The RINGO2 and DIPOL Optical Polarisation Catalogue of Blazars
H. Jermak, I. A. Steele, E. Lindfors, T. Hovatta, K. Nilsson, G. P. Lamb, C. Mundell, U. Barres de Almeida, A. Berdyugin, V. Kadenius, R. Reinthal, L. Takalo
aa r X i v : . [ a s t r o - ph . H E ] A ug Mon. Not. R. Astron. Soc. , 1– ?? (2014) Printed 8 November 2018 (MN LaTEX style file v2.2) The RINGO2 and DIPOL Optical Polarisation Catalogue of Blazars
H. Jermak, I. A. Steele, E. Lindfors, T. Hovatta, , K. Nilsson, G. P. Lamb, C. Mundell, U. Barres de Almeida, A. Berdyugin, V. Kadenius, R. Reinthal, L. Takalo, Astrophysics Research Institute, Liverpool John Moores University, Brownlow Hill, Liverpool, UK, L3 5RF. Tuorla Observatory, Department of Physics and Astronomy, University of Turku, V¨ais¨al¨antie 20, 21500 Piikki¨o, Finland. Aalto University Mets¨ahovi Radio Observatory, Mets¨ahovintie 114, 02540 Kylm¨al¨a, Finland. Aalto University Department of Radio Science and Engineering,P.O. BOX 13000, FI-00076 AALTO, Finland. Finnish Center for Astrophysics with ESO, University of Turku, V¨ais¨al¨antie 20, 21500 Piikki¨o, Finland. Department of Physics, Bath University, Bath, UK, BA2 7AY. Centro Brasileiro de Pesquisas Fisicas, Rua Dr. Xavier Sigaud 150, Urca, Rio de Janeiro, RJ 22290-160, Brazil.
Released 2016 Xxxxx XX
ABSTRACT
We present ∼ ∼ γ -ray brightblazars over a period of 936 days (11/10/2008 - 26/10/2012) using data from the Tuorlablazar monitoring program (KVA DIPOL) and Liverpool Telescope (LT) RINGO2 polarime-ters (supplemented with data from SkyCamZ (LT) and Fermi-LAT γ -ray data). In 11 outof 15 sources we identify a total of 19 electric vector position angle (EVPA) rotations and95 flaring episodes. We group the sources into subclasses based on their broadband spectralcharacteristics and compare their observed optical and γ -ray properties. We find that (1) theoptical magnitude and γ -ray flux are positively correlated, (2) EVPA rotations can occur inany blazar subclass, 4 sources show rotations that go in one direction and immediately ro-tate back, (3) we see no difference in the γ -ray flaring rates in the sample; flares can occurduring and outside of rotations with no preference for this behaviour, (4) the average degreeof polarisation (DoP), optical magnitude and γ -ray flux are lower during an EVPA rotationcompared with during non-rotation and the distribution of the DoP during EVPA rotations isnot drawn from the same parent sample as the distribution outside rotations, (5) the numberof observed flaring events and optical polarisation rotations are correlated, however we findno strong evidence for a temporal association between individual flares and rotations and (6)the maximum observed DoP increases from ∼
10% to ∼
30% to ∼
40% for subclasses withsynchrotron peaks at high, intermediate and low frequencies respectively.
Key words: galaxies: active – techniques: polarimetric –instrumentation: polarimeters –galaxies: jets – gamma-rays: galaxies.
The centre of most, if not all galaxies, contains at least one super-massive black hole (Kormendy & Richstone 1995; Magorrian et al.1998). If the matter in the vicinity of the compact object is closeenough to become accreted onto the compact object then it isclassified as an active galactic nucleus (AGN). The viewing an-gle of an AGN often determines its observational classification.Those AGN viewed within a small opening angle of the jet axisare classified as blazars (Urry & Padovani 1995). Blazars are de-fined by rapid flux variability with large amplitudes, high appar-ent luminosities, greater brightness temperatures than typical AGN,high polarisation and superluminal motion of ejected componentsin the jet. The apparent superluminary properties are caused bythe relativistic beaming of the jet emission towards the observer(Blandford & Rees 1978). Questions of the formation, collimation and acceleration ofblazar jets from the regions close to the supermassive black holeare still unsolved, however progress can be made by exploring thesignatures of the magnetic field in polarised light. Using the linearStokes Parameters to calculate the angle and degree of polarisa-tion we can explore how the optical synchrotron emission evolvesduring a γ -ray flare and whether rotations in the electric vector po-sition angle (EVPA) correspond with low- or high- states in theoptical and γ -ray emission. Changes in the EVPA and the degreeof polarisation can afford information about the structure and orderof the underlying magnetic field (Kikuchi et al. 1988).Blazars are the most energetic of the AGN classes andhave characteristic ‘double-humped’ spectral energy distributions(SEDs) that span the entire electromagnetic spectrum. The firsthump, peaking in the Infrared-optical is attributed to synchrotronemission, whereas the higher energy peak (X-ray to γ -ray) is c (cid:13) H. Jermak et al. thought to be produced by Inverse- or synchrotron self-Comptonscattering of jet or external photons. Blazars can be sub-divided ac-cording to the location of the synchrotron emission peak in theirspectral energy distribution (SED). Flat spectrum radio quasars(those sources originally identified to have optical emission lineequivalent widths ≥ ν < Hz (IR), λ & ≤ ν ≤ (optical/IR, λ ∼ ≥ Hz (UV, λ . ◦ .Nalewajko (2010) continue with this idea and suggest that a sym-metric emitting region on a bent jet could produce a gradual EVPArotation.In this paper we present the results of a polarimetric and pho-tometric campaign on a sample of fifteen blazars and present thecorrelations between these data and Fermi γ -ray data. We also ex-plore the relationship between γ -ray flares and polarisation anglerotations. In § γ -ray flares and EVPArotations used in this analysis. In § § § . The polarisation monitoring program with RINGO2 is a contin-uation of a program that was running at the KVA-60 telescopein 2009-2011. The KVA-60 telescope is used for optical sup-port observations of the MAGIC telescopes and has a relativelysmall mirror diameter of 60 cm. The source sample originallyconsisted of 8 γ -ray bright blazars that had an optical magnitudeof R <
16 and were known to show strong ( > The KVA telescope, operated remotely from Finland, consists oftwo tubes; 35cm and 60cm. The KVA 35 cm is used for the R-band photometric observations of the Tuorla blazar monitoring pro-gram . The observations are coordinated with the MAGIC ImagingAir Cherenkov Telescope and while the monitoring observationsare typically performed two to three times a week (the weatherallowing), during MAGIC observations the sources are observedevery night. The data are analysed using standard aperture pho-tometry procedures with the semi-automatic pipeline developed inTuorla by K. Nilsson. The pipeline presents the user with a graph-ical image of each frame to allow the rapid identification of thetarget object and comparison stars. The magnitudes are measuredusing the differential photometry and comparison star magnitudesfound in the footnotes , . The magnitudes are converted into Jan-skys using the standard formula S = 3080 × − ( mag / . . For mostof the sources, the contribution of the host galaxy to the measuredflux is insignificant, the exceptions being Mrk 421 and Mrk 501.If the host galaxy has been detected, its contribution has been sub-tracted from the measured fluxes (Nilsson et al. 2008). Finally, themeasured fluxes were corrected for the galactic absorption usingthe values from NED .While for many of the sources there is > years of data, inthis paper, we only use the data that is from the same observingperiods as our DIPOL and RINGO2 polarisation measurements. http://vizier.u-strasbg.fr/viz-bin/VizieR http://users.utu.fi/kani/1m http://ned.ipac.caltech.edu ptical Polarization Catalogue of Blazars Name z Type R Mag. range Pol. range (%) Fermi range Observation Period (MJD) Absent data3C 66A − -1.1x10 − S5 0716+714 − -1.3x10 − OJ 287 − -1.4x10 − − - 9.7x10 − Mrk 421 − -1.3x10 − Mrk 180 − - 2.6x10 − ON 231 − -8.6x10 − PKS 1222+216 − -1.3x10 −
3C 279 − -2.7x10 − PKS 1510-089 − -2.1x10 − PG 1553+113 < − -1.0x10 − Mrk 501 − -1.4x10 − BL Lac − -1.5x10 − Table 1.
The full RINGO2 catalogue with redshift, source type, R band magnitude range, Polarisation range, Fermi range, observation period information anddetails of absent/unavailable data (see Section 2.1.4). References for the redshift values can be found in Section 3.
RINGO2 was a fast-readout imaging polarimeter with a V+R hy-brid filter (covering 460-720 nm) constructed from a 3mm SchottGG475 filter cemented to a 2mm KG3 filter. RINGO2 used arapidly rotating ( ∼ ∼ The SkyCamZ camera consists of a 200mm diameter telescope thatparallel points with the Liverpool Telescope in order to providephotometric monitoring during observations with other instrumentsand also carry out a synchronous variability survey of the northernsky. The Z denotes a ‘zoomed’field-of-view (1 ◦ ) and the instru-ment can detect sources down to ∼
16 mag. When the enclosure isopen, the camera takes a 10 second exposure automatically onceper minute. All data are automatically dark subtracted, flat-fieldedand fitted with a world co-ordinate system (WCS) by the STILTpipeline (Mawson, Steele & Smith 2013). The data are then intro-duced to the same pipeline used to reduce the RINGO2 data (seeSection 2.1.2). The pipeline runs source extractor on the data andusing a pre-identified secondary star (with its literature magnitudecoming from the USNO-B1 catalogue) performs differential pho-tometry.
Fermi-LAT (Large Area Telescope) is a space-based pair produc-tion telescope with an effective area of 6500cm on axis for > γ -rays with energies in the rangeof 20 MeV to above 300 GeV (Atwood et al. 2009). To producethe Fermi-LAT light curves the reprocessed Pass 7 data was down-loaded and analysed using the ScienceTools version v9r32p5. In theevent selection the LAT team recommendations were followed forPass 7 data . We modelled a 15 degree region around each sourceusing the instrument response function P7REP SOURCE V15, http://fermi.gsfc.nasa.gov/ssc/data/analysis/documentation/Pass7REP usage.html H. Jermak et al.
Galactic diffuse model gll iem v05 rev1, and isotropic backgroundmodel iso source v05.The light curves were binned using an adaptive binningmethod (Lott et al. 2012), with estimated 15% statistical flux un-certainty in each bin. The flux in each bin was then estimated usingthe unbinned likelihood analysis and the tool gtlike. All sourceswithin 15 degrees of the target that are listed in the 2FGL cata-logue (Nolan et al. 2012) were included in the likelihood model.The spectral index of all sources are frozen to the values reportedin 2FGL, and for sources more than 10 degrees from the target alsofluxes are frozen to the 2FGL values. The sources Mrk 180 and1ES1426+428 are too faint to produce adequate Fermi light curvesfor this analysis as the bin sizes would be too large.
The KVA polarisation monitoring program began in December2008 using the Kungliga Vetenskapsakademien (KVA) telescopelocated on the Canary Island of La Palma. The KVA telescope con-sists of two telescopes; a 35 cm Celestron and a 60 cm Schmidt re-flector. The larger of the two, DIPOL, a 60 cm reflector, is equippedwith a CCD polarimeter capable of polarimetric measurementsin BVRI bands using a plane-parallel calcite plate and a super-achromatic /2 retarder (Piirola et al. 2005).The observations typically took place 1-2 times a week. Thetypical observation time per source was 960s and the observationswere performed without a filter to improve the signal-to-noise.There are several gaps in the cadence when the source has been toofaint (R >
15) and/or too weakly polarised (1-2%) to be detectablewith KVA. In total, 10 to ∼
100 polarisation measurements persource were collected. During some of the nights, polarised stan-dard stars from Turnshek et al. (1990) were observed to determinethe zero point of the position angle. The instrumental polarisationof the telescope has been found to be negligible.The data analysis is performed following the standard aper-ture photometry procedures with the semi-automatic software thathas been developed for monitoring purposes. The sky-subtractedtarget counts were measured in the ordinary and extraordinarybeams using aperture photometry. The normalised Stokes param-eters and the degree of polarisation and position angle were calcu-lated from the intensity ratios of two beams using standard formula(e.g. Landi Degl’Innocenti, Bagnulo & Fossati 2007).
Optical polarimetry was obtained using the novel RINGO2 fast-readout imaging polarimeter (Steele et al. 2010) on the LiverpoolTelescope (LT) (see Section 2.1.2). RINGO2 was mounted on thetelescope in the period 2010 August 1 – 2012 October 26. Dur-ing this period observations were obtained of the blazar samplewith a typical cadence of ∼ γ ray state.The measured target counts were then corrected for instru-mental polarisation by division by the corresponding mean valueof the counts for the same Polaroid angle measured from allof the zero-polarised standard star observations (averaging over a period of time within which the polarimeter has not been re-moved from the telescope or altered). These corrected target countsand errors were then combined using the equations presented byClarke & Neumayer (2002) to calculate q, u and their associatederrors by standard error propagation. Analysis of the scatter in the q, u polarisation values derived from the zero-polarised standardsallowed us to estimate the stability of that correction as having anassociated q and u errors of . %, which we therefore combinedin quadrature to our final error estimate. Next we combine q and u to estimate an initial value of degree of polarisation ( p ): p = p q + u . (1)This measured value was then corrected for an instrumentaldepolarisation factor 0.76 ± For both the DIPOL and RINGO2 polarimetric data, to correct p forthe statistical bias associated with calculating errors from squareroots (where positive and negative q and u values are possible butonly positive p values can result) we used the methodology pre-sented by Simmons & Stewart (1985) to calculate 67% confidencelimits and the most likely p value. As a check on this procedure wealso ran a Monte Carlo simulation taking as input Gaussian distri-butions of q and u values with standard deviations equal to theircalculated errors and examined the resulting distribution of p . Theresults were identical.The electric vector position angle (EVPA) in degrees was cal-culated as EVPA = atan2(u , q) + ROTSKYANGLE + PA (2)where the function atan2() calculates the arctangent of u/q with a correct calculation of the sign and returns an angle between -180 and 180 degrees, ROTSKYANGLE is the angle of the telescopemount with respect to the sky when the image was taken and PA isa calibration constant derived from repeated measurements of theEVPA of the polarised standard stars. Errors on EVPA were calcu-lated according to the prescription in Naghizadeh-Khouei & Clarke(1993), and again confirmed by Monte Carlo simulation. There are no exact definitions of what consists of a flare or a flaringperiod. By eye it is possible to identify datapoints that appear tobe flaring however producing a sample wide condition that selectsthese points is difficult. It is not possible to assign a single level ofquiescence above which data are considered to be flaring due to thevarying baseline in γ -ray blazars. For clarity, we detail the condi-tions of the code used to identify the peak of γ -ray flaring periods.It is also necessary to identify what is considered to be an EVPA ro-tation or swing, this is detailed in this section and follows on fromdefinitions in other EVPA studies. A summary of the results of thisanalysis is presented in Table 2. ptical Polarization Catalogue of Blazars A blazar flare may be associated with quasi-stationary, high densityregions within the jet caused by magnetic field irregularities or itmay be associated with a knot or blob of emission moving along thejet (Bradt 2014). The definition of a γ -ray flare is complicated dueto it being relative to the (varying) baseline of the γ -ray emissionat the time prior to the flare. The method used to identify flaringevents was 2 fold. First we establish an initial level of increasedactivity using a moving window which defines ‘active’ points asthose which are twice the standard deviation of the five precedingpoints. Then for the points that meet this criterion the conditionof a flare is such; active points that are greater than five times thestandard deviation of the preceding five active points are considered‘flare’. In addition, due to the nature of the moving window, for thefirst five points in the light curve if the flux is greater than the meanflux for the whole light curve then the points are classed as flares.Once the flare points are identified, a flare episode is defined bythose flare points that are within 20 days of each other. These flaringepisodes can often contain more than one peak, these are so closetogether that we define them as one event. The flaring episodes arerepresented in the light curves by vertical blue lines covering allflares within the 20 day range. For the analysis in Section 4.2, the centre of this flaring episode is used. The recurrent episodes of optical electric vector position angle(EVPA) rotations that are seen from AGN jets have been inter-preted in a number of ways. Usually here we are referring to large-amplitude ( > ◦ ), smooth and long-lasting rotation events whichseem to signal some coherent process developing within the jet. Al-though random walks in the Stokes plane, driven by turbulent mag-netic fields, have been demonstrated to be able to explain long ro-tations (Jones et al. 1985; Marscher 2014), it cannot, for example,explain preferred rotation directions within some specific sources(which goes against the stochastic nature of the process). Nor (asshown via Monte Carlo simulations by Blinov et al. (2015)) canthey answer for an entire population of rotations observed. Otherinterpretations of the EVPA rotation which link them to coherent jetfeatures, such as (a) plasma following a helical path due to a large-scale helicoidal magnetic field configuration of the jet, resultingin long, slow rotations of the EVPA Marscher et al. (2008, 2010);Zhang, Chen & B¨ottcher (2014); (b) a bend or curvature in the jetwhich leads to a projection effect on the plane of the sky akin toa rotation, which can invert its rotation due to relativistic effectsresulting from the collimated emission (Abdo et al. 2010b).Since the EVPA has a 180 ◦ ambiguity, long gaps in the po-larisation light curves can lead to confusion when interpreting theEVPA rotations. To avoid introducing an incoherent view to theprocess with random 180 ◦ jumps being added to the EVPA dataset,we chose to interpret the observed EVPA light curves followinga continuity hypothesis. We assume that variations proceed in thesmoothest way possible with no sudden jumps. Although there isno predefined limit to the length of gaps in the data, we decideto apply correction only when the difference between consecutivedata points is <
30 days. In this work we define an EVPA rotationso that are results are consistent with those of the RoboPol group(Blinov et al. 2015), therefore an EVPA rotation is ‘any continuouschange of the EVPA curve with a total amplitude of ∆ θ max > ◦ ,which is comprised of at least four measurements with significantswings between them’. The flexibility of monitoring with a robotic telescope such as theLiverpool Telescope allows the user to increase monitoring of aparticular source if its activity is deemed interesting. The main sam-ple of blazars was added to over the period of the RINGO2 obser-vations according to reported flaring and thus some sources havemore seasons and more data than others. The multi-wavelengthlight curves of the individual sources in this sample can be foundin the Appendix.
3C 66A is a well-known BL Lac at redshift z > = γ -rays (E >
100 GeV)(Aliu et al. 2009; Acciari et al. 2009). There are many polarimet-ric monitoring observations of this source (e.g. Takalo & Sillanpaa(1993) and references therein). In these data the polarisation de-gree is always high, typically between 10 - 20 % with the max-imum value measured 33 %. In the historical data the EVPA issignificantly variable but shows a preferred position angle around20 - 40 ◦ , which is perpendicular to the direction of the VLBA jet.Ikejiri et al. (2011) and Itoh et al. (2013) also report a rotation ofEVPA of > ◦ (at MJD ∼ ∼ ∼ ◦ rotation of the EVPA,however, during November 2009 to January 2010 (MJD 55151 to55220) our data appear to show another 180 ◦ rotation (see Table2 and Figure 9 in the Appendix). The nature of the ± ◦ ambi-guity and the smooth rotation selection of the EVPA data meansthat the absence of even one data point can be the difference be-tween a rotation (our data) or a slight peak (Figure 4 in Itoh et al.(2013)). However we have combined our data with that of Itoh(priv. comm.) and we see the combined data suggest a rotation.Itoh et al. (2013) describe a polarisation degree which is system-atically different among the four periods due to a long-term slowchange. We continue to see this behaviour in our data beyond theirfourth period. The source enters a relatively quiescent phase afterlate July 2010 (MJD ∼ ∼ γ -ray flux stays low with the exception of asmall flare at MJD ∼ The BL Lac object S5 0716+714 has been studied intensively atall frequencies. It has no spectroscopic redshift but constraintsfrom intervening absorption systems give z < ± γ -rays. The source is thought to be observed very close to the lineof sight of the jet allowing an excellent view down the jet itself(Impey et al. 2000). There are several dedicated studies of the opti-cal polarisation behaviour of the source (Uemura et al. 2010). In theoptical band the source shows extremely fast brightness and degree H. Jermak et al. of polarisation variations. Intra-night variability of the polarisationhas been reported by Impey et al. (2000)and Villforth et al. (2009)with significant variations on timescales of 10-15 minutes. In ourdata we also see fast brightness and degree of polarisation varia-tions across the four seasons with a variation of ∼ ∼ > ◦ (Ikejiri et al. 2011). In the historical data the range of the degreeof polarisation is from ∼ ± ◦ ambiguity in our data indicates either the EVPA exhibits a rapidrotation of ∼ ◦ in March 2009 (MJD ∼ & ◦ in October 2010 (MJD ∼ ∼ ◦ in March 2012 (MJD ∼ ∼ γ -ray and optical data show correlations in late2009 and early 2011, see the Discussion for more details. OJ 287 (z=0.305) is a BL Lac object and one of the most fa-mous blazars as it hosts a supermassive binary black hole sys-tem at its centre (Sillanpaa et al. 1988). It is bright in HE γ -rays(Acciari et al. 2009) but has not been detected in VHE γ -rays(e.g. (Seta et al. 2009)). The dedicated studies of the optical po-larisation behaviour (D’arcangelo et al. 2009; Villforth et al. 2010;Uemura et al. 2010) have shown that there is a strong preferredposition angle for the polarisation which is perpendicular to theflow of the jet. The polarisation is strong (maximum 35%). Oc-casionally the EVPA also shows rapid rotations with durations of10-25 days. This behaviour has been interpreted as a signature oftwo components (Holmes et al. 1984; Villforth et al. 2010), station-ary polarisation core and chaotic jet emission. Occasionally flareswith a negative correlation between flux and polarisation degreehave been observed (Ikejiri et al. 2011). Our data include the pe-riod observed by Agudo et al. (2011) and we see similar behaviourof the polarisation properties, particularly the rotation in April 2009(MJD ∼ ∼
30% to <
5% during the second season following a γ -ray flare. The optical flux is variable and ranges between 13.5 and15.5 magnitudes. γ -rays Albert et al. (2007a),was little studied. It is bright in HE γ -rays (Acciari et al. 2009)and has little optical polarisation literature data. From 1987 there isone archival polarisation observation (Wills, Wills & Breger 2011) which shows low polarisation ∼
2% and an EVPA ∼ ◦ . The KVAand RINGO2 data presented here (taken for multi-wavelength cam-paigns (Ahnen et al. 2016a,b) show similar results to the archivalobservations. Low polarisation of <
10% and an EVPA at ∼ ◦ . Mrk 421 (z=0.03 (de Vaucouleurs et al. 1991) is a nearby BL Lacobject that was the first extragalactic VHE γ -ray emitter to be dis-covered (Punch et al. 1992). Its optical polarisation behaviour hasbeen studied extensively in the past (e.g. Hagen-Torn et al. (1983);Tosti et al. (1998a) and references therein). In these data for themajority of the time the source shows rather low polarisation < ∼ ◦ . Ikejiri et al. (2011) found over-all significant correlation between optical brightness and degreeof polarisation and during a large optical flare in the winter of1996-1997 (MJD ∼ ∼ ◦ from May to October 1995(MJD ∼ ∼ ◦ rotationof the position angle, along with a steady increased in the opti-cal magnitude. This behaviour precedes an unprecedentedly large γ -ray flare which occurs after June 2012 (MJD ∼ Mrk 180 (z=0.045 (Falco et al. 1999)) is a nearby BL Lac ob-ject that was detected in VHE γ -rays in 2006 (MJD ∼ ∼ ∼ ∼ ◦ and an R band magnitude of ∼ γ -rays. Its optical flux has varied overthe last ∼
10 years from 15.2-16.4 magnitudes in the R band (fromTuorla blazar monitoring campaign ). There are very few polar-isation measurements in the literature, Jannuzi, Smith & Elston(1994) report a degree of polarisation of ∼ ∼ ptical Polarization Catalogue of Blazars ON 231, also commonly known as W Comae, (z=0.102Weistrop et al. (1985)) is a HE and VHE γ -ray bright BL Lac ob-ject (Acciari et al. 2008, 2009). Observations taken in 1981-1982(MJD ∼ ∼ ◦ (Wills, Wills & Breger 2011). The source underwentthree major outbursts in March 1995 (MJD ∼ ∼ ∼ ∼ ∼ ◦ . They also found significant nega-tive correlation between flux and polarisation degree (see Figure 28of Ikejiri et al. (2011)). Sorcia et al. (2014) presented results fromFebruary 2008 to May 2013 and find a gradual decrease in meanflux over the ∼ ∼ ± ∼ ◦ which coincided with a γ -ray flare in June 2008.The KVA-60 and RINGO2 data in this work show a degreeof polarisation and EVPA consistent with the source in a low state.We see slightly brighter optical and γ -ray fluxes in the first season(see Figure 18 in the Appendix) and optical magnitude starts todecrease with the increase in degree of polarisation at the end ofour last observing season. PKS 1222+216 (z=0.435, Veron-Cetty & Veron (2006)) is a flatspectrum radio quasar (FSRQ), and therefore LSP, which hasreceived a lot of attention since its discovery in VHE γ -rays(Aleksi´c et al. 2011). Very little optical polarisation data are avail-able in the literature. A single measurement from (Ikejiri et al.2011) shows a degree of polarisation of 5.9 %. The data presentedhere were taken in the 2011-2012 (MJD ∼ γ -rays. The EVPA shows very little variation and the degree of po-larisation is low ( <
3C 279 (z=0.536, Burbidge & Rosenberg (1965)) was one of thefirst extragalactic γ -ray sources discovered (Hartman et al. 1992)and is one of the first flat spectrum radio quasars to be detectedin very high energy γ -rays (Albert et al. 2008). Over ∼
10 yearsof observations the source showed variability ranging from 13-16magnitudes in the R band. In the space of ∼
100 days the sourcebecame fainter by 3 magnitudes (Larionov et al. 2008b) and fromMJD 54120-54200 showed a rotation which they conclude is in-trinsic to the jet. This rotation was coincident with a low degreeof polarisation which was higher before and after the rotation (at23%). The low polarisation during the rotation is attributed by theauthors to the symmetry of the toroidal component of the helicalmagnetic field. In the period prior to the the start of the RINGO2program the source showed a rapid decline in magnitude over theperiod of ∼ ◦ rotation of the position angle ((Abdo et al.2010b). Kiehlmann et al. (2013), using data from RINGO2 andKVA-60 amongst other instruments, showed there was an increasein flux and degree of polarisation along with a ∼ ◦ rotation of the position angle in May 2011 (MJD 55700) the addition ofFermi data showed that during this period of ∼ γ -rayflux decreases by ∼
100 [10 − cm − s − ] (Aleksi´c et al. 2014a).Aleksi´c et al. (2014a) interpreted this optical outburst with a rota-tion of the position angle and the increase in the degree of polarisa-tion as geometric and relativistic aberration effects such as an emis-sion knot’s trajectory bending such that it crosses the observer’s lineof sight (for full description see Aleksi´c et al. (2014a)).We have three seasons of polarimetric data and four seasons ofphotometric data. Having the same data as Kiehlmann et al. (2013)we see the same behaviour. In the third and fourth seasons we seerotations that rotate in one direction and then back on themselves.We see an additional rotation which is followed by a lack of data.The source drops in brightness at the start of the observing periodand is at its highest in polarisation ( ∼ γ -ray source (Petry et al. 2000; Horan et al. 2002).Jannuzi, Smith & Elston (1994) report an optical degree of polari-sation of < ∼ ∼ ◦ . The optical flux maintains a fairly constant value of be-tween 16-17 in the R band. We find the degree of polarisation staysbelow ∼ γ -rays to be significantly detectedwithin the analysed time window. PKS 1510-089 is a γ -ray bright (Acciari et al. 2009) FSRQ/LSP atz=0.36 (Burbidge & Kinman 1966). The source has shown brightflares in optical, radio, x-ray and HE γ -rays at the beginning of2009 (MJD ∼ γ -rays (Abramowski et al. 2013). During the γ -ray flaring fromMJD 54950 - MJD 55000 (April 2009 onwards) the optical elec-tric vector position angle (EVPA) was reported to rotate by > ◦ and during the major optical flare the optical polarisation degreeincreased to <
30% (Marscher et al. 2010). In early 2012 (MJD55960 and onwards) it again showed high activity in HE γ -raysand was also detected in VHE γ -rays. Again there was a > ◦ ro-tation of the EVPA following this flare but the polarisation degreestayed low ∼
2% (Aleksi´c et al. 2014b).The polarimetric RINGO2 data we present for this source hasbeen averaged over 5-day bins to account for the scatter in the data(see Section 2.2.2). Our data show the above mentioned γ -ray andoptical flaring activity from the end of 2008 into 2009, we see a ro-tation in the EVPA at this time but due to interpretation of our datausing the EVPA tracing code (see Section 2.3.2) we do not report arotation as great as 720 ◦ , rather a rotation of 333 ◦ . The difference isdue to the data sampling and thus highlights the need for intensiveoptical monitoring during γ -ray activity. For clarity we include azoomed region of Figure 10 (see Figure 1) for comparison with thebottom panel of Figure 4 in Marscher et al. (2010). The red pointsshow the EVPA data point at it is measured and at the +180 posi-tion. The EVPA trace code in this work selected the lower of the H. Jermak et al.
MJD
EVPA ( D eg r ee s ) Figure 1.
Zoomed view of PKS 1510-089 light curve (full light curveshown in Appendix, Figure 10). This plot can be compared with Figure4 in Marscher et al. (2010) where the polarisation angle data are interpretedas showing a 720 ◦ rotation. Here we report a rotation of less than half that(333 ◦ ) and this is due to the interpretation of this particular dataset by theEVPA trace code which, to account for the ± two points as it is closer to the previous point, had there been in-termediate points the rotation might have shown to continue at asteeper gradient which would result in a ∼ ◦ rotation measure-ment.Ikejiri et al. (2011) report a correlation between V band mag-nitude and degree of polarisation. We see similar results in our anal-ysis (see Table 5), however, because we lack data when the R bandmagnitude was the brightest we are unable to populate the brighterend of the magnitude-degree of polarisation plot. PG 1553+113 is a γ -ray bright BL Lac object at z > γ -rays(Albert et al. 2007a; Aharonian et al. 2006) which has triggeredseveral multi-wavelength studies of the source (e.g. Aleksi´c et al.(2010)). However, only few campaigns have included polari-metric observations. Polarisation observations were reported inAlbert et al. (2007a), Andruchow et al. (2011) and Ikejiri et al.(2011) with the maximum value for polarisation degree of 8.2%.The observations of Ikejiri et al. (2011), which cover the longestperiod of time, do not show a clear preferred angle for the EVPA.In 2008 (MJD ∼ ∼ ◦ whilethe later observations (in 2009 and 2010 (MJD ∼ ∼ ◦ . RINGO2 and KVA-60 data suffer from poor sampling but agree with literature, show-ing an EVPA which is ∼ ◦ until March 2012 when there is arotation over a period of a few months which coincides with a flaredetected in HESS and MAGIC but not in Fermi (Abramowski et al.2015; Aleksi´c et al. 2015). Mrk 501 (z=0.0337, Ulrich et al. (1975)) is a BL Lac type sourcewhich was discovered as a VHE γ -ray source in 1996 (Quinn et al.1996) and above 1.5 TeV (Bradbury et al. 1997). The source wasobserved during a period of high activity in 1997 (MJD ∼ ∼ ∼ ◦ . In the available data we see norotations of the EVPA (see definition of EVPA in Section 2.3.2) anda very stable optical flux, the degree of polarisation reaches ∼ γ -ray flare, however, for thelarger γ -ray flare the source was not visible from La Palma. BL Lac (z=0.069, Vermeulen et al. (1995)) is a bright source ofHE γ -rays and occasionally of VHE γ -rays (Albert et al. 2007b;Arlen et al. 2013). Its polarisation has been extensively studiedwith the two long-term studies presented in Hagen-Thorn et al.(2002b) and Hagen-Thorn et al. (2002a). In these publications ob-servations from 1969 to 1991 (MJD ∼ ∼ ◦ which is close to the direction of the jet in very long base-line interferometry (VLBI). In the second half of their data (1980-1991, MJD ∼ q (Hagen-Thorn et al. 2002b). The polarisation degree for this 22year period varied from <
1% to ∼ ∼ ∼ ∼ ◦ which is nearly aligned with theradio core EVPA and mean jet direction.The EVPA tracing code presented in this paper (see Section2.3.2) identifies four polarisation angle rotations in the BL Lac data,however, only two of these can be classified as ‘true ’rotations ac-cording to the condition that the rotation must consist of 4 or moremeasurements with significant swings between them. The degreeof polarisation varies between values of ∼
25% and little or no po-larisation signal at all. The drop to a degree of polarisation of ∼ γ -ray flare, along with anincrease in the optical magnitude, and is consistent with previouslyreported behaviour. In this section comparisons are made between the polarisationproperties, optical flux and γ -ray flux for those sources with rea-sonable sampling. This sample is subject to selection biases andtherefore the results in this work cannot be generalised to thelarger blazar population. For those sources which have been ob-served only for a short period of time, which have sparse data sam-pling or lack sufficient multi-wavelength information (Mrk 180,1ES 1011+496, 1ES 1426+496 and 1ES 1218+304) only optical-optical analysis and their light curves (see Appendix) are presented ptical Polarization Catalogue of Blazars and they are excluded from the γ -ray analysis. The following sec-tion explores correlations between the optical data and the γ -rayflux, along with the frequency of flares in relation to optical polari-sation rotations. Correlations between the optical and γ -ray data can give informa-tion about the emission regions and magnetic field structure withinthe jets of the different blazars. Optical flux (lacking a strong po-larisation signal) can also originate from outside the jet.Due to observational constraints from ground-based tele-scopes, along with weather and observing priorities, we have im-perfectly sampled optical data. While it is possible that the con-tinuous Fermi γ -ray data could be binned to coincide with opti-cal monitoring this could not be possible with the adaptive binningcode used in this work as it automatically sets the bin sizes accord-ing to the gamma-ray brightness of the source (see Section 2.1.4).The binning of γ -ray data according to optical observations is likelyto dilute flaring behaviour (which is displayed in more detail withthe adaptive binning method) and also involve difficulties in estab-lishing bin sizes because the optical observations only take ∼ γ -ray data points (which areof course not completely synchronous) we explore two methods. Inmethod one we use each of the dates associated with optical ob-servations and interpolate a value from the γ -ray light curve forthis date by fitting a gradient to the nearest neighbouring γ -raypoints and calculating the matched γ -ray flux using the equationfor a straight line. The plots in Figure 2 are produced by such amethod and show the overall behaviour of the sources according totheir different subclasses.It is also possible to match the optical and γ -ray data by us-ing the same bins as the γ -ray data to bin the optical data. Thismethod produces correlation plots (see Figure 3) which are lessdense than those produced by leading with optical data sampling(compare with Figure 2). Binning of optical data in this way re-sults in higher temporal frequency optical activity being averagedout, As the focus of this paper is the optical data we therefore useour first method in the following analysis. The same method is ap-plied using the optical polarisation degree dates. When data fromone wavelength do not change over a period in which data fromthe other wavelength does change then these periods appear on thecorrelation plots as straight horizontal or vertical lines.We use the Spearman Rank Coefficient test to determine thecorrelation of the data. The null hypothesis states that the two vari-ables are not correlated. If p < . then the null hypothesis can berejected. Significant correlations are indicated by p > . (no cor-relation) or p < . (correlation). For the analysis that involvesmagnitude, the values of the correlation coefficient ρ have been cal-culated so that the reverse nature of the parameter is appropriatelyused.For exploring the distributions of Spearman Rank test results,we will use the Kolmogorov Smirnoff (KS) test where the null hy-pothesis is that the two samples are drawn from the same popula-tion where p=1 suggests there is a strong probability that the sam-ples come from the same parent distribution. γ -ray flux correlations Figure 2 shows 6 plots of γ -ray flux against optical magnitude forall sources and 5 different subclasses; BL Lacs and FSRQs (iden- ρ p NAll -0.153 6.4x10 − FSRQ − BL Lac < − LSP -0.539 < − ISP < − HSP < − Table 3.
Spearman rank correlation coefficient ρ , p value and number ofsources for each blazar subclass for the γ -ray and optical flux correlations(See Figure 2 for plots). tified according to the presence/size of optical emission lines) andHSP, ISP and LSP sources (classified according to the location ofthe synchrotron peak in their SEDs). It is evident in the plot ofall sources that there are two visible subclasses, both with positivecorrelations but different ranges in optical and γ -ray fluxes. Thesetwo subclasses are shown to be the FSRQ and BL Lac sources inthe next two plots. So not only do these sources show differencesin the strength of their optical emission lines, they also cover dif-ferent ranges in γ -ray flux and optical magnitude. The bottom 3plots in Figure 2 show that the sources split by the location of theirsynchrotron peak also cover different ranges in γ -ray and opticalflux. LSP sources are brightest in γ -ray flux, there is a decrease inmaximum γ -ray flux as the spectral peak moves toward higher fre-quencies, with the HSP sources in this sample having a much lowerrange in γ -ray fluxes compared with LSPs.In Table 4.1 the results of the Spearman Rank analysis of thisdata are presented and it is demonstrated that (due to differing dis-tances of the sources), the “whole sample” approach is not particu-larly useful. Rather we must consider the properties of each source(which are differently coloured in the figure) individually. In addi-tion in order to investigate the properties of the individual sourcesthe data were first separated into observing seasons (to avoid falseperiods of apparently stable behaviour introduced by long periodsof non-visibility (Itoh et al. 2013)). The number of seasons for eachsources depends on the availability of the optical data and vary from1 to 4 seasons.For each of the season datasets a Spearman Rank Coefficienttest was performed to measure the statistical dependence of oneflux against the other. A summary of these results are presented,along with those for other correlations, in Table 4. Results for indi-vidual source seasons can be found in Table 5 in the Appendix. Wefind that 68% of source seasons (25/37) show a positive a correla-tion ¯ ρ = 0.46 with significant p values (i.e. p ≤ ≥ ¯ ρ = 0.36) with p values ranging between 0.000 - 0.988.A Kolmogorov Smirnov (KS) test was performed on the HSP& ISP, ISP & LSP and HSP & LSP sources respectively to testwhether the distribution of the Spearman Rank Coefficient ρ valuesfor different blazar subclasses suggests that the subclasses originatefrom the same parent population. The mean of the distributions foreach subclass are ¯ ρ = 0.30, 0.43 and 0.34 for HSP, ISP and LSPsrespectively. The p values from the KS test indicate the probabilityof the HSP and ISP sources being from the same parent populationis 56%, for ISP and LSP sources the probability is 58% and forthe HSP and LSP sources the probability is 84%. It is not possibleto distinguish these probabilities from each other and none has asignificant p value to either indicate the subclass results are or arenot drawn from the same parent distribution. H. Jermak et al.
1. Source 2. Rot ↑
3. Rot ↓
4. Flares 5. Type 6. Fermi mon. 7. Max. 8. Flare rate 9. Days between 10. Flares(anti-c-wise) (c-wise) period (days) deg. flares/day (year) rot & flare during rot3C 66A
S5 0716+714
OJ 287
Mrk 421
ON231
PKS 1222+216
3C 279
PKS 1510-089
PG 1553+113
Mrk 501
BL Lac
Table 2.
Tabulated data of the upward and downward EVPA rotations and γ -ray flares for different blazar subclasses for the 11 sources that have EVPArotation/ γ -ray flare events. Also included are the length of the Fermi monitoring period in days, the maximum degree of polarisation, the flare rate (and meanflare rate), days between rotations and flares (and mean of this value) and number of flares during a rotation. − . − . − . − . − . − . − . l og ga mm a − r a y pho t on f l u x [ ^ − ph c m ^ − s ^ − ] All
Gamma−ray flux and R Magnitude
FSRQ BL Lac pks1510BL_Lacpks1222s50716pg15533c279oj2873c66amrk501on231mrk421
18 17 16 15 14 13 12 − . − . − . − . − . − . − . l og ga mm a − r a y pho t on f l u x [ ^ − ph c m ^ − s ^ − ] LSP
18 17 16 15 14 13 12
R mag
ISP
18 17 16 15 14 13 12
HSP
Figure 2.
Fermi γ -ray data plotted against magnitude for 11/15 sources (those which have > γ -ray datapoints) (each with a separate colour) and subsequentblazar subclasses: FSRQs, BL Lacs, LSPs, ISPs and HSPs. The γ -ray data points are interpolated to match the date of the optical data points (see Section 4.1).Black squares show the mean γ -ray and optical value for each source. ptical Polarization Catalogue of Blazars − . − . − . − . − . − . − . l og ga mm a − r a y pho t on f l u x [ ^ − ph c m ^ − s ^ − ] All
Gamma−ray flux & R Mag.(binned)
FSRQ BL Lac pks1510BL_Lacpks1222s50716pg15533c279oj2873c66amrk501on231mrk421
18 17 16 15 14 13 12 − . − . − . − . − . − . − . l og ga mm a − r a y pho t on f l u x [ ^ − ph c m ^ − s ^ − ] LSP
18 17 16 15 14 13 12
R mag
ISP
18 17 16 15 14 13 12
HSP
Figure 3.
The sample plots as Figure 2 with optical data points binned according to the range of the Fermi bins. There are fewer data but the overall trends aresimilar.
All P p ≤ ρ Mean p Mean ρ Quantity Quantity Range ρ Mean ρ Quantity Quantity- ρ ,+ ρ - ρ ,+ ρ mag-gam HSP 2.20x10 − - 0.620 -0.0929 - 0.745 0.210 0.299 1,6 7 0.502 - 0.745 0.608 0,3 3ISP 4.97x10 − - 0.524 -0.067 - 0.718 0.115 0.429 0,16 16 0.287 - 0.718 0.567 0,10 10LSP 0.000 - 0.988 -0.600 - 0.711 0.141 0.337 2,12 14 -0.600 - 0.711 0.390 2,10 12ALL 0.000 - 0.988 -0.600 - 0.745 0.142 0.369 3,34 37 -0.600 - 0.745 0.487 2,23 25gam-deg HSP 1.89x10 − - 0.419 -0.121 - 0.633 0.196 0.231 1,4 5 0.160 - 0.633 0.397 0,2 2ISP 6.68x10 − - 0.946 -0.560 - 0.411 0.370 -0.0382 8,8 16 -0.560 - 0.272 -0.229 2,1 3LSP 1.52x10 − - 0.925 -0.249 - 0.556 0.340 0.0619 6,5 11 0.360 - 0.556 0.426 0,3 3ALL 1.89x10 − - 0.946 -0.560 - 0.633 0.332 0.038 15,17 32 -0.560 - 0.633 0.173 2,6 8deg-mag HSP 1.98x10 − - 0.695 0.0876 - 0.549 0.312 0.268 0,7 7 0.468 - 0.513 0.525 0,2 2ISP 1.53x10 − - 0.0754 -0.485 - 0.395 0.0212 0.0334 2,2 4 -0.460 - 0.403 0.0697 1,2 3LSP 0.0607 - 0.999 5.47x10 − - 0.270 0.472 0.154 0,4 4 NA NA NA 0ALL 1.53x10 − - 0.999 -0.485 - 0.549 0.277 0.175 2,13 15 -0.460 - 0.513 0.252 1,4 5 Table 4.
Summary of results from the Spearman Rank correlation test showing the p and ρ values for different subclasses for optical vs γ -ray data, degree ofpolarisation vs γ -ray data and optical flux vs optical degree of polarisation. The full dataset is presented in Table 5 in the Appendix. H. Jermak et al. γ -ray flux correlations Figure 4 shows the γ -ray flux against optical degree of polarisationfor all sources (each coloured individually) and the 5 subclasses.The γ -rays are plotted on a logarithmic scale for visualisation pur-poses. The horizontal lines are caused by the polarisation varyingduring a wide γ -ray flare bin, usually in low γ -ray states. The ver-tical lines are caused by the γ -ray flux varying when the degreeof polarisation is very low. For the γ -ray and degree of polarisa-tion plots it is not possible to distinguish the FSRQ and BL Lacsubclasses from each other. The FSRQs exhibit higher γ -ray fluxesthan the BL Lacs. The spectral peak subclasses differ in their γ -rayflux value (as already shown in the previous section), however theyalso differ in their maximum degree of polarisation value. The LSPsources can exhibit polarisation degrees up to ∼ ∼ ∼ ρ coefficientsas histograms. The peak of the overall distribution is close to zero(as shown in Table 4. However the peak of the ρ value distributionsfor the LSP and HSP sources are positive and for ISP sources, neg-ative. All HSP and LSP source seasons show positive correlationswith p ≤ ≤ ρ val-ues. The HSP and ISP distributions have 44% probability of beingfrom the same parent distribution and the HSP and LSP distributionhave a 42% probability. The probability of the ISP and LSP sourcesbeing from the same distribution is 48%. None of these p values issignificant. Figure 6 shows plots of the degree of polarisation against the opticalmagnitude separated by object type. Here we plot all 15 sources inour sample (i.e. including those without Fermi data). Those sourcesthat do not have synchronous magnitude and degree of polarisation(Mrk 180 and PKS 1222+216) have their points interpolated fromneighbouring data where available. In addition as the data are syn-chronous they are not split into seasons but compared across thewhole available dataset. As already shown in the previous corre-lation plots, the HSP sources are limited to degree of polarisationvalues < ∼
40% and ∼
30% re-spectively. The HSP sources show tighter groupings than the LSPand ISP sources.Table 4 shows the Spearman Rank Coefficient ρ and probabil-ity values for the optical flux and degree of polarisation data. 87%(13/15) of sources show weak positive correlations between the op-tical flux and the optical degree of polarisation with ¯ ρ = 0.18. Inaddition, 4 sources show weak positive correlations with p ≤ ≤ Figure 7.
Number of observed optical rotations (corrected for the observingduty cycle) versus number of observed γ -ray flares for those sources whichhave sufficient γ -ray data (11/15 sources) - note there are two points at x=5,y=0. γ -ray properties during EVPA rotations We have identified 95 γ -ray flare events (see Section 3 for thedescription of a flaring event) in 11 sources. In the sample, therate of flaring is between 0.0022 - 0.017 flares per day (0.8 - 6.2per year). The mean flare rates (and standard deviations) for eachsubclass are HSP = . ± . , ISP = . ± . andLSP = . ± . flares per day, equivalent to HSP = 2.0 ± ± ± ρ = 0 . , p = 0 . is apparent. It thereforeappears that there is at least some link between a propensity for γ -ray flaring and that for optical polarisation rotations.Due to the visibility of the sources, 67 of the γ -ray flaring pe-riods occur when we lack coincident optical data or there are nodata between the data point and the nearest flare. Of the remain-ing 28 γ -ray flares that have optical photometry and polarimetry incoincident periods with the γ -ray data there are 17 that occur dur-ing rotation of the EVPA (see Table 2, Column 10). However wenote that this statistic is dominated by one source (PKS 1510-089)which has the highest mean flare rate and multiple flares within asingle long EVPA rotation. In addition we can associate 11 flaresthat occur outside an EVPA rotation with the closest in time EVPArotation (i.e. the nearest lying rotation to a flare where there areno missing data in between). There are 5 flares that occurred < <
116 days be-fore the rotation (see Table 2, Column 9). Even though we do notanalyse flares that occur during periods when we lack optical data,we must be cautious: the average observing season is ∼
180 dayswhich means that it may be possible that flares could be associ-ated with closer lying rotations that occur when we are unable toobserve them.In order to investigate the γ -ray and optical properties during ptical Polarization Catalogue of Blazars − . − . − . − . − . − . − . l og ga mm a − r a y pho t on f l u x [ ^ − ph c m ^ − s ^ − ] All
Gamma−ray flux and R Deg. of Pol.
FSRQ BL Lac pks1510BL_Lacpks1222s50716pg15533c279oj2873c66amrk501on231mrk421 − . − . − . − . − . − . − . l og ga mm a − r a y pho t on f l u x [ ^ − ph c m ^ − s ^ − ] LSP
R deg
ISP
HSP
Figure 4.
Fermi γ -ray flux against optical degree of polarisation for all sources and each blazar subclass, a different colour for each source separately. Blacksquares show where the mean of the source lies on the plot. Horizontal and vertical lines in the data show periods during which the γ -ray/optical data(respectively) are constant while the other continues to vary. LSP rho F r equen cy −1.0 −0.5 0.0 0.5 1.0 ISP rho F r equen cy −1.0 −0.5 0.0 0.5 1.0 HSP rho F r equen cy −1.0 −0.5 0.0 0.5 1.0 Figure 5.
Histograms showing the distribution of ρ values from the Spearman Rank Coefficient test for the optical degree of polarisation and γ -ray flux. Fromleft to right: LSPs (red), ISPs (blue) and HSPs (green). The dotted histograms are the distribution of the total sample and the black vertical lines show wherethe sample mean lies. The mean of the subclasses are shown as a vertical line in their respective colours. H. Jermak et al. D eg r ee o f po l a r i s a t i on ( % ) ALL FSRQ BL Lac pks1510BL_Lacpks1222s50716pg15531es10113c2791es12181es1426oj2873c66amrk501on231mrk421mrk180
17 16 15 14 13 12 D eg r ee o f po l a r i s a t i on ( % ) LSP
17 16 15 14 13 12
ISP
R mag
17 16 15 14 13 12
HSP
Figure 6.
The optical degree of polarisation against optical magnitude for all 15 sources. For those sources that do not have synchronous points(i.e. Mrk 421and PKS 1222) we interpolate the nearest lying point from the neighbouring datapoints. Each source is coloured separately and black boxes show where themean of that source lies on the plot. and outside of rotations we separated the data for each source intotwo periods: during (a) rotation and (b) non-rotation. The first twohistograms in the top panel of Figure 8 show the degree of polari-sation for all sources during those periods. The data are presentedas a percentage of the full range of the degree of polarisation for aparticular source and the whole histogram has been divided by theratio of the number of points in the larger dataset (outside of EVPArotations) over the number of points in the smaller dataset (duringEVPA rotations), this removes rare events from the analysis andtakes into account any selection effects. After this normalising wefind that the distributions do not change and each bin still has ≥ ¯ DoP ) during a rota-tion is 0.34 and outside of a rotation ¯ DoP = 0.46. On average thedegree of polarisation is therefore 26% lower during a rotation. AKS test was performed on the data to establish the probability thatthe degree of polarisation during rotation and non-rotation eventsare from the same parent distribution. There is a very low prob-ability (p < >
90% of the total flux during a rotationwhereas outside of rotations the are ∼
300 points that have polari-sation values >
90% of the total flux. The mean of the distributionduring a rotation is ¯ R = 52% and outside of rotation periods themagnitude is ¯ R = 59%. On average the degree of polarisation istherefore 17% lower during a rotation. The results from the KStest show, as for the degree of polarisation, the two distributions ofmagnitude during and outside of rotation periods have a very low(p < γ -ray flux during- and outside of- EVPArotations. During the rotations the γ -ray flux never rises above 59%of the total γ -ray flux. Outside of the rotations the γ -ray flux hasa longer high- γ flux tail, with the maximum brightness occurringoutside of a rotation event. The KS test results show that the like-lihood of the rotation and non-rotation γ -ray flux to be from thesame parent population is 24%. This means the null hypothesis,that the samples are from the same distribution, cannot be formallyrejected. However we note that during rotations, the mean of the γ -ray flux distribution ( ¯ γ = 10%) is 42% lower compared to thatoutside of a rotation ( ¯ γ = 17%).The third column in Figure 8 shows, for each individualsource, the mean ratio of degree of polarisation (top), R magnitude(middle) and -ray flux (bottom) during and outside of a rotation. For ptical Polarization Catalogue of Blazars deg during EVPA rotation (a) F r equen cy deg outside of EVPA rotation (b) F r equen cy Ratio of means − polarisation F r equen cy . . . . . R mag during EVPA rotation (a) F r equen cy R mag outside of EVPA rotation (b) F r equen cy Ratio of means − Magnitude F r equen cy gam during EVPA rotation (a) F r equen cy gam outside of EVPA rotation (b) F r equen cy Ratio of means − Gamma−ray flux F r equen cy . . . . . Figure 8.
The degree of polarisation (top), optical magnitude (middle) and γ -ray flux (bottom) displayed as a fraction of the normalised range for a) duringEVPA rotations (white) and b) outside of EVPA rotations (grey) and c) as a ratio of the mean of each property during a rotation over the mean of each propertyoutside of a rotation for each individual source (see Section 4.2 for more details). The black vertical lines in the first two columns show the mean of thehistograms. the degree of polarisation, 5/8 sources have lower values (i.e. ratios <
1) during rotations. For the R magnitude there are 7/8 sourcesthat have lower values during a rotation. For the γ -rays there are5/8 sources which are less bright in γ -rays during a rotation. There are important caveats to consider before making conclud-ing remarks about the results presented in this paper. Firstly, theRINGO2 blazar monitoring survey was designed to follow-upsources detected by MAGIC, the original sample size has increased,but the essence of the sample is that the sources are all γ -ray brightand have exhibited some kind of flaring activity (hence the rea-son they are added to the sample). Thus, due to this selection bias,the presented sample averaged results (Section 4) cannot be gener-alised to the larger blazar population, however the correlations forindividual sources (Section 3) are robust. In Section 3 we presented a detailed discussion of the be-haviour of the individual sources in our sample. Comparing sourceto source the orientation of the rotation (i.e. whether it is upwardor downward) does not afford any information as the rotation di-rection is presumably subject to the arbitrary sense of the mag-netic field and its properties which will vary from blazar to blazar.However, in four sources 3C279, PKS 1510-089, PG 1553+113 andS5 0716+714 we observe upward then downward rotations and inthe case of S5 0716+714 we see the EVPA rotate upward, down-ward and then upward again (see light curves for these sources inthe Appendix). We also have cases in four sources PKS 1510-089,S5 0716+714, PG 1553+113 and Mrk 421 in which there is a γ -ray flare, during or temporally close, associated with a rotation.Such behaviour of the EVPA is potentially important in studyingthe magnetic field and/or the orientation of the jet/emission blobwithin the jet with respect to the observer. Monte Carlo analysis byBlinov et al. (2015) suggests that a single EVPA rotation event can H. Jermak et al. be caused by a random walk of the EVPA, but it was unlikely that all rotations are due to random walk.In Section 4 we carried out a statistical study of the generalproperties of these sources without considering their individual be-haviour. We found the following principal results:(i) The maximum observed degree of optical polarisation for theLSP sources was ∼ %. For ISP sources it was ∼
30% and forHSP sources ∼ %. It is natural to attribute the low maximumpolarisation degree in HSP sources to their optical light being dom-inated by non-synchrotron emission which could originate from theaccretion disk or emitting regions outside of the jet. This explana-tion also accords with the low optical variability in these sources,however it could also be a signature of low-ordered magnetic fieldsin the jet. It must also be noted that these results cannot be appliedto the larger blazar population.(ii) On average the optical degree of polarisation and γ -ray fluxare not strongly correlated. ISP and LSP sources show no strongpreference for either positive or negative correlations. HSP sourcesshow a stronger (yet still weak) positive correlation.(iii) In 92% (34/37) of source seasons we found a positive cor-relation ( ¯ ρ = 0.37) between optical and γ -ray flux. In over halfof the seasons (25/37 = 68%) the probability of correlation is sig-nificant (i.e. p ≤ γ -ray emitting regions in blazars. We find no significantevidence to determine if the HSP, ISP and LSP distributions of thecorrelation coefficient were similar or not. This suggests, on aver-age, a common mechanism connects γ -ray flaring and optical po-larisation in these different blazar subclasses.(iv) There is a weak positive ( ¯ ρ = 0 . ) correlation betweenoptical flux and degree of polarisation in 13/15 source seasons. In5/15 cases the probability of correlation ( ¯ ρ = 0.25) is significant(i.e. p ≤ γ -ray flaring and EVPA rota-tions. There is a significant correlation ( ρ = 0.59, p = 0.05) betweenthe number of flares and the number of EVPA rotations in a givenobject. We do not, however, find any systematic difference by classin γ -ray flaring rate or number of EVPA rotations.(vi) γ -ray flaring episodes can occur during and outside of rota-tion events. The distribution of lead and lag values between flaresand rotations show that there is no preference for either behaviour.The association of the γ -ray flare and the EVPA rotation could pro-vide evidence for the cause of the rotation and the flare originatingfrom the same shock region, whereby the shock provides electronsfor up-scattering photons via Inverse-Compton processes and thetangled magnetic field providing the structure for the EVPA rota-tion. However, optical and gamma-ray flaring is not always syn-chronous with an observed rotation, suggesting that other mecha-nisms are involved in some instances.(vii) The mean degree of polarisation as a percentage of the totalrange of polarisation is 26% lower during periods of rotation com-pared to periods of non-rotation, Blinov et al. (2016) also report adecrease in polarisation during rotations. The mean optical flux is17% lower during a rotation compared with outside rotations andthe mean γ -ray flux is 41% lower during a rotation compared withoutside a rotation. The lower degree of polarisation during a rota-tion can be interpreted as a difference in the degree of ordering ofthe magnetic field during a rotation compared with non-rotation.Alternatively it could be evidence for their association with emis- sion features or shocks travelling along helical magnetic field lines(Marscher et al. 2008; Zhang et al. 2015). We would like to thank the reviewer for their careful reading of thepaper and constructive comments. H. Jermak is supported by theScience and Technology Facilities Council (STFC) funding and aRoyal Astronomical Society grant. T. Hovatta was supported by theAcademy of Finland project number 267324. C. Mundell acknowl-edges support from the Royal Society, the Wolfson Foundation andthe STFC. UBA is partially funded by a CNPq Research Produc-tivity grant number 309606/2013-6 from the Ministry of Science,Technology and Innovation of Brazil. We thank Asaf Pe’er for fruit-ful discussions during his visit to the ARI, also theoretical discus-sions with Shiho Kobayashi and Drejc Kopac and statistical dis-cussions with Chris Collins. The authors thank Benoit Lott for pro-viding the adaptive binning light curve analysis code for the FermiGamma-Ray Telescope data. The Liverpool Telescope is operatedon the island of La Palma by Liverpool John Moores University inthe Spanish Observatorio del Roque de los Muchachos of the Insti-tuto de Astrofisica de Canarias with financial support from the UKSTFC. ptical Polarization Catalogue of Blazars REFERENCES
Abdo A. A. et al., 2010a, ApJ, 716, 30Abdo A. A. et al., 2010b,
Nature , 463, 919Abramowski A. et al., 2013, A&A, 554, A107Abramowski A. et al., 2015, ApJ, 802, 65Acciari V. A. et al., 2009, ApJL, 693, L104Acciari V. A. et al., 2008, ApJL, 684, L73Adelman-McCarthy J. K., et al., 2009, VizieR Online Data Cata-log, 2294, 0Agudo I. et al., 2011, ApJL, 726, L13Aharonian F. et al., 2006, A&A, 448, L19Aharonian F. A., 2000, NewAstronomy, 5, 377Ahnen M. L. et al., 2016a, A&A, 590, A24Ahnen M. L. et al., 2016b, MNRAS, 459, 2286Albert J. et al., 2006a, ApJL, 642, L119Albert J. et al., 2006b, ApJL, 648, L105Albert J. et al., 2007a, ApJL, 654, L119Albert J. et al., 2007b, ApJL, 666, L17Albert J. et al., 2008, ApJ, 681, 944Aleksi´c J. et al., 2010, A&A, 515, A76Aleksi´c J. et al., 2014a, A&A, 567, A41Aleksi´c J. et al., 2014b, A&A, 569, A46Aleksi´c J. et al., 2015, MNRAS, 450, 4399Aleksi´c J. et al., 2011, ApJL, 730, L8Aliu E. et al., 2009, ApJL, 692, L29Anderhub H. et al., 2009, ApJL, 704, L129Andruchow I., Combi J. A., Mu˜noz-Arjonilla A. J., Romero G. E.,Cellone S. A., Mart´ı J., 2011, A&A, 531, A38Arlen T. et al., 2013, ApJ, 762, 92Arnold D. M., Steele I. A., Bates S. D., Mottram C. J., SmithR. J., 2012, in Society of Photo-Optical Instrumentation Engi-neers (SPIE) Conference Series, Vol. 8446, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference SeriesAtwood W. B. et al., 2009, ApJ, 697, 1071Barres de Almeida U., 2011, International Cosmic Ray Confer-ence, 8, 79Bertin E., Arnouts S., 1996, A&AS, 117, 393Bjornsson C.-I., 1982, ApJ, 260, 855Blandford R. D., Levinson A., 1995, ApJ, 441, 79Blandford R. D., Rees M. J., 1978, Physica Scripta, 17, 265Bła˙zejowski M., Sikora M., Moderski R., Madejski G. M., 2000,ApJ, 545, 107Blinov D. et al., 2015, MNRAS, 453, 1669Blinov D. et al., 2016, MNRAS, 457, 2252Bloom S. D., Marscher A. P., 1996, ApJ, 461, 657B¨ottcher M., Reimer A., Zhang H., 2013, in European PhysicalJournal Web of Conferences, Vol. 61, European Physical JournalWeb of Conferences, p. 5003Bradbury S. M. et al., 1997, A&A, 320, L5Bradt H., 2014, Astrophysics Processes. Cambridge UniversityPressBurbidge E. M., Kinman T. D., 1966, ApJ, 145, 654Burbidge E. M., Rosenberg F. D., 1965, ApJ, 142, 1673Clarke D., Neumayer D., 2002, A&A, 383, 360Cohen D. P., Romani R. W., Filippenko A. V., Cenko S. B., LottB., Zheng W., Li W., 2014, ApJ, 797, 137Costamante L. et al., 2001, A&A, 371, 512Danforth C. W., Keeney B. A., Stocke J. T., Shull J. M., Yao Y.,2010, ApJ, 720, 976Danforth C. W., Nalewajko K., France K., Keeney B. A., 2013,ApJ, 764, 57 D’arcangelo F. D. et al., 2009, ApJ, 697, 985de Vaucouleurs G., de Vaucouleurs A., Corwin, Jr. H. G., ButaR. J., Paturel G., Fouqu´e P., 1991, Third Reference Catalogue ofBright Galaxies. Volume I: Explanations and references. VolumeII: Data for galaxies between 0 h and 12 h . Volume III: Data forgalaxies between 12 h and 24 h .Dermer C. D., Schlickeiser R., 1993, ApJ, 416, 458Dermer C. D., Schlickeiser R., Mastichiadis A., 1992, A&A, 256,L27Dermer C. D., Sturner S. J., Schlickeiser R., 1997, ApJS, 109, 103Falco E. E. et al., 1999, PASP, 111, 438Furniss A., Fumagalli M., Danforth C., Williams D. A., ProchaskaJ. X., 2013, ApJ, 766, 35Georganopoulos M., Kazanas D., 2003, ApJL, 594, L27Ghisellini G., Madau P., 1996, MNRAS, 280, 67Ghisellini G., Tavecchio F., 2008, MNRAS, 386, L28Ghisellini G., Tavecchio F., Foschini L., Ghirlanda G., 2011, MN-RAS, 414, 2674Hagen-Thorn V. A., Larionov V. M., Jorstad S. G., LarionovaE. G., 2002a, AJ, 124, 3031Hagen-Thorn V. A., Larionova E. G., Jorstad S. G., Bj¨ornsson C.-I., Larionov V. M., 2002b, A&A, 385, 55Hagen-Torn V. A., Marchenko S. G., Smekhacheva R. I., Iakovl-eva V. A., 1983, Astrofizika, 19, 199Hartman R. C. et al., 1992, ApJL, 385, L1Holmes P. A. et al., 1984, MNRAS, 211, 497Horan D. et al., 2002, ApJ, 571, 753Hovatta T. et al., 2014, MNRAS, 439, 690Ikejiri Y. et al., 2011, PASJ, 63, 639Impey C. D., Bychkov V., Tapia S., Gnedin Y., Pustilnik S., 2000,AJ, 119, 1542Itoh R. et al., 2013, PASJ, 65, 18Jannuzi B. T., Smith P. S., Elston R., 1994, ApJ, 428, 130Jermak H., 2016, Robotic Polarimetry of Blazars. PhD Thesis,Liverpool John Moores UniversityJones T. W., Rudnick L., Aller H. D., Aller M. F., Hodge P. E.,Fiedler R. L., 1985, ApJ, 290, 627Joshi U. C., Baliyan K. S., Ganesh S., Deshpande M. R., Bhat-tacharyya S., Kaul R. K., Kaul C. L., Bhat C. L., 2000, Bulletinof the Astronomical Society of India, 28, 409Kiehlmann S. et al., 2013, in European Physical Journal Web ofConferences, Vol. 61, European Physical Journal Web of Con-ferences, p. 6003Kikuchi S., Mikami Y., Inoue M., Tabara H., Kato T., 1988, A&A,190, L8Kormendy J., Richstone D., 1995, ARA&A, 33, 581Landi Degl’Innocenti E., Bagnulo S., Fossati L., 2007, in As-tronomical Society of the Pacific Conference Series, Vol. 364,The Future of Photometric, Spectrophotometric and Polarimet-ric Standardization, Sterken C., ed., p. 495Larionov V., Konstantinova T., Kopatskaya E., Larionova L., Efi-mova N., Blinov D., Melnichuk D., Troitsky I., 2008a, The As-tronomer’s Telegram, 1502, 1Larionov V. M. et al., 2008b, A&A, 492, 389Lorenz E., Martinez M., 2005, Astronomy and Geophysics, 46,6.21Lott B., Escande L., Larsson S., Ballet J., 2012, A&A, 544, A6Magorrian J. et al., 1998, AJ, 115, 2285Mannheim K., Biermann P. L., 1992, A&A, 253, L21Maraschi L., Ghisellini G., Celotti A., 1992, ApJL, 397, L5Marcha M. J. M., Browne I. W. A., Impey C. D., Smith P. S., 1996,MNRAS, 281, 425 H. Jermak et al.
Marscher A. P., 2014, ApJ, 780, 87Marscher A. P., Gear W. K., 1985, ApJ, 298, 114Marscher A. P. et al., 2008,
Nature , 452, 966Marscher A. P. et al., 2010, ApJL, 710, L126Mawson N. R., Steele I. A., Smith R. J., 2013, AstronomischeNachrichten, 334, 729Monet D. G. et al., 2003, AJ, 125, 984M¨ucke A., Protheroe R. J., 2001, Astroparticle Physics, 15, 121M¨ucke A., Protheroe R. J., Engel R., Rachen J. P., Stanev T., 2003,Astroparticle Physics, 18, 593Naghizadeh-Khouei J., Clarke D., 1993, A&A, 274, 968Nalewajko K., 2010, International Journal of Modern Physics D,19, 701Nilsson K., Pursimo T., Sillanp¨a¨a A., Takalo L. O., Lindfors E.,2008, A&A, 487, L29Nolan P. L. et al., 2012, ApJS, 199, 31Petry D. et al., 2000, ApJ, 536, 742Piirola V., Berdyugin A., Mikkola S., Coyne G. V., 2005, ApJ,632, 576Punch M. et al., 1992,
Nature , 358, 477Quinn J. et al., 1996, ApJL, 456, L83Raiteri C. M. et al., 2013, MNRAS, 436, 1530Robbins M., Hadwen B., 2003, Electron Devices, IEEE Transac-tions on, 50, 1227R¨ugamer S. et al., 2011, ArXiv e-printsSchmidt G. D., Elston R., Lupie O. L., 1992, AJ, 104, 1563Seta H. et al., 2009, PASJ, 61, 1011Sikora M., Begelman M. C., Rees M. J., 1994, ApJ, 421, 153Sillanpaa A., Haarala S., Valtonen M. J., Sundelius B., Byrd G. G.,1988, ApJ, 325, 628Simmons J. F. L., Stewart B. G., 1985, A&A, 142, 100Sorcia M., Ben´ıtez E., Hiriart D., L´opez J. M., Cabrera J. I.,M´ujica R., 2014, ApJ, 794, 54Steele I. A., Bates S. D., Guidorzi C., Mottram C. J., MundellC. G., Smith R. J., 2010, in Society of Photo-Optical Instrumen-tation Engineers (SPIE) Conference Series, Vol. 7735, Societyof Photo-Optical Instrumentation Engineers (SPIE) ConferenceSeriesTakalo L. O., Sillanpaa A., 1993, Ap&SS, 206, 191Tavecchio F., Maraschi L., Ghisellini G., 1998, ApJ, 509, 608Tosti G. et al., 1998a, A&A, 339, 41Tosti G. et al., 1998b, A&AS, 130, 109Tosti G., Massaro E., Takalo L. O., Villata M., Efimov Y., 1999, inAstronomical Society of the Pacific Conference Series, Vol. 159,BL Lac Phenomenon, Takalo L. O., Sillanp¨a¨a A., eds., p. 149Turnshek D. A., Bohlin R. C., Williamson, II R. L., Lupie O. L.,Koornneef J., Morgan D. H., 1990, AJ, 99, 1243Uemura M. et al., 2010, PASJ, 62, 69Ulrich M.-H., Kinman T. D., Lynds C. R., Rieke G. H., EkersR. D., 1975, ApJ, 198, 261Urry C. M., 2000, in American Institute of Physics ConferenceSeries, Vol. 522, American Institute of Physics Conference Se-ries, Holt S. S., Zhang W. W., eds., pp. 299–306Urry C. M., Padovani P., 1995, PASP, 107, 803Vermeulen R. C., Ogle P. M., Tran H. D., Browne I. W. A., CohenM. H., Readhead A. C. S., Taylor G. B., Goodrich R. W., 1995,ApJL, 452, L5Veron-Cetty M.-P., Veron P., 2006, VizieR Online Data Catalog,7248, 0Villforth C. et al., 2010, MNRAS, 402, 2087Villforth C., Nilsson K., Østensen R., Heidt J., Niemi S.-M., PforrJ., 2009, MNRAS, 397, 1893 Weistrop D., Shaffer D. B., Hintzen P., Romanishin W., 1985,ApJ, 292, 614Wills B. J., Wills D., Breger M., 2011, ApJS, 194, 19Young A., 2010,
Nature , 463, 886Zhang H., Chen X., B¨ottcher M., 2014, ApJ, 789, 66Zhang H., Deng W., Li H., B¨ottcher M., 2015, ArXiv e-prints
We present here the fifteen light curves covering the RINGO2 pe-riod of monitoring. Along with the light curves we discuss the his-torical behaviour of the sources and how this is relates or differsfrom the RINGO2 observations. The four windows show (fromtop to bottom) Fermi γ -ray data, optical EVPA, optical degree ofpolarisation and optical flux density. In the Fermi window the ar-eas of rotations are shown (pink for upwards rotation, green fordownwards rotation), along with the areas that lack correspondingoptical polarisation data (grey) and the Fermi flares (blue) identi-fied using the automated code. For the core sample of 8 sourcesalong with Mrk 421 a large quantity of polarimetric observationsare available in the literature. We have reviewed observations fromcatalogues as well as papers dedicated to single sources and com-pare our data with the historical behaviour. We describe the γ -rayemission as High Energy (HE: E >
100 MeV) or Very High Energy(VHE: E >
100 GeV) regimes. p ti c a l P o l a r i z a ti on C a t a l ogu e o f B l a z a rs Log F l u x * - Log F l u x * - EVPA ( deg ) EVPA ( deg ) P ( % ) P ( % ) MJD R m Jy Figure 9.
All γ -ray and optical data for 3C 66A. Top panel shows the Fermi γ -ray light curve. The errors on the x axis represent the bins used for the Fermi data. Grey vertical sections show periods where nosynchronous optical data available, pink vertical sections highlight regions where optical polarisation angle rotations occur in the upwards direction, light green sections show downward rotations. Flaring episodesare identified by vertical blue lines (see Section 2.3.1 for definition of a flare). The second panel shows the optical polarisation angle or electric vector position angle (EVPA), the grey points are RINGO2 data andthe black points KVA-DIPOL data. The black line traces the temporally closest EVPA points, showing the most likely behaviour of the EVPA. The third panel shows the optical degree of polarisation, and the fourthpanel the optical magnitude, all point colours are the same as those for panel 2. H . Je r m a ke t a l . Log F l u x * - pks1510 Log F l u x * - EVPA ( deg ) EVPA ( deg ) P ( % ) P ( % ) MJD R m Jy Figure 10.
All γ -ray and optical data for PKS 1510-089. Top panel shows the Fermi γ -ray light curve. The errors on the x axis represent the bins used for the Fermi data. Grey vertical sections show periods where nosynchronous optical data available, pink vertical sections highlight regions where optical polarisation angle rotations occur in the upwards direction, light green sections show downward rotations. Flaring episodesare identified by vertical blue lines (see Section 2.3.1 for definition of a flare). The second panel shows the optical polarisation angle or electric vector position angle (EVPA), the grey points are RINGO2 data andthe black points KVA-DIPOL data. The black line traces the temporally closest EVPA points, showing the most likely behaviour of the EVPA. The third panel shows the optical degree of polarisation, and the fourthpanel the optical magnitude, all point colours are the same as those for panel 2. p ti c a l P o l a r i z a ti on C a t a l ogu e o f B l a z a rs Log F l u x * - Log F l u x * - EVPA ( deg ) EVPA ( deg ) P ( % ) P ( % ) MJD R m Jy Figure 11.
All γ -ray and optical data for 3C 279. Top panel shows the Fermi γ -ray light curve. The errors on the x axis represent the bins used for the Fermi data. Grey vertical sections show periods where nosynchronous optical data available, pink vertical sections highlight regions where optical polarisation angle rotations occur in the upwards direction, light green sections show downward rotations. Flaring episodesare identified by vertical blue lines (see Section 2.3.1 for definition of a flare). The second panel shows the optical polarisation angle or electric vector position angle (EVPA), the grey points are RINGO2 data andthe black points KVA-DIPOL data. The black line traces the temporally closest EVPA points, showing the most likely behaviour of the EVPA. The third panel shows the optical degree of polarisation, and the fourthpanel the optical magnitude, all point colours are the same as those for panel 2. H . Je r m a ke t a l . Log F l u x * - oj287 Log F l u x * - EVPA ( deg ) EVPA ( deg ) P ( % ) P ( % ) MJD R m Jy Figure 12.
All γ -ray and optical data for OJ287. Top panel shows the Fermi γ -ray light curve. The errors on the x axis represent the bins used for the Fermi data. Grey vertical sections show periods where nosynchronous optical data available, pink vertical sections highlight regions where optical polarisation angle rotations occur in the upwards direction, light green sections show downward rotations. Flaring episodesare identified by vertical blue lines (see Section 2.3.1 for definition of a flare). The second panel shows the optical polarisation angle or electric vector position angle (EVPA), the grey points are RINGO2 data andthe black points KVA-DIPOL data. The black line traces the temporally closest EVPA points, showing the most likely behaviour of the EVPA. The third panel shows the optical degree of polarisation, and the fourthpanel the optical magnitude, all point colours are the same as those for panel 2. p ti c a l P o l a r i z a ti on C a t a l ogu e o f B l a z a rs Log F l u x * - s50716 Log F l u x * - EVPA ( deg ) EVPA ( deg ) P ( % ) P ( % ) MJD R m Jy Figure 13.
All γ -ray and optical data for S5 0716. Top panel shows the Fermi γ -ray light curve. The errors on the x axis represent the bins used for the Fermi data. Grey vertical sections show periods where nosynchronous optical data available, pink vertical sections highlight regions where optical polarisation angle rotations occur in the upwards direction, light green sections show downward rotations. Flaring episodesare identified by vertical blue lines (see Section 2.3.1 for definition of a flare). The second panel shows the optical polarisation angle or electric vector position angle (EVPA), the grey points are RINGO2 data andthe black points KVA-DIPOL data. The black line traces the temporally closest EVPA points, showing the most likely behaviour of the EVPA. The third panel shows the optical degree of polarisation, and the fourthpanel the optical magnitude, all point colours are the same as those for panel 2. H . Je r m a ke t a l . Log F l u x * - BL_Lac
Log F l u x * - EVPA ( deg ) EVPA ( deg ) P ( % ) P ( % ) MJD R m Jy Figure 14.
All γ -ray and optical data for BL Lac. Top panel shows the Fermi γ -ray light curve. The errors on the x axis represent the bins used for the Fermi data. Grey vertical sections show periods whereno synchronous optical data are available, pink vertical sections highlight regions where optical polarisation angle rotations occur in the upwards direction, light green sections show downward rotations. Flaringepisodes are identified by vertical blue lines (see Section 2.3.1 for definition of a flare). The second panel shows the optical polarisation angle or electric vector position angle (EVPA), the grey points are RINGO2data and the black points KVA-DIPOL data. The black line traces the temporally closest EVPA points, showing the most likely behaviour of the EVPA. The third panel shows the optical degree of polarisation, andthe fourth panel the optical magnitude, all point colours are the same as those for panel 2. p ti c a l P o l a r i z a ti on C a t a l ogu e o f B l a z a rs Log f l u x * ^ − mrk421 Log f l u x * ^ − −2000200400 EVPA ( D eg r ee s ) EVPA ( D eg r ee s ) −20002004000246810 D o P ( % ) P ( % ) . . . MJD R m ag . m Jy Figure 15.
All γ -ray and optical data for 3Mrk 421 Top panel shows the Fermi γ -ray light curve. The errors on the x axis represent the bins used for the Fermi data. Grey vertical sections show periods where nosynchronous optical data available, the green vertical section highlights the region where the optical polarisation angle rotates in the downwards direction. Flaring episodes are identified by vertical blue lines (seeSection 2.3.1 for definition of a flare). The second panel shows the optical polarisation angle or electric vector position angle (EVPA), the grey points are RINGO2 data and no KVA-DIPOL data are available. Theblack line traces the temporally closest EVPA points, showing the most likely behaviour of the EVPA. The third panel shows the optical degree of polarisation (again no KVA-DIPOL data are available), and thefourth panel shows the optical magnitude; photometric calibration of the RINGO2 data was not possible due to the lack of suitable secondary stars in the frame, we instead present SkyCamZ data (open circles) tocomplement the KVA-DIPOL data. H . Je r m a ke t a l . Mrk 180
Mar 2012 Apr 2012 May 2012 Jun 2012 Jul 2012 Aug 2012 Sep 2012 Oct 2012
EVPA ( D eg r ee s ) EVPA ( D eg r ee s ) P ( % ) P ( % ) MJD R m Jy Figure 16.
All optical data for Mrk 180 (the source is too faint in Fermi data). Top panel shows the optical polarisation angle or electric vector position angle (EVPA), the grey points are RINGO2 data and there areno polarisation data points from KVA-DIPOL. The second panel shows the optical degree of polarisation, and the third panel the optical magnitude, all point colours are the same as those for Figure 15. p ti c a l P o l a r i z a ti on C a t a l ogu e o f B l a z a rs Log F l u x * - mrk501 Log F l u x * - EVPA ( deg ) EVPA ( deg ) P ( % ) P ( % ) MJD R m Jy Figure 17.
All γ -ray and optical data for Mrk 501. Top panel shows the Fermi γ -ray light curve. The errors on the x axis represent the bins used for the Fermi data. Grey vertical sections show periods where nosynchronous optical data are available. Flaring episodes are identified by vertical blue lines (see Section 2.3.1 for definition of a flare). The second panel shows the optical polarisation angle or electric vector positionangle (EVPA), the grey points are RINGO2 data, there are no KVA-DIPOL data for this source. The black line traces the temporally closest EVPA points, showing the most likely behaviour of the EVPA. There areno EVPA rotations (i.e. > ◦ ). The third panel shows the optical degree of polarisation (grey points are RINGO2 and no KVA-DIPOL data available). The fourth panel shows the optical magnitude (black pointsKVA-DIPOL, grey points RINGO2). H . Je r m a ke t a l . Log F l u x * - on231 Log F l u x * - EVPA ( deg ) EVPA ( deg ) P ( % ) P ( % ) MJD R m Jy Figure 18.
All γ -ray and optical data for ON 231. Top panel shows the Fermi γ -ray light curve. The errors on the x axis represent the bins used for the Fermi data. Grey vertical sections show periods where nosynchronous optical data available. There are no flaring episodes identified in ON 231 during this period of time. The second panel shows the optical polarisation angle or electric vector position angle (EVPA), thegrey points are RINGO2 data and the black points KVA-DIPOL data. The black line traces the temporally closest EVPA points, showing the most likely behaviour of the EVPA. The third panel shows the opticaldegree of polarisation, and the fourth panel the optical magnitude, all point colours are the same as those for panel 2. p ti c a l P o l a r i z a ti on C a t a l ogu e o f B l a z a rs Log F l u x * - pg1553 Log F l u x * - EVPA ( deg ) EVPA ( deg ) P ( % ) P ( % ) MJD R m Jy Figure 19.
All γ -ray and optical data for PG 1553+113. Top panel shows the Fermi γ -ray light curve. The errors on the x axis represent the bins used for the Fermi data. Grey vertical sections show periods where nosynchronous optical data available, pink vertical sections highlight regions where optical polarisation angle rotations occur in the upwards direction, light green sections show downward rotations. Flaring episodesare identified by vertical blue lines (see Section 2.3.1 for definition of a flare). The second panel shows the optical polarisation angle or electric vector position angle (EVPA), the grey points are RINGO2 data andthe black points KVA-DIPOL data. The black line traces the temporally closest EVPA points, showing the most likely behaviour of the EVPA. The third panel shows the optical degree of polarisation, and the fourthpanel the optical magnitude, all point colours are the same as those for panel 2. H . Je r m a ke t a l . Log F l u x * - pks1222 Log F l u x * - EVPA ( deg ) EVPA ( deg ) P ( % ) P ( % ) MJD R m Jy Figure 20.
All γ -ray and optical data for PKS 1222+216. Top panel shows the Fermi γ -ray light curve. The errors on the x axis represent the bins used for the Fermi data. Grey vertical sections show periods whereno synchronous optical data available. Flaring episodes are identified by vertical blue lines (see Section 2.3.1 for definition of a flare). The second panel shows the optical polarisation angle or electric vector positionangle (EVPA), the grey points are RINGO2 data and the black point is KVA-DIPOL data. The black line traces the temporally closest EVPA points, showing the most likely behaviour of the EVPA. The third panelshows the optical degree of polarisation, and the fourth panel the optical magnitude, all point colours are the same as those for panel 2. p ti c a l P o l a r i z a ti on C a t a l ogu e o f B l a z a rs Log F l u x * - Log F l u x * - EVPA ( deg ) EVPA ( deg ) P ( % ) P ( % ) MJD R m Jy Figure 21.
All γ -ray and optical data for 1ES 1011+496. Top panel shows the Fermi γ -ray light curve. The errors on the x axis represent the bins used for the Fermi data which, due to the faintness of the source, arequite large. No flare analysis was performed on this source due to the lack of Fermi data. The second panel shows the optical polarisation angle or electric vector position angle (EVPA), the grey points are RINGO2data and no KVA-DIPOL data are available. The black line traces the temporally closest EVPA points, showing the most likely behaviour of the EVPA. The third panel shows the optical degree of polarisation, andthe fourth panel the optical magnitude, all point colours are the same as those for panel 2. H . Je r m a ke t a l . −0.20.00.20.40.6 Log F l u x * - −0.20.00.20.40.6Mar 2012 Log F l u x * - EVPA ( deg ) EVPA ( deg ) P ( % ) P ( % ) MJD R m Jy Figure 22.
All γ -ray and optical data for 1ES 1218+304. Top panel shows the Fermi γ -ray light curve. The errors on the x axis represent the bins used for the Fermi data which, due to the faintness of the source, arequite large. No flare analysis was performed on this source due to the lack of Fermi data. The second panel shows the optical polarisation angle or electric vector position angle (EVPA), the grey points are RINGO2data and no KVA-DIPOL data are available. The black line traces the temporally closest EVPA points, showing the most likely behaviour of the EVPA. The third panel shows the optical degree of polarisation, andthe fourth panel the optical magnitude, all point colours are the same as those for panel 2. p ti c a l P o l a r i z a ti on C a t a l ogu e o f B l a z a rs May 2012 Jun 2012 Jul 2012 Aug 2012
EVPA ( D eg r ee s ) EVPA ( D eg r ee s ) P ( % ) P ( % ) MJD R m Jy Figure 23.
All optical data for 1ES 1426+428 (the source is too faint in Fermi data). Top panel shows the optical polarisation angle or electric vector position angle (EVPA), the black points are KVA-DIPOL dataand there are no polarisation data points from RINGO2. The second panel shows the optical degree of polarisation, and the third panel the optical magnitude, all point colours are the same as those for Figure 15. H. Jermak et al.
Source/Season MJD range ρ (gam-mag) ρ (gam-deg) ρ (deg-mag) p (gam-mag) p (gam-deg) p (deg-mag)3C66Aa < − > < > < > < − > < − > < > − < − > < − > < > − > > < − < > < − b > < > < − > − < > − < > < > − > − ...1ES1426a < < − > < > < − > − − ...PG1553+113a < − b > < − ... ...c > < > − − ...Mrk501a < > < − b > < − > < − > − Table 5.
Full table of Spearman Rank Correlation results ( ρρ