From koji.kawabata@nao.ac.jp Tue Jul 09 08:24:00 2002 Dear David, I send you the revised reply to the referee and also the revised paper in this mail. The former is still almost as you wrote, and changed by me only at minor points: They are put between the asterisks (***). I also think that we should sufficiently explain the required issues regardless the length of the text. For the latter, I dropped many sentences and cut the Figure 1 down for the Letter page limitation. (Most of them are kept in the TeX source by commenting out.) Compared with the length estimation reported by the editorial office and the result of the ApJ-emulating TeX compile, the present version may be within the 4 page limitation narrowly or slightly exceeds (by less than 0.1 page). The PS-file version can be downloaded from http://optik2.mtk.nao.ac.jp/~kawabtkj/tmp/ms.ps I would like you to check and revise them. They will still include many errors and inferior expressions, also include what you could not agree with possibly. I will also ask Nomoto and some co-authors for comments on them in a few days. I expect that we can submit the revised paper next week. I also hope that you have sufficient time to read and reply them besides the educational works in this and next weeks... Sincerely yours, Koji -------------------------- reply to the referee ---------------------- *** We made some revisions, especially for the interpretation of the observation, in response to the referee's comments. We also shortened the length of the text for the 4 page limitation of the Letter. Although many sentences are deleted, we believe that the reader will have enough information to judge the validity of the arguments from the current version of the manuscript. We reply to the referee's comments below. *** Major concern 1 We should first say our proposal of a jet is very speculative and we do say this in our abstract. The purpose of the jet is to explain a component of the polarization of the supernova data for Feburary that we identify from the data. This component measured as polarized flux appears to be a polarized, redshifted reflection of the flux spectrum: the redshift being that caused by a velocity of order 0.23c. The identification cannot be certain because there is not exact agreement between the flux spectrum redshifted and scaled to the observed polarized flux. And in fact we would not expect exact agreement. If a jet or jets are thrown out of a supernova during the explosion phase, the supernova is likely to exhibit other asymmetries that contribute to polarization. We note that Leonard et al. (2002, astro-ph/0206368) agrees with us that the flux spectrum redshifted and appropriately scaled does provide a partial match to the observed polarized flux spectrum. In Section 4.3 ***of the previous version***, we suggested some jet parameters (i.e., 0.05 M_sun at of order 0.23c, etc.) that can yield the right order of polarization component we have identified and account for its disappearance by our later observations. The parameters we ***suggested*** are not a unique choice, and ***this point was emphasized also*** in the paper. The particular choice we made was partially made because one of us (Jeffery) had the results in hand from a calculation that preceded the discovery of SN 2002ap. According to the referee and the paper by Berger et al. (2002, astro-ph/0206183), that reports the radio observations and modeling of SN 2002ap, our suggested jet parameters and the radio observations and modeling are inconsistent. The referee asks us to address this inconsistency. We note that we are not experts in radio observations and modeling ourselves, and so here we will not challenge the conclusions of Berger et al. Consistency can perhaps be achieved by reducing the speed of the jet model. The redshift of 0.23c is what our data suggests, but this can be obtained without having the characteristic scattering location of the jet moving a 0.23c. If the jet is not at 90 degrees to the line of sight but at a smaller angle $\theta$ and behind the supernova, then there are in effect two redshifts: one from going into the jet frame and one from coming out of it in the somewhat opposite direction: the redshift of the scattered polarized flux would then be given by \lambda'=\lambda[ 1 + (1+\cos\theta)(v/c) ] If theta is smallish, then our requirement on (v/c) can be made smaller, *** i.e., the lower limit of v is 0.23c / 2 = 0.115c***. Of course, the polarization by Thomson scattering decreases as \theta is made smaller. It is left to later work if we can get the required polarization with v significantly smaller than 0.23c. In this regard it would be useful to find out how large a (v/c) the Berger et al. observations and modeling can tolerate. We have decided to drop from the paper all mention of parameters for the jet, except the essential velocity 0.23c causing the redshift that we use to explain the polarization component we have speculatively identified. Our original set of parameters are inconsistent with the Berger et al. observations and modeling as noted above. Proposing another, more consistent set would take more modeling which is not appropriate for letter of which our jet speculation is only part. Also it seems that any set we propose may be taken as a claim to have established a reliable set no matter what we say to the contrary. For example, Berger et al. imply that we made a claim for reliability for our original set when we explicitly did not. Major Concern 2 If there is a jet is thrown out of the core of the exploding supernova, then it is plausible that it carries some radioactive Ni-56. The gamma-rays from decay would keep the jet ionized to some degree just as they later keep the nebular phase bulk ejecta ionized. We now mention this possibility briefly. We could also say here that the jet is illuminated from below by the supernovae and this may provide some ionization. X-rays from the circumstellar interaction might also contribute. However, the modeling of the ionization state of a jet is beyond the scope of our letter. Major Concern 3 We disagree that it is really critical to present more details of the jet model that we are thinking terms of. Since we are dropping the mention of possible parameters, the details of the model would become rather superfluous. Our paper is a letter of which the jet hypothesis is only a part. Also the basic picture is clear enough. A fraction of the supernova flux is redshifted, polarized, and reflected into the line of sight by a jet (or a clump) that breaks the symmetry of the supernova. The polarized flux is diluted by the flux from the bulk of the supernova (which is itself somewhat polarized). The net polarization level turns out to be smallish: of order 0.4 % or less in the continuum. However, even such small polarization requires significant asymmetry of some sort. As we remark in the paper, the ellipsoidal ejecta polarization model would require an axis ratio on the plane of the sky that differed from 1 by of order 10 %. For the referee's