arXiv:1409.4620v1 [astro-ph.IM] 16 Sep 2014 o.Nt .Ato.Soc. Astron. R. Not. Mon. .Bagnulo S. the for of tool characterisation diagnostic and new classification a spectro-polarimetry: Linear h hsclpoete fterflcigsurface. about reflecting the information of reveal scattering properties may the physical measurement on the its and polariza- and surface and the reflecting angle, structure the the of on of state depends re- composition light radiation The light scattered the surface. the the than of brighter polarized that tion more a expect is by can surface surface reflected darker one darker a one, a by brighter by flected a damped by efficiently through down more than and is up surface moving by electron the produced an radiation of the oscillations Since the beams). light scattered the h ieto aallt h ufc n epniua ot to ( perpendicular and polarized plane surface scattering linearly the to partially parallel is radiatio direction the electron the it Accordingly, the surface it. by the to re-emitted to more parallel perpendicular is direction than wave the rather electromagnetic in a an oscillate in by to sitting free hit intu- electron and an be surface that may planar thinking This by polarized. understood is itively surfaces by scattered Light INTRODUCTION 1 o in 08; September 2014 Received 14. September 2014 Accepted c 2 1 3 NF-Osraoi srfiiod oio -02 ioTori Pino I-10025 Torino, UK. di Astrofisico 9DG, Osservatorio BT61 - Armagh INAF Hill, College Observatory, Armagh uoenSuhr bevtr,Karl-Schwarzschild-Stra Observatory, Southern European 04RAS 2014 1 i.e. .Cellino A. , h ln otiigteicdn and incident the containing plane the , 000 – 21)Pitd1 oebr21 M L (MN 2018 November 13 Printed (2014) 1–5 , 59 us,lctdvr ls oteWtoi aiy samme of member a is sp family, Watsonia that the discovered asteroids. ta to serendipitously Barbarian and close of have albedo very we located the Finally, Luisa, with (599) objects. correlated linear the is of phase-angle of variation with the stud that and previous an found wavelength few we polarization a polarimetry, with linear broadband agreement colour In what of correlated. to fraction contrast positively in be the is, may that considerations, We violated, physical asteroids. be may of basic sugges law classification Umov This the the refine cases spectra. to some used polarization be different may totally polarimetry have may optical) leoadtxnmccassado w ml ein ttelm fth of limb the at regions small two compo of a and and dozen albedo classes a taxonomic including of and surfaces, measurements albedo the spectro-polarimetric of optical properties obtained too w the variatio sensing and particular wavelength-dependent remote tion the In between a measurements. relationships as reflectance possible spectro-polarimetry traditional of to use addition the explore We ABSTRACT hudb ie tepann ohrflcac n oaiainspect polarization and words: reflectance Key both explaining at aimed be should 2 efudta bet ihmrial ieetrltv reflectance relative different marginally with objects that found We esgetta uuemdligatmt ftesraestruct surface the of attempts modelling future that suggest We n .F Sterzik F. M. and oaiain–mnrpaes seod:gnrl–. – general asteroids: , minor – polarization s ,D878Grhn,Germany. Garching, D-85748 2, sse self he in n iia om21 uy24 July 2014 form riginal -al [email protected] E-mail: ee Italy. nese, 3 rdtoal eerdt ihtesmhwcnuigtr o term confusing somehow the is with which scatterin to phenomenon, simple referred This traditionally the above. to out contrast sketched in mechanism plane, para is scattering polarization the linear tha to of is plane the properties phase-angles polarimetric small of feature prising r etitdt ml nevl typically interval, small observed a be to may Sun they restricted the which are from at phase-angles distance the longer aster- Earth, significantly than for a main-belt at and Since orbit therein), 2005). teroids al. references (Penttil¨a (see et and classification determinations oid 2012, albedo and of al. purposes object) et the Cellino target for resulting the used be the from may the seen of between as morphology angle observer plot- the (the the phase-angle usually the and are of sun filters function mea- a char- optical BBLP as the system. standard ted for solar the tool our in of sensing surements objects remote the a of as acterisation used been long have nl a ehge,wl bv 40 above phase- well attainable higher, maximum be the can angle objects near-Earth of case rabn ierplrzto BL)measurements (BBLP) polarization linear Broadband -al [email protected] E-mail: A T E tl l v2.2) file style X -al [email protected] E-mail: hs-oaiaincurves phase-polarization ◦ ehp h otsur- most the Perhaps . flna polariza- linear of n triso different of steroids r neetdin interested are e e ae nmulti- on based ies lofudta in that found also oaiainwith polarization nlrc asteroid inel-rich h reflectance the d o seod in asteroids for l sepce from expected is sta spectro- that ts iin ehave We sition. r fasteroids of ure ∼ ra. pcr i the (in spectra ooi class xonomic h aeclass rare the 0 Moon. e − 30 ◦ nthe In . at t llel as- g f 2 S. Bagnulo et al.

◦ − ◦ negative polarization, is normally seen in the 0 20 phase- Table 1. BBLP values in the Bessel V RI filters from PQ spectra. angle range (usually referred to as the negative branch of the Photon-noise is negligible, and accuracy is limited by instrumen- phase-polarization curve) and may be explained in terms of tal polarization, which we estimate 6 0.1 %. The double taxon- coherent backscattering (Muinonen et al. 2002). omy classification given in col. 2 are from Tholen (1984) (left) and A widely adopted remote-sensing tool for the physi- Bus & Binzel (2002) (right). Asteroid observations were obtained cal characterization of small bodies is spec- from September 2013 to March 2014. (1) was observed with troscopy. Similarly to what happens in stellar spectroscopy, ISIS, all the remaining targets with FORS. The Moon was ob- asteroid reflectance spectra are classified into distinct taxo- served with FORS in April and June 2011. nomic classes. Taxonomy based on multi-band optical pho- tometry was first developed by several authors in the ’70s, Object Class α V R I and culminated in the classical work by Tholen (1984). More (%) (%) (%) recently, broadband photometry has evolved in full-fledged spectroscopy using spectrographs equipped with CCDs. A (1) Ceres G/C 22.4◦ 1.17 1.21 1.25 commonly adopted taxonomic classification based on spec- (2) Pallas B/B 27.5◦ 2.25 2.29 2.33 tra at visible wavelengths was published by Bus & Binzel 22.9◦ 0.99 1.00 1.03 (2002), and an extension to the near IR region was more (7) Iris S/S 26.9◦ 0.58 0.52 0.48 recently proposed by DeMeo et al. (2009). ◦ In this paper we want to assess whether spectro- 27.5 0.68 0.62 0.56 28.2◦ 0.75 0.68 0.64 polarimetry may be used to complement and refine the ob- serving techniques of spectroscopy and broadband polarime- (8) Flora S/S 28.4◦ 0.78 0.68 0.60 try, that so far have been only separately considered. For this (21) Lutetia M/Xk 14.6◦ −1.19 −1.23 −1.23 reason, we have started a survey of spectro-polarimetry of ◦ asteroids, to our knowledge the first of its kind. (24) Themis C/B 14.0 −1.23 −1.18 −1.12 The taxonomic classifications of reflectance spectra by (44) Nysa E/Xc 9.1◦ −0.27 −0.30 −0.32 Tholen (1984) and Bus & Binzel (2002) were based on Prin- 24.2◦ 0.23 0.24 0.25 cipal Component Analysis of hundreds of objects. So far, our . ◦ − . − . − . spectro-polarimetric dataset is far too small to allow us any (51) Nemausa CU/Ch 15 7 1 11 1 10 1 06 systematic classification. This paper presents therefore the (208) Lacrimosa S/Sk 13.7◦ −0.46 −0.47 −0.50 results of a pilot project aimed at assessing the usefulness ◦ (236) Honoria S/L 7.1 −1.00 −1.08 −1.17 of further investigations using this technique. (433) Eros S/S 42.0◦ 1.99 1.87 1.86 (599) Luisa S/K 26.9◦ −0.39 −0.30 −0.16 2 OBSERVATIONS Moon E n.a. 81.7◦ 9.86 8.28 7.07 We have obtained spectro-polarimetric measurements of Moon M n.a. 78.3◦ 5.81 4.99 4.36 a sample of asteroids using the FORS2 instrument (Appenzeller et al. 1998) of the ESO Very Large Tele- scope (VLT), and the ISIS instrument of the William Herschel Telescope (WHT) of the Isaac Newton Group tracted then wavelength calibrated using IRAF routines, of Telescopes. During an earlier VLT-FORS visitor mode and then combined with FORTRAN routines. Throughout run dedicated to the observations of the Earthshine this paper we will refer to the reduced Stokes parameter (Sterzik, Bagnulo, & Pall´e2012) we have also observed the PQ(λ) = Q/I representing the flux perpendicular to the sunlit limb of the Moon. plane Sun-Object-Earth (the scattering plane) minus the The instruments employed in our measurements are slit- flux parallel to that plane, divided by the sum of the two fed and are equipped with similar polarimetric optics, con- fluxes. For symmetric reasons, Stokes U is expected to be sisting of a retarder waveplate and a beam-splitter polarizer: zero. From the spectro-polarimetric data we calculated syn- a Wollaston prism in case of FORS2, and a Savart plate in thetic BBLP values (see Table 1). Approximate reflectance case of ISIS. The retarder waveplates may be set at fixed po- spectra r(λ) were obtained by dividing the intensity spec- sition angles, allowing one to exploit the advantages of the tra by the spectrum of solar analogue HD 30246 observed “beam-swapping” technique (Bagnulo et al. 2009). Thanks on 2014-01-30, but without taking into account wavelength to the beam-swapping technique, to the fact that both in- dependent slit losses, and then normalised to λ = 550 nm. struments are slit-fed, and that the light reflected by the Data were rebinned to a spectral bin of ∼ 11 nm. target reaches the polarimetric optics without oblique reflec- Polarization spectra of our targets are shown in Fig. 1. tions, we were able to obtain very accurate measurements of As expected, we found positive polarization (i.e., perpendic- the continuum polarization. Observations with the FORS in- ular to the scattering plane) at phase-angles α & 20◦, and strument were obtained using grism 300V with and without negative polarization (i.e., parallel to the scattering plane) ◦ order-sorting filter GG435, covering the wavelength range at phase-angles α . 20 . Remarkably, there is one exception: 435–930 nm and 390–930 nm, respectively. ISIS observations in spite of having been observed at a phase-angle as large ◦ were obtained using grism R158R and order-sorting filter as ∼ 27 , asteroid (599) Luisa exhibits a negative polariza- GG495, covering the spectral range 480 nm to 975 nm. tion. This makes it a new member of the class of the so-called Reductions of FORS data were performed with the aid Barbarians (Cellino et al. 2006), i.e., asteroids displaying an of the ESO FORS pipeline (Izzo et al. 2010), and dedicated anomalous phase-polarization curve, characterized by a very ◦ FORTRAN routines. Spectra obtained with ISIS were ex- wide negative polarization branch, extending up to α ∼ 30 .

c 2014 RAS, MNRAS 000, 1–5 Linear spectro-polarimetry of asteroids 3

Figure 2. pq spectra of asteroids (i.e., PQ spectra normalised to the value at λ = 550 nm). The symbol (+) means that the spectrum was obtained in the positive branch, while the symbol (–) means that it was obtained in the negative branch.

We therefore introduce the polarization spectra normalised to the value at λ = 550 nm:

PQ(λ, α) pq(λ, α)= . PQ(λ = 550 nm, α) Figure 1. Polarization spectra of 12 asteroids (bottom panel) The introduction of this new quantity allows us to compare and of two regions of the Moon (top panel). data of different objects obtained at different phase-angles, under the approximation that, at least to first-order, the 3 DISCUSSION dependence of the polarization upon phase-angle may be separated from the dependence upon wavelength, in which To discuss the diagnostic power of spectro-polarimetry, we case we have pq(λ, α) ≃ pq(λ). We note that unless the PQ are going to address the following inter-related questions. spectra cross the zero, pq is always positive. Answering questions ii) and iii) is equivalent to ex- i) Do polarization spectra depend on the phase-angle? plicitly addressing the issue of whether spectro-polarimetry ii) Do asteroids of a given taxonomic class have identical brings additional information than spectroscopy. Figure 1 polarization spectra? suggests that in the specific case of asteroids (7) Iris and iii) Do asteroids of different taxonomic classes have differ- (8) Flora – both S-class in the Bus & Binzel (2002) system, ent polarization spectra? and both observed at α ∼ 28◦ – the answer to question ii) iv) What is the relationship between polarization spectra is yes. To address question iii) we may consider that aster- and reflectance spectra? oids (2) Pallas (B-class), (7) Iris (S-class), and (599) Luisa Firm answers require observations of several asteroids per (K-class) which were all all observed close to α ∼ 27◦, show taxonomic class with a homogeneous sampling of the phase- rather different PQ spectra. To proceed further, we can only angle range. However, even our limited dataset suggests compare observations of different asteroids obtained at dif- some tentative answers, and, most importantly, guides us ferent phase-angles, making use of the normalised polariza- on how to refine the strategy for future observations. tion spectra pq introduced above. We already know from classical BBLP measurements The top panel of Fig. 2 shows the pq spectra of B- and that the fraction of linear polarization does depend on phase- C-type asteroids. Asteroids (2) Pallas and (1) Ceres are both angle. In this analysis, however, we are more interested in the observed in the positive branch, and share similar pq spectra. shape of the polarization spectra. Observations of (2) Pallas Asteroids (21) Themis and (51) Nemausa are both observed and (7) Iris suggest that in the positive branch, at least in the negative branch, and also share similar pq spectra. In within limited phase-angle ranges, the shape of the PQ spec- Fig. 1 we see that the PQ spectra of B- and C-type asteroids tra does not change, although observations of (44) Nysa sug- always have a negative wavelength gradient. We note that gest that the shape of the PQ spectra obtained in the positive since in the negative branch the gradients of pq and PQ branch may differ from that obtained in the negative branch. spectra have opposite sign, B- and C-type asteroids have

