2 Observations and data reduction

 

 

Table 1: List of observed globular cluster candidates.

Slitlet ID	Originb	RA J2000.0 Dec		V (HST)	(V-I) (HST)	(V-I) (spec)c	cz	S/N	member
1	-	No object in slit.
2	-	10$^{\rm h}$05$^{\rm m}$06 $.\!\!^{\prime\prime}$5	-7$^\circ$45$^{\rm m}$01 $^{\prime \prime }$	-	-	1.183	$408\pm37$	9	$\surd$
3	-	10$^{\rm h}$05$^{\rm m}$06 $.\!\!^{\prime\prime}$1	-7$^\circ$44$^{\rm m}$30 $^{\prime \prime }$	-	-	1.142	$439\pm17$	23	$\surd$
4	-	10$^{\rm h}$05$^{\rm m}$09 $.\!\!^{\prime\prime}$9	-7$^\circ$45$^{\rm m}$01 $^{\prime \prime }$	-	-	0.883	$0\pm11$	72	-
5	WHT09	10$^{\rm h}$05$^{\rm m}$11 $.\!\!^{\prime\prime}$4	-7$^\circ$45$^{\rm m}$01 $^{\prime \prime }$	-	-	0.980	$379\pm21$	32	$\surd$
6	-	10$^{\rm h}$05$^{\rm m}$09 $.\!\!^{\prime\prime}$2	-7$^\circ$43$^{\rm m}$52 $^{\prime \prime }$	-	-	-	z=0.39a	-	-
7	HST08	10$^{\rm h}$05$^{\rm m}$12 $.\!\!^{\prime\prime}$6	-7$^\circ$44$^{\rm m}$28 $^{\prime \prime }$	$20.282\pm0.006$	$1.178\pm0.008$	1.148	$291\pm16$	28	$\surd$
8	HST03	10$^{\rm h}$05$^{\rm m}$12 $.\!\!^{\prime\prime}$6	-7$^\circ$44$^{\rm m}$09 $^{\prime \prime }$	$19.794\pm0.006$	$1.051\pm0.009$	1.029	$393\pm13$	61	$\surd$
9	HST02	10$^{\rm h}$05$^{\rm m}$13 $.\!\!^{\prime\prime}$1	-7$^\circ$43$^{\rm m}$59 $^{\prime \prime }$	$19.742\pm0.005$	$1.077\pm0.007$	1.068	$494\pm16$	38	$\surd$
10	HST21	10$^{\rm h}$05$^{\rm m}$13 $.\!\!^{\prime\prime}$0	-7$^\circ$43$^{\rm m}$41 $^{\prime \prime }$	$21.072\pm0.012$	$0.979\pm0.018$	1.050	$456\pm37$	8	$\surd$
11	HST39	10$^{\rm h}$05$^{\rm m}$14 $.\!\!^{\prime\prime}$7	-7$^\circ$43$^{\rm m}$56 $^{\prime \prime }$	$21.539\pm0.014$	$1.199\pm0.018$	1.218	$702\pm31$	8	$\surd$
12	HST13	10$^{\rm h}$05$^{\rm m}$14 $.\!\!^{\prime\prime}$4	-7$^\circ$43$^{\rm m}$35 $^{\prime \prime }$	$20.544\pm0.008$	$0.979\pm0.011$	0.982	$582\pm35$	19	$\surd$
13	HST18	10$^{\rm h}$05$^{\rm m}$15 $.\!\!^{\prime\prime}$2	-7$^\circ$43$^{\rm m}$34 $^{\prime \prime }$	$20.970\pm0.010$	$0.989\pm0.013$	1.015	$546\pm28$	12	$\surd$
14a	HST27	10$^{\rm h}$05$^{\rm m}$16 $.\!\!^{\prime\prime}$6	-7$^\circ$43$^{\rm m}$29 $^{\prime \prime }$	$21.266\pm0.011$	$1.105\pm0.015$	1.117	$321\pm25$	11	$\surd$
14b	HST12	10$^{\rm h}$05$^{\rm m}$17 $.\!\!^{\prime\prime}$2	-7$^\circ$43$^{\rm m}$20 $^{\prime \prime }$	$20.522\pm0.007$	$1.159\pm0.009$	1.138	$696\pm15$	23	$\surd$
15	HST17	10$^{\rm h}$05$^{\rm m}$16 $.\!\!^{\prime\prime}$8	-7$^\circ$43$^{\rm m}$09 $^{\prime \prime }$	$20.946\pm0.010$	$0.978\pm0.013$	1.021	$739\pm23$	16	$\surd$
16	HST32	10$^{\rm h}$05$^{\rm m}$18 $.\!\!^{\prime\prime}$0	-7$^\circ$43$^{\rm m}$15 $^{\prime \prime }$	$21.467\pm0.012$	$0.913\pm0.016$	0.878	$599\pm54$	12	$\surd$
17	HST09	10$^{\rm h}$05$^{\rm m}$17 $.\!\!^{\prime\prime}$2	-7$^\circ$42$^{\rm m}$49 $^{\prime \prime }$	$20.393\pm0.007$	$1.147\pm0.010$	1.167	$813\pm17$	23	$\surd$
18	HST31	10$^{\rm h}$05$^{\rm m}$18 $.\!\!^{\prime\prime}$4	-7$^\circ$42$^{\rm m}$45 $^{\prime \prime }$	$21.392\pm0.012$	$1.190\pm0.016$	1.236	$210\pm23$	9	$\surd$
19	HST57	10$^{\rm h}$05$^{\rm m}$17 $.\!\!^{\prime\prime}$3	-7$^\circ$42$^{\rm m}$08 $^{\prime \prime }$	$21.921\pm0.018$	$0.891\pm0.028$	0.983	$736\pm111$	5	$\surd$
20	HST46	10$^{\rm h}$05$^{\rm m}$17 $.\!\!^{\prime\prime}$6	-7$^\circ$41$^{\rm m}$57 $^{\prime \prime }$	$21.802\pm0.015$	$0.919\pm0.023$	0.946	$661\pm55$	7	$\surd$
21	-	10$^{\rm h}$05$^{\rm m}$20 $.\!\!^{\prime\prime}$1	-7$^\circ$42$^{\rm m}$16 $^{\prime \prime }$	-	-	0.971	$846\pm42$	12	$\surd$
22	-	10$^{\rm h}$05$^{\rm m}$19 $.\!\!^{\prime\prime}$0	-7$^\circ$41$^{\rm m}$47 $^{\prime \prime }$	-	-	0.957	$670\pm41$	7	$\surd$
23	WHT15	10$^{\rm h}$05$^{\rm m}$20 $.\!\!^{\prime\prime}$2	-7$^\circ$41$^{\rm m}$45 $^{\prime \prime }$	-	-	1.140	$682\pm15$	34	$\surd$
24	WHT16	10$^{\rm h}$05$^{\rm m}$21 $.\!\!^{\prime\prime}$2	-7$^\circ$41$^{\rm m}$38 $^{\prime \prime }$	-	-	0.992	$1008\pm18$	42	$\surd$
25a	-	10$^{\rm h}$05$^{\rm m}$23 $.\!\!^{\prime\prime}$4	-7$^\circ$41$^{\rm m}$59 $^{\prime \prime }$	-	-	0.832	$202\pm42$	19	$\surd$
25b	-	10$^{\rm h}$05$^{\rm m}$23 $.\!\!^{\prime\prime}$7	-7$^\circ$41$^{\rm m}$55 $^{\prime \prime }$	-	-	0.890	$721\pm30$	46	$\surd$
26a	-	10$^{\rm h}$05$^{\rm m}$22 $.\!\!^{\prime\prime}$7	-7$^\circ$41$^{\rm m}$13 $^{\prime \prime }$	-	-	1.068	$803\pm24$	10	$\surd$
26b	WHT17	10$^{\rm h}$05$^{\rm m}$22 $.\!\!^{\prime\prime}$9	-7$^\circ$41$^{\rm m}$07 $^{\prime \prime }$	-	-	0.934	$769\pm25$	19	$\surd$

^a

Background galaxy where [O II] emission is visible.

^b

HSTxx denotes globular clusters selected from the HST study of Kundu & Whitmore (1998), WHTxx are objects from the investigation of Kavelaars (1998) while "-'' indicates objects selected from our own VLT-FORS2 image in order to fill the mask.

^c

Pseudo (V-I) colour derived from the globular cluster spectra. The mean error is 0.037 mag. See Sect. 3 for details.

The candidate GCs were selected from the HST/WFPC2 investigation of Kundu & Whitmore (1998) who detected 144 globular cluster candidates in the central region of NGC 3115. In order to keep integration times reasonably short only clusters with V<22 (the peak of the GC luminosity function is at  $V=22.37\pm0.05$; Kundu & Whitmore 1998) were selected while keeping a balance between red ( (V-I) < 1.06) and blue ( $(V-I) \ge 1.06$) clusters. In order to utilize the full field-of-view of our multi-object spectrograph we supplemented this list with GC candidates from a low resolution spectroscopy survey reported in Kavelaars (1998) and also placed some slits on promising objects without prior information (see Table 1). In total two masks with each 26 slitlets covering more than 50 different GC candidates were manufactured.

The observations were carried out 26/27 Feb. 2001 with FORS2 on VLT using the blue 600 l/mm grism and 1 $^{\prime \prime }$ wide slitlets giving a resolution of $\sim$4.8 Å (FWHM) sampled at 1.2 Å pixel^-1. The seeing was generally ${\it FWHM}\le 1\hbox{$.\!\!^{\prime\prime}$ }0$. The total exposure time for the first mask was 12 440 s split up in six individual exposures of varying length. Due to bad weather the second mask was only exposed for 2700 s which was not long enough to be of use for this study. The flux standard GD 71 was observed through a long-slit (2 $.\!\!^{\prime\prime}$5 width) to enable us to correct the continuum shape of the spectra.

The standard data reduction procedures (bias subtraction, flat-fielding, wavelength calibration, sky-subtraction and continuum correction to a relative flux scale) were performed with a combination of MIDAS and IRAF tasks. For each slit a 2-dimensional subsection of the CCD was extracted and then treated as a long-slit observation. The extraction region was defined by tracing the globular cluster spectrum along the wavelength direction and extracting the corresponding sections for the flat-field and arc-lamp observations. The cosmic rays were removed with the routine lacos_spec (van Dokkum 2001). The wavelength calibration was accurate to <0.2 Å. After sky-subtraction GCs were extracted from the individual exposures to achieve an optimal S/N (Horne 1986). Finally, the spectra of the individual exposures were combined.

We used the spectroscopic flux standard GD 71 to correct the continuum shape of our spectra to a relative flux scale and applied a reddening correction of E(B-V) = 0.146 (Schlegel et al. 1998). In order to transform our observations onto the Lick/IDS system we convolved our spectra with a wavelength-dependent Gaussian kernel (taking into account small variations of the instrumental resolution with position on the chip) thereby reproducing the Lick/IDS spectral resolution (Worthey & Ottaviani 1997).

As a first analysis step we measured the recession velocity (with fxcor in IRAF) and S/N of the GC candidates. The first slit (Slitlet ID: 1) did not show any object which we ascribe to a field distortion towards the edges of the field-of-view. Basic information for the remaining 28 globular cluster candidates is listed in Table 1. The first column in this table is an object identifier, the second column indicates from which source the object was selected from. The following columns list the J2000.0 coordinates and V, I photometry of Kundu & Whitmore (1998). The seventh column shows pseudo (V-I) colours derived from our spectroscopy (see Sect. 3 for details). Columns eight and nine list the radial velocity and the mean S/N per pixel respectively. Finally, column ten indicates whether the GC candidate is a spectroscopically confirmed member of the NGC 3115 system (see next paragraph).

The radial velocity of NGC 3115 is listed as $720\pm5$ km s^-1 by Smith et al. (2000). Kavelaars (1998) find a mean velocity of 620 km s^-1 and a velocity dispersion of $\sigma = 177$ km s^-1 from low resolution spectroscopic observations of 22 globular clusters. We consider all objects with radial velocities between 200 km s^-1 and 1300 km s^-1 as members of the NGC 3115 system. Object 6 is a background galaxy with a redshift of z=0.39, while object 4 is likely to be a galactic star. In total 26 out of 28 GC candidates (93%) turned out to be objects which are dynamically associated with NGC 3115. This low contamination fraction is to be expected for a sample of candidates predominantly selected by morphological criteria from HST data.

Figure 1: Rest-frame spectra of globular clusters corrected to the Lick/IDS resolution and sorted by increasing strength of our mean metallicity indicator [MgFe] (from top to bottom, see also Sect. 4). Note the change in overall continuum shape (metal poorer globular clusters are bluer) and also the change in absorption strength of the Balmer lines (H$\delta$ at 4101 Å, H$\gamma$ at 4340 Å and H$\beta$ at 4861 Å). The slitlet ID for each spectrum is indicated at the right hand side of the plot (see also Table 1).

In order to allow a proper line-strength analysis we restrict our sample to GCs with a mean $S/N\ge 12$ per pixel yielding a final sample of 17 GCs in NGC 3115. Below this signal-to-noise cut the line-strength errors are too large to allow a proper age-metallicity analysis. The mean S/N of our final sample is 27 per pixel (ranging from 12 to 61). The GCs have a mean velocity of 600 km s^-1 (1$\sigma$scatter: 215 km s^-1). The spectra of these 17 GCs broadened to the Lick/IDS resolution are shown in Fig. 1.

We note that the globular cluster 14a (HST27, S/N=11) shows clear [O III] $\lambda 5007$ emission which is however redshifted by $\approx$190 km s^-1 with respect to the recession velocity of the globular cluster itself. Therefore the emission cannot be associated with the globular cluster itself, although the spatial extent of the emission is consistent with the size of the GC on the sky.

Lick indices (for index definitions see Worthey & Ottaviani 1997; Trager et al. 1998) were measured from the resolution and continuum corrected spectra covering a common observed wavelength range of 3670-5500 Å. Uncertainties in the indices were derived by Monte-Carlo simulations which take into account the photon noise, read-out noise of the CCD and the errors in the velocity determination. The Lick system is based on non-flux calibrated spectra, so one expects small offsets (e.g., Kuntschner 2000) in the zero-point of the line indices. Since no stars in common with the Lick stellar library were observed during our observing run we use the offsets established by Vazdekis (1999) for a large flux-calibrated sample of stars ("Jones library'') to transform our index measurements onto the Lick system. Since typical metallicities for GCs are between $-2.0 < \mbox{[Fe/H]} < 0.0$, we calculated the offset by averaging the values for $\rm [Fe/H] = -0.7$ and -0.4 given by Vazdekis (1999 see his Table 2). Note, that some indices (e.g., Mg₂) show strong evidence for a metallicity dependent offset, which can introduce systematic offsets as a function of line-strength. Here in this paper we use mainly indices where the former problem is only a second order effect. The measured index values and their associated errors are listed in Tables 2 and A.1 in the Appendix.

The positions of the GCs with respect to the parent galaxy are shown in Fig. 2. Due to the optimisation of the target efficiency and the observational constraints of the FORS2 instrument the GCs lie in a narrow band $\sim$25 $^{\prime \prime }$ eastwards of the galaxy centre parallel to the major axis of NGC 3115. Both the red and blue sub-samples were evenly distributed across the CCD.

Figure 2: Positions of spectroscopically confirmed globular clusters with respect to the centre of NGC 3115 (triangles represent blue clusters and circles red clusters, see Sect. 3). Approaching and receding globular clusters are indicated by open symbols and filled symbols respectively, while the symbol size indicates the extent of the velocity difference with respect to the mean velocity of the whole sample. The ellipse encloses half of the integrated light and indicates the position and orientation of the main galaxy itself (data from Michard & Marchal 1994). North is up and east to the left.

In Fig. 3 we assess the kinematics of all globular clusters which have recession velocities consistent with being members of the NGC 3115 system. There is a strong signature of GC rotation along the major axis of the galaxy. We confirm here the weak signal of rotation for the red cluster population detected by Kavelaars (1998). In our sample there is no clear difference between red and blue clusters, both show an equally strong signal of rotation. Since our sample of GCs is dominated by clusters close to the location of the major axis (see Fig. 2) we note that the sample is probably biased to globular clusters associated with the disk-formation of NGC 3115.

Figure 3: Assessment of rotation for NGC 3115 globular clusters. The triangles and circles represent blue and red clusters, respectively. Filled symbols indicate the high S/N sample of 17 globular clusters for which we present a line-strength analysis in this paper. Panel  a): The vertical axis shows the radial velocity of the globular clusters relative to the mean velocity of the sample. The horizontal axis is the position angle between a globular cluster and the galaxy centre, where a PA of 0 degrees represents north and positive is east of north. The sinusoid is a simple least-squares fit to the full sample with one iteration to reject outliers. Panel  b): radial velocity is plotted against the projected major axis radius. Positive numbers are towards north-east. The small dots represent the stellar rotation curve of NGC 3115 along the major axis (Capaccioli et al. 1993).