arXiv:astro-ph/0202212v2 19 Feb 2002 eaudne oain n lse yais(e uiPecc Fusi (see dynamics l cluster mass and age, rotation, include abundance, i candidates He morphology pa- possible HB of govern clusters; role that globular the “sec metallicity) understand the (besides to to rameters seeks tied which is controversy debate, metallicity parameter” The Brown 1997). 1992; Pinsonneault al. O’Connell, & et Dorman, Demarque, Horch, 1990; 1995; Chio Renzini Rood Bressan, & & 1990; Greggio el t al. 1994; of reflecting et Fagotto populations metallicity, Brocato dominant high (e.g., cooler, at liptical the lie possib in Another found EHB composition the 1997). that Lee meta is wide & a ity Park of possibility tail (e.g., metal-poor One distribution the in licity two reside nature. been stars EHB their have stars the regarding there that EHB thought and origin, of candidate, EHB schools theoretical an strong of a proof were observational the before dif may populations cases. the two but (sdO) the O field, in and Galactic considerably (sdB) local B the subdwarf of the Thes of stars progeny. analogs be their to and ex- appear stars 2000) of stars (EHB) population al. minority branch et a horizontal (Brown from treme imaging comes and emission UV 1997) this al. et that (Ferguson et O’Connell spectroscopy Brown UV (see 1991; from origin al. know upturn we today UV man but the 1999), covered for debate candidates Early possible ol as populations. co galaxies early-type ellipticals passively-evolving of in picture component traditional im- stellar the the hot dicted and a 1969), discov- “UV of (Code existence major the astronomy plied first dubbed extragalactic the UV Å, among in 2700 was eries of phenomenon The shortward upturn upturn.” strong a hibit rpittpstuigL using typeset Preprint T PERIN APPEAR O h pcr felpia aaisadsia aaybulges spiral and galaxies elliptical of spectra The h Vutr sntcmltl nesod oee.Long however. understood, completely not is upturn UV The bnacsfl nterneosre o udafBsasof stars B subdwarf for conseque observed natural range a the are in anomalies fall abundan abundance abundances The Such ell upturn. solar. any 2% UV of at the observations within in lines time absorption composi first metallic chemical the the For however, debate. progeny; well-establishe standing their now is and It stars Å. (HB) 2500 of NGC shortward ellipticals, rise quiescent spectral all Of Explorer. Spectroscopic htpeitdb ipecoigflwmdl fteNC19 X- 1399 NGC the of headings: models Subject flow cooling simple by predicted that h pe ii to limit upper The T HTSHRCAUDNE FTEHTSASI G 39ADLIMI AND 1399 NGC IN STARS HOT THE OF ABUNDANCES PHOTOSPHERIC lhuhNC19 sa h etro h onxcutr efin we cluster, Fornax the of center the at is 1399 NGC Although gala elliptical giant the of spectroscopy far-UV present We HE A STROPHYSICAL A T pc eecp cec nttt,30 a atnDie B Drive, Martin San 3700 Institute, Science Telescope Space E tl mltajv 04/03/99 v. emulateapj style X eateto srnm,Uiest fVrii,PO o 3 Box P.O. Virginia, of University Astronomy, of Department 1. INTRODUCTION oe51 AAGdadSaeFih etr rebl,M 20 MD Greenbelt, Center, Flight Space Goddard NASA 551, Code J λλ aais bnacs–glxe:elpia aais st galaxies: – elliptical galaxies: – abundances galaxies: OURNAL 1032 L , ETTERS 08eiso s3 is emission 1038 oapa nTeAtohsclJunlLetters Journal Astrophysical The in appear To T HOMAS LSE OLN FLOW COOLING CLUSTER .B M. R . ,cool, d, OBERT 9 R × i & si, ntra- AYMOND ROWN oss, ond ABSTRACT ex- 10 fer he il- is l- n y e - i − .O’C W. 15 H , 1 lioe D228 bonssieu [email protected] [email protected], 21218. MD altimore, r s erg .O G. ENRY htti msincmsfo o oiotlbranch horizontal hot from comes emission this that d 39hstesrnetkon“Vutr”–asharp a – upturn” “UV known strongest the has 1399 k 92 usene l 98,ee huhtesetaof spectra the s very though qualitatively even are The 1988), wavelengths longer al. at et ellipticals Capacciol Burstein (Bertola, 1982; galaxies Oke early-type mag 2.05–4.50 quiescent from nearby (ranging variation strong surprisingly Mg elzii1997). Bellazzini & aaya h etro h onxcutr thstestronge the elliptical has It giant the cluster. Fornax is the 1399 of NGC Explorer center the Spectroscopic 2000). at Ultraviolet al. galaxy Far et Moos the (FUSE; with 1399 are NGC Gyr model. 8 (1994) al. as et low Bressan as metal-rich ages the while in hypothesis, allowed metal- (1997) the Lee under & required t Park are exceeding clusters Ages globular Galactic galaxies. more of these in to ages populations different lead stellar imply the which hypotheses These emissio of UV population. in the both increase from subsequent (RGB), a and rate branch stars mass-loss EHB giant metal-rich higher red a in- the possibly d at on and metallicity, redder abundance high become He at to reversed increased distribution be can HB metallicity the ar creasing for school in metal-rich trend emission the the UV contrast, more that In producing galaxies. thus old age, massive with becom- HB para blue metal-poor “second more metal- the dominant ing with the the morphology, is in HB age driving stars scenario, eter” from this In comes tail. upturn poor the though galaxies even such dices, earlier; UV formed strong higher and with have massive galaxies more scena that de- are own reasons metallicity their emission of EHB school proof the metal-poor indic further in as The color sides correlation this optical Both see of bate 1988). behavior al. et the (Burstein to opposite strengths), pia aay u pcr lal hwphotospheric show clearly spectra our galaxy, iptical ONNELL 1,Caltevle A293 [email protected] 22903. VA Charlottesville, 818, eo sa 5 oa,S sa 3 oa,adCis C and solar, 13% at is Si solar, 45% at is N of ce − h aatcfield. Galactic the hrceie ythe by Characterized ihti eaei id eotie bevtosof observations obtained we mind, in debate this With ino hs tr a entesbeto long- a of subject the been has stars these of tion c fgaiainldfuin hs photospheric These diffusion. gravitational of nce HL 2 1 yNC19,otie ihteFrUltraviolet Far the with obtained 1399, NGC xy m .F C. cm ieasrto nthe in absorption line 1550 a luminosity. ray oeiec o O for evidence no d − ERGUSON 2 − [email protected] 771. qiaett 0.14 to equivalent , V mean oo spstvl orltdwt h teghof strength the with correlated positively is color la otn lrvoe:galaxies ultraviolet: – content ellar ealct,a rce hog pia in- optical through tracked as metallicity, m 1550 VI SO H FORNAX THE ON TS V − M oln o emission. flow cooling ad(.. le thge line higher at bluer (i.e., band V ⊙ oo,teU punshows upturn UV the color, yr − 1 n esthan less and , u imilar. eto ue hose poor gues ,& i, in ) rio. for m- es st n 2

UV upturn of any quiescent measured to date nated by a population of EHB stars and their post-HB progeny, (m1550 −V = 2.05 mag; Burstein et al. 1988). Although the UV the AGB-Manqu´e stars (Brown et al. 1997). upturn correlates well with optical metallicity indices, such in- To analyze the FUSE spectra of NGC 1399, we integrated dices track the composition of the cool population (RGB and the best-fitting EHB model from Brown et al. (1997), using main sequence stars). Our far-UV spectroscopy is intended to new high-resolution synthetic spectra calculated at represen- measure the photospheric abundances in the hot stars produc- tative points along the EHB evolutionary path. The track as- ing the UV upturn – measurements not possible in earlier spec- sumes a small HB envelope mass (0.016 M⊙), high metal- troscopy of elliptical galaxies (Brown et al. 1997; Ferguson et licity ([Fe/H]=0.71), and high main-sequence He abundance al. 1991), due to the low signal-to-noise and resolution. In ad- (Y = 0.45). Note that approximately half of the far-UV flux dition, X-ray observations suggest NGC 1399 is the host to a in this track comes from the EHB itself, with the rest coming significant cooling flow (Bertin & Toniazzo 1995; Rangarajan from the AGB-Manqu´e phase. We computed solar abundance et al. 1995), making it an interesting target for O VI emission atmospheres using the ATLAS9 code (Kurucz 1993), under the measurements. We present here an analysis of our FUSE data assumption of local thermodynamic equilibrium (LTE). Syn- for NGC 1399; future papers will present our observations of thetic spectra were then calculated using the SYNSPEC code other ellipticals, which are not yet complete. (Hubeny, Lanz, & Jeffery 1994), again assuming LTE, but with varying abundances of C, N, Si, Fe, and Ni (holding other el- 2. OBSERVATIONS ements at solar abundance). Note that this code has been used We observed NGC 1399 on 12–13 Dec 2000 and 3 Oct 2001 previously to model FUSE and HUT data of sdB stars (Ohl, for 28 ksec, with 20 ksec during orbital night. The spectro- Chayer, & Moos 2000; Brown et al. 1996). We used the Ku- scopic resolution is limited by the velocity dispersion of the rucz (1993) line list in our spectral synthesis, including lines galaxy (331 km s−1; Burstein et al. 1988), so we scheduled ob- predicted from atomic physics but not yet measured in the lab- servations in the default 30′′ × 30′′ low-resolution square aper- oratory. Weak lines that are experimentally unconfirmed are ture. The data were obtained in time-tag mode, recording the potentially prone to position and wavelength errors, but they arrival time and detector position of each photon. provide the necessary pseudocontinuum. We reprocessed the data using the CALFUSE software, ver- No Galactic foreground reddening is applied to the mod- sion 2.0.5. Scattered light and airglow emission lines both in- els. Both the maps of Burstein & Heiles (1984) and Schlegel, − crease during orbital day, so we rejected daytime data. We Finkbeiner, & Davis (1998) suggest that E(B V) is nearly also implemented pulse-height screening appropriate for very 0 mag along a line of sight toward NGC 1399. faint targets, by raising the minimum pulse height in CALFUSE To compare the composite models to the FUSE data, we de- from 0 to 4, thus discarding more of the detector dark counts fined spectral indices of C, N, and Si (Table 1). Each index with little loss of source counts. measures the ratio of the mean flux in a line region to that in Useful data were obtained in all detector segments, each with continuum regions, selected to be free of strong features from its own wavelength bins and instrumental resolution. Because other elements. The C and Si indices come from the ratio in the velocity dispersion limits the useful resolution, we coad- one section of the spectrum, whereas the N index comes from ded all of the data onto a linear wavelength scale (0.25 Å bins), the average in two sections. with propagation of statistical errors. The bins in our coad- ded data are much larger than the nominal FUSE wavelength TABLE 1: Composition Indices Cont. Line Dominant Transitions bins (∼ 0.01 Å), but they adequately sample the resolution ele- Ion (Å) (Å) (Å) ments in the spectrum (> 1 Å for velocity-broadened features). We masked bad pixels when coadding the data, because CAL- C III 1170–1173 1174–1177 1174.93 1175.261175.59 FUSE does not account for small-scale flat-field features. Note 1177–1180 1175.711175.991176.37 that the wavelength solution in this version of CALFUSE is im- N III 986–988 989–993 989.80991.51991.58 proved significantly, although these corrections are unimportant 994–996 in our large wavelength bins. Furthermore, although fixed pat- N II 1074–1078 1083–1087 1083.99 1084.561084.58 tern noise can be an issue in high-resolution FUSE data, it is 1092–1096 1085.531085.551085.70 negligible for our large bins and coadded detector segments. Si III 1105–1107 1107–1114 1108.36 1109.941109.97 1114–1116 1113.171113.201113.23 3. STELLAR MODELS 4. RESULTS Our FUSE spectra offer the first clear detection of metal- lic absorption lines from the hot stars responsible for the UV 4.1. Photospheric Abundances upturn. Previously, Brown et al. (1997) fit composite mod- By varying the abundances in the synthetic spectra, we find els to the spectra of elliptical galaxies observed with the Hop- that the composition that best reproduces the line indices mea- kins Ultraviolet Telescope (HUT), including NGC 1399. Those sured in the data is C at 0.02 ± 0.01 solar, N at 0.45 ± 0.15 so- spectra were obtained through large apertures, with coverage lar, and Si at 0.13 ± 0.06 solar (Figure 1). The errors represent from 900–1800 Å, but the low signal-to-noise and resolution statistical uncertainties in each line index, transformed to abun- hampered detection of individual absorption lines from ele- dance uncertainties. This result does not significantly depend ments heavier than H. Brown et al. (1997) integrated synthetic upon our chosen evolutionary track; although the best-fitting spectra over EHB evolutionary tracks from Dorman, Rood, & track from Brown et al. (1997) has enhanced He and metal- O’Connell (1993), using synthetic spectra appropriate for the licity, our derived abundances remain unchanged (within 1σ) HUT resolution (Brown, Ferguson, & Davidsen 1996). The if we use other unenhanced tracks that reproduce the far-UV models reproduce well the entire far-UV spectral energy distri- spectrum. bution, and show that the light from elliptical galaxies is domi- Fe and Ni lines blanket the entire FUSE region, but 3

5 ) −1

Å 4 −2 cm

−1 3 CrII + NII SiIII SiIII FeII erg s FeII −14 NII 2 NIII FeIII FeII SiII CaII CIII

Flux (10 + NII 1 NI OI NIII 990 1000 1080 1090 1100 1110 1120 1130 1140 1150 1160 1170 1180 Observed Wavelength (Å) FIG. 1– Sections of the FUSE NGC 1399 spectrum (black histogram). Photospheric (roman) and Galactic ISM (italics) features are easily distinguished by their widths. The composite model is shown with photospheric Fe and Ni at solar (thick grey curve) and 5 times solar (thin curve) abundance. Thick black lines at the top and bottom of the plot denote the regions used for spectral indices (Table 1). The model normalization changes by 10% at the break (the absolute flux calibration varies by ∼10% between detector segments). do not significantly affect the derived abundances of the light 4.2. Cooling Flow Emission elements. To demonstrate the systematic errors from Fe and NGC 1399 lies at the center of the . The Ni assumptions, we varied the abundances of Fe and Ni over a X-ray luminosity of the galaxy implies a large cooling flow, extremely large range: from 0.01 solar to 5.0 solar. Through- −1 with a mass deposition of at least 2 M⊙ yr (Bertin & Toni- out this range, the derived abundances of C, N, and Si remain azzo 1995; Rangarajan et al. 1995). Cooling flow models pre- within their 1σ statistical uncertainties. We show the same evo- dict that this X-ray emission should be accompanied by O VI lutionary model, when Fe and Ni are enhanced, in Figure 1; the λλ1031.9,1037.6Å emission, radiated as the hot gas cools from increased Fe and Ni severely depressed the continuum, so the ∼ 107 K to ∼ 105 K. Bregman, Miller, & Irwin (2001) did not curve has been renormalized upward. Note that diffusion pro- find such emission in their FUSE observations of NGC 1404 cesses tend to deplete the photospheric abundances of the light (upper limit 0.3 M yr−1), an elliptical galaxynear the center of elements in sdB stars (Lamontagne et al. 1985; Lamontagne, ⊙ the Fornax cluster, but its X-ray luminosity suggests a smaller Wesemael, & Fontaine 1987), but Fe can be radiatively sup- cooling flow deposition rate than that for NGC 1399 (Bertin & ported; e.g., Fe enhancement is thought to be the mechanism Toniazzo 1995). Thus, NGC 1399 offers a better opportunity for the pulsating sdB stars (Charpinet et al. 1997), which show for finding the warm gas from the Fornax cooling flow. no photospheric Fe depletion (Heber, Reid, & Werner 2000). We see no O VI emission in our NGC 1399 spectrum (Fig- The C and Si line indices rely upon lines that are not res- onance transitions, and thus should not be contaminated from ure 2); indeed, the observed flux at the location of the 1032 Å any interstellar medium (ISM) in NGC 1399 – i.e., the features (rest frame) component is slightly lower than the expected flux in the FUSE spectrum should be completely photospheric. In from the stellar continuum. To set upper limits, we first fit an contrast, the N index comes from series of resonance lines. Be- EHB model with the abundances determined above. The fit cause the NGC 1399 spectrum is redshifted by 1424 km s−1, region was from 1035–1045 Å, and we included C II and O I narrow absorption features from these same transitions in the absorption lines from our own Galactic ISM. Note that these Galactic ISM can be seen in the FUSE spectrum, offset by strong Galactic features are not accompanied by equivalent features in the NGC 1399 frame, again implying a lack of a ∼ 5 Å to the blue. If NGC 1399 has a significant cold ISM, cold ISM. Once the best fit was achieved, these components the redshifted N features could have both ISM and photospheric were fixed, and then two components representing the O VI contributions. However, when looking at purely interstellar features in the

FUSE data, we see no evidence for absorption from a cold ) 4 OVI λλ1031.9,1037.6 ISM in NGC 1399. The N II and N III features in Table 1 can −1 Å arise from both hot stellar photospheres and the ISM, but other −2 3 resonance transitions will arise only from an ISM. These can cm be used to discern how much contamination of photospheric −1 features we should expect from an ISM in NGC 1399. E.g., 2 erg s

strong absorption from N I (1134.2 Å, 1134.4 Å, and 1135.0 Å) −14 II and Ca (1135.5 Å and 1135.6 Å) from the Galactic ISM is 1 present in the FUSE spectrum (narrow lines in Figure 1), but OI there is no corresponding strong absorption in the NGC 1399 Flux (10 CII 0 frame (1140 Å, with a width of 1.3 Å). The same can be said 1036 1038 1040 1042 1044 II I of C λ1036 Å and O λ1039 Å (only Galactic features are Observed Wavelength (Å) present). Thus, the N II and N III spectral indices have little, if FIG. 2– The NGC 1399 spectrum at the expected location of O VI any, contribution from an ISM in NGC 1399, and should track emission (histogram with error bars). The curves show the stellar photospheric abundances only. model (light grey), including Galactic ISM absorption (italic labels), and the 3σ upper limit to the redshifted 0 VI emission (dark grey). 4 doublet were added. The wavelengths and widths were fixed with their primordial abundances. The photospheric abundance at the NGC 1399 and velocity dispersion, and the flux of a given element can be affected by both gravitational settling ratio was fixed at 2:1 for the 1032 and 1038 Å components. and radiative levitation, and thus the photospheric abundance −15 −1 −2 We find a 3σ upper limit of fλ1032 = 2.6 × 10 erg s cm can be either reduced or enhanced relative to the primordial −15 −1 −2 (and thus fλλ1032,1038 = 3.9 × 10 erg s cm for the total abundances that drive the evolution of a . Some calculations O VI emission). As in Bregman et al. (2001), we assume that have shown that abundance enhancements via radiative levita- 39 ˙ −1 Lλ1032 = 0.9 × 10 M erg s (a compromise between the re- tion can be especially strong in the low-metallicity regime, be- lations for isochoric and isobaric cooling). Thus, the 3σ upper cause the absorption lines are not saturated (Michaud, Vauclair, limit to the mass deposition rate implied from the O VI emis- & Vauclair 1983). −1 sion is 0.14 M⊙ yr , for the 3 × 3 kpc region sampled by the Taken at face value, our photospheric abundances do not re- FUSE aperture (assuming m − M = 31.52 mag; Ferrarese et al. solve the metallicity debate for the origin of the UV upturn; 2000). The limit becomes twice as large if the fit is restricted to they are neither extremely metal-poor nor extremely metal-rich. the cleaner region around 1038 Å. Models of the X-ray lumi- Tying these photospheric abundances to formation abundances −1 nosity imply a larger rate of ≈ 0.4 M⊙ yr within the FUSE will require further theoretical and observational work. In par- aperture (Rangarajan et al. 1985). ticular, UV observations of sdB stars in globular clusters are needed to demonstrate how the light elements are affected by 5. DISCUSSION diffusion processes in the metal-poor regime, showing whether We have obtained the first clear detection of metallic absorp- gravitational settling or radiative levitation dominates. How- tion features in the UV upturn. The photospheric composition ever, given the tendency for light element abundances to be di- of the hot population in NGC 1399 appears to reflect that seen minished by diffusion, it seems unlikely that the UV emission in the field sdB stars of the . In the nearby Galac- in NGC 1399comes from EHB stars with extremelylow forma- − − tic field population, sdB and sdO stars often show moderate to tion abundances (e.g., at 2.4 < [Fe/H] < 1.0, as suggested severe depletions of the light elements, attributed to diffusion. by Park & Lee 1997); it seems more plausible that these stars Typically, N appears near solar abundance, while C and Si are started with high abundances that were then depleted, but at this depleted (Lamontagne et al. 1985; Lamontagne et al. 1987). point such conclusions are very speculative. Several other processes could skew the photospheric abun- NGC 1399 is thought to be the center of a cooling flow in dances in EHB stars. Greggio & Renzini (1990) noted that hot Fornax. Our upper limit to the O VI emission in the FUSE aper- HB stars of high metallicity would have very thin H envelopes; ture is lower than expected from simple cooling flow models of −10 these could be wind-ejected at mass-loss rates of ∼ 10 M⊙ the X-ray emission. Bregman et al. (2001) suggested that the yr−1, resulting in surface compositions rich in He and N and lack of O VI emission in NGC 1404 could be due to an addi- poor in C and O. CN processing on the RGB is another possi- tional source of heating, but noted that NGC 1404 has no de- ble effect (Sweigart & Mengel 1979). Assuming initial abun- tected central radio source. NGC 1399 hosts a two-lobed radio dances at the solar C:N ratio and C+N nuclei conservation, the source (Sadler, Jenkins, & Kotanyi 1989), which may help ex- observed C and N abundances would imply primordial abun- plain the lack of O VI emission. dances of 0.12 solar, in the absence of diffusion. Galactic field sdB stars tend to show depletion of the light elements from their original near-solar abundances, but little is Support for this work was provided by NASA through the known about the effects of diffusion on metal-poor EHB stars, FUSE Guest Investigator program. The authors gratefully because EHB stars in globular clusters are too distant to ob- acknowledge support from NASA grant NAS 5-9696 to the tain spectroscopy easily. Diffusion is the largest uncertainty Catholic University of America. The authors wish to thank A. when associating the photospheric abundances of EHB stars Sweigart and W. Landsman for useful insight and discussions.

REFERENCES

Bertin, G., & Toniazzo, T. 1995, ApJ, 451, 111 Greggio, L., & Renzini, A. 1990, ApJ, 364, 35 Bertola, F., Capaccioli, M., & Oke, J. B. 1982, ApJ, 254, 494 Heber, U., Reid, I.N., & Werner, K. 2000, A&A, 363, 198 Bregman, J.N., Miller, E.D., & Irwin, J.A. 2001, ApJ, 553, L125 Horch, E., Demarque, P., & Pinsonneault, M. 1992, ApJ, 388, L53 Bressan, A., Chiosi, C., & Fagotto, F. 1994, ApJS, 94, 63 Hubeny, I., Lanz, T., & Jeffery, C. S. 1994, in Newsletter on Analysis of Brocato, E., Matteucci, F., Mazzitelli, I., & Tornambe, A. 1990, ApJ, 349, 458 Astronomical Spectra No. 20, ed. C. S. Jeffery (CCP7; St. Andrews: St. Brown, T.M., Bowers, C.W., Kimble, R.A., Sweigart, A.V., & Ferguson, H.C. Andrews Univ.), 30 2000, ApJ, 532, 308 Kurucz, R. L. 1993, CD-ROM 13, ATLAS9 Stellar Atmosphere Programs and Brown, T.M., Ferguson, H.C., & Davidsen, A.F. 1996, ApJ, 472, 327 2 km/s Grid (Cambridge: Smithsonian Astrophys. Obs.) Brown, T.M., Ferguson, H.C., Davidsen, A.F., & Dorman, B. 1997, ApJ, 482, Lamontagne, R., Wesemael, F., & Fontaine, G. 1987, 318, 844 685 Lamontagne, R., Wesemael, F., Fontaine, G., & Sion, E.M. 1985, 299, 496 Burstein, D., Bertola, F., Buson, L. M., Faber, S. M., & Lauer, T. R. 1988, ApJ, Michaud, G., Vauclair, G., & Vauclair, S. 1983, ApJ, 267, 256 328, 440 Moos, H.W., et al. 2000, ApJ, 538, L1 Burstein, D., & Heiles, C. 1984, ApJS, 54, 33 O’Connell, R. W. 1999, in ARA&A, 37, 603 Charpinet, S., Fontaine, G., Brassard, P., Chayer, P., Rogers, F.J., Iglesias, C.A., Ohl, R.G., Chayer, P., & Moos, H.W. 2000, ApJ, 538, L95 & Dorman, B. 1997, ApJ, 483, 123 Park, J.-H., & Lee, Y.-W. 1997, 476, 28 Code, A.D. 1969, PASP, 81, 475 Rangarajan, F.V.N., Fabian, A.C., Forman, W.R., & Jones, C. 1995, MNRAS, Dorman, B., O’Connell, R. W., & Rood, R. T. 1995, ApJ, 442, 105 272, 665 Dorman, B., Rood, R.T., & O’Connell, R.W. 1993, ApJ, 419, 596 Sadler, E.M., Jenkins, C.R., & Kotanyi, C.G. 1989, MNRAS, 240, 591 Ferguson, H.C., et al. 1991, ApJ, 382, L69 Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500, 525 Ferrarese, L., et al. 2000, ApJ, 529, 745 Sweigart, A.V., & Mengel, J.G. 1979, ApJ, 229, 624 Fusi Pecci, F., & Bellazzini, M. 1997, in The Third Conference on Faint Blue Stars, ed. A.G.D. Phillip, J. Liebert, R.A. Saffer, & D.S. Hayes (Schenectady: L. Davis Press), 255