interest, the polarization jet model (Jeffery 2002, in preparation) is a very simple, semi-analytic, non-relativistic model. The purpose of the model is to allow a semi-quantitative understanding of the polarization a jet would cause including it's time dependence and to allow a 1st order analyis of the observed data. The model has a number of free parameters which unfortunately are hard to set independently. Because it is a very simple and semi-analytic model, it cannot span the set of all possible jet models. The model was in fact invented in 2001 and the paper on it has not been finished merely because of time constraints on the author. Relativistic effects were not included in the model because they are a complication and because Jeffery does not believe highly relativistic jets could contribute to the optical polarization. The referee is correct that moderate relativistic effects would make it harder (although not overwhelmingly so) to obtain simultaneously the redshift and polarization of the scattered flux that we require from the jet model for SN 2002ap. If we do decrease jet velocity below 0.23c, the relativistic problem is decreased. Major Concern 4 The referee has not understood the effect of redshift in the model. Say that the bulk of the supernova ejecta emits a flux F(lambda) toward the observer. A fraction f of this flux is redshifted, scattered and polarized by the jet emitted in the direction of the observer. Assuming f is rather small and the non-relativistic approximation can be made and the bulk supernova flux is unpolarized, then the polarization of the net flux at wavelength lambda is P(lambda) = f P(theta) F( lambda(1+(1+cos\theta)(v/c) ) / F(lambda) , where theta is the angle of the jet from the line of sight and P(theta) is the polarization of the scattered flux (100 % for 90 degree scattering). The polarizaed flux at lambda comes from a bluer part of the spectrum. If v/c=0, then the polarization would be constant: such a jet is not tenable: the jet must be well outside of the photosphere for the approximation to be made that jet and photosphere are point-like. For v/c of significant size, the polarization will vary with wavelength because of the variation of F(lambda) with wavelength. The variation would be particularly strong about line profiles. We have added a clarifying remark in the text. (David, I add the equation of the polarization level due to the scattered, redshifted light spectrum, which will be helpful for the readers.) Minor Concern 1 We do not reference the papers of Cropper et al (1988) and Barrett et al (1988) because the former does not correct the data for interstellar polarization (ISP) and the latter uses an outdated ISP value. Our references to the SN 1987A polarization data are to support effects that are only reliably seen after a good ISP correction. We use a reference that uses the best ISP correction we know of for SN 1987A data: i.e., Jeffery, 1991, ApJS, 77, 405. The Wang et al. (1996) reference is an important one and we ***include*** it. Minor Concern 2 We agree with the referee that it would be interesting to see if the data from the February and March epochs can be analyzed on a day by day basis. However, for a 1st order analysis in a letter, using February and March averages seemed reasonable since the day to day changes in those epochs are not striking. We do, in fact, present all the day to day data in Figure 1. A day by day analysis would be particularly enhanced if our data were combined with that from other groups (Leonard et al. 2002, astro-ph/0206368; Wang et al 2002, astro-ph/0206386). *** About other revisions 1. We deleted many sentences over the whole text in order to shorten the text length within the 4 page limitation of the Letter. 2. We shorten the subtitle from "Evidence for a high velocity asymmetric explosion of a hypernova" to "A high velocity asymmetric explosion" because some words in the previous subtitle are not appropriate. 3. Additionally we introduce the observation on June 9, two weeks after the submission of this paper. It is cited only for the confirmation of the previously estimated ISP parameters, and does not affect any conclusion of the discussion. However, we consider the presentation of the June data is still important. 4. The panels for the averaged values in the Figure 1 (d,e) are deleted for the shortening of the text length. *** -------------------- TeX source of the manuscript -------------------- \documentclass{aastex} %\documentclass[12pt,preprint]{aastex} % \documentclass[preprint2]{aastex} %\newcommand{\vdag}{(v)^\dagger} %\newcommand{\myemail}{skywalker@galaxy.far.far.away} %\slugcomment{Not to appear in Nonlearned J., 45.} %\usepackage{emulateapj5} %\submitted{Submitted to ApJL} \shorttitle{Optical Spectropolarimetry of SN~2002ap} \shortauthors{Kawabata et al.} \begin{document} \title{Optical Spectropolarimetry of SN~2002ap: A High Velocity Asymmetric Explosion\footnotemark[1]} \footnotetext[1]{Based on data obtained at the Subaru Telescope, which is operated by the National Astronomical Observatory of Japan (NAOJ)} \author{K.~S.~Kawabata\altaffilmark{2,3}, D.~J.~Jeffery\altaffilmark{4}, M.~Iye\altaffilmark{2,5}, Y.~Ohyama\altaffilmark{6}, G.~Kosugi\altaffilmark{6}, N.~Kashikawa\altaffilmark{2}, N.~Ebizuka\altaffilmark{7}, T.~Sasaki\altaffilmark{6}, K.~Sekiguchi\altaffilmark{6}, K.~Nomoto\altaffilmark{8,9}, P.~Mazzali\altaffilmark{9,8,10}, J.~Deng\altaffilmark{9,8}, K.~Maeda\altaffilmark{8}, K.~Aoki\altaffilmark{6}, Y.~Saito\altaffilmark{2}, T.~Takata\altaffilmark{6}, M.~Yoshida\altaffilmark{11}, R.~Asai\altaffilmark{8}, M.~Inata\altaffilmark{2}, K.~Okita\altaffilmark{11}, K.~Ota\altaffilmark{8,2}, T.~Ozawa\altaffilmark{12}, Y.~Shimizu\altaffilmark{11}, H.~Taguchi\altaffilmark{13}, Y.~Yadoumaru\altaffilmark{12}, T.~Misawa\altaffilmark{8,2}, F.~Nakata\altaffilmark{8,2}, T.~Yamada\altaffilmark{2}, I.~Tanaka\altaffilmark{2}, and T.~Kodama\altaffilmark{14} } \altaffiltext{2}{Opt. \& IR Astron. Div., NAOJ, Mitaka, Tokyo 181-8588, Japan} \altaffiltext{3}{E-mail: koji.kawabata@nao.ac.jp} \altaffiltext{4}{Department of Physics, New Mexico Institute of Mining and Technology, Socorro, NM 87801, USA} \altaffiltext{5}{Department of Astronomy, Graduate University for Advanced Studies, Mitaka, Tokyo 181-8588, Japan} \altaffiltext{6}{Subaru Telescope, NAOJ, 650 North A'ohoku Place, Hilo, HI 96720, USA} \altaffiltext{7}{RIKEN, Wako, Saitama 351-0198, Japan} \altaffiltext{8}{Department of Astronomy, University of Tokyo, Bunkyo-ku, Tokyo 113-0033, Japan} \altaffiltext{9}{Research Center for the Early Universe, University of Tokyo, Bunkyo-ku, Tokyo 113-0033, Japan} \altaffiltext{10}{Osservatorio Astronomico, Via Tiepolo, 11, 34131 Trieste, Italy} \altaffiltext{11}{Okayama Astrophysical Observatory, NAOJ, Asakuchi-gun, Okayama 719-0232, Japan} \altaffiltext{12}{Misato Observatory, Amakusa-gun, Wakayama 640-1366, Japan} \altaffiltext{13}{Dep. of Astron \& Earth Sci., Tokyo Gakugei University, Koganei, Tokyo 184-8501, Japan} \altaffiltext{14}{Theory Division, NAOJ, Mitaka, Tokyo 181-8588, Japan} \begin{abstract} We present spectropolarimetry of the Type Ic supernova SN 2002ap, taken at two epochs, close to and one month later than the visual maximum (2002 February 8). In addition we present a later spectropolarimetry being supplementary to the analysis. The data show the development of linear polarization. Distinct polarization profiles were seen only in the \ion{O}{1}~$\lambda 7773$ multiplet/\ion{Ca}{2}~IR triplet absorption trough at maximum light and in the \ion{Ca}{2}~IR triplet absorption trough a month later, with the latter showing a peak polarization as high as $\sim 2$ \%. %We estimate the interstellar polarization toward the supernova to be %Serkowski law polarization with %$p_{\rm max}=0.64\pm 0.20$ \%\ and %$\theta_{\rm ISP}=120\arcdeg\pm 10\arcdeg$. The intrinsic polarization shows three clear position angles: $80\arcdeg$ for the February continuum, $120\arcdeg$ for the February line feature, and $150\arcdeg$ for the March data. %Since the position angle varies, the supernova cannot be %exactly axisymmetric. We conclude that there are multiple asymmetric components in the ejecta. We suggest that the supernova has a bulk asymmetry with an axial ratio projected on the sky that is different from 1 by of order $10$ \%. %, together with a polarizing jet or jets moving at $0.23c$. Furthermore, we suggest very speculatively that a high velocity component moving at faster than $\sim 0.115c$ contribute to polarization in the February epoch. \end{abstract} \keywords{polarization --- supernovae: individual (SN 2002ap)} \section{INTRODUCTION} SN~2002ap was discovered in the nearby spiral galaxy M74 ($=$ NGC 628) on 29 January 2002 \citep{nak02} and reached its maximum of $V\sim 12.4$ mag on February 8 \citep{gal02}. It has been classified as a Type~Ic supernova (SN~Ic) and recognized as a hypernova (but at the low-energy end of the sequence of hypernovae) from the fact that its early spectra had very broad absorption lines (\citealt{maz02} and references therein). On the other hand, \citep{ber02} suggest that SN~2002ap is rather an ordinary Type Ib/c SN from their analysis of the radio emission. Because of its brightness, SN~2002ap provided us with a unique opportunity to carry out multi-epoch, high-quality spectropolarimetry of the peculiar supernova. %using the Subaru Telescope. %Unfortunately, spectropolarimetric observations are demanding in %terms of telescope time, and only two epochs (February 9--12 and March %8--10) could be observed. %In addition, Unfortunately, the supernova was lost behind the Sun from mid-March to early June, limiting our ability to observe it in the brightest phase. %In \S~2 of this Letter, we review current knowledge on SNe~Ic, %hypernovae, and supernova polarization. %\S~3 reports our new observations and the data reduction process. %Our results and discussion appear in \S~4 and conclusions in \S~5. \section{SNe~IC, HYPERNOVAE, SUPERNOVA POLARIZATION} A SN Ic is thought to be the result of the core collapse of a massive star that has either lost its hydrogen and helium envelopes prior to the explosion or has an invisible helium envelope due to low excitation. The details of the explosion mechanism are still under discussion (\citealt{nom95,bra01} and references therein). It has recently been recognized that there is a subgroup of SNe Ic: Members of this subgroup exhibit very broad absorption lines in their early spectra. The spectra have been successfully modeled as the hyperenergetic explosion of a massive C$+$O star, with an explosive kinetic energy exceeding $\sim 5$--$10$ times as much as that of normal core-collapse SNe (\citealt{iwa98,nom01} and references therein). %This is much larger than the kinetic energy of normal core-collapse %SNe, which are $\sim 10^{51}$ ergs. SN 1998bw, the most luminous and energetic `hypernova' to date, has been particularly well studied, and its probable connection with the $\gamma$-ray burst GRB 980425 has been pointed out (e.g., \citealt{gal98}). An aspherical explosion has been suggested to explain the slowly-declining light curve of SN 1998bw and the narrowness of the \ion{O}{1} $\lambda$6300 emission line in the nebular phase \citep{maz01,nak01,mae02}. Alternatively, \citet{hoe99} suggested that the observed behavior could even be explained by a moderate explosion ($2\times 10^{51}$ ergs) if the ejecta had a prolate asphericity with an axial ratio of about 2 and were viewed close to the symmetry axis. %Because during the photospheric epoch the optical continuum opacity %is dominated by electron scattering, supernova %atmospheres are intrinsically highly polarizing. Since SNe are unresolved, local polarization is geometrically canceled out as long as the explosion is spherically symmetric: the presence of any polarization thus reveals asymmetry. It has been found that core-collapse supernovae are generally polarized at levels of $p\simeq 0.5$--$4$ \%\ and that the level of polarization increases after optical maximum light (e.g., \citealt{jef91b,wan96,wan01,leo01}); however, the polarization falls to zero at very late times, when the electron scattering opacity becomes very low (e.g., \citealt{jef91b}). %In the photospheric epoch, the expanding SN envelope gives rise %to broad P~Cygni line profiles consisting of line-centered emissions %and blueshifted absorption troughs. %The polarization level at the P~Cygni absorption troughs can be much %higher than in the continuum, due to the removal of unpolarized flux %that comes straight from the photosphere. %At the P~Cygni line emission feature, on the other hand, %polarization can fall below the continuum %value owing to the dilution of electron-scattered polarized light %by line-scattered, unpolarized photospheric flux, and/or NLTE %line-emitted unpolarized flux. %Moreover, line scattering can depolarize the electron scattered flux, %further reducing polarization at the flux maximum \citep{how01}. The typical line profile of the polarization level, predicted theoretically \citep{jef89} and to some degree confirmed observationally in SN~1987A and other supernovae \citep{jef91a,jef91b,leo01}, is %therefore an inverted P~Cygni profile. However, line blending and other intrinsic effects may affect these profiles. %Moreover, interstellar polarization (ISP) can completely %distort them, and even interchange the location of polarization maxima %and minima. %To understand the intrinsic polarization and the asymmetry of %supernovae, the effects of ISP must be corrected for (see \S~4.1). For SN 1998bw, an intrinsic optical polarization of $0.4$--$0.6$ \%\ was found, suggesting an asymmetry of less than $2/1$ in the axial ratio of the ejecta \citep{pat01,kay98}. However, no distinct polarization feature was revealed associated with the spectral lines, probably due to the poor S/N or the relatively narrow wavelength range of those observations. \section{OBSERVATIONS AND DATA REDUCTION} The spectropolarimetric observations were made with the 8.2-m Subaru Telescope equipped with the Faint Object Camera and Spectrograph (FOCAS, \citealt{kas00,yos00}). The observing log is shown in Table \ref{tbl-1}. The linear polarimetric module consists of a rotating superachromatic half-wave plate and a crystal quartz Wollaston prism, and both the ordinary and the extraordinary rays are simultaneously recorded on two MIT/LL CCDs (2k$\times$4k$\times$15\micron ). %The dispersion angle between {\it o-} and {\it e-}rays is about %0\fdg 5, and we have to use a focal plane slit of length shorter %than 10.3 mm (corresponding to $21\farcs 2$ in the projected sky) %to avoid overlaps of {\it o-} and {\it e-}spectra. %For the observations in February we used a slit of $0\farcs 4$ width %$\times 20\farcs 6$ length and a grism of 300 gr/mm and 5500 \AA\ %blaze, which gives a resolution of %For the observations in March we used a slit of $0\farcs 8$ width %$\times 20\farcs 6$ length and a grism of 300 gr/mm and 7500 %\AA\ blaze, which gave a resolution of approximately 11 \AA. %To suppress contamination due to second-order spectra we used %order-cut filters. A typical observing sequence consisted of four integrations at the $\psi=0\arcdeg, 45\arcdeg, 22\fdg 5$ and $67\fdg 5$ positions of the half-wave plate. %The frames were reduced according to standard procedures for %CCD spectroscopy. %Since no field object was seen in the images obtained, %we estimated the sky background component simply by %an interpolation of the spectra at both sides of the object %along the slit. Stokes $Q/I$ and $U/I$ were calculated as in \S 6.1.2 of \citet{tin96}. For polarimetric calibration, we obtained data for unpolarized and polarized standard stars, including measurements of flatfield lamps through fully-polarizing filters. Although the stability of instrumental polarization and depolarization in FOCAS on the Subaru Telescope have not yet been fully calibrated, our results indicate that the instrumental polarization ($\lesssim 0.1$ \%) and the depolarization factor ($\lesssim 0.05$) are negligible at all wavelengths. %The results of observations for strongly polarized standard stars %are tabulated in Table \ref{tbl-2}. %Thus, we give no correction for them. %The zero-point of the position angle on the sky was determined %from the observations of the polarized standard stars. %All our target observations were done through the atmospheric %dispersion correction (ADC) optics at the Cassegrain focus of the %Subaru Telescope. %Our calibration observation suggests that the influence of the %ADC on spectropolarimetry is also negligible. %(cf. Table \ref{tbl-2}). The flux was calibrated using observations of G191B2B and BD+28$\arcdeg$4211 \citep{oke90} using the same slit with the target observation, and then was multiplied by a constant to match the VSNET\footnote{\url{http://www/kusastro.kyoto-u.ac.jp/vsnet}} photometric data. \section{RESULTS AND DISCUSSION} Figure \ref{fig1} shows the observed flux and polarization spectra. Several blueshifted broad absorption lines can be identified in the February spectrum \citep{maz02}. %\ion{Fe}{2} $\lambda\lambda$4924, 5018, 5169, %\ion{Na}{1} $\lambda\lambda$5890, 5896, %\ion{Si}{2} $\lambda\lambda$6347, 6371, %and probably the blended lines of %\ion{O}{1} $\lambda$7773, %\ion{O}{1} $\lambda8446$ and %\ion{Ca}{2} $\lambda\lambda$8498, 8542, 8662 %(i.e., the \ion{Ca}{2} IR triplet). For March and June spectra a detailed analysis has yet to be done. However, we note that the March spectra resemble that of SN~1997ef at day 67 \citep{maz00}, including the onset of net emission in \ion{Ca}{2} $\lambda\lambda$8498, 8542, 8662 (i.e., the \ion{Ca}{2} IR triplet). %The SN 1997ef analysis implies that the lines in the March spectra %are essentially the same as those present in the February spectra, %and confirms that SN~2002ap evolves at about twice the rate as %SN~1997ef. The significant emission line at $\sim 6300$\AA\ in the June spectrum could be considered as [\ion{O}{1}] $\lambda$6300, 6363 \citep{maz01}. %In the march spectra the \ion{O}{1} $\lambda$7773 and the %\ion{Ca}{2} IR triplet troughs are clearly separated: they are visible at %$\sim 7600$ \AA\ and $\sim 8250$ \AA, respectively. The flux is generally polarized by $p\gtrsim 0.5$ \%\ at a position angle (PA) of $\theta =120\arcdeg \pm 20\arcdeg$ over the observed wavelengths. It is noted no significant variation is seen in the polarization spectra except for some noisy points within each month. In this Letter, the daily variation is not discussed, which is left for a future work. %there is apparently little variation in %polarization within each month, %we have averaged the data for each month to derive higher S/N %polarization spectra. \subsection{Interstellar Polarization} The level of an interstellar polarization (ISP) varies slowly with wavelength in the optical, and is well approximated by the empirical formula: $p_{\rm ISP}(\lambda ) = p_{\rm max} \cdot \exp [ -1.15 \ln^{2}( \lambda_{\rm max}/\lambda) ]$, where $p_{\rm max}$ is the peak polarization level occurring at wavelength $\lambda_{\rm max}$ \citep{ser75}. %A good method for estimating the effects of ISP is to use the %observed polarization of flux that %is known {\it a priori} to be of low intrinsic polarization. In March the emission feature of the \ion{Ca}{2} IR triplet line profile shows strong net emission due to NLTE processes. Such NLTE line flux is necessarily unpolarized on emission. Since the line profile is still broad (absorption peak $\sim -14,000$ km s$^{-1}$, corresponding to an enclosed mass of 2 M$_{\odot}$ in model CO100/4, which has a total mass of 2.4 M$_{\odot}$), much of the emission is probably coming from far out in the ejecta, where the electron optical depth is low. We conclude that the flux from this emission line is mostly unscattered by electrons and unpolarized, and that it dilutes the polarized electron scattered flux. If this were the only effect, then the intrinsic polarization should show a distinct minimum exactly at the wavelength of the flux emission maximum. %In the observed polarization, such a minimum would show up %as a feature that varied relatively rapidly with wavelength. Since, in fact, the polarization is roughly constant across the P~Cygni emission feature ($8400$--$9000$ \AA ), apart from small variations that may be mostly noise, we conclude that the line is not only diluting the polarized flux but is also strongly depolarizing it. Thus the intrinsic polarization across the emission feature is probably close to zero, and the observed level of polarization in this region is close to the ISP level. We will assume that the observed polarization in this region is all due to ISP: thus $p_{\rm ISP}(8600\;\AA )\approx 0.5$ \%. Now \citet{ser75} find that $\lambda_{\rm max}$ for 30 stars with $p_{\rm max} / E_{B-V}\geq 7.0$ has a median value of $5370$ \AA\ and an rms deviation of $400$ \AA . We adopt this median value to derive our ISP estimate from a non-linear regression: $p_{\rm max}=0.64\pm 0.20$ \%\ and $\theta_{\rm ISP}=120\arcdeg\pm 10\arcdeg$. (The uncertainties are crude estimates based on the alternative assumption that only line flux dilution, and not line depolarization, occurs in the region of the \ion{Ca}{2} IR triplet emission feature.) Since \citet{tak02} derive a color excess for SN~2002ap of $E_{B-V}=0.09$ (a sum of $0.07$ within our Galaxy and $0.02$ within M74) from interstellar Na D absorptions, our assumption of the \citet{ser75} $\lambda_{\rm max}$ is consistent: $p_{\rm max} / E_{B-V} = 0.64 / 0.09 \approx 7$. The estimated ISP is consistent with other observational indications. In the recently compiled ISP catalog \citep{hei00}, 16 stars are recorded within $10\arcdeg$ of SN~2002ap. The data for these stars suggest a possible positive correlation between polarization level and the distance along the line of sight toward SN~2002ap. The two most distant stars among them, HD8919 ($d=525$ pc) and HD9560 ($d=437$ pc) show $(p,\ \theta) = (0.32\pm 0.10$ \mbox{ \%}, $99\arcdeg\pm 9\arcdeg)$ and $(0.48\pm 0.09$ \mbox{ \%}, $123\arcdeg\pm 5\arcdeg)$, respectively. On the other hand, it has been found that $p_{\rm max} (\% )$ has an empirical upper limit of $9E_{B-V}$ \citep{ser75}. From the derived $E_{B-V} = 0.09$, an upper limit on the ISP toward the supernova is $0.81$ \%. The estimated ISP is nicely sandwiched between the possible lower bounds of the cited stars and the empirical upper limit. The estimated ISP position angle is also consistent with the position angle of the spiral arm in M74 at the position of SN~2002ap, $110\arcdeg$--$140\arcdeg$ (cf. DPOSS images). %It will be possible to obtain a more definitive ISP from %spectropolarimetry of the supernova when it has evolved well %into the nebular phase and the electron scattering optical depth %has become negligible. %Unfortunately, the supernova will then be relatively dim, %and measurement correspondingly more difficult. The polarization spectrum obtained on June 9 has only low S/N. It is noted, however, that the polarization values at the peak of the strong emission flux at the [\ion{O}{1}] $\lambda$6300, 6363 forbidden line seems roughly consistent with the estimated ISP (Figure \ref{fig1}c). In the following, we concentrate our attention to the higher S/N polarization spectra in the earlier two epochs. \subsection{Intrinsic Polarization} Figure \ref{fig2} shows the intrinsic polarization (calculated using the estimated ISP) plotted on a {\it QU} diagram: the polarization points are connected according to their wavelength ordering. Given the uncertainty in the estimated ISP, points within $0.2$ \%\ of the origin must be considered very uncertain when drawing conclusions about the intrinsic position angle. If one assumes that the intrinsic supernova polarization is produced by a single axisymmetric component in the ejecta, then the intrinsic polarization plotted on a {\it QU} diagram should lie on a line passing through the origin. %ISP can displace the line but, because of its slow wavelength %dependence, cannot strongly distort it. %Figure \ref{fig2} shows that the SN~2002ap intrinsic polarization %cannot arise from a single axisymmetric component. It can be seen that the polarization in February has two clear position angles, PA less than or $\sim 120\arcdeg$ (associated with the \ion{O}{1}/\ion{Ca}{2} line trough) and PA$\sim 80\arcdeg \pm 20\arcdeg$ (associated with the continuum from $\sim 5700$--$8200$ \AA), joined by a somewhat complicated transition. %The slow wavelength dependence of the position angle is also seen %clearly in Figure \ref{fig3}c. The polarization in March has a clear position angle PA$\sim 150\arcdeg$ associated with the \ion{Ca}{2} line trough and with at least some of the continuum. We must conclude that there are multiple asymmetric components in the supernova, and that their contribution varies with time. It is likely that the recession of the supernova photosphere uncovers different asymmetries. \subsection{Possible Models} Figures \ref{fig3}a, b, and c show the flux corrected for heliocentric redshift [$v_{\rm helio}=+631 \mbox{\rm km s}^{-1}$ \citep{sma02}] and interstellar extinction [$E_{B-V}=0.09$ \citep{tak02}] and the polarization due to the estimated ISP. The figures show that polarization is low and barely significant (given the uncertainty in the estimated ISP), except in the regions $\sim 6700$--$8000$ \AA\ for February and $\sim 6700$--$8300$ \AA\ for March. %The position angle outside the regions of high polarization varies %wildly in places: the variations are probably not significant. The low level of the continuum polarization blueward of $\sim 6700$ \AA\ in both epochs may be due to the depolarizing effect of lines: in supernova spectra, lines generally become stronger further to the blue. The level of the continuum polarization, where it is significant (a relatively small region) is $\sim 0.4$ \% in both epochs. If the asymmetry is assumed to be an axisymmetric, global prolate or oblate asymmetry, then a continuum polarization of this level can be explained by an axial ratio (assuming there is a main axis) projected on the sky that is different from 1 by of order $10$ \%. This asymmetry estimate is a crude one based on realistic, but parameterized, calculations \citep{hoe91,hoe95}. The estimate is also crude because, as noted above, the asymmetry cannot be completely axisymmetric. The estimated asymmetry is not large, compared to those estimated for some other supernovae (e.g., \citealt{wan01}). %It may be possible that the actual asymmetry is large but its %projection on the sky is small. %However, this would imply that we are viewing the SN from %close to the `jet' axis, which is in apparent contradiction with %the absence of an observed GRB. Therefore, any global asymmetry is %probably not extremely large. %Another possibility is that the asymmetry is due to elemental %inhomogeneities in the ejecta. %These would affect the ionization balance differently %in different regions and hence the %electron scattering optical depth would be different in different directions. %The ionization balance could be especially affected by %radioactive $^{56}$Ni spread far out into the ejecta, %as found by \citet{mae02} in an asymmetric explosion %calculation. %Radioactive, non-thermal ionization would probably enhance the electron %scattering opacity. %On the other hand, iron-peak elements have many lines %in the optical that would tend to increase line %depolarization. The three distinct line polarization profiles seen in Figure~\ref{fig3} (at the \ion{O}{1}/\ion{Ca}{2} flux absorption in February and the \ion{O}{1} and \ion{Ca}{2} flux absorptions in March) can partially be accounted for by the inverted P~Cygni profile %discussed in \S~2: (cf. \S~2): the polarization maxima associated with line trough features are clear. Without detailed modeling more information probably cannot be extracted from these profiles. %The redshifting and narrowing of the \ion{O}{1}/\ion{Ca}{2} %flux absorption around maximum light %account for the similar behavior in the polarization %maximum associated with this feature \citep{wan02,kaw02}. For the polarized continuum observed in February, we can suggest a radically different origin from the bulk asymmetry assumed in the prolate/oblate models or element inhomogeneity models (see below). It is possible that this continuum polarization results from scattering off electrons (the scattering coefficient is independent of the wavelength) in a rapidly expanding ejecta. In Figure~\ref{fig3}d we show the intrinsic polarized flux ($p\times F$) compared to the observed flux scaled down by a factor of $0.0023$ and redshifted by a velocity of $0.23c$. There is fair agreement over the range $\sim 5000$---$8000$ \AA. This agreement suggests that a large component of polarized flux comes from scattering by an ionized blob (e.g., a jet) moving at $\gtrsim 0.115c$. (The light scattered off is redshifted by $v_{\rm sca}\sim v_{\rm jet}(1\pm \cos i)$, where $v_{\rm sca}$ is the characteristic velocity of scatterer, $i$ is the inclination angle of the traveling direction of the scatterer to the line of sight, and the plus and minus cases are for a jet moving away from and coming toward the observer, respectively.) The existence of high velocity jet-like clumps has been proposed in some hydrodynamic explosion models for SNe, hypernovae and GRB's (e.g., \citealt{nag97,mac99}). We call the rescaled flux the jet model continuum polarization. If the jet is thrown out of the core of the exploding supernova, then it is plausible that it carries some radioactive $^{56}$Ni. The gamma-rays from decay would keep the jet ionized to some degree just as they kep the nebular phase bulk ejecta ionized. %Recall that blueward of $\sim 6700$ \AA\ the intrinsic polarization %is very low and uncertain. %, which varies %directly with the moving velocity for right angle scattering, %$\alpha=90\arcdeg$.) %Bipolar jets at $90^{\circ}$ to the line of sight, with %an opening angle of $10^{\circ}$, a length comparable to the %distance from the bulk ejecta, and consisting of about $0.05M_{\odot}$ %of doubly ionized C/O can produce a continuum polarization of %the total supernova emission of order $\sim 0.4$ \% %at about 13 days after explosion \citep{jef02}. %(The explosion of SN~2002ap occurred on about January~28 \citep{maz02} %or about 13 days before our February polarization observations.) %The cited parameters are only suggested as reasonable characteristic %values that can account for the continuum polarization: a %large choice of jet parameters can yield a similar polarization. %The jet mass that we suggest is not unreasonable: %the total mass of ejected matter in SN~2002ap has been %estimated to be $2.5$--$5$ M$_{\sun}$ \citep{maz02}. %If hypernovae are to account for GRB's, the jet velocity would need to %be close to $c$. %However, the value $0.23c$ that we use could plausibly be the speed of %the slow tail of such a highly relativistic jet. % %With the parameters given above, the jet polarization contribution %would be only of order $0.05$\% by the March polarization observation %epoch (i.e., about 40 days after explosion) and so would be an almost %insignificant contribution to the net polarization. %The continuing expansion of the jet gives its optical depth %an inverse square dependence on the time since explosion. %(The optical depth will also depend on any variation %in the jet's ionization.) % %There is nothing inconsistent in assuming an early jet contribution to %polarization, which subsequently vanishes, along with a continuing net %polarization. The ejection of fast jets in the explosion would very %likely be accompanied by a bulk asymmetry that would continue to cause %polarization long after the jets had become optically thin and %their polarization effects negligible. The bulk ejecta, %which contain much more matter than the jets, become optically thin and %non-polarizing much later than do the jets. For the jet picture to be %correct in our case, the polarization due to the bulk ejecta would be %expected to increase with time after the February observation epoch. %This is possible, since polarization increasing with time has been %observed in other supernovae (e.g., \citealt{jef91b,wan01,leo01}) However, from the radio observations \citep{ber02} suggested that the energy in relativistic electrons and magnetic fields is only $\approx 1.5\times10^{45}$ ergs in ejecta (with a velocity $\approx 0.3c$). The results apparently restrict the the mass and the velocity of the possible jet. A consistent modeling of the jet polarization is required, whereas it is beyond the scope of this Letter. The analysis will be presented in the forthcoming paper \citep{jef02}. If the jet picture is indeed correct, then the position angle of the jet on the sky is $\sim 170^{\circ}$ since electron scattering polarizes perpendicularly to the scattering plane. %and the position angle of the February continuum polarization is $\sim 80^{\circ}$. %The angle made by the jet to the line of sight, $i$, cannot %be determined without further information. %The jet is observationally maximally polarizing for $i=90^{\circ}$ %and, if the jet is truly axisymmetric, unpolarizing for $i=0^{\circ}$. %(The electron scattering polarization of natural light varies %from $0$ to $100$ \% as the scattering angle increases %from $0^{\circ}$ to $90^{\circ}$.) The \ion{O}{1}/\ion{Ca}{2} line polarization maximum in the February data cannot easily be associated with the jet. %In a simple jet picture, the position angle of this feature would %have to be $80^{\circ}$, matching the continuum. In reality, the observed position angle makes an excursion from $\sim 80^{\circ}$ up to $120^{\circ}$ across the polarization maximum. Some of the line polarization would therefore have to arise in the bulk asymmetry of the supernova. It is possible that the position angle of $\sim 120^{\circ}$ is the net result of a jet polarizing at $\sim 80^{\circ}$ and a bulk asymmetry polarizing at $\sim 150^{\circ}$ (i.e., at the position angle observed in the March data). To test this model we have subtracted the jet model continuum polarization, $p_{\rm jet}(\lambda) = F(\lambda (1-\beta ))/F(\lambda)\cdot (1-\beta) \cdot p_{c}$, from the February intrinsic polarization, where $F(\lambda)$ is the corrected flux (Figure \ref{fig3}a), $\beta =1-v_{\rm sca}/c$, $v_{\rm sca}=0.23c$, and $p_{c}=0.0023$. The residual polarization and position angle spectra are plotted in Figures \ref{fig3}e,f. The position angle of the residual polarization for February from the region of significant polarization (i.e., $\sim 6700$--$8000$ \AA) is now approximately centered on $150^{\circ}$ and deviates by more than $30^{\circ}$ only in a few isolated points. The results in Figure~\ref{fig3}f are thus consistent with the jet model. The jet may not be a completely separated amount of ejecta, but rather a blob rich in $^{56}$Ni moving at $\gtrsim 0.115 c$. A high-velocity, $^{56}$Ni-rich region is required both in theoretical hypernova explosion models (\citealt{nak01,mae02}), and in SN~2002ap (\citealt{maz02}) in order to reproduce the light curve. Note that, the maximum ejecta velocity indicated by spectrum synthesis of SN~2002ap is $\sim 0.22 c$. %, which is insignificantly %different from the possible redshift of the polarized continuum. A blob would affect the ionization balance differently in different regions of the outer ejecta and hence the electron scattering optical depth would be different in different directions. The ionization would likely be increased by radioactive $^{56}$Ni and this would make the blob more polarizing than other parts of the ejecta at the same velocity. %On the other hand, iron-peak elements have many lines %in the optical that would tend to increase line depolarization %particularly on the blue side of the optical. The net effect of a blob is difficult to estimate. Other chemical inhomogeneities in the ejecta at varying velocities are also possible \citep{mae02} and would affect polarization in complicated ways. %As discussed in \S4.3, the electron density in a $^{56}$Ni-rich region could %be increased as a consequence of nonthermal ionization, leading to higher %polarization. %Finally, the fact that only the spectrum between rest wavelength 3000-6000\AA\ %shows a polarized continuum may be consistent with the large number of active %Fe lines in that wavelength range. Fe lines would naturally form in a %$^{56}$Ni-rich region. \section{CONCLUSIONS} In this Letter we have presented spectropolarimetry for SN~2002ap and given its first order interpretation. We suggest that the supernova has a bulk asymmetry with an axial ratio projected on the sky that is different from 1 by of order $10$ \% together with a polarizing jet(s) moving at $\gtrsim 0.115c$. The jet(s) makes a significant contribution to the polarization only in the February observational epoch. Undoubtedly more realistic modeling is necessary for a further definitive understanding of the polarization. %Given the expected orientation of the jet, this model predicts %that little or no gross asymmetries will be visible in the nebular %spectrum (fast Fe is roughly tangential to the line of sight). The degree of the bulk asymmetry predicted in this model may be tested with the line widths and their ratios in the nebular spectrum \citep{mae02}. %Our observations are representative of the high-quality %spectropolarimetry of supernovae that will increasingly become %available from very large telescopes. %With detailed analysis, these data will give further %insights into these energetic explosions. \acknowledgments We are grateful to the staff members at the Subaru Telescope for their kind help and their rearrangement of the telescope maintenance schedule for our observation on March 8. This work has been supported in part by the grant-in-Aid for Scientific Research (12640233, 14047206, 14540223) and COE research (07CE2002) of the Ministry of Education, Science, Culture, Sports, and Technology in Japan. \begin{thebibliography}{} \bibitem[Berger, Kulkarni, \& Chevalier(2002)]{ber02} Berger, E., Kulkarni, S. R., \& Chevalier, R. A. 2002, submitted to ApJL \bibitem[Branch(2001)]{bra01} Branch, D. 2001, in Supernovae and Gamma-Ray Bursts, ed. M. Livio, N. Panagia and K. Sahu (Cambridge: Cambridge Univ. Press), 96 %\bibitem[Brown \& McLean(1977)]{bro77} Brown, J. C., \& McLean, % I. S. 1977, \aap, 57, 141 \bibitem[Cardelli, Clayton, \& Mathis(1989)]{car89} Cardelli, J. A., Clayton, G. C., \& Mathis, J. S. 1989, \apj, 345, 245 %\bibitem[Cropper et al.(1988)]{cro88} Cropper, M., Bailey, J., % McCowage, J., Cannon, R. D., Couch, W. J., Walsh, J. R., Strade, J. O., % \& Freeman, F. 1988, \mnras, 231, 695 %\bibitem[Filippenko(1997)]{fil97} Filippenko, A. V. 1997, \araa, 35, 309 \bibitem[Galama et al.(1998)]{gal98} Galama, T. J., et al. 1998, \nat, 395, 670 \bibitem[Gal-Yam, Ofek, \& Shemmer(2002)]{gal02} Gal-Yam, A., Ofek, E. O., \& Shemmer, O. 2002, \mnras, 332, L73 %\bibitem[Harkness(1987)]{har87} Harkness, R. P., et al. 1987, % \apj, 317, 355 %\bibitem[Heiles(1976)]{hei76} Heiles, C. 1976, \araa, 14, 1 \bibitem[Heiles(2000)]{hei00} Heiles, C. 2000, \aj, 119, 923 \bibitem[H\"oflich(1991)]{hoe91} H\"oflich, P. 1991, \aap, 246, 481 \bibitem[H\"oflich(1995)]{hoe95} H\"oflich, P. 1995, \apj, 440, 821 \bibitem[H\"oflich, Wheeler, \& Wang(1999)]{hoe99} H\"oflich, P., Wheeler, J. C. \& Wang, L. 1999, \apj, 521, 179 \bibitem[Howell et al.(2001)]{how01} Howell, D. A., H\"oflich, P., Wang, L., \& Wheeler, J. C. 2001, \apj, 556, 302 \bibitem[Iwamoto et al.(1998)]{iwa98} Iwamoto, K., et al. 1998, \nat, 395, 672 %\bibitem[Iwamoto et al.(2000)]{iwa00} Iwamoto, K., et al. 2000, % \apj, 534, 660 \bibitem[Jeffery(1989)]{jef89} Jeffery, D. J. 1989, \apjs, 71, 951 \bibitem[Jeffery(1991a)]{jef91a} Jeffery, D. J. 1991a, \apj, 375, 264 \bibitem[Jeffery(1991b)]{jef91b} Jeffery, D. J. 1991b, \apjs, 77, 405 \bibitem[Jeffery(2002)]{jef02} Jeffery, D. J. 2002, in preparation \bibitem[Kashikawa et al.(2000)]{kas00} Kashikawa, N., et al. 2000, Proc. SPIE, 4008, 104 %\bibitem[Kawabata et al.(2002)]{kaw02} Kawabata, K. S., Kosugi, G., %Sasaki, T., Ohyama, Y., Kashikawa, N., Saito, Y., Iye, M., \& %Nomoto, K. 2002, \iaucirc, No. 7835 \bibitem[Kay et al.(1998)]{kay98} Kay, L. E., Halpern, J. P., Leighly, K. M., Heathcote, S., \&\ Magalhaes, A. M. 1998, \iaucirc, No. 6969 %\bibitem[Kinugasa et al.(2002)]{kin02} Kinugasa, K., Kawakita, H., % Ayani, K., Kawabata, T., \& Yamaoka, H. 2002, \iaucirc, No. 7811 %\bibitem[Kulkarni et al.(1998)]{kul98} Kulkarni, S. R. 1998 % \nat, 395, 663 \bibitem[Leonard et al.(2001)]{leo01} Leonard, D. C., Filippenko, A. V., Ardila, D. R., \& Brotherton, M. S. 2001, \apj, 553, 861 \bibitem[MacFadyen \& Woosley(1999)]{mac99} MacFadyen, A. I., \& Woosley, S. E. 1999, \apj, 524, 262 \bibitem[Maeda et al.(2002)]{mae02} Maeda, K., Nakamura, T., Nomoto, K., Mazzali, P. A., Patat, F., \& Hachisu, I. 2002, \apj, 565, 405 \bibitem[Mazzali, Iwamoto, \& Nomoto(2000)]{maz00} Mazzali, P. A., Iwamoto, K., \& Nomoto, K. 2000, \apj, 545, 407 \bibitem[Mazzali et al.(2001)]{maz01} Mazzali, P. A., Nomoto, K., Patat, F., \& Maeda, K. 2001, \apj, 559, 1047 \bibitem[Mazzali et al.(2002)]{maz02} Mazzali, P. A., et al. 2002, ApJ, 572, L61 %\bibitem[McLean \& Clarke(1979)]{mcl79} McLean, I. S., \& % Clarke, D. 1979, \mnras, 186, 245 \bibitem[Nagataki et al.(1997)]{nag97} Nagataki, S., Hashimoto, M., Sato, K., \& Yamada, S. 1997, \apj, 486, 1026 \bibitem[Nakamura et al.(2001)]{nak01} Nakamura, T., Mazzali, P. A., Nomoto, K., \& Iwamoto, K. 2001, \apj, 550, 991 \bibitem[Nakano et al.(2002)]{nak02} Nakano, S., \& Hirose, Y. % Kushida, R., Kushida, Y., \& Li, W. 2002, \iaucirc, No. 7810 \bibitem[Nomoto, Iwamoto, \& Suzuki(1995)]{nom95} Nomoto, K., Iwamoto, K., \& Suzuki, T. 1995, \physrep, 256, 173 \bibitem[Nomoto et al.(2001)]{nom01} Nomoto, K., et al. 2001, in Supernovae and Gamma-Ray Bursts, ed. M. Livio, N. Panagia and K. Sahu (Cambridge: Cambridge Univ. Press), 144 \bibitem[Oke(1990)]{oke90} Oke, J. B. 1990, \aj, 99, 1621 \bibitem[Patat et al.(2001)]{pat01} Patat, F., et al. 2001, \apj, 555,900 %\bibitem[Schlegel, Finkbeiner, \& Davis(1998)]{sch98} Schlegel, D. J., % Finkbeiner, D. P., \& Davis, M. 1998, \apj, 500, 525 %\bibitem[Schmidt-Kaler(1958)]{sch58} Schmidt-Kaler, Th. 1958, % Zs. for Astrophys. 46, 145 %\bibitem[Schmidt, Elston, \& Lupie(1992)]{sch92} Schmidt, G. D., % Elston R., \& Lupie O. L. 1992, \aj, 104, 1563 %\bibitem[Serkowski(1973)]{ser73} Serkowski, K. 1973, in IAU Symp., No. % 52, Interstellar Dust and Related Topics, eds. J. M. Greenberg \& % H. C. van de Hulst (Dordrecht: Reidel), 145 %\bibitem[Serkowski(1974)]{ser74} Serkowski, K. 1974, % in Methods of Experimental Physics, Vol. 12, Part A. Astrophysics: % Optical and Infrared, ed. N. Carleton (New York: Academic Press), 361 \bibitem[Serkowski, Mathewson, \& Ford(1975)]{ser75} Serkowski, K., Mathewson, D. S., \& Ford, V. L. 1975, \apj, 196, 261 %\bibitem[Stathakis et al.(2000)]{sta00} Stathakis, R. A., et al. %2000, \mnras, 314, 807 \bibitem[Smartt \& Meikle(2002)]{sma02} Smartt, S., \& Meikle, P. 2002, \iaucirc, No. 7822 %\bibitem[Smartt et al.(2002b)]{sma02b} Smartt, S. J., Vreeswijk, P. M., %Ramirez-Ruiz, E., Gilmore, G. F., Meikle, W. P. S., Ferguson, A. M. N., %\& Knapen, J. H. 2002, \apj, 572, L147 \bibitem[Takada-Hidai, Aoki, \& Zhao(2002)]{tak02} Takada-Hidai, M, Aoki, W., \& Zhao, G. 2002, in preparation %\bibitem[Trammell, Hines, \& Wheeler(1993)]{tra93} Trammel, S. R., % Hines, D. C., \& Wheeler, J. C. 1993, \apj, 414, L21 \bibitem[Tinbergen(1996)]{tin96} Tinbergen, J. 1996, Astronomical Polarimetry (New York: Cambridge Univ. Press) %\bibitem[de Vaucouleurs et al.(1991)]{vau91} de Vaucouleurs, G., % de Vaucouleurs, A., Corwin Jr., H.G., Buta, R. J., Paturel, G., % \& Fouque, P. 1991, in Third Reference Catalogue of Bright Galaxies, % Sky and Telescope, 82, 621 \bibitem[Wang et al.(1996)]{wan96} Wang, L., Wheeler, J. C., Li, Z., \& Clocchiatti, A. 1996, \apj, 467, 435 \bibitem[Wang et al.(2001)]{wan01} Wang, L., Howell, D. A., H\"oflich, P., \& Wheeler, J. C. 2001, \apj, 550, 1030 %\bibitem[Wang et al.(2002)]{wan02} Wang, L., Baade, D., Fransson, C., %H\"oflich, P., Lundqvist, P., \& Wheeler, J. C. 2002, \iaucirc, No. 7820 %\bibitem[Wang et al.(2002)]{wan02} Wang, L., Baade, D., %H\"oflich, P., Wheeler, J. C., Fransson, C., \& Lundqvist, P. 2002, %submitted to ApJ %\bibitem[Wilking, Lebofsky, \& Rieke(1982)]{wil82} Wilking, B. A., % Lebofsky, M. J., \& Rieke, G. H. 1982, \aj, 87, 695 %\bibitem[Whittet et al.(1992)]{whi92} Whittet, D. C. B., % Martin, P. G., Hough, J. H., Rouse, M. F., Bailey, J. A., \& Axon, % D. J. 1992, \apj, 386, 562 %\bibitem[Woosley \& Eastman(1997)]{woo97} Woosley, S. E., \& % Eastman, R. G. 1997, in Thermonuclear Supernovae, eds, % P. Ruiz-Lapuente, R. Canal, \& J. Isern (Dordrecht: Kluwer), 821 \bibitem[Yoshida et al.(2000)]{yos00} Yoshida, M., et al. 2000, Proc. SPIE, 4009, 240 \end{thebibliography} \clearpage {\scriptsize \begin{deluxetable}{lccrr} \tabletypesize{\scriptsize} %\rotate \tablenum{1} \tablewidth{0pt} \tablecaption{Log of Observations for SN~2002ap\label{tbl-1}} \tablehead{ \colhead{Date (UT)} & \colhead{Grism\tablenotemark{a}} & \colhead{$\lambda\lambda$ (\AA )\tablenotemark{b}} & \colhead{$\lambda/\Delta\lambda$} & \colhead{$\Delta t$ (s)} } \startdata 2002 Feb 9.2 & 300/5500 & 4750--8300 & 1200 & $1200$ \\ 2002 Feb 9.3 & 300/5500 & 3850--6050 & 1200 & $2800$ \\ 2002 Feb 10.3 & 300/5500 & 4750--8300 & 1200 & $1200$ \\ 2002 Feb 11.3 & 300/5500 & 4750--8300 & 1200 & $1600$ \\ 2002 Feb 11.3 & 300/5500 & 3850--6050 & 1200 & $1600$ \\ 2002 Feb 12.3\tablenotemark{c} & 300/5500 & 4750--8300 & 1200 & $1200$ \\ 2002 Mar 8.2 & 300/7500 & 4850--9050 & 650 & $960$ \\ 2002 Mar 10.2 & 300/7500 & 4850--9050 & 650 & $1080$ \\ 2002 Jun 9.6 & 300/5500 & 4750--8300 & 1200 & $1200$ \\ \tablenotetext{a}{Grooves per millimeter/central wavelength in angstroms} \tablenotetext{b}{Effective wavelength range of the observation, which depends on the combination of the grism and the order-cut filter used.} \tablenotetext{c}{On February 12 we could not carry out the whole sequence of polarimetry because of unstable weather, and so obtained only flux data.} %\tablerefs{} %\tablecomments{} \enddata \end{deluxetable} } \clearpage \figcaption[Kawabata.fig1.eps]{Flux and polarization spectra of SN 2002ap. Heliocentric redshift, interstellar extinction and polarization have not been corrected for. From top to bottom, we plot (a) total flux, (b, c) polarization level $p$ and position angle $\theta$ on each observation night. %, and (d,e) $p$ and $\theta$ averaged for each month. The polarimetric data are binned to a constant photon noise of $0.05$ \% % (b,c) or $0.04$ \% (d,e) which is shown by the error bars of polarization points. %In (a), flux data in March multiplied by three are also plotted %for comparison. %For the element indicated at each broad absorption feature, %we referred to by Mazzali et al. (2002). %Considerable polarization features, which also changed with %time significantly, were found near broad absorption features %around 7500 \AA and 8200 \AA . The estimated ISP component is shown by a dashed curve in (b,c). \label{fig1}} \figcaption[Kawabata.fig2.eps]{{\it QU}-diagram of the monthly-averaged intrinsic polarization for February and March epochs. The estimated ISP has been removed and the data are binned to a constant photon noise of $0.04$ \$. It can be seen that the polarization in February has, at least, two preferred axes: PA$\sim 120\arcdeg$ (associated with the \ion{O}{1}/\ion{Ca}{2} line trough) and PA$\sim 80\arcdeg$ (associated with the continuum). The polarization in March has a clear position angle PA$\sim 150\arcdeg$ associated with the \ion{Ca}{2} line trough and the significantly polarized continuum. These position angles are indicated by thick arrows. Note that the position angle on the sky is half the angular location on a {\it QU} diagram. \label{fig2}} \figcaption[Kawabata.fig3.eps]{Polarization spectra corrected for heliocentric redshift and interstellar extinction. The estimated ISP component has also been removed. From top to bottom, we plot (a) total flux in erg s$^{-1}$ cm$^{-2}$ \AA$^{-1}$, (b) polarization level $p$, (c) position angle $\theta$, (d) polarized flux, and (e,f) $p$ and $\theta$ of the residual polarization after the jet model continuum polarization has been subtracted from the February data. The February flux is the mean of Feb 9 and 11, and the March flux is the mean of Mar 8 and 10. We adopt a heliocentric redshift $+631$ km s$^{-1}$ for M74 (Smartt \& Meikle 2002), a color excess of $E_{B-V}=0.07$ in our Galaxy and $0.02$ in M74 (Takada-Hidai et al. 2002) and the normal interstellar extinction curve (Cardelli, Clayton, \& Mathis 1989). Deep absorption bands due to the terrestrial atmosphere and the interstellar medium have been removed by interpolation using nearby continuum levels. The polarimetric data are binned in the same manner as in Fig. 2. The solid curve in (d) is the February flux multiplied by $0.0023$ and redshifted by $+0.23c$ (see \S 4.3). \label{fig3}} \plotone{f1.eps} \plotone{f2.eps} \plotone{f3.eps} %\clearpage \end{document}