1: %DOCUMENT CLASS
2: %\documentclass[12pt,preprint]{aastex}
3: \documentclass[apj]{emulateapj}
4: \usepackage{lscape,apjfonts}
5: %\usepackage{apjfonts}
6: % NatBib declarations
7: \citestyle{apj}
8:
9: % NEW COMMANDS AND DEFINITIONS %
10:
11: \newcommand{\co} {{\rm $^{12}$CO}}
12: \newcommand{\hi} {\mbox{\rm \ion{H}{1}}}
13: \newcommand{\hii} {\mbox{\rm \ion{H}{2}}}
14: \newcommand{\oiii} {[\mbox{\rm \ion{O}{3}}] $\lambda$4363~\AA}
15: \newcommand{\jone} {($J=1\rightarrow0$)}
16: \newcommand{\kms} {km~s$^{-1}$}
17: \newcommand{\gpercmcu} {g~cm$^{-3}$}
18: \newcommand{\ha} {\mbox{H$\alpha$}}
19: \newcommand{\mlk} {\mbox{$M_{\odot}/L_{\odot, K}$}}
20: \newcommand{\mlr} {\mbox{$M_{\odot}/L_{\odot, R}$}}
21: \newcommand{\mlrs} {\mbox{$M_{\odot}/L_{\odot, r'}$}}
22: \newcommand{\mslk} {\mbox{$M_{*}/L_{K}$}}
23: \newcommand{\msl} {\mbox{$M_{*}/L$}}
24: \newcommand{\mslr} {\mbox{$M_{*}/L_{R}$}}
25: \newcommand{\mslrs} {\mbox{$M_{*}/L_{r'}$}}
26: \newcommand{\hr} {\mbox{$^h$}}
27: \renewcommand{\min} {\mbox{$^m$}}
28:
29: \def\spose#1{\hbox to 0pt{#1\hss}}
30: \def\lta{\mathrel{\spose{\lower 3pt\hbox{$\mathchar"218$}}
31: \raise 2.0pt\hbox{$\mathchar"13C$}}}
32: \def\gta{\mathrel{\spose{\lower 3pt\hbox{$\mathchar"218$}}
33: \raise 2.0pt\hbox{$\mathchar"13E$}}}
34:
35:
36: % OTHER STUFF
37: \shorttitle{Resolving Abundance Gradient Discrepancies in M33}
38: \shortauthors{Rosolowsky \& Simon}
39:
40:
41: % PAPER
42: \begin{document}
43:
44: \title{The M33 Metallicity Project: Resolving the Abundance Gradient
45: Discrepancies in M33\altaffilmark{1}}
46: %The Abundance Gradient and Intrinsic Scatter in M33
47:
48: \author{Erik Rosolowsky}
49:
50: \affil{Harvard-Smithsonian Center for Astrophysics, 60 Garden St.,
51: MS-66, Cambridge, MA 02138; erosolow@cfa.harvard.edu}
52:
53: \and
54:
55: \author{Joshua D. Simon}
56:
57: \affil{Department of Astronomy,
58: California Institute of Technology, 1200 E. California Blvd.,
59: MS 105-24, Pasadena, CA 91125; jsimon@astro.caltech.edu}
60:
61: %\email{jsimon@astro.caltech.edu}
62: \altaffiltext{1}{Data presented herein were obtained at the
63: W. M. Keck Observatory, which is operated as a scientific
64: partnership among the California Institute of Technology, the
65: University of California, and the National Aeronautics and Space
66: Administration. The Observatory was made possible by the generous
67: financial support of the W. M. Keck Foundation.}
68:
69:
70: \begin{abstract}
71: We present a new determination of the metallicity gradient in M33,
72: based on Keck/LRIS measurements of oxygen abundances using the
73: temperature-sensitive emission line [\ion{O}{3}]~$\lambda$4363~\AA\ in
74: 61 \hii\ regions. These data approximately triple the sample of
75: direct oxygen abundances in M33. We find a central abundance of 12 +
76: log(O/H) = $8.36 \pm 0.04$ and a slope of $-0.027 \pm
77: 0.012$~dex~kpc$^{-1}$, in agreement with infrared measurements of the
78: neon abundance gradient but much shallower than most previous oxygen
79: gradient measurements. There is substantial intrinsic scatter of
80: 0.11~dex in the metallicity at any given radius in M33, which imposes
81: a fundamental limit on the accuracy of gradient measurements that rely
82: on small samples of objects. We also show that the ionization state
83: of neon does not follow the ionization state of oxygen as is commonly
84: assumed, suggesting that neon abundance measurements from optical
85: emission lines require careful treatment of the ionization
86: corrections.
87: \end{abstract}
88:
89: \keywords{galaxies: abundances --- galaxies: M33 --- galaxies: spiral
90: --- HII regions --- ISM: abundances --- ISM: evolution}
91:
92:
93: \section{INTRODUCTION}
94: \label{intro}
95:
96: M33 was one of the first galaxies in which a radial abundance gradient
97: was recognized \citep{searle71}. Beginning with \citet{smith75}, a
98: series of studies showed that \hii\ region abundances in M33 decline
99: by approximately an order of magnitude over the visible extent of the
100: galaxy \citep{ka81,vilchez88,garnett92}, with a simple exponential
101: providing an accurate fit to the gradient. A typical value for the
102: gradient from this work is $-0.12 \pm 0.02$~dex~kpc$^{-1}$
103: \citep[][hereafter V88]{vilchez88}. More recently, \citet[][hereafter
104: C06]{crockett06} presented new metallicity measurements for a small
105: sample of \hii\ regions that indicated nearly constant oxygen
106: abundances across M33, with a best-fit gradient of only $-0.012 \pm
107: 0.011$~dex~kpc$^{-1}$. If confirmed, this very shallow gradient would
108: call into question a variety of results concerning, e.g., the
109: metallicity dependence of the Cepheid period-luminosity relation
110: \citep{lee02} and the CO-H$_{2}$ conversion factor \citep{eros03} that
111: depend on the abundance gradient in M33. To complicate matters
112: further, infrared measurements of the neon abundance gradient, which
113: should be less sensitive to systematic effects such as extinction and
114: the temperature of the \hii\ region, yield an intermediate gradient of
115: $-0.034 \pm 0.015$~dex~kpc$^{-1}$ that is not in very good agreement
116: with any of the published oxygen gradients from optical data
117: \citep[][hereafter WNP02]{wnp02}.
118:
119: In this paper, we present new, direct metallicity measurements based
120: on detections of the temperature-sensitive \oiii\ line in 61 \hii\
121: regions in M33, approximately tripling the sample available in the
122: literature for M33 (or any other galaxy). We use these data to
123: provide an improved measurement of the abundance gradient in M33. In
124: \S \ref{observations} we describe the observations, data reduction,
125: and derivation of metallicities. In \S \ref{results} we determine the
126: oxygen abundance gradient in M33, attempt to derive neon abundances,
127: and discuss some of the implications of our results. We briefly
128: summarize the paper in \S \ref{conclusions}.
129:
130:
131: \section{OBSERVATIONS AND DATA REDUCTION}
132: \label{observations}
133:
134: \subsection{Observations}
135:
136: The M33 Metallicity Project is a long-term program in which we plan to
137: obtain deep spectroscopy of $\sim1000$ \hii\ regions in M33 drawn from
138: the unified catalog of \citet{lgatlas}. These data will eventually
139: yield a high-resolution, two-dimensional metallicity map covering the
140: entire galaxy, with a mean sampling rate of $\gta2$ measurements per
141: square kpc, and sampling down to $\lta50$~pc in some regions. At
142: present, the data set is approximately 20\%\ of its expected final
143: size.
144:
145: The observations discussed in this paper were obtained with the LRIS
146: spectrograph \citep{oke95} on the Keck I telescope on 2004 September
147: 22-23 and 2004 October 21. Weather conditions during the September
148: nights were clear, while the October observations were affected by
149: clouds at times. The spectrograph was configured with the 600/4000
150: grism on the blue side, a dichroic at 5600~\AA, and the 900/5500 and
151: 400/8500 gratings on the red side (in separate exposures). The
152: blue-side spectral resolution was $\sim5$~\AA, and spectra typically
153: covered $3600-5400$~\AA\ with some variation resulting from slit
154: placement on the mask. Slits for which the wavelength coverage missed
155: the [\ion{O}{2}]~$\lambda$3727~\AA\ line have been excluded from our
156: analysis. The slits were 1\arcsec\ wide and had a minimum length of
157: 10\arcsec. We used integration times of 15 minutes per exposure, with
158: $1-4$ exposures per slitmask. The slits were generally aligned within
159: $\sim10\degr$ of the parallactic angle during the observations. We
160: observed 11 slitmasks in the southwest half of the galaxy, with an
161: average of $\sim20$ slits on each mask. For the purposes of this
162: paper we use only the blue-side data, which include all of the lines
163: ([\ion{O}{2}] $\lambda\lambda$3726, 3729\AA\ and
164: [\ion{O}{3}]$\lambda\lambda$ 4363, 4959, 5007 \AA) necessary to derive
165: oxygen abundances.
166:
167: \subsection{Data Reduction}
168: \label{data-reduction}
169: The data were reduced using an IDL pipeline originally developed for
170: the Sloan Digital Sky Survey spectroscopic data (D. J. Schlegel et al., in
171: preparation) and then modified to deal appropriately with multi-slit
172: LRIS spectra. The images were bias-subtracted and then flat-fielded
173: using either dome flat exposures or the internal quartz lamp. We used
174: the flatfield exposures to identify the location of each slit on the
175: CCD, and the remainder of the processing was carried out on each slit
176: individually. We produced a two-dimensional wavelength solution for
177: each slit using exposures of Hg, Cd, Zn, Ne, and Ar arc lamps. We
178: then removed cosmic rays from the images and coadded the frames
179: together.
180:
181: After these basic reduction steps, each slit was examined
182: interactively and the targets were identified based on peaks in the
183: spatial profile of the H$\beta$ line. We selected an extraction
184: region around each target by following the H$\beta$ profile down to
185: the background level (which in most cases includes emission from the
186: diffuse ionized medium of M33) and setting the extraction limits where
187: the H$\beta$ intensity reached the background value on either side of
188: the peak. A b-spline fit to the remainder of the slit produced a 2D
189: sky model covering the entire slit. Finally, we subtracted the sky
190: model from the data and extracted the \hii\ region spectrum with a
191: boxcar algorithm, yielding a 1D spectrum and associated uncertainties.
192: The spectrum was corrected for atmospheric extinction and
193: flux-calibrated by comparison to an exposure of the spectrophotometric
194: standard star G191B2B \citep{massey88}. An example calibrated 1D
195: spectrum is shown in Figure \ref{examplespec}.
196:
197: \begin{figure*}
198: %\epsscale{1.25}
199: \plotone{f1.eps}
200: \caption{\label{examplespec} Example spectrum of \hii\ region 60 in
201: Table \ref{linetable} (BCLMP 266). This spectrum has the median S/N
202: (8.4) for the [\ion{O}{3}]~$\lambda$4363~\AA\ line in our sample.
203: $F_{\lambda}$ is given in units of
204: $10^{-16}$~erg~s$^{-1}$~cm$^{-2}$~\AA$^{-1}$.}
205: \end{figure*}
206:
207:
208: Our procedure for measuring emission line fluxes in the 1D spectra
209: began by dividing each spectrum up into a series of short
210: ($\sim200$~\AA) spectral windows, each containing at least one bright
211: line (except for the regions between 4500~\AA\ and 4800~\AA\ and
212: beyond 5100~\AA\ where no such lines exist) and performing a
213: constrained linear+Gaussian fit to all of the emission lines contained
214: within each window. In each window the line separations were fixed at
215: their known values, and the widths of the lines were all set equal to
216: the width of the brightest line in the window (except for blends such
217: as the partially resolved [\ion{O}{2}]~$\lambda$3727~\AA\ doublet).
218: We then measured fluxes by integrating the observed intensity over a
219: $\pm1.5$ FWHM region centered on each line. Uncertainties in the
220: measured line fluxes were calculated by propagating the uncertainty
221: associated with each pixel in the one-dimensional spectrum. We
222: corrected the measured line fluxes for reddening and Balmer absorption
223: from the underlying stellar population using the method outlined by
224: \citet{os01}, including all of the Balmer emission lines for which we
225: had high S/N detections. We used the Galactic reddening law of
226: \citet{ccm89} and assumed that the equivalent widths of all of the
227: Balmer absorption lines in a given \hii\ region were the same. We
228: experimented with setting the ratios of the equivalent widths of the
229: absorption lines equal to their mean values averaged over time for the
230: solar metallicity, continuous star formation models of \citet{g-d99},
231: but we found that the resulting metallicity changes are always less
232: than 0.01~dex. With such minimal changes and with possible
233: differences between the stellar populations simulated by \citet{g-d99}
234: and those present in M33 \hii\ regions, we proceeded using the simpler
235: model of constant Balmer equivalent widths. Corrected line fluxes for
236: each \hii\ region with a detection of
237: [\ion{O}{3}]~$\lambda$4363~\AA\ are given in Table \ref{linetable}.
238: The listed uncertainties in the line fluxes have been propagated
239: through the reddening and absorption corrections. Balmer line ratios
240: as well as derived reddening and absorption values are given in Table
241: \ref{hlinetable}.
242:
243: %\clearpage
244: %\end{landscape}
245:
246:
247:
248: \subsection{Derivation of \hii\ Region Temperatures and Abundances}
249: We measure oxygen abundances using the {\it nebular} module
250: \citep{sd95} of the STSDAS package in IRAF. We model the nebula as
251: consisting of two ionization zones with the low- and medium-ionization
252: zones defined by the presence of O$^{+}$ and O$^{+2}$, respectively.
253: The temperature of the medium ionization zone is calculated using {\it
254: nebular} based on the standard ratio of the
255: [\ion{O}{3}]~$\lambda\lambda 4959,5007$~\AA~doublet to the
256: [\ion{O}{3}]~$\lambda$4363~\AA\ auroral line. To determine the
257: temperature of the low-ionization zone, we adopt the scaling of
258: \citet{campbell86}: $T([\mbox{\ion{O}{2}}]) =
259: 0.70~T([\mbox{\ion{O}{3}}])+3000~\mathrm{K}.$ We assume that all
260: \ion{H}{2} regions are in the low density limit ($n_e \lta
261: 10^2~\mbox{cm}^{-3}$). With these densities and temperatures, we
262: measure ionic abundances relative to H$^+$ for O$^{+}$ and O$^{+2}$
263: and assume that all oxygen is found in these two ionization states.
264: This analytic approach is common to many extragalactic abundance
265: studies \citep[e.g.,][]{ken_m101}, but relies on several (admittedly
266: well-justified) assumptions. We will examine these assumptions in
267: future work with the complete sample of \ion{H}{2} regions including
268: the full spectral coverage of LRIS.
269:
270: We restrict our analysis to the 61 \ion{H}{2} regions that have
271: detections of the [\ion{O}{3}]~$\lambda$4363~\AA\ line with S/N$ > 3$
272: (more conservative selections do not change our results; see \S
273: \ref{grad-sec}). The locations of these \hii\ regions are
274: superimposed on an H$\alpha$ image of the galaxy in Figure
275: \ref{slitloc}. All of our targets from this phase of the M33
276: Metallicity Project are in the southwest portion of the galaxy.
277: Previous work \citep[with the notable exception of][]{magrini07} has
278: focused primarily on giant \ion{H}{2} complexes (e.g., NGC~604,
279: NGC~588) in the northeast half of the galaxy. The
280: [\ion{O}{3}]~$\lambda$4363~\AA\ line is detected in $\sim 1/3$ of our
281: targets and this fraction shows no significant variation with
282: galactocentric radius. Contaminating supernova remnants, emission
283: line stars and planetary nebulae are removed from the sample based on
284: identification of characteristic spectral line features and
285: cross-matching with published catalogs of these objects.
286:
287: As noted in \S\ref{data-reduction}, uncertainties in the line fluxes
288: are propagated from single pixel error values. The uncertainties in
289: $c(\mathrm{H}\beta)$ and the equivalent width of the Balmer absorption
290: have been included in the errors for the dereddened line fluxes. The
291: reported errors in the physical parameters are determined by a Monte
292: Carlo propagation of errors through the entire metallicity
293: determination process. The Monte Carlo process is particularly
294: well-suited for gauging the widths of skewed uncertainty distributions
295: in the derived properties from low signal-to-noise lines. For each
296: \ion{H}{2} region, we generate 100 sets of trial line fluxes by adding
297: a normal deviate times the flux uncertainty to each line flux and
298: re-analyzing the spectrum using {\it nebular}. We also include a
299: random fluctuation in the temperature of the low ionization zone
300: derived from the relationship of \citet{campbell86}. Based on the
301: magnitude of the uncertainties found in the work of \citet{ken_m101},
302: the 1$\sigma$ error in $T$([\ion{O}{2}]) is 300~K. The final
303: uncertainties on the abundances are defined as the 1$\sigma$ widths of
304: the Monte Carlo abundance distributions (see also C06). For
305: reasonably strong [\ion{O}{3}]~$\lambda$4363~\AA~emission (S/N$ >
306: 10$), precisions in the derived metallicity are $\lta 0.08$ dex. For
307: the four \ion{H}{2} regions in our sample that have been studied by
308: previous observers, the abundance determinations agree to within the
309: uncertainties (see Table \ref{comparison_table}).
310:
311: \begin{deluxetable*}{lccccl}
312: \tablewidth{0pt}
313: \tablecolumns{6}
314: \tablecaption{Comparison with Literature Metallicity Measurements}
315: \tablehead{
316: \colhead{\hii\ region} & \colhead{KA81} &
317: \colhead{V88} & \colhead{C06} & \colhead{M07} &
318: \colhead{Our} \\
319: \colhead{} & \colhead{Abundance} & \colhead{Abundance} &
320: \colhead{Abundance} & \colhead{Abundance} & \colhead{Abundance}}
321: \startdata
322: NGC~588 & \nodata & $8.30 \pm 0.06$ & \nodata & \nodata & $8.315 \pm 0.061$ \\
323: MA~2 & 8.38 & $8.44 \pm 0.15$ & \nodata & \nodata & $8.334 \pm 0.083$ \\
324: BCLMP~90 & \nodata &\nodata & $8.50 \pm 0.06$ & \nodata & $8.503 \pm 0.057$\tablenotemark{a} \\
325: BCLMP~218 & \nodata &\nodata & \nodata & $8.25 \pm
326: 0.05$ & $8.157 \pm 0.059$
327: \enddata
328: \tablerefs{KA81 = \citealt{ka81}; V88 = \citealt{vilchez88}; C06 =
329: \citealt{crockett06}; M07 = \citealt{magrini07}}
330: \tablenotetext{a}{Weighted average of two independent measurements at
331: the same position; see Table \ref{linetable}.}
332: \label{comparison_table}
333: \end{deluxetable*}
334:
335:
336: \subsection{\ion{He}{2} Zones}\label{heii}
337: We detect \ion{He}{2} $\lambda 4686$~\AA\ from 5 of the 61 \ion{H}{2}
338: regions in our sample at a typical level of 2.5\% of the H$\beta$ flux
339: (objects 3, 8, 16, 18, and 29 in Table \ref{linetable}). The emission
340: is almost certainly nebular as it is spectrally unresolved but
341: spatially resolved in these sources (C06). The presence of
342: \ion{He}{2} emission indicates that the simple two ionization zone
343: model adopted in the metallicity estimates is somewhat inaccurate. A
344: high ionization region implies the presence of O$^{+3}$, which is not
345: accounted for in our abundance determination. Consequently, the
346: derived oxygen abundances will be systematically low by an unknown
347: amount. To estimate the magnitude of this effect, we calculate
348: He$^{+2}$/He$^{+}$ based on the ratio of the \ion{He}{2} $\lambda
349: 4686$~\AA\ and \ion{He}{1} $\lambda 4471$~\AA\ lines. This ratio is
350: only approximate because the \ion{He}{1} line is likely contaminated
351: by stellar absorption. The derived ratio is roughly equal to
352: O$^{3+}$/(O$^{+}$+O$^{2+}$), providing an estimate of the contribution
353: of O$^{3+}$. We therefore expect that we have underestimated the
354: oxygen abundance by $\sim 0.01-0.06$ dex in these regions. Since we
355: are not sure of the influence of stellar absorption lines, we add (in
356: quadrature) this derived increase to the reported oxygen uncertainty
357: for these five regions.
358:
359: \section{RESULTS}
360: \label{results}
361:
362: \subsection{The Oxygen Gradient}
363: \label{grad-sec}
364: Figure \ref{gradient} shows the radial distribution of gas phase
365: oxygen abundances in M33. We have assumed a thin disk geometry based
366: on a distance to M33 of 840 kpc \citep{Freedman2001}, $i=52^{\circ}$
367: and PA = $22^{\circ}$ \citep{cs00}. An ordinary, linear least-squares
368: fit to the data, weighting each point according to its inverse
369: variance, produces a gradient of $12+\log(\mathrm{O/H}) = (8.38 \pm
370: 0.03) - (0.032 \pm 0.006)~R_{\rm kpc}$, but the residuals from this
371: fit are much larger than would be expected from the measurement
372: uncertainties ($\widetilde{\chi^{2}} = 3.2$). Since it is evident
373: that an exponential decline with radius is a reasonable functional
374: form to describe the data, the high $\widetilde{\chi^{2}}$ value
375: indicates the presence of substantial intrinsic scatter around the
376: relation.
377:
378:
379:
380: \begin{figure}
381: %\epsscale{1.25}
382: \plotone{f3.eps}
383: \caption{\label{gradient} Abundances of 61 \ion{H}{2} regions in M33
384: as a function of galactocentric radius. A linear gradient with a
385: slope of $-0.027$~dex~kpc$^{-1}$ is fit to the data (solid
386: line). Regions with significant \ion{He}{2}~$\lambda
387: 4686$\AA\ emission are
388: indicated with open symbols.}
389: \end{figure}
390:
391:
392: In this situation, accurately estimating the fit parameters must be
393: done using the method of \citet[][hereafter AB96]{bces}. We perform a
394: linear regression with the AB96 technique for weighted least-squares
395: fitting in the presence of intrinsic scatter in the data. The AB96
396: technique estimates the intrinsic variance and adds this variance to
397: that of the data to produce the appropriate weights for the fit.
398: Since the intrinsic variance is larger than the statistical
399: uncertainties for the M33 abundance data, there is less variation in
400: the weights than in the measurement uncertainties. Even though
401: [\ion{O}{3}] $\lambda4363$~\AA\ detections with S/N as low as 3 result
402: in relatively large uncertainties in the derived abundances, including
403: these data improves the quality of the fit by increasing the sample
404: size. This technique yields an abundance gradient of
405: \begin{equation}
406: 12+\log(\mathrm{O/H}) = (8.36 \pm 0.04) - (0.027 \pm 0.012)~R_{kpc}.
407: \end{equation}
408:
409: The implied intrinsic variance in the population (i.e., the scatter in
410: the metallicities at any given radius) is 0.11 dex, larger than the
411: precision of most of the measurements. We check the uncertainties on
412: the central abundance and the slope of the abundance gradient by
413: bootstrapping the data \citep{numrec}, which yields uncertainties
414: $\lta 10\%$ higher than those derived from the AB96 formalism.
415: Although all the included data appear to have a statistically (and
416: visibly) significant [\ion{O}{3}] $\lambda 4363$~\AA\ detection,
417: performing the analysis only on the 23 \ion{H}{2} regions with
418: S/N$>10$ in this line yields an indistinguishable gradient of
419: $12+\log(\mathrm{O/H}) = (8.41 \pm 0.07) - (0.039 \pm 0.021)~R_{kpc}$.
420: We note that all of our targets are located in southwest half of the
421: galaxy (see Figure \ref{slitloc}), and therefore that the other half
422: of the galaxy could conceivably have a different gradient, but we
423: regard this possibility as unlikely. Finally, we have correlated our
424: derived metallicities and their residuals around a radial gradient
425: against numerous parameters in our analysis such as L(H$\alpha$) taken
426: from the catalog of \citet{lgatlas}, $c$(H$\beta)$, the equivalent
427: widths of the Balmer absorption and H$\beta$ emission, and the oxygen
428: ionization correction factor. We find no correlation with the
429: residuals that would indicate systematic errors in our methods.
430:
431: \begin{figure*}
432: %\epsscale{1.25}
433: \plotone{f2.eps}
434: \caption{\label{slitloc} Locations of the 61 \ion{H}{2} regions with
435: oxygen abundance determinations superimposed on a
436: continuum-subtracted H$\alpha$ image of the southwest portion of
437: M33. The image is produced from the survey data of
438: \citet{massey-m33m31}. Contours of constant galactocentric radius
439: are shown as dotted ellipses with a spacing of 1 kpc.}
440: \end{figure*}
441:
442:
443: \subsection{Complications in Measuring the Neon Gradient}
444: The slope of the neon abundance gradient is expected to be equal to
445: that of oxygen since these elements are produced in nearly equal
446: proportions for all channels of stellar nucleosynthesis. WNP02
447: derived neon abundances from \emph{Infrared Space Observatory} spectra
448: and found a gradient consistent with our oxygen data. Optical
449: observers commonly derive a neon gradient based on the
450: [\ion{Ne}{3}]~$\lambda$3869~\AA\ line, yielding an Ne$^{+2}$
451: abundance. The fraction of neon found in the Ne$^{+2}$ state is often
452: assumed to be the same as the fraction of oxygen in O$^{+2}$, so that
453: an identical ionization correction factor (ICF) can be used for the
454: two elements \citep[C06]{stasinska01}. Using the simple ICF results,
455: we found a significant gradient in the Ne/O ratio, contrary to
456: expectations. We find cause for concern with this assumption for the
457: neon ICF when we derive the abundance of Ne$^{+2}$ using the standard
458: method and plot the Ne$^{+2}$ to O$^{+2}$ ratio as a function of the
459: fraction of oxygen found in the doubly ionized state (Figure
460: \ref{neon-problem}). This ratio should be invariant with oxygen
461: ionization fraction and equal to the Ne/O ratio, yet the plot shows a
462: significant slope, with Ne$^{+2}$/O$^{+2}$ varying by a factor of
463: $\sim4$ over the observed range. Since the relationship between the
464: two variables is not known a priori, we fit a linear regression using
465: the BCES method of AB96 to account for the uncertainties in both
466: directions. The resulting fit gives:
467: \begin{equation}
468: \log(\mbox{Ne$^{+2}$/O$^{+2}$}) = (-1.45 \pm 0.10) +
469: (1.27 \pm 0.22)~\mbox{O$^{+2}$/O}.
470: \label{neonfit}
471: \end{equation}
472: This analysis finds that the slope of the relationship is strongly
473: inconsistent with the value of zero expected from using the same ICF
474: for neon and oxygen. The fraction of doubly-ionized oxygen is a proxy
475: for the ionization state of the nebula, so the trend indicates less
476: Ne$^{+2}$ relative to O$^{+2}$ in low ionization nebulae. This is the
477: expected direction for this trend since the third ionization potential
478: of neon (41 eV) is larger than that of oxygen (35 eV). More recent
479: work by \citet{izotov04,izotov06} suggests that applying the oxygen
480: ICF to neon may not be valid in lower ionization regions because of
481: charge transfer reactions between O$^{2+}$ and atomic hydrogen. These
482: discrepancies could also be caused by a failure to account
483: appropriately for photons with $h\nu \sim 40$~eV in the O-star
484: atmospheres used in photoionization codes (C06).
485:
486: \citet{perez-montero07} propose a nonlinear ICF based on their own
487: photoionization models:
488: \begin{equation}
489: \mathrm{ICF(Ne^{2+})}= 0.753+0.142x+\frac{0.171}{x},
490: \end{equation}
491: where $x=\mathrm{O^{2+}/(O^++O^{2+})}$. Adopting this ICF, we
492: recalculate neon abundances and find no significant gradient in Ne/O
493: with radius. For our sample of \hii\ regions, we measure
494: $\langle\mathrm{log(Ne/O)}\rangle=-0.69\pm 0.07$, consistent with the
495: results of \citet{perez-montero07} for giant extragalactic \ion{H}{2}
496: regions. We conclude that using the same ICF for neon and oxygen is
497: inappropriate for the \ion{H}{2} regions we observed and that more
498: sophisticated ICFs seem to produce reasonable results for generic
499: \ion{H}{2} regions in M33.
500:
501:
502: \begin{figure}
503: %\epsscale{1.25}
504: \plotone{f4.eps}
505: \caption{\label{neon-problem} Ratio of doubly-ionized neon to
506: doubly-ionized oxygen plotted as a function of O$^{+2}$ ionization
507: fraction. Based on photoionization models, this value should be
508: constant and equal to the Ne/O ratio. However, there is a clear
509: trend of the ratio between the doubly ionized species and the degree
510: of ionization in real \ion{H}{2} regions. The dashed line shows the
511: fit given in Equation \ref{neonfit}. Regions with significant
512: \ion{He}{2}~$\lambda 4686$\AA\ emission are indicated with open
513: symbols.}
514: \end{figure}
515:
516:
517: %\section{DISCUSSION}
518: %\label{discussion}
519:
520: \subsection{Comparison to Previous Gradient Measurements}
521:
522: Our measured gradient of $-0.027$~dex~kpc$^{-1}$ is consistent within
523: the uncertainties with the neon gradient of WNP02 ($-0.034\pm 0.015$
524: dex kpc$^{-1}$) and marginally so with the oxygen gradient of C06
525: ($-0.012\pm 0.011$ dex kpc$^{-1}$). However, it is not consistent
526: with that of V88 ($-0.05\pm 0.01$ dex kpc$^{-1}$ for the outer regions
527: and $-0.10$ dex kpc$^{-1}$ overall after rescaling to our assumed
528: distance) and various older studies. We note that the steep inner
529: gradient measured by V88 is driven by abundances determined from
530: photoionization models, which may be systematically high compared to
531: electron temperature abundances. The apparent discrepancy in
532: abundance gradient likely results, at least in part, from an
533: underestimation of the uncertainties imparted by intrinsic scatter in
534: the \ion{H}{2} region abundances (see \S\ref{variance}). The
535: intrinsic scatter in our optical \hii\ region abundances (0.11 dex) is
536: comparable to the value of $0.07-0.10$~dex reported by WNP02.
537:
538: WNP02 suggested that a linear gradient may not be an appropriate
539: description for the enrichment of M33, arguing for a flat distribution
540: of abundances with a step down at a galactocentric radius of 4~kpc.
541: We fit a linear gradient to the 38 \ion{H}{2} regions inside $R_{\rm
542: gal}=4$~kpc and find a slope of $-0.015 \pm 0.024$~dex~kpc$^{-1}$,
543: which is consistent with no gradient. \citet{stan06-pne} also claimed
544: an abundance plateau in the inner galaxy based on several sub-solar
545: metallicity planetary nebulae in the region. We measure a mean
546: metallicity in the inner galaxy ($R_{\rm gal}<4$~kpc) of 8.30; in the
547: outer galaxy, the mean is 8.21, implying a jump of the same order as,
548: but somewhat smaller than, that of WNP02 if a step function
549: metallicity profile holds.
550:
551: There now appears to be a growing consensus that the oxygen abundance
552: gradient in \hii\ regions in M33 is relatively shallow, in conflict
553: with the decades-long assumption of a gradient at least as steep as
554: $-0.10$~dex~kpc$^{-1}$. If we ignore the innermost \hii\ region
555: observed by V88, which does not have a detection of [\ion{O}{3}]
556: $\lambda 4363$~\AA\ or any other temperature-sensitive lines, then all
557: of the modern \ion{H}{2} region studies report gradients of
558: $-0.06$~dex~kpc$^{-1}$ or shallower \citep[V88, C06,][]{magrini07}.
559: We show in \S \ref{variance} that the results of these studies, which
560: include $6-14$ \hii\ region abundances, are consistent with our
561: measurements once their small sample sizes are taken into account.
562:
563: Other types of sources besides \hii\ regions are increasingly being
564: employed to study abundances in M33. \citet{urbaneja05} used
565: quantitative spectroscopy of OB supergiants to determine an oxygen
566: gradient of $-0.06 \pm 0.02$~dex~kpc$^{-1}$. Since these stars are
567: very young, they are expected to trace the interstellar medium (ISM)
568: abundances. If the OB stars have a similar intrinsic abundance
569: scatter to the \hii\ regions we measure, then this result would be
570: compatible with our findings. \citet{magrini07} argue for a
571: steepening of the gradient at small radii based on the young star
572: metallicities in the inner 2 kpc of the galaxy \citep{urbaneja05}, but
573: we find no evidence for such a steepening based on the \ion{H}{2}
574: region data alone. Planetary nebulae \citep{magrini04} still yield a
575: steep oxygen gradient ($-0.14$~dex~kpc$^{-1}$), but this result is
576: dependent on a single nebula at large radius; the remaining objects
577: clearly have a much shallower gradient. Finally, supernova remnants
578: have also shown a shallow oxygen gradient, although the gradients in
579: other elements appear to be steeper \citep{bk85}.
580:
581:
582: \subsection{Intrinsic Scatter and the Determination of Gradients}
583: \label{variance}
584: Even after accounting for uncertainties in the stellar absorption and
585: reddening corrections, we find that there is an intrinsic scatter of
586: 0.11 dex around the gradient that is unexplained by the measurement
587: uncertainties. Such scatter is commonly seen in gradient
588: determinations \citep[e.g., in the Milky Way;][]{gal-abund-grad} and
589: may result from metallicity fluctuations in the interstellar medium.
590: Regardless of its source, gradient determinations made in the face of
591: significant scatter coupled with a limited number of observations may
592: produce widely varying results. The historical evolution of the
593: gradient determination in M33 (one of the best-studied galaxies),
594: ranging over nearly an order of magnitude, serves as a cautionary
595: example.
596:
597: Only large numbers of measurements can overcome the uncertainties
598: engendered by the intrinsic variance and relatively shallow gradient
599: in M33. In Figure \ref{gradient-distn}, we show the likelihood of
600: measuring a particular value of the gradient if only 10 \hii\ regions
601: from our sample were used in the determination of the gradient. We
602: model the behavior for 10 objects since this is a typical number of
603: targets studied in estimates of galactic abundance gradients, such as
604: those of V88 and C06. The broad distribution of the derived gradients
605: (standard deviation of 0.03~dex~kpc$^{-1}$) suggests that
606: uncertainties in the gradients are systematically underreported in
607: galaxies with significant scatter around their gradients. Even with a
608: sample several times larger than previous studies, we can only report
609: the presence of a gradient in M33 with 2.3$\sigma$ confidence! Other
610: galaxies may have better established gradients owing to smaller
611: intrinsic scatter and/or a steeper abundance gradient \citep[e.g.,
612: M101;][]{ken_m101}.
613:
614:
615: \begin{figure}
616: %\epsscale{1.25}
617: \plotone{f5.eps}
618: \caption{\label{gradient-distn} Distribution of the abundance gradients
619: that would be measured for samples of 10 \ion{H}{2} regions drawn
620: randomly from our sample of 61. The large width of distribution
621: shows the uncertainty imprinted upon the abundance determinations by
622: the underlying variance. The measurements of \citet[V88]{vilchez88}
623: and \citet[C06]{crockett06} are indicated and are consistent with
624: the distribution given their small sample sizes. }
625: \end{figure}
626:
627:
628: \section{SUMMARY AND CONCLUSIONS}
629: \label{conclusions}
630:
631: Using data from the M33 Metallicity Project, we have obtained direct
632: oxygen abundance measurements based on detections of the [\ion{O}{3}]
633: $\lambda 4363$~\AA\ line for 61 \ion{H}{2} regions in the southwest
634: half of M33. This data set approximately triples the sample of M33
635: \hii\ region metallicities in the literature and makes M33 by far the
636: best-studied external galaxy in terms of its ISM abundances. We
637: presented a refined oxygen gradient for M33 and found an exponential
638: abundance profile with a gradient of $-0.027 \pm 0.012$~dex~kpc$^{-1}$
639: and a central abundance of $12+\log(\mathrm{O/H})=8.36\pm 0.04$.
640: There is an intrinsic scatter of 0.11 dex around the trend, which
641: complicates the measurement of the shallow gradient in the galaxy.
642: When this scatter is accounted for, previous studies (which generally
643: derived steeper gradients) were shown to be consistent with our
644: measurements. We also noted that there is a significant correlation
645: of the Ne$^{+2}$/O$^{+2}$ ratio with the fraction of oxygen found in
646: the doubly ionized state, which suggests that neon determinations
647: based solely on optical lines require more sophisticated ionization
648: correction factors than are typically assumed.
649:
650: \acknowledgements{The authors wish to recognize and acknowledge the
651: very significant cultural role and reverence that the summit of
652: Mauna Kea has always had within the indigenous Hawaiian community.
653: We are most fortunate to have the opportunity to conduct
654: observations from this mountain. We thank the anonymous referee for
655: comments that improved the paper. ER acknowledges support from an
656: NSF AAP Fellowship (AST-0502605), and JDS acknowledges the support
657: of a Millikan Fellowship provided by Caltech. We thank Leo Blitz
658: for support of the early stages of this project, Joe Hennawi for
659: assistance with our data reduction code, and Jos{\'e} V{\' i}lchez
660: for helpful conversations. We also acknowledge the many useful
661: discussions with the participants in the Metals 2007 Conference
662: whose advice has improved this work. This research has made use of
663: NASA's Astrophysics Data System Bibliographic Services and the
664: SIMBAD database, operated at CDS, Strasbourg, France.}
665:
666:
667: {\it Facilities:} \facility{Keck:I (LRIS)}
668:
669:
670: %\bibliographystyle{apj}
671: %\bibliography{refs}
672:
673: \begin{thebibliography}{}
674:
675: \bibitem[{{Afflerbach} {et~al.}(1997){Afflerbach}, {Churchwell}, \&
676: {Werner}}]{gal-abund-grad}
677: {Afflerbach}, A., {Churchwell}, E., \& {Werner}, M.~W. 1997, \apj, 478, 190
678:
679: \bibitem[{{Akritas} \& {Bershady}(1996)}]{bces}
680: {Akritas}, M.~G. \& {Bershady}, M.~A. 1996, \apj, 470, 706 (AB96)
681:
682: \bibitem[{{Blair} \& {Kirshner}(1985)}]{bk85} {Blair}, W.~P., \&
683: {Kirshner}, R.~P.\ 1985, \apj, 289, 582
684:
685: \bibitem[{{Boulesteix} {et~al.}(1974){Boulesteix}, {Courtes}, {Laval},
686: {Monnet}, \& {Petit}}]{bou74}
687: {Boulesteix}, J., {Courtes}, G., {Laval}, A., {Monnet}, G., \& {Petit}, H.
688: 1974, \aap, 37, 33
689:
690: \bibitem[Campbell et al.(1986)]{campbell86} Campbell, A.,
691: Terlevich, R., \& Melnick, J.\ 1986, \mnras, 223, 811
692:
693: \bibitem[Cardelli et al.(1989)]{ccm89} Cardelli, J.~A., Clayton,
694: G.~C., \& Mathis, J.~S.\ 1989, \apj, 345, 245
695:
696: \bibitem[{{Corbelli} \& {Salucci}(2000)}]{cs00}
697: {Corbelli}, E. \& {Salucci}, P. 2000, \mnras, 311, 441
698:
699: \bibitem[Crockett et al.(2006)]{crockett06} Crockett, N.~R.,
700: Garnett, D.~R., Massey, P., \& Jacoby, G.\ 2006, \apj, 637, 741 (C06)
701:
702: \bibitem[Courtes et al.(1987)]{cou87} Courtes, G.,
703: Petit, H., Petit, M., Sivan, J.-P., \& Dodonov, S.\ 1987, \aap, 174,
704: 28
705:
706: \bibitem[{{Freedman} {et~al.}(2001){Freedman}, {Madore}, {Gibson}, {Ferrarese},
707: {Kelson}, {Sakai}, {Mould}, {Kennicutt}, {Ford}, {Graham}, {Huchra},
708: {Hughes}, {Illingworth}, {Macri}, \& {Stetson}}]{Freedman2001}
709: {Freedman}, W.~L., {et~al.} 2001, \apj, 553, 47
710:
711: %\bibitem[{{Garnett}(1992)}]{garnett92a}
712: %{Garnett}, D.~R. 1992, \aj, 103, 1330
713:
714: \bibitem[Garnett et al.(1992)]{garnett92} Garnett, D.~R.,
715: Odewahn, S.~C., \& Skillman, E.~D.\ 1992, \aj, 104, 1714
716:
717: \bibitem[Gonz{\'a}lez Delgado et al.(1999)]{g-d99}
718: Gonz{\'a}lez Delgado, R.~M., Leitherer, C., \& Heckman, T.~M.\ 1999,
719: \apjs, 125, 489
720:
721: %% \bibitem[{{Hodge} {et~al.}(1999){Hodge}, {Balsley}, {Wyder}, \&
722: %% {Skelton}}]{hodge99}
723: %% {Hodge}, P.~W., {Balsley}, J., {Wyder}, T.~K., \& {Skelton}, B.~P. 1999, \pasp,
724: %% 111, 685
725:
726: \bibitem[Hodge et al.(2002)]{lgatlas} Hodge, P.~W., Skelton, B.~P., \&
727: Ashizawa, J.\ 2002, An Atlas of Local Group Galaxies.~By
728: Paul.~W.~Hodge, Brooke P.~Skelton, Joy Ashizawa.~University of
729: Washington, Astronomy Department, Seattle, U.S.A.~Astrophysics and
730: Space Science Library, Volume 221.~Kluwer Academic Publishers,
731: Dordrecht, 2002.
732:
733: \bibitem[{{Izotov} {et~al.}(2004){Izotov}, {Papaderos}, {Guseva}, {Fricke}, \&
734: {Thuan}}]{izotov04}
735: {Izotov}, Y.~I., {Papaderos}, P., {Guseva}, N.~G., {Fricke}, K.~J., \& {Thuan},
736: T.~X. 2004, \aap, 421, 539
737:
738: \bibitem[{{Izotov} {et~al.}(2006){Izotov}, {Stasi{\'n}ska}, {Meynet}, {Guseva},
739: \& {Thuan}}]{izotov06}
740: {Izotov}, Y.~I., {Stasi{\'n}ska}, G., {Meynet}, G., {Guseva}, N.~G., \&
741: {Thuan}, T.~X. 2006, \aap, 448, 955
742:
743: \bibitem[{{Kennicutt} {et~al.}(2003){Kennicutt}, {Bresolin}, \&
744: {Garnett}}]{ken_m101}
745: {Kennicutt}, R.~C., {Bresolin}, F., \& {Garnett}, D.~R. 2003, \apj, 591, 801
746:
747: \bibitem[Kwitter \& Aller(1981)]{ka81} Kwitter, K.~B., \&
748: Aller, L.~H.\ 1981, \mnras, 195, 939
749:
750: \bibitem[Lee et al.(2002)]{lee02} Lee, M.~G., Kim, M., Sarajedini, A.,
751: Geisler, D., \& Gieren, W.\ 2002, \apj, 565, 959
752:
753: \bibitem[Magrini et al.(2004)]{magrini04} Magrini, L., Perinotto,
754: M., Mampaso, A., \& Corradi, R.~L.~M.\ 2004, \aap, 426, 779
755:
756: \bibitem[Magrini et al.(2007)]{magrini07} Magrini, L., Vilchez,
757: J.~M., Mampaso, A., Corradi, R.~L.~M., \& Leisy, P.\ 2007, \aap, in
758: press (arXiv e-prints, arXiv:0705.3116)
759:
760: \bibitem[Massey et al.(1988)]{massey88} Massey, P., Strobel, K.,
761: Barnes, J.~V., \& Anderson, E.\ 1988, \apj, 328, 315
762:
763: \bibitem[Massey et al.(2006)]{massey-m33m31} Massey, P., Olsen,
764: K.~A.~G., Hodge, P.~W., Strong, S.~B., Jacoby, G.~H., Schlingman, W.,
765: \& Smith, R.~C.\ 2006, \aj, 131, 2478
766:
767: \bibitem[Mayall \& Aller(1942)]{ma42} Mayall, N.~U., \&
768: Aller, L.~H.\ 1942, \apj, 95, 5
769:
770: \bibitem[Oke et al.(1995)]{oke95} Oke, J.~B., et al.\ 1995, \pasp,
771: 107, 375
772:
773: \bibitem[Olive \& Skillman(2001)]{os01} Olive, K.~A., \&
774: Skillman, E.~D.\ 2001, New Astronomy, 6, 119
775:
776: \bibitem[{{P{\'e}rez-Montero} {et~al.}(2007){P{\'e}rez-Montero}, {H{\"a}gele},
777: {Contini}, \& {D{\'{\i}}az}}]{perez-montero07}
778: {P{\'e}rez-Montero}, E., {H{\"a}gele}, G.~F., {Contini}, T., \& {D{\'{\i}}az},
779: {\'A}.~I. 2007, \mnras, 779
780:
781:
782: \bibitem[{{Press} {et~al.}(1992){Press}, {Teukolsky}, {Vetterling}, \&
783: {Flannery}}]{numrec}
784: {Press}, W.~H., {Teukolsky}, S.~A., {Vetterling}, W.~T., \& {Flannery}, B.~P.
785: 1992, {Numerical recipes in C. The art of scientific computing} (Cambridge:
786: University Press, 1992, 2nd ed.)
787:
788: \bibitem[Rosolowsky et al.(2003)]{eros03} Rosolowsky, E.,
789: Engargiola, G., Plambeck, R., \& Blitz, L.\ 2003, \apj, 599, 258
790:
791: \bibitem[Searle(1971)]{searle71} Searle, L.\ 1971, \apj, 168,
792: 327
793:
794: \bibitem[{{Shaw} \& {Dufour}(1995)}]{sd95}
795: {Shaw}, R.~A. \& {Dufour}, R.~J. 1995, \pasp, 107, 896
796:
797: %\bibitem[{{Skillman} \& {Kennicutt}(1993)}]{sk93}
798: %{Skillman}, E.~D. \& {Kennicutt}, Jr., R.~C. 1993, \apj, 411, 655
799:
800: \bibitem[Smith(1975)]{smith75} Smith, H.~E.\ 1975, \apj, 199,
801: 591
802:
803: \bibitem[{{Stasi{\'n}ska} {et~al.}(2001){Stasi{\'n}ska}, {Schaerer}, \&
804: {Leitherer}}]{stasinska01}
805: {Stasi{\'n}ska}, G., {Schaerer}, D., \& {Leitherer}, C. 2001, \aap, 370, 1
806:
807: \bibitem[{{Stasi{\'n}ska} {et~al.}(2006){Stasi{\'n}ska}, {V{\'{\i}}lchez},
808: {P{\'e}rez}, {Gonzalez Delgado}, {Corradi}, {Mampaso}, \&
809: {Magrini}}]{stan06-pne}
810: {Stasi{\'n}ska}, G., {V{\'{\i}}lchez}, J.~M., {P{\'e}rez}, E., {Gonzalez
811: Delgado}, R.~M., {Corradi}, R.~L.~M., {Mampaso}, A., \& {Magrini}, L. 2006,
812: in Planetary Nebulae Beyond the Milky Way, ed. L.~{Stanghellini}, J.~R.
813: {Walsh}, \& N.~G. {Douglas}, 234
814:
815: \bibitem[Urbaneja et al.(2005)]{urbaneja05} Urbaneja, M.~A., Herrero,
816: A., Kudritzki, R.-P., Najarro, F., Smartt, S.~J., Puls, J., Lennon,
817: D.~J., \& Corral, L.~J.\ 2005, \apj, 635, 311
818:
819: \bibitem[V{\' i}lchez et al.(1988)]{vilchez88} V{\' i}lchez, J.~M.,
820: Pagel, B.~E.~J., D{\' i}az, A.~I., Terlevich, E., \& Edmunds,
821: M.~G.\ 1988, \mnras, 235, 633 (V88)
822:
823: \bibitem[Willner \& Nelson-Patel(2002)]{wnp02} Willner, S.~P., \&
824: Nelson-Patel, K.\ 2002, \apj, 568, 679 (WNP02)
825:
826: %% \bibitem[{{Wyder} {et~al.}(1997){Wyder}, {Hodge}, \& {Skelton}}]{wyder97}
827: %% {Wyder}, T.~K., {Hodge}, P.~W., \& {Skelton}, B.~P. 1997, \pasp, 109, 927
828:
829: \end{thebibliography}
830:
831:
832:
833:
834: \clearpage
835: \LongTables
836: \begin{landscape}
837: \begin{deluxetable}{lcccccccccc}
838: \tabletypesize{\tiny}
839: %\tabletypesize{\small}
840: %\tabletypesize{\footnotesize}
841: \tablecaption{\label{linetable} \hii\ Region Line Fluxes}
842: \tablewidth{0pt}
843: \tablehead{ \colhead{Obj.} &
844: \colhead{Coordinates\tablenotemark{a}} & \colhead{Name\tablenotemark{b}} &\colhead{$R_{gal}$} &
845: %\colhead{$c$(H$\beta$)} &
846: %\colhead{EW Abs.} &
847: \colhead{[\ion{O}{2}]\tablenotemark{c}} &
848: \colhead{[\ion{Ne}{3}]\tablenotemark{c}} &
849: \colhead{[\ion{O}{3}]\tablenotemark{c}} &
850: \colhead{[\ion{O}{3}]\tablenotemark{c}} &
851: \colhead{[\ion{O}{3}]\tablenotemark{c}} &
852: \colhead{$T_{e}$([\ion{O}{3}])} &
853: \colhead{$12+\log([\mathrm{O/H}])$}\\
854: \colhead{} &
855: \colhead{($\alpha_{2000},\delta_{2000}$)} &
856: \colhead{} &
857: \colhead{(kpc)} &
858: %\colhead{} &
859: %\colhead{(\AA)} &
860: \colhead{$\lambda 3727$~\AA} &
861: \colhead{$\lambda 3869$~\AA} &
862: \colhead{$\lambda 4363$~\AA} &
863: \colhead{$\lambda 4959$~\AA} &
864: \colhead{$\lambda 5007$~\AA} &
865: \colhead{(K)} &
866: \colhead{}
867: }
868: \startdata
869: \input{tab1.tex}
870: \enddata
871:
872: \tablenotetext{a}{Actual position observed based on targets identified
873: in the Local Group Survey images of \citet{massey-m33m31}.}
874: \tablenotetext{b}{Closest matching cataloged \ion{H}{2} region for
875: observed coordinates. Catalog abbreviations are: MA = \citet{ma42},
876: S = \citet{searle71}, B = \citet{bou74}, C = \citet{cou87}. Full
877: designations of substructure within individual \hii\ regions
878: (indicated by the letters at the end of the catalog names) are taken
879: from \citet{lgatlas}.}
880: \tablenotetext{c}{Extinction corrected line fluxes relative to H$\beta=1$.}
881: \tablenotetext{d}{The same object was observed on multiple slitmasks.}
882: \end{deluxetable}
883:
884:
885: \begin{deluxetable}{ccccllllcllll}
886: \tabletypesize{\scriptsize}
887: \tablewidth{0pt}
888: \tablecolumns{13}
889: \tablecaption{Balmer Series Line Ratios and Derived Properties}
890: \tablehead{
891: \colhead{Obj.} & \colhead{EW H$\beta$\tablenotemark{a}} & \colhead{c(H$\beta$)} &
892: \colhead{Stellar Absorption EW} & \multicolumn{4}{c}{Observed} & \colhead{} & \multicolumn{4}{c}{Extinction / Absorption Corrected}\\
893: \cline{5-8} \cline{10-13}
894: \colhead{} & \colhead{(\AA)} & \colhead{} & \colhead{(\AA)} &
895: \colhead{H$\gamma$} & \colhead{H$\delta$} &
896: \colhead{H7} & \colhead{H9} & \colhead{} & \colhead{H$\gamma$} &
897: \colhead{H$\delta$} & \colhead{H7} & \colhead{H9}
898: }
899: \startdata
900: \input{tab2.tex}
901: \enddata
902: \tablecomments{\ Line fluxes are given relative to H$\beta$=1 and rounded
903: to the precision of one significant figure in the error.}
904: \tablenotetext{a}{The equivalent width of H$\beta$ in emission.}
905: \tablenotetext{b}{Because there is no discernible continuum emission
906: associated with the nebular line emission from this \hii\ region, the
907: equivalent width of the H$\beta$ emission is undefined.}
908: \label{hlinetable}
909: \end{deluxetable}
910: \clearpage
911: \end{landscape}
912:
913:
914:
915: \end{document}
916: