclearlyvisibleat14.5GHz. (Teräsrantaetal.2004). Figure1showsthesingle-dishradio Figure 1 shows that the flare propagated through the lightcurvesof 3C111at4.8GHz, 8GHz, and14.5GHz ob- spectrum as qualitatively expected by standard jet theory tained from the UMRAO radio-flux-densitymonitoring pro- (e.g.,Marscher&Gear1985):high-frequencyradioemission gram (Aller,Aller,&Hughes 2003). These data show that comesfromthemostcompactregionsofthejet,theemission from early 1996 on, the radio-flux density of 3C111 was peak shifts to lower frequenciesas a newly ejected jet com- rising at 14.5GHz, reaching its maximum in late 1996. At ponenttravelsdownthe jet andbecomesoptically thin. The 4 Kadleretal. peakfluxdensityshiftedwithfrequencyatabout10GHzyr- 1. The evolution of the spectral index, α (S ∼ να), for 15 AB (14.5/8.0)GHzand(8.0/4.8)GHzisshowninFig.2. Before C D 1996,thesamplingwastoosparsetoderivethechangeofthe E E1 spectralindexinthe(14.5/8.0)GHzband. Between,8.0GHz E2 i1wann9igdt9h6t4ha.ce8omGrparrHexesi-zpm1,o9utnh9mde6essdpppeteerocciottarrdaas.lluiibTnnsddheeeeqxxur,eawαndati∼soflaafl0pt,utpeixrnno-idxtnheigmenos(aif1ttye4thl.yo5eu/-s8tp0b.0e.u7)crGtsdrtuHuimrnz- Distance [mas]10 EEFGHH341 band reached in mid 1996. In the post-outburst period be- Core HH23 tween 1998 and 2004, α was typically in the range - 0.5 5 H4 I to - 0.7 between 14.5GHz and 8.0GHz and slightly steeper J (- 0.7to- 0.9)inthe(8.0/4.8)GHzband. Theoverallsteeper KL spectralindexat lowerfrequenciescan be understoodas the M N contribution of optically thin large-scale emission from the 0 radiolobesof3C111tothesesingle-dishlightcurves. 1995 2000 2005 Epoch [yr] FIG. 4.— Core separation ofmodel-fit components vs. time. Crosses 3.2. VLBAMonitoringResults representcomponents whichcouldnotbecrossidentified overtheepochs. Figure 3 shows the variable parsec-scale structure of Positionuncertaintieshavebeenestimatedfromtheinternaldeviationsofthe datafromlinearmotionsforeachcomponent.Theuncertaintiesfortheposi- 3C111at18differentepochsofVLBAobservationsbetween tionofE1havebeendeterminedseparatelyforthepre-2004andpost-2004 1995.26 and 2005.73. The variable source structure can be epochsbecauseofthepartialresolutionofthiscomponentafter2004.Uncer- describedbyaclassicalone-sidedcore-jetmorphologyinthe taintiesforcomponentswithlessthanthreeepochswereestimatedfromother firsttwoepochswithtypicalvelocitiesoftheoutwardmoving componentsatsimilarpositionsinthejetandwithsimilarfluxdensities. jetcomponentsofabout1.4to1.7masyr- 1 correspondingto about5c. In1996.82anewjetcomponent,evenbrighterthan the core, dominated the source structure. By 1997.19, this C new componentwas even brighter (∼3.4Jy) and in the fol- 1 C CCAore C C C C C C C C C CC C C C CC C lowingepochsittraveledalongthejetwhileitbecamegradu- B allymorestretchedoutalongthejet-ridgeline. C D y] E mMododeellfiFticttoinmg:p—oneInntsFfirgo.m4,tthheecroardeiaisldshisotwannceasoaftfhuencvtaiorinouosf nsity [J 0.1 EEE123 time. Thecomponentidentificationwasbasedonacompari- ux De EFG4 sonofthepositionsandfluxdensities,andalinearregression Fl H H1 ofthedistancesfromthecoreasafunctionoftimewasused 0.01 H2 H3 todeterminethekinematics. Thederivedcomponentveloci- H4 tiesaretabulatedinTable3. Theearlyouterjetcomponents (A,B,C, D)ofthe1995.26epochcanbetracedovertwoto 0.001 fourepochsbeforetheirfluxdensitiesfallbelowthedetection 1995 2000 2005 threshold(compareFig.5). Inlate1996andearly1997,the Epoch [yr] source structure was dominated by the emission of the core FIG. 5.—Flux-densityevolutionofthecoreandthejetcomponentswith andthenewlyformedjetcomponentsE andF, with E being time. Forclarity, onlycomponentsejectedbefore2001.50areshown. The the leading component. The two components traveled out- fluxdensitiesofE1,E2,E3,andE4wereaddedforthepost-1999epochs wardswithameanapparentvelocityof(1.00±0.02)masyr- 1 andaflux-densityweightedeffectivepositionwascalculated todisplaythe flux-densityevolutionoftheblendedfeaturethatwouldbevisibleatlower and (0.64±0.07)masyr- 1, respectively. Before mid 1997, resolution. NotethatcomponentsE4andGareblendedinepoch2004.27 component F was substantially brighter than component E andthatthefluxdensityofE4maybeoverestimatedforthisepoch. butafterthat,itsfluxdensitydroppedsteeply. Fwasnotde- tectedatanyepochlaterthan1998.18,whileEwasstillabout wasseen from2004.80on. In thefollowing,we referto the 800mJy atthattime. Thelightcurvesof E andFreproduce componentsE1,andH1asthe“leadingcomponents”andto qualitativelythetwo-componentshapeoftheflux-densityout- E2, E3, E4, H2, H3, andH4 as the “trailing components” burst in Fig. 1 with component F being responsible for the ofEandH,respectively. narrowerandhigher-amplitudepeakbetweenearly1996and For the pc-scale jet of 3C111, the ejection epochs of the late1997andcomponentEdominatingtheslower-decreasing individualjetcomponentscanbedeterminedfromthelinear tail of the outburst after late 1997 (compare Fig. 5 and dis- regression by back-extrapolating the component trajectories