DYNAMICS OF LYMAN BREAK GALAXIES AND THEIR HOST HALOS

The velocities and velocity limits we measure may simply not reflect orbital motion under gravity. Peculiar motion of galaxies or subgalactic clumps in unvirialized systems such as loose groups or very young clusters could play a significant role. Chance near superposition of physically unrelated sources could also lead to erroneous mass estimates. Since many of our targets are indeed clumps of knots or close pairs, these factors may strongly affect our results and interpretation.

To try to assess the significance of the non-virialization scenario, we turn to the "phenomenological" n-body and semianalytic ΛCDM galaxy formation models of van Kampen & Crawford (2007). Their "Model 3" represents a combination of two scenarios: star formation dominated by quiescent star formation in disks, usually just the central galaxy, and star formation dominated by merger-induced starbursts (both galaxy–galaxy and halo–halo mergers). One run of that model produced 13,257 simulated LBGs at z ∼ 3 "detected" in 1 deg² with criteria similar to those used by Steidel et al. (2003). We searched the catalog for projected close pairs of LBGs and examined their characteristics. We find 1206 pairs with separation r < 7''. In Figure 5, we plot the relative velocity Δv in each pair versus the projected separation Δr on the sky, where the velocities are calculated using the two galaxies' redshifts and their peculiar velocities. We focus on the simulated pairs with redshift difference Δz < 0.001 (where "redshift" means cosmological redshift only, without considering peculiar velocity), which we take to be truly physically related. There are 103 such pairs in the simulated catalog, or 8.5% of all pairs with r < 7''; these are represented as open triangles in Figure 5. We also plot the values for the four close pairs in our sample (see Section 4).

Zoom In Zoom Out Reset image size

We calculate for each simulated close pair of LBGs a dynamical mass

$$\begin{equation} M_{\rm dyn} = r_{\rm dyn} v_{\rm rot}^2 / {\rm G}, \end{equation} \tag{ 1 }$$

where v_rot is the maximum observed circular velocity, which in this case we take as Δv, and r_dyn is the radius at which the velocity is measured, in this case the separation between the pair. We also calculate the total (dark matter) halo mass of each pair as the sum of the two individual halo masses listed for the galaxies in the simulation catalog.

Figure 6 shows calculated dynamical mass versus total halo mass for the close pairs in the simulation of van Kampen & Crawford (2007). Surprisingly, there is no strong correlation seen. This implies that the velocities in fact do not (or not always) reflect true virial velocities. They may instead reflect infall velocities of subclumps or galaxies streaming into a potential well; such infall velocities are likely larger than virial velocities by at least a factor of 2. We conclude that, if the simulation of van Kampen & Crawford (2007) is correct, the pairwise velocities in our LBG sample may provide at best weak constraints on the true total masses of our LBG systems.

Zoom In Zoom Out Reset image size

A second caveat is that we may be sampling the kinematics of only a small region, either central to a larger potential well or embedded but not centered in a large massive protogalactic cloud or even a large but UV-faint galaxy (Idzi et al. 2004).

As discussed above in Section 1, several recent groups have used optical and NIR photometry to study the rest-optical SEDs of LBGs in the HDF-N, and found that their stellar masses were typically M < 10¹⁰M_☉ (Papovich et al. 2001; Dickinson et al. 2004; Sawicki & Yee 1998; Rigopoulou et al. 2006; Labbe et al. 2005; Shapley et al. 2005; Huang et al. 2005). These stellar masses provide a firm lower limit to the total dynamical mass.

In the local universe, rotation curves of massive disk galaxies rapidly reach maximum rotational velocity v_max, typically within a few kpc (Sofue & Rubin 2001), similar to the half-light radii of the LBGs studied here. Therefore, if LBG kinematics behave at all like those of local disks then we might not in fact be missing higher rotation velocities from faint extended regions below our detection limit (although we could be missing faint emission at the same velocity from larger radii, which would raise the mass estimates). Similarly, the central velocity dispersions σ of massive local elliptical galaxies are generally excellent indicators of the total galactic virial mass as measured by independent methods (de Zeeuw & Franx 1991). Again this implies that even if LBGs reside in larger systems, the total mass may be well sampled by the LBG kinematics.

Finally, we can compare LBGs at z ∼ 3 to luminous compact blue galaxies (LCBGs) at redshifts z < 1, which have many properties similar to those of LBGs such as small half-light sizes r_1/2 < 3 kpc, high luminosities L ∼ L* (where L* is the characteristic luminosity in a Schechter luminosity function of galaxies today), blue optical colors, diverse, irregular morphologies, asymmetries A ∼ 0.3, SFRs 10 < SFR < 100 M_☉ yr⁻¹, and narrow optical emission lines, σ < 100 km s⁻¹ (Guzmán et al. 1996, 1997, 2003; Phillips et al. 1997; Rix et al. 1997; Pisano et al. 2001; Lowenthal et al. 2005).

Barton et al. (2006) found from deep optical imaging that more than half of the 27 compact narrow emission line galaxies (CNELGs, close cousins of LCBGs) at 0.1 < z < 0.7 in their sample had sizes consistent with small local dwarfs, arguing against massive underlying host galaxies. Similar results derive from studies of nearby compact UV-luminous galaxies using Galaxy Evolution Explorer (Basu-Zych et al. 2007; Heckman et al. 2005; Hoopes et al. 2007). The overall picture is that LCBGs at intermediate redshift are galaxies with luminosity L ∼ L* but masses only 1/10th that of a typical L* galaxy, and dynamical/stellar mass ratios <3.

Constraints on stellar mass do not address the question of extended massive dark matter or gas halos surrounding the LBGs, of course. However, like measurements of local galaxy halo masses using satellite galaxy kinematics (Zaritsky et al. 1993), our constraints on the dynamical masses of close pair LBG systems are insensitive to mass-to-light ratio or stellar population uncertainties.

The third significant caveat regarding our mass estimates is that the Lyα emission line widths we use to estimate dynamical masses of individual LBGs and subclumps are subject to the influence of bulk inflows and outflows of gas, strong absorption by dust, and multiple scattering by neutral hydrogen, all of which can change the line profile. Such effects have been studied theoretically by several groups (e.g., Wolfe 1986; Verhamme et al. 2006; Neufeld 1990; Tenorio-Tagle et al. 2004). Observational evidence includes the systematic asymmetry and redshifting of the Lyα emission line—presumably due to a combination of outflows and absorption by foreground dust—in LBGs and local starbursts (Pettini et al. 2001; Erb et al. 2004, 2006; Forster Schreiber et al. 2006), and Östlin et al. (2009) find that the Lyα emission of local starburst galaxies is spatially more extended than the continuum emission, presumably due to resonant scattering. We searched for but could not find any published or unpublished direct comparison between observed line widths of Lyα and those of Hα or other non-resonance nebular emission lines. Therefore, our mass estimates based on Lyα emission line widths must be regarded as tentative.

Finally, we generally lack constraints on the inclination angle i of our targets, so each mass has associated with it an unknown sin i correction factor. Apart from estimating or measuring inclinations of individual LBGs (as done by, e.g., Bouché et al. 2007), this problem can be addressed only by assuming an average inclination or else assembling a large enough sample of kinematic measurements that all inclinations are statistically well represented.

This relatively bright, elongated source has a disky isophote and captivated our interest as a potential high-redshift disk. Cohen et al. (2000) report for their object No. 2 a redshift z = 0.904, with low confidence (Quality =9, where 11 is the lowest confidence). Our spectrum shows no sign of Mg ii absorption at that redshift in either 2D or 1D, and no C iii (although at 3634 Å, our spectrum has low sensitivity). We conclude that the redshift assignment z = 0.904 seems unlikely.

However, the final coadded spectrum shows weak emission lines at 4499, 5869, and 6045 Å, plausibly matching [O ii]3727, Hβ, and [O iii]5007, respectively, at redshift z = 0.207. We therefore conclude (though with low confidence, given the lines' weakness) that the source is a low-redshift interloper that slipped through our color-selection filter.

This compact, high-surface-brightness source is the brightest target in our survey. Lowenthal et al. (1997) were unable to measure a redshift, but found weak evidence for Mg ii absorption at z = 1.0155 and z = 0.879. Cohen et al. (2000) report z = 0.882 with low confidence (Quality =9, where 11 is the lowest confidence), and cite Cohen et al. (1996) as the redshift source, but we are unable to find the target in that reference. Fernández-Soto et al. (2001) argued that the published spectrum does not support that redshift, and added that their photometric redshift technique favors z_phot = 0.00. Meanwhile, Budavári et al. (2000) report a photometric redshift using HST/WFPC2+NICMOS z_phot = 2.67.

We accumulated 24.5 ks of integration time observing the source with LRIS. Nevertheless, we find no strong emission or absorption lines, and we are unable to obtain an unambiguous redshift for the source, nor can we measure any kinematic signature.

No strong continuum break is visible in our spectrum, which extends down to 3320 Å with continuum detectable by eye in the 2D image down to 3920 Å. Since no strong Lyα emission or absorption is seen redward of that wavelength, we conclude that the emission redshift must be z < 2.17.

We again find weak evidence for Mg ii absorption at z = 1.0155 and z = 0.879, as reported in Lowenthal et al. (1997). Unfortunately, the Mg ii lines for the lower redshift coincide with night sky lines, complicating their detection and measurement. The lower redshift, which is consistent with the z = 0.882 value of Cohen et al. (2000), is supported by possible Fe ii absorption. If real, either or both sets of redshifted absorption lines could be due either to intrinsic absorption from the emitting galaxy itself, or to intervening absorption if the emitting galaxy is at higher redshift. The redshift of the galaxy GOODS J123640.85+621203.4, with a position centroid 22 from hd4_1076_1847, is reported by Cohen et al. (1996) to be z = 1.010. GOODS J123640.85+621203.4 has a highly linear morphology extending at least 15 in the HDF, aligned to within 10° of the separation vector between the two sources. The close agreement between its emission redshift and the tentative Mg ii absorption redshift we find in the spectrum of hd4_1076_1847 suggests that it is either responsible for the absorption, or else associated with another as yet unidentified absorbing galaxy—perhaps hd4_1076_1847 itself.

We are therefore unable to confirm any of the previously published redshifts or derive a new one for this source, apart from the lower and upper limits mentioned above. No kinematic information is available from our spectrum.

Object C4-09 in the catalog of Steidel et al. (1996a) is a remarkable source consisting of four bright knots of emission with nearly identical colors, all within an area barely more than 1'' across. Zepf et al. (1997) investigated and finally rejected the source as a possible gravitational lens system. As reported by Steidel et al. (1996a), the spectrum shows strong emission with two peaks at 5118.7 and 5132.6 Å. Continuum is clearly detected redward of the emission but not blueward, supporting the interpretation of the emission lines as Lyα at redshifts 3.211 and 3.222, respectively, slightly lower than the z = 3.226 reported by Steidel et al. (1996a). Weak absorption lines are seen at 5314, 5489, 5624, and 5633 Å, corresponding to Si ii, O i, C ii, and Si iv, respectively, at an average absorption redshift z = 3.216, between the redshifts of the emission line peaks.

We observed C4-09 with two PAs, one (98°) aligned with the longest axis of the parallelogram of knots, and the other (137°) closer to the short axis. The four knots visible in the HDF image are so close together, with spacings between 057 and 117, that the continuum and the red side of the emission line appear unresolved in both of our 2D spectra. The blue side of the Lyα emission line, however, appears to be slightly extended spatially in the long-axis spectrum, with FWHM = 17, compared with the seeing FWHM = 11, and offset toward the east with respect to the redder line and the continuum. No such spatial extent or offset is visible in the short-axis spectrum. No sign of ordered rotation is visible.

The spatial extent and offset of the blue peak with respect to the red peak and the continuum are consistent with a scenario in which some subset—one, two, or three knots—of the four emission knots is responsible for the blue peak, and another subset at a different velocity gives rise to the red peak. The blue and red Lyα peaks are well fit in the spectral direction by Gaussian profiles with FWHM 5.5 and 7.0 Å, respectively, i.e., unresolved or marginally resolved. After deconvolution with the 5 Å resolution, those widths correspond to <135 km s⁻¹ and 290 km s⁻¹, respectively. The 14 Å separation between the two peaks, however, corresponds to a velocity difference of 820 km s⁻¹. If the wavelength separation is indeed due to radial velocity differences among two or more of the four knots (rather than, e.g., an absorption line superposed on a broad emission line), then we can use it to estimate a system dynamical mass. Given the maximum spatial separation (which will yield maximum mass) of 117, we derive a dynamical mass from Equation (1) M_dyn < 1.3 × 10¹² M_☉.

Alternately, the double-peaked emission profile is also well fit by a single narrow absorption line with λ = 5125.7 Å and FWHM = 8 Å superposed on a broad emission line with λ = 5126.5 Å and FWHM = 14 Å. Such a broad emission line could be caused by strong gas outflows, and the absorption line could be caused by a foreground clump or screen of gas, even dust-free, that resonantly scatters Lyα photons out of the line center. In this scenario, the emission profile FHWM of 800 km s⁻¹ would reflect outflow velocity, perhaps coupled with resonant scattering, rather than circular velocity of a virialized system. The redshifts from the Lyα emission and absorption lines would then be z = 3.2164 and 3.2170, in excellent agreement with the redshift z = 3.216 from the interstellar absorption lines but contrary to the commonly redshifted Lyα emission reported by Shapley et al. (2003) and Erb et al. (2004), which they cite as evidence of outflows in LBGs. Thus an outflow model for C4-09 appears somewhat problematic.

We are not able with the current data to distinguish definitively between the two scenarios. If outflow is indeed responsible for the line profile, then the true dynamical mass is most likely significantly lower than the value derived above.

This source is compact and relatively bright, but continues to elude efforts to obtain a definitive redshift.

Cohen et al. (1996) list a redshift z = 2.268 and Cohen et al. (2000) report a corrected redshift z = 2.500 with high confidence (Quality =2, where 1=highest confidence and 11 = lowest). However, we see no Al ii 1671 Å absorption matching either redshift, only possible O i 1302 Å/Si ii 1304 Å and possible N v 1243 Å in P-Cygni profile matching z = 2.500; no C iv 1550 Å is detected. Budavári et al. (2000) give a photometric redshift z_phot = 1.44.

Continuum is clearly detected in our spectrum with a spatial profile matching an unresolved PSF. No strong emission or absorption lines are detected. Weak emission lines may be present at 5925 and 5976 Å. If these were [O iii] at z = 0.194, we would expect Hβ at 5804 Å and/or [O ii] at 4450 Å, but neither is seen. Possible weak absorption lines appear at 4566, 4678, 5069, and 5447 Å, but none is well detected, nor were we able to discern a pattern among them indicating a source redshift. We find no support for the tentative redshift z = 2.04 reported by Lowenthal et al. (1997) based on cross-correlation, apart from the possible line at 5069 Å, which would correspond to Al ii 1671 Å.

Without solid absorption or emission line features available, we also find no kinematic indicators in our spectrum.

hd4_0298_0744 is another elusive target. Cohen et al. (2000) report z = 2.801 with high confidence, while Budavári et al. (2000) find z_phot = 1.76. Continuum is clearly visible in the 2D spectrum even before sky subtraction down to the blue limit of our spectrum, with no strong break visible. No strong emission or absorption features are seen in the final 2D or 1D extracted spectrum. We were therefore unable to confirm the redshift reported by Cohen et al. (2000), which we would have expected to produce a strong Lyα emission or absorption line or continuum break visible at 4621 Å. Because no strong continuum break is visible down to 3900 Å in our spectrum, we conclude that the redshift is most likely z < 2.2.

We cannot extract any dynamical information on the source.

C4-06 from the catalog of Steidel et al. (1996a) is one of the brightest, largest sources in our sample. The HDF image shows a linear structure—dubbed the "Hot Dog Galaxy" (Bunker et al. 1998; Bunker 2001)—roughly 1'' × 3'' in extent, with two major knots and several subknots. The LRIS mask slit was aligned along that structure.

Strong, spatially resolved continuum is clearly detected in the 2D spectrum, although it appears as a single spatially continuous source, rather than the separate knots seen with HST resolution. The spatial extent of the continuum emission, measured by compressing the 2D spectrum into a 1D spatial profile, is 32, consistent with the HDF image.

No emission lines are visible, but several strong absorption lines are detected matching redshifted Lyα, Si ii, O i, C ii, and Si iv. The Lyα line is broad (78 Å FWHM), while the other lines are unresolved. Additional marginal detections include Si ii, N i, and Si iii. The redshift derived from averaging the strong narrow absorption line redshifts is z = 2.794, while cross-correlating the spectrum with that of a template average of 12 LBGs yields z = 2.802, similar to the redshift z = 2.803 reported by Steidel et al. (1996a).

The narrow absorption lines show no sign of any velocity shift across the extended continuum (see Figure 3). To quantify any subtle velocity gradient, we extracted three 1D spectra at the middle and the two ends of the linear source and intercompared them. The central extraction aperture was 12 pixels (26) wide and the apertures at the two ends were 6 pixels (13) wide, centered 8 pixels apart, so that the two end apertures had no overlap.

To constrain the velocity shift between the two end apertures, we selected regions of the spectra containing strong absorption lines and free from residual noise from strong sky emission lines and cross-correlated them. The cross-correlation yielded a strong peak at δλ = 0.8 Å. Assuming a minimum relative wavelength accuracy of 1.0 Å, the velocity shift between the two apertures is then δv < 60 km s⁻¹, consistent with zero within our measurement errors. The dynamical mass limit implied by the observed constraint on line-of-sight velocity gradient δv < 60 km s⁻¹ over 322 is M_dyn < 2.6 × 10⁹M_☉, the lowest dynamical mass estimate in our sample (modulo, of course, the unknown inclination and three-dimensional morphology of the source).

We interpret the lack of observed velocity gradient over more than 20 kpc as evidence that we are observing not an edge-on disk with ordered rotation, but rather a truly linear, perhaps filamentary source. The object may of course be collapsing or even expanding perpendicular to the line of sight; collapse is more natural under most current galaxy formation scenarios.

This galaxy appears as a very close pair in the HDF image, with a separation of only 03. Cohen et al. (2000) report a redshift for the source z = 2.969. Our mask slitlet was aligned along the same PA, 73°, as the two emission knots; however, 1'' ground-based seeing certainly blurs the two into a single unresolved source in our LRIS observations.

The spectrum shows very strong emission at 4823.2 and weaker emission at 4812.3 Å. The stronger line appears extended both spatially and spectrally: the spatial FWHM of the line is 12, comparable to the seeing, but the emission profile's faint wings extend to cover 33. The spectral FWHM is 7.0 Å, compared to the resolution of 5 Å, and the total detected emission spans 26.4 Å, including the two peaks and their wings.

The weaker line's spatial position centroid is offset 025 from that of the brighter line, closely matching the separation of the pair in the HDF image.

Weak continuum is detected both redward and blueward of the emission lines, but appears to be stronger on the red side. No other emission or absorption features are seen.

It is highly unlikely that this double emission line corresponds to an optical line emitted by a low-redshift (z < 1) system. The [O ii] doublet would, at z = 0.293, be separated by 3.5 Å, not the 11 Å we observe. Any [O iii] or Balmer line should be accompanied by other optical lines as well. We conclude that the most likely interpretation is Lyα at z = 2.967, confirming with only a slight revision the redshift reported by Cohen et al. (2000).

The double emission line profile may be due to a single emission line with strong absorption superposed by an intervening cloud or galaxy. Alternatively, as for C4-09, it may be caused by radial velocity differences between the two emitting knots. Given the close match between the spatial separations in the image and in the spectrum, we adopt the latter scenario. The velocity difference between the emission peaks then corresponds to 677 km s⁻¹, while the entire 26.4 Å span of the emission line corresponds to ∼1650 km s⁻¹, and the spectral FWHM of 7.0 Å corresponds, after deconvolving with the 5 Å instrumental resolution, to 4.9 Å, or ∼300 km s⁻¹.

To derive a mass estimate for the system, we assume each knot in the pair emits one emission line, and we therefore adopt r = 03 (2.3 kpc) and σ = 677 km s⁻¹ to obtain M = 2.4 × 10¹¹ M_☉. The large total range in observed emission velocity, 1650 km s⁻¹, may be due to bulk gas outflow induced by merging of two or more subunits and/or by the subsequent starburst.

Despite the relative brightness of this target, its redshift remains elusive. Lowenthal et al. (1997) reported a tentative redshift z = 2.47 based on a few possible absorption lines. Thompson (2003) estimated a photometric redshift z_phot = 2.80, while Budavári et al. (2000) give none. The source is smooth, slightly extended, and elongated in the HDF image. Continuum is easily detected along the entire length of our spectrum. However, no emission lines are visible, nor are any strong absorption features. There are possible absorption lines at 4386, 4528, 4794, 4816, 5168, and 5573 Å, but we were unable to discern any convincing pattern matching our redshifted template spectra or our previous tentative redshift z = 2.47. Cross-correlation with the template spectra likewise revealed no robust redshift. The lack of a strong continuum break redward of 4000 Å implies z < 2.29, contrary to the photometric redshift z_phot = 2.80 calculated by Thompson (2003).

With no strong absorption or emission features, we were also unable to derive any kinematic information from the spectrum.

The HDF image shows the source hd4_0367_0266 to be an elongated object with two bright knots separated by 06 and an extended tail terminating in a fainter knot, all nearly aligned. The redshift of z = 2.931 reported by Lowenthal et al. (1997) was based on several absorption lines. The total extent of the source shown in the HDF is 23; the LRIS mask slitlet was aligned to cover all three knots as well as the tail.

Our spectrum shows weak extended continuum emission with FWHM ∼ 34. No emission lines are detected. Several possible weak absorption lines appear in the extracted 1D spectrum at 4956, 5117, and 5478 Å. These lines would correspond to Si ii, O i, and Si iv, respectively, at the previously reported redshift z = 2.931. A possible broad absorption trough at 4757 Å could be a damped Lyα absorber at z = 2.913. Overall our confidence in the redshift is somewhat lower than reported in Lowenthal et al. (1997), and we downgrade the redshift quality to Q_z = 3.

To search for velocity gradients or shifts across the galaxy, we extracted two spatially independent 1D spectra centered 24 apart and cross-correlated them, using only the clean regions of the spectra that were free from sky noise residuals and showed evidence of absorption features. The cross-correlation function displays a weak peak consistent with δv = 0, but unfortunately the S/N was too low to allow any robust measurement of kinematics.

The source hd2_1928_1041 (C2-06 in the catalog of Steidel et al. 1996a) is separated by only 20 spatially from C2-05, and we consider them here as a close pair. Each of the galaxies is a compact source with one bright knot and a small cloud of extended emission as seen in the HDF. Steidel et al. (1996a) report z = 2.845 for C2-05, but this is revised to z = 2.005 by Cohen et al. (2000). For hd2_1928_1041, Papovich et al. (2001) report z = 2.009, citing Lowenthal et al. (1997); we here slightly revise that redshift.

Both targets are covered in a single LRIS slitlet on our slit masks. Strong, unresolved continuum is detected from both objects. No emission lines are visible, but numerous absorption lines corresponding to interstellar species such as Si ii, O i, C ii, Si iv, C iv, and Al ii are well detected in the two spectra, corresponding to redshifts z = 2.005 (hd2_1928_1041) and z = 2.008 (C2-05). The velocity difference implied by the redshifts is δv = 390 km s⁻¹. Cross-correlation of absorption line regions in the two 1D extracted spectra produces a strong peak at δv = 364 ± 87 km s⁻¹ (where the error is estimated by dividing the FWHM of the cross-correlation peak by 10), consistent with the simple redshift difference.

The observed velocity difference and spatial separation provide a dynamical mass estimate M_dyn ∼ 5.8 × 10¹¹ M_☉ for the two-source system.

A highly elongated, multiple-knot source embedded in extended emission, extending over at least 26 along the long dimension, to which our LRIS slitlet was aligned.

Spectroscopic redshifts reported in the literature include z_spec = 2.72 (Lowenthal et al. 1997) with low confidence and z_spec = 1.980 (Cohen et al. 2000) with high confidence (Quality =3, where 1=best). Fernández-Soto et al. (1999) cite z_spec = 2.002, but we could not find the source of that measurement in the literature. Reddy et al. (2006b) report z_spec = 1.872 based on absorption lines. Wirth et al. (2004) reported that a redshift was not measurable in their spectrum of the source.

Photometric redshifts calculated for hd2_1739_1258 include z_phot = 1.640 (Fernández-Soto et al. 1999), z_phot = 2.854 (Thompson 2003), and z_phot = 1.34, 2.07 (Budavári et al. 2000).

Strong, extended continuum emission is detected in our 39 ks integration with a spatial FWHM∼ 9 pixels, or 19 and a total extent of 39. The continuum dies below about 4075 Å, implying z ∼ 2.35 if the dip is due to Lyα blanketing. However, we also detect two possible weak absorption lines at 5305, 5318 Å, which match Mg ii at z = 0.897. These absorption lines could be due to intervening gas, but they could also be intrinsic to the target. We detect no features that support either the redshift z = 1.980 (Cohen et al. 2000) or z_spec = 2.72 (Lowenthal et al. 1997). Given the lines we do detect weakly, we tentatively conclude that hd2_1739_1258 in fact lies at z = 0.897 rather than at z>2.

We extracted two spatially independent 1D spectra from the extended 2D spectrum and cross-correlated them in an attempt to measure velocity gradients across the source, but no statistically significant cross-correlation signal was detected.

This galaxy consists of a single compact core with two short wings of extended emission. Cohen et al. (2000) report z = 2.237.

Our LRIS slitlet was aligned with the long axis of the system. The spectrum shows a single weak emission line at 3934.2 Å with FWHM ∼ 6 Å, i.e., barely resolved, with clear continuum redward but none blueward. Interpreting the line as Lyα, which is supported by the continuum break, we derive a redshift z = 2.236. No other strong emission lines are detected, although possible weak C iv and Al ii absorption and He ii emission are visible. The width of the line corresponds to a deconvolved velocity width of 240 km s⁻¹, which we adopt as an upper limit.

Neither the continuum nor the strong emission line shows any spatial extent in our 2D spectrum, so we are unable to constrain kinematics for this system.

C3-02 is a small source with a single compact bright knot of emission and a diffuse tail extending barely 1'', along which our LRIS slit was roughly oriented. Steidel et al. (1996a) reported a tentative redshift z = 2.775, Fernández-Soto et al. (1999) report a photometric redshift z_phot = 1.720, and Budavári et al. (2000) report z_phot = 2.218.

Spatially unresolved continuum is easily detected with average S/N ∼11 per resolution element over the entire length of our 25.5 ks spectrum, which extends from 3650 to 6285 Å. No emission lines are visible. Several possible weak absorption lines appear at 4317, 4663, and 4608 Å, the first of those being the strongest (EW_obs ∼ 8 Å). None of those wavelengths matches features expected from galaxies at z = 2.775. Both the photometric and spectroscopic redshifts previously reported therefore seem unlikely to be correct. There is a possible match to O i1302, Si ii1304, C ii 1335/1336, and Si iv 1402/1403 at z = 2.316. No Lyα emission is visible at the expected wavelength for that redshift, but the continuum shape is consistent with a moderate break there. Given the weakness of the lines, we adopt this redshift only tentatively (Q_z = 2). Kinematic measurements of this source are also impossible from our spectrum.

hd2_0698_1297 consists of two distinct knots embedded in an asymmetrical blob of diffuse emission. The knots are 05 apart, and the diffuse emission visible in the HDF F814W image extends for 12. The system is 27 from hd2_0705_1366, an unresolved point source (measured FWHM = 014) in the HDF image.

Because of slit mask orientation constraints, we observed this pair at two different PAs: for 25.5 ks at 42°, and for 10.2 ks at 58°. Due to the close proximity of the two sources, the 11 wide slitlets at both PAs include emission from both targets.

Both spectra show weak continuum with S/N ∼ 1 per resolution element and a single emission line from each source. The emission line from hd2_0698_1297 is stronger in the spectrum with PA = 42°, while that from hd2_0705_1366 is stronger at PA = 58°, presumably because of slit placement with respect to each source. We measure the wavelengths of the emission lines from the stronger spectrum in each case: 5396 Å (hd2_0698_1297) and 5312 (hd2_0705_1366), each with FWHM = 8 Å, corresponding for redshifted Lyα to z = 3.439 and 3.370, respectively, within δz < 0.01 of the values reported in Lowenthal et al. (1997). There is evidence for an absorption trough bluewards of the emission line in hd2_0698_1297, supporting our identification of the line as Lyα. We cannot confidently rule out the possibility that each emission line is in fact spectrally unresolved, so we adopt the measured FWHM as an upper limit in each case.

The continuum and the emission lines are all spatially unresolved, except perhaps continuum from hd2_0698_1297, which shows a slight spatial extent FWHM∼ 15, although this may be due at least partly to small spatial registration errors in co-adding the 18 individual exposures.

Given the wavelength difference 84 Å and projected separation 27 between the two sources, and assuming the wavelength difference to reflect gravitationally induced dynamics, we derive a system dynamical mass M_dyn = 1 × 10¹⁴M_☉, the largest estimated dynamical mass in our sample.

See hd2_0698_1297 above.

See hd2_0624_1688 below.

This triple-knot compact source is part of a close pair with hd2_0529_1567, 625 away. Both were covered by the slitlet at PA = 289 that targeted hd2_0725_1818 and hd2_0743_1844 (see below). hd2_0624_1688, for which Lowenthal et al. (1997) reported z = 2.419, was in our original kinematic sample target list, while hd2_0529_1567 was not: at I_814,AB = 27.233, it offered little hope of providing a useful spectrum.

However, our 2D spectrum shows clear continuum and a single strong emission line from each source. The continuum emission from hd2_0624_1688 shows a significant drop blueward of the emission line. Continuum from the fainter hd2_0529_1567 is of course much weaker, and the proximity of the emission line to the blue end of the spectrum prevents us from assessing any continuum break across the line.

The single emission line from hd2_0624_1688 lies at 4162.7 Å, yielding redshift z = 2.424 assuming that the emission line is Lyα, compared with z = 2.419 reported by Lowenthal et al. (1997). No other absorption or emission lines are visible, and Lyα is the only plausible interpretation. The emission line width is FWHM = 6.3 Å, corresponding to 3.8 Å or ∼275 km s⁻¹ after deconvolution with the 5 Å resolution, and we adopt that value as an upper limit.

The emission line from hd2_0529_1567 lies at 3973.3 Å, yielding z = 2.268, again under the assumption that the line is redshifted Lyα. The deconvolved line width is FWHM = 3.5 Å or ∼ 270 km s⁻¹, which we adopt as an upper limit. No other absorption or emission features are seen.

Together with the projected separation of 625 (50 kpc), the observed velocity difference of 1394 km s⁻¹ yields a dynamical mass estimate for the pair system M_dyn = 2.2 × 10¹³M_☉.

This small (r_1/2 = 026) source lies only 13 from hd2_0743_1844. Lowenthal et al. (1997) reported a spectroscopic redshift z = 2.233 for hd2_0725_1818 based on a Lyα emission line, but only a tentative redshift z = 2.39 for hd2_0743_1844. The two sources have similar colors, with hd2_0743_1844 bluer by only 0.27 mag in B₄₅₀ − I₈₁₄ and redder by only 0.05 mag in U₃₀₀ − B₄₅₀, less than the spread in the colors of LBGs reported by Papovich et al. (2005); hd2_0743_1844 is fainter by 0.8 mag in I₈₁₄ (all measurements from the catalog of Williams et al. 1996).

Our LRIS spectrum covers not only both of those sources, but also a spiral galaxy to the north with a redshift z = 1.148 reported by Phillips et al. (1997) and, to the south, hd2_0624_1688 and hd2_0529_1567 (see above).

Continuum from both galaxies is easily visible in the 2D spectrum, but with some overlap of their spatial profiles, given their mere 13 separation. The flux from hd2_0725_1818 is two to three times stronger than that from hd2_0743_1844, consistent with their photometric measurements. Several absorption features are apparent in the 2D image; while they seem to span spatially the continuum from both sources, the weakness of the fainter source makes this appearance difficult to confirm visually.

We extracted a 1D combined spectrum of the pair. To assess their redshifts and velocities independently, we also extracted two separate 1D spectra in the following way: first, we extracted a 1D spectrum of the brighter source, hd2_0725_1818, using only those pixels in the 2D spectrum extending spatially from the peak of the source away from hd2_0743_1844, i.e., the side of the spatial profile that is less contaminated by flux from the fainter source. We then used this 1D extraction to model the entire 2D spectrum and subtracted the scaled model from the 2D image. This left a clean, isolated spectrum of hd2_0743_1844, which we then extracted to 1D. We repeated the process in reverse, modeling and subtracting the 2D spectrum of hd2_0743_1844 to produce a clean, isolated and uncontaminated spectrum of hd2_0725_1818.

The absorption lines in the 1D spectrum of hd2_0725_1818 are well matched by interstellar Si ii, C ii, and Al ii, and stellar Si iv, C iv at z = 2.232, thus confirming and slightly revising the redshift reported by Lowenthal et al. (1997). We also find a tentative detection of C iii in emission, similar to their earlier result. Lyα lies at the blue end of the spectrum, which is clean but low S/N; no emission or break is visible in either 2D or 1D.

The spectrum of the fainter source, hd2_0743_1844, also shows absorption corresponding to Si ii, C iv, and O i. No Al ii is detected, though the S/N is low in that part of the spectrum. The implied redshift is z = 2.230 ±  0.004, consistent with that of hd2_0725_1818, and we adopt that redshift with quality Q_z = 3 rather than 4 given the source's extreme faintness.

To measure any small velocity difference between the two sources, we cross-correlated the independent 1D spectra against each other, using only those parts of the spectra free of bright sky emission line residuals. The result was a strong correlation peak (correlation strength = 0.2) at δv = 9 km s⁻¹, consistent with δv = 0 km s⁻¹ within our nominal velocity resolution of 60 km s⁻¹. For comparison, we also cross-correlated each 1D spectrum with a 1D spectrum of blank sky extracted from the same 2D image. No strong correlation peak is seen anywhere near δv = 0, with a maximum of only 0.017 within 3000 km s⁻¹ of δv = 0, and a maximum correlation strength 0.1 for the whole spectrum. We thus conclude that the correlation between hd2_0725_1818 and hd2_0743_1844 is significant, and that they have identical redshifts within our uncertainties.

Adopting an upper limit δv < 60 km s⁻¹ and the projected separation 13 or 10.9 h⁻¹ kpc, we then derive a dynamical mass upper limit M_dyn < 8.6 × 10⁹M_☉, the lowest derived mass in our sample.

See hd2_0725_1818 above.

An elongated, wispy source consisting of two clumpy wings extending over 21, with a 05 gap separating them, and aligned roughly east–west. The source's total isophotal colors in the HDF Version 1 catalog satisfy the B-dropout criteria of Lowenthal et al. (1997). However, in the HDF Version 2 catalog, the two sides' colors are significantly different, with the western side bluer by 1.30 mag in B_450,AB − V_606,AB and 0.52 mag in V_606,AB − I_814,AB. Furthermore, in the Version 2 catalog photometry, neither component individually satisfies either the U-dropout or the B-dropout criterion.

Our LRIS slit covered both wings, as well as the source hd3_1824_1945 74 away and two other sources, one on either end of the slit. Cohen et al. (2000) report z = 4.050 for the source.

Only extremely very faint continuum is detected at the position of hd3_1633_1909 in the 2D spectrum, with no absorption lines discernible. A single emission line appears, however, at 6145 Å. The emission line, though apparently spectrally resolved with FWHM = 8.4 Å, is not well fit by the [O ii] doublet at z = 0.649; the peaks of the two components would be separated by 4.1 Å, while the line in the spectrum is well fit by a single Gaussian. If the line is Lyα, the source's redshift is z = 4.056, the highest redshift in our sample and consistent with the redshift reported by Cohen et al. (2000). The resolution-deconvolved velocity width of the emission line is FWHM = 330 km s⁻¹.

No spatially extended kinematic information is available from our spectrum.

This compact double-knot source was covered by the same slit that included hd3_1633_1909 (see above), 74 away. Cohen et al. (2000) report z_sp = 2.050 (object 145 in their online list).

Our spectrum shows strong continuum with spatial FWHM = 16 extending without any obvious break virtually to the blue limit of the spectrum at 3750 Å. No strong emission lines are visible; only a few weak possible absorption lines are present. We see no compelling evidence to support the redshift z = 2.050 reported by Cohen et al. (2000); for example, Al ii 1671 Å, C iv, and Si iv, which often appear in strong absorption in LBG spectra, are not seen.

The lack of a strong Lyα continuum break down to 3750 Å implies z < 2.2, in conflict with the photometric redshifts z_phot = 2.30 calculated by Fernández-Soto et al. (2001) and z_phot = 2.395 by Budavári et al. (2000). Meanwhile, the lack of strong emission lines including [O ii] 3727 Å implies z>0.75.

No kinematic information is available from our spectrum.