c 2014 RAS, MNRAS 000, 1–5 4 S. Bagnulo et al. dpq/dλ < 0 in the negative branch, and dpq/dλ < 0 in the positive branch (see Fig. 2). The mid panel shows the pq spectra of four S-type as- ◦ teroids: (7) Iris (observed three times around α ∼ 28 ), ◦ (433) Eros (a near-Earth asteroid observed at α = 42.0 ), (8) Flora (observed at α = 28.4◦) and (208) Lacrimosa (ob- ◦ served in the negative branch at α = 13.7 ). The three pq spectra of (7) Iris overlap each other well. The pq spectrum of (433) Eros exhibits a marginally more pronounced con- cavity than that of (8) Flora and (7) Iris, but we are not able to say whether this (small) difference comes from the fact that Eros observations were obtained at a quite differ- ◦ ent phase-angle (∼ 14 larger than those of Flora and Iris), or because we are observing objects with different surface structure. We note that the PQ spectra obtained in the pos- itive branch have a negative gradient. The PQ spectrum of (208) Lacrimosa, the only S-class asteroid observed in the negative branch, also has a negative gradient (which corre- sponds to a positive gradient for pq). We therefore conclude that the intermediate albedo S-class asteroids exhibit a po- larimetric behaviour opposite to that of low-albedo B- and C-class asteroids, i.e., the gradient of their PQ spectra is always negative. Figure 3. Normalised polarization spectra pq (thick solid lines) The bottom panel of Fig. 2 shows that the pq spectra of and reflectance r spectra (thin dashed lines) of two regions of the high-albedo Xc-class asteroid (44) Nysa are somewhat sim- Moon and of three asteroids of different taxonomic classes. ilar both in the negative and in the positive branch, and similar to the other X-class asteroid (21) Lutetia. The slope of the PQ spectra of (44) Nysa is negative in the negative Moon are steeper than that of S-type asteroid (7) Iris (blue branch, and positive in the positive branch, therefore it must solid lines), but less steep than K-type asteroid (599) Luisa change its sign somewhere around the inversion angle. We (red solid line). It is likely that the difference in slope are de- may speculate that this feature is common to all high-albedo termined by a remarkable diversity of the surface structures asteroids, but more data are needed to confirm this. and compositions. We now consider the polarization spectra of two regions In general, we expect that higher albedo corresponds at the limb of the Moon, one close to the Grimaldi crater, to smaller polarization, and lower albedo to higher polar- and one close to the Mare Crisium, which are plotted in ization. Indeed the behaviour of lunar spectra confirms the the top panel of Fig. 1. Due to the high phase-angle value, results by Dollfus et al. (1971) that for lunar regions, the both lunar PQ spectra have a much higher amplitude than polarization and reflectance spectra obey to the Umov law that observed for asteroids. Compared among themselves, (Umov 1905), i.e., PQ(λ) ∝ 1/r(λ). The Umov law is also the two lunar spectra show a similar trend but, due to the valid for asteroids (7) Iris and (599) Luisa, but in the case different phase-angle at which they were obtained, have a of asteroid (236) Honoria, both the absolute value of the po- quite different amplitude. Once they are normalised, they larization and the reflectance have a positive gradient, i.e. nearly overlap each other, as shown with the light blue and both polarization and albedo increase with wavelength. A magenta solid lines in Fig. 3. similar behaviour is exhibited by (21) Lutetia (not shown in Figure 3 allows us to compare spectropolarimetric data the Figure), that in the Tholen system had been classified as of asteroids of different taxonomic classes and of the Moon, M-type. This is another aspect of the phenomenon discussed and also to make some considerations about the reflectance by Belskaya et al. (2009) who discovered that in M-type and spectra, that are shown with dashed lines (normalised to S-type asteroids, which have higher albedo in the red than in λ = 550 nm). First we consider the three asteroids (7) Iris, the blue, the minimum of the polarization curves is deeper (236) Honoria, and (599) Luisa, which, although belonging in the red than in the blue. Umov law is rooted on the basic to different classes in the Bus & Binzel (2002) system, were mechanism described by the Fresnel laws. Perhaps it is not all classified as S-type in the Tholen system. It is remark- surprising that it is violated in those conditions when Fresnel able that, while their reflectance spectra appear similar to laws cannot even explain the orientation of the polarization. each other (which explains their common classification in This phenomenon deserves further observational and theo- the Tholen system), the pq spectra appear completely dif- retical investigation. In particular it would be interesting to ferent from each other! In particular, both (236) Honoria assess if it manifests itself only at small phase-angles, when and (599) Luisa were observed in the negative branch, but the polarization is parallel to the scattering plane (being display polarization spectra with opposite gradients: the pq perhaps linked to the coherent backscattering mechanism), ◦ spectrum of (599) Luisa (observed at α ∼ 27 ) has a strong or if it may be observed also at large phase-angles. negative gradient, i.e., the absolute value of the polariza- (236) Honoria is a known Barbarian. Our discovery that tion decreases with wavelength; viceversa, the absolute value also (599) Luisa is a Barbarian is particularly interesting. ◦ of the polarization of (236) Honoria (observed at α ∼ 7 ) In the space of orbital proper elements this asteroid is lo- strongly increases with wavelength. The pq spectra of the cated in a high-inclination region where other Barbarians are

c 2014 RAS, MNRAS 000, 1–5 Linear spectro-polarimetry of asteroids 5 also present, i.e., (387) Aquitania, (980) Anacostia and (729) be perfectly reproduced even with different instruments. Watsonia. The latter is the lowest-numbered member of Spectro-polarimetric techniques still allow us to simultane- a dynamical family (Novakovi´c, Cellino, & Kneˆzevi´c2011; ously obtain reflectance spectra, provided that the usual cal- Milani et al. 2014) which has been found by Cellino et al. ibrations are performed. (2014) to be a reservoir of small Barbarians. Moreover, spec- We suggest that spectro-polarimetric analysis of aster- troscopic data show that (387) Aquitania, (980) Anacostia oids should complement traditional spectroscopic measure- and (599) Luisa have peculiarly high abundances of the ments and classification. In the longer term, any physical spinel mineral, up to 30 % (Sunshine et al. 2008). The link model capable of reproducing the observed reflectance spec- between the Barbarian polarimetric behaviour and a com- tra should also be tested against its capability to reproduce position rich in spinel is therefore further confirmed by our the observed spectro-polarimetric data. discovery that the spinel-rich asteroid (599) Luisa is also a Barbarian. We remind that (236) Honoria and (599) Luisa display opposite polarimetric gradients (see Fig 3). Perhaps ACKNOWLEDGMENTS these differences are due to the large gap in phase-angle at which the observations were taken (though both in the SB and AC gratefully acknowledge support from COST negative branch), which would imply that the wavelength Action MP1104 “Polarimetry as a tool to study the so- gradient of the polarization changes its sign in the negative lar system and beyond”. Observations were performed with branch. If instead the difference of the polarization spectra ING Telescopes under programme W/2014A/5 and with reflects a difference in structure surface and composition, we ESO Telescopes at the La Silla-Paranal Observatory under may have found a hint to the existence of different categories programme IDs 087.C-0040(A) and 092.C-0639. We thank of Barbarians. More data are needed to confirm this. WHT staff, the ESO Paranal SCIOPS Team and User Sup- port Division for their excellent support to the observations.

4 CONCLUSIONS REFERENCES

We have obtained a number of polarization spectra of as- Appenzeller, I., Fricke, K., Furtig, W., et al. 1998, The teroids and the Moon. From their analysis we tentatively Messenger, 94, 1 suggest that PQ spectra of low albedo asteroids always have Bagnulo, S., Landolfi, M., Landstreet, J.D., et al. 2009, a positive gradient, and intermediate albedo asteroids al- PASP, 121, 993 ways have a negative gradient: this would be a confirmation Belskaya, I., Levasseur-Regourd, A.-C., Cellino, A., et al. of preliminary results obtained in the pioneering works by 2009, Icarus, 199, 97 Lupishko & Kiselev (1995) and Belskaya et al. (2009), based Bus, S.J., & Binzel, R. 2002, Icarus, 158, 146 on multi-colour BBLP data. Polarization spectra of high- Cellino, A., Belskaya, I. N., Bendjoya, Ph., et al. 2006, albedo asteroid (44) Nysa has a positive gradient in the pos- Icarus, 180, 565 itive branch, and a negative gradient in the negative branch, Cellino, A., Gil-Hutton, R., Dell’Oro, A, et al. 2012, but more observations are needed to check if this result can JQSRT, 113, 2552 be generalized to a wide class of asteroids. Cellino, A., Bagnulo, S., Tanga, P., Novakovi´c, B., Delbo, We have found strong evidence that the Umov law M. 2014, MNRAS, 439, 75 may be violated: observed in the negative branch, both re- DeMeo, F., Binzel, R.P., Slivan, S.M., & Bus, S.J. 2009, flectance and polarization of asteroid (236) Honoria strongly Icarus, 202, 160 increase with wavelength. We have also discovered that Dollfus, A., Bowell, E., & Titulaer, C. 1971, A&A, 10, 450 (599) Luisa is a member of the Barbarian class of asteroids. Izzo, C., de Bilbao, L., Larsen, J., et al. 2010, SPIE, 7737, We have shown that two objects belonging to the S- 773729 class observed at the same phase-angle have nearly identical Lupishko, D.F., & Kiselev, N.N., 1995, Bull. Am. Astron. polarization spectra, but we have also found that three ob- Soc. 27, 1064. jects, (7) Iris, (236) Honoria and (599) Luisa, have relatively Milani, A., Cellino, A., Kneˇzevi´c, et al. 2014, Icarus, 239, similar optical reflectance spectra but totally different polar- 46 izations spectra. Particularly puzzling is the difference be- Muinonen, K., Piironen, J., Shkuratoc, Yu. G., tween the polarization spectra of Honoria and Luisa, which Ovcharenko, A., Clark, B. E. 2002, in Asteroids III are both members of the Barbarian class of asteroids. We do (W.F. Bottke, A. Cellino, P. Paolicchi, R. P. Binzel, not know if this diversity is a consequence of the fact that Eds.), 123, University of Arizona Press these objects were observed at a different phase-angle, or if Novakovi´c, B., Cellino, A., & Kneˆzevi´c, Z. 2011, Icarus, it originates from a remarkably different surface structure. 216, 69 In asteroid spectroscopy, the choice of the solar ana- Penttil¨a, A., Lumme, K., Hadamcik, E., & Levasseur- logue used for the normalisation of the intensity spectra, Regourd, A.-C. 2005, A&A, 432, 1081 and the quality of the calibration of the atmospheric ex- Sterzik, M.F., Bagnulo, S., & Pall´e, E. 2012, Nature, 483, tinction play a crucial role on the final data quality, and, 64 ultimately, on the spectral classification of asteroids. By con- Sunshine, J. M., Connolly, H. C., McCoy, T. J.; Bus, S. J.; trast, spectro-polarimetric measurements are robust, nearly La Croix, L. M. independent of atmospheric conditions, and do not require Tholen, D.J. 1984, PhD thesis, University of Arizona. any calibration with a solar analogue . Provided that in- Umov, E. 1905, Physik. Z. 6, 674 strumental polarization is low and under control, they may

c 2014 RAS, MNRAS 000, 1–5