cussionbelow).Inthefollowingepochs,Esplitintofourdis- to the core. In Fig. 1, these ejection epochs and the associ- tinct components(E1, E2, E3, E4) at distances of 3.5mas ateduncertaintiesareindicatedasshadedareas. Itisapparent to4.5masfromthecore. E2,E3,andE4allmovedatsub- that the ejection of the components E and F coincides with sequently slower speeds than E1, resulting in an elongated theonsetofthemajorflux-densityoutburstin1996described morphologicalstructureoftheassociatedemissioncomplex. above. Thefollowingmajorcomponentejections(G,H,and In later epochs, new components have been ejected from thecombinedM/Nevent)allhavedirectcounterpartsinlocal the core into the jet. The two strongest components (G,H) maximaoftheradiolightcurve,especiallyat14.5GHz. Fig- canbetracedthroughthefollowingeightandninemonitoring ure2showsthatalltheseejectionepochscoincidedwithlocal epochs,respectively.ComponentHsplitintothreeindividual maximaof the spectralindexin the 14.5/8.0GHz band. Be- components in 2004.27 and a fourth associated component tween2002and2004,anumberofminorcomponentejections TheTrailsofSuperluminalJetComponentsin3C111 5 1e+12 E (pre 1999) E (post 1999) F 1e+11 1 E1 E2 Flux Density [Jy] 0.1 EE34 ghtness Temperature [K]111eee+++100098 0.01 Bri 1e+07 0.001 1e+06 0 5 10 15 1 10 Core Distance [mas] Core Distance [mas] FIG. 6.—Flux-densityevolutionofcomponentEanditstrailingcompo- FIG. 8.—Brightnesstemperaturesofmodel-fitcomponentsasafunction nentsvs. thedistance traveledfromthecore. NotethatcomponentE4is of their distance to the core. The brightness temperatures of components blendedwithcomponentGinepoch2004.27andthatitsfluxdensitymaybe belongingtotheE–,F–andH–componentsareindicatedbyfilledblackcir- overestimatedforthisepoch. cles. Thesolidlinerepresents aleast-squares fittoallbuttheE–,F–and H–components.Theslopeoftheregressioncurveis- 2.4 0.2. ± G H H1 E H2 1e+11 E1 1 H3 E2 H4 E3 E4 K]1e+10 F Flux Density [Jy] 0.1 ness Temperature [1e+09 I ht1e+08 0.01 g Bri 1e+07 II 0.001 0 5 10 15 1e+06 Core Distance [mas] 1 10 Core Distance [mas] FIG. 7.—Flux-density evolution ofcomponentGandHanditstrailing componentsvs.distancetraveledfromthecore. FIG.9.—BrightnesstemperaturesofcomponentE,itsleadingandtrailing components, and component F as a function oftheir distance to the core. Thetworegimesofbrightness-temperaturegradientdiscussedinthetextare took place but the regression-fit quality (due to the neareby indicatedwithdashedlines. components, the low flux densities and the small time base- line)onlymoderatlyconstrainstheejectionepochs. Inaddi- the factthatitdoesnotappearto have splitinto leadingand tion,thetimesamplingofUMRAOobservationsinthistime trailing components like E and H, did exhibit a pronounced range is relatively poor, in particular from mid 2001 to mid fluxdensitymaximumafter2002,aswell, about6masfrom 2003. thecore. FluxDensityEvolution:— Figure5showsthebrightnessevo- The Tb gradient along the jet:— Following Kadleretal. lution of the core and the jet components that have been (2004); Kadler (2005), the power-law index s, which de- ejected priorto 2001.5. Apparently,the trailing components scribes the brightness temperature gradient via Tb ∝rs, can E4 and H4 appearedfirst in a rising state, i.e., theyfirst in- beparametrizedas creased in flux density before they became fainter in later s = l+n+b(1- α) (1) epochs. Component F showed an extraordinary steep de- crease in brightnessin 1997–1998. In Fig. 6 and Fig. 7, the where l, n and b are the power law indicesthat describe the flux-densityevolutionofthecomponentsE,G,andHandthe gradientsofjetdiameterd∝rl,particledensityn ∝rn,and e associated leading and trailing components are shown with magneticfield B∝rb with distance r from the core, respec- distancetraveledfromthecore,respectively. Theejectafirst tively. Therefore,measuringthebrightnesstemperaturegra- roseinfluxdensitywithintheinner1masfromthecore,then dientprovidesamethodtoconstrainthecriticalphysicalprop- they showed a decline about almost three orders of magni- ertiesalongthejetandabruptchangesintheT -gradientcan b tudeinthefollowingdecade,exhibitingaplateauorbroadlo- highlightregionsinthejetwherethedensity,magneticfield, calmaximumin1998–2000atadistancefrom2–4masfrom orjetdiameterchangerapidly. the core. ComponentH and its leading and trailing compo- Figure8showsthebrightnesstemperaturesofalljetcom- nentsexhibitedasimilarbehavioralthoughonaboutanorder ponentsintheparsec-scalejetof3C111at2cmwavelength ofmagnitudelowerflux-densitylevelsandatslightlyfurther between 1995 and 2005as a functionof their core distance. downstream,4–6masfromthecore.ComponentG,inspiteof Ingeneral, thebrightnesstemperatureofallcomponentsde- 6 Kadleretal. creased as the componentstraveledoutwardsbut an approx- imationwithasimplepowerlawdoesnotyielda goodfitto H thefulldataset(χ2 =1.8,115degreesoffreedom[d.o.f.]). red Oct. 2002 Visual inspection of Fig. 8 shows that this is due to the E–, F–andH–componentsandtheirleadingandtrailingcompo- nents, respectively. This behavior is different than expected for a straight and stable jet geometry in which the power- H law dependences of the particle density, the magnetic field strengthandthejetdiameteronthecoredistancepredictsthat Aug. 2003 thebrightnesstemperaturealongthejetcanbedescribedwith awell-definedpower-lawindexs. Mostextragalacticparsec- H scale jets which do not show pronounced curvature, show a power-law decrease with increasing distance from the core and power-law indices typically around - 2.5 (Kadler 2005). April 2004 Infact,excludingtheE–,F–andH–componentsfromthefit yields a statistically better result (χ2 =1.3, 52 d.o.f.) and H red a gradient of - 2.4±0.2. In Sect. 4.1 and Sect. 4.5, we dis- cusspossible physicalreasonsforthe differentbehaviorand I Oct. 2004 natureof thesecomponents. Themeasuredrelationbetween component sizes and distance along the jet is affected by a large degreeof scatter and doesnot provideindependentin- H formation from the flux density and brightness temperature I plots described above. Therefore we do not show plots of Jan. 2005 componentsizeversusjetdistance. The brightness-temperature gradient of component E was H firstflatorinvertedimmediatelyafterthecreationofthisnew component within approximately 1mas from the core and then reached steep values of - 2.5 to - 2.8 (regime I; com- Sept. 2005 pareFig.9)through1997whenthecomponenttraveledfrom 1masto2mas. Between2masand4mas,thedetermination ofthebrightness-temperaturegradientrequiresanidentifica- tion of component E with either component E1 or E3 (see below). Independently of this identification, the brightness- temperaturegradienteventuallychangedtoverysteepvalues FIG. 10.—Naturallyweightedimagesofthelinear-polarization structure (s<- 5)beyond5masfromthecore(regimeII).Component of3C111between2002and2005. Therestoring-beamdimensionsandori- entationsforeachepochareindicatedbyacrosstotheleftofeachStokes-I F began its very rapid decline in brightness temperature at image. StokesIcontoursstartat1mJy/beamandincreasebyfactorsof2. a very small distance from the core (<0.7mas) with an ex- Fractionalpolarizationisover-plottedontheStokes-Icontoursincolor. To tremelysteepTb-gradient(s<- 8). therightoftheStokes-Iimagesarethepolarizationintensitycontoursstarting at1mJy/beamandincreasingbyfactors√2. Thepolarizationcontoursare LinearPolarization:— From1995to2002,2cm-Surveyob- over-plottedwithtick-marksrepresentingtheelectricvectorpositionangle. Asingle Stokes-I1mJycontour surroundsthepolarization imagetoshow servations were done in left circular polarization only, so registration. The dotted line marks the distance of 3.3mas from the core nolinear-polarizationinformationcan bederivedfromthese wherethemostpronouncedchangesofthepolarizationpropertiestakeplace data. MOJAVEobservations(from2002on)aredoneinfull- (seetext). polarimetric mode. Figure 10 shows our polarization data ofHwasapproximatelyperpendiculartothejet. throughSeptember2005. After component H passed through this region (epochs InOctoberof2002,componentHwasabout2.5masfrom April2004throughSeptember2005),itsplitintoanumberof thebaseofthejetandshowedafractionalpolarizationof5% subcomponentsasdescribedearlier,anditspolarizationgrad- to10%increasingtowardsthedownstreamsideofthecompo- ually became more uniform. Consistent fractional polariza- nent. Theelectricvectorpositionangle(EVPA)displayedby tion of 5% to 10% was approachedwith the electric vectors thecomponentwasapproximatelyalignedwiththejet. Inthis approximatelyperpendiculartothelocaljetdirection. epoch, the jet material just downstream of component H at ∼3.3masfromthebaseofthejetwasmorehighlypolarized, The much weaker component, I, developed polarization verysimilartoHasitpassedthroughthesameregion,about exceeding 20% fractional polarization on the jet’s southern side,andtheEVPAofthepolarizationturnedtobeabout45◦ 3.3masfromthecore,withfractionalpolarizationexceeding 20%towardthesouthernsideofthejetwithanEVPAatap- tothemainjetdirection. proximately 45◦ to the main jet axis. This is also the same ByAugustof2003,componentHhadenteredaregionap- regionofthejetinwhichcomponentEhadbrokenupintoa proximately3.3masfromthecoreanditsobservedpolariza- number of sub-components. In future epochs, we will have tionwasnowsimilartotheemissionin thissameregionob- the opportunity to follow component K as it passes through servedinthepreviousepoch. Theobservedfractionalpolar- thissameregion. izationofHnowclimbedsharplytovaluesinexcessof20% toward the jet’s southern side while there was no detectable 4. DISCUSSION polarizationfromthenorthernsideofH.TheobservedEVPA ofHhadrotatedfurthertobeapproximately60◦ tothemain In this section, we discuss the aftermath of the major out- jetdirection. However,the EVPA onthe southern-mostside burst in 3C111 in 1996 and the following component ejec- TheTrailsofSuperluminalJetComponentsin3C111 7 tionsthrough2005. We organizethe subsectionsof our dis- requiresan accelerationof this component(see Fig. 4) from cussionaccordingtothedownstreamdistancefromtheVLBI β =3.26±0.07toβ =5.5±0.1between1998.18and app,E app,E1 corewhereweobservetheeffectofinterest. 1999.38.Thismaybeinterpretedintermsofanexpansionof the jet in a rarefied medium. Taking an angle to the line of 4.1. Within1pc: ForwardandReverseStructures sightof 19◦ (see AppendixA), the componentwouldbe ac- Numerical simulations (Aloyetal. 2003) show that an celeratedfrom β =0.956(γ∼3.4) to β =0.995(γ ∼10.3). abrupt perturbation of the fluid density at the jet injection Theincreaseofvelocityislessatsmallerviewingangles. point during a short time propagates downstream, evolves Analternativemodelfortheaccelerationandbrighteningof spreadingasymmetricallyalongthejetandfinallysplitsinto componentE would be a changeof the jet inclinationto the twodistinctregions.Bothofthesetworegionshaveenhanced lineofsightfromabout24◦toabout11◦atthislocationinthe energy density with respect to the underlying jet, and they jetasobservedinthecaseofthequasar3C279(Homanetal. emitsynchrotronradiation. Theleading(forwardshock)and 2003). However,weseenosignificantchangeinjetposition the following region (reverse shock) have higher and lower angleintheskywhichwouldbeexpectedtoaccompanysuch Lorentz factors, respectively, than the underlying jet. Thus, a large change in jet direction. Moreover, subsequent com- theyshouldseparatewithtimeastheypropagatedownstream ponents, particularly G, do not show the same kind of large inthejet. accelerationinthisregion. ComponentFmatchesthedescriptionofabackwardmov- Direct identification of component E with component E1 ing wave associated with the major injection into the jet of is not straightforward in the frame of expansion, as compo- 3C111 after the flux-density outburst of 1996. It follows nentE1 in epoch 1999.38was smaller than componentE in the trail of component E but at a lower speed. If compo- 1998.18 (see Table 2). However, component E3 in epoch nent F is identified with a reverse shock and component E 1999.38is larger than componentE in 1998.18. We can in- with a forwardshock,it ispossiblethen tocomputethe size terpretthisascomponentEincludingcomponentsE1andE3 of the shocked region (Peruchoetal. 2007). In 1996.82and (andmaybeE4). Thesecomponentswouldbeindistinguish- 1997.19,EandFwerebothverybrightandseparatedbyonly ableinourobservationsbefore1999.38.Infact,Jorstadetal. ∼0.3pc in projected distance. During these two epochs, F (2005) monitored 3C111 between 1998 and 2001 with the was 300mJy to 500mJy brighter than the leading compo- VLBA at 43GHz. They find an emission complex, that can nent E. Following Aloyetal. (2003), a backward shock can be identified with our componentE, that graduallystretches be brighter than a forward shock if the latter is beamed in a out as it travels from ∼ 2 mas from the core in 1998 to cone smaller than the viewing angle due to its larger speed. roughly between 5mas and 8mas from the core in 2001. WehaveexaminedtheDopplerfactorsofcomponentsEand TheirleadingcomponentC1canbeidentifiedwithourcom- Ffortherangeofpossibleviewingangles(seeAppendixA) ponent E1, their component c2 with E2 and their c1 with andthemeasuredvelocitiesandconcludethatthisalonecan- E3. At their higher angular resolution, Jorstad et al. can not explain the brightness difference between component E separate components C1 and c1 already in early 1998. In and F because the difference in apparent speed is not large agreement with our analysis at 15GHz, they detect c2 (E2) enough. Jorstadetal.(2005)pointoutthatbackwardshocks about a year after they detect c1 (E3). They do not detect can be brighter than forward shocks as long as the distur- a component corresponding to E4 but this may be an effect bance is prolonged and there is a continuous supply of par- of partially resolvingout the jet structure at their higherob- ticles enteringfrom the underlyingjet throughthe shock re- serving frequency, particularly in later epochs. It is further gion.Withinhalfayear,between1997.19and1997.66,Flost interesting to note that the observed speeds at both frequen- abouthalfofitsbrightness.Thisextraordinarilyfastdimming cies agree well. For E1(C1), µ =1.69±0.04masyr- 1 app,2cm ofthebackwardshockcanbecausedbythelackofinputof at 15 GHz and µ = 1.77±0.06masyr- 1 at 43 GHz; app,7mm particlesfrombehind,i.e., a lowerplasmaejectionrateafter for E2(c2), µ = 1.29±0.06masyr- 1 at 15 GHz and app,2cm theprimaryinjectionpossiblyduetoadepletionoftheinner µ =1.23±0.04masyr- 1 at 43 GHz; and for E3(c1), accretiondisk (Marscheretal. 2002;Marscher 2006) which app,7mm µ = 1.22±0.05masyr- 1 at 15 GHz and µ = feedstheplasmainjection. app,2cm app,7mm ComponentFcanalsobeinterpretedasararefactionpropa- 1.07±0.02masyr- 1at43GHz.Thediscrepancyinthespeeds gatingbackwardsinthereferenceframeoftheejectedblobof measuredforE3andc1seemstobeduetoaslightaccelera- gas.Ararefactionisproducedwhentheblobisoverpressured tionofE3after2002.Afittothe15GHzdataofE3between with respect to the jet, as this overpressure causes the front 1999and2002aloneyieldsaslowerspeedof∼1.0masyr- 1 toaccelerateinthejet,thusleavingararefiedregionbetween similartothespeedofc1inthesametimeperiodat43GHz. theheadoftheblob(forwardshock)anditsrearpart,whichis Intheirwork,Jorstadetal. donotreportaccelerationofcom- stillslower(itmoveswiththeinjectionvelocity).Inthiscase, ponentsfrom2masto4mas.However,thisislikelyduetothe theemissionincomponentFcouldbeassociatedtothedenser factthattheirobservationsstartedinearly1998,thusmissing andoverpressuredgasintheblobwhichhasstillnotbeenrar- thefirstobservationsofcomponentEpresentedinthispaper, efied. Thisgaswouldceasetoemitassoonasitreachesthe whenitsspeedhasbeenmeasuredtobesmaller. rarefiedregion,which mayalso explainthe suddendecrease 4.3. Between2pcand6pc: RecollimationoftheJet in brightnessof this component. An extendeddiscussion on thenatureofcomponentFandtheevolutionofitsbrightness Inspection of Fig. 9 shows that the back-extrapolation of willbegiveninPeruchoetal.(2007). thebrightnesstemperatureofcomponentEfromregimeIIto regime I is at least two orders of magnitude too high if this 4.2. Between2pcand4pc: ExpansionandAcceleration extrapolationis basedon the gradientgivenby E1. Thelow Itisnota-prioriclearwithwhichpost-split-upcomponent brightnesstemperatureofcomponentEinregimeIcannotbe theoriginalfeatureEshouldbeidentifiedafter1999.Anatu- explainedby opacityeffectsbecausethe radio-lightcurvein ralidentificationwouldbetheleadingcomponentE1butthat Fig.1showsthatthesourcewasopticallythinfrom1997on. 8 Kadleretal. Moreover,ifweidentifycomponentEwithE1,itisDoppler- ering an inhomogeneous external Faraday screen. Such a deboostedfromepoch1998.18to1999.38duetotheaccelera- screen could produce the observed differential rotation of tionandarelativelylargeviewingangle;thus,wearenotable the EVPA while a componenttravelsthrougha givenregion to explain the increase in brightnesstemperaturein terms of alongthe jet. Zavala&Taylor (2002) observed3C111 with Doppler boosting. However, compact sub-components may theVLBAandproducedaFaradayrotation-measuremapbe- havelargerbrightnesstemperatures,sothattheT valuesplot- tween8GHzand15GHz. TheyfindstrongFaradayrotation, b ted in Fig. 9 for E in regime I inward of about 3mas may ∼730radm- 2, at the same distance from the core (3.3mas) represent lower limits for compact components already em- where our observations show the swing of the EVPA of the beddedintheunresolvedstructure. componentHandsteeplydecreasingFaradayrotationfurther Not only E/E1 but also components G and H show downstream. However,wenotethat730radm- 2 translatesto an increase in total flux density several milliarcseconds 17◦ of rotation at 15GHz which alone is not enough to ex- downstream. Compared to E/E1, these somewhat weaker plainthe changein EVPA thatwe observewhile component components exhibit their flux-density maxima at somewhat Htravelsthroughthisregion.Ontheotherhand,thesteepde- larger distances from the core (compare Fig. 6 and Fig. 7). creaseoftheFaradayrotationmeasuredup-anddownstream This can be explained if the gas in the components trav- ofthisregionbyZavala&Taylor(2002)againagreeswitha els through a mild standing shock in a recollimation re- changeoftheexternalgasdensityatthispoint,whichinturn gion. This effect has been observed in numerical simula- may be identified with the pressure gradient responsible for tionsof parsec (Gómezetal. 1997) and kiloparsecscale jets thecomponentexpansionandaccelleration. (Perucho&Martí 2007). The material in the componentsis AcombinationofinhomogeneousFaradayrotationandan expectedtobeoverpressuredwithrespecttoitsenvironment, interaction between the jet plasma and its ambient medium thus expandinginto it. After the initial expansion, the com- appears most likely to explain our observations of the vary- ponentsbecomeunderpressuredwith respect to the underly- ing linear polarizationstructure; however, both explanations ing flow. The resulting recollimation leads to the formation pointtotheroleoftheexternalmedium,eitherthroughadis- of a shock, whose strength depends on the initial degree of creteinteractionorarapiddecreaseinexternalgaspressure, overpressureof the materialin the component. Thisprocess inshapingthejetflowdownstreamofthislocation. explainstheincreaseinfluxdensityandbrightnesstempera- 4.5. Between3pcand5pc: FormationofTrailing tureasduetocompressionofthegasintherecollimation. In Components Figs.6and7,weseethatthefluxdensityofcomponentEin- creasesclosertothecorethanforcomponentGandH,which ThecomponentsE2,E3,andE4canbeinterpretedastrail- isconsistentwiththeformerbeingslowerthanthelatter,thus ingcomponentsforminginthewakeoftheleadingE1which recollimatingearlier(seePerucho&Martí2007). Italsoex- is identical with the original component E. This scenario is plainswhyweseeasignificantaccelerationonlyinthefaster attractive because the basic concept of trailing components expanding,brightercomponentE/E1. as introduced by Agudoetal. (2001) predicts the formation Finally, after this mild recollimation, the fluid becomes of trailing features in the wake of the initial perturbation in overpressuredwithrespecttoitsenvironment,thusfurtherex- the jet flow. Such a behaviour has first been found both as- pandingandacceleratingdownstream. sociatedwithbrightsub-andsuperluminaljetcomponentsin Centaurus A and 3C 120 (Tingay,Preston,&Jauncey 2001; 4.4. Near3pc: TheRoleoftheExternalMedium Gómezetal.2001). Jorstadetal.(2005)reporttrailingcom- The polarization behavior of components H and I can be ponentsinfouradditionalsources(3C273,3C345,CTA102 understoodintermsofaninteractionbetweenthejetandthe and3C454.3)andin3C111(seebelow). externalmediumatadistanceof 3.3mas(=3.3pc)inthejet. Theinteractionoftheexternalmediumwithastrongshock Assuming no Faraday rotation, the EVPA of component H pinches the surface of the jet, leading to the production of withinapproximately3.3masfromthecoreindicatesatrans- pinch-bodymodeKelvin-Helmholtzinstabilities:thetrailing verse magnetic field order as might be expected for a trans- features. Hence,a singlestrongsuperluminalshockejection verse shock propagating down the jet. The change in the fromthe jet nozzlemay lead to the productionof a multiple fractional polarization, its north-south gradient, and the ro- setofemissionfeaturesthroughthismechanism.Thetrailing tation of the EVPA suggest that a contactsurface persists at features have a characteristic set of properties, which make the southern boundary of the jet beam at a distance of ap- them recognizable with high resolution VLBI: they form in proximately3.3mas downstreamthe jet core. If the bulk jet thewakeofstrongcomponentsinsteadofbeingejectedfrom materialflows faster than the flow at the southernboundary, thecoreofVLBIjets,theyarerelatedtoobliqueshocks,they themagneticfieldisstretchedthroughshear. Ouroverallpic- arealwaysslowerthantheleadingfeature,and(iftheunder- turethenisofanoriginallytransverseshockinteractingwith lyingjethasacertainopeningangle)theyshouldbegenerated the jet on the southern side of the jet at 3.3mas from the withawiderangeofapparentspeeds(fromalmoststationary core.Theinteractionchangesthecomponent’smagneticfield nearthecoretosuperluminalfurtherdownstream).Moreover, throughsome combinationof oblique shock and differential Agudoetal. (2001) showed that the separation between the flow resulting in a magnetic field approximately parallel to trailing components increases downstream due to their mo- thejetaxisinthelaterepochs. Nostrongshockisneededat tiondownapressuregradient. thislocation in the jetbut thisregionmaybe identifiedwith Allthisis in agreementwith whatwe observein thetrail- therecollimationregion(seeSect.4.3)ataboutthesamepo- ing componentsof E1 and with our interpretation of an ex- sition in the jet). In this picture, the jet-medium interaction pansion of the jet in a density decreasing ambient medium. mayformaneffectivenozzlewhichacceleratesthejetonone Forthetimerangecoveredbytheirobservations(1998.23to edgerelativetotheother. 2001.28),Jorstadetal.(2005)alsoidentifiedthetrailingphe- An alternative explanation for the observed polarization nomenologyinthissource. structure and dynamics of 3C111 can be found by consid- The north-south gradients detected in the linear- TheTrailsofSuperluminalJetComponentsin3C111 9 polarization emission in the region where the trailing onthenatureandcharacteristicsoftheseinitialcomponents. features are formed, is in agreement with an oblique shock Severalparsecsdownstream,thejetplasmaentersaregionof structure. The steep brightness-temperaturegradients of the rapidlydecreasingexternalpressure,expandsintothejetam- trailing components indicate that the particle and magnetic bient medium and accelerates. In the following, the plasma field density associated with these components evolve in getsrecollimatedandtrailingfeaturesareformedinthewake a different way compared to the “normal" jet flow. These oftheleadingcomponent. shocked regions may be more overpressured with respect A particularly interesting aspect of the source 3C111 to their environment, making them expand rapidly. This in the light of this and other recent works is that it is fast expansion implies a larger positive value of l, which, one of the very rare non-blazar gamma-ray bright AGN. however, is compensated by an even larger (negative) value Besides CentaurusA (Sreekumaretal. 1999) and the pos- of n and b(1- α) in equation 1, resulting in a very steep sible identification of NGC6251 with the EGRET source brightnesstemperaturegradient(regimeII). 3EGJ1621+8203(Mukherjeeetal.2002),3C111istheonly Pinching modes of the Kelvin-Helmholtz instability were AGN whose jet-system is inclinedat a relativelylargeangle shown to couple to the trailing components observed in the tothelineofsightandthathasa reliableEGRETidentifica- simulations in Agudoetal. (2001). In the case of compo- tion: Sgueraetal. (2005) reconsidered the possible identifi- nentsE2-E4,the distancebetweenthemrangesfrom0.7-0.8 cation of the EGRET source 3EGJ0416+3650with 3C111, masatthefirstepochsinwhichtheyareobserved,toalmost which was first suggested by Hartmanetal. (1999) but con- 2.0 mas in the latest epochs. Taking into account that: a) sidered unlikely because of the poor positional coincidence. theirFWHMisofthesameorder(seeTable2);b)thatthese Very recently, R.C. Hartman & M. Kadler (in prep.) found wavelengths have to be corrected for geometrical and rela- that 3EGJ0416+3650 is composed out of at least two dis- tivistic effects, resulting in a maximum intrinsic wavelength tinctcomponents. Oneofthemisthedominantsourceabove of∼0.7mas,andc)thatthesizeofthecomponentscanbeof 1GeVandisinexcellentpositionalagreementwiththeloca- the order or smaller than the jet radius (Perucho&Lobanov tionof3C111.Comparedtoblazars,thelargeinclinationan- 2007),thisimpliescouplingofthepinchingtowavelengthsof gleandtherelativelysmalldistanceof3C111allowustore- theorderorsmallerthanthejetradius. Peruchoetal.(2007) solvestructuresalongthejetthatareassmallasparsecsinde- have shown that resonant Kelvin-Helmholtz instabilities as- projectionandwhichwouldbeheavilyblendedwithadjacent sociated to high-order body modes appear in sheared jets at featuresinblazarjets. Asdemonstratedinthispaper,VLBA thesewavelengths.Thesemodeshavelargergrowthratesthan observations of 3C111 probe a variety of physically differ- low-order body modes or surface modes, and their growth ent regions in a relativistic extragalactic jet such as a com- bringsthejettoafinalquasi-steadystateinwhichitremains pactcore,superluminaljetcomponents,recollimationshocks well-collimatedandgeneratesahotshear-layerwhichshields andregionsofinteractionbetweenthejetanditssurrounding thecoreofthejetfromtheambientmedium.Interestingly,the medium, which are all possible sites of gamma-ray produc- jetin3C111isknowntobewell-collimateduptokiloparsec tion. From early 2008 on, the gamma-ray satellite GLAST scales. Further research in this direction is needed in order (Lottetal. 2007) is goingto monitorthe sky. If detected by tochecktheinfluenceoftheresonantmodesinthelongterm GLAST,3C111maybecomeakeysourceinthequestforan evolutionofthisjet. understandingof the origin of gamma-raysfrom extragalac- A by-productof the interpretationof these componentsas tic jets. In addition, the combinationof GLAST and VLBA Kelvin-Helmholtzinstabilities is the fact that it allows us to data with spectral data at intermediate wavelengths(optical, put constraints to the velocity of the jet. We can regard IR,X-ray)mayallowabetterdeterminationofjetparameters the wave speed as the minimum speed of the jet flow, as andrelativisticbeamingeffectsthanin mostblazarsbecause KH modes have an upper limit in their wave speeds that is ofthehigherlinearresolutionofferedbythisnearbyandonly precisely the velocity of the flow in which they propagate weaklyprojectedjetsystem. (Peruchoetal.2006).Theupperlimitisgivenbythespeedof Ourobservationsof3C111arequalitativelyinremarkable E1,interpretedasashockwave,thathastobethusfasterthan agreement with numerical relativistic hydrodynamic struc- theunderlyingflow. Inthispicture,wewouldhavethestruc- tural and emission simulations of jets such as the ones pre- tureE1movingwithLorentzfactorγ∼8.3throughajetwith sented by Agudoetal. (2001) and Aloyetal. (2003). Fur- Lorentzfactor8.3>γ ≥4.6in theacceleratedregion(post ther progress is being made in the transition from two- j 1999.38),wherethelowerlimitisgivenbytheLorentzfactor dimensional to three-dimensional simulations of relativistic ofcomponentE2,thefastestofthethreetrailingcomponents jets and in the development of new methods considering identifiedhere. magnetic fields (Leismannetal. 2005; Mizunoetal. 2007; Roca-Sogorbetal.2008,e.g.,),theequationofstateforrela- 5. SUMMARYANDCONCLUSIONS tivisticgases(Perucho&Martí2007),andradiativeprocesses Inthispaper,wehaveinvestigatedtheparsec-scalejetkine- (e.g., Mimicaetal. 2004, 2007, and Mimica et al. in prepa- maticsandtheinteractionofthejetwithitsambientmedium ration). Butsofarneitherobservationaldatanorsimulations in the broad-line radio galaxy 3C111. Our analysis has havereachedanadequatelevelofdetailandcompletenessin demonstratedthatavarietyofprocessesinfluencethejetdy- orderto allow usa quantitativedirectcomparisonof numer- namics in this source: a plasma injection into the jet beam icalmodelsand observedrelativistic jet structure andevolu- associatedwithamajorflux-densityoutburstleadstothefor- tion. Inparticular,itisnotfeasibletodaytofititerativelythe mationofmultipleshocksthattravelatdifferentspeedsdown- parametersofrelativisticmagneto-hydro-dynamical(RMHD) stream and interact with each other and with the ambient jetsimulationsto match the brightnessdistributionobserved medium. The primary perturbation causes the formation of for any individual source. The main reasons for this are a) a forward and a backward shock (or rarefaction). The latter the immense computationalpower required to conducta re- fadesawaysofastthatislikelytoremainundetectedinminor alistic (i.e., sufficiently detailed) modern 3D jet simulation ejections. A separate work by Peruchoetal. (2007) focuses and b) the highly non-linear nature of RMHD plasmas and 10 Kadleretal. their evolution. Simulation results depend critically on the ported under National Science Foundation grant 0406923- starting conditions like the exact velocity, composition, and AST. The Very Long Baseline Array is operated by the Na- transversalstructureoftheflow,thestructureandstrengthof tional Radio Astronomy Observatory, a facility of the Na- the magnetic field and the jet environment. Future develop- tionalScienceFoundationoperatedundercooperativeagree- mentofcomputationalpowerwillallowustouselargerres- ment by Associated Universities, Inc. UMRAO is partially olutionsto decrease the numericalviscosities, and to imple- supportedby a series of grants from the NSF, most recently mentnonlinearandmicrophysicsprocessesintosimulations. AST-0607523, and by funds from the University of Michi- VLBA observations are capable of putting hard quantitative gan. MK has been supported in part by a Fellowship of the constraintsontheinputparametersforRMHDjetsimulations InternationalMaxPlanckResearchSchoolforRadioandIn- if they are densely sampled over several years. Polarimet- fraredAstronomyandinpartbyanappointmenttotheNASA ric observationsat multiple radio frequenciesmay allow the PostdoctoralProgramattheGoddardSpaceFlightCenter,ad- effects of jet-intrinsic magnetic-field variations and external ministered by Oak Ridge Associated Universities through a Faraday-screen inhomogenities or temporal variations to be contract with NASA. MP acknowledges support in part by disentangled.Suchdataat15GHzareontheway,e.g.,aspart theSpanishDirecciónGeneraldeEnseñanzaSuperiorunder ofthenextphaseoftheMOJAVEprogram,inwhichrapidly grantAYA2004-08067-C03-01and in partby a postdoctoral evolving sources like 3C111 are being observed every two fellowshipof the GeneralitatValenciana(BecaPostdoctoral months. d’Excel·lència). YYK is a Research Fellow of the Alexan- der von HumboldtFoundationand was supported in part by the Russian Foundation for Basic Research (project 05-02- We would like to acknowledge the support of the rest of 17377).DCHwassupportedbygrantsfromResearchCorpo- the MOJAVE Team, who have contributed to the data used ration and the National Science Foundation (AST-0707693) in this paper, in particular we would like to thank Christian IA has been supported in part by an I3P contract with the Frommforhishelpwiththeproductionofthefiguresforthe SpanishConsejoSuperiordeInvestigacionesCientíficasand paper. We thankDharamVir Lal and Silke Britzen for their in part by a contract with the German Max-Planck-Institut careful reading of the manuscript and their comments. We für Radioastronomie (through the ENIGMA network, con- also thank the referee for his very constructive suggestions, tractHPRN-CT-2002-00321),whichwerepartiallyfundedby which have helped to improve this paper. We are grateful theEU. to Greg Taylor, who provided complementary 2cm VLBA data for an additional epoch. The MOJAVE project is sup- APPENDIX A. THE JET INCLINATIONANGLE The1996radiooutburstof3C111putsstrongconstraintsontheangletothelineofsightforthissource,ifoneassumesthata similarlybrightcomponentasEhasbeenejectedinthecounterjet,aswell. DuetodifferentialDopplerboosting,thefluxdensity ratiobetweenthejet-andcounter-jetemissionis S 1+βcosθ 2- α J = . (A1) S (cid:18)1- βcosθ(cid:19) CJ Thus,foragivenjettocounter-jetratiox= SJ SCJ x- 1 βcosθ= . (A2) x+1 Withα=0.3,S =3.4Jy(componentsEandFin1997.19),S <10mJyandβ<1,θ<21◦. Forarealisticjetspeedof,e.g., J CJ β=0.956(γ=3.4),theangletothelineofsightis: θ=19◦. Anestimateclosetothisvaluecanbederivedfromthevariability Doppler factor measured by Lähteenmäki&Valtaoja (1999) and the apparentsuperluminaljet speed. As outlined in detail in Cohenetal.(2007),thisleadstoavalueofθ∼15◦. Itisimportanttonotethatthiscalculationimplicitlyassumessymmetrybetweenthejetandcounter-jet,whichinprojection does not have to be the case if the counter-jet is covered by an obscuring torus as it is well-established for systems at larger inclinationangles(e.g.,NGC1052:see Kadleretal.2004). Indeed,Faradayrotationmeasurementstowardsthe3C111pc-scale jet (Zavala&Taylor 2002; see also Sect. 4.4) and X-ray spectral observations(Lewisetal. 2005) suggest substantialamounts ofobscuringmaterial. Free-freeabsorptioncouldalsosubstantiallylowerthecounter-jetradioemissionandallowforlargerjet anglestothelineofsight. Anindependentlowerlimitontheinclinationangleofθ>21◦wasgivenbyLewisetal.(2005)assumingthatthedeprojected size of the largescale 3C111 double-lobestructure is smaller than 500h- 1kpc. This discrepancy implies that either 3C111 is unusuallylargeorthereisamisalignementbetweenthelarge-scalejet-axisandtheparsec-scalejetaxisinclinationtothelineof sight,althoughtheprojectedpositionanglesofthelarge-scalejet(63◦)andtheparsec-scalejet(∼65◦)arealmostthesame.