1: \documentclass[preprint,12pt]{aastex}
2: \usepackage{natbib}
3:
4: \newcommand\dunits{h^3\;{\rm Mpc}^{-3}\;{\rm mag}^{-1}}
5: \newcommand\Rab{R_{\rm AB}}
6:
7: \shorttitle{Lyman Break Galaxy Candidates in SDSS DR1}
8: \shortauthors{Bentz, Osmer, \& Weinberg}
9:
10:
11: %\received{2003 September 23}
12: \begin{document}
13: \title{Bright Lyman Break Galaxy Candidates in the Sloan Digital Sky Survey
14: First Data Release}
15: \author{Misty C. Bentz, Patrick S. Osmer, and David H. Weinberg}
16: \affil{Department of Astronomy, The Ohio State University}
17: \affil{140 W. 18th Ave, Columbus, OH 43210-1173}
18: \email{bentz,posmer,dhw@astronomy.ohio-state.edu}
19:
20: \begin{abstract}
21: We report the discovery of six compact, starburst galaxy candidates with
22: redshifts $2.3 < z < 2.8$ and $r$-band magnitudes 19.8$-$20.5 in the
23: Quasar Catalog of the SDSS DR1. The SDSS spectra resemble the composite
24: spectrum of Lyman Break Galaxies (LBGs) at $z \approx 3$ (albeit with
25: some differences and broader spectral lines), but the objects are $4-5$
26: magnitudes brighter than an ``$L_*$'' LBG. These objects could be
27: extremely luminous LBGs, normal LBGs amplified by gravitational lensing,
28: or a rare class of BAL quasars. In the first case, star formation rates
29: inferred from the UV continuum luminosities, with no correction for dust
30: extinction, are $\sim$ 300$-$1000 ${\rm M}_{\odot}$ yr$^{-1}$, similar
31: to those of ultraluminous infrared galaxies, but in these UV-bright
32: objects the star formation is evidently not obscured by high dust column
33: densities. The SDSS images show no evidence of multiple imaging or
34: foreground lensing structures. The spectra have fairly broad absorption
35: features and prominent high-ionization absorption but do not show
36: high-ionization emission lines and do not resemble known BAL quasars. A
37: rough estimate of the luminosity function of the objects is above a
38: Schechter function extrapolation from LBGs at fainter magnitudes.
39: Improved optical spectra and observations at X-ray, IR, and sub-mm
40: wavelengths could help determine the nature of these remarkable objects.
41: \end{abstract}
42: \keywords{galaxies: starburst --- surveys}
43:
44:
45: %\keywords{galaxies: starburst --- surveys}
46:
47: \section{INTRODUCTION}
48: One of the most important recent developments in observational cosmology
49: is the discovery of a large population of star-forming galaxies at $z
50: \approx 3$ (e.g., \citealt{1996ApJ...462L..17S};
51: \citealt{1997ApJ...481..673L}). These so-called Lyman break galaxies
52: (LBGs) can be identified by their characteristic colors in deep imaging
53: surveys. Spectroscopic surveys of LBGs have transformed our
54: understanding of the history of cosmic star formation (e.g.,
55: \citealt{1996MNRAS.283.1388M}; \citealt{1999ApJ...519....1S}) and of
56: galaxy clustering in the early universe (e.g.,
57: \citealt{2003ApJ...584...45A}). At $\Rab \approx 24.5$, the surface
58: density of LBGs is $\sim 1$ arcmin$^{-2}$, and the luminosity function
59: is well described by a \citet{1976ApJ...203..297S} function with a
60: characteristic luminosity corresponding to an R$_{AB}$ apparent
61: magnitude $m_* = 24.54$ at $z \approx 3 $ (Adelberger \& Steidel 2000).
62:
63: Most known LBGs have been discovered in deep images covering small areas
64: of the sky. The survey by \citet{2003ApJ...592..728S}, for example,
65: contains more than 2000 spectroscopically confirmed galaxies, but the 17
66: fields in the study cover only $\sim$ 0.38 deg$^2$. The
67: brightest galaxies in this survey have $\Rab \approx 23$. The only
68: substantially brighter object known, MS 1512-cB58 with $\Rab \approx
69: 20.4$, was discovered serendipitously by \citet{1996AJ....111.1783Y} as
70: part of the CNOC-1 redshift survey, and is now understood to be
71: magnified by a factor $\sim 30$ as a result of gravitational lensing by
72: a foreground galaxy cluster \citep{1998MNRAS.298..945S}. Detailed
73: studies of the stellar populations and interstellar gas properties of
74: LBGs rest either on this single object (e.g.,
75: \citealt{2000ApJ...533L..65T}; \citealt{2000ApJ...528...96P}) or on
76: composite spectra constructed from many individual galaxies
77: \citep{2003ApJ...588...65S}. The limited area of existing LBG surveys
78: also means that the very bright end of the LBG luminosity function is
79: unconstrained.
80:
81: The imaging survey of the Sloan Digital Sky Survey (SDSS,
82: \citealt{2000AJ....120.1579Y}) complements existing LBG imaging surveys
83: by covering a much wider area to a brighter limiting magnitude.
84: Furthermore, because the SDSS quasar selection algorithm
85: \citep{2002AJ....123.2945R} identifies unresolved sources with non-stellar
86: colors, rather than just objects with expected quasar colors, it {\it
87: also} selects the most luminous LBGs as quasar candidates and targets
88: them for spectroscopy. In a search for quasars with anomalously low
89: \ion{C}{4} emission, \citet{paper1} found one example of an apparently
90: luminous, $z \approx 2.5$ starburst candidate that had been classified
91: as a quasar by the SDSS spectroscopic pipeline in the Early Data Release
92: (EDR, \citealt{2002AJ....123..485S}) Quasar Catalog
93: \citep{2002AJ....123..567S}. In this {\it Letter},
94: we describe a detailed search through the SDSS First Data Release (DR1, \citealt{dr1preprint})
95: and the discovery of five additional objects with similar properties.
96: These objects have intriguing properties that necessitate a variety of
97: additional observations, and, if confirmed as starbursts, they would
98: provide the first constraints on the space density of extremely luminous
99: (in rest-frame UV) star-forming galaxies at high redshifts.
100:
101: \section{SPECTRAL ANALYSIS}
102: As described by \citet{dr1qsopreprint}, the SDSS DR1 Quasar Catalog
103: covers $\sim$ 1360 $\rm{deg^2}$ of the sky and contains 16,713 objects
104: with rest-frame absolute magnitudes $M_i < -22$ (for $h \equiv H_0/ 100
105: \;{\rm km}\;{\rm s}^{-1}\;{\rm Mpc}^{-1} = 0.7$, $\Omega_M = 0.3$,
106: $\Omega_{\Lambda} = 0.7$, and a power law (frequency) continuum index of
107: $\alpha_Q=-0.5$), at least one emission line with a FWHM larger than
108: 1000 km s$^{-1}$, and reliable redshifts. For this investigation, we
109: searched the SDSS DR1 Quasar Catalog for high-$z$ starburst galaxies
110: that might have been classified as quasars, limiting the catalog to $z >
111: 2.3$ in order to probe the high-$z$ regime of traditional LBG surveys.
112: Color cuts were then applied to the 1658 objects satisfying the redshift
113: cut. Although the filter systems are different, those objects with
114: optical colors satisfying
115: \begin{equation}
116: g - r \leq 1.2,\ u - g \geq g - r + 1
117: \end{equation}
118: are analogous to the objects that would be detected by the LBG color
119: criteria of \cite{2003ApJ...588...65S}. A total of 591 objects met the
120: color criteria, and they were all visually inspected for absorption line
121: features similar to those in the LBG composite spectrum of
122: \citet{2003ApJ...588...65S}, and for the absence of characteristic
123: quasar emission lines from highly ionized species. Six objects
124: (including the original object discovered in the EDR by
125: \citealt{paper1}) displayed these spectral signatures of a starburst
126: galaxy. Table 1 lists the six candidate objects and their properties.
127: Figure 1 shows the individual spectra of the six candidates taken with
128: the 2.5$-$m SDSS telescope alongside the composite spectrum of 811
129: galaxies obtained with the 10$-$m Keck telescope by
130: \citet{2003ApJ...588...65S}.
131:
132: The six candidates are red relative to normal quasars, but because they
133: have non-stellar colors and are unresolved at the resolution of SDSS
134: imaging, they meet the SDSS selection criteria for high-redshift quasar
135: candidates \citep{2002AJ....123.2945R} and were targeted for
136: spectroscopy. We checked the SDSS DR1 spectral database to confirm that
137: there were no objects classified as galaxies with $z > 2.3$ and no
138: high-$z$ quasar candidates satisfying our selection criteria that might
139: have been rejected from the Quasar Catalog because they failed to pass
140: the 1000 km s$^{-1}$ emission line cut.
141:
142: Table 1 lists the redshifts and broad-band photometric properties of the
143: six candidates, taken from the SDSS DR1 Quasar Catalog
144: \citep{dr1qsopreprint}. All six objects are 4-5 magnitudes brighter
145: than an $L_*$ LBG, and 2-3 magnitudes brighter than the brightest LBG
146: detected by \citet{2003ApJ...592..728S}.
147:
148: \section{LENSED LBGS, ULTRALUMINOUS LBGS, OR BAL QUASARS?}
149: From the absence of emission lines of highly ionized species, such as
150: \ion{N}{5} and \ion{C}{4}, it is obvious that these objects are not
151: normal AGN. But are they normal LBGs that have been amplified by
152: gravitational lensing, extremely luminous LBGs, or unusual broad
153: absorption line (BAL) quasars?
154:
155: The objects were found by searching for absorption and Ly$\alpha$
156: features similar to those of the LBG composite, so by definition they share
157: many spectral similarities. There is strong evidence in most of the
158: candidates for absorption from \ion{Si}{2} $\lambda$1260, \ion{C}{2}
159: $\lambda$1334, \ion{Si}{4} $\lambda\lambda$ 1393,1402, \ion{C}{4}
160: $\lambda\lambda$1548, 1550, and \ion{Al}{3} $\lambda\lambda$1854, 1862.
161: In a few of the candidates, there is even noticeable absorption from
162: \ion{O}{1} $\lambda$1302, \ion{Si}{2} $\lambda$1304, \ion{Si}{2}
163: $\lambda$1526, \ion{Fe}{2} $\lambda$1608, and \ion{Al}{2} $\lambda$1670.
164: The precedent of MS 1512-cB58 suggests that the high apparent brightness
165: of these objects could be a consequence of lensing. We inspected the
166: SDSS images of the six candidates for evidence of multiple images or
167: possible foreground galaxies or clusters, but all of the systems are
168: consistent with lone, unresolved point sources with unexceptional
169: foreground environments. We also counted the numbers of objects
170: classified as extended in the DR1 imaging catalog within a 5\arcmin\
171: radius of five of the six objects,\footnote{We were not able to carry
172: out the counts for J1553+0056 because of a difficulty with the
173: database.} and similarly for control fields offset by $1\degr - 4\degr$.
174: The counts did not show any statistically significant excess of extended
175: sources around the candidates, except possibly for SDSS J1444+0134. The
176: SDSS data are not sufficient to rule out sub-arcsecond image splitting
177: or the presence of foreground systems at likely lens redshifts of $z
178: \sim 1$.
179:
180: At present, the main evidence against the lensing hypothesis comes from
181: the differences between the LBG composite spectrum and the spectra of
182: the galaxy candidates. The candidates are redder than most LBGs, and
183: most of their absorption features are broader, as are their Ly$\alpha$
184: profiles.\footnote{The FWHMs of the Ly$\alpha$ emission lines range
185: from $\sim 1100-3000$ km s$^{-1}$. SDSS J1340+6344 and SDSS J1432-0001
186: have fairly symmetric Ly$\alpha$ profiles, while the LBG composite has a
187: P-Cygni type asymmetry suggestive of outflows.} While we would not
188: expect such differences if these were normal LBGs, they might arise if
189: the objects have extremely high star formation rates (SFRs). If we
190: assume that lensing and AGN contributions are unimportant, we can follow
191: the standard practice of estimating SFRs from the continuum luminosity
192: at $\lambda$1500 \AA, which is produced mainly by young O and B stars.
193: We estimate the rest-frame $\lambda$1500
194: \AA\ flux directly from the calibrated spectra and convert to luminosity
195: assuming a cosmological model with $\Omega_M=0.3$, $\Omega_\Lambda=0.7$,
196: and $h=0.7$. Assuming a Salpeter IMF with mass limits of 0.1 to 100
197: M$_{\odot}$ and a 10$^8$ yr old continuous star formation model,
198: \citet{Kennicutt1998} derives the relation
199: \begin{equation}
200: \rm SFR\ M_{\odot}\ yr^{-1} = 1.4 \times 10^{-28} L_{\nu}(\lambda 1500)
201: \end{equation}
202: between the star formation rate and the luminosity density $L_\nu$ (in
203: ergs s$^{-1}$ Hz$^{-1}$) at $\lambda=1500$\AA. The implied SFRs are
204: tabulated in Table 2, along with the measured flux in the observed frame
205: at $\lambda$1500 \AA\ $\times (1+z)$. As expected for objects as bright
206: as these, the SFRs are very high, ranging from $\sim$ 300 M$_{\odot}$
207: yr$^{-1}$ to over 1000 M$_{\odot}$ yr$^{-1}$.
208:
209: The values in Table~2 assume no dust extinction and are therefore lower
210: limits to the ``true'' SFRs, at least if the assumed IMF is correct.
211: Lyman break galaxies in typical spectroscopic surveys have similarly
212: estimated SFRs in the range $\sim 50-100 h_{70}^{-2}\;{\rm
213: M}_\odot\;{\rm yr}^{-1}$ {\it after} correction for an average factor of
214: $\sim$ 7 attenuation by dust \citep{2003ApJ...588...65S}. As previously
215: mentioned, our candidates are redder than the LBGs in the
216: \citet{2003ApJ...592..728S} survey, which suggests that their extinction
217: corrections should be larger, pushing the corrected SFRs to thousands of
218: M$_\odot\;{\rm yr}^{-1}$. The stellar winds and interstellar turbulence
219: associated with such extreme SFRs could account for the relatively broad
220: absorption features and the strength of high excitation absorption seen
221: in some of our candidates.
222:
223: The brightest sub-mm galaxies are inferred to have comparable or, in
224: some cases, even higher SFRs at these redshifts (see, e.g.,
225: \citealt{2003Natur.422..695C}), but in these objects most of the energy
226: from massive young stars is absorbed and re-radiated by dust. The SCUBA
227: source in N2 850.4 \citep{Smail2003}, with an estimated SFR $\sim 300$
228: M$_{\odot}$ yr$^{-1}$, has an optical spectrum reminiscent of those in
229: Figure~1, though our objects do not show P-Cygni profiles.
230: High-redshift galaxies with such SFRs {\it should} appear in the SDSS if
231: they have moderate UV extinction, and our candidates could plausibly
232: represent a later evolutionary stage of luminous sub-mm galaxies, with
233: young stellar populations that have burned through enough of their dusty
234: envelopes to become visible in the rest-frame UV.
235:
236: Finally, there is the possibility that these luminous objects are
237: powered by black hole accretion rather than star formation, and that
238: they represent a very unusual class of BAL quasars. In favor of this
239: interpretation, two of the spectra show evidence for broad \ion{C}{3}]
240: emission, and some have very broad absorption lines, up to $\sim 7600$
241: km s$^{-1}$. The strengths of the absorption features from highly
242: ionized species are also much stronger than the low ionization
243: absorption features. However, the candidates do not show typical BAL
244: profiles, and we are not aware of any analogs of these objects among
245: previously discovered BALs. There is significant variation among the
246: spectra in Figure~1, and the relative importance of star formation and
247: AGN activity could be different from object to object.
248:
249: \section{THE BRIGHT END OF THE LBG LUMINOSITY FUNCTION}
250: If we assume that the ultraluminous LBG interpretation is correct, and
251: that we can therefore ignore lensing amplification and AGN
252: contributions, then we can constrain the bright end of the rest-frame UV
253: luminosity function of high-redshift galaxies. Our estimate here is
254: necessarily crude, (1) because we have only six objects, and (2) because
255: assessing completeness as a function of luminosity and redshift would
256: require detailed simulation of the SDSS quasar target selection
257: algorithm. We make the conservative assumption that the selection
258: efficiency is unity in the redshift range $2.3 < z < 2.8$ over the 1360
259: deg$^2$ area of DR1, for objects above the $i=20.2$ apparent magnitude
260: limit of the quasar catalog. Our objects lie within a 1-magnitude bin,
261: with median redshift $z = 2.55$ and median absolute magnitude
262: $M_{\rm UV} \approx -25.4$.\footnote{We calculate the absolute magnitude
263: from the $i$-band apparent magnitude, so it corresponds to rest-frame
264: wavelength 7600\AA/$(1+z) \approx 2100$\AA.} We estimate $\Phi(M) =
265: \sum_{j=1}^6 1/V_{a,j} = 2.52 \times 10^{-9}
266: \dunits$, where $V_{a,j}$ is the comoving accessible volume for galaxy
267: $j$, which we take to run from $z=2.3$ to the smaller of $z=2.8$ and
268: $z_{\rm max}$, the redshift at which the galaxy's apparent magnitude
269: would hit the $i=20.2$ limit. The $z=2.8$ upper limit is based on the
270: maximum redshift of our candidates, and while the selection function may
271: continue beyond this redshift, it is likely to decline as Ly$\alpha$
272: moves past the peak of the $g$-filter. If we assumed instead that each
273: galaxy could be seen all the way to $z_{\rm max}$, then our density
274: estimate would drop by 20\%.
275:
276: Figure 2 shows this estimate in the context of the LBG luminosity
277: function estimated by \citet{2000ApJ...544..218A} from a combination of
278: ground-based data and the Hubble Deep Field. Circles show the
279: \citet{2000ApJ...544..218A} data points, with no correction for dust
280: extinction, and the solid line shows their Schechter function fit. The
281: open triangle shows our above estimate at $z = 2.55$, with an error bar
282: that corresponds to the Poisson error on six objects. We plot our point
283: at the median apparent magnitude $r \approx 20.3$, ignoring the
284: difference between the Sloan $r$-band and Adelberger \& Steidel's
285: $R$-band. Our assumption of unit selection efficiency is probably too
286: generous, so the most likely value for the space density is above the
287: data point. Since the
288: \citet{2000ApJ...544..218A} data are given in terms of apparent
289: magnitude and their galaxies have a median $z \approx 3$, we also plot a
290: filled triangle that shows the effect of moving our galaxies from
291: $z=2.55$ to $z=3$, using a $K$-correction based on the median spectral
292: slope of $\alpha = -2.16$, with $f_{\nu} \propto \nu^{\alpha}$, derived
293: by fitting the $griz$ magnitudes.
294:
295: Our estimate of the luminosity function at $r \approx 20.3$ is above the
296: Schechter function extrapolation from fainter magnitudes. However,
297: there is no reason to expect that the high-redshift, rest-frame UV
298: luminosity function should be described by a Schechter function over
299: such a wide range in luminosity, especially given the importance of
300: recent star formation history and the geometry of extinction in
301: determining a galaxy's luminosity at $\lambda 1500$\AA. A power law
302: connecting our data point (offset to $z=3$) to the brightest data point
303: from \citet{2000ApJ...544..218A} has a slope of $\Rab^{-1.93}$ (or
304: $L^{-2.93}$).
305:
306: A variety of follow-up observations could help to determine the nature
307: of these remarkable objects. High-resolution imaging to search for
308: image splitting and deeper imaging to search for foreground structures
309: could test the lensing hypothesis. Better optical spectra would allow
310: better measurements of absorption profiles and tighter constraints on
311: broad emission lines, and they might reveal high-excitation photospheric
312: lines that would be the signature of starburst activity. Near-IR
313: spectroscopy, sub-mm imaging, and X-ray measurements could better
314: quantify the importance of dust extinction and AGN activity. If these
315: objects are confirmed as extreme LBGs, they demonstrate that the
316: stupendous starbursts so often veiled by thick layers of dust can
317: sometimes emerge to show their true colors.
318:
319: \acknowledgements
320: We would like to thank Patrick Hall, Ian Smail, Charles Steidel, and an
321: anonymous referee for helpful comments and conversations. Misty Bentz
322: is supported by a Graduate Fellowship of The Ohio State University.
323:
324: The discovery of $z=2.5$ galaxies by a 2.5-m telescope is a tribute to
325: the power of a large area, multi-color imaging survey, to the efficiency
326: of the SDSS spectrographs and spectroscopic pipeline, and to the utility
327: of the SDSS Archive. The SDSS Web Site, http://www.sdss.org , lists the
328: Participating Institutions and the project's other sources of funding,
329: which include the Sloan Foundation, the NSF, NASA, DOE, the Japanese
330: Monbukagakusho, and the Max Planck Society.
331:
332: \bibliographystyle{apj}
333: \bibliography{ms}
334:
335: \clearpage
336: \begin{deluxetable}{lcccccccc}
337: \tabletypesize{\footnotesize}
338: \tablecolumns{9}
339: \tablewidth{0pc}
340: \tablecaption{Properties of Starburst Candidates}
341: \tablehead{
342: \colhead{} &
343: \colhead{} &
344: \multicolumn{7}{c}{Colors and Magnitudes\tablenotemark{a}}\\
345: \cline{3-9}\\
346: \colhead{Object} &
347: \colhead{z\tablenotemark{a}} &
348: \colhead{u$-$g} &
349: \colhead{g$-$r} &
350: \colhead{r$-$i} &
351: \colhead{i$-$z} &
352: \colhead{r Obs.} &
353: \colhead{i Obs.} &
354: \colhead{i Abs.\tablenotemark{b}}}
355: \startdata
356: SDSS J024343.77$-$082109.9 & 2.59 & 2.181 & 0.390 & 0.319 & 0.500 & 20.403 & 20.084 & -25.916 \\
357: SDSS J114756.00$-$025023.5\tablenotemark{c} & 2.56 & 3.629 & 1.170 & 0.532 & 0.679 & 19.825 & 19.293 & -26.678 \\
358: SDSS J134026.44+634433.2 & 2.79 & 2.453 & 0.689 & 0.465 & 0.340 & 19.823 & 19.358 & -26.792 \\
359: SDSS J143223.10$-$000116.4\tablenotemark{d} & 2.47 & 1.885 & 0.615 & 0.399 & 0.609 & 20.514 & 20.115 & -25.810 \\
360: SDSS J144424.55+013457.0 & 2.66 & 2.405 & 0.622 & 0.433 & 0.349 & 20.510 & 20.077 & -26.028 \\
361: SDSS J155359.96+005641.3 & 2.63 & 2.719 & 0.562 & 0.523 & 0.447 & 20.199 & 19.676 & -26.524 \\
362: \enddata
363: \tablenotetext{a} {Values taken from \citet{dr1qsopreprint}.}
364: \tablenotetext{b} {As determined by \citet{dr1qsopreprint}, with
365: $H_{0} = 70$ km s$^{-1}$ Mpc$^{-1}$, $\Omega_M = 0.3$,
366: $\Omega_{\Lambda} = 0.7$; note that this absolute
367: magnitude is extrapolated to rest-frame $i$-band
368: assuming $\alpha_Q = -0.5$, while our objects are
369: much redder. }
370: \tablenotetext{c} {SDSS J114756.00$-$025023.5 is also a 2MASS
371: object \citep{2003yCat.2246....0C}.}
372: \tablenotetext{d} {SDSS J143223.10$-$000116.4 is also an EDR
373: object (\citealt{2002AJ....123..567S};
374: \citealt{paper1}).}
375: \end{deluxetable}
376:
377:
378: \begin{deluxetable}{lcc}
379: \tabletypesize{\footnotesize}
380: %\renewcommand{\arraystretch}{.6}
381: \tablecolumns{3}
382: \tablewidth{0pc}
383: \tablecaption{Star Formation Rates}
384: \tablehead{
385: \colhead{} &
386: \colhead{F$_{\lambda}$($\lambda 1500 \times (1+z)$)\tablenotemark{a}} &
387: \colhead{SFR\tablenotemark{b}}\\
388: \colhead{Object} &
389: \colhead{(10$^{-17}$ erg s$^{-1}$ cm$^{-2}$ \AA$^{-1}$)} &
390: \colhead{(M$_{\odot}$ yr$^{-1}$)}}
391: \startdata
392: SDSS J0243$-$0821 & 2.7 & 550 \\
393: SDSS J1147$-$0250 & 2.7 & 530 \\
394: SDSS J1340+6344 & 4.1 & 1100 \\
395: SDSS J1432$-$0001 & 1.8 & 320 \\
396: SDSS J1444+0134 & 3.7 & 820 \\
397: SDSS J1553+0056 & 2.6 & 560 \\
398: \enddata
399: \tablenotetext{a}{Flux in observed frame at $\lambda = 1500 \times (1+z)$,
400: with a characteristic uncertainty of $0.3 \times
401: 10^{-17}$ erg s$^{-1}$ cm$^{-2}$ \AA$^{-1}$ }
402: \tablenotetext{b}{Assumes a Salpeter IMF with mass limits between 0.1
403: and 100 M$_{\odot}$ and a 10$^8$ year old continuous
404: star formation model \citep{Kennicutt1998}. No
405: corrections for dust have been applied.}
406: \end{deluxetable}
407:
408: \clearpage
409:
410: \begin{figure}
411: \figurenum{1}
412: \epsscale{0.6}
413: \plotone{f1.eps}
414: \caption{Rest frame spectra of the Lyman break composite (Fig. 1a,
415: \citealt{2003ApJ...588...65S}) and the six candidate starburst
416: galaxies. The Sloan spectra have been smoothed over five pixels
417: and are as follows: b.) SDSS J0243-0821, c.) SDSS J1147-0250,
418: d.) SDSS J1340+6344, e.) SDSS J1432-0001, f.) SDSS J1444+0134,
419: and g.) SDSS J1553+0056}
420: \end{figure}
421:
422: \begin{figure}
423: \figurenum{2}
424: \epsscale{1}
425: \plotone{f2.eps}
426: \caption{Luminosity function of starburst galaxies at redshifts
427: $z \approx$ 3 (for $\Omega_M$ = 0.3, $\Omega_{\Lambda}$ = 0.7).
428: The circles are data points from \citet{2000ApJ...544..218A},
429: and the solid curve shows their Schechter function fit with
430: parameters $\Phi_* = 4.4\times 10^{-3} \dunits$, $m_* = 24.54$,
431: $\alpha=-1.57$. The open triangle is the additional data point
432: based on the objects described in this work, with errorbars
433: showing the Poisson noise for six objects. The filled triangle
434: shows the effect of changing from $z=2.55$, the center of our
435: redshift bin, to $z=3$, the typical redshift of the
436: \citet{2000ApJ...544..218A} objects.}
437: \end{figure}
438:
439:
440:
441: \end{document}
442: