1: \documentclass[12pt,preprint]{aastex}
2:
3: %
4: % define symbols for this paper
5: %
6: \def\Msun{M_\odot}
7: \def\Lstar{L_{b_J}^\ast}
8:
9:
10: \slugcomment{Submitted to The Astrophysical Journal Letters}
11: \lefthead{Agustsson \& Brainerd}
12: \righthead{Satellite Galaxies are Radially Aligned}
13:
14:
15: \begin{document}
16:
17: \title{The Orientation of Satellite Galaxies: Evidence of Elongation
18: in the Direction of the Host}
19:
20: \author{Ingolfur Agustsson \& Tereasa G. Brainerd}
21: \affil{Boston University, Institute for Astrophysical Research,
22: 725 Commonwealth Ave., Boston, MA 02215}
23: \email{ingolfur@bu.edu, brainerd@bu.edu}
24:
25: \begin{abstract}
26: We use the fourth data release of the Sloan Digital Sky Survey to
27: investigate the orientations of 4289 satellite
28: galaxies with respect to their hosts.
29: The orientation of the satellites is inconsistent
30: with a random distribution at the 99.94\% confidence level, and
31: the satellites show a preference for elongation in the
32: direction of their hosts. Further, on
33: scales $< 50$~kpc the major axes of the host galaxies and their satellites
34: are preferentially aligned.
35: Phrased in the terminology of weak lensing, the
36: images of the satellites have a mean shear of
37: $\gamma_T = -0.045 \pm 0.010$, averaged over scales $10~{\rm kpc} \le
38: r_P \le 50~{\rm kpc}$.
39: In a galaxy--galaxy lensing study where
40: lenses and sources are separated solely on the basis of
41: apparent magnitude, we estimate that on
42: scales $\lesssim 250$~kpc satellite galaxies account for
43: between 10\% and 15\% of the objects that are identified as sources.
44: In such studies,
45: elongation
46: of the satellites will
47: cause a reduction
48: of the galaxy--galaxy lensing shear by of order 25\% to 40\%.
49: Hence, the elongation of satellite galaxies in the direction of their hosts is
50: a potentially important effect for precision studies
51: of galaxy--galaxy lensing, and argues
52: strongly in favor of the use of accurate photometric redshifts
53: in order to identify
54: lenses and sources in future studies.
55: \end{abstract}
56:
57: \keywords{galaxies: dwarf ---
58: galaxies: fundamental parameters ---
59: galaxies: halos --- gravitational lensing
60: }
61:
62: \section{Introduction}
63:
64: Galaxy--galaxy lensing, hereafter GG lensing,
65: has become a premiere tool for constraining
66: the nature of the dark matter halos of galaxies (e.g., Brainerd
67: 2004a, Brainerd \& Blandford 2002, and references therein).
68: Recent investigations of GG lensing have
69: moved beyond the most basic constraints on the nature of the ``average''
70: dark matter halo, demonstrating that there are real physical
71: differences between the halos of early--type and late--type galaxies,
72: and that the halos are non-spherical (e.g., Hoekstra et al.\ 2004;
73: Kleinheinrich et al.\ 2004; Sheldon et al.\ 2004; Mandelbaum et al.\
74: 2005a; Seljak et al.\ 2005). From its
75: earliest days, however, GG lensing has been haunted by
76: the possibility that a number of
77: genuine satellite galaxies, in orbit about
78: the lenses, could be mistakenly
79: identified as
80: sources.
81: If such satellites are randomly oriented with
82: respect to the lenses, their presence introduces noise
83: in the measurement of the lensing signal. If
84: the satellites are non--randomly oriented with respect to the lenses,
85: this alters the observed
86: lensing signal compared to the true signal that would be measured in
87: the absence of such false sources.
88: Non--random orientations
89: could be caused by tidal distortions at
90: small distances from the host or, at larger distances,
91: by the tendency of galaxies to form in preferential alignment within
92: filaments (e.g., Catelan et al.\ 2000;
93: Croft \& Metzler 2000; Heavens et al.\ 2000;
94: Lee \& Pen 2000, 2001; Crittenden et al.\ 2001;
95: Brown et al.\ 2002; Jing 2002; Heymans et al.\ 2004; Mandelbaum et al.\
96: 2005b)
97:
98: Phillips (1985), Tyson (1985),
99: and Brainerd et al.\ (1996) used the clustering strength
100: of faint galaxies to place limits on contamination of the
101: GG lensing signal caused by
102: satellites and concluded that the contamination was sufficiently small
103: to be ignored. Bernstein \& Norberg (2002), hereafter BN, found that
104: on scales $< 500$~kpc, the mean
105: tangential ellipticity of satellites in
106: the Two Degree Field Galaxy Redshift Survey (2dFGRS; Colless et al.\ 2001,
107: 2003)
108: was consistent with zero,
109: and concluded that the contamination of the GG lensing
110: signal was $< 20$\%. Hirata et al.\ (2004) used photometric redshifts
111: in an
112: analysis of Sloan Digital Sky Survey data
113: (SDSS\footnote{\url{http://www.sdss.org}};
114: Fukugita et al.\ 1996; Hogg et al.\ 2001;
115: Smith et al.\ 2002; Strauss et al.\ 2002; York et al.\ 2002)
116: and concluded that
117: on scales of 30$h^{-1}$~kpc to 446$h^{-1}$~kpc, the mean intrinsic
118: shear of satellite galaxies was consistent with zero and that the
119: contamination of the GG lensing signal due to satellites
120: was $\lesssim 15$\%.
121:
122: Here we use the
123: fourth data release (DR4) of the SDSS (Adelman--McCarthy et al.\ 2005) to
124: revisit the question of whether satellite galaxies have a preferred
125: orientation with respect to their hosts.
126: %Details of the selection of hosts
127: %and satellites are presented in \S2. The orientation of the satellites
128: %and null tests are presented in \S3, and a
129: %discussion of our results is
130: %presented in \S4.
131: Throughout we adopt $H_0 = 70$~km~sec$^{-1}$~kpc$^{-1}$,
132: $\Omega_{m0} = 0.3$, and $\Omega_\Lambda = 0.7$.
133:
134: \section{Host and Satellite Galaxies in the SDSS DR4}
135:
136: Hosts and satellites are selected by requiring: [1]
137: the hosts are relatively isolated and [2]
138: host--satellite pairs are nearby to one another
139: in terms of projected separation, $r_P$, and radial
140: velocity difference, $|dv|$. Specifically, hosts must be 2.5 times
141: more luminous than any other galaxy that falls within $r_P \le 700$~kpc
142: and $|dv| \le 1000$~km~sec$^{-1}$. Satellites must be
143: at least 4 times less luminous than their host, and must be located within
144: $r_P \le 250$~kpc and $|dv| \le 500$~km~sec$^{-1}$.
145: %Further, in order
146: %to reject a small number of hosts with a large number of satellites
147: %(and which may, therefore, be associated with clusters of galaxies)
148: %we require the sum total of the luminosities of the satellites of a
149: %given host to be less than the luminosity of the host.
150: We use
151: only galaxies with redshift confidence parameter
152: ${\rm zconf} > 0.9$ and, to avoid systematics due to overlapping
153: isophotes, we use only those satellites that are located at radii
154: larger than three times the scale radius, $r_s$, of their host.
155: Lastly, we visually inspect the images of all candidate satellites
156: that are found within $r_p = 50$~kpc, and we reject those
157: which have been misidentified as ``galaxies'' in the
158: database. Only 6.5\% of the candidate satellites within
159: 50~kpc are rejected this way,
160: and these consist of either a star or a bright blue knot that has been
161: misidentified as a small, faint galaxy.
162: Implementation of all our
163: criteria yields a final sample of
164: 3180 hosts and 4289 satellites. The median redshift of the hosts is
165: $z_{\rm med} = 0.058$. Distributions of observed $r$--band apparent
166: magnitudes, the difference in observed $r$ magnitude, and
167: K--corrected colors are shown
168: in Fig.~1. K--corrections were obtained from version 3.2 of
169: Michael Blanton's IDL
170: code\footnote{\url{http://cosmo.nyu.edu/blanton/kcorrect/}} (e.g., Blanton et
171: al.\ 2003).
172: Since the satellites are large and
173: well--resolved, their images are not greatly affected by local anisotropies
174: in the PSF and, therefore, we make no corrections to their image shapes
175: below.
176:
177: \begin{figure}
178: \centerline{
179: \scalebox{0.85}{%
180: \includegraphics{f1.eps}%
181: }
182: }
183: %\vskip -4.5cm
184: \caption{Top panels:
185: Distribution of observed $r$ magnitude for the hosts and satellites
186: (left) and distribution of difference in observed
187: $r$ magnitude (right). Bottom panels:
188: Distribution of K--corrected colors for the hosts and satellites.
189: }
190: \label{fig1}
191: \end{figure}
192:
193:
194: \section{Orientation of the Satellites}
195:
196: We use the $r$-band position angles of the satellites to compute $\theta$,
197: the angle between the major axes of the satellites and
198: the direction vectors on the sky that connect the centroids of the
199: hosts and their satellites. We restrict $\theta$ to
200: the range $\left[ 0^\circ , 90^\circ \right]$, where $\theta = 0^\circ$
201: corresponds to a radial orientation of the satellite in the direction of
202: its host and $\theta = 90^\circ$ corresponds to a tangential
203: orientation.
204: Shown in the top panels of Fig.~2 are the differential probability
205: distribution, $P(\theta)$, and continuous cumulative probability
206: distribution, $P(\theta \le \theta_{\rm max})$, for the orientations
207: of the satellites. The data in both panels are inconsistent
208: with random distributions.
209: A $\chi^2$ test performed on $P(\theta)$
210: rejects the random distribution at the 99.93\% confidence level, while
211: a Kolmogorov--Smirnov (KS) test performed on $P(\theta \le \theta_{\rm max})$
212: rejects a random distribution at the
213: 99.94\% confidence level. From the top panels of Fig.~2, then, there
214: is a preference for the satellites to be elongated in the direction
215: of their hosts.
216:
217: \begin{figure}
218: \centerline{
219: \scalebox{0.85}{%
220: \includegraphics{f2.eps}%
221: }
222: }
223: %\vskip -3.5cm
224: \vskip 0.1cm
225: \caption{
226: Left panels: Differential probability distribution
227: for the orientation of satellite galaxies (top) and
228: stars with $16.5 \le r \le 18.5$ (bottom),
229: relative to the locations of the host galaxies.
230: Dotted lines show $P(\theta)$ for a random distribution.
231: Formal rejection confidence levels from the $\chi^2$ test are
232: shown in the panels.
233: Right panels: Cumulative probability distribution for the
234: orientation of the satellite galaxies (top) and stars
235: (bottom). Dotted lines show
236: $P(\theta \le \theta_{\rm max})$ for a random distribution.
237: Formal
238: rejection confidence levels from the KS test are shown in the panels.
239: }
240: \label{fig2}
241: \end{figure}
242:
243: Shown in the bottom panels of Fig.~2 is a null test in which
244: unsaturated stars with magnitudes similar to those of the
245: satellite galaxies
246: are used
247: to compute $P(\theta)$ and
248: $P(\theta \le \theta_{\rm max})$.
249: The stellar sample consists of 92,489 stars in the SDSS
250: photometric database that have $16.5 \le r \le 18.5$ and are
251: found within projected radii $3r_s \le r_p \le 250$~kpc of each host galaxy.
252: Here $r_s$ is again the scale radius of the host galaxy.
253: Both $P(\theta)$ and $P(\theta \le \theta_{\rm max})$ for the stars
254: are consistent
255: with random distributions and we are, therefore, confident that the
256: non--random orientation of the satellites shown in the
257: top panels of Fig.~2 is unlikely to be caused by systematics in the
258: imaging (e.g., drift--scanning, overlapping
259: image isophotes, and/or classical aberrations).
260: We specifically do not perform a null test that
261: is sometimes performed in GG lensing: the substitution of
262: the images of the hosts for those of the satellites. That is,
263: if the ``lens'' galaxies are foreground
264: objects and the ``source'' galaxies are background
265: objects, the images of the sources should
266: be tangentially aligned with respect to the lenses but the
267: images of the lenses should be randomly oriented with respect to
268: the sources. This
269: pre--supposes that the centroids of the sources are distributed uniformly
270: around the lenses; however, this is not the case for our satellites.
271: Brainerd (2005) showed that satellites
272: are found preferentially close to the major axes of their hosts. That is,
273: the major axes of the hosts point preferentially toward the
274: locations of their satellites (i.e., the hosts are radially
275: aligned along the direction vectors that connect the centroids of the
276: hosts with their satellites). Indeed, the satellites in
277: our present sample have a mean location angle
278: of $ \left< \phi \right> = 42.4^\circ \pm 0.4^\circ$
279: relative to the major axes of their hosts, consistent with the
280: results of Brainerd (2005).
281:
282: The combination of the results of Brainerd (2005) and our results in
283: Fig.~2 lead to the conclusion that host
284: galaxies and their satellites are intrinsically aligned on small
285: scales. We quantify
286: this by computing a two--point correlation function of the shapes of the
287: hosts and their satellites:
288: $C_{\gamma \gamma} (r_P) \equiv \left< \vec{\gamma}_{h} \cdot
289: \vec{\gamma}_{s}^\ast \right>_{r_P}
290: $. This is analogous to a function used in weak lensing to
291: measure
292: correlated distortions in the images of lensed galaxies
293: as a function of the separation of the images on the sky (e.g., Blandford
294: et al.\ 1991). Here $\vec{\gamma}_h$ and $\vec{\gamma}_s$ are
295: shape parameters for
296: the hosts and satellites, respectively, where $\vec{\gamma} \equiv \epsilon
297: e^{2i \varphi}$, $\varphi$ is the position angle of a galaxy,
298: $a$ and $b$ are its major and minor axes, and $\epsilon \equiv
299: (a-b)/(a+b)$.
300: The mean, denoted by angle brackets,
301: is computed over all pairs of hosts and satellites separated by
302: projected radii $r_P \pm 0.5 ~dr_P$.
303: Host and satellite images that are uncorrelated
304: yield $C_{\gamma \gamma} (r_P) = 0$, while host and satellite images that
305: are aligned yield positive values of
306: $C_{\gamma \gamma} (r_P)$. Solid circles in the top panel of Fig.~3 show
307: $C_{\gamma \gamma} (r_P)$
308: for our hosts and satellites, where
309: it is clear that on scales $< 50$~kpc the images of the hosts and satellites
310: are aligned with each other. On scales $> 50$~kpc, the images of the hosts
311: and satellites show no apparent correlation, consistent with the lack of
312: large scale
313: intrinsic alignment of SDSS galaxies reported by Mandelbaum et al.\
314: (2005b). For comparison, open circles in the top panel of Fig.~2 show
315: $C_{\gamma\gamma}(r_P)$ computed using the shape parameters of the hosts
316: and the stars with $16.5 \le r \le 18.5$ that are nearby to the hosts.
317: Unlike the images of the hosts and satellites,
318: the images of the hosts and nearby stars are uncorrelated, and we conclude
319: that the apparent correlation of host and satellite images on scales
320: $< 50$~kpc is unlikely to be caused by imaging systematics.
321:
322: \begin{figure}
323: \centerline{
324: \scalebox{0.85}{%
325: \includegraphics{f3.eps}%
326: }
327: }
328: \vskip 0.1cm
329: \caption{Top panel: Two--point correlation function of host and satellite
330: galaxy image shapes (solid circles), as well as host and
331: star image shapes (open circles). Positive values of $C_{\gamma \gamma}(r_p)$
332: indicate alignment of host and satellite images. Bottom panel:
333: Mean tangential shear for the SDSS satellite
334: galaxies (solid squares) and stars (open squares). Negative
335: tangential shear indicates a systematic distortion in the radial direction.
336: Dotted line indicates $\gamma_T(r_P) = 0$. Dashed, solid, and
337: dot--dash lines show the theoretical tangential shear for isothermal sphere
338: lenses with velocity dispersion $\sigma_v$.
339: }
340: \label{fig3}
341: \end{figure}
342:
343: Since our satellites
344: are not randomly oriented with respect to their hosts, and
345: because a non--random orientation has potentially important implications
346: for GG lensing, we use the satellites to compute the
347: standard weak lensing quantity known as the mean
348: tangential shear, $\gamma_T$.
349: The tangential ellipticity for each satellite, $j$, is
350: computed as
351: $\gamma_j = \epsilon_j \cos(2 \alpha_j)$ where again
352: $\epsilon_j \equiv (a_j - b_j)/(a_j + b_j)$ and $\alpha_j$ is the
353: angle between the major axis of the satellite and the tangent to the
354: direction vector that connects the centroids of the host and satellite.
355: A simple,
356: unweighted mean of the individual values of $\gamma_j$ is used to
357: compute $\gamma_T$. A positive value of $\gamma_T$ indicates tangential
358: orientation of the satellite images while a negative value of
359: $\gamma_T$ indicates radial orientation.
360: Solid squares in the bottom panel of Fig.~3 show
361: $\gamma_T$ for the SDSS satellites, computed
362: as a function of $r_P$. Open squares show $\gamma_T$ for
363: stars with $16.5 \le r \le 18.5$ that are nearby to the hosts.
364: As expected from
365: Fig.~2, the satellites have a negative value of $\gamma_T$
366: (most notably on scales $< 100$~kpc) and $\gamma_T$ for the stars
367: is consistent with zero.
368: In particular, for
369: satellites with
370: $10~{\rm kpc} \le r_P \le 50~{\rm kpc}$, the mean tangential shear is
371: $\gamma_T = -0.045 \pm 0.010$.
372: Although the sign of $\gamma_T$ is the opposite of what one expects
373: in gravitational lensing, the magnitude of $\gamma_T$ is
374: comparable to what one would expect from galaxy--mass lenses with similar
375: impact parameters. Shown in Fig.~3 for comparison are the
376: expected functions $\gamma_T(r_P)$ for four simple scenarios in which
377: lens galaxies are modeled as
378: singular isothermal spheres with velocity dispersion
379: $\sigma_v$. The lenses are located at redshift $z_L$ and the sources
380: are located at redshift $z_S$. The values of $z_L$ and $z_S$ are similar to
381: the actual median redshifts in current GG lensing studies, and
382: the values of $\sigma_v$ roughly span the range of values
383: that have been inferred for lens galaxies from GG lensing.
384:
385: If a significant number of unidentified
386: satellite galaxies are present amongst the ``source'' galaxies in a
387: GG lensing data set, a substantial reduction of the
388: true shear could result. Contamination
389: by satellites is most likely to occur when
390: lenses and sources have been identified solely on the basis of apparent
391: magnitude (i.e., ``bright'' galaxies are identified as lenses and
392: ``faint'' galaxies are identified as sources).
393: In these cases, the typical
394: difference in apparent magnitude between the ``sources'' and the ``lenses''
395: is of order 2 to 2.5 magnitudes, which is similar to the magnitude
396: difference between the hosts and satellites in our sample (see also
397: BN). In addition, the vast majority of the SDSS
398: satellites
399: have apparent magnitudes that are fainter than the vast majority of the
400: SDSS hosts, so it is reasonable to estimate the degree to which bright
401: satellites might contaminate the GG lensing signal in a data set
402: that has a similar range of magnitude differences amongst its ``bright''
403: and ``faint'' galaxies. In the magnitude range $16.5 \le r \le 18.5$, then,
404: we find that genuine satellite galaxies account for $\sim 10$\% to
405: $\sim 15$\% of the {\it total} number of SDSS galaxies that surround the
406: host galaxies over scales $10~{\rm kpc} \le r_P \le r_{\rm max}$, where
407: $25~{\rm kpc} \le r_{\rm max} \le 250~{\rm kpc}$. If we use a conservative
408: estimate of 10\% for the satellite contamination of the GG lensing signal,
409: then, over scales $r_P \lesssim 250$~kpc our
410: observed elongation of
411: satellites in the direction of
412: their hosts reduces the true tangential shear for the
413: model lenses in Fig.~3 by an amount between 25\%$\pm$5\% (high--redshift
414: lens with $\sigma_v = 155$~km~sec$^{-1}$) and 40\%$\pm$10\% (low--redshift
415: lens with $\sigma_v = 135$~km~sec$^{-1}$).
416:
417:
418: \section{Discussion}
419:
420: %We have shown that large, bright satellites of isolated host galaxies
421: %in the SDSS DR4 have a tendency to be oriented radially with respect
422: %to the location of their host galaxy. Phrased in terms of the standard
423: %shear measurement of gravitational lensing, the images of the satellites
424: %have a tangential shear of $\gamma_T = -0.030 \pm 0.007$, where the negative
425: %sign indicates radial alignment. On scales $\lesssim 250$~kpc,
426: %a conservative estimate of the degree
427: %to which the GG lensing shear may be reduced by the presence of
428: %unidentified satellite galaxies ranges from $\sim 25$\% to $\sim 40$\%,
429: %depending upon the velocity dispersion of the lenses and the depth
430: %of the sample.
431:
432: The cause of the elongation of the satellites in the direction of their hosts
433: is likely to be twofold. On small scales, distortions caused by the
434: gravitational interaction of the satellites with their hosts may
435: occur, leading to tidal streams. We have examined the images
436: of $\sim 300$ of the brightest hosts and a handful ($< 10$) do
437: seem to show the presence of faint tidal streams that connect
438: the hosts and satellites. On large scales, the orientation of
439: the satellites most
440: likely reflects the tendency for galaxies to form in rough alignment
441: along filaments.
442:
443: The only previous study to which our work is directly comparable is
444: that of BN, who selected hosts and satellites in a manner similar to ours.
445: BN, however,
446: concluded that the tangential ellipticity of the
447: 2dFGRS satellites was consistent with zero. There are a number
448: of factors that contribute to this discrepancy.
449: First, our work is based on a larger number of
450: satellites (4289 vs.\ 1819). The difference in the number of satellites
451: is caused largely by the fact that BN's satellites are much fainter than
452: the majority of the satellites used here.
453: Second, BN computed the tangential ellipticity over
454: a large aperture of radius 500~kpc, and we find that the preferential
455: alignment of the satellites is restricted to $r_P < 100$~kpc.
456: Third, the fraction of ``interlopers'' (i.e., galaxies
457: identified as satellites but which are not dynamically associated with
458: the host) is likely to be larger in BN than here. This
459: is due to a combination of the facts that the radial velocity errors are
460: larger in the 2dFGRS than they are in the SDSS ($\sim 85$~km~sec$^{-1}$
461: vs.\ $\sim 25$~km~sec$^{-1}$) and that the interloper fraction increases
462: substantially with projected radius (e.g., Prada et al.\ 2003; Brainerd 2004b).
463: Finally, the resolution of the satellite images used by BN is likely to
464: have been lower than the resolution of the SDSS satellite images used
465: here. This is in part due to the fact that BN restricted their satellites
466: to be at least 7.6 times fainter than their host (so, on average, their
467: satellites would have subtended a smaller angle on the sky than our
468: SDSS satellites). In addition, the imaging of BN's satellites was based on
469: scans of glass plates with a pixel scale of $\sim 0.7''$, compared to
470: a CCD pixel scale of $\sim 0.4''$ in the SDSS.
471:
472: To compare to BN, we have converted SDSS magnitudes to the
473: $b_J$--band using the photometric transformation of Norberg et al.\ (2002),
474: $b_J = g + 0.155 + 0.152(g-r)$, and have identified isolated hosts
475: and their satellites in the DR4 using the same criteria as BN.
476: Specifically, hosts
477: must be at least 7.6 times brighter than any other galaxy located within
478: $r_P \le 500$~kpc and $|dv| \le 1000$~km~sec$^{-1}$, and satellites consist
479: of all galaxies found within $r_P \le 500$~kpc and $|dv| \le 500$~km~sec$^{-1}$
480: of a host. Implementing these criteria yields 1074~hosts
481: and 1467 satellites, from which we compute $P(\theta)$ and
482: $P(\theta \le \theta_{\rm max})$. When all satellites with
483: $r_P \le 500$~kpc are used, both $P(\theta)$ and
484: $P(\theta \le \theta_{\rm max})$ are consistent with random distributions
485: ($\chi^2$ rejection confidence level of 61.1\% and KS rejection confidence
486: level of 49.3\%). When only those satellites with
487: $r_P \le 250$~kpc (i.e., the maximum $r_P$ for our
488: satellites) are used, the total number of satellites is reduced by $\sim 50$\%
489: and both $P(\theta)$ and $P(\theta \le \theta_{\rm max})$ remain
490: consistent with random distributions
491: ($\chi^2$ rejection confidence level of 48.2\% and KS rejection confidence
492: level of 83.8\%). Thus, when selected in a comparable
493: manner,
494: the SDSS galaxies yield
495: results that are consistent with BN's.
496: In addition,
497: we have applied our selection criteria from \S2 to the final data release
498: of the 2dFGRS, resulting in a sample of 2054 hosts and 2663 satellites. A
499: preliminary analysis of the images of the satellites shows a
500: weak tendency for them to be elongated in the direction
501: of their
502: hosts; a uniform distribution
503: for the 2dFRGS satellites is rejected
504: at the 95.4\% confidence level by the $\chi^2$ test and at
505: the 91.8\% confidence level by the KS test. Averaged over scales
506: $10~{\rm kpc} \le r_P \le 50~{\rm kpc}$,
507: the mean tangential shear of the
508: 2dFGRS satellites is $\gamma_T = -0.019 \pm 0.010$, in modest agreement
509: with our measurement for the SDSS DR4 satellites. We will further
510: explore the distortion of satellite images in both the SDSS and 2dFGRS,
511: including possible dependencies on host luminosity and spectral type,
512: in a future paper (Brainerd et al.\ 2006).
513:
514: In conclusion, we find that
515: on scales $< 100$~kpc satellite galaxies are elongated in the
516: direction of their
517: hosts. In addition, on scales $< 50$~kpc the images of the hosts
518: and their satellites are aligned with each other.
519: A decade ago the first
520: tentative detections of GG lensing yielded only noisy constraints
521: on the shear, and the presence of non--randomly oriented satellites amongst
522: the lensed sources
523: could be largely ignored. Our results here, however, suggest that in the future
524: great care must be taken
525: to reject satellite galaxies (via, e.g., accurate photometric
526: redshifts) in order for ``precision shear''
527: observations of GG lensing
528: to result in truly precision constraints on the nature of
529: dark galaxy halos.
530:
531: \section*{Acknowledgments}
532:
533: We are grateful to an anonymous referee for helpful suggestions
534: that improved the manuscript, and for support
535: under NSF contract AST-0406844.
536: Funding for the SDSS has been provided by the
537: A.\ P.\ Sloan Foundation, the Participating Institutions, NASA,
538: the NSF,
539: the US Dept.\ of Energy, the Japanese Monbukagakusho, and the Max
540: Planck Society.
541: The SDSS is managed by the Astrophysical Research Consortium for
542: the Participating Institutions
543: (Univ.\ of Chicago, Fermilab, the Institute for Advanced Study, the
544: Japan Participation Group, Johns Hopkins Univ., Los Alamos National
545: Laboratory, the Max-Planck-Institute for Astronomy, the
546: Max-Planck-Institute for Astrophysics, New Mexico State Univ.,
547: Univ.\ of Pittsburgh, Princeton Univ., the US Naval
548: Observatory, and Univ.\ of Washington).
549:
550:
551: \begin{thebibliography}
552:
553: \bibitem{} Adelman--McCarthy, J.\ et al.\ 2005, ApJS submitted
554: (astro--ph/0507711)
555:
556: \bibitem{} Bernstein, G. M. \& Norberg, P. 2002, AJ, 124, 733 (BN)
557:
558: \bibitem{} Blandford, R. D., Saust A. B.\ et al 1991, MNRAS, 251, 600
559:
560: \bibitem{} Blanton, M. R., Lin H.\ et al.\ 2003, AJ, 125, 2276
561:
562: \bibitem{} Brainerd, T. G., Blandford, R. D. \& Smail, I. 1996, ApJ, 466, 623
563:
564: \bibitem{} Brainerd, T. G. \& Blandford, R. D. 2002, in ``Gravitational
565: Lensing: An Astrophysical Tool'', Lecture Notes in Physics vol.\ 608,
566: eds.\ F. Courbin \& D. Minniti, 96
567:
568: \bibitem{} Brainerd, T. G. 2004a, in ``The New Cosmology'', AIP Conf.\
569: Proc.\ vol.\ 743, eds.\ R. E. Allen, D. V. Nanopoulos, \& C. N. Pope, 129
570:
571: \bibitem{} Brainerd, T. G. 2004b, ApJ submitted (astro--ph/0409374)
572:
573: \bibitem{} Brainerd, T. G. 2005, ApJ, 628, 101
574:
575: \bibitem{} Brainerd, T. G., Agustsson, I. \& McConnell, N. 2006, in prep.
576:
577: \bibitem{} Brown, M. L., Taylor, A. N. et al.\ 2002, MNRAS, 333, 501
578:
579: \bibitem{} Catelan, P., Kamionkowski, M. et al.\ 2000,
580: MNRAS, 320, L7
581:
582: \bibitem{} Colless, M. et al.\ 2001, MNRAS, 328, 1039
583:
584: \bibitem{} Colless, M. et al.\ 2003, astro--ph/0306581
585:
586: \bibitem{} Crittenden, R. G., Natarajan, P. et al.\ ApJ, 559, 552
587:
588: \bibitem{} Croft, R. A. C. \& Metzler, C. A. 2000, ApJ, 545, 561
589:
590: \bibitem{} Fukugita, M., Ichikawa, T., et al.\
591: 1996, AJ, 111, 1748
592:
593: \bibitem{} Heavens, A., Refregier, A. \& Heymans, C. 2000, MNRAS, 319, 649
594:
595: \bibitem{} Heymans, C. Brown, M. Heavens, et al.\ 2004, MNRAS, 347, 895
596:
597: \bibitem{} Hirata, C. M., Mandelbaum, R. et al.\ 2004, MNRAS, 353, 529
598:
599: \bibitem{} Hoekstra, H., Yee, H. K. C. \& Gladders, M. D., 2004,
600: ApJ, 606, 67
601:
602: \bibitem{} Hogg, D. W., Schlegel, D. J.\ et al.\ 2001, AJ,
603: 122, 2129
604:
605: \bibitem{} Jing, Y. P. 2002, MNRAS, 335, L89
606:
607: \bibitem{} Kleinheinrich, M. et al.\ 2004, AA submitted
608: (astro--ph/0412615)
609:
610: \bibitem{} Lee, J. \& Pen, U. 2000, ApJ, 532, L5
611:
612: \bibitem{} Lee, J. \& Pen, U. 2001, ApJ 555, 106
613:
614: \bibitem{} Norberg, P., Cole, S.\ et al.\ 2002, MNRAS, 336, 907
615:
616: \bibitem{} Mandelbaum, R.\ et al.\ 2005a, MNRAS submitted
617: (astro--ph/0507108)
618:
619: \bibitem{} Mandelbaum, R.\ et al.\ 2005b, MNRAS submitted
620: (astro--ph/0509026)
621:
622: \bibitem{} Phillips, S. 1985, Nature, 314, 721
623:
624: \bibitem{} Prada, F., Vitvitska, M.\ et al.\ 2003, ApJ, 598, 260
625:
626: \bibitem{} Seljak, U., Makarov, A. et al.\ 2005, PRD, 71, 3511
627:
628: \bibitem{} Sheldon, E. S., Johnston, D. E. et al.\ 2004, AJ, 127,2544
629:
630: \bibitem{} Smith, J. A., Tucker, D. L.\ et al.\ 2002, AJ, 123, 2121
631:
632: \bibitem{} Strauss, M. A., Weinberg, D. H.\ et al.\ 2002, AJ, 124, 1810
633:
634: \bibitem{} Tyson, J. A. 1985, Nature, 316, 799
635:
636: \bibitem{} York, D. G., Adelman, J.\ et al.\ 2000, AJ, 120, 1579
637:
638: \end{thebibliography}
639:
640: \end{document}
641: