THE PROPER MOTION OF NGC 6522 IN BAADEÏS WINDOW1

Transcription

1 THE ASTRONOMICAL JOURNAL, 115:1476È1482, 1998 April ( The Amerian Astronomial Soiety. All rights reserved. Printed in U.S.A. THE PROPER MOTION OF NGC 6522 IN BAADEÏS WINDOW1 DONALD M. TERNDRUP, PIOTR POPOWSKI, AND ANDREW GOULD2 Department o Astronomy, Ohio State University, 174 West 18th Avenue, Columbus, OH 4321; terndrup=astronomy.ohio-state.edu, popowski=astronomy.ohio-state.edu, gould=astronomy.ohio-state.edu R. MICHAEL RICH Department o Astronomy, Pupin Laboratories, Columbia University, 538 West 12th Street, New York, NY 127; rmr=astro.olumbia.edu AND ELAINE M. SADLER Shool o Physis, University o Sydney, Sydney, NSW 26, Australia; ems=physis.usyd.edu.au Reeived 1997 Otober 3; revised 1997 Deember 16 ABSTRACT We have deteted seven stars with a ommon proper motion that are loated within 2@.5 o the globular luster NGC 6522 in the BaadeÏs window Ðeld o the Galati bulge. We argue that these stars are members o the luster, and derive a weighted mean proper motion and helioentri radial veloity o k6 mas yr~1, mas yr~1 and km s~1. We rederive the l \ 1.4 ^.2 k6 \[6.2 ^.2 v6 \[28.5 ^ 6.5 distane to NGC 6522 [(.91 b ^.4)R, where R is the Galatoentri distane] and metalliity ([Fe/H] \[1.28 ^.12), making use o reent revisions in the oreground extintion toward the luster (A \ 1.42 ^.5). We Ðnd the spatial veloity o the luster and onlude that the luster stays lose to the Galati enter, and may have experiened signiðant bulge/disk shoking during its lietime. Key words: astrometry È Galaxy: abundanes È globular lusters: individual (NGC 6522) È stars: kinematis 1. INTRODUCTION Three-dimensional spae motions o individual stars and stellar systems throughout the Galaxy are an important probe o the Milky WayÏs gravitational potential and o the kinematis o di erent populations within it. Most o what we know, or example, about the disk and halo o the Galaxy has ome rom studies o the radial veloities and proper motions o stars in the loal volume o spae near the Sun. In reent years, this has been steadily extended to samples throughout the Galaxy, partiularly or globular lusters (Cudworth & Hanson 1993; Dauphole et al. 1996) and the Large Magellani Cloud and other Galati satellites (see, e.g., Jones, Klemola, & Lin 1994; Shweitzer & Cudworth 1996; Ibata et al. 1997). A potentially important sample or in situ studies o the Galati bulge is the proper-motion survey o Spaenhauer, Jones, & Whitord (1992, hereater SJW). This survey ontains relative proper motions o 427 stars in the BaadeÏs window Ðeld o the nulear bulge at (l, b)\ (1., [3.9), whih samples stellar orbits relatively lose to the Galati enter (minimum line-o-sight distane o B55 p). The proper motions presented by SJW are with respet to the mean motion o the entire sample (i.e., they set k6 l 4, k6 b 4 ), sine no extragalati reerene rame is available. We have been onduting an extensive photometri and spetrosopi ollow-up o the SJW survey. In our Ðrst paper (Terndrup, Sadler, & Rih 1995), we presented methods o analysis and demonstrated that the sample ontains a onsiderable number o stars loated in ront o the bulge. In the seond paper (Sadler, Rih, & Terndrup 1996), ÈÈÈÈÈÈÈÈÈÈÈÈÈÈÈ 1 Based in part on data obtained at the Anglo-Australian Observatory and at Cerro Tololo Inter-Amerian Observatory, National Optial Astronomy Observatories, operated by the Assoiation o Universities or Researh in Astronomy, In., under ooperative agreement with the National Siene Foundation. 2 Alred P. Sloan Fellow we derived individual metalliities and distanes to the SJW stars. In preparing or a ull analysis o the spae motions in BaadeÏs window (Rih, Terndrup, & Sadler 1998), we have disovered that a number o stars with proper motions in BaadeÏs window are likely to be members o the globular luster NGC In 2, we desribe the proedure by whih the stars are identiðed as luster members, measure the proper motion o the luster relative to the mean proper motion o the SJW sample, and estimate the lusterïs radial veloity. In 3, we analyze the luster olor-magnitude diagram (CMD) and derive estimates or the luster metalliity and distane. In 4, we estimate the spae motion o the luster and disuss its orbit. 2. IDENTIFICATION OF CLUSTER MEMBERS We begin by seleting the 46 stars that have radial veloities (Terndrup et al. 1995) out o the 427 stars in the SJW survey.3 We Ðrst demonstrate that members o NGC 6522 are present in the SJW survey by noting that there are several stars with a ommon proper motion that have radial veloities near the value previously determined or the luster. The radial veloity o NGC 6522 has been reported several times in the literature. Rih (199) obtained the radial veloity o two stars near the luster enter rom spetra o resolution 1.9 A ; the average helioentri veloity o these was v \[25.6 ^ 1.5 km s~1, idential to the value o [25 ^ 16 km s~1 derived rom Fabry-Perot observations at Ha by Smith, Hesser, & Shawl (1976). Averaging a variety o observations, inluding the Smith et al. (1976) value, Webbink (1981) obtained v \]8^16 km s~1. As Webbink himsel noted, however, the veloities he inluded in the average or NGC 6522 ormed a bimodal distribution; it is thereore likely that severaulge stars were ÈÈÈÈÈÈÈÈÈÈÈÈÈÈÈ 3 SJW tabulated 429 stars, but listed the stars Arp and 4-27 twie.

2 PROPER MOTION OF NGC inluded in the omputation. In his ompilation o globular luster veloities, Zinn (1985) repeats the Webbink (1981) value. Pryor & Meylan (1993) quote the radial veloity as [1.4 ^ 1.5 km s~1 rom an apparently unpublished paper. Reently, Rutledge et al. (1997) obtained Ca II spetra o 18 stars near the luster and derived v \[18.3 ^ 9.3 km s~1, where the unertainty is the standard error o the mean. In the top panel o Figure 1, we display the propermotion vetor point diagram or the omplete SJW survey, while in the bottom panel we show those stars loated within 2@.5 o the luster enter with radial veloities statistially onsistent at the 1 p level with the Smith et al. (1976) value o [25 ^ 16 km s~1. The error bars in the lower let-hand orner o eah panel in Figure 1 show the mean error in proper motion or the stars in the sample. In the bottom panel, one lumping o points is present at (k, k ) D (1.5, [6) mas yr~1. As we will show, this is the most likely proper motion o the luster, beause stars near this lump lie on the giant branh or NGC 6522 and have absorption-line strengths unlike most o the stars in the bulge; other possible lumps o stars in Figure 1 do not yield this a posteriori onsisteny with the luster CMD and metalliity. To identiy luster members, we employ an iterative sheme that searhes or stars in the SJW sample that are simultaneously within 1.5 p o the mean luster motion (k6 in eah proper-motion omponent and in l,, k6 b,, v6 ) radial veloity. For example, or helioentri radial veloity omponent, we demand that o v [ v6 i o ¹ 1.5 p i, (1) where v is the veloity o an individual star, v6 is the mean i veloity o luster members, p2 4 v2]p2(v), (2) i i v is the measurement error, and p (v) is the estimated i satter aused by the internal veloity dispersion o the luster. Similar expressions are used or k and k. We adopt p (v) \ 6.7 km s~1 (Pryor & Meylan 1993). For a distane to NGC 6522 o 7.1 kp (see below), this orresponds to p (k ) \ p (k ) \.2 mas yr~1, whih we adopt or the dis- persion l in eah b omponent o proper motion. (The results o this proedure are not very sensitive to the hoie o p, beause v º p or most o the sample.) We began i the iteration with a ew stars near the lump in the bottom panel o Figure 1. Then, at eah step the list o stars near the mean luster motion was used to reompute the weighted means (k6 with weights in eah l,, k6 b,, v6 ), 1/p i 2 omponent. This proess quikly onverged on seven stars, whih are listed in Table 1. The star names in the Ðrst µ b (mas yr 1 ) µ b (mas yr 1 ) µ l (mas yr 1 ) FIG. 1.ÈProper-motion vetor point diagrams or all stars in the BaadeÏs window survey (SJW; top) and those stars loated within 2@.5 o the luster enter that have radial veloities near that o NGC 6522 (bottom). The units are milliarseonds per year. The error bar in the lower let-hand orner o eah panel shows the mean error o measurement in k and k, plotted as though they were independent. Beause the proper motions l in b l and b were obtained rom the proper motion in right asension and delination through a oordinate rotation, the errors in k and k are in at orrelated. olumn are rom Arp (1965). Columns (2)È(7) show the radial veloity and photometry (Terndrup et al. 1995). The next our olumns ontain the proper motion and errors rom SJW, where the units are mas yr~1. The last olumn shows the angular distane o eah star in arminutes rom the enter o NGC TABLE 1 PROBABLE MEMBERS OF NGC 6522 Star v h v(v h ) v( ) [I v( [I) k l v(k l ) k b v(k b ) d (1) (2) (3) (4) (5) (6) (7) (8) (9) (1) (11) (12) [ [ [ [ [ [ [ [ [ [ [ [ [ [

3 1478 TERNDRUP ET AL. ol. 115 Membership probability (%) Membership probability (%) Distane rom luster enter (armin) FIG. 2.ÈPerentage membership probabilities, as deðned in the text, plotted against the distane rom the enter o NGC 6522 (top), and against the magnitude (bottom). The stars that we identiðed as luster members have P º 85%. At this point we examined the errors in proper motion and radial veloity to onðrm that they were estimated orretly, or i the seleted stars truly are members o the luster the satter in projeted motion should be almost entirely due to observational error. ia a s2 test, we determined that the satter in proper motions about the mean value is onsistent with the errors, but the errors or the radial veloities are too large by a ator o D1.7. The reason or this is not ompletely lear. As disussed by Terndrup et al. (1995), the radial veloity errors were estimated rom the width o the peak in the ross-orrelation between the bulge stars and a set o radial veloity templates; interomparisons between multiple measures allowed them to onðrm that the error estimates were, in general, approximately right. The possible luster members, however, happen to be among the most metal-weak o the stars in the SJW sample (see below), whih ould have led to an overestimate o the error beause o the resulting mismath between these stars and the veloity standards. An examination o the estimated errors as a untion o metalliity (Sadler et al. 1996) reveals that this explanation is plausible: the mean estimated error in radial veloity or stars with [Fe/H] \ [.7 is 25 km s~1, ompared with 15 km s~1 or stars with [Fe/H] [ [.7. We then saled the radial veloity errors downward by a ator o 1.7 (but not below a reasonable minimum value o 7kms~1) and repeated the iterations, obtaining the same list o andidate members o NGC The weighted mean luster motion and standard errors o the mean rom this sample are k6 \]1.2 ^.2 mas yr~1, l, k6 \[6. ^.2 mas yr~1, b, v6 \[28.5 ^ 6.5 km s~1. (3) The radial veloity we derive is in statistial agreement with the value o [18.3 ^ 9.3 km s~1 rom spetra o 18 stars near the luster reported by Rutledge et al. (1997). Finally, in order to explore the sensitivity o our seletion method, we ompute membership probabilities or the SJW stars with radial veloities, assuming a Gaussian distribution or the luster and Ðeld (mostly the bulge) and taking into aount the errors in the individual measurements (see Dinesu et al. 1996). We write the mean apparent motion o the Ðeld as Mk6 and the dispersion as l,, k6 b,, v6 N Mp (k l ), p (k ), p (v)n. Denoting with the subsript the equivalent quantities b or the luster, determined above, we deðne the membership probability or star i as P \ o i,, (4) i o ] o i, i, where the densities o and o o luster and Ðeld stars are i, i, represented as the Gaussian distributions N o \ i, (2n)3@2p C (k )p (k )p (v) i l i b i D ] exp [ (k l,i [ k6 l, )2 [ (k b,i [ k6 b, )2 [ (v i [ v6 )2, (5) 2p2(k ) 2p2(k ) 2p2(v) i l i b i N o \ i, (2n)3@2p (k )p (k )p (v) l b ] exp C [ (k l,i [ k6 l, )2 2p 2(k l ) [ (k b,i [ k6 b, )2 2p 2(k b ) D [ (v i [ v6 )2. (6) 2p2(v) Note that in the expression or the luster density untion we ombined the errors o the individual stars and the adopted luster dispersion as in equation (2). The values or the Ðeld were derived rom the SJW stars with veloities exluding the seven stars in Table 1, and were k6 l, \.4 mas yr~1, p (k ) \ 3.1 mas yr~1, k6 mas yr~1,, \.18 p (k ) \ 2.7 mas yr~1, v6 km s~1, and b \[2.1 p (v) \ 15 km s~1. As opposed to the luster members, the Ðeld starsï dispersions are dominated by the intrinsi veloity dispersions and not observational errors. Figure 2 displays the membership probabilities, expressed as perentages, against the distane rom the enter o NGC 6522 (top) and the magnitude (bottom). The stars in Table 1 are those seven stars with the highest membership probabilities; all seven have P º 85%. There are only three other stars with P º 1%, and most o the sample has P very nearly zero. (The SJW sample ontains almost no stars within 1@ o the luster, beause this region was deliberately avoided in their survey.) The stars with the highest membership probabilities are all within 2@.5 o the luster enter; this is well within the tidal radius o the luster, whih is not aurately measured but is on the order o 1@È15@ (Peterson & Reed 1987; Harris 1996).

4 No. 4, 1998 PROPER MOTION OF NGC DISTANCE TO NGC 6522 To support our laim that we have ound members o NGC 6522 in the SJW survey, we now examine photometry and line-strength indexes (Terndrup et al. 1995) or the stars in Table 1. We also derive the distane to NGC 6522, needed or the onversion o proper motion into a spae veloity, rom its CMD and by diret omparison o its horizontal-branh magnitude with that o RR Lyrae stars in BaadeÏs window. In this proess, we obtain a photometri metalliity estimate or omparison with reent results rom spetrosopy. The distane to NGC 6522 is not known aurately: although there are several RR Lyrae stars near the luster, it is diiult to determine whih are members and whih are in the bulge (Blano 1984; Walker & Mak 1986; Walker & Terndrup 1991; Carney et al. 1995). Consequently, the only distane estimates have ome rom analysis o the lusterïs CMD, and these are sensitive to the assumed metalliity o and extintion toward the luster. For example, Terndrup & Walker (1994) derived [Fe/H] \[1.6 ^.2 and (m [ M) \ 14.8 ^.3 (d \ 9.1 `1.4 kp) rom omparison ~1.2 o the luster CMD with the giant branhes o globular lusters in Da Costa & Armandro (199). This derivation was or E( [I) \.65 ^.7, and the Da Costa & Armandro sequenes were tied to the Lee, Demarque, & Zinn (199) alibration o RR Lyrae absolute magnitudes o M (RR) \.82 ].17[Fe/H]. The metalliity that Tern- drup & Walker (1994) derived was in agreement within the errors to the Zinn & West (1984) value, [Fe/H] \[1.44, rom integrated olors. Nearly simultaneously with Terndrup & Walker (1994), Barbuy, Ortolani, & Bia (1994) presented their own CMD o the luster and derived extintion values and metalliity by omparison with the CMD o NGC 6752, whih has [Fe/H] \[1.54 (Da Costa & Armandro 199 and reerenes therein). They Ðnd E(B[ ) \.55 ^.5, whih orresponds to E( [I) \.68 ^.6 (i.e., lose to the Terndrup & Walker 1994 value) and a distane modulus o (m [ M) \ (d \ 6.2 kp). They assume a horizontal- branh luminosity o M \].6 and obtain a very approximate metalliity estimate o [Fe/H] B [1.. Together, these two estimates imply that they were using a horizontal-branh luminosity sale that is slightly (.5 mag) more luminous than the Lee et al. (199) sale employed by Terndrup & Walker (1994). Note, however, that the di erene in horizontal-branh luminosity sale is ar too small to explain the.8 mag di erene in the distane determinations; the largest e et omes rom the hoie o horizontal-branh magnitudes on the CMD in the two analyses. To derive the metalliity o the luster, we reanalyze the photometry o Terndrup & Walker (1994) using the method outlined by Sarajedini (1994). Their method uses polynomial Ðts to the Da Costa & Armandro (199) giant branh sequenes to measure simultaneously the metalliity o and extintion toward a luster rom CMDs in (, [I). In using this method, we make two hanges to reñet reent developments. First, we use the Stanek (1996) reddening map or BaadeÏs window, whih allows us to orret the photometry o the luster (and o the probable luster members in Table 1) or the onsiderable variation in extintion with position. Seond, we adopt the Carney, Storm, & Jones (1992) alibration o the RR Lyrae lumi- nosity (M \ 1.1 ].16[Fe/H]) instead o the Lee et al. (199) sale impliit in the Sarajedini (1994) approah; sine the metalliity slope is almost the same or both alibrations, this has the e et o reduing the derived distane by.2 mag. We adopt the Carney et al. (1992) sale so that our distane or NGC 6522 is on the same system as Galati enter distane rom RR Lyrae stars in BaadeÏs window (Carney et al. 1995). This way, the relative distane between the luster and the Galati enter an be ound independently o the zero point in the alibration o RR Lyrae luminosity. In employing the Stanek (1996) map, we Ðrst add an o set o [.1 to A (i.e., the reddening is less than in StanekÏs map), as disussed by Gould, Popowski, & Terndrup (1998) and onðrmed by Alok et al. (1998a). Next we interpolate aross a region o radius 2@ around NGC 6522, whih is not inluded in the map. An inspetion o photographs o BaadeÏs window (see, e.g., Blano 1984) shows that, at least near the luster, the gradient in extintion is primarily north-south, with the extintion lower to the north o the luster. In Figure 3, we show the Stanek reddenings (with the orreted zero point) or the stars in the SJW survey as a untion o the delination di erene between eah star and the luster enter (positive values indiate stars north o the luster enter). The straight line shows a linear Ðt to the reddenings; the rms satter about this line is.6 in A. From this plot and the Stanek (1996) relation A \ 2.49E( [I), we onlude that the total and seletive extintions toward the enter o the luster are A \ 1.42 ^.5 and E( [I) \.57 ^.2. In Figure 4 (let), we display a CMD in (, [I) or stars within 1@.5 o NGC 6522 (dots) along with the photometry or the luster members in Table 1 (irles). The right panel shows the photometry or stars arther than 2@.5 rom the A δ rom NGC 6522 (armin) FIG. 3.Èisual extintion near NGC Shown are the values o the visual extintion A or stars in the proper-motion survey, as derived rom the Stanek (1996) extintion map or BaadeÏs window, plotted against the di erene in delination between eah star and the enter o NGC The extintion values have been redued by [.1 rom the Stanek map, ollowing the realibration o the zero point o the extintion in Gould et al. (1998). The straight line is a least-squares Ðt to the data, and was used to ompute reddenings within 2@ o the luster enter, an area not inluded in the Stanek (1996) extintion map.

5 148 TERNDRUP ET AL. ol I I FIG. 4.ÈCMDs in BaadeÏs window. L et, photometry or stars within 1@.5 o NGC 6522; right, photometry or stars arther than 2@.5 rom the luster enter. The large irles are or the probable luster members in Table 1. All photometry has been orreted to A \ 1.42, a value appropriate or the luster enter, as desribed in the text. The urved line is a quadrati Ðt to the luster giant branh. luster enter. In this Ðgure, the photometry is rom the soures listed in Terndrup & Walker (1994) and Terndrup et al. (1995); all the olors and magnitudes have been orreted to the reddening adopted or the luster enter using the linear Ðt to the reddening with position in Figure 3. The urved line is a quadrati Ðt to giant branh stars in the let panel o Figure 4 (not only the possible proper-motion members) using a 3 p rejetion riterion. Note that, in support o our seletion method, the possible luster members are near the lusterïs giant branh and well away rom the majority o bulge stars (the satter about mean giant branh is signiðantly worse without the use o the Stanek 1996 reddening map). Returning to the distane determination or the luster, we note that the level o the horizontaranh at a representative olor or RR Lyrae stars (B[ \.8, at the reddening to the luster) is \ 16.5 ^.15. This is onsistent with the horizontal-branh level on the muh better CMD o NGC 6522 rom the Hubble Spae Telesope in Sosin et al. (1997); interpolating rom their CMD, we Ðnd \ ^.7. Using this value or, we apply the Sarajedini HB (1994) method and derive [Fe/H] HB \[1.28 <.12. The metalliity we derive agrees with the Rutledge et al. (1997) value o [Fe/H] \[1.21 ^.4, derived rom inrared Ca II triplet measures transormed to the highresolution abundane sale o Carretta & Gratton (1997). The method simultaneously returns an estimate o the luster reddening o E( [I) \.58 ^.3, in exellent agreement with the value E( [I) \.57 ^.2 derived above rom the orreted Stanek (1996) extintion map, and an estimate o the luster distane modulus o 14.3 ^.1 on the Carney et al. sale (see also below). In Figure 5, we plot the Lik Mg indexes as a untion o 2 olor or the BaadeÏs window sample (Terndrup et al. 1995). The irles with error bars are or the stars in Table 1. With the possible exeption o Arp 2-69, whih has a value o Mg about 1.8 p above that o the others, all the possible 2 luster members have low Mg indexes at a given olor, 2 unlike those or the bulge stars in the SJW sample. Sine the alibration o the Lik indexes with metalliity is unertain or stars o low metalliity (Sadler et al. 1996), we do not attempt to measure the abundane o NGC 6522 rom these Mg indexesèwe simply note that the probable members are 2 among the lowest metalliity stars in BaadeÏs window, onsistent with both the photometri and spetrosopi determinations o the abundane o NGC A diret alulation o the relative distane between the luster and the Galati enter is as ollows: An analysis o the magnitudes o RR Lyrae stars in BaadeÏs window shows that the dereddened visual magnitude o the horizontal branh at typial olors o RR Lyrae stars is \ ^.3 (Alok et al. 1998b); there is an additional,bw error rom the unertainty in extintion that does not enter here, beause we are omparing the luster horizontaranh and the bulge horizontaranh in the same Ðeld. Taking the dependene o the horizontal-branh level as M,HB \a]b[fe/h] (7) and using the deðnition o distane modulus, we write the di erene in distane moduli between the luster and the Galati enter as *,BW \( HB, [ A [ M, ) [ (,BW [ M,BW ), (8)

6 No. 4, 1998 PROPER MOTION OF NGC Mg I FIG. 5.ÈMg indexes or SJW stars rom Terndrup et al. (1995) as a untion o [I 2 olor. The irles with error bars are or the likely members o NGC 6522 in Table 1, while the rosses are or the rest o the SJW sample. where the subsript reers to the luster and the subsript BW to the RR Lyrae stars in BaadeÏs window. With the adopted reddening to NGC 6522, this beomes * \([.23 ^.8) ] b([fe/h] [ [Fe/H] ). (9),BW BW We ound [Fe/H] \[1.28 ^.12 (above), and we adopt [Fe/H] \[1.1 (Walker & Terndrup 1991). Setting BW b \.16 ^.1, we have * \[.2 ^.8. (1),BW For a distane to the Galati enter o 7.8 kp, the luster is thereore at a distane o 7.1 ^.3 kp. We will adopt this distane in the analysis that ollows. 4. SPACE MOTION Beause the SJW survey was not tied to an extragalati reerene rame, some are is required to transorm this measurement into an estimate o the lusterïs spae motion. We begin by writing the observed vetor proper motion l as obs l \ l,m [ _,M[,M [ _,M[*l, (11) obs d d l where d and are the distane to the luster and its veloity l transverse l,m to the line o sight, d and are the mean distane and transverse veloity o bulge Ðeld,M stars in BaadeÏs window, is the veloity o the Sun transverse to the line o sight, _,M and *l is a orretion to be disussed below. (Note that 1 mas yr~1 \ 4.74 km s~1 kp~1.) Next we solve or and write the result in a way that allows easy identiðation l,m o the role o various unertainties, l,m \ d l A l ] *l ],M d B ] _,M A 1[ d l d B. (12) We now estimate *l, the mean proper motion o bulge stars relative to the proper-motion rame o the SJW stars. There are two distint soures o ontamination that lead to an o set: oreground disk stars and the luster stars themselves. To estimate the o set due to oreground disk stars, we ous on the subsample o 31 stars (out o a total o 427) with spetrosopi distane estimates rom Sadler et al. (1996). We ompare the 241 stars with distanes greater than 4 kp with the ull subsample o 31 stars and Ðnd an o set (*k, *k ) \ (.16, [.5) mas yr~1. We assume that this o set l is representative b o all 427 stars. The ull SJW sample ontains D33 stars at distanes greater than 4 kp, o whih about seven are luster members. The o set due to luster stars is thereore 7/33 D 2% o the value given in equation (3), or (*k, *k ) \ (.2, [.12) mas yr~1. Within the auray o our l determinations, b the ombined o set rom these two soures o bias is thereore (*k, *k ) \ (.2, [.2) mas yr~1. (13) The distane to the Galati enter is unertain (see, e.g., Reid 1993), but or present purposes, we treat it as being Ðxed at the distane given by the Carney et al. (1992, 1995) sale, 7.8 kp, and sale all results to this value. Reall that we have also estimated the distane to the luster on this sale. Sine BaadeÏs window is very lose to the minor axis o the bulge (at l \]1 ), a air sample o bulge stars should have \. However, there are two ompeting biases,m that an ause a deviation rom this expeted value. First, the underlying SJW sample is biased toward brighter stars and hene toward stars on the near side o the bulge. Seond, the one o observation is wider on the ar side o the bulge than the near side and so ontains more ar-side stars. By examining the distribution o spetrosopi distanes, we estimate that the median o the distribution lies.3 ^.3 kp behind the peak, whih we identiy with the Galatoentri distane. The SJW sample is thereore slightly biased toward the kinematis o ar-side stars. Sine there are D2 stars between the peak and the median, we estimate this bias as D2/241 o the di erene in mean proper motion o the near-side and ar-side subsamples, whih we evaluate as (*k, *k ) \ ([.2,.2) mas yr~1. Sine this orretion is an order o magnitude smaller than the observational errors, we ignore it and adopt \. We use equation (11) and apply the orretion,m derived in equation (13) to the values o the observed (relative) proper motion given in equation (3) to obtain the proper motion with respet to a star at rest in the nulear bulge, (k, k ) \ (1.4 ^.2, [6.2 ^.2) mas yr~1. (14) l,l l,b We note that the zero point o the rame is unertain by.17 mas yr~1 in eah diretion beause o the shot noise o the D33 stars, with satter D3 mas yr~1 eah. Assuming that the total solar motion is \ (9, 232, 7) km s~1 in the (U,, W ) reerene rame (U _ being positive inward), we evaluate equation (12) and Ðnd (v, v ) \ (68 ^ 18, [28 ^ 23) km s~1, (15) l,l l,b where we have inluded the distane errors but not the unertainty in the overall distane sale. Finally, we evaluate the 3-spae motion in the (U,, W ) reerene rame, \ ([3 ^ 7, 68 ^ 18, [28 ^ 23) km s~1. (16) l That NGC 6522 has a signiðant motion away rom the Galati plane indiates immediately that it is on a halo

7 1482 TERNDRUP ET AL. orbit (Cudworth & Hanson 1993), onsistent with most other globular lusters that have [Fe/H] \ [1 (Zinn 1985; Armandro 1989). The lusterïs urrent distane on the Carney et al. (1995) sale (7.1 ^.3 kp) is very lose to the Ðxed Galatoentri distane o 7.8 kp, so the luster must be on an orbit that takes it quite near (\1 kp) the enter o the Galaxy. We explore possible past orbits or NGC 6522 using our measurements o the lusterïs position and spatial motion. For the inner Galaxy, we assume an oblate logarithmi potential o the orm '(x, y, z) \ 1v2 ln [x2]y2](kz)2] ] onst, (17) 2 where v is the irular veloity in the equatorial plane (taken as 22 km s~1) and k is a Ñattening term that we set in the range k \ 1È3. We Ðnd that about 2 ] 16 yr ago, the luster passed the Galati enter at a distane as lose as 4È55 p, and that it generally does not ahieve an apogalati distane in exess o 15 p. Our alulation also shows that the timesale or signiðant hanges to the orbital energy through dynamial rition is at least a ew times 11 yr. Clusters that pass within a ew hundred parses o the Galati enter are likely to experiene signiðant shoking rom the time-variable gravitational potential o the bulge and inner disk (see, most reently, Gnedin & Ostriker 1997; Murali & Weinberg 1997) and are unlikely to survive muh longer than another Hubble time. NGC 6522 is also a oreollapsed luster (see, e.g., Djorgovski & King 1986; Lugger, Cohn, & Grindlay 1995), another sign that the luster may have experiened signiðant shoks (see, e.g., Gnedin & Ostriker 1997 and reerenes therein). We thereore suggest that NGC 6522 would be a good target or detailed kinemati studies o high auray, to inrease the number o veloity-seleted members and to obtain better estimates o the mass and veloity dispersion. Work by A. G. and P. P. was supported in part by grant AST rom the NSF. D. M. T. thanks Ruth C. Peterson or her helpul insights. REFERENCES Alok, C., et al. 1998a, ApJ, 494, 396 Pryor, C., & Meylan, G. 1993, in ASP Con. Ser. 5, Struture and ÈÈÈ. 1998b, ApJ, in preparation Dynamis o Globular Clusters, ed. S. G. Djorgovski & G. Meylan (San Armandro, T. E. 1989, AJ, 97, 375 Franiso: ASP), 357 Arp, H. 1965, ApJ, 141, 45 Reid, M. J. 1993, ARA&A, 31, 345 Barbuy, B., Ortolani, S., & Bia, E. 1994, A&A, 285, 871 Rih, R. M. 199, ApJ, 362, 64 Blano, B. M. 1984, AJ, 89, 1836 Rih, R. M., Terndrup, D. M., & Sadler, E. M. 1998, in preparation Carney, B. W., Fulbright, J. P., Terndrup, D. M., Suntze, N. B., & Walker, Rutledge, G. A., Hesser, J. E., Stetson, P. B., Mateo, M., Simard, L., Bolte, A. R. 1995, AJ, 11, 1674 M., Friel, E. D., & Copin, Y. 1997, PASP, 19, 883 Carney, B. W., Storm, J., & Jones, R , ApJ, 386, 663 Sadler, E. M., Rih, R. M., & Terndrup, D. M. 1996, AJ, 112, 171 Carretta, E., & Gratton, R. G. 1997, A&AS, 121, 95 Sarajedini, A. 1994, AJ, 17, 618 Cudworth, K. M., & Hanson, R. B. 1993, AJ, 15, 168 Shweitzer, A. M., & Cudworth, K. M. 1996, in ASP Con. Ser. 92, Forma- Da Costa, G. S., & Armandro, T. E. 199, AJ, 1, 162 tion o the Galati HaloÈInside and Out, ed. H. L. Morrison & Dauphole, B., Ge ert, M., Colin, J., Duourant, C., Odenkirhen, M., & A. Sarajedini (San Franiso: ASP), 532 Tuholke, H.-J. 1996, A&A, 313, 119 Smith, M. G., Hesser, J. E., & Shawl, S. J. 1976, ApJ, 26, 66 Dinesu, D. I., Girard, T. M., van Altena, W. F., Yang, T.-G., & Lee, Y.-W. Sosin, C., Rih, R. M., Piotto, G., Djorgovski, S. G., King, I. R., Dorman, 1996, AJ, 111, 125 B., Liebert, J., & Renzini, A. 1997, in Abundanes in Stellar Evolution, Djorgovski, S., & King, I. R. 1986, ApJ, 35, L61 ed. R. T. Rood & A. Renzini (Cambridge: Cambridge Univ. Press), in Gnedin, O. Y., & Ostriker, J. P. 1997, ApJ, 474, 223 press Gould, A., Popowski, P., & Terndrup, D. M. 1998, ApJ, 492, 778 Spaenhauer, A., Jones, B. F., & Whitord, A. E. 1992, AJ, 13, 297 (SJW) Harris, W. E. 1996, AJ, 112, 1487 Stanek, K. Z. 1996, ApJ, 46, L37 Ibata, R. A., Wyse, R. F., Gilmore, G., Irwin, M. J., & Suntze, N. B. 1997, Terndrup, D. M., Sadler, E. M., & Rih, R. M. 1995, AJ, 11, 1774 AJ, 113, 634 Terndrup, D. M., & Walker, A. R. 1994, AJ, 17, 1786 Jones, B. F., Klemola, A. R., & Lin, D. N. C. 1994, AJ, 17, 1333 Walker, A. R., & Mak, P. 1986, MNRAS, 22, 69 Lee, Y.-W., Demarque, P., & Zinn, R. 199, ApJ, 35, 155 Walker, A. R., & Terndrup, D. M. 1991, ApJ, 378, 119 Lugger, P. M., Cohn, H. N., & Grindlay, J. E. 1995, ApJ, 439, 191 Webbink, R. F. 1981, ApJS, 45, 259 Murali, C., & Weinberg, M. D. 1997, MNRAS, 288, 749 Zinn, R. 1985, ApJ, 293, 424 Peterson, C. J., & Reed, B. C. 1987, PASP, 99, 2 Zinn, R., & West, M. J. 1984, ApJS, 55, 45