1: \documentstyle[12pt,aaspp4]{article}
2: %\documentstyle [12pt,aasms4]{article}
3: %\documentstyle [emulateapj]{article}
4: %\documentstyle [aaspp4]{article}
5: \begin{document}
6:
7: \title{Anisotropies in the Motions and Positions of the Galactic Globular
8: Clusters}
9:
10: \author{F.D.A. Hartwick}
11:
12: \affil{Department of Physics and Astronomy, \linebreak University of Victoria,
13: Victoria, BC, Canada, V8W 3P6}
14:
15: \begin {abstract}
16: The velocity ellipsoid for 38 globular clusters with [Fe/H]$\leq-1.0$ is
17: derived and shown to be significantly anisotropic with major axis directed
18: towards low Galactic latitude. Principal axes of the spatial distribution
19: of different groups of clusters are derived and compared with the velocity
20: ellipsoid. The metal poor cluster spatial distribution is significantly
21: flattened along an axis which coincides within the uncertainties with the
22: major axis of the velocity ellipsoid. Given the observed steep age-metallicity
23: relation for metal poor clusters, one speculative interpretation of the
24: data is that an initially flattened filament underwent a relatively rapid
25: initial transverse collapse forming satellite galaxies and metal poor
26: globular clusters while the protogalaxy collapsed and assembled more
27: slowly along the filament acquiring and/or redistributing angular momentum in
28: the process.
29: \end{abstract}
30:
31: \keywords{Galaxy: halo --- Galaxy: formation --- globular clusters: general}
32:
33: \section {Introduction}
34:
35: Globular clusters provide a unique probe of the earliest phases of the
36: evolution of the Galaxy. From the early work of Kinman (1959),
37: Zinn (1985), and others it is clear that the clusters can be divided into at
38: least two groups -- the spatially extended, metal poor blue clusters and
39: the more centrally concentrated towards the Galactic center, metal rich red
40: clusters. This work is
41: concerned primarily with the former group, and the main motivation is the
42: recent availability of space motions for a significant number of these
43: clusters as published by Dinescu et al.\ (1999). Dinescu (private
44: communication) has also provided
45: this author with data for 4 additional clusters. These data are used to
46: determine the velocity ellipsoid for the metal poor group in order to allow
47: a comparison with the spatial distribution. In a previous paper (Hartwick
48: 2000, hereafter H2000) the spatial distribution of various halo samples was
49: examined. There it was found that the metal poor group of globular clusters
50: formed an oblate distribution whose minor axis was directed to a low Galactic
51: latitude. This was in contrast to the metal rich group which exhibited a
52: triaxial distribution with minor axis toward the Galactic pole. We re-examine
53: the spatial distributions with modified samples of clusters and compare the
54: results with the kinematics.
55:
56: \section{The Velocity Dispersion Tensor of the Halo Clusters}
57:
58: Danescu et al.\ (1999) have compiled space motions for 38 globular
59: clusters, and 35 of these have [Fe/H] values less than $-1.0$.
60: Since that time 3 more halo clusters (NGC6522, NGC7006, \& Pal 13)
61: have been added to the compilation.
62: This sample of 38 clusters is the basis for the following kinematical
63: investigation. The tabulated
64: UVW components of space motion (corrected solar peculiar motion and for
65: Galactic rotation of 220 km~sec$^{-1}$) are referred to to the Galactic
66: center where $V_{\pi}$ is the radial component in the Galactic plane
67: directed positive away from the center, $V_{\phi}$ is the tangential
68: component positive in the direction of Galactic rotation and $V_{z}$ is
69: the vertical component directed positive in the direction of the north
70: Galactic pole.
71:
72: The average of each of these components for the entire sample was removed
73: before forming the following six moments: $\overline{V_{\pi}V_{\pi}},
74: \overline{V_{\phi}V_{\phi}}, \overline{V_{z}V_{z}},
75: \overline{V_{\pi}V_{\phi}}, \overline{V_{\pi}V_{z}}$, and
76: $\overline{V_{\phi}V_{z}}$. These six moments are the components of the
77: symmetrical velocity dispersion tensor. The eigenvalues of the matrix
78: were then determined in the standard way and are given below. The units
79: of the eigenvalues denoted by $\sigma$ are km~sec$^{-1}$ and the one sigma
80: uncertainties throughout were computed by the bootstrap method.
81:
82: \begin{center}
83: Clusters with [Fe/H]$<-1.00$\ (n=38)
84: \end{center}
85: \vspace{-0.3 in}
86: \begin{eqnarray}
87: \sigma_{a} = 161^{+18.3}_{-15.0} & l = 338^{+10.5}_{-6.65} & b =
88: -5.65^{+13.6}_{-15.0} \nonumber \\
89: \sigma_{b} = 120^{+9.44}_{-15.5} & l = 87.2^{+155.}_{-30.0.} & b =
90: -73.5^{+27.9}_{-6.63} \\
91: \sigma_{c} = 95.6^{+2.66}_{-17.6} & l = 66.1^{+11.2}_{-10.8} & b =
92: 15.4^{+22.2}_{-21.1} \nonumber \\
93: & \overline{V_{\phi}} = 59.6^{+13.5}_{-17.9} \nonumber
94: \end{eqnarray}
95:
96: \noindent
97: As a check on the sensitivity of the above result to the [Fe/H] cutoff, we
98: repeated the procedure for the 26 clusters with [Fe/H]$<-1.5$ and find:
99:
100: \begin{center}
101: Clusters with [Fe/H]$<-1.50$\ (n=26)
102: \end{center}
103: \vspace{-0.3 in}
104: \begin{eqnarray}
105: \sigma_{a} = 172^{+17.9}_{-19.2} & l = 330^{+11.4}_{-8.11} & b =
106: -6.31^{+14.9}_{-14.5} \nonumber \\
107: \sigma_{b} = 122^{+8.45}_{-27.7} & l = 179^{+91.7}_{-112.} & b =
108: -82.8^{+32.6}_{+6.58} \\
109: \sigma_{c} = 94.7^{+5.76}_{-27.8} & l = 60.8^{+12.1}_{-12.1} & b =
110: -3.48^{+30.1}_{-21.3} \nonumber \\
111: & \overline{V_{\phi}} = 49.3^{+23.7}_{-28.1} \nonumber
112: \end{eqnarray}
113:
114: In both of the above solutions the mean streaming motions in the $\pi$ and $z$
115: directions differed from zero by less than 1.5 sigma. Also the above values of
116: $\overline{V_{\phi}}$ are independent of distance from the Galactic center.
117:
118: Asymmetric error ellipsoids in the components of space motion can be expected
119: due to relatively small errors in radial velocity and relatively large errors
120: in proper motion. In order to ensure that such effects were not unduly
121: influencing solution (1) the errors in radial velocity, proper motion and
122: distance given by Dinescu et al.\ (1999) were used as dispersions in assumed
123: gaussian distributions and solutions were performed on space motions
124: determined by the propagation of randomly chosen uncertainties in each of the
125: above quantities. The mean and standard deviation of the major axis from 100
126: trials with 38 clusters is $\sigma_{a} = 167 \pm 7.98$ towards $l = 332 \pm
127: 6.28$ and $b = -5.94 \pm 7.64$. Results for the other two axes are $\sigma_
128: {b} = 122 \pm 10.9$ and $\sigma_{c} = 108 \pm 10.1$. This independent
129: assessment of uncertainties thus strengthens the case for the generally
130: $prolate$ velocity ellipsoid given by solution (1).
131:
132: An examination of the spatial distribution of the clusters from solution
133: (1) shows that only 7 of the 38 clusters were in the hemisphere
134: beyond the Galactic center. In order to assess the effect of this spatial
135: imbalance on our result, kinematic
136: solutions were made using the above 7 clusters combined with 7 independently
137: chosen clusters from the hemisphere containing the sun. Thirty solutions
138: were made with 14 clusters each and the median results for each aspect of
139: the major axis are: $\sigma_{a} = 166^{+20}_{-9},\ \ l = 348^{+8}_{-7},\ \ $
140: and $b = -13^{+29}_{-8}$.
141: While the one sigma uncertainties of solution (1) do overlap with the above
142: result, perhaps the most important implication of this exercise is a clear
143: need for more cluster space motion determinations.
144:
145: \section{The Spatial Distribution of the Globular Clusters}
146:
147: The spatial distribution of the outlying satellites and the globular
148: clusters was determined in H2000. There it was found that the 99 metal
149: poor globular clusters ([Fe/H]$\leq-1.0$) formed an oblate distribution
150: with minor axis directed towards $l=307^{\circ}$,$b=-3.3^{\circ}$ quite
151: different from the short axis of the distribution of the satellite
152: galaxies. The result from the kinematics described above prompted a
153: re-examination of this result. It is believed that the pole of the orbit
154: of the Sagittarius system is highly inclined to that of the remaining
155: outlying satellite galaxies (Lynden-Bell \& Lynden-Bell 1995). As well
156: these authors tentatively assign the 6 clusters M54, NGC2419, Arp 2, Pal 2,
157: Terzan 7 and Terzan 8 to the Sagittarius stream. Given that the present
158: orbit of
159: Sagittarius is probably not the one that it had at formation (c.f. Zhao
160: 1998, also Johnston et al. 2002) we removed the above clusters from the
161: original sample and performed new solutions. (Note that the Harris (1996)
162: web-based compilation contains 147 clusters of which 99 have [Fe/H]$\leq-
163: 1.0$. Culling 5 clusters from this sub-group still leaves us with 95\% of the
164: original sample. Furthermore, none of these five clusters was included in
165: the above kinematic solutions.) The analysis procedure followed here is
166: slightly different from that in H2000. Here the tensor components
167: $\overline{x_ix_i}, \overline{x_iy_i}$ etc were calculated both without
168: weighting and with weighting by 1/d$_i$ where the $x_i$, $y_i$, and $z_i$
169: are the projections of d$_i$, the distance from the Galactic center
170: (R$_0=8$ kpc assumed). Both the weighted and unweighted solutions are
171: given since the true distribution probably lies somewhere in between.
172: The eigenvalues of the resulting matrix are denoted e$_a$, e$_b$, and
173: e$_c$ and have units of kpc.
174:
175: \begin{center}
176: Cluster Sample with [Fe/H]$\leq-1.0$\ (weighted)\ (n=94)
177: \end{center}
178: \vspace{-0.3 in}
179: \begin{eqnarray}
180: e_{a} = 5.30^{+1.77}_{-1.11} & l = 246^{+33.6}_{-11.1} & b =
181: -40.0^{+24.0}_{-18.2} \nonumber \\
182: e_{b} = 5.04^{+1.31}_{-0.742} & l = 44.1^{+19.2}_{-12.1} & b =
183: 48.0^{+12.4}_{-28.3} \\
184: e_{c} = 3.22^{+0.241}_{-0.464} & l = 326^{+13.9}_{-12.2} & b =
185: -10.8^{+7.22}_{-13.5} \nonumber \\
186: & c/a = 0.61^{+0.13}_{-0.14} \nonumber
187: \end{eqnarray}
188:
189: \begin{center}
190: Cluster Sample with [Fe/H]$\leq-1.0$\ (unweighted)\ (n=94)
191: \end{center}
192: \vspace{-0.3 in}
193: \begin{eqnarray}
194: e_{a} = 19.9^{+4.03}_{-5.32} & l = 236^{+8.20}_{-24.8} & b =
195: -63.2^{+25.6}_{-11.6} \nonumber \\
196: e_{b} = 15.5^{+2.90}_{-5.25} & l = 52.3^{+21.3}_{-8.04} & b =
197: -26.8^{+15.8}_{-22.6} \\
198: e_{c} = 8.18^{+0.306}_{-2.01} & l = 323^{+11.7}_{-9.19} & b =
199: 1.53^{+4.06}_{-18.6} \nonumber \\
200: & c/a = 0.41^{+0.093}_{-0.091} \nonumber
201: \end{eqnarray}
202:
203: \noindent
204: As a check on the robustness of the result, solutions were also made for the
205: 60
206: remaining clusters with [Fe/H]$\leq-1.5$.
207:
208:
209: \begin{center}
210: Cluster Sample with [Fe/H]$\leq-1.5$\ (weighted)\ (n=60)
211: \end{center}
212: \vspace{-0.3 in}
213: \begin{eqnarray}
214: e_{a} = 5.76^{+1.66}_{-0.721} & l = 115^{+66.3}_{-36.5} & b =
215: -61.3^{+24.2}_{-4.81} \nonumber \\
216: e_{b} = 5.34^{+1.57}_{-1.07} & l = 60.2^{+22.9}_{-20.0} & b =
217: 17.3^{+29.0}_{-23.2} \\
218: e_{c} = 3.67^{+0.249}_{-0.511} & l = 337^{+10.6}_{-39.4} & b =
219: -22.1^{+12.7}_{-10.3} \nonumber \\
220: & c/a = 0.64^{+0.061}_{-0.15} \nonumber
221: \end{eqnarray}
222:
223: \begin{center}
224: Cluster Sample with [Fe/H]$\leq-1.5$\ (unweighted)\ (n=60)
225: \end{center}
226: \vspace{-0.30 in}
227: \begin{eqnarray}
228: e_{a} = 19.4^{+4.67}_{-7.93} & l = 238^{+10.0}_{-33.8} & b =
229: -48.5^{+10.8}_{-2.93} \nonumber \\
230: e_{b} = 15.1^{+4.12}_{-5.89} & l = 66.2^{+31.9}_{-8.99} & b =
231: -41.3^{+11.8}_{-3.54} \\
232: e_{c} = 8.29^{+0.112}_{-2.11} & l = 333^{+18.2}_{-27.5} & b =
233: -3.90^{+8.86}_{-23.0} \nonumber \\
234: & c/a = 0.43^{+0.13}_{-0.11} \nonumber
235: \end{eqnarray}
236:
237: \noindent
238: We note that for both samples the short axis of the (nearly oblate)
239: cluster distribution is very similar to the $\it{long}$ axis of the
240: kinematic solution. In order to insure that our result is not affected by
241: incompleteness or uncertain absorption corrections, solutions were also
242: performed for only those
243: clusters whose absolute Galactic latitudes were larger than 10$^{\circ}$.
244:
245: \begin{center}
246: Cluster Sample with [Fe/H]$\leq-1.0$\ and $\left|b\right|>10$\ (weighted)\
247: (n=66)
248: \end{center}
249: \vspace{-0.3 in}
250: \begin{eqnarray}
251: e_{a} = 7.93^{+2.32}_{-0.618} & l = 192^{+54.2}_{-76.1} & b =
252: -76.4^{+33.2}_{+0.863} \nonumber \\
253: e_{b} = 7.11^{+0.568}_{-1.55} & l = 58.6^{+21.9}_{-11.9} & b =
254: -9.44^{+29.4}_{-33.0} \\
255: e_{c} = 4.64^{+0.268}_{-0.591} & l = 327^{+14.4}_{-11.3} & b =
256: -9.65^{+6.99}_{-12.6} \nonumber \\
257: & c/a = 0.58^{+0.011}_{-0.13} \nonumber
258: \end{eqnarray}
259:
260: \begin{center}
261: Cluster Sample with [Fe/H]$\leq-1.0$\ and $\left|b\right|>10$\ (unweighted)\
262: (n=66)
263: \end{center}
264: \vspace{-0.30 in}
265: \begin{eqnarray}
266: e_{a} = 23.6^{+6.37}_{-6.36} & l = 236^{+34.1}_{-20.1} & b =
267: -65.1^{+22.1}_{-7.44} \nonumber \\
268: e_{b} = 17.8^{+2.33}_{-4.45} & l = 50.3^{+15.3}_{-10.7} & b =
269: -24.8^{+21.1}_{-20.0} \\
270: e_{c} = 9.43^{+0.372}_{-1.97} & l = 321^{+9.03}_{-9.78} & b =
271: 2.05^{+6.38}_{-11.1} \nonumber \\
272: & c/a = 0.40^{+0.071}_{-0.092} \nonumber
273: \end{eqnarray}
274:
275: The quoted uncertainties in the above solutions were determined by the
276: bootstrap method. It is instructive to consider the effects on the solution
277: of errors in the individual distances to each cluster. Randomly chosen errors
278: drawn from a gaussian distribution with a sigma of $10\%$ of each distance
279: were propagated through and the mean and standard deviation of the minor axis
280: of the spatial distribution from 100 trials with 66 clusters was found to be
281: $e_{c} = 4.67 \pm 0.0949$ towards $l = 327 \pm 1.28$ and $b = -9.91 \pm 1.02$.
282: The amplitudes of the other two axes are $e_{a} = 7.98 \pm 0.146$ and $e_{b}
283: = 7.11 \pm 0.148$. The good agreement with solution (7) strengthens the case
284: for the generally $oblate$ spatial distribution given by solution (7).
285:
286:
287: For illustration, the distribution of clusters in this new coordinate
288: system (solution (7)) is shown in Fig. 1 where x$^{\prime}$ are projections of
289: Galactocentric distance along the major axis and z$^{\prime}$ are projections
290: along the minor axis. As summarized in Table 1, the short axes of the spatial
291: distributions are remarkably similar to the major axes of the kinematic
292: solutions of the previous section. Furthermore, the minor axis of the
293: satellite galaxy distribution is also quite similar.
294:
295: In H2000 the spatial distribution of the 34 most metal rich globular
296: clusters ([Fe/H]$\geq-0.7$) was also determined. The distribution was
297: found to be triaxial with the long axis directed slightly off ($\sim17^
298: {\circ}$) the
299: sun-center line and the short axis now pointing to high Galactic latitude.
300: We repeat the process below but with one less cluster (Terzan 7).
301:
302: \begin{center}
303: Cluster Sample with [Fe/H]$\geq-0.7$\ (weighted)\ (n=33)
304: \end{center}
305: \vspace{-0.3 in}
306: \begin{eqnarray}
307: e_{a} = 2.02^{+0.431}_{-0.251} & l = 338^{+6.19}_{-29.0} & b =
308: -6.09^{+4.63}_{-2.92} \nonumber \\
309: e_{b} = 1.51^{+0.273}_{-0.228} & l = 68.1^{+14.5}_{-12.7} & b =
310: 0.430^{+9.99}_{-9.13} \\
311: e_{c} = 0.761^{+0.0968}_{-0.174} & l = 334^{+65.2}_{-60.2} & b =
312: 83.9^{+0.907}_{-9.95} \nonumber \\
313: & c/a = 0.38^{+0.049}_{-0.094} \nonumber
314: \end{eqnarray}
315:
316: \begin{center}
317: Cluster Sample with [Fe/H]$\geq-0.7$\ (unweighted)\ (n=33)
318: \end{center}
319: \vspace{-0.30 in}
320: \begin{eqnarray}
321: e_{a} = 3.78^{+1.00}_{-1.29} & l = 335^{+4.86}_{-11.9} & b =
322: -8.50^{+7.51}_{-2.76} \nonumber \\
323: e_{b} = 2.29^{+0.336}_{-0.472} & l = 65.7^{+6.05}_{-9.34} & b =
324: -3.51^{+12.9}_{-10.1} \\
325: e_{c} = 1.11^{+0.0.0711}_{-0.413} & l = 358^{+37.5}_{-85.6} & b =
326: -80.8^{+0.623}_{-9.86} \nonumber \\
327: & c/a = 0.29^{+0.088}_{-0.12} \nonumber
328: \end{eqnarray}
329:
330: \noindent
331: A triaxial distribution is still present, but now the major axis is
332: $\sim26^{\circ}$ off of the sun-center line while the minor axis is directed
333: to high latitude. Again it seems reasonable to suggest that we are observing a
334: globular cluster counterpart to the `COBE' bar (c.f. Binney \& Merrifield,
335: 1998, p. 616). Should future work confirm the remarkably
336: close alignment between
337: the major axis of the spatial distribution and the major kinematic
338: dispersion axis (see Table 1), an intimate connection between halo (metal poor
339: clusters) and bulge (metal rich clusters) would be implied. One possible
340: argument against such a connection
341: is the recent suggestion that bars may be less common at high redshift than
342: at z=0 (van den Bergh et al. 2002).
343: Further elucidation will come from the determination of the space motions
344: for these metal rich clusters.
345:
346:
347: \section{Discussion}
348:
349: Both the velocity ellipsoid and the spatial distribution of the most metal
350: poor Galactic globular clusters appear to be anisotropic. It is
351: interesting that the major axis of the velocity ellipsoid is directed
352: towards low Galactic latitude, consistent with the direction of the short
353: axis of the spatial distribution and not far from the pole of the outlying
354: satellite galaxy distribution. A quantitative comparison for the
355: different samples is given in Table 1 where for reference we take the
356: major axis of the kinematic solution (1).
357:
358: \begin{deluxetable}{ccccccc}
359: \footnotesize
360: \tablenum{1}
361: \tablecolumns{7}
362: \tablecaption{Comparison of the Axes of the Various Samples}
363: \tablehead{
364: \colhead{Sample} &
365: \colhead{Number} &
366: \colhead{Solution} &
367: \colhead{ Axis } &
368: \multicolumn{2}{c}{Galactic Coordinates} &
369: \colhead{$\Delta\theta$\tablenotemark{a}}
370: \\
371: \colhead{} &
372: \colhead{n} &
373: \colhead{number} &
374: \colhead{Major/Minor} &
375: \colhead{~~~~~$l$} &
376: \colhead{~~~~~$b$} &
377: \colhead{}
378: }
379: \startdata
380: Velocity ellipsoid & 38 & 1 & Major
381: &~$338^{+10.5}_{-6.65}$&~$-5.65^{+13.6}_{-15.0}$& $0^{\circ}$ \nl
382: [Fe/H]$\leq-1.0$ & & & & & & \nl
383: \\ Velocity ellipsoid & 26 & 2 & Major
384: &~$330^{+11.4}_{-8.11}$&~$-6.31^{+14.9}_{-14.5}$&$8.0^{\circ}\pm13$ \nl
385: [Fe/H]$\leq-1.5$ & & & & & & \nl
386: \\ Clusters & 94 & 3 & Minor
387: &~$326^{+13.9}_{-12.2}$&~$-10.8^{+7.22}_{-13.5}$&$13^{\circ}\pm16$ \nl
388: [Fe/H]$\leq-1.0$ & & & & & & \nl
389: \\ Clusters & 60 & 5 & Minor
390: &~$337^{+10.6}_{-39.4}$&~$-22.1^{+12.7}_{-10.3}$&$16^{\circ}\pm18$ \nl
391: [Fe/H]$\leq-1.5$ & & & & & & \nl
392: \\ Clusters & 66 & 7 & Minor
393: &~$327^{+14.4}_{-11.3}$&~$-9.65^{+6.99}_{-12.6}$&$11^{\circ}\pm16$ \nl
394: [Fe/H]$\leq-1.0$, $|b|>10^{\circ}$ & & & & & & \nl
395: \\ Outlying Satellites & 10 & H2000 & Minor
396: &~$336^{+8.28}_{-3.79}$&~$11.1^{+3.89}_{-5.05}$&$17^{\circ}\pm15$ \nl
397: \\ Clusters & 33 & 9 & Major
398: &~$338^{+6.19}_{-29.0}$&~$-6.09^{+4.63}_{-2.92}$&$0.44^{\circ}\pm15$ \nl
399: [Fe/H]$\geq-0.7$ & & & & & & \nl
400: \enddata
401:
402: \tablenotetext{a}{$\Delta\theta$ is the angular difference in degrees
403: between axes of the first sample and each of the subsequent samples. }
404: \end{deluxetable}
405:
406: There would appear to be no straight forward explanation for the
407: observations above. If the gravitational potential was spherical, one can
408: argue that the direction of highest velocity dispersion would also be that of
409: the longest spatial axis and vice versa. The above result then apparently
410: rules out a spherical potential but would be consistent with one that was
411: flattened in the direction of highest velocity dispersion. Recalling that
412: the age-metallicity relation for the
413: most metal poor globular clusters is extremely steep (vandenBerg 2000), a
414: naive picture is that an initially flattened filament underwent a
415: rapid collapse tranverse to the long axis during which time satellite
416: galaxies, the metal poor clusters, and the
417: triaxial structure discussed in H2000 were formed. Simultaneously the
418: Galaxy was being assembled by a slower collapse along the filament during
419: which time most of its angular momentum was being induced and/or
420: redistributed. While triaxial
421: structures are formed in cold dark matter simulations, the scenario outlined
422: above may be more consistent with a warm dark matter theory for
423: structure formation where the lowest mass objects form from the
424: fragmentation of caustics (c.f. Bode, Ostriker, \& Turok 2001).
425:
426: \acknowledgments
427:
428: The author wishes to thank Dr.\ Dana Dinescu for providing the latest
429: compilation of cluster space motions in convenient form, and Ray Carlberg,
430: Scott Tremaine, and Sidney van den Bergh for their helpful comments on an
431: earlier draft. He also wishes to acknowledge financial support from an NSERC
432: of Canada operating grant.
433:
434: \begin{references}
435:
436: \reference{}
437: Binney, J., \& Merrifield, M. 1998, Galactic Astronomy, (Princeton: PUP)
438:
439: \reference{}
440: Bode, P., Ostriker, J.P. \& Turok, N. 2001, \apj, 556, 93
441:
442: \reference{}
443: Dinescu, D.I., Girard, T.M., \& van Altena, W.F. 1999, \aj, 117, 1792
444:
445: \reference{}
446: Harris, W.E. 1996, \aj, 112, 1487
447:
448: \reference{}
449: Hartwick, F.D.A. 2000, \aj, 119, 2248 (H2000)
450:
451: \reference{}
452: Johnston, K.V., Spergel, D.N., \& Haydn, C. 2002, \apj, 570, 656
453:
454: \reference{}
455: Kinman, T. D. 1959, \mnras, 119, 538
456:
457: \reference{}
458: Lynden-Bell, D., \& Lynden-Bell, R.M. 1995, \mnras, 275, 429
459:
460: \reference{}
461: vandenBerg, D. A. 2000, \apjs, 129, 315
462:
463: \reference{}
464: van den Bergh, S., Abraham, R.G.,Whyte, L.F., Merrifield, M.R.,
465: Eskridge, P.B., Frogel, J.A. \& Pogge, R. 2002, \aj, 123, 2913
466:
467: \reference{}
468: Zhao, H. 1998, \apj, 500, L149
469:
470: \reference{}
471: Zinn, R. 1985, \apj, 293, 424
472: \end{references}
473:
474: \clearpage
475:
476: \begin{figure}
477: \plotone{f1.eps}
478: \figcaption{
479: The spatial distribution of 66 globular clusters with [Fe/H]$\leq-1.0$
480: and $|b|>10^{\circ}$ in coordinates transformed to the major (x$^{\prime}$) and
481: minor (z$^{\prime}$) axes of solution (7) illustrating the flattening of the
482: distribution. For reference the minor axis is directed to $l=327^{\circ}$,
483: $b=-9.65^{\circ}$.}
484: \end{figure}
485:
486: \end{document}
487:
488: