1: \documentclass[preprint2]{aastex}
2: %\documentstyle[12pt,aasms4]{article}
3:
4: \begin{document}
5:
6: \title{FAR ULTRAVIOLET IMAGERY OF THE EDGE-ON SPIRAL GALAXY NGC~4631}
7:
8: \author{Andrew M. Smith\altaffilmark{1}, Nicholas R. Collins\altaffilmark{2},
9: William H. Waller\altaffilmark{3},
10: Morton S. Roberts\altaffilmark{4}, \\ Denise A. Smith\altaffilmark{5},
11: Ralph C. Bohlin\altaffilmark{5}, K.-P. Cheng\altaffilmark{6},
12: Michael N. Fanelli\altaffilmark{7},
13: Susan G. Neff\altaffilmark{1}, Robert W. O'Connell\altaffilmark{8},
14: Ronald A. Parise\altaffilmark{9},
15: Eric P. Smith\altaffilmark{1},
16: and Theodore P. Stecher\altaffilmark{1}}
17:
18: \altaffiltext{1}{Laboratory for Astronomy and Solar Physics, Code 681, NASA/GSFC, Greenbelt MD 20771}
19: \altaffiltext{2}{Raytheon STX Corporation, Code 682, NASA/GSFC, Greenbelt MD 20771}
20: \altaffiltext{3}{Raytheon ITSS Corporation, Harvard-Smithsonian Center for Astrophysics and Tufts University. Current Address: Department of Physics and Astronomy, Tufts University, Medford MA 02155}
21: \altaffiltext{4}{National Radio Astronomy Observatory, Edgemont Road, Charlottesville, VA 22903}
22: \altaffiltext{5}{Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218}
23: \altaffiltext{6}{Department of Physics, California State University, Fullerton, 800 N. State College Ave., Fullerton, CA 92634}
24: \altaffiltext{7}{Department of Physics, University of North Texas, P.O. Box 305370, Denton, TX 76203}
25: \altaffiltext{8}{Department of Astronomy, University of Virginia, P.O. Box 3018, Charlottesville, VA 22903}
26: \altaffiltext{9}{Computer Sciences Corp., Code 686.9, NASA/GSFC, Greenbelt MD 20771}
27:
28:
29: \begin{abstract}
30:
31: Far ultraviolet ($FUV$) imagery of the edge-on, Sc/SBd galaxy,
32: NGC~4631 reveals very strong $FUV$ emission, resulting from active
33: star formation, uniformly distributed along the galactic mid-
34: plane. Multi-band imagery, \ion{H}{1} and
35: \ion{H}{2} position-velocity curves and extinction considerations all imply
36: that the emission is from the outer edges of the visible galaxy. The
37: overall $FUV$ morphology of this edge-on disk system is remarkably
38: similar to those of the so-called ``chain galaxies'' evident at high
39: redshift, thus suggesting a similar interpretation for at least some
40: of those distant objects. $FUV, U, B$ and $V$ magnitudes, measured
41: for 48 star forming regions, along with corresponding $H\alpha$ and
42: $H\beta$ measurements are used to construct diagnostic color-color
43: diagrams. Although there are significant exceptions, most of the star
44: forming regions are less massive and older than 30~Doradus. Comparison
45: with the expectations from two star formation models yields ages of
46: 2.7 to 10~Myr for the instantaneous burst (IB) model and star
47: formation cut-off ages of 0 to 9~Myr for the continuous star formation
48: (CSF) model. Interpreted in terms of the IB model the photometry
49: implies a total created mass in the 48 star forming regions of
50: 2.5$\times$10$^{7}$~M$_{\sun}$. When viewed as resulting from constant star formation the
51: photometry implies a star formation rate of 0.33~M$_{\sun}$~yr$^{-1}$. These
52: results are compared to those derived from FIR and radio
53: observations. Corrections for $FUV$ emission reprocessed by interstellar
54: grains are estimated.
55:
56: A large ring, $\sim$3~kpc in diameter, of 14 star forming regions is
57: concentrically located with an expanding \ion{H}{1} shell \citep{ran93}
58: toward the eastern end of the galaxy. Our
59: observations imply that the shell may have been generated primarily by
60: supernovae arising from 5.3$\times$10$^{4}$ OB stars in a massive star forming
61: region beginning about 20~Myr ago, and that the presently observed $FUV$ bright
62: emission is due to second generation stars.
63:
64: \end{abstract}
65:
66: \keywords{galaxies: individual (NGC~4631)---galaxies: ISM---galaxies: starburst---galaxies: star clusters---ultraviolet: galaxies}
67:
68: \section{Introduction}
69:
70: The group of galaxies including NGC~4631 provides an outstanding
71: example of a galaxy interaction accompanied by intensive star
72: formation. Far Ultraviolet ($FUV$) imagery of the Sc/SBd galaxy
73: NGC~4631, recorded by the Ultraviolet Imaging Telescope (UIT),
74: exhibits very bright $FUV$ emission from sources corresponding to the
75: \ion{H}{2} regions cataloged by \citet{cri69}
76: along with substantial diffuse $FUV$ light. The observed
77: emission is, in our experience, extraordinary in that NGC~4631 is
78: observed nearly edge-on, and strong attenuation of $FUV$ light by dust
79: in the galactic disk would be anticipated. Such is the case for
80: NGC~891, another highly inclined galaxy and M82, an edge-on starburst
81: galaxy -- both of which show very little detectable $FUV$ emission.
82: (Images of NGC~891 and M82 are accessible at
83: http://archive.stsci.edu/astro/uit/index.html). Together with the
84: $H\alpha$ imagery obtained by \citet{ran92}
85: the $FUV$ results signify widespread intensive star formation along
86: the observable disk of NGC~4631. The overall $FUV$ morphology is
87: remarkably similar to those of the so-called ``chain galaxies'' that
88: have been observed at high redshift \citep{cow95}.
89: It may be that a significant fraction of these distant
90: $FUV$ sources is composed of highly inclined disk galaxies rather than
91: dynamically unstable chains of sub-galactic clumps. See \citet{smi97}
92: for further details.
93:
94: Observations of the radio halo imply that the galactic-wide star
95: formation in NGC~4631 is sufficiently intense as to supply cosmic rays
96: to the halo through bursting supershells formed by supernovae and
97: stellar winds in the disk \citep{dah95,gol94a}.
98: Optical images reveal a
99: patchy distribution of dust rather than a well defined dust
100: lane. Thus, the extensive $FUV$ emission may result from the combined
101: dispersal of dust by these bursting supershells and by forces
102: generated through tidal interactions NGC~4631 and NGC~4656
103: \citep{com78}.
104:
105: There is also evidence for an inner disk or torus (r$\sim$50$\arcsec$) revealed
106: most clearly in CO(1-0) and CO(2-1) line emissions
107: \citep{gol94b,sof89,sof90}. The radio continuum emission coincides spatially
108: with the CO emission and has a substantial non-thermal component
109: \citep{dur82}.
110: There also appears to be a region of enhanced star
111: formation near the center (R$<$360~pc) of NGC~4631 \citep{dur82,sof89} coinciding
112: closely with the center of the galaxy as measured in the infrared by
113: \citet{aar78}. More recently, \citet{wan95} reported
114: the ROSAT detection of soft X-rays extending out to 8~kpc above the galactic
115: midplane. When interpreted in terms of the burst supershell model, this
116: observation implies substantial star formation near the center of NGC~4631.
117: Indeed, considering mostly the radio data, \citet{gol94b}
118: view NGC~4631 as a galaxy with a {\it mild} central
119: starburst.
120:
121: Of interest is a supershell with a diameter of 3~kpc detected at 21~cm
122: by \citet{ran93}. The shell
123: corresponds to a distortion in the disk near the eastern end of the
124: galaxy appearing in all optical images and in a particularly obvious
125: way in both the $FUV$ and $H\alpha$ images. A preliminary explanation
126: for the shell is that it is produced by the combined effects of
127: stellar winds and supernovae from OB associations containing
128: (1-3.5)$\times$10$^{4}$ OB stars \citep{ran93}.
129: Alternatively, these authors suggest that the
130: eastern cavity may have been punched into the galactic disk by a high
131: velocity cloud (HVC) which itself originated from the tidal stripping
132: of the galactic disk during a prior interaction with NGC~4656. A
133: collision with a dwarf galaxy could have also produced the same
134: result.
135:
136: In this paper we present the $FUV$ imagery of NGC~4631. Using these data together
137: with recently obtained optical imagery, we infer the star formation history of
138: the bright $FUV$ emitting associations including those circumscribed by the Rand
139: and van der Hulst super shell. The implications of these results on the
140: ultraviolet emission morphology, local and global star formation, and the
141: viability of the starburst shell and HVC impact hypothesis are discussed. In
142: all subsequent discussions, we place the galactic center at
143: RA(2000)~=~12$^{h}$39$^{m}$ 39$\fs$8, DEC(2000)~=~+32\arcdeg 48\arcmin
144: 48\arcsec, adopt a galactic distance of 7.5~Mpc \citep{gol94a}
145: and abbreviate the \ion{H}{2} regions observed by
146: \citet{cri69} as CM(\#).
147:
148: \section{Observations}
149:
150: NGC~4631 was observed by the UIT through the B1 filter
151: ($\lambda_{eff}$~=~152.0~nm; $\Delta\lambda$~=~35.4~nm) for
152: 1140~seconds on March 10, 1995 during the Spacelab ASTRO-2
153: mission. The instrument is completely described in
154: \citet{ste92,ste97}. $FUV$ images,
155: intensified and down-converted to visible wavelengths by a two stage
156: magnetically focussed intensifier, were recorded on IIa-O film,
157: digitized with a PDS microdensitometer, and reduced as described in
158: \citet{ste92,ste97}.
159: The plate scale of 1\arcmin~mm$^{-1}$ together with a
160: microdensitometer aperture of 0.021~mm resulted in a pixel size of
161: 1.26\arcsec. The FWHM of the point spread function was 3.4\arcsec~
162: over a circular 40\arcmin~ field-of-view.
163:
164: Ground based images of NGC~4631 were also recorded in the standard $U, B, V,$
165: and $R$ bandpasses and in $H\alpha$ and $H\beta$ line emission. The
166: observations are summarized in Table~1. The $FUV,$ $H\alpha$, $U, B,$ and
167: $V$-band images are shown in Figure~1.
168:
169: %\begin{table}
170: %\dummytable\label{tab:1}
171: %\end{table}
172:
173:
174: The absolute sensitivity of the UIT was determined by observations of stars the
175: spectra of which were measured by the IUE
176: \citep{ste97}. This procedure resulted in an uncertainty in the
177: absolute flux measurements of $\sim$10\%. Relative sensitivity throughout the
178: B1 passband was measured in the laboratory. Relative transmissions anywhere in
179: the passband were accurate to 2\% at transmission levels above 1\% of the
180: maximum transmission. At 320~nm the instrument response is reduced from the
181: maximum response by more than 5~dex and at 330~nm by more than 8~dex.
182:
183: The UIT image was corrected for geometric image (``S'') distortion according to
184: the formulation developed by \citet{gre94}.
185: Each of the ground based images was then transformed to the same plate
186: scale and orientation as the UIT image.
187:
188: The absolute sensitivity of the $U, B, V,$ and $R$ band images was determined
189: using observations of the spectrophotometric standard star BD+33~2642. After
190: establishing the calibration using the spectrophotometry of
191: \citet{sto77}, the $U, B,$ and $V$ magnitudes were cross checked
192: against the broad band measurements recorded by
193: \citet{kle62}. The values match within 5\%.
194:
195: Astrometry of the broad band $U, B, V, R$ and $H\beta$ images was determined
196: using stars in common with the {\sl Digitized Guide Star Catalog} of \citet{las90}.
197: The astrometric solution of the UIT image
198: was based on positions of star forming knots in common with the KPNO U-band
199: image yielding a positional uncertainty of $\pm$~0.74~arcseconds.
200:
201: Continuum subtraction in the $H\alpha$ image was performed using foreground
202: stars in both on-line and off-line images. Observations of the planetary
203: nebula NGC~7027 were used for absolute calibration. Astrometry for the
204: $H\alpha$ image was adopted from \citet{ran92},
205: and differs from the broad-band astrometry by no more than 1.1~arcseconds.
206:
207: The on-line and off-line $H\beta$ images were separately calibrated using the
208: spectrophotometric standard star, Kopff~27 \citep{sto77}. After
209: absolute calibration, a scale factor of 1.22 was applied to the off-line image
210: to allow removal of continuum sources, i.e. star images from the field. After
211: continuum subtraction was performed, the flux of the UV-bright region in the
212: disk of NGC~4631, CM-67, as measured through a 21\arcsec~ diameter synthetic
213: aperture was 2.34$\times$10$^{-13}$ ergs s$^{-1}$ cm$^{-2}$. This value is 6\%
214: greater than the equivalent measurement by \citet{roy91}.
215:
216:
217:
218: \section{Morphology}
219:
220: Inspection of the V and B images shows dust silhouetted against the
221: red stars of the nuclear bulge. Most of the dust appears to be
222: concentrated near the galactic center; its morphology corresponds
223: roughly to the distribution of CO in an interior disk as measured by
224: \citet{gol94b}, \citet{sof90}, and of the dust as observed by
225: \citet{bra95}, in 1.3~mm continuum emission. Its
226: distribution is chaotic extending above the galactic midplane by as
227: much as 1.3~kpc. The B, V and $H\alpha$ images show that the cores of
228: some of the CM \ion{H}{2} regions are silhouetted against the dust
229: patches, implying that much of the bright $FUV$ emission originates
230: edgeward from the central 4~kpc of NGC~4631.
231:
232: This impression is reinforced by the measured \ion{H}{1} and \ion{H}{2}
233: velocity-position curves. The \ion{H}{1} velocity-position curves
234: \citep{wel78,sof90} imply a
235: constant \ion{H}{1} velocity at distances greater than $\sim$1\arcmin~ from
236: the galactic center. However, at these distances the $H\alpha$ data of
237: \citet{cri69} show a linear increase in the
238: observed velocity as a function of increasing distance from the galactic
239: center. Such behavior would result if the \ion{H}{2} gas velocity were
240: constant beyond $\sim$1\arcmin~ from the galactic center, and the emission
241: was from gas located at an approximately constant galactocentric radius. Since the bright
242: FUV/\ion{H}{2} regions extend 24~kpc along the {\it entire} major axis of the imaged
243: galaxy, we deduce that most of the \ion{H}{2} regions are located near the
244: galactic periphery.
245:
246: We also note that if the presently observed $FUV$ emission regions were to be
247: further obscured by dust in an interstellar gas of density 0.5~cm$^{-3}$, as
248: indicated by \citet{ran94} 21~cm measurements of the NGC~4631
249: midplane, an effective gas depth of 1.2~kpc would cause more than half of the
250: observed UV-bright regions to be undetectable by the UIT. To obtain this
251: result, the relationship
252:
253: \begin{displaymath}
254: \langle N(HI + H_{2}) / E(B-V) \rangle = 5.8\times10^{21}~atoms~cm^{-2}~mag^{-1}
255: \end{displaymath}
256: \noindent
257: \citep{boh78}, determined from measurements
258: along 75 lines-of-sight in the Galaxy, is assumed to be valid. We do not
259: include the molecular gas density in our estimate since this quantity is
260: unknown; its inclusion would further reduce the effective gas depth. We also
261: assume that a detection would be at the 2$\sigma$ level. Even though the dust
262: is observed to be patchy, the fact that the entire major axis is delineated by
263: bright $FUV$ emission requires that many of the $FUV$ emission regions are
264: near the visual periphery of the galaxy.
265:
266: Finally we note that peripheral \ion{H}{2} regions and $FUV$ emitting sources
267: are commonplace in many Sc-Sd type galaxies, including M101
268: \citep{wal97a}, M74 \citep{cor94} and M83 \citep{boh83}.
269:
270: Generally, the $FUV$ emission coincides closely with the $H\alpha$
271: emission, much of which shows significant departures from the midplane
272: as defined by the symmetric distribution of \ion{H}{2} regions about
273: the major axis (see \citet{cri69}). If this
274: coincidence is related to the distribution of young, massive stars,
275: and contributions to the observed $FUV$ light from distant regions of
276: the galaxy are masked by the UV opacity of the displaced gas then star
277: formation must be occurring at substantial distances
278: ($\leq$~36$\arcsec$, 1.3~kpc) above the galactic midplane. This does
279: not include the even greater extent of Rand's shell 1 perpendicular to
280: the galactic midplane (z~$\lesssim$~1.7~kpc). Some unknown fraction of
281: the high diffuse $FUV$ emission at high altitude is likely due to dust
282: scattered light.
283:
284: Comparison of the $FUV$ and $H\alpha$ images indicates that the sizes
285: of the \ion{H}{2} and star forming regions are about the same as well
286: as being co-spatial. This implies that the B-type stars which
287: account for most of the $FUV$ light have not moved far from their
288: birth place. Typically, a star forming region which appears to be
289: a single rounded area unblended with other regions in the UIT
290: image subtends an angle of about 8~arcseconds corresponding to a
291: linear distance of 291~pc at the assumed distance of 7.5~Mpc.
292: Moving with a speed of 10~km~s$^{-1}$ stars would disperse to a width
293: of 291~pc in 14~Myr. Although some of the $FUV$ bright regions are
294: clearly blends of smaller regions making geometric comparisons
295: impossible, we expect that many of the star forming regions must
296: be on the order of 14~Myr old or younger.
297:
298:
299: \section{Photometry}
300:
301: \subsection{$FUV$ and Visual Photometry of 48 Regions}
302:
303: Figure~2 is the $FUV$ image which shows hand-drawn boundaries and numbering of
304: the apertures in which photometry was performed. The criteria for
305: establishing the numbered boundaries were (1) that each should include at
306: least one unresolved $FUV$ emission region that is also seen clearly in the B
307: image, (2) that each should include all the $FUV$ reflection nebulosity
308: belonging to the selected emission regions, and (3) that each should include
309: all the compact \ion{H}{2} regions and $H\alpha$ filaments apparently
310: belonging to the selected associations and not to the general background. The
311: boundaries thus determined were also applied to the $U$, $B$, $V$, $H\alpha$ and
312: $H\beta$ images. The intent was to delineate those data which will yield a
313: consistent model of star formation within each boundary. Similarly, a large
314: aperture drawn around the entire galactic image in each broad band and
315: emission line permitted global quantities to be computed from spatially
316: integrated fluxes.
317:
318: The sky background, subtracted from each image, is determined from the
319: responses in 50 rectangular apertures distributed over the entire
320: field-of-view and located in clear regions between stellar images. The
321: apertures are 31$\times$31~pixels$^{2}$ and the standard deviation
322: of the distribution of the pixel responses within an aperture were
323: taken to indicate the rms variation in the pixel-to-pixel response. In
324: flux density units the pixel-to-pixel variation is
325: 4.3~$\times$~10$^{-18}$~ergs~cm$^{-2}$~s$^{-1}$~\AA$^{-1}$ in the $FUV$
326: image and
327: $<$~1.4~$\times$~10$^{-18}$~ergs~cm$^{-2}$~s$^{-1}$~\AA$^{-1}$ in the
328: $U$, $B$ and $V$ images. The sky background is the average of the responses
329: in the 50 apertures. The standard deviation of the average responses
330: in the 50 apertures is taken to indicate the uniformity of the sky
331: over the field-of-view. This variation is
332: $<$~5.1$\times$10$^{-19}$~ergs cm$^{-2}$ s$^{-1}$~\AA$^{-1}$ in the
333: FUV image and
334: $<$~3.6$\times$10$^{-19}$~ergs~cm$^{-2}$~s$^{-1}$~\AA$^{-1}$ in the
335: $U$, $B$ and $V$ images. The variation in sky uniformity makes a negligible
336: contribution to the total random error in the derived flux
337: densities. However, the contribution of the pixel-to-pixel variation
338: can be as large as 28\%, although it is generally less than 10\%.
339:
340: Correction for scattered light and sky background within the
341: individual apertures was achieved in the following manner. Each
342: aperture was expanded by 2~pixels (2.5~arcseconds) at its edge to create
343: annuli between the original and expanded apertures. Histograms of the
344: pixel response to the background included in the annulus were made
345: from which the mean value was taken to be representative of the
346: background from all sources underlying the emission from the recently
347: formed stars and the gas excited by those stars. The total background
348: was simply the mean value multiplied by the number of pixels
349: encompassed by the original aperture.
350:
351: Other than the correction for the sky background, no further
352: correction was made to the integrated flux density of each image
353: within the large aperture. Expanding and shrinking the large aperture
354: by 8~pixels or 20~arcseconds changes the integrated flux by 1\% or less
355: in all of the broad-band images. Thus, the integrated flux is quite
356: insensitive to the actual position of the large aperture on the sky.
357:
358: The contribution to the observed $FUV$ continuum arising from two photon
359: emission in the ionized gas was evaluated and found to be generally at the 1\%
360: level with a maximum of 4\% of the observed continuum. This effect was
361: neglected in subsequent calculations.
362:
363: Error will be made to the extent that the stellar emission, associated
364: reflected light and line emission from the individual regions
365: cannot be uniquely isolated from light emitted from the rest of the
366: galaxy. This is particularly true for the $B$, $V$, and $R$ images and is
367: exacerbated by the edge-on aspect of the object. The problem is so
368: severe for the $R$-band image that the aperture photometry is very
369: inaccurate, and is for this reason not included in the subsequent
370: analysis. In addition, uncertainties
371: arising from the background correction procedure can be
372: substantial, particularly for faint sources. Some indication of
373: the magnitude of this systematic error can be had by expanding
374: and contracting the apertures by 2~pixels (2.5~arcseconds). The
375: result is that sky corrected flux densities change by an average
376: factor of 1.5 for the broad-band photometry and factors of 1.4 and
377: 1.7 respectively for the $H\alpha$ and $H\beta$ line photometry.
378:
379: Listed in Table~2 are the photometry results corrected for background
380: and extinction in the Galaxy. The latter correction was accomplished
381: using the Galactic value of E(B-V)~=~0.007 \citep{dev91}
382: (see note b, Table~2) with
383: the \citet{sea79} extinction law. The aperture number as in Figure~2
384: and CM \ion{H}{2} regions included in each aperture are listed in
385: column 1 and 2 respectively. The observed $FUV$ (152~nm), $U, B,$ and
386: $V$ fluxes in units of 10$^{-15}$~ergs~cm$^{-2}$~s$^{-1}$~\AA$^{-1}$
387: are listed in columns 3, 4, 5 and 6 respectively, and the $H\alpha$
388: and $H\beta$ fluxes in units of 10$^{-15}$~ergs~cm$^{-2}$~s$^{-1}$ are
389: listed in columns 7 and 8 respectively. Values of $E(B-V)$ computed
390: from the Balmer Decrement are given in column 9. The second number in
391: each column is the 1$\sigma$ standard deviation due to random errors
392: in the measurements only. The next to last row shows the sums of each
393: column for columns 2 through 7, i.e. the sum of the fluxes in all 48
394: apertures. The last row shows the flux values integrated over the
395: entire galactic image inside the large aperture.
396:
397: %\begin{table}
398: %\dummytable\label{tab:2}
399: %\end{table}
400:
401: The $B-V$ color excesses were computed from Balmer Decrement
402: and from the difference between the measured $B-V$
403: colors and $B-V$ colors determined from models of evolving star
404: forming clusters. The models (see Section~5, below) describe the
405: evolution of clusters in which the stars are all coeval, sometimes
406: called ``instantaneous burst'', or IB models. In principle, the derivation
407: of $E(B-V)$ in this way is more realistic than using the Balmer
408: Decrement since it is related to the extinction of stellar continuum
409: light rather than the extinction of emission from nearby excited
410: gas. However, the method suffers from the the need for
411: {\it a priori} knowledge of the correct IB model parameters in order to
412: infer $E(B-V)$. With the present data-set, the models were characterized
413: only by their burst ages which were determined from the ratio of the
414: number of Lyman continuum photons to the $FUV$ luminosity, and these
415: latter quantities were determined from the $FUV$ and $H\alpha$
416: observations corrected for dust extinction. Values for all other model
417: parameters were assumed as described below in Section~5.
418:
419: The distribution of $E(B-V)$ values determined from the Balmer
420: Decrements is shown by the red line in Figure~3. A single
421: gaussian fit to the distribution results in a mean value of $E(B-V) =$~0.39.
422: Though the distribution is sharply peaked, there are
423: broad wings at low occurrence levels extending to unrealistic
424: negative $E(B-V)$ values. We take these broad wings to be the
425: result principally of significant errors in the $H\beta$ flux density
426: measurements. The less important part of the error is
427: statistical; the most important part, as discussed above, is due
428: to our inability to identify the relevant emission, and is
429: estimated by the variable aperture photometry. Because of these
430: errors we conclude that use of the individual color excesses to
431: correct the measured flux densities for extinction is
432: inappropriate. Rather, we determine a mean value to be applied to
433: all the photometry by means of an iterative procedure. As an
434: initial step, we use the mean value of the distribution computed
435: from the Balmer Decrement to correct the $FUV$ and $H\alpha$ measurements
436: in each region for extinction using the extinction law of
437: \citet{cal94}. From these corrected values we compute models
438: for each region which are characterized by unique burst times.
439: Subsequently, $E(B-V)$ values are derived from the modeled and
440: observed colors for each region from which a new mean value is
441: determined. This process is repeated until the mean values of
442: $E(B-V)$ converge. The final distribution of $E(B-V)$ values, shown
443: as a black line in Fig 3, exhibits occurrence levels in the wings
444: lower than those associated with the Balmer Decrement
445: measurements, and a mean value of $E(B-V) =$~0.28 which is less by
446: a factor of 0.7 than the value derived from the Balmer Decrement
447: measurements. Qualitatively, this difference in derived values of
448: $E(B-V)$ is characteristic of stars in regions of active star
449: formation. However, the factor by which results of the two
450: methods differ is usually smaller, being about 0.44
451: \citep{cal94,fan88}.
452: The 2$\sigma$ width of the Gaussian function
453: fitted to the final distribution is 0.36~mag.
454:
455: Several photometric quantities, which have been corrected for internal
456: extinction, are listed in Table~3. As in Table~2, the aperture number appears
457: in column 1. Magnitudes in the $FUV, U, B, $ and $V$ bands are listed in
458: columns 2, 3, 4 and 5 respectively. The $H\alpha$ luminosity, the log of the
459: ratio of the number of Lyman continuum quanta to the $FUV$ luminosity,
460: log$_{10}$(N$_{Lyc}$/L$_{152}$), and the
461: M$_{152}$-$U$, M$_{152}$-$B$ and M$_{152}$-$V$ colors are listed in columns 6,
462: 7, 8, 9 and 10 respectively. The associated formal random measurement errors
463: (1$\sigma$),excluding those arising from the extinction
464: corrections are listed adjacent to each quantity in each column.
465: Uncertainty in the extinction correction will result in
466: significant uncertainty in the listed quantities. Increasing or
467: decreasing the mean value of $E(B-V)$ by the standard deviation in
468: the $E(B-V)$ distribution changes the $FUV$, $U$, $B$ and $V$ magnitudes by
469: 0.80, 0.50, 0.43 and 0.36 mag respectively, the $H\alpha$ luminosity by
470: a factor of 2 and the M$_{152}-U$, M$_{152}-B$ and M$_{152}-V$ colors by an average
471: of 0.37~mag. The change in aperture sizes results in an average
472: change in broad-band magnitudes of 0.56 mag, a change in the H$\alpha$
473: luminosity by 35\%, and an average change in the M$_{152}-U$, M$_{152}-B$ and
474: M$_{152}-V$ colors of 0.37~mag. In general, random errors in quantities
475: expressed as magnitudes are comparable to the changes induced by
476: variation of the aperture sizes. The errors do not include the
477: systematic errors associated with the absolute calibration or the
478: choice of the reddening law.
479:
480: %\begin{table}
481: %\dummytable\label{tab:3}
482: %\end{table}
483:
484:
485: \subsection{Integrated Photometry of NGC~4631}
486:
487: The last line in Table~2 gives the integrated photometry of that part
488: of NGC~4631 which can be observed, in particular all of the $FUV$ image inside
489: the large aperture. A comparison with two possible analog galaxies,
490: M101 (Sc(s)I) and M83 (SBc(s)II)), both observed face-on is given in
491: Table~4. Results from integrated UIT photometry of two additional
492: face-on galaxies, M74 (Sc(s)I) and M51 (Sbc(s)I-II), are also
493: listed. Excepting NGC~4631, the $FUV$ luminosities have been derived
494: from the results of \citet{wal97b},
495: with corrections only for extinction arising in the Galaxy, and with
496: modifications to take into account differences in the galactic
497: distances listed by \citet{wal97b} and
498: those used in this paper. The NGC~4631 152.1~nm flux is a factor of
499: 1.1 brighter than the 155.0~nm flux
500: (9.81$\times$10$^{-13}$~ergs~cm$^{-2}$~s$^{-1}$~\AA$^{-1}$) measured
501: by \citet{cod82} using the University of
502: Wisconsin experiment on the first {\sl Orbiting Astronomical
503: Observatory} (OAO). The measured NGC~4631 $FUV$ luminosity is dimmer
504: by factors of 1.2 and 4.0 than the measured luminosities of M83 and
505: M101 respectively.
506:
507: Far infrared (FIR) luminosities after
508: \citet{ric88}, adjusted for the differences in distances
509: between that publication and this paper, are listed in column 5. The
510: ratios of FIR to $FUV$ luminosity are listed in column 6. The
511: comparatively high L$_{FIR}$/L$_{FUV}$ ratio of NGC~4631 compared to
512: M101, M74 and M51 is expected because of its edge-on aspect. If, when
513: viewed face-on, the L$_{FIR}$/L$_{FUV}$ ratio for NGC~4631 were
514: similar to that of M101, i.e $\sim$4, then we can surmise that the
515: $FUV$ luminosity would be $\sim$2.3$\times$10$^{9}$~L$_{\sun}$ . If
516: the face-on L$_{FIR}$/L$_{FUV}$ ratio were more nearly like that of
517: M74 or M51, then the $FUV$ luminosity would be
518: $\sim$1.1$\times$10$^{9}$~L$_{\sun}$. The exceptional galaxy in this
519: group of five is M83 which shows a comparatively large
520: L$_{FIR}$/L$_{FUV}$ ratio (nearly equal to that of NGC~4631) even
521: though it is viewed face-on. It may be that the dust in M83 provides
522: especially efficient conversion of $FUV$ light into FIR radiation
523: primarily because of its distribution relative to the hot, young
524: stars. The distribution of dust in M101 may also account for the
525: comparatively small L$_{FIR}$/L$_{FUV}$ ratio in that galaxy.
526:
527: %\begin{table}
528: %\dummytable\label{tab:4}
529: %\end{table}
530:
531: \section{Cluster Ages, Masses, and Star Formation Rates}
532:
533: The photometric results can be combined with models of star forming clusters to
534: explore some aspects of the star formation processes in NGC~4631. The cluster
535: models we will use are discussed in detail in
536: \citet{hil94}. Here, we will summarize that discussion. The models
537: are those of \citet{lan92}, which are based on the work of
538: \citet{leq81} who derive spectral energy
539: distributions and masses of evolving star clusters. The models use the
540: Geneva stellar evolutionary tracks \citep{mey94} and the
541: models of \citet{kur91} at log$_{10}$(Z/Z$_{\sun}$)~=~-0.3. In
542: our work, the form of the IMF is $\Psi$(M) = dn(M)/d(lnM)~$\propto$~M$^{-x}$
543: for 1.8~M$_{\sun}$~$<$~M~$<$~120~M$_{\sun}$ and M$^{-0.6}$ for
544: 0.007~M$_{\sun}$~$<$~M~$<$~1.8~M$_{\sun}$. We assume an upper IMF slope of
545: x~=~1.35 conforming to the \citet{sal55} IMF. For our
546: purposes, an instantaneous burst (IB) model of a cluster of co-eval stars is a
547: spectral energy distribution (SED), L$_{IB}$($\lambda,t$), corresponding to a
548: given elapsed time from a burst of star formation. It is the sum of the SEDs of
549: all the stars in the cluster at age $t$ weighted by the assumed IMF and
550: normalized to an astrated mass of 1~M$_{\sun}$. Its units are
551: ergs~s$^{-1}$~\AA$^{-1}$~M$_{\sun}$$^{-1}$. The ionizing photon luminosity is
552: computed by integrating L$_{IB}$($\lambda,t$)/hc over $\lambda$ shortward of
553: 912~\AA. The observed ionizing photon luminosity is given by
554:
555: \begin{equation}
556: N_{Lyc} = 2.206~L_{H\alpha,0}\lambda_{H\alpha}/hc
557: \end{equation}
558: \noindent
559: where L$_{H\alpha,0}$ is the $H\alpha$ luminosity obtained from the
560: observed flux corrected for both extinction and distance. The factor
561: 2.206 is derived under the assumption of case B recombination at
562: 10,000K. The $FUV$ luminosity, L$_{IB}$(1520~\AA,$t$), is defined as
563: L$_{IB}$($\lambda,t$) integrated over the $FUV$ response curve of the
564: UIT. The age of a cluster is simply determined by finding that value
565: of $t$ at which the observed and model ratios of
566: N$_{Lyc}$/L$_{IB}$(1520~\AA,$t$) are equal. The total astrated mass
567: inferred from the IB model is simply the observed luminosity divided
568: by the normalized IB model SED integrated over the response curve for
569: each broad band.
570:
571: If continuous star formation (CSF)has occurred at an essentially
572: constant rate over a period of time, the star formation rate, SFR, can
573: be computed by integrating the IB models over time; i.e.
574:
575: \begin{eqnarray}
576: SFR_{\lambda} = {{L_{obs,\lambda}}\over{\int_{t} L_{IB}(\lambda,t)dt }}~ M_{\sun} yr^{-1}
577: \end{eqnarray}
578: \noindent
579: where L$_{obs,\lambda}$ is the measured spectral luminosity in
580: ergs~s$^{-1}$~\AA$^{-1}$ and L$_{IB}$($\lambda,t$) is the IB model defined
581: above. The integration in time is over the interval of star formation.
582:
583: \subsection{Instantaneous Burst Models}
584:
585: Plots of the observed colors, M$_{152}$-$U$, M$_{152}$-$B$, and
586: M$_{152}$-$V$, (corrected for extinction),
587: vs. log$_{10}$(N$_{Lyc}$/L$_{152}$) are shown in Figure~4, a, b, and
588: c. Numbers adjacent to the data points in Figure~4c identify the
589: aperture corresponding to each datum. Error bars indicate the
590: 1$\sigma$ random photometric errors. The points which have been
591: plotted meet the conditions that the signal to noise ratios in the
592: M$_{152}-V$ colors are greater than 3 and the color changes due to
593: variations in the aperture sizes are less than the 1$\sigma$ errors.
594:
595: The triangle symbols represent the colors computed from IB models
596: where each triangle corresponds to a time in the past when the star
597: burst occurred in integer units of 1~Myr beginning at 2~Myrs on the
598: right and ending at 11~Myrs on the left. The arrow indicates the
599: direction points would move if $E(B-V)$ were to increase by 0.18~mag
600: which is the 1$\sigma$ width of the relevant $E(B-V)$ distribution
601: (black line) shown in Figure~3. All three colors derived from the
602: observations of the selected star forming regions tend to be more
603: positive than those derived from the models. This behavior may be due
604: to our use of an inappropriate reddening curve or possibly to the
605: assumptions regarding the model input parameters. However, in view of
606: the random and systematic errors such distinctions are
607: problematical. For example, increasing the mean value of $E(B-V)$ by
608: about 0.11~mag, which is the adopted uncertainty in this quantity for
609: integrated values as explained below, brings the data in reasonably
610: good agreement with the models. A comparison of the observational
611: results with the IB model indicates that if the star formation occurs
612: in bursts then the bursts occurred from 2.7 to 10~Myrs in the past. It
613: will be noted that the burst times are strongly dependent on
614: log$_{10}$(N$_{Lyc}$/L$_{152}$). On average, the change in aperture
615: size will move the points to the left or right an amount equivalent to
616: 1~Myr. The implication of the uncertainty in the extinction correction
617: is that star formation for some regions may have occurred 2.7~Myr
618: further in the past, or 2.3~Myr toward the present.
619:
620: Individual burst times, determined from the measured
621: log$_{10}$(N$_{Lyc}$/L$_{152}$) values are computed for each
622: aperture. The average variation in burst times attributable to the
623: photometric standard deviation is 0.71~Myr, which is about the same as
624: variation due to changes in the aperture sizes. The total astrated
625: mass is determined for all apertures from photometry in the four wide
626: bands as outlined above. The masses summed over the 48 apertures for
627: each broad band are similar; the largest deviation, 20\%, is
628: associated with the $FUV$ photometry. This result is remarkable in
629: that the mass of the stars producing the $FUV$ light constitutes a
630: small part of the total mass of stars producing light in the optical
631: bands, and it implies that an average of the summed masses in each
632: band is justified. Random errors in the masses summed over all the
633: apertures for each band are on the order of a percent whereas changes
634: in the aperture sizes cause the summed masses to vary by a factor of
635: 1.41. For integrated quantities corrected for extinction internal to
636: NGC~4631 we adopt, somewhat arbitrarily, an uncertainty in $E(B-V)$ of
637: 0.11, which is equal to the difference between the mean values of the
638: $E(B-V)$ distributions derived from the Balmer Decrements and from the
639: stellar continua. This difference results in an uncertainty in the
640: summed masses by of factor of 2.8. When the aperture size variation is
641: combined with the extinction uncertainty the summed burst mass is
642: uncertain by a factor of 3.9.
643:
644: \subsection{Continuous Star Formation Models}
645:
646: \citet{cal97} has suggested that in active star forming regions such
647: as described here, B type stars, which supply large amounts of $FUV$
648: radiation but little $H\alpha$, migrate away from the dust clouds of
649: their origin where star formation is continuing to produce massive O
650: stars. A typical migration time of 50~Myr corresponds to a linear
651: dimension of 500~pc and a stellar velocity of 10~km~s$^{-1}$. Such a
652: picture can explain why the color excesses measured from stellar
653: continuum colors are often less than those measured from Balmer
654: Decrements in the gas surrounding Lyman continuum emitting stars.
655: From our observations we estimate that the smallest, most compact
656: UV-bright regions in NGC~4631 can not have begun forming stars
657: substantially more than 14~Myr ago purely on the basis of the
658: morphological similarity of the H$\alpha$ and FUV images. The
659: crossing time of the largest region is equivalent to a turn-on time of
660: 48~Myr in the past. However, the FUV image must be composed in part
661: of many spatially unresolved clusters, and with the available data we
662: can not disentangle effects due to the diffusion of stars from those
663: due to the distribution of stellar birth sites. The impression of
664: comparative youth is enhanced by the similarity of the mean E(B-V)
665: values determine by the Balmer Decrements and the integrated stellar
666: continua of each star forming region. Nevertheless, there is likely
667: to be a spread in turn-on times among the UV-bright regions. The analysis
668: presented in Section 6 implies that the large eastern shell may have
669: resulted from a star burst which ocurred 20~Myr ago. Somewhat
670: arbitrarily, we assume that the most representative turn-on time is
671: 14~Myr in the past, but that this representative turn-on time could
672: vary by at least $\pm$~5~Myr.
673:
674: Shown in Figure~4 a, b, and c are colors determined from CSF models
675: with star formation beginning 14~Myrs ago and ending at times
676: indicated by circles plotted in 1~Myr intervals from the present
677: (0~Myrs) on the right to 9~Myrs ago on the left. The data points
678: indicate a similar range in turn-off times. Star formation turn-off
679: times and astration rates computed from relation (2) for each passband
680: and the $H\alpha$ line are listed in Table~5, columns 8 through
681: 13. The rate averaged over all four passbands and the $H\alpha$ is
682: listed in column~14. Omitted from this compilation are the rates for
683: region 28, which shows a turn-off time so close to the onset of star
684: formation that the rate is unrealistically high. Stated differently,
685: region 28 must be considered a star burst. The average star SFR
686: summed over all the star formation regions is
687: 0.33~M$_{\sun}$~yr$^{-1}$. If the representative turn-on time was
688: 20~Myr ago, then the total SFR would be about half this rate, or
689: 0.15~M$_{\sun}$~yr$^{-1}$. The average variation of the summed SFR due
690: to the change in the aperture sizes is $\pm$0.17
691: M$_{\sun}$~yr$^{-1}$, whereas error in the extinction correction (0.11~mag)
692: results in an uncertainty in the summed SFR of a factor of 5.5. The
693: random error in the summed average SFR is about 1\%.
694:
695: %\begin{table}
696: %\dummytable\label{tab:5}
697: %\end{table}
698:
699: The diagnostic diagrams derived for the CSF models are very similar to
700: those derived for the IB models, their differences being significantly
701: smaller than the dispersion of the data. The observed colors agree
702: with the model colors within the limits defined by the estimated
703: errors. The slope displayed by the CSF models is flatter and less
704: variable than that of the IB models, particularly at the most recent
705: turn-off times (or burst times). This circumstance arises from the
706: fact that most of the FUV flux is provided by B-type stars with
707: lifetimes longer than 14~Myrs. Therefore, the relative proportion of
708: B and later type stars does not vary significantly over the 14~Myr
709: baseline, thus yielding approximately constant UV colors which vary
710: smoothly with turn- off time.
711:
712:
713: \section{The Eastern Supershell}
714:
715: \citet{mcc87} have developed a
716: quantitative picture of the dynamics of supershells, which can be used
717: to interpret our results and reinterpret those of
718: \citet{ran93}. The picture is one in which the
719: supershell was formed and driven to its present size primarily by
720: supernovae occurring in an extraordinarily large collection of
721: associations and clusters resembling those we observe. Combining
722: equations (3) and (4) of \citet{mcc87} gives an expression
723:
724: \begin{equation}
725: t_{7} = 0.0588(km~s^{-1}~pc^{-1})~(R_{s}/V_{s})
726: \end{equation}
727: \noindent
728: where t$_{7}$, in units of 10~Myr, is the time required for the shell
729: to expand to a radius, R$_{s}$, with a terminal velocity of V$_{s}$.
730: Adopting R$_{s}$ = 1.5~kpc and V$_{s}$~=~45~km~s$^{-1}$
731: \citep{ran93} we find t$_{7}$~=~1.96 or about 20~Myr. During this
732: interval, only those stars with mass equal to or greater than M$_{*}$
733: will become supernovae and contribute to the shell expansion. M$_{*}$
734: is given approximately by the relation
735: M$_{*}$~$\sim$10~M$_{\sun}$~(t$_{7}$/3)$^{-0.625}$
736: \citep{mcc87}, and is 13~M$_{\sun}$ for the assumed
737: conditions. The total number of stars, N$_{*}$, contributing to shell
738: expansion is given by
739:
740: \begin{equation}
741: N_{*}~\times~E_{51} = n_{0}~(R_{s}/82 pc)^{5}~t_{7}^{-3}
742: \end{equation}
743: \noindent
744: where E$_{51}$ is the energy input to the ISM from each supernovae
745: event in units of 10$^{51}$~ergs, here assumed to be unity, and
746: n$_{0}$ is the density of the ambient ISM, which is assumed to be
747: uniform. Substituting the numerical values for R$_{s}$ and t$_{7}$
748: gives N$_{*}$~=~2.7$\times$10$^{5}$~n$_{0}$ stars with masses
749: $\geq$~13~M$_{\sun}$. If n$_{0}$~=~0.2~cm$^{-3}$, consistent with the
750: estimate of \citet{ran93}, then
751: N$_{*}$~=~5.3$\times$10$^{4}$ stars (M $\geq$ 13 M$_{\sun}$). This
752: number is somewhat larger than the estimate of
753: \citet{ran93} (1 - 3.5)$\times$10$^{4}$ stars, and in
754: absolute terms is extraordinarily large. For example, \citet{ken84}
755: has estimated that in 30~Doradus the total mass
756: of OB stars between 10 and 100~M$_{\sun}$ is 50,000~M$_{\sun}$.
757: Assuming a \citet{sal55} IMF, the total number of
758: OB stars between 13~M$_{\sun}$ and 100~M$_{\sun}$ is
759: 1.54$\times$10$^{3}$, a factor of about 34 less than our estimated
760: number of OB stars responsible for the large shell. We find that the
761: light emitted from region 20, which includes the giant \ion{H}{2}
762: region CM~67, is produced by stars with a total mass of
763: 2.7$\times$10$^{6}$~M$_{\sun}$. If our assumed IMF is valid, then
764: this region contains approximately 1.3$\times$10$^{4}$ stars between
765: 13~M$_{\sun}$ and 120~M$_{\sun}$, a number which approaches that
766: inferred for the ancestral stars of the large shell. We infer an even
767: larger number of stars (1.5$\times$10$^{4}$) for region 28.
768:
769: In this picture of shell formation, the UV-bright regions associated
770: with the shell, namely those delineated by apertures 3 through 16,
771: must arise from second generation star formation. As
772: \citet{mcc87} point out, the shell will probably
773: burst at high galactic latitudes before third generation stars are
774: formed. In fact, as \citet{ran93}
775: suggest on the basis of a break in the \ion{H}{1} emission observed on
776: the north side of the shell (their Figure~3), this breakout may have
777: already occurred. These authors also point out that a shell of 3~kpc
778: diameter can be produced initially only if there is significant gas
779: pressure and/or magnetic pressure at high z. These conditions seem
780: plausible when taking into account the obvious tidal effects of the
781: interaction between NGC~4631 and NGC~4656 manifested in the \ion{H}{1}
782: emission maps of \citet{wel78} and \citet{ran93}.
783:
784: If the stars in the vicinity of the supershell were made from a reservoir of
785: gas, most of which was swept up into the shell, then it is possible to make an
786: estimate of a star formation efficiency, here defined as
787: M$_{stellar}$/(M$_{stellar}$ + M$_{gas}$), in the supershell region by
788: referring to the \citet{ran93} estimate of the hydrogen supershell
789: mass. The average total stellar mass in the shell region (apertures 3 through
790: 16) inferred from our observations in the $FUV$, $U$, $B$, and $V$ bands is
791: 1.6$\times$10$^{7}$~M$_{\sun}$ based on the IB model. Comparison with the
792: estimated hydrogen shell mass of (1-2)$\times$10$^{8}$~M$_{\sun}$ implies an
793: efficiency of $\sim$0.047. This efficiency represents an upper limit since the
794: contribution of a molecular gas has not been taken into account.
795:
796:
797: \section{Discussion}
798:
799: Interpreted in terms of an instantaneous burst model, the light in all the
800: numbered apertures and in all four wavelength bands implies a total created
801: mass of 2.5$\times$10$^{7}$~M$_{\sun}$ within a factor of 3.9.
802: This number may be underestimated by a comparatively small amount due
803: to the absorption of light by interstellar grains as discussed below. Our
804: estimate can be compared to the equivalent number of
805: 5.2$\times$10$^{6}$~M$_{\sun}$ in the Magellanic irregular, NGC~4449, which
806: exhibits intense star formation \citep{hil94}. The
807: difference between the two galaxies is even greater than indicated here
808: because NGC~4449 is observed nearly face-on revealing more of its
809: star-forming regions than does NGC~4631 which is observed almost edge-on.
810:
811: The interior disk of NGC~1068 exhibits numerous starburst knots surrounding the
812: nucleus. \citet{bru91} have estimated the total
813: mass of nine of the brightest of these knots, assuming an IB scenario, to be
814: 2.4$\times$10$^{7}$~M$_{\sun}$, a number comparable to that derived for our observed
815: 48 regions in NGC~4631. Thus, while the masses deduced for the
816: star-forming regions of NGC~4631 are extraordinarily large, they are not
817: uniquely so.
818:
819: We also compare the star formation activity of the individual star
820: forming regions with the prototypical giant \ion{H}{2} region,
821: 30~Doradus. Using an aperture of 370~pc (d=50~kpc),
822: \citet{ken84}, obtained L$_{H\alpha}~=~$1.5$\times$10$^{40}$~ergs~s$^{-1}$
823: and an estimated total mass of 50,000~M$_{\sun}$ for all stars with
824: individual masses between 10 and 100~M$_{\sun}$ . Extending this
825: Kennicutt estimate over a larger mass range by using the Salpeter IMF
826: we find a total stellar mass between 0.007 and 120~M$_{\sun}$ of
827: 1.11$\times$10$^{6}$~M$_{\sun}$. If it is assumed that this mass was
828: produced in a burst, we can compare the total mass estimate with the
829: IB model mass estimates for the individual star forming regions listed
830: in columns 3, 4, 5 and 6 of Table~5. Of the 48 aperture delineated
831: \ion{H}{2} regions listed, 6 exhibit derived total masses, averaged
832: between the four wide-band images, in excess of the 30~Doradus value.
833: Taking into account our estimated error, this number can be increased
834: to 26. Of these 6 regions, none exhibit $H\alpha$ luminosities comparable to
835: 30~Doradus. Again, the photometric uncertainty allows this number to be
836: increased to 3. The $H\alpha$ luminosity of region 20 is about two-thirds that
837: of 30~Doradus, but its mass is about a factor of 2.5 greater. Region 20
838: incorporates the bright \ion{H}{2} region CM~67 and is associated with a
839: giant molecular cloud, both of which have been tentatively identified
840: with the tip of a bar \citep{roy91}. Regions 41 and 47
841: show both $H\alpha$ luminosities and masses comparable to 30~Doradus, and we note
842: also that the $H\alpha$ luminosity of region 41 is comparable to that of the
843: supergiant \ion{H}{2} region in M101 (NGC~5461), which has an $H\alpha$ luminosity
844: uncorrected for extinction of 2.7$\times$10$^{40}$~ergs~s$^{-1}$ \citep{wal90}.
845: However, with some significant exceptions such as those
846: highlighted above, most of the NGC~4631 star forming regions are less
847: massive and older than 30~Doradus. Even when the upper limit on $E(B-V)$ is
848: used to correct for internal extinction only 7 regions exhibit $H\alpha$
849: luminosities about equal to or exceeding that of 30~Doradus.
850:
851: A ``short term'' star formation rate (SFR) averaged over the range in
852: burst times exhibited by the data, i.e. 7.3~Myrs, can be computed using
853: the created masses summed over all 48 apertures and averaged over the
854: four broad bands. This rate is $\sim$3.5 M$_{\sun}$~yr$^{-1}$ which is
855: substantial. As discussed below, it is about equal to the global SFR determined
856: from FIR measurements, and a factor of 7 less than the
857: total SFR measured by \citet{dah95}.
858:
859: In summary, when the $FUV$ and optical imagery is interpreted in terms of the
860: instantaneous burst paradigm it is obvious that NGC~4631 incorporates
861: exceptionally large star formation activity.
862:
863: Interpretation of our results in terms of CSF models points to a
864: galaxy of a more ordinary variety. The total average star formation
865: rate determined from the $FUV$ data summed over the 48 apertures for
866: NGC~4631 is 0.33~M$_{\sun}$~yr$^{-1}$, which is a factor of 2.7 less
867: than than the equivalent NGC~4449 rate (0.88~M$_{\sun}$~yr$^{-1}$).
868: However, our assumed uncertainty in the mean value of $E(B-V)$ allows
869: our derived SFR to exceed the NGC~4449 rate by a factor of 2. Thus,
870: that part of NGC~4631 which is observable and has incorporated star
871: formation within the past 14~Myr creates stars at a rate comparable to
872: the total rate of the face-on starburst galaxy, NGC~4449. Assuming
873: that the turn-on time was 20~Myr ago, which
874: results in an SFR equal to 0.15~M$_{\sun}$~yr$^{-1}$, does not change
875: this conclusion. The NGC~4631 results can also be compared to the
876: high rate of 2.4~M$_{\sun}$~yr$^{-1}$ found by
877: \citet{bru91} when interpreting in terms of a CSF scenario
878: their observations of the star forming regions found in the disk of
879: NGC~1068. Assuming constant star formation, our observations indicate
880: a total mass created in 14~Myr of about 4.6$\times$10$^{6}$
881: M$_{\sun}$, which is less than the mass inferred from the IB model
882: (5.1$\times$10$^{7}$ M$_{\sun}$) by a factor of 5.5. This factor
883: increases to 8.4 if the turn-on time was 20~Myr ago.
884:
885: \citet{dah95} have related radio continuum
886: observations of the halos of several edge-on galaxies, including NGC~4631, to
887: the rates of supernovae occurring within the galactic disks. The
888: inferred supernova rate for NGC~4631 is 0.23~yr$^{-1}$. Assuming
889: equilibrium between the birth rate and death rates of supernovae producing
890: stars and a \citet{sal55} IMF over the mass range
891: between 0.007 and 120~M$_{\sun}$ we find that this result implies a star
892: formation rate of 24~M$_{\sun}$~yr$^{-1}$.
893:
894: A significantly smaller rate is derived from the FIR luminosity.
895: For stars with masses greater than 10~M$_{\sun}$
896:
897: \begin{equation}
898: SFR_{FIR} (M_{\sun}~yr^{-1}) = 1.4 \times 10^{-10} L_{FIR} (L_{\sun})
899: \end{equation}
900: \noindent
901: \citep{sau92}. In the case of NGC~4631,
902: SFR$_{FIR}$(M$_{\sun}$ $>$ 10) = 1.3~M$_{\sun}$~yr$^{-1}$. Employing
903: the Salpeter IMF described in Section~5, the SFR for stellar masses
904: between 0.007~M$_{\sun}$ and 120~M$_{\sun}$ is
905: 8.1~M$_{\sun}$~yr$^{-1}$ or 4.5~M$_{\sun}$~yr$^{-1}$ if 45\% of the
906: FIR luminosity is from the diffuse interstellar medium.
907:
908: The total continuous SFR, summed over the 48 apertures in our data
909: set, 0.33~M$_{\sun}$~yr$^{-1}$, is far smaller than the
910: rate inferred from Dahlem's radio observations. It, of course,
911: refers only to the star forming activity delineated by the 48
912: apertures. If it is assumed that the recorded $FUV$ emission
913: from the whole galaxy arises from star formation, then multiplication of the
914: 0.33~M$_{\sun}$~yr$^{-1}$ rate by the ratio of the observed $FUV$
915: flux in the large aperture to the observed flux summed over all 48
916: small apertures increases the continuous SFR to
917: 2.2~M$_{\sun}$~yr$^{-1}$, still a very small SFR compared to the Dahlem value
918: but one that approaches the FIR rate.
919:
920: It is possible that a large portion of the newly created stars are so
921: heavily shielded by dust that there is no indication of their
922: existence in any of our wavelength bands. For example,
923: \citet{bru91} estimate that the emerging UV
924: light from $FUV$ bright knots in the disk of NGC~1068 implies an SFR
925: which is only 7\% of the true value. This estimation was based on the
926: relative FIR and UV knot luminosities and the assumption that most of
927: the FIR emission is reprocessed UV radiation from hot stars absorbed
928: and reradiated by interstellar grains. Unfortunately, no far infrared
929: observations of NGC~4631 exist at a spatial scale which would allow
930: the isolation of FIR emission from the galactic background for any of
931: the star formation regions. An alternative procedure for approximating
932: the relative contribution of reprocessed UV radiation in NGC~4631
933: star-forming regions is to measure the ratio of integrated $FUV$ and
934: FIR luminosities of face-on galaxies. The galaxy M74 (NGC~628) is
935: ideal for this purpose since the required luminosities are known and
936: $E(B-V) = $0.33 over much of the galaxy
937: \citep{cor94}, a value which is similar to the mean value of the color excesses
938: measured for the NGC~4631 star-forming regions. The ratio of the total
939: integrated FIR luminosity from IRAS measurements
940: \citep{ric88} to the integrated $FUV$ luminosity from
941: UIT measurements \citep{wal97b} is 8.5
942: (see Table~4). Similar ratios for other face-on galaxies are also
943: listed in Table~4.
944:
945: It is likely that the dust emitting the $IR$ radiation is heated in part by
946: the interstellar radiation field, with an approximate contribution to the
947: total $IR$ heating budget of 45\% \citep{sau92};
948: though see contrary arguments of \citet{dev92}
949: and \citet{dev94} who claim that almost all of
950: the $IR$ radiation is reprocessed $FUV$ emission. Assuming that
951: such processes are occurring in NGC~4631, and adopting the M74
952: $FIR$/$FUV$ luminosity ratio of 8.5 (the average $FIR$/$FUV$ luminosity
953: ratio for the face-on spiral galaxies M83, M101, M74, and M51 is 8.8)
954: and an interstellar to total $IR$ radiation energy density
955: ratio of 0.45, we find that the SFR determined from the
956: {\it extinction corrected FUV} luminosity will be increased by a factor
957: of only 1.21 when taking into account
958: the stellar $FUV$ light absorbed by the interstellar grains.
959: Thus, the total SFR for the 48
960: star forming regions alone becomes 0.40~M$_{\sun}$~yr$^{-1}$ with a large
961: uncertainty. If the
962: integrated $FUV$ emission within the large aperture is assumed to arise
963: entirely from star formation then the total estimated SFR {\it in that part
964: of NGC~4631 which we are able to observe} becomes 2.7~M$_{\sun}$ yr$^{-1}$.
965:
966: The correction is not sensitive to assumptions about the IR
967: heating budget. For example, if all the observed FIR light was
968: reprocessed $FUV$ light from star formation instead of only 55\% as
969: we have assumed then the correction factor would increase by only
970: 16\%. On the basis of observations of M83 and M101 the uncertainty
971: in the value of R, i.e. the applicable value of the ratio of
972: observed FIR to $FUV$ light, induces an uncertainty in the
973: correction factor of 12\%. However, the correction is still quite
974: uncertain since FIR observations of NGC~4631 with spatial
975: resolution comparable to those of the UIT have not been made.
976:
977: If all the FIR light is reprocessed $FUV$ light, then the SFR inferred
978: from the $FUV$ light in the large aperture becomes
979: 3.1~M$_{\sun}$~yr$^{-1}$, again with large uncertainties, which should
980: be compared to the SFR computed from the observed FIR emission via
981: relation (5) of 8.1~M$_{\sun}$~yr$^{-1}$. The result that the SFR
982: summed over the 48 apertures is from factors of 11 to 25 less than the
983: SFR determined from the FIR observations is not surprising considering
984: the edge-on aspect of NGC~4631. Although the amount of created stellar
985: mass differs by a factor of 5.5 there is no reason for distinguishing
986: between the applicability of the IB and CSF models on the basis of the
987: diagnostic diagrams in Figure~4. The fact that the large aperture
988: $FUV$ photometry leads to an SFR about equal to that derived from the
989: FIR within a factor of 2 is surprising, and may indicate that most of
990: the FUV light arises from short term star bursting activity, i.e. that
991: the exclusive application of the CSF model is not appropriate.
992:
993: By contrast, the SFR inferred from the radio halo
994: \citep{dah95} is a factor of at least 5 larger than that
995: associated with either the large aperture FUV photometry or the FIR
996: results (assuming only 55\% of the observed FIR is reprocessed FUV
997: light). We speculate that the SFR derived from the radio observations
998: is too large because the assumption of equilibrium between the birth
999: and death rates of the supernovae producing stars is invalid, and that
1000: the radio halo is due to a starburst and not continuous star
1001: formation.
1002:
1003: Finally, to the extent that the IB paradigm is valid, our
1004: conclusions regarding the formation of the supershell in NGC~4631 indicates
1005: that strong starburst activity has occurred in that part of NGC~4631
1006: approximately 20~Myrs ago.
1007:
1008: \section{Summary}
1009:
1010: A far ultraviolet image of the galaxy NGC~4631 centered on 152.1~nm was
1011: recorded by the Ultraviolet Imaging Telescope. The $FUV$ image, combined with
1012: similar images in the $U, B, V,$ and $R$ bands and in $H\alpha$ and $H\beta$,
1013: is used to study star formation in this late type (Sc/SBd) galaxy, viewed
1014: nearly edge-on, with the following results.
1015:
1016: 1. For a disk galaxy seen edge-on, NGC~4631 is extraordinarily bright at $FUV$
1017: wavelengths. The attenuation of light from 48 star forming regions located
1018: uniformly along the whole length of the visible galactic disk is low with the
1019: mean value of $E(B-V) = $0.28. Dust is seen silhouetted against the cool star
1020: population, and the star forming regions appearing in $H\alpha$ emission are
1021: seen silhouetted against the dust. These facts together with our
1022: interpretation of previously measured \ion{H}{1} and \ion{H}{2}
1023: position-velocity curves indicate that the $FUV$ emission arises
1024: near the edge of the observable galaxy, as is observed in the
1025: face-on Sc/SBc galaxies M101 and M83.
1026:
1027: The overall $FUV$ morphology of this edge-on disk galaxy is remarkably similar
1028: to that of the so-called chain galaxies that have been recently observed at
1029: high redshift, thus implying a similar interpretation for at least some of
1030: those distant $FUV$ sources (see \citet{smi97}).
1031:
1032: 2. Photometric data is presented in the form of diagnostic
1033: diagrams showing the M$_{152}-U$, M$_{152}-B$ and M$_{152}-V$ colors plotted
1034: against log$_{10}$(N$_{Lyc}$/L$_{152}$). The data exhibit considerable dispersion
1035: due to errors made in the determination of the relevant
1036: background and extinction corrections. Both instantaneous burst
1037: (IB) models with burst ages ranging from 2.7 to 10~Myr and
1038: continuous star formation (CSF) models with star formation
1039: beginning 14~Myr ago and terminating from 0 to 9~Myr ago fit the
1040: data equally well. The uncertainties permit the burst ages and
1041: turn-off times to be extended 1 to 3 Myrs to earlier times.
1042:
1043: 3. The ratio of the mean value of $E(B-V)$ determined from the
1044: continua of stars to that found from measurements of Balmer
1045: Decrements is 0.7, and is thus greater than the value of 0.44
1046: quoted by \citet{cal97} as being typical of regions exhibiting
1047: active star formation. This result is compatible with our
1048: conclusion on the basis of $FUV$ emission morphology that the
1049: observed star formation is young, i.e. that the stars have not
1050: had much time to migrate from their birth site.
1051:
1052: 4. If the observational results are interpreted with the IB models,
1053: the inferred stellar masses created in each star forming region are
1054: substantial. As determined by our analysis, the inferred stellar
1055: masses of 6 regions exceed the mass of 30~Doradus. Errors permit this
1056: number to increase or decrease by up to 20. The region of the second
1057: largest derived mass, 2.7$\times$10$^{6}$~M$_{\sun}$ , incorporates
1058: the bright \ion{H}{2} region CM~67 and is associated with a giant
1059: molecular cloud, both of which have been tentatively identified with
1060: the tip of a bar \citep{roy91}. Of these 6
1061: regions, none exhibit $H\alpha$ luminosities equal to that of
1062: 30~Doradus. Photometric uncertainties permit this number to increase
1063: to 3. The $H\alpha$ luminosity of region 41 is comparable to that of
1064: the supergiant H II region in M101 (NGC 5461). The total created
1065: stellar mass deduced from the light emitted from the star forming
1066: regions is 2.6$\times$10$^{7}$~M$_{\sun}$ to within a factor of 2.3,
1067: which is about equal to the stellar mass
1068: (2.4$\times$10$^{7}$~M$_{\sun}$) of the UV emitting knots in the disk
1069: of NGC~1068 \citep{bru91} and is a factor of 5 greater than the stellar
1070: mass (5.2$\times$10$^{6}$~M$_{\sun}$) associated with the $FUV$
1071: emitting knots in NGC~4449 \citep{hil94}.
1072: Using the range in burst times of 7.3 Myrs, we derive an average
1073: ``short term'' star formation rate of 3.5~M$_{\sun}$~yr$^{-1}$.
1074:
1075: 5. Based on previously obtained data, application of the shell
1076: generation theory of \citet{mcc87} shows
1077: that the number of supernova producing stars which could generate the
1078: 3~kpc diameter ring of $FUV$ and \ion{H}{1} emission near the east end
1079: of the galactic image is 5.3$\times$10$^{4}$, which is somewhat
1080: greater than the range (1 to 3.5)$\times$10$^{4}$ estimated by
1081: \citet{ran93} based on the dynamics of a
1082: hydrogen supershell surrounding the stellar ring. While this number is
1083: extraordinarily large, it is comparable to the same number,
1084: 1.3$\times$10$^{4}$, associated with the intense star forming region
1085: delineated by aperture 20, a region that includes the giant \ion{H}{2}
1086: region CM67. Those FUV bright regions on and within the periphery of
1087: the large shell (apertures 3 through 16) must represent second
1088: generation star formation, and this conclusion together with the shell
1089: analysis implies that strong star formation occurred approximately
1090: 20~Myrs ago in that part of NGC~4631 incorporating the large shell.
1091:
1092: 6. Analysis of the $FUV$, $U$, $B$, and $V$ photometry in the context
1093: of the CSF paradigm yields a total SFR for the 48 individual star
1094: forming regions equal to 0.33~M$_{\sun}~$~yr$^{-1}$, uncertain by a
1095: factor of 8.2. Approximate accounting for UV light which is absorbed
1096: by interstellar grains leads to a corrected rate of
1097: 0.40~M$_{\sun}~$~yr$^{-1}$. Under the assumption that 55\% of the
1098: observed FIR is from reprocessed $FUV$ light this corrected rate is
1099: about a factor of 11 less than the SFR inferred from the measured
1100: total $FIR$ (4.5~M$_{\sun}$~yr$^{-1}$), which is not surprising
1101: considering the edge-on aspect of NGC~4631. If the observed $FUV$
1102: light integrated over that part of the galaxy we are able to observe
1103: is assumed to arise in constant star formation, then the SFR equals
1104: 2.6~M$_{\sun}$~yr$^{-1}$ when approximate accounting for the
1105: reprocessing of $FUV$ light by interstellar grains is taken into
1106: account. This is a surprising result in that one would expect the SFR
1107: inferred from the integrated $FUV$ data to be significantly less than
1108: the global SFR determined from the FIR observations, and the
1109: photometric uncertainties exacerbate the problem. We interpret this
1110: paradox to indicate that the $FUV$ emission should not be seen
1111: exclusively in terms of the CSF model, but that much of the observed
1112: global $FUV$ light arises in star bursts. This interpretation is
1113: consistent with the inference of star bursting activity near the
1114: galactic center drawn by \citet{gol94b}
1115: and in an inner disc (r$\approx$50$\arcsec$) by \citet{dur82}.
1116:
1117: \acknowledgements
1118:
1119: We thank the many people who made the ASTRO 2 mission a success. We would also
1120: like to thank Robert Hill for his advice on many aspects of image data
1121: reduction and Richard Rand for providing us with a calibrated $H\alpha$ image
1122: of NGC~4631. Funding for the UIT project was through the Spacelab Office at
1123: NASA Headquarters under project 440-51. This research has made use of the
1124: digitized version of the National Geographic Palomar Sky Survey produced at the
1125: Space Telescope Science Institute under U.S. Government Grant NAGW-2166. The
1126: Oschin Schmidt Telescope is operated by the California Institute of Technology
1127: and Palomar Observatory. W.W.W. acknowledges partial support from NASA through
1128: the Astrophysical Data program (071-969dp).
1129:
1130:
1131: \begin{thebibliography}{}
1132: \bibitem[Aaronson(1978)]{aar78} Aaronson, M. 1978, \pasp, 90, 28
1133: \bibitem[Bohlin et al.(1983)]{boh83} Bohlin, R. C., Cornett, R. H.,
1134: Hill, J. K., Smith, A. M., \& Stecher, T. P. 1983, \apjl, 274, L53
1135: \bibitem[Bohlin et al.(1978)]{boh78} Bohlin, R. C., Savage, B. D.,
1136: \& Drake, J. F. 1978, \apj, 224, 132.
1137: \bibitem[Bothun \& Thompson(1988)]{bot88} Bothun, G. D., Thompson, I. B. 1988, \aj, 96, 877
1138: \bibitem[Braine et al.(1995)]{bra95} Braine, J., Krugel, E., Sievers, A.,
1139: \& Wielebinski, R. 1995, \aap, 295, L55
1140: \bibitem[Bruhweiler et al.(1991)]{bru91} Bruhweiler, F. C., Truong, K. Q.,
1141: \& Altner, B. 1991, \apj, 379,596
1142: \bibitem[Burstein \& Heiles(1984)]{bur84} Burstein, D., \& Heiles, C. 1984,
1143: \apjs, 54, 33
1144: \bibitem[Calzetti(1997)]{cal97} Calzetti, D. 1997, in AIP Conf. Proc. 408,
1145: The Ultraviolet Universe at Low and High Redshift:
1146: Probing the Progress of Galaxy Evolution,
1147: ed. W. H. Waller, M. N. Fanelli, J. E. Hollis, \& A. C. Danks
1148: (Woodbury: AIP),403
1149: \bibitem[Calzetti et al.(1994)]{cal94} Calzetti, D., Kinney, A. L.,
1150: \& Storchi-Bergmann, T. 1994, \apj, 429, 582
1151: \bibitem[Code \& Welch(1982)]{cod82} Code, A. D. \& Welch, G. A. 1982,
1152: \apj, 256,1
1153: \bibitem[Combes(1978)]{com78} Combes, F. 1978, \aap, 65,47
1154: \bibitem[Cornett et al.(1994)]{cor94} Cornett, R. H., O'Connell, R.W.,
1155: Greason, M. R., Offenberg, J. D.,
1156: Angione, R. J., Bohlin, R. C., Cheng, K. P., Roberts, M. S.,
1157: Smith, A. M., Smith, E. P., Talbert, F. D., \& Stecher, T. P. 1994,
1158: \apj, 426, 553
1159: \bibitem[Crillon \& Monnet(1969)]{cri69} Crillon, R. \& Monnet, G. 1969,
1160: \aap, 2, 1
1161: \bibitem[Cowie et al.(1995)]{cow95} Cowie, L. L., Hu, E. M., \& Songaila, A.
1162: 1995, \aj, 110, 1576
1163: \bibitem[Dahlem et al.(1995)]{dah95} Dahlem, M., Lisenfeld, U., \& Golla, G.
1164: 1995, \apj, 444, 119
1165: \bibitem[Devereux \& Young(1992)]{dev92} Devereux, N. A., Young, J. S. 1992 \aj, 103, 784
1166: \bibitem[Devereux \& Scown(1994)]{dev94} Devereux, N. A., Scowen, P. A. 1994 \aj, 108, 1244
1167: \bibitem[de Vaucouleurs et al.(1991)]{dev91} de Vaucouleurs, G., de Vaucouleurs, A., Corwin, H.G.,
1168: Buta, R.J., Paturel, G., \& Fouque, P. 1991, Third Reference Catalogue of Bright Galaxies (RC3)
1169: (New York: Springer-Verlag).
1170: \bibitem[Duric et al.(1982)]{dur82} Duric, N., Crane, P. C., \&
1171: Seaquist, E. R. 1982, \aj, 87, 1671
1172: \bibitem[Fanelli et al.(1988)]{fan88} Fanelli, M. N., O'Connell, R. W., \&
1173: Thuan, T. X. 1988, \apj, 334, 665
1174: \bibitem[Feldmeier \& Ciardullo(1997)]{fel1997} Feldmeier, J. J., \&
1175: Ciardullo, R. 1997, \apj, 479, 231
1176: \bibitem[Fitzpatrick(1985)]{fit85} Fitzpatrick, E. L. 1985, \apj, 299, 219
1177: \bibitem[Golla \& Hummel(1994)]{gol94a} Golla, G. \& Hummel, E.
1178: 1994, \aap, 284, 777
1179: \bibitem[Golla \& Wielebinski(1994)]{gol94b} Golla, G. \& Wielebinski, R.
1180: 1994, \aap, 286, 733
1181: \bibitem[Greason et al.(1994)]{gre94} Greason, M. R., Offenberg, J. D.,
1182: Cornett, R. H., Hill, R. S., \& Stecher, T. P. 1994, \pasp, 106, 1151
1183: \bibitem[Hill et al.(1994)]{hil94} Hill R.S., Home A. T., Smith A. M.,
1184: Bruhweiler, F. C., Cheng, K. P., Hintzen, P. M. N., \& Oliversen, R. J.
1185: 1994, \apj, 430, 568
1186: \bibitem[Home(1997)]{hom97} Home, A. T. 1997, Ph.D. Thesis,
1187: Catholic University of America.
1188: \bibitem[Kennicutt(1984)]{ken84} Kennicutt, R. C. 1984, \apj, 287, 116
1189: \bibitem[Klemola(1962)]{kle62} Klemola, A. 1962, \aj, 67, 740
1190: \bibitem[Kurucz(1991)]{kur91} Kurucz, R. L. 1991, in The Stellar Populations of
1191: Galaxies, ed. B. Barbuy \& A. Renzini (Dordrecht: Kluwer), 225
1192: \bibitem[Landsman(1992)]{lan92} Landsman, W. 1992, private communication
1193: \bibitem[Lasker et al.(1990)]{las90} Lasker, B. M., Sturch, C. R.,
1194: McLean, B. J., Russell, J. L., \& Jenker, H. 1990, \aj, 99, 2019
1195: \bibitem[Lequeux et al.(1981)]{leq81} Lequeux, J., Maucherat-Joubert, M.,
1196: Deharveng, J. M., \& Kunth, D. 1981, \aap, 103, 305
1197: \bibitem[McCray \& Kafatos(1987)]{mcc87} McCray, R. \& Kafatos, M.
1198: 1987, \apj, 317, 190
1199: \bibitem[Meynet et al.(1994)]{mey94} Meynet, G., Maeder, A., Schaller, G., Schaerer, D.
1200: \& Charbonnel, C. 1994, \aap, 103, 97
1201: \bibitem[Pierce(1994)]{pie94} Pierce, M. J. 1994, \apj, 430, 53
1202: \bibitem[Rand et al.(1992)]{ran92} Rand, R. J., Kulkarni, S. R., \&
1203: Hester, S. J. 1992, \apj, 396,97
1204: \bibitem[Rand \& van der Hulst(1993)]{ran93} Rand, R. J.,
1205: \& van der Hulst, J. M. 1993, \aj, 105, 2098
1206: \bibitem[Rand's(1994)]{ran94} Rand, R. J. 1994, \aap, 285, 833
1207: \bibitem[Rand \& Stone(1996)]{ran96} Rand, R. J., \& Stone, J. M.
1208: 1996, \aj, 111, 190
1209: \bibitem[Rice et al.(1988)]{ric88} Rice, W., Lonsdale, C. J., Soifer, B. T.,
1210: Neugebauer, G., Kopan, E. L., Lloyd, L. A., de Long, T., \& Habing, H. J.
1211: 1988, \apjs, 68, 91
1212: \bibitem[Roy et al.(1991)]{roy91} Roy, J. R., Wang, J., \& Arsenault, R.
1213: 1991, \aj, 101, 825
1214: \bibitem[Salpeter(1955)]{sal55} Salpeter, E.E. 1955, \apj, 121, 161
1215: \bibitem[Sauvage \& Thuan(1992)]{sau92} Sauvage, M. \& Thuan, T. X.
1216: 1992, \apjl, 396, L73
1217: \bibitem[Seaton(1979)]{sea79} Seaton, M. 1979, \mnras, 187, 73P
1218: \bibitem[Sharina et al.(1996)]{sha96} Sharina, M. E., Karachetsev, D., \&
1219: Tikhonov, N. A. 1996, \aap, 119, 499
1220: \bibitem[Smith et al.(1997)]{smi97} Smith, A. M., Collins, N. R.,
1221: Waller, W. H., Fanelli, M. N., \& Stecher, T. P. 1997,
1222: in AIP Conf. Proc. 408, The Ultraviolet Universe at Low and
1223: High Redshift: Probing the Progress of Galaxy Evolution,
1224: ed. W. H. Waller, M. N. Fanelli, J. E. Hollis, \& A. C. Danks
1225: (Woodbury: AIP), 439
1226: \bibitem[Sofue et al.(1990)]{sof90} Sofue, Y., Handa, T., Golla, G., \&
1227: Wielebinski, R. 1990, \pasj, 42, 745
1228: \bibitem[Sofue et al.(1989)]{sof89} Sofue, Y., Handa, T., \& Nakai, N.
1229: 1989, \pasj, 41, 937
1230: \bibitem[Stecher et al.(1992)]{ste92} Stecher, T. P. {\it et al.}
1231: 1992, \apjl, 395, L1
1232: \bibitem[Stecher et al.(1997)]{ste97} Stecher, T. P. {\it et al.}
1233: 1997, \pasp, 109, 584
1234: \bibitem[Stone(1977)]{sto77} Stone, R. 1997, \apj, 218, 767
1235: \bibitem[Waller(1990)]{wal90} Waller, W. H., 1990, Ph.D. Thesis,
1236: University of Massachusetts
1237: \bibitem[Waller et al.(1997a)]{wal97a} Waller, W. H, Bohlin, R. C., Cornett, R. H.,
1238: Fanelli, M. N., Freedman, W. L., Hill, J. K., Madore, B. F., Neff, S. G.,
1239: Offenberg, J.D., O'Connell, R.W., Roberts, M.S., Smith
1240: A. M., \& Stecher, T. P. 1997a, \apj, 481, 169
1241: \bibitem[Waller et al.(1997b)]{wal97b} ----------, Fanelli, M. N., Collins, N. R., Cornett, R. H.,
1242: Offenberg, J. D., Marcum, P. M., Stecher, T. P., \& the UIT
1243: Science Team 1997b, in AIP Conf. Proc. 408, The Ultraviolet Universe at Low and
1244: High Redshift: Probing the Progress of Galaxy Evolution,
1245: ed. W. H. Waller, M. N. Fanelli, J. E. Hollis, \& A. C. Danks
1246: (Woodbury: AIP), 39
1247: \bibitem[Wang et al.(1995)]{wan95} Wang, Q. D., Walterbos, R. A. M., Steakley, M. F., and
1248: Norman, C. A. 1995, \apj, 439, 176
1249: \bibitem[Weliachew et al.(1978)]{wel78} Weliachew, L., Sancisi, R., and
1250: Guelin, M. 1978, \aap, 65, 37
1251: \bibitem[Zombeck(1990)]{zom90}Zombeck, M. V. 1990, Handbook of Astronomy and
1252: Astrophysics, Second Edition (Cambridge, UK: Cambridge University Press).
1253: \end{thebibliography}
1254:
1255: \clearpage
1256:
1257: \begin{figure}
1258: \figurenum{1}
1259: \begin{minipage}{6.25in}
1260: \plotone{f1.eps}
1261: %\plotone{f1.orig.eps}
1262: \caption{Far-ultraviolet (FUV), $H\alpha$,
1263: $U$, $B$, and $V$, images of NGC~4631. North is up, east to the left.
1264: Portions of the dwarf elliptical galaxy, NGC~4627, which is part of
1265: the interacting system composed of NGC~4631, NGC~4656 and NGC~4627,
1266: can be seen on the north side of the $U$, $B$, and $V$
1267: images.}
1268: \label{fig1}
1269: \end{minipage}
1270: \end{figure}
1271:
1272: \clearpage
1273:
1274: \begin{figure}
1275: \figurenum{2}
1276: \begin{minipage}{6.25in}
1277: \plotone{f2.eps}
1278: %\plotone{f2.orig.eps}
1279: \caption{$FUV$ image of NGC~4631 as in Figure~1 with aperture
1280: boundaries indicated and numbered. The apertures are numbered, 1 to
1281: 48, from left to right, but because of crowding only the units number
1282: is shown except for apertures numbered 10, 20, 30 and 40. The large
1283: aperture boundary encompassing the whole image is used to compute
1284: global properties.}
1285: \label{fig2}
1286: \end{minipage}
1287: \end{figure}
1288:
1289: \clearpage
1290:
1291: \begin{figure}
1292: \figurenum{3}
1293: \begin{minipage}{6.25in}
1294: \plotone{f3.eps}
1295: \figcaption{Histograms showing distribution of $E(B-V)$ values of
1296: each star forming region. {\it Red line}: distribution based on the
1297: measured Balmer Decrements. {\it Black line}: distribution based on the
1298: difference between modeled colors and observed colors. Single
1299: Gaussian fits yield mean values of $E(B-V)$ equal to 0.39~mag and
1300: 0.28~mag respectively, indicated by vertical dashed lines. The 2$\sigma$
1301: width of the Gaussian function fitted to the distribution represented
1302: by the black line is 0.36~mag.}
1303: \label{fig3}
1304: \end{minipage}
1305: \end{figure}
1306:
1307: \clearpage
1308:
1309: \begin{figure}
1310: \figurenum{4a}
1311: \begin{minipage}{6.25in}
1312: \plotone{f4a.eps}
1313: \caption{Color vs. log$_{10}$(N$_{Lyc}$/L$_{152}$) diagrams.
1314: Colors M$_{152}$-$U$, M$_{152}$-$B$ and M$_{152}$-$V$ are plotted in panels a,
1315: b and c respectively. Error bars indicate 1$\sigma$ errors. Numbers in
1316: panel c identify associated apertures. Filled circles: observational results
1317: corrected for dust attenuation using the formalism of
1318: \protect\citet{cal94}.
1319: Open triangles: values derived from instantaneous burst models
1320: computed at integer burst times beginning at 2 Myr on right and
1321: increasing to 11 Myr on left. Open circles: values derived from
1322: continuous star formation models computed at integer turn-off times
1323: beginning at 0 Myr on right and increasing to 9 Myr on left. The arrow
1324: indicates the effect of increasing the correction for extinction on
1325: the observed colors arising from an increase in $E(B-V)$ of 0.18 mag.}
1326: \label{fig4a}
1327: \end{minipage}
1328: \end{figure}
1329:
1330: \clearpage
1331:
1332: \begin{figure}
1333: \figurenum{4b}
1334: \begin{minipage}{6.25in}
1335: \plotone{f4b.eps}
1336: \caption{M$_{152}$-$B$ vs. log$_{10}$(N$_{Lyc}$/L$_{152}$) diagram.}
1337: \label{fig4b}
1338: \end{minipage}
1339: \end{figure}
1340:
1341: \clearpage
1342:
1343: \begin{figure}
1344: \figurenum{4c}
1345: \begin{minipage}{6.25in}
1346: \plotone{f4c.eps}
1347: \caption{M$_{152}$-$V$ vs. log$_{10}$(N$_{Lyc}$/L$_{152}$) diagram.}
1348: \label{fig4c}
1349: \end{minipage}
1350: \end{figure}
1351:
1352: \clearpage
1353:
1354: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1355:
1356:
1357: \begin{deluxetable}{lrrcr}
1358: \tablecolumns{5}
1359: %\tabletypesize{\scriptsize}
1360: \tablewidth{0pc}
1361: %
1362: %
1363: \tablenum{1}
1364: \tablecaption{Observations\label{tab:1}}
1365: %
1366: \tablehead{
1367: %
1368: \colhead{Instrument} &
1369: \colhead{Date} &
1370: \colhead{Exposure} &
1371: \colhead{Bandpass} &
1372: \colhead{Resolution}\\
1373: \colhead{} &
1374: \colhead{} &
1375: \colhead{(seconds)} &
1376: \colhead{$\lambda_{0} / \Delta\lambda$} &
1377: \colhead{(FWHM)}}
1378:
1379: \startdata
1380:
1381: UIT & 03/10/1995 & 1140 & FUV, 152.1/35.4 nm & 3.4\arcsec \\
1382: & & & & \\
1383: Kitt Peak 0.9m & 05/16/1996 & 600 & U & 3.0\arcsec \\
1384: & 05/16/1996 & 300 & B & 1.6\arcsec \\
1385: & 05/16/1996 & 300 & V & 1.1\arcsec \\
1386: & 05/16/1996 & 180 & R & 1.6\arcsec \\
1387: & & & & \\
1388: Palomar\tablenotemark{a} 1.5m & 05/1990 & 2000 & H$\alpha$, 658.5/1.5 nm & 1.2\arcsec \\
1389: & & & & \\
1390: Mt. Laguna & 05/23/1996 & 600 & H$\beta$, 487.8/5.0 nm & 4.5\arcsec \\
1391:
1392:
1393: \tablenotetext{a}{The H$\alpha$ data was provided by Dr. R. J. Rand. See Rand et al. (1992).}
1394:
1395:
1396: \enddata
1397:
1398: \end{deluxetable}
1399:
1400: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1401:
1402: \begin{deluxetable}{lccrrr}
1403: \tablecolumns{6}
1404: \tablewidth{0pc}
1405: \small
1406: %
1407: %
1408: \tablenum{4}
1409: \tablecaption{FUV and IR Luminosities of Galaxies\label{tab:4}}
1410: %
1411: \tablehead{
1412: %
1413: \colhead{Name} &
1414: \colhead{AB\tablenotemark{a}} &
1415: \colhead{Distance\tablenotemark{b}} &
1416: \colhead{L$_{FUV}$\tablenotemark{c}} &
1417: \colhead{L$_{IR}$\tablenotemark{d}} &
1418: \colhead{L$_{IR}$/L$_{FUV}$} \\
1419: \colhead{} &
1420: \colhead{(Mag)} &
1421: \colhead{(Mpc)} &
1422: \colhead{L$_{\sun}$} &
1423: \colhead{L$_{\sun}$} &
1424: \colhead{}}
1425:
1426: \startdata
1427:
1428: NGC~4631 & 0.02 & 7.5 & 6.99$\times$10$^{8}$ & 9.34$\times$10$^{9}$ & 13.40 \\
1429: NGC~5236 (M83) & 0.14 & 4.8 & 8.68$\times$10$^{8}$ & 1.20$\times$10$^{10}$ & 13.80 \\
1430: NGC~5457 (M101) & 0.12 & 7.7 & 2.77$\times$10$^{9}$ & 1.12$\times$10$^{10}$ & 4.04 \\
1431: NGC~628 (M74) & 0.12 & 7.8 & 3.34$\times$10$^{8}$ & 2.85$\times$10$^{9}$ & 8.53 \\
1432: NGC~5194 (M51) & 0.00 & 8.4 & 1.23$\times$10$^{9}$ & 1.08$\times$10$^{10}$ & 8.78 \\
1433:
1434: \tablenotetext{a}{Foreground Galactic extinction: NGC~4631, NGC~~628, NGC~5194, Burstein and Heiles 1984;
1435: NGC~5236, AND NGC~5457, Hill {\it et al.}(1994).}
1436: \tablenotetext{b}{NGC~5236, Pierce (1993); NGC~5457 and NGC~5194, Feldmeir and Ciardullo (1997);
1437: NGC~628, Sharina (1996).}
1438: \tablenotetext{c}{Except for NGC~4631, luminosities are computed from Waller
1439: (1997b), and are adjusted for the distances given in Table~4.}
1440: \tablenotetext{d}{IRAS luminosities are from (Rice {\it et al.} 1988) and are
1441: adjusted for distances given in Table~4.}
1442:
1443:
1444: \enddata
1445:
1446: \end{deluxetable}
1447:
1448: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1449:
1450: \end{document}
1451:
1452: