1: \documentclass[10pt,preprint]{aastex}
2: \usepackage{psfig}
3:
4: \def\kms{{\rm km\,s^{-1}}}
5: \def\eq#1{equation (\ref{#1})}
6: \def\Eq#1{Eq.~\ref{#1}}
7: \def\km{{\rm km}}
8: \def\au{{\rm AU}}
9: \def\kms{{\rm km~s^{-1}}}
10: \def\gaia{{\it Gaia}}
11: \def\deg{^\circ}
12: \def\msun{M_\odot}
13: \def\mearth{M_\oplus}
14: \def\mjup{M_{\rm jup}}
15: \def\muas{\mu{\rm as}}
16: \def\mx{M_X}
17: \def\dx{D_X}
18: \def\db{D_b}
19: \def\pix{\Pi_X}
20: \def\pis{\Pi_S}
21: \def\mux{\mu_X}
22: \def\pirel{\Pi_{rel}}
23: \def\be{\begin{equation}}
24: \def\ee{\end{equation}}
25: \def\thetae{\theta_{\rm E}}
26: \def\bth{{\bf \theta}}
27: \def\bu{{\bf u}}
28: \def\siga{\sigma_{ast}}
29: \def\sn{{\rm S/N}}
30: \def\sntot{(\sn)_{\rm tot}}
31: \def\ave#1{\langle#1\rangle}
32: \def\fsky{f_{\rm sky}}
33:
34: \begin{document}
35:
36: \title{Astrometric Microlensing Constraints on a Massive Body in the Outer
37: Solar System with \gaia}
38: \author {B.\ Scott Gaudi\altaffilmark{1} and Joshua S.\
39: Bloom\altaffilmark{1,2,3}}
40:
41: \affil{$^1$ Harvard-Smithsonian Center for Astrophysics, MC 20, 60 Garden
42: Street, Cambridge, MA 02138, USA}
43:
44: \affil{$^2$ Harvard Society of Fellows, 78 Mount Auburn Street, Cambridge, MA
45: 02138 USA}
46:
47: \affil{$^3$ Department of Astronomy, 601 Campbell Hall, University of
48: California at Berkeley, Berkeley, CA 94720}
49:
50: \begin{abstract} A body in Solar orbit beyond the Kuiper belt exhibits an
51: annual parallax that exceeds its apparent proper motion by up to many
52: orders of magnitude. Apparent motion of this body along the
53: parallactic ellipse will deflect the angular position of background
54: stars due to astrometric microlensing (``induced parallax''). By
55: synoptically sampling the astrometric position of background stars
56: over the entire sky, constraints on the existence (and basic
57: properties) of a massive nearby body may be inferred. With a simple
58: simulation, we estimate the signal-to-noise ratio for detecting such a
59: body -- as function of mass, heliocentric distance, and ecliptic
60: latitude -- using the anticipated sensitivity and temporal cadences
61: from \gaia\ (launch 2011). A Jupiter-mass ($\mjup$) object at 2000 AU
62: is detectable by \gaia\ over the whole sky above $5 \sigma$, with even
63: stronger constraints if it lies near the ecliptic plane. Hypotheses
64: for the mass ($\sim 3 \mjup$), distance ($\sim 20,000$ AU) and
65: location of the proposed perturber (``Planet X'') which gives rise to
66: long-period comets may be testable.
67: \end{abstract}
68:
69: \keywords{gravitational
70: lensing --- methods: data analysis --- astrometry --- Oort Cloud}
71:
72: \section{Introduction\label{sec:intro}}
73:
74: Observations of long-period comets in the inner Solar System suggest not only a
75: substantial population of comets at 50,000 to 100,000 AU (the Oort Cloud;
76: \citealt{oort50}), but a mechanism for effectively perturbing the orbits of
77: these comets. Such a perturber must be massive enough to hold considerable
78: gravitational influence on the Oort cloud. Galactic tidal perturbations could
79: be the cause of a steady steam of cometary infall \citep{byl83} while close
80: encounters with passing stars would cause a more punctuated cascade
81: \citep{hills81}. Punctuated (and perhaps periodic; \citealt{hae87}) cometary
82: showers into the inner Solar system could also be caused by a perturber that
83: is bound to the Sun. Specific predictions of the mass and orbit
84: ($\sim 0.003 M_\odot$, $d \approx 1-10 \times 10^{4}$ AU) of such a perturber
85: depend on whether its existence is invoked to explain temporal features in mass
86: extinctions on Earth (``Nemesis''; e.g.,~\citealt{dhm84},\citealt{wj84}, and
87: \citealt{vs93}) and/or the trajectories of anomalous streams of comets
88: (``Planet X''; see \citealt{murr99} and \citealt{mww99}, but see a more
89: cautious view from \citealt{he02}).
90:
91: There are some direct constraints on the existence of any massive (planetary or
92: larger) perturber in the outer Solar System. To have eluded detection by all-sky
93: synoptic surveys like {\it Hipparcos} and Tycho-2 \citep{hog00}, any massive
94: body in the outer Solar System but must be fainter than $V \sim 11$\, mag,
95: corresponding to absolute magnitude $M_V > 21$\,mag for $d < 0.1$ pc.
96: This constraint rules out main sequence stars above the hydrogen-burning limit.
97:
98: Detection of a massive perturber through reflected Solar light grows
99: increasingly difficult with increasing distance due to $r^{-4}$ dimming. In
100: reflected light, at current sensitivity limits and angular size coverages,
101: discoveries of objects in the Kuiper Belt at $\approx$40 AU have only recently
102: become routine (e.g., \citealt{btr04}). Yet even with an all-sky synoptic
103: survey to limiting magnitudes of $R=24$\, mag (e.g.,
104: Pan-STARRS\footnotemark\footnotetext{{\tt
105: http://pan-starrs.ifa.hawaii.edu/project/reviews/PreCoDR/documents/
106: scienceproposals/sol.pdf}}), massive planets like Neptune would be undetectable
107: via reflected light beyond $\sim$800 AU and a 0.1 $M_\odot$ perturber with a
108: density of 1 g cm$^{-3}$ would be undetected with $d > 2000$\,AU.
109:
110: Old and cooled degenerate stars (emitting thermally) could be faint enough to
111: have gone undetected. The oldest neutron star (NS) known with an apparent
112: thermal emission component is B0950$+$08 with $M_B \approx (20.0 \pm 0.2)$\,mag
113: \citep{zsk+02} ($d \approx 260~{\rm pc}$; age = 10$^{7.2}$ yr). At $d=90,000$
114: AU, the source would be $B \approx 13$\,mag, likely detectable with the next
115: generation synoptic surveys. However, with a cooling time that of the age of
116: the Solar System, we would expect a NS perturber to have cooled considerably,
117: likely to $T < 10^3$ K from $T \approx 10^5$ K (extrapolating from
118: \citealt{plp+04}) and so would be significantly fainter than current detection
119: levels. Constraints on the existence of even colder distant planetary-mass
120: objects from the
121: lack of detection of their thermal infrared emission with the {\it Infrared
122: Astronomical Satellite (IRAS)} are largely superseded by constraints from the
123: ephemerides of the outer planets~\citep{hogg91}. An infrared survey with
124: significantly higher spatial resolution and sensitivity may provide interesting
125: constraints on distant objects.
126:
127: Surveys that monitor distant stars with high cadence to search for
128: occultations by foreground objects are in principle sensitive to
129: objects of mass as low as $\sim 0.01~\mearth$ out to the Galactic
130: tidal radius of the solar system at $\sim 10^5~{\rm AU}$. However, the
131: probability that any one object will occult a sufficiently bright
132: background star to be detectable is very low. Therefore, in order to
133: detect any occultation events at all, a large number of objects must be
134: present. Thus such surveys can only constrain the existence of a substantial
135: population of objects, and will place essentially no constraints on
136: the existence of individual bodies in the outer solar system.
137:
138: Clearly, the limits on faint massive objects in the outer Solar System must be
139: probed with a fundamentally different technique than through reflected, thermally
140: emitted, or occulted light. Here we suggest an indirect search for massive
141: outer Solar System bodies by observing the differential astrometric
142: microlensing signature that such bodies would impart on the distant stars. As
143: the apparent position of the lens moves on the sky, astrometric monitoring of
144: background sources in the vicinity of the lens (with the appropriate
145: sensitivity) will reveal a complex pattern of apparent motion of those
146: background sources. In \S \ref{sec:ip} we introduce the microlensing formalism
147: in the regime of interest. Detecting the astrometric microlensing signature of
148: a lens requires either the background stars to move and/or the lens to move.
149: Nearby objects exhibit extremely large parallaxes and so the apparent position
150: of the lens, regardless of whether it can be detected directly in reflected
151: light, sweeps out a large area of influence on the sky even if the proper
152: motion of lens is small. Indeed, parallax dominates the apparent motion of
153: objects in Solar orbit beyond the Kuiper Belt. In \S \ref{sec:sn} we estimate
154: the detectability of a nearby massive perturber using the data from the \gaia\ mission\footnotemark\footnotetext{Launch excepted June 2011; {\tt
155: http://astro.estec.esa.nl/GAIA/}} using a Monte Carlo simulation. Finally, we
156: highlight some improvements in the detectability estimate for future work.
157:
158: \section{Properties of Induced Parallax} \label{sec:ip}
159:
160: Consider a distant source with parallax $\pis$ with an (angular) separation
161: $\bth$ from a foreground massive body with parallax $\pix$. The foreground
162: body will deflect the apparent position of the centroid of the background
163: source relative to its unlensed position by, \be \Delta \bth =
164: \frac{\bu}{u^2+2}\thetae, \ee where $\bu=\bth/\thetae$ is the angular
165: separation of lens and source in units of the angular Einstein ring radius, \be
166: \thetae=(\kappa \mx \pirel)^{1/2}. \label{eqn:thetae} \ee Here
167: $\kappa=4G/c^2\au=8.144~{\rm mas}/\msun$, and $\pirel=\pix-\pis$ is the
168: relative lens-source parallax. For the cases considered here, $\pis \ll \pix$.
169: For $u\gg 1$, $|\Delta \bth| = \thetae^2/\theta$.
170:
171: Due to parallax, the apparent position of the massive body will trace out an
172: ellipse on the sky over the course of a year. In addition, it will have a
173: proper motion $\mux$ due to its intrinsic motion. In ecliptic coordinates, the
174: position of the lens at time $t$, relative to time $t_0$ has components, \be
175: \delta\lambda_X(t)=\pix \sin(\omega[t-t_0]) + \mux(t-t_0)\cos\gamma \ee \be
176: \delta\beta_X(t)=-\pix\sin(\beta)\cos(\omega[t-t_0]) + \mux(t-t_0)\sin\gamma,
177: \label{eqn:lambet} \ee where $\beta$ is the ecliptic latitude of the object,
178: and $\gamma$ is the angle of the proper motion with respect to the ecliptic
179: plane. For orbits with zero inclination (in the plane of the ecliptic),
180: $\gamma=0$. We have also assumed $\pix\ll 1~{\rm rad}$.
181:
182: \begin{figure} \centerline{\psfig{file=f1.eps,width=6.9in,angle=90}}
183: \caption{\label{fig:one} Illustration of the effects of a nearby lens on
184: background sources as the lens sweeps through its parallactic ellipse (black
185: solid line). The unlensed positions of the sources, assumed to be fixed and
186: with no parallax, are marked with small squares. The proper motion of the lens
187: is suppressed for illustration. Colors in the deflection tracks indicate the
188: relative amount of the maximum deflection (longer wavelengths correspond to
189: less departure from the unlensed position). This realization assumes a $M_{\rm
190: X}$ = 1.4 $M_{\odot}$ NS at a low ecliptic latitude ($b=1.7^{\circ}$) with a
191: heliocentric distance $d_{\rm X}$ = 10000 AU; the actual density of stars to
192: the sensitivity limit of \gaia\ is typically lower than shown here. For
193: other configurations of the lens, the ordinate scales as $10000$ AU/$d_{\rm X}$
194: and the abscissa scales as $\sin(b/1.7^\circ)/0.0297 \times (10000$ AU/$d_{\rm
195: X})$. The normalization of the deflection angles scales as $M_{\rm X}/1.4
196: M_{\odot}$, and, for the other Solar System deflector objects of interest,
197: would be several orders of magnitude smaller than shown here.} \end{figure}
198:
199: The deflection tracks of background stars that are astrometrically microlensed
200: by the motion of lens parallax (hereafter ``induced parallax'') can exhibit a
201: variety of shapes depending on the angular position with respect to the
202: parallactic ellipse of the lens. Figure \ref{fig:one} shows a realization of
203: several tracks around a neutron star at 10,000 AU. For sources at large impact
204: parameter to the lens, the apparent positions over the year trace out a curved
205: path along a distortion angle approximately parallel with the direction of
206: motion of the lens at the minimum impact of the source along the parallactic
207: ellipse. Near the position of maximal parallactic position of the lens, these
208: curves resemble ``tear drop'' shapes. For impacts comparable to the semi-minor
209: axis of the parallactic ellipse ($\pix \sin \beta$), the deflection tracks take
210: the appearance of ``crescent'' shapes or a ``circle-within-circle''. This is
211: due to comparable deflection during the nearest impact and the distant
212: opposite-side impact months later; these such types of deflection paths are
213: obviously more common at smaller $|\beta|$. Sources interior to the parallactic
214: ellipse are deflected near maximally twice a year, resulting in shapes
215: resembling a ``figure eight''.
216:
217: Although we call the deflection tracks due to parallactic motion of the lens
218: ``induced parallax,'' the deflection tracks generally do not resemble the
219: traditional parallactic ellipse. First, the eccentricity of the tracks does not
220: generally scale with $\cos b$. Second, the direction of motion along the
221: tracks is {\it retrograde} with respect to the parallactic motion of the lens.
222: Moreover, unlike traditional parallax (where the date of maximum departure is
223: fixed by the ecliptic azimuth), the time of maximum departure from the unlensed
224: positions depends only on the time of minimum impact of the source to the lens.
225: In these ways, the source parallactic motion may be distinguished from the
226: effects due to induced parallax in principle. However, in practice the presence
227: of intrinsic source proper motion and parallax, which are typically much larger
228: than the signals we are concerned with here, as well as poor sampling and
229: signal-to-noise ratio, may cause considerable degradation of the detectability. We
230: consider these issues in more detail below.
231:
232:
233: \section{Estimating the Lens Mass-Distance Sensitivity of an Astrometric Survey}
234: \label{sec:sn}
235:
236: Figure \ref{fig:one} shows a rather dramatic effect of a nearby neutron star
237: upon a background field, with deflections of many background sources more than
238: arcseconds from unlensed positions. Since the magnitude of the deflection
239: tracks scales as the mass of the lens, all-sky astrometric missions could, in
240: principle, probe to masses significantly smaller than $M_\odot$. We now
241: quantify what mass/distance configurations would give rise to a detectable
242: signal in the presence of astrometric uncertainty and a finite number of
243: position samples of the background sources. Though the deflection of a single
244: background source may not be detectable, clearly neighboring sources will
245: exhibit similar, correlated deflection; therefore, the presence of a nearby
246: massive lens can be inferred at a statistically significant level by
247: aggregating a collection of statistically insignificant deflections.
248:
249: Consider a massive body in solar orbit with mass $\mx$ and heliocentric
250: distance $\dx$. This body will have a parallax $\pix=\au/\dx$, and a proper
251: motion $\mux=v_X/\dx$, where $v_X$ is its transverse velocity. If we assume
252: that the body is in a circular orbit, and that $\dx\gg \au$ (so that projection
253: effects are small), then $v_X=v_\oplus \pix^{1/2}$.
254:
255: Now consider that the body is moving in front of a background screen of source
256: stars with surface density $\Sigma_*$, and that series of $N$ astrometric
257: measurements of these stars are taken at times $t_j$. At each time $t_j$, we
258: can compute the deflection due to the lens $\Delta \bth_k(t_j)=[\Delta
259: \theta_{\lambda,k}(t_j),\Delta \theta_{\beta,k}(t_j)]$, for each source $k$,
260: using the expressions presented in \S\ref{sec:ip}. Assuming all the source
261: stars have the same (one-dimensional) astrometric uncertainty $\siga$, we can
262: estimate the total signal-to-noise ratio $\sn$ with which the deflection of the
263: massive body is detected as, \be (\sn)^2 = \frac{1}{2\siga^2} \sum_k \sum_j
264: (\Delta \theta_{\lambda,k}(t_j)-\ave{\Delta \theta_{\lambda,k}})^2+ (\Delta
265: \theta_{\beta,k}(t_j)-\ave{\Delta \theta_{\beta,k}})^2. \label{eqn:snone} \ee
266: Here $\ave{\Delta \theta_{\lambda,k}}$ and $\ave{\Delta \theta_{\lambda,k}}$
267: are the average deflections, i.e.\ $\ave{\Delta \theta_{\lambda,k}} \equiv
268: N^{-1}\sum_j \theta_{\lambda,j}$. These are the average positions of the source
269: determined over the course of the \gaia\ mission relative to some external
270: reference grid well away from the deflector. Adopting this $\sn$ criterion for
271: detection is in some sense conservative, in that it only defines the
272: significance with which the positions of the background stars differ from the
273: null hypothesis of no deflections. The effective $\sn$ will likely be
274: increased by fitting a model to the data which implicitly accounts for the
275: shape of the deflection track, as well as the correlation between neighboring
276: sources. We note that, using the median deflections in \eq{eqn:snone}, rather
277: than the mean, increases the $\sn$ by $\sim 10\%$.
278:
279: We estimate the $\sn$ using a simple Monte Carlo\footnotemark\footnotetext{Note
280: that it is possible, using some simplifying assumptions and by analyzing the
281: problem in limiting regimes, to make significant analytical progress and arrive
282: at simple expressions for the signal-to-noise ratio as a function of the mass
283: and distance to the perturber, as well as the surface density and astrometric
284: accuracy of the source stars. We have chosen not to present these analytic
285: expressions here, as there are not fully general, and so one ultimately must
286: resort to numerical evaluations to determine the detectability in all relevant
287: regimes. We note that these analytic results generally confirm the numerical
288: results we now present.}. We create a random screen of stars, and simulate a
289: series of $N$ uniformly sampled measurements. We then calculate $\sn$ using
290: \eq{eqn:snone}. Under our assumptions, the $\sn$ depends on the parameters of
291: the lens, $\mx,\pix,\beta,t_0,\gamma$, as well as the properties of the source
292: stars, $\Sigma_*,\siga$. We calculate $\sn$ for many different realizations
293: of the positions of the background source stars, and we also vary the input
294: parameters. We find the following approximate expression for the signal-to-noise
295: ratio, \be
296: \sn \simeq \left\{ \begin{array}{ll} \frac{10\muas}{\sqrt{2}\siga}
297: \left(\frac{\mx}{\mearth}\right) \left(\frac{\dx}{10^3~\au}\right)^{-1}
298: \left(\frac{\Sigma_*}{10^{-3}~{\rm arcsec}^{-2}}\right)^{1/2}
299: \left(\frac{N}{40}\right)^{1/2} \left(1+\sin{\beta}\right) & {\rm if}\,\,
300: \Sigma_*\pi\pix^2 \ge 1 \\
301: \frac{10\muas}{\sqrt{2}\siga}
302: \left(\Sigma_*\pi\pix^2\right)^{1/2}
303: \left(\frac{\mx}{\mearth}\right) \left(\frac{\dx}{10^3~\au}\right)^{-1}
304: \left(\frac{\Sigma_*}{10^{-3}~{\rm arcsec}^{-2}}\right)^{1/2}
305: \left(\frac{N}{40}\right)^{1/2} \left(1+\sin{\beta}\right) & {\rm if}\,\,
306: \Sigma_*\pi\pix^2 < 1 \end{array}\right.. \label{eqn:snscale}
307: \end{equation}
308: The two regimes in \eq{eqn:snscale} correspond to the strong, `collisional'
309: regime where there is on average one star in the parallactic ellipse, and the
310: weak, `tidal' regime where there is typically less than one star in the
311: ellipse. Equation \ref{eqn:snscale} is generally accurate to considerably
312: better than the variance at fixed values of the parameters due to Poisson fluctuations
313: in the number density and location of source stars, for most parameter
314: combinations. The $\sn$ can vary by a large amount due to Poisson noise
315: depending on the parameters, and especially so in the tidal regime for low
316: $\Sigma_*$. Note that, as reflected in \eq{eqn:snscale}, we find that the $\sn$
317: does not depend on $t_0$ or $\gamma$ to within the Poisson fluctuations.
318:
319: \subsection{Application to \gaia}
320:
321: In order to provide a quantitative estimate of the mass-distance
322: sensitivity of an astrometric survey to massive objects in the outer
323: solar system, we adopt parameters appropriate for the \gaia\
324: mission. \gaia\ will monitor the entire sky synoptically for five
325: years, acquiring astrometric measurements for $O(10^9)$ stars down
326: to apparent magnitudes of $V\sim 20$. For bright stars $(V\le 12)$,
327: \gaia\ will have a single-measurement astrometric precision limit of
328: $30~\muas$, whereas at $V\sim 20$, the astrometric accuracy will be
329: $\sim 1400~\muas$. Typically, each star will have $100-200$
330: astrometric measurements, grouped in clusters of 2 to 5 measurements
331: each.
332:
333: To proceed with our estimate,
334: we adopt a model of the surface density of source stars on the sky as a
335: function of magnitude, Galactic latitude and longitude, and a model of the
336: expected astrometric performance of \gaia. This allows
337: us to predict the total $\sn$ with
338: which a object of a given mass and distance would be detected with \gaia,
339: at a given
340: location in the sky.
341:
342: The expected performance of \gaia\ has and will continue to evolve, and the
343: final mission astrometric accuracy is therefore impossible to access currently.
344: For definiteness, we assume that the (one-dimensional) astrometric uncertainty
345: of each measurement is given by, \be\label{eqn:sig1d} \sigma_{1D}^2(V)=\left\{
346: \begin{array}{ll} \sigma^2_{sys}\qquad &{\rm if}\qquad V\le 12.5 \\ \sigma^2_s
347: 10^{0.4(V-12.5)} + \sigma^2_b 10^{0.8(V-20)} \qquad &{\rm if} \qquad V > 12.5
348: \end{array},\right. \ee with $\sigma_{sys}=\sigma_{s}=30~\muas$ and
349: $\sigma_b=1000\muas$. This form was chosen to reproduce the astrometric
350: accuracies from Table 1 of \citet{be02}. \gaia\ will not make
351: astrometric measurements uniformly across the sky; certain ecliptic
352: latitudes will be sampled a larger number of times than others. We
353: assume that the number of samples as a function of ecliptic latitude
354: $\beta$ is given by, \be N_{samp} =
355: 100+300\exp\left[-\left(\left|\frac{|\beta|-35^\circ}{10^\circ}\right|\right)^{
356: 1/2}\right]. \label{eqn:nsamp} \ee This form was chosen to
357: qualitatively reproduce Figure 5 of \citet{be02}. We assume that the
358: samples are clustered into groups of $n_c$ points, and so the
359: effective number of points is $N=N_{samp}/n_c$, and the effective
360: astrometric accuracy of each point is
361: $\siga=\sigma_{1D}(V)/\sqrt{n_c}$. This assumes that the
362: single-measurement errors can be reduced by root-$n$ averaging. This
363: may not be the case: the measurement errors in any given cluster may
364: be correlated, or there may exist systematic errors that are not
365: reducible. Since it is difficult to anticipate the behavior of the
366: astrometric errors in advance, we will adopt the assumption of
367: root-$n$ averaging for simplicity. We adopt $n_c=5$ \citep{be02}.
368: For other values of $n_c$, the $\sn$ for any given star, as well as
369: the integrated $\sn$, will scale as $\sqrt{n_c/5}$.
370:
371: We determine the surface density of source stars as a function of position and
372: magnitude using a simple model for the Galaxy. For the density distribution of
373: sources, we adopt the double-exponential disk plus barred bulge model of
374: \citet{hg95,hg03}. We assume that the dust column is independent of
375: Galactocentric radius and has an exponential distribution in height above the
376: plane with a scale height of $120~{\rm pc}$. We normalize the midplane column
377: density so that the $V$-band extinction is $A_V=1~{\rm mag}~(D_s/{\rm kpc})$,
378: where $D_s$ is the distance to the source. We also will show results assuming
379: the dust model of \citet{be02}, which is similar to ours for $\beta \ga
380: 20^\circ$, but differs in detail for latitudes closer to the plane. Finally,
381: we assume a $V$-band luminosity function that is independent position and is
382: equal to the Bahcall-Soneira \citep{bs80} luminosity function for $M_V\le 10$,
383: and is constant for $10\le M_V \le 20$.
384:
385: The surface density of stars down
386: to $V=20$ in our model ranges
387: from $\sim 10^{-5}~{\rm arcsec}^{-2}$ near the Galactic poles, to
388: $\sim 10^{-3}~{\rm arcsec}^{-2}$ near the Galactic anticenter, to
389: a maximum of $\sim 0.1~{\rm arcsec}^{-2}$ within a few degrees of the Galactic
390: center. Therefore, regions of the sky near the Galactic plane and especially the Galactic
391: center will have greater sensitivity to lower mass and/or more distant perturbers
392: for fixed $\sn$. The total number of stars in the sky
393: with $10\le V\le 20$ in this model is $1.3\times 10^9$ for our standard dust
394: extinction model, and $1.0\times 10^9$ for the \citet{be02} dust model.
395: Thus the average surface density is $\sim 10^{-3}~{\rm arcsec}^{-2}$.
396:
397: Figure \ref{fig:two} shows the distribution of $\sn$ for
398: an object with $M=3000\mearth\sim 10~\mjup$
399: and $D=10^4~\au$ located in three different locations on the sky:
400: near the Galactic bulge, anticenter and north Galactic pole. The source
401: densities in these three locations vary considerably, from $\Sigma_*
402: \sim 10^{-2}~{\rm arcsec}^{-2}$ near the Galactic bulge to $10^{-5}~{\rm arcsec}^{-2}$ near the pole. The
403: shape of the distribution of $\sn$ depends on the location on the sky,
404: through the distribution of source densities as a function of
405: magnitude (and so astrometric accuracy). For the locations near the
406: Galactic plane with high source densities, the distribution of $\sn$
407: has a tail toward higher values, and so the total $\sn$ is generally
408: dominated by one or two stars. For the location near the Galactic
409: pole, a larger number of stars contribute significantly to the total
410: $\sn$. The total $\sn$ (integrated over $V$-magnitude from $V=10$ to $V=20$)
411: for these three locations are $\sntot=94.4$
412: (bulge), 16.5 (anticenter), and 1.4 (pole).
413:
414:
415: \begin{figure} \plotone{f2.eps} \caption{\label{fig:two}
416: Distribution of signal-to-noise ratios ($\sn$) for
417: detecting a
418: massive object of mass $M=3000\mearth\simeq 10\mjup$ and distance
419: $\dx=10^4~\au$ with \gaia, for
420: various
421: locations on the sky. The histograms show the number of stars with $10\le V\le 20$ as
422: a function of the $\sn$, for three different locations for the massive
423: object with very different background source
424: densities: near the Galactic bulge (solid, $l=10^\circ,b=10^\circ$, $\Sigma_*\sim 10^{-2}$),
425: near the Galactic anticenter (dotted, $l=180^\circ,b=10^\circ$, $\Sigma_*\simeq 10^{-3}$), and
426: near the north Galactic pole (dashed, $l=180^\circ,b=80^\circ$, $\Sigma_*\simeq 10^{-5}$).
427: For these three locations, an object with $M=3000\mearth\simeq 10\mjup$ and
428: $\dx=10^4~\au$ would be detected at $\sntot=94.4$ (bulge), 16.5 (anticenter), and 1.4 (pole).
429: }\end{figure}
430:
431:
432: \begin{figure} \plotone{f3.eps} \caption{\label{fig:three}
433: All sky map in Galactic coordinates of the $\sn$ for detecting a
434: massive object of mass $M=3000\mearth\simeq 10\mjup$ and distance
435: $\dx=10^4~\au$ with \gaia. The dotted lines show
436: lines of constant Galactic latitude and longitude at $15^\circ$ intervals.
437: The Galactic center is located at
438: the center of the figure. Contours of constant $\sn$ are in grey, at
439: levels of $\sn=3,5,10,20,40$ (lighter to heavier). We also show lines
440: of constant ecliptic latitude for $\beta=\pm 35^\circ, \pm 10^\circ,
441: 0$ (solid black lines). The oval shaded region brackets the uncertainty in the inferred
442: position of ``Planet X'' from the clustering of cometary aphelion
443: distances \citep{murr99}.
444: }\end{figure}
445:
446: Figure \ref{fig:three} shows contours of constant $\sntot$ for a
447: object with $M=3000\mearth\sim 10~\mjup$ and $D=10^4~\au$. The
448: distribution of $\sntot$ on the sky is highly non-uniform: objects of
449: a given $M$ and $\dx$ located toward certain regions of the sky will
450: be detected with higher $\sntot$ than if they are located in other
451: regions. The $\sntot$ is primarily driven by the surface density of
452: stars, and therefore regions of the sky near the Galactic plane and
453: especially the Galactic center are preferred. However, it is also the
454: case that the number of samples $N_{samp}$ depends on ecliptic
455: latitude, such that stars with ecliptic latitude $\sim \pm 35\deg$
456: will have several times more astrometric measurements than stars near
457: the ecliptic poles. Therefore locations near ecliptic
458: latitudes of $\pm 35\deg$ will also have higher $\sntot$ for fixed perturber
459: mass and distance.
460:
461: Figure \ref{fig:four}
462: shows the fraction of the sky enclosed by contours of a given $\sntot$, i.e.\
463: the fraction of the sky over which an object of mass $M=3000\mearth$ and
464: distance $\dx=10^4~\au$ would be detected with $\sn$ greater than a given
465: value. We determine the fraction of sky above a given $\sntot$ for a range of
466: masses and distances. Objects with mass greater than a minimum mass \be M_{min}
467: \simeq (290,490,750)M_\oplus \left\{ \begin{array}{ll}
468: \left(\frac{\dx}{10^4~\au}\right) \left[\frac{(\sn)_{\rm th}}{5}\right] & {\rm
469: if}\, \dx\le\db \\ \left(\frac{\dx}{\db}\right)
470: \left(\frac{\dx}{10^4~\au}\right) \left[\frac{(\sn)_{\rm th}}{5}\right] & {\rm
471: if}\, \dx> \db \end{array}\right\},\, {\rm for}~\fsky=(10\%,50\%,100\%).
472: \label{eqn:mscale} \end{equation} can be detected with $\sn \ge (\sn)_{\rm
473: th}$, where $\fsky$ is the fraction of the sky. Here $\db$ is the `break
474: distance', and has values of $\db=(4470,1550,780)\au$ for
475: $\fsky=(10\%,50\%,100\%)$. These limits are shown in Figure~\ref{fig:five}.
476:
477: \begin{figure} \plotone{f4.eps} \caption{\label{fig:four} Fraction of the sky
478: over which an object of mass $M=3000\mearth\simeq 10\mjup$ and distance
479: $\dx=10^4~\au$ would be detected with $\sn$ greater than a given value. The
480: solid curves show the fraction assuming our fiducial model for the dust
481: distribution, whereas the dotted lines shows the alternative model of
482: \citet{be02}. Upper curves are for the entire sky, the bottom curves are for
483: ecliptic latitudes $|\beta|<10^\circ$. } \end{figure}
484:
485: Figure \ref{fig:four} also shows the fraction of the sky within $10^\circ$ of
486: the ecliptic plane enclosed by contours of a given $\sntot$. Since the ecliptic
487: plane fortuitously passes near the Galactic bulge, the slope for this curve is
488: shallower than that for the entire sky, resulting in a relatively larger
489: fraction of the area for which a high $\sn$ detections are possible.
490:
491: There are several obvious limitations of our calculations. One is that we have
492: neglected the motion of background stars due to parallax. To the extent that
493: these motions correlate with the microlensing signal, they will tend to degrade
494: the signal-to-noise ratio, by effectively allowing one to partially `fit out' the
495: anomalous excursions. Motions of stars in binaries could also confound a clean
496: measurement of induced parallax. Also, since we adopted the simple scaling
497: relation in \eq{eqn:snscale} when integrating over the magnitude distribution
498: of source stars, we have neglected the effect of the Poisson fluctuations of
499: the surface density and location of stars on the total signal-to-noise ratio.
500: To provide a rough
501: estimate of the magnitude of these effects, we have performed a few simulations
502: where we determine the signal-to-noise ratio for stars of a given magnitude
503: directly from the Monte Carlo simulation (which {\it per force} includes
504: Poisson fluctuations), while explicitly fitting for the parallax of the source
505: stars. Since these calculations are extremely time intensive, we have not
506: performed a comprehensive exploration, but rather checked only a few cases.
507: For these few cases, we find that fitting for the parallax of the source
508: does indeed reduce $\sntot$, but by a relatively small factor, $\sim 10\%$.
509: On the other hand, we find that the effect of Poisson fluctuations causes us to
510: {\it underestimate} $\sntot$, by as much as $\sim 75\%$.
511:
512:
513: \begin{figure} \plotone{f5.eps} \caption{\label{fig:five} The lines show
514: constraints on the mass and distance of an object that can be detected by
515: \gaia\ at various $\sn$ thresholds over various fractions of the sky. The red
516: line shows where the angular size $\theta_X$ of an object with density of
517: $1~{\rm g~cm^{-3}}$ is equal to its Einstein ring radius $\thetae$; objects to
518: the left of this line have $\theta_X\ge \thetae$. Objects with parameters in
519: the shaded region will occult at least one measurement of at least one
520: background source, assuming a typical background source density of $2\times
521: 10^{-3}~{\rm arcsec}^{-2}$.
522: }
523: \end{figure}
524:
525: \section{Discussion and Conclusions}
526:
527: We have shown that a substantial, as yet unexplored, region of mass-distance
528: parameter space of nearby massive bodies will be accessible with the current
529: incarnation of the datastream from the \gaia\ experiment. We have focused
530: on the effect of ``induced parallax'' caused only by the parallax of the lens
531: as it sweeps through the parallactic ellipse. Based on our albeit simplistic
532: simulation, the search for massive bodies in the outer Solar System by the
533: observation of induced parallax has a reasonable chance of uncovering the
534: proposed perturber of cometary orbits in the Oort cloud (Figs.~\ref{fig:four}
535: and \ref{fig:five}). In particular, we believe that the non-detection of a massive
536: body in the \gaia\ dataset using the proposed technique would relegate the proposed
537: mass-distances of Planet X to a significantly smaller
538: parameter space then the currently allowed space\footnotemark\footnotetext{It
539: is noteworthy that \citet{he02} also appeal to \gaia\ for constraining the
540: existence of Planet X, but by making use of ephemeris data of $\sim$1000
541: long-period comets that would be discovered by \gaia\ relatively uniformly
542: over the sky.}.
543:
544:
545: \begin{figure}
546: \epsscale{0.8}
547: \plotone{f6.eps} \caption{\label{fig:six} All-sky constraints
548: on distant massive solar-system objects using various methods. Masses and
549: distances to the upper left of the lines are currently excluded by these
550: methods (`Comets,' `Planet Orbits,' and `Timing'),
551: or can be excluded in the future (`Reflected Light,' `Gaia,' and `Occultation').
552: Limits marked `Comets' and `Planet
553: Orbits' were taken from \citet{hogg91}; the
554: limit marked `Timing' is from \citet{zakamska05}; the `Occultation' limit is derived
555: following \citet{gaud04}. The line marked `Gaia' shows the mass and distance of
556: an object that can be detected by \gaia\ at $\sn=5$ over 50\% of the sky. The
557: line marked ``SuperGaia'' shows the same limit for a hypothetical experiment
558: with astrometric sensitivity that is two orders of magnitude better than \gaia\
559: with the same limiting magnitude. The small points show the masses and
560: distances of known Kuiper Belt Objects (KBOs) and minor planets, where the masses were
561: derived from their absolute magnitude assuming an albedo of $4\%$ and a density
562: of $1~{\rm g~cm^{-3}}$. The open circles show the masses of the three recently-discovered
563: bright KBOs 2003 EL$_{61}$, 2003 UB$_{313}$ and 2005 FY$_9$ (see \citealt{brown05} and references therein), under the same assumptions. The large squares show the masses and distances of
564: Uranus, Neptune, and Pluto. The shaded region at the extreme right shows the
565: tidal radius of the solar system, $\sim 10^5~{\rm AU}$. The shaded
566: rectangles show the range of masses and distances inferred for `Planet X'
567: and `Nemesis.' Since the proposed ecliptic longitude and
568: latitude of `Planet X' \citep{murr99} is near the Galactic plane, the
569: limiting masses
570: probed in that region are smaller than over the sky as a whole: the short lines
571: passing through the shaded region labeled `Planet X' show the lower limits on the mass and
572: distance of an object that yields $\sn=1,3,5$ (lighter to heavier) at the
573: \citet{murr99} position.
574: }
575: \end{figure}
576:
577: \citet{murr99} made specific predictions for the current position of ``Planet
578: X'' on the sky, based on the clustering of cometary aphelion distances. Since
579: the $\sn$ map of the sky is not uniform, it is interesting to ask with what
580: $\sn$ one would expect to detect ``Planet X'' with the allowed mass and
581: distances, at its expected position. Figure \ref{fig:three} shows the positional
582: error ellipse from \citet{murr99}. The expected $\sn$ for $\mx=10^3\mearth$
583: and $\dx=10^4\au$ ranges from $\sntot\simeq 3$ to $\sntot \simeq 12$. The
584: mass/distance limit for thresholds of $(\sn)_{th}=1,3$, and $5$ in this error
585: ellipse is shown in Figure \ref{fig:six}; roughly 25\% the allowed parameter
586: space could be excluded at $3\sigma$ with a non-detection.
587:
588: Hypothesis for the mass ($\sim 0.03\msun$) and distance ($\sim
589: 10^5~\au$) of Nemesis will likely be difficult to test with \gaia\ (see
590: Fig.~\ref{fig:six}), due primarily to the large distance and thus
591: small size of the parallactic ellipse. However, specific predictions
592: for the current position of Nemesis might be testable using a targeted
593: astrometric satellite with higher astrometric precision than \gaia,
594: such as the {\it Space Interferometry Mission (SIM)}. Of course,
595: constraints on smaller-mass objects at any distance could be obtained with
596: with an all-sky synoptic experiment that has improved astrometric
597: accuracy but with a similar limiting magnitude (``SuperGaia'',
598: Fig.~\ref{fig:six}) or by probing more stars to fainter levels with
599: \gaia-like astrometric accuracies.
600:
601: Should a significant detection be made, what can be learned about the lens? In
602: principle, the astrometric data alone provide an estimate of the mass,
603: position, distance, and proper motion of the lens for high-$\sn$ detections of
604: induced parallax for stars very near to the parallactic ellipse. Orbit
605: determination will generally be difficult, unless there is a significant
606: acceleration over the five year mission lifetime; this is only expected for
607: relatively nearby lenses. For more modest $\sn$ detections, or detections in
608: the tidal regime where the source stars are quite distant from the parallactic
609: ellipse, the information will be seriously degraded, and degeneracies between
610: the mass, distance, and angular separation from the lens arise. In the extreme
611: case where only one distant star is significantly perturbed, the detection may
612: yield very little information about the lens. Exploration of the information
613: that can be extracted from these various classes of detections is beyond the
614: scope of this paper, but is an interesting topic for future study.
615:
616: Further follow-up of potential candidates may be possible with a variety of
617: methods. Astrometric follow-up of individual background sources may be
618: possible with {\it SIM} with higher
619: astrometric precision and cadence than possible with \gaia; such measurements
620: may improve on the determination of the lens parameters. Direct detection of
621: the reflected light from some candidates may be possible with ultra-deep
622: imaging using very large aperture, next generation, ground-based,
623: optical/near-infrared telescopes such as the {\it Giant Magellan Telescope
624: (GMT)}, the {\it Thirty Meter Telescope (TMT)}, or the {\it Overwhelmingly
625: Large Telescope (OWL)}. Finally, the {\it James Webb Space Telescope (JWST)}
626: should have the sensitivity to detect the thermal emission from essentially all
627: objects detected astrometrically by \gaia.
628:
629: A similar astrometric microlensing search with \gaia\ for massive stellar
630: remnants in the Solar neighborhood ($d \sim 150$\,pc) was proposed by
631: \citet{be02} but with several important differences compared to the present
632: work. First, we considered the detectability of an object significantly closer
633: to Earth so that the lens parallax is $\sim10^{5-7}$ larger than the typical
634: source parallaxes whereas that difference is only $10^{1-2}$ for Solar
635: neighborhood lenses. We also focused on Solar System lenses in Solar orbit
636: where the parallax motion dominates proper motion; the motion of Solar
637: neighborhood objects are dominated by proper motion. Both these different
638: regimes result in significantly different microlensing tracks of a single
639: background star (compare our Fig.~\ref{fig:one} with Fig.~1 of \citealt{be02}).
640: Second, we focus on detection of objects with a planet-scale mass whereas the
641: analysis technique of \citealt{be02} is optimized to constrain the mass
642: function of stellar-mass objects in the Solar neighborhood (see, e.g., Fig.\
643: \ref{fig:three}) with $M > 0.1 M_\odot$. Last, and conceptually the most
644: distinct, we consider the detectability of a single massive object using the
645: aggregate induced parallax signatures of thousands of stars whereas
646: \citeauthor{be02} focused on constraining the properties of a large population
647: of faint stellar-mass objects, where the mass of each object is inferred using
648: the astrometric microlensing ``event'' a single background source. Ultimately,
649: though, both analyzes make use of the same datastream and act toward
650: complimentary goals.
651:
652: We have assumed that our lenses are point-like, and so have ignored the effects
653: of occultation of the background sources by the lens. If the angular size of
654: the lens $\theta_X$ is an appreciable fraction of its angular Einstein ring
655: radius $\thetae$, then both occultation and lensing effects can potentially be
656: important \citep{agol02,taka03}. In Figure \ref{fig:five}, we show the locus of
657: mass and distance where $\theta_X=\thetae$. Objects with $M_X\la 10^3~\mearth$
658: will have angular sizes that are larger than their Einstein ring radii provided
659: they are closer than $\sim 4000~\au$; for such objects, complete occultations
660: are possible. However, an occultation will obviously only occur if a
661: background source happens to be located within an angular radius of the lens
662: when a measurement is taken. This condition is met when the number of
663: measurements satisfies $N\Sigma_* \pi \theta_X^2\sim 1$. Figure \ref{fig:five}
664: shows the region of parameter space for which at least one measurement will be
665: occulted by the lens, for typical background source densities of
666: $\Sigma_*=2\times 10^{-3}~{\rm arcsec}^{-2}$. Clearly, for most lenses,
667: occultation effects are negligible.
668:
669: In our simulation, we assumed the perturber is in a circular orbit around the
670: Sun. However, we found that our results are essentially independent of the
671: proper motion of the lens. Furthermore, realistic motions along the
672: line-of-sight are unlikely to alter our signal-to-noise ratio estimates
673: substantially for the distances considered herein. Therefore, the assumption
674: that the lens is on a circular orbit or indeed even bound to the Sun is
675: immaterial to our conclusions.
676:
677: As we have discussed, an obvious shortcoming of our estimation is that we have
678: neglected the motion of background stars due to parallax, proper motion, and
679: orbits. These motions will tend to degrade the signal-to-noise ratio, effectively
680: introducing more free parameters to help explain away anomalous excursions.
681: Still our preliminary calculations show that source parallax is not likely to
682: degrade the $\sn$ substantially, however these calculations were admittedly
683: limited. We hope to perform a more comprehensive study to quantify the effect
684: of a realistic background screen in future work.
685:
686: Our simplistic simulation for S/N estimation also neglects another feature of
687: data that could be exploited to {\it improve} the S/N. Any nearby foreground
688: massive source will lens multiple source background stars differently in the
689: course of a 5 year mission. Moreover, neighboring background sources will be
690: lensed similarly. So the expectation of correlated deflection paths (which are
691: fixed for a given lens mass, distance, and proper motion) could be used to
692: create a ``matched filter'' for improving the sensitivity of detecting a nearby
693: massive lens. Though computationally very expensive, one can envision applying
694: such a filter to the \gaia\ dataset for all possible nearby lens masses at
695: all possible distances and positions on sky to search for a signal. Aside from
696: the need to simultaneously constrain the parallax, proper motion, and orbital
697: parameters of all background sources, the matched filter search may also need
698: to search for a possible changing parallax of the lens over the mission
699: lifetime: a massive object passing nearby that is unbound to the Sun with $|v|
700: \approx 30$ km s$^{-1}$ would travel $\approx$30 AU over 5 yr, with some of
701: this motion in the radial direction from the Sun.
702:
703: Finally, the choice of the appropriate $\sn$ threshold for a robust detection
704: deserves some discussion. Here one must not only consider the astrometric
705: noise properties of the sources, but also the total number of independent
706: trials performed in searching the data with a matched filter. This latter
707: factor can be quite crucial in the current context, given the fact that one is
708: performing a blind search over the entire sky with $O(10^9)$ source stars, with
709: many independent filters corresponding to varying lens locations, distances,
710: masses, and proper motions.
711:
712: While a high signal-to-noise ratio measurement of the entire induced parallax of
713: single star will yield the lens mass, sky position, proper motion, and
714: distance, the likelihood of such a special configuration is rare. Instead,
715: each of these lens events will contribute individually to constraints on the
716: lens properties at different times, leading to the possibility of improving the
717: signal-to-noise ratio of the lens properties through the matched filter. Another
718: utility of global astrometric filtering of the \gaia\ data is that the
719: masses and ephemerides of known Solar System objects might be determined {\it a
720: priori}, based solely on measurements of the astrometric microlensed
721: background; whether the masses determined thusly will be more precise than
722: measured by other means remains to the be seen.
723:
724: \acknowledgments BSG supported by a Menzel Fellowship from the Harvard College
725: Observatory. JSB was partially supported by a grant from the
726: Harvard-Smithsonian Center for Astrophysics. We would like to thank Avi Loeb
727: for comments on the manuscript, and the anonymous
728: referee for a prompt and helpful report. We would like to extend special thanks to Andy
729: Gould for his detailed and comprehensive comments and insightful discussions
730: which led to a much improved paper. Lastly, JSB thanks Eugene Chiang and Josh
731: Eisner for their enthusiasm during the early stages of this work.
732: \begin{thebibliography}{24}
733: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
734:
735: \bibitem[{{Agol}(2002)}]{agol02}
736: {Agol}, E. 2002, \apj, 579, 430
737:
738: \bibitem[{{Bahcall} \& {Soneira}(1980)}]{bs80}
739: {Bahcall}, J.~N. \& {Soneira}, R.~M. 1980, \apjs, 44, 73
740:
741: \bibitem[{{Belokurov} \& {Evans}(2002)}]{be02}
742: {Belokurov}, V.~A. \& {Evans}, N.~W. 2002, \mnras, 331, 649
743:
744: \bibitem[{{Brown} {et~al.}(2004){Brown}, {Trujillo}, \& {Rabinowitz}}]{btr04}
745: {Brown}, M.~E., {Trujillo}, C., \& {Rabinowitz}, D. 2004, \apj, 617, 645
746:
747: \bibitem[{{Brown} {et~al.}(2005){Brown}, {Trujillo}, \& {Rabinowitz}}]{brown05}
748: ---. 2005, \iaucirc, 8577, 1
749:
750: \bibitem[{{Byl}(1983)}]{byl83}
751: {Byl}, J. 1983, Moon and Planets, 29, 121
752:
753: \bibitem[{{Davis} {et~al.}(1984){Davis}, {Hut}, \& {Muller}}]{dhm84}
754: {Davis}, M., {Hut}, P., \& {Muller}, R.~A. 1984, \nat, 308, 715
755:
756: \bibitem[{{Gaudi}(2004)}]{gaud04}
757: {Gaudi}, B.~S. 2004, \apj, 610, 1199
758:
759: \bibitem[{{Han} \& {Gould}(1995)}]{hg95}
760: {Han}, C. \& {Gould}, A. 1995, \apj, 447, 53
761:
762: \bibitem[{{Han} \& {Gould}(2003)}]{hg03}
763: ---. 2003, \apj, 592, 172
764:
765: \bibitem[{{Hills}(1981)}]{hills81}
766: {Hills}, J.~G. 1981, \aj, 86, 1730
767:
768: \bibitem[{{H{\o}g} {et~al.}(2000){H{\o}g}, {Fabricius}, {Makarov}, {Urban},
769: {Corbin}, {Wycoff}, {Bastian}, {Schwekendiek}, \& {Wicenec}}]{hog00}
770: {H{\o}g}, E., {Fabricius}, C., {Makarov}, V.~V., {Urban}, S., {Corbin}, T.,
771: {Wycoff}, G., {Bastian}, U., {Schwekendiek}, P., \& {Wicenec}, A. 2000, \aap,
772: 355, L27
773:
774: \bibitem[{{Hogg} {et~al.}(1991){Hogg}, {Quinlan}, \& {Tremaine}}]{hogg91}
775: {Hogg}, D.~W., {Quinlan}, G.~D., \& {Tremaine}, S. 1991, \aj, 101, 2274
776:
777: \bibitem[{{Horner} \& {Evans}(2002)}]{he02}
778: {Horner}, J. \& {Evans}, N.~W. 2002, \mnras, 335, 641
779:
780: \bibitem[{{Hut} {et~al.}(1987){Hut}, {Alvarez}, {Elder}, {Kauffman}, {Hansen},
781: {Keller}, {Shoemaker}, \& {Weissman}}]{hae87}
782: {Hut}, P., {Alvarez}, W., {Elder}, W.~P., {Kauffman}, E.~G., {Hansen}, T.,
783: {Keller}, G., {Shoemaker}, E.~M., \& {Weissman}, P.~R. 1987, \nat, 329, 118
784:
785: \bibitem[{{Matese} {et~al.}(1999){Matese}, {Whitman}, \& {Whitmire}}]{mww99}
786: {Matese}, J.~J., {Whitman}, P.~G., \& {Whitmire}, D.~P. 1999, Icarus, 141, 354
787:
788: \bibitem[{{Murray}(1999)}]{murr99}
789: {Murray}, J.~B. 1999, \mnras, 309, 31
790:
791: \bibitem[{{Oort}(1950)}]{oort50}
792: {Oort}, J.~H. 1950, \bain, 11, 91
793:
794: \bibitem[{{Page} {et~al.}(2004){Page}, {Lattimer}, {Prakash}, \&
795: {Steiner}}]{plp+04}
796: {Page}, D., {Lattimer}, J.~M., {Prakash}, M., \& {Steiner}, A.~W. 2004, \apjs,
797: 155, 623
798:
799: \bibitem[{{Takahashi}(2003)}]{taka03}
800: {Takahashi}, R. 2003, \apj, 595, 418
801:
802: \bibitem[{{Vandervoort} \& {Sather}(1993)}]{vs93}
803: {Vandervoort}, P.~O. \& {Sather}, E.~A. 1993, Icarus, 105, 26
804:
805: \bibitem[{{Whitmire} \& {Jackson}(1984)}]{wj84}
806: {Whitmire}, D.~P. \& {Jackson}, A.~A. 1984, \nat, 308, 713
807:
808: \bibitem[{{Zakamska} \& {Tremaine}(2005)}]{zakamska05}
809: {Zakamska}, N.~L. \& {Tremaine}, S. 2005, \aj, submitted (astro-ph/0506548)
810:
811: \bibitem[{{Zharikov} {et~al.}(2002){Zharikov}, {Shibanov}, {Koptsevich},
812: {Kawai}, {Urata}, {Komarova}, {Sokolov}, {Shibata}, \& {Shibazaki}}]{zsk+02}
813: {Zharikov}, S.~V., {Shibanov}, Y.~A., {Koptsevich}, A.~B., {Kawai}, N.,
814: {Urata}, Y., {Komarova}, V.~N., {Sokolov}, V.~V., {Shibata}, S., \&
815: {Shibazaki}, N. 2002, \aap, 394, 633
816:
817: \end{thebibliography}
818:
819:
820: \end{document}
821:
822:
823:
824:
825: