1: \documentclass[12pt,preprint2]{aastex}
2:
3: \begin{document}
4:
5: \title{New Elemental Abundances for V1974 Cygni}
6:
7: \author{K.M. Vanlandingham}
8: \affil{Columbia University, Department of Astronomy, New York, NY 10027}
9: \email{kmv14@columbia.edu}
10:
11: \author{G.J. Schwarz}
12: \affil{Steward Observatory, University of Arizona, Tucson, AZ 85721}
13: \email{gschwarz@as.arizona.edu}
14:
15: \author{S.N. Shore}
16: \affil{Dipartimento di Fisica "Enrico Fermi", Universit\`a di Pisa,
17: largo Pontecorvo 3, Pisa 56127, Italy}
18: \email{shore@df.unipi.it}
19:
20: \author{S. Starrfield} \affil{Department of Physics \& Astronomy,
21: Arizona State University, Tempe, AZ 85287-1504}
22: \email{sumner.starrfield@asu.edu}
23:
24: \and
25:
26: \author{R.M. Wagner} \affil{Steward Observatory, University of
27: Arizona, Tucson, AZ 85721} \email{rmw@as.arizona.edu}
28:
29: \begin{abstract}
30: We present a new analysis of existing optical and ultraviolet spectra
31: of the ONeMg nova V1974 Cygni 1992. Using these data and the
32: photoionization code Cloudy, we have determined the physical
33: parameters and elemental abundances for this nova. Many of the
34: previous studies of this nova have made use of incorrect analyses and
35: hence a new study was required. Our results show that the ejecta are
36: enhanced, relative to solar, in helium, nitrogen, oxygen, neon,
37: magnesium and iron. Carbon was found to be subsolar. We find an
38: ejected mass of $\sim 2\times10^{-4}$M$_{\odot}$. Our model results
39: fit well with observations taken at IR, radio, sub-millimeter and
40: X-ray wavelengths.
41:
42: \end{abstract}
43:
44: \keywords{novae, binary stars,stars:individual (V1974 Cyg), stars:abundances}
45:
46: \section{Introduction}
47:
48: Novae explosions are the result of a thermonuclear runaway (TNR)
49: occurring on the surface of a white dwarf (WD) in a close binary
50: system. Material is transferred from the secondary star, a late-type
51: main-sequence star, through the inner Lagrangian point to an accretion disk and then
52: onto the WD. Once enough material is accreted nuclear fusion begins
53: in the surface layers of the WD. Since the WD is degenerate, this
54: leads to a TNR that then results in the ejection of the accreted
55: material. Analysis of the ejecta provides information about the
56: physics of the nova process and the WDs on which they take place.
57:
58: V1974 Cygni 1992 (hereafter Cyg 92) was discovered on 1992 February 19
59: by Collins (1992) (taken to be t$_0$ in this paper). At that time it
60: was the brightest nova since V1500 Cygni and hence was one of the most
61: extensively observed novae in history with observations spanning the
62: entire spectral range from gamma-rays to radio. Ultraviolet (UV)
63: observations made with the International Ultraviolet Explorer
64: satellite (IUE) show that Cyg 92 was a "neon" nova - one that takes
65: place on an ONeMg WD.
66:
67: Three detailed abundance analyses have been carried out for Cyg 92: Austin
68: et al. (1996, hereafter A96), Hayward et al. (1996) \& Moro-Mart\'{i}n et al. (2001).
69: Unfortunately these analyses were based on data that was dereddened incorrectly. In the
70: initial analysis by A96 the reddening correction was applied in the
71: wrong direction for the optical spectra. In other words, the optical spectra
72: were {\it reddened} rather than {\it dereddened}. The UV spectra were
73: dereddened correctly, however since the analyses relied on the flux ratios
74: relative to H$\beta$ the UV ratios were ultimately affected by this mistake.
75: Hence a new analysis is required.
76: Using optical and UV data we have determined physical
77: parameters and elemental abundances for Cyg 92. While we have used the same
78: photoionization code as A96 and Moro-Mart\'{i}n, we have developed a two-component
79: model to simulate the inhomogeniety of the nova shell which is more physically
80: accurate than these other analyses. Hayward et al. also used a multi-component
81: model in their work however we have included significantly more data in our
82: analysis than they did. We have compared our results
83: with X-ray, infrared (IR), sub-millimeter, and radio data in the literature and
84: find them consistent.
85:
86: In Section 2 we briefly describe the observations used in our
87: analysis. Section 3 reviews the reddening to the nova. An overview
88: of our analysis technique is given in Section 4. Section 5 contains a
89: detailed description of our model results and these results are
90: compared to others in the literature in Section 6. Our conclusions
91: and summary are given in Section 7.
92:
93: \section{Observations}
94:
95: The UV data were obtained with IUE. Low-dispersion large-aperture
96: data were taken with both the Short Wavelength Primary (SWP:
97: 1200-2000\AA) and the Long Wavelength Primary (LWP: 2000-3400\AA)
98: cameras (resolution 7\AA). The data were reduced at the Goddard
99: Space Flight Center (GSFC) Regional Data Analysis Facility (RDAF)
100: using the NEWSIPS IUE software. The optical data were taken with the
101: Perkins 1.8-m telescope at Lowell Observatory using the Ohio State
102: University Boller \& Chivens spectrograph (resolution 6\AA, wavelength
103: range 3200-8450\AA). For a detailed description of the data and
104: spectral evolution see A96. The data we are re-analyzing are the
105: spectra taken roughly 300, 400 and 500 days after the outburst.
106:
107: The optical spectra were not absolutely flux calibrated. In order to
108: combine them with the UV data we had to scale the optical flux to
109: match in the overlap region. This was done using the ratio of the
110: \ion{He}{2} lines at 1640\AA\ and 4686\AA. Osterbrock (1989) gives
111: the theoretical ratio of these lines as 6.79. After correcting the
112: spectra for reddening, we scaled the optical flux until the 1640/4686
113: ratio was equal to 6.79. The dereddened line fluxes are given in
114: Table 1.
115:
116: %\clearpage
117: \begin{table}
118: \caption{Dereddened Observed Line Fluxes\tablenotemark{a}}
119: \vspace{1mm}
120: \begin{tabular}{@{}lrrr}
121: \hline
122: Line ID&Day 300&Day 400&Day 500\\
123: \hline
124: \ion{N}{5} 1240\tablenotemark{b}&1616.5&7203.98&2284.19\\
125: \ion{O}{4}$]$ 1402&526.25&117.49&26.92\\
126: \ion{N}{4}$]$ 1486&865.56&202.34&49.46\\
127: \ion{C}{4} 1549&309.94&63.06&18.17\\
128: $[$\ion{Ne}{5}$]$ 1575&147.08&34.67&10.30\\
129: $[$\ion{Ne}{4}$]$ 1602&403.26&96.16&21.71\\
130: \ion{He}{2} 1640&233.66&56.89& 18.29\\
131: \ion{O}{3}$]$ 1663&221.33&36.22&13.08\\
132: \ion{N}{3}$]$ 1750&379.42&64.90&17.98\\
133: \ion{C}{3}$]$ 1909&105.75&19.13&4.50\\
134: $[$\ion{Ne}{4}$]$ 2424&66.69&64.70& -\\
135: \ion{Mg}{2} 2798&221.02&59.73&5.9\\
136: $[$\ion{Ne}{5}$]$ 2976&100.11&57.96 &5.1\\
137: $[$\ion{Ne}{5}$]$ 3346\tablenotemark{c}&1627.95&612.36&372.28\\
138: $[$\ion{Ne}{5}$]$ 3426&4411.55&1833.24&1123.39\\
139: $[$\ion{Ne}{3}$]$ 3869&937.85&273.665&151.16\\
140: $[$\ion{Ne}{3}$]$ 3968&299.56&80.33&43.48\\
141: \ion{H}{1} 4102&62.64&13.25&6.44\\
142: \ion{H}{1} 4340&35.74&9.53&6.47\\
143: $[$\ion{O}{3}$]$ 4363&308.89&90.33&53.12\\
144: \ion{He}{2} 4686&53.91&18.59&15.80\\
145: $[$\ion{Ne}{4}$]$ 4721&199.00&62.34&31.63\\
146: \ion{H}{4} 4861&85.72&24.03&13.94\\
147: $[$\ion{O}{3}$]$ 4959&199.60&73.29&58.23\\
148: $[$\ion{O}{3}$]$ 5007&568.70&201.02&156.35\\
149: $[$\ion{N}{2}$]$ 5755&16.62&9.87&9.63\\
150: \ion{He}{1} 5876&10.06&1.64&1.17\\
151: $[$\ion{Fe}{7}$]$ 6087&16.72&8.13&9.94\\
152: $[$\ion{O}{1}$]$ 6300&4.58&1.53&1.96\\
153: \ion{H}{1} 6563&250.79&52.20&31.11\\
154: $[$\ion{O}{2}$]$ 7325&26.26&6.47&3.42\\
155: \hline
156: \tablenotetext{a}{Lines have been dereddened using E(B-V)=0.32.}
157: \tablenotetext{b}{UV fluxes are 10$^{-12}$ erg sec$^{-1}$ cm$^{-2}$}
158: \tablenotetext{c}{Optical fluxes are relative. Scale factors for day 300, 400
159: and 500 are 6.4$\times10^{-13}$, 4.5$\times10^{-13}$, and 1.7$\times10^{-13}$,
160: respectively.}
161: \end{tabular}
162: \end{table}
163:
164:
165: \section{Reddening}
166:
167: A summary of reddening values used by other groups is given in Chochol
168: et al. (1997, their Table 1). Published values for E(B-V) range from
169: 0.17 to 0.42. For our analysis we have used two methods to determine
170: the reddening. First we used the interstellar absorption feature at
171: $2175$\AA. UV spectra taken with the Faint Object Spectrograph (FOS)
172: on the Hubble Space Telescope (HST) on 1995 November 30 clearly show
173: this feature. We have taken these data and applied several different
174: reddening values using the extinction curve of Seaton (1979) and fit a
175: line through the continuum. The spectra are shown in Figure 1. These
176: data clearly favor higher reddening values with the best value being
177: E(B-V)=0.36. For a second estimate of the reddening we looked at the
178: bolometric lightcurve. If we apply a reddening correction of 0.36 we
179: find that there is a deviation from constant flux. If we instead
180: lower our reddening value to 0.32 we are able to maintain a constant
181: flux while still removing most of the interstellar absorption feature
182: in the UV. Therefore we have chosen E(B-V)=0.32 as our reddening
183: value for our analysis. If we apply this reddening value and the range
184: of distances to the nova from the literature (1.5-3 kpc) to the bolometric
185: flux (Shore et. al 1994) we find the luminosity of the nova to be 1-4$\times10^{38}$ erg
186: s$^{1}$ cm$^{-2}$. This is roughly the Eddington luminosity for a 1M$_{\odot}$
187: WD.
188:
189: \begin{figure}
190: \plotone{f1.eps}
191: \caption{The FOS spectra are shown here dereddened using several
192: different E(B-V) values. A rough line is fit to the continuum in
193: each plot. These data clearly favor the higher reddening
194: values. \label{fig1}}
195: \end{figure}
196:
197: The FOS spectrum also provided serendipitous support for our previous
198: study of the X-ray turnoff for V1974 Cyg (Figure 2). The FOS observation
199: occurred soon after our last GHRS low
200: resolution spectrum (1995 Sept. 28) (Shore et al. 1997). Comparing
201: this GHRS large aperture (2 arcsec) low resolution spectrum with the FOS
202: data shows that the emission line fluxes have changed significantly
203: due to recombination following the X-ray turnoff,
204: which occurred before the GHRS spectrum was obtained.
205: At this stage in the evolution of the ejecta
206: ($\sim$1300 days after outburst), the helium recombination timescale
207: was approximately one month for the densities of about
208: 1$\times10^6$cm$^{-3}$. The integrated line flux of \ion{He}{2}
209: 1640\AA\ decreases by about 20\% between these two observations that
210: are roughly two months apart,
211: from 1.26$\times 10^{-13}$ (GHRS) to 1.06$\times 10^{-13}$ erg
212: s$^{-1}$cm$^{-1}$ (FOS). Since
213: both spectra were obtained with the large aperture, which completely
214: contained the then-resolved ejecta, and the continua agree in
215: intensity and slope, the change cannot be merely instrumental in
216: origin. It thus appears that the bulk of the emission came from
217: regions with densities that are characteristic of the clumps and
218: there was little emission from the diffuse gas (see also below).
219:
220:
221: \begin{figure}
222: \plotone{f2.eps}
223: \caption{A comparision of the FOS and GHRS spectra. The GHRS spectrum
224: (solid line) was taken on 1995 Sept. 28 and the FOS spectrum (dashed
225: line) was taken on 1995 Nov. 30. The flux in the emission lines has
226: decreased significantly between the two observations. No reddening
227: correction has been applied.}
228: \end{figure}
229:
230:
231: \section{Analysis}
232:
233: We use Cloudy 94.00 (Ferland et al. 1998) to model the relative line
234: strengths of Cyg 92 on three epochs independently. In the past we
235: used this method to determine the physical characteristics of many
236: other novae (Vanlandingham et al. 1996, 1997, 1999; Schwarz et
237: al. 1997, 2001, 2002). Cloudy
238: simultaneously solves the equations of statistical and
239: thermal equilibrium for specified initial physical conditions; the
240: model parameters are the spectral energy distribution of the
241: continuum source, its temperature and luminosity, the hydrogen
242: density, the density law for the ejecta (given by $\alpha$, where
243: $\rho \propto r^\alpha$), the inner and outer radii of the shell, the
244: geometry of the shell, the covering and filling factors of the shell,
245: the filling factor law (defined in the same way as the density law)
246: and the elemental abundances (relative to solar). We adopt a blackbody
247: to model the incident continuum. Our previous work with other novae has
248: shown that using a model other than a blackbody for the underlying
249: continuum source results in little difference in the model fit for
250: this interval in the outburst (Schwarz 2002). We ran models for one of
251: the dates using a NLTE model planetary nebula nuclei (Rauch 1997) as
252: the continuum source instead of a blackbody and were able to achieve
253: the same fit to the data with changing only the temperature of the
254: source by less than 10\%. While it has been found by Balman et. al (1998)
255: that blackbodies cannot be used to fit the soft X-ray observations,
256: we are not attempting to reproduce the spectral energy distribution of
257: the incident source so blackbodies give adequate results.
258: We can constrain the radiation temperature of the source using published
259: X-ray observations (Balman
260: et. al 1998), and the hydrogen density by observing the
261: relative strengths of various ionization stages of a given element.
262: The FWHM of the emission lines and the terminal velocities of the P
263: Cygni profiles provide the minimum and maximum velocities of the
264: ejecta and at any time since outburst, these are used to
265: determine an inner and outer radius of the nova shell. Based on the
266: luminosities derived for other ONeMg novae, we choose a starting
267: value of 1$\times 10^{38}$ erg s$^{-1}$ and then allow the luminosity
268: to vary with successive iterations of the code. We assume a
269: spherical geometry for the shell, and start with a covering factor of
270: unity. We choose an initial value for the filling factor of 0.1,
271: since previous studies have shown that novae do not eject homogeneous
272: shells but rather clumps of gas imbedded in a diffuse gas (Shore et
273: al. 1993).
274:
275: To determine the goodness of the fit of the model spectrum to the
276: observed spectrum we use the $\chi^2$ of the model:
277: \begin{equation}
278: \chi^2 = \sum {\frac{(M_i-O_i)^2}{\sigma_i^2}}
279: \end{equation}
280: where $M_i$ is the modeled line flux ratio, $O_i$ is the observed line
281: flux ratio, and $\sigma_i$ is the error in the measurement of the
282: observed flux for each line (A96). The error is determined by
283: measuring the line flux several times and looking at the variation of
284: the measurements. The variation between measurements is primarily
285: due to the placement of the continuum and therefore the weakest lines
286: have the largest errors. The flux for blended lines was estimated
287: using the 'deblend' option in the IRAF 'splot' package. These lines
288: also have higher errors than the average. These are typically on the
289: order of 20\% for the strongest lines but may be as high as 50\% for
290: the weakest or blended lines. From our observations we typically have
291: $\sim$30 emission lines on which to base our fit. Of the 24 input
292: parameters in Cloudy, we fix 13: the density power law, the filling
293: factor and its power law, the inner and outer radii, the geometry of
294: the shell, and 7 abundances for which we had no data. This usually
295: left us with $\sim$11 free parameters and $\sim$19 degrees of
296: freedom. A model is considered a good fit if it has a reduced
297: $\chi^2$ (defined as the $\chi^2$ divided by the degrees of freedom)
298: equal to one.
299:
300: \section{Modeling the Spectra}
301:
302: We modeled the same spectra described in A96 roughly corresponding to
303: 300, 400 and 500 days after outburst. The day 300 analysis is based
304: on the IUE spectra of 1992 December 4 combined with an optical
305: spectrum from 1992 December 15. Day 400 is represented by the IUE
306: spectra from 1993 April 4 and an optical spectrum from 1993 March 16.
307: We encountered an apparent calibration error in the LWP spectrum for
308: this date. In Figure 3 we have plotted the dereddened SWP and LWP
309: spectra for Day 400 with a line fit to the SWP continuum. The LWP
310: continuum appears to be too low. We can be somewhat confident that
311: the problem is with the LWP spectrum rather than the SWP spectrum
312: since there are two SWP spectra taken on this date and they agree with
313: one another. If we multiply the LWP spectrum by a factor of ten then
314: the fit is much better. While this is an eye estimate, the two
315: spectra match reasonably well if a multiplication factor anywhere
316: between 8 and 12 is used. This uncertainty in the calibration of the
317: LWP spectrum results in a 20\% uncertainty in the line flux ratio for
318: the three emission lines obtained from this spectrum. Lastly, our Day
319: 500 analysis consists of the IUE spectrum from 1993 July 2 and an
320: optical spectrum from 1993 July 17. There appeared to be a problem
321: with the LWP spectrum on July 2, most likely due to scattered light,
322: so only the SWP spectrum was used for this date. The dates for the UV
323: and optical spectra do not match exactly for each of these pairings
324: however, due to the fact that the nova is evolving very slowly this
325: late after the outburst, this does not present a problem.
326:
327: \begin{figure}
328: \plotone{f3.eps}
329: \caption{The top box shows the SWP+LWP spectra for Day 400 with a
330: reddening correction of 0.32 applied. A rough line has been fit to
331: the SWP continuum. The bottom box shows the same spectra but with the
332: LWP flux multiplied by a factor of 10. The fit to the LWP continuum
333: is much improved. \label{fig2}}
334: \end{figure}
335:
336: \subsection{One-Component Model}
337:
338: The first date we modeled was Day 300.
339: We used 29 emission lines in our fit (see Table 2). The resulting values for
340: the physical parameters and elemental abundances are given in Table 4.
341: All parameters are in cgs unit. The luminosity is given in erg s$^{-1}$, the density
342: in g cm$^{-3}$, and the radii are in cm. All abundances are given by number relative to
343: hydrogen relative to solar. The fit has a
344: $\chi^2$=28 which gives a reduced $\chi^2$ of 1.6. The number in parentheses next to
345: the abundance in Table 4 is the number of spectral lines that were used to determine that abundance.
346: The greater the number of lines the more constrained the value
347: of the abundance. Thus the abundances of magnesium and iron are much more uncertain being
348: based solely on the fit to one spectral line, although they are probably accurate to within
349: a factor of 5-10.
350:
351: \clearpage
352: \begin{table}
353: \caption{Day 300 Observed and Model Line Flux Ratios\tablenotemark{a}}
354: \vspace{1mm}
355: \begin{tabular}{@{}lrrrrr}
356: \hline
357: &Observed&1-component&&2-component&\\
358: Line ID&Ratio&Model Ratio&$\chi^2$&Model Ratio&$\chi^2$\\
359: \hline
360: \ion{N}{5} 1240& 29.33& 25.14&0.23&30.90&0.03\\
361: \ion{O}{4}$]$ 1402& 9.58&6.74&0.55&8.81&0.04\\
362: \ion{N}{4}$]$ 1486& 15.77&16.97&0.07&15.43&0.01\\
363: \ion{C}{4} 1549& 5.64&6.00&0.05&5.78&0.01\\
364: $[$\ion{Ne}{5}$]$ 1575& 2.68&2.08&0.55&2.50&0.05\\
365: $[$\ion{Ne}{4}$]$ 1602& 7.34&7.00&0.02&7.60&0.01\\
366: \ion{He}{2} 1640& 4.26&4.33&0.003&4.81&0.19\\
367: \ion{O}{3}$]$ 1663& 4.03&6.00&2.65&5.90&2.39\\
368: \ion{N}{3}$]$ 1750& 6.90&7.41&0.06&5.94&0.22\\
369: \ion{C}{3}$]$ 1909& 1.92&1.99&0.01&1.70&0.16\\
370: $[$\ion{Ne}{4}$]$ 2424& 1.22&0.21&2.73&1.43&0.13\\
371: \ion{Mg}{2} 2798& 4.02&3.85&0.02&3.99&0.00\\
372: $[$\ion{Ne}{5}$]$ 2976& 1.82&0.70&4.23&0.84&3.26\\
373: $[$\ion{Ne}{5}$]$ 3346&18.94& - & 10 &17.19&0.21\\
374: $[$\ion{Ne}{5}$]$ 3426&51.37& - & 10 &47.06&0.18\\
375: $[$\ion{Ne}{3}$]$ 3869& 10.93&14.40&2.52&11.32&0.03\\
376: $[$\ion{Ne}{3}$]$ 3968& 3.50&4.34&1.45&3.41&0.02\\
377: \ion{H}{1} 4102& 0.73&0.29&1.43&0.28&1.50\\
378: \ion{H}{1} 4340& 0.42&0.50&0.48&0.49&0.36\\
379: $[$\ion{O}{3}$]$ 4363& 3.61&2.58&0.91&2.69&0.71\\
380: \ion{He}{2} 4686& 0.63&0.56&0.14&0.64&0.01\\
381: $[$\ion{Ne}{4}$]$ 4721& 2.31&1.61&1.01&1.75&0.65\\
382: \ion{H}{4} 4861& 1.00&1.00&0.00&1.00&0.00\\
383: $[$\ion{O}{3}$]$ 4959& 2.33&0.67&3.18&0.83&2.60\\
384: $[$\ion{O}{3}$]$ 5007& 6.63&1.93&3.14&2.39&2.56\\
385: $[$\ion{N}{2}$]$ 5755& 0.19&0.16&0.27&0.14&0.98\\
386: \ion{He}{1} 5876& 0.12&0.12&0.002&0.07&0.90\\
387: $[$\ion{Fe}{7}$]$ 6087& 0.19&0.15&0.36&0.18&0.04\\
388: $[$\ion{O}{1}$]$ 6300& 0.05&0.06&0.18&0.09&3.65\\
389: \ion{H}{1} 6563& 2.93&3.26&0.13&2.99&0.00\\
390: $[$\ion{O}{2}$]$ 7325& 0.31&0.20&1.24&0.24&0.58\\
391: &&&&&\\
392: Total $\chi^2$&&&28\tablenotemark{b}&&22\\
393: \hline
394: \tablenotetext{a}{Ratios are relative to H$\beta$}
395: \tablenotetext{b}{$[$\ion{Ne}{5}$]$ 3346, 3426\AA\AA\ are not included
396: in the total $\chi^2$ of this model}
397: \end{tabular}
398: \end{table}
399:
400: \clearpage
401:
402: \begin{table}
403: \caption{Day 400 \& 500 Observed and Model Line Flux Ratios\tablenotemark{a}}
404: \vspace{1mm}
405: \begin{tabular}{@{}lrrrrrr}
406: \hline
407: &Observed&Day 400&&Observed&Day 500&\\
408: Line ID&Ratio&Model Ratio&$\chi^2$&Ratio&Model Ratio&$\chi^2$\\
409: \hline
410: \ion{N}{5} 1240&66.64 &84.80&0.83&96.37&92.58&0.02\\
411: \ion{O}{5} 1371&1.78&1.15&0.78&-&-&-\\
412: \ion{O}{4}$]$ 1402& 10.86 &9.99&0.04&11.36&4.29&2.42\\
413: \ion{N}{4}$]$ 1486& 18.71 &17.52&0.05&20.87&18.24&0.18\\
414: \ion{C}{4} 1549& 5.83 &6.67&0.23&7.67&7.57&0.00\\
415: $[$\ion{Ne}{5}$]$ 1575& 3.21 &2.71&0.26&4.34&1.72&4.04\\
416: $[$\ion{Ne}{4}$]$ 1602& 8.90 &4.19&3.11&9.16&2.80&5.36\\
417: \ion{He}{2} 1640& 5.26 &6.16&0.32&7.72&8.27&0.06\\
418: \ion{O}{3}$]$ 1663& 3.35 &4.39&1.07&5.52&2.68&2.94\\
419: \ion{N}{4} 1719&1.29&0.35&3.34&1.39&0.33&3.63\\
420: \ion{N}{3}$]$ 1750& 6.00 &4.12&1.09&7.58&5.51&0.83\\
421: \ion{C}{3}$]$ 1909& 1.77 &0.92&2.55&1.90&1.14&1.76\\
422: $[$\ion{Ne}{4}$]$ 2424& 5.98 &0.45&3.43&-&-&-\\
423: \ion{Mg}{2} 2798& 5.52 &3.28&1.03&-&-&-\\
424: $[$\ion{Ne}{5}$]$ 2976& 5.36 &0.91&7.67&-&-&-\\
425: $[$\ion{Ne}{5}$]$ 3346&25.50&21.49&0.62&26.70&16.48&3.66\\
426: $[$\ion{Ne}{5}$]$ 3426&76.32&58.83&1.31&80.58&45.13&4.84\\
427: $[$\ion{Ne}{3}$]$ 3869& 11.39 &11.58&0.01&10.84&10.18&0.09\\
428: $[$\ion{Ne}{3}$]$ 3968& 3.34 &3.49&0.05&3.12&3.07&0.01\\
429: \ion{H}{1} 4102& 0.55 &0.28&0.98&0.46&0.27&0.66\\
430: \ion{H}{1} 4340& 0.40 &0.49&0.58&0.46&0.48&0.02\\
431: $[$\ion{O}{3}$]$ 4363& 3.76 &3.96&0.03&3.81&2.53&1.24\\
432: \ion{He}{2} 4686& 0.77 &0.82&0.03&1.13&1.10&0.01\\
433: $[$\ion{Ne}{4}$]$ 4721& 2.60 &0.97&4.38&2.27&0.64&5.70\\
434: \ion{H}{4} 4861& 1.00&1.00&0.00&1.00&1.00&0.00\\
435: $[$\ion{O}{3}$]$ 4959& 3.05 &2.33&0.35&4.12&1.93&1.77\\
436: $[$\ion{O}{3}$]$ 5007& 8.37 &6.73&0.24&11.22&5.57&1.59\\
437: $[$\ion{N}{2}$]$ 5755& 0.41 &0.13&5.13&0.69&0.20&5.64\\
438: \ion{He}{1} 5876& 0.07 &0.08&0.18&0.08&0.07&0.27\\
439: $[$\ion{Fe}{7}$]$ 6087& 0.34 &0.40&0.21&0.71&0.88&0.32\\
440: $[$\ion{O}{1}$]$ 6300& 0.06 &0.11&3.10&0.14&0.13&0.02\\
441: \ion{H}{1} 6563& 2.17 &2.94&1.40&2.23&2.88&0.93\\
442: $[$\ion{O}{2}$]$ 7325& 0.27 &0.23&0.21&0.25&0.20&0.39\\
443: &&&&&\\
444: Total $\chi^2$&&&45&&&48\\
445: \hline
446: \tablenotetext{a}{Ratios are relative to H$\beta$}
447: \end{tabular}
448: \end{table}
449:
450: \clearpage
451:
452: \begin{table}
453: \caption{Nova Cyg 92 Model Parameters}
454: \vspace{1mm}
455: \begin{tabular}{@{}lrrrr}
456: \hline
457: &One-Component&Two-Component&Two-Component&Two-Component\\
458: Parameter&Day 300&Day 300&Day 400&Day 500\\
459: \hline
460: log(T$_{BB})$ K&5.52&5.52&5.67 &5.65\\
461: log(L) erg s$^{-1}$ &38.06&38.06&38.37 &38.1\\
462: log(H$_{den})$(diffuse) g cm$^{-3}$&-&7.097&6.99&6.79\\
463: log(H$_{den})$(clump) g cm$^{-3}$&7.75&7.75&7.39&7.1\\
464: $\alpha$&-3.0&-3.0&-3.0 &-3.0\\
465: log(R$_{in}$) cm &15.317&15.317&15.44 &15.54\\
466: log(R$_{out}$) cm &15.715&15.715&15.84 &15.94\\
467: Fill&0.1&0.1&0.1 &0.1\\
468: Power&0.0&0.0&0.0 &0.0\\
469: Cover(diffuse)&-&0.5&0.62&0.47\\
470: Cover(clump)&0.3&0.32&0.33&0.18\\
471: He\tablenotemark{a}&1.1(3)&1.0 (3)&1.3 (3) &1.4 (3)\\
472: C\tablenotemark{a}&0.40(2)&0.69 (2)&0.87 (2) &0.59 (2)\\
473: N\tablenotemark{a}&21.4(4)&35.5 (4)&57.6 (5)&41.7 (5)\\
474: O\tablenotemark{a}&3.9(7)&12.3 (7)&19.5 (8)&6.5 (7)\\
475: Ne\tablenotemark{a}&38.2(7)&56.5 (9)&44.9 (8)&23.0 (7)\\
476: Mg\tablenotemark{a}&1.3(1)&2.6 (1)&6.5 (1)&1.0 (0)\\
477: Fe\tablenotemark{a}&1.7(1)&1.3 (1)&3.8 (1)&9.5 (1)\\
478: $\chi^2$&28&22 &45 &48\\
479: &&&&\\
480: Total \# of lines &29& 31 & 33 & 29\\
481: \# of free parameters &11& 13 & 13 & 12\\
482: DOF &18& 18 & 20 & 17\\
483: Reduced $\chi^2$&1.6& 1.2 & 2.3 & 2.8\\
484: \hline
485: \tablenotetext{a}{Elements are given by number relative to hydrogen relative to solar.}
486: \end{tabular}
487: \end{table}
488:
489: \clearpage
490:
491: Despite the low $\chi^2$, there are several problems with this model.
492: First, the model is unable to adequately reproduce the high ionization
493: lines seen in the observations and in particular the $[$\ion{Ne}{5}$]$
494: 3324, 3426\AA\ lines. The $\chi^2$ quoted for the model is with these
495: two lines removed from the fit, each of which have an individual
496: $\chi^2$ of $\sim$10. This is disconcerting since these are the
497: strongest lines in the optical spectrum and hence should have small
498: measurement errors associated with them. There is some intrinsic
499: error introduced by the fact that these lines are at the bluest end of
500: the spectrum where the sensitivity of the CCD drops off. However, the
501: magnitude of the discrepancy between the observed flux and the model
502: flux is too large to be explained by this. There are also other
503: groups who report the flux ratios for these lines
504: (Moro-Mart\'{i}n et al. 2001) and our measurements agree with
505: theirs. The difference between the model and observations is more
506: likely a shortcoming of the model itself. In addition to the
507: $[$\ion{Ne}{5}$]$ lines, our initial model is also unable to
508: reproduce the $[$\ion{Ne}{6}$]$ 76$\mu$/$[$\ion{Ne}{2}$]$ 128$\mu$
509: line ratio reported by Hayward et al. (1996). Our model predicts a
510: value of 0.8 for this ratio while they measured a value of $\sim45$.
511: Clearly, the one-component model consistently under-predicts the high
512: ionization lines in the spectra.
513:
514: \subsection{A Two-Component Model}
515:
516: Novae ejecta are not uniform in density but rather are clumpy, with
517: knots of high density material embedded in a more diffuse gas (Shore
518: et al. 1993). This is seen quite clearly in HST images of Cyg 92.
519: As a one-dimensional model,
520: Cloudy is not well suited to represent this type of environment. To
521: overcome this shortcoming, we have created a two-component model, one
522: component being the clumps and the other a diffuse gas, where the
523: resulting line fluxes from the two components are then combined.
524: While this model is more realistic than a simple one-component model,
525: it is still not perfect since the two components are handled
526: separately by Cloudy when in reality they are not separate. Ideally,
527: we would like to be able to embed the clumpy component within the
528: diffuse component but this is beyond the abilities of Cloudy.
529: However, we feel that our two-component model is a reasonable
530: approximation until a better model is found. Since our initial
531: one-component model fit a majority of the lines, we added a second
532: component to increase the flux of the high ionization lines. Most of
533: the model parameters for the two components should be the same. For
534: instance, the elemental abundances are not expected to vary between
535: the clumps and the diffuse component. However, the physical
536: parameters, such as the density, filling factor and covering factor,
537: will necessarily be different for each component. Going to a
538: two-component model increases the number of free parameters and makes
539: the task of finding a solution more difficult. In our previous
540: analyses of other novae, one-component models fit the available
541: observations quite well. This analysis of Cyg 92 has been the first
542: time a one-component model has had difficulty in fitting the
543: observations. This is due to the wealth of data at wavelengths beyond
544: the optical and UV that we are able to use to constrain our models of
545: Cyg 92.
546:
547: We added a diffuse component with a density that was less than the
548: original model (now considered the 'clump' component). We again
549: adjusted the free parameters to obtain the best fit to the
550: observations. The addition of the second component increased the
551: number of free parameters by 2 since we now have a second density and
552: a second covering factor. The other parameters (shell radii,
553: elemental abundances, etc.) were set to the same value as the first
554: component. The fit to the individual emission lines for our best
555: two-component model is shown in Table 3. The parameters of the model
556: are given in Table 4. The fit has the same temperature and
557: luminosity for the underlying source as the initial one-component
558: model. The elemental abundances are slightly enhanced relative to
559: the one-component model.
560:
561: The best values of the covering factors for the clump and diffuse
562: components are found to be 0.32 and 0.5, respectively. The reduced
563: $\chi^2$ of the fit is 1.3 which is better than our original
564: one-component model and we now include the $[$\ion{Ne}{5}$]$ lines in
565: the $\chi^2$. The two-component model also improves our fit to the
566: IR lines ratios. We find a ratio of
567: $[$\ion{Ne}{6}$]$/$[$\ion{Ne}{2}$]$=22 which is only a factor of 2 too
568: small instead of a factor of 50. This shows that the two-component
569: model is significantly more realistic than the simple one-component
570: model used previously. Woodward et al. (1995) also found that
571: $[$\ion{Mg}{8}$]$ 30$\mu$/$[$\ion{Al}{6}$]$ 36$\mu$ $\sim$4 and our
572: two-component model predicts a ratio of 13. The fact that this is now
573: higher than what is observed is most likely because we have set the
574: aluminum abundance to solar. We have found in our work with other
575: ONeMg novae that aluminum is typically enhanced (Vanlandingham et
576: al. 1996, 1997, 1999). We do observe lines of aluminum
577: ($[$\ion{Al}{6}$]$ 2601\AA\ and \ion{Al}{2}$]$ 2665\AA) in the Cyg 92
578: spectra however they are too weak to measure reliably so we have not
579: used them in our models. If we increase the abundance of aluminum to
580: 2.4 times solar then this ratio matches the observations.
581:
582: As an additional check of our models, we can use the radio and
583: sub-millimeter observations of Cyg 92. The observations pertinent to
584: this analysis range from 1 to 500 GHz and were obtained within $\pm$
585: 50 days of the Day 300 optical and UV observations (Hjellming 1996;
586: Ivison et al. 1993; Eyres, Davis, \& Bode 1996). To compare our
587: model to the observations, the luminosity was scaled to distances of
588: 1.5 and 3 kpc corresponding to the range of distances reported in the
589: literature. Figure 4 shows the comparison of the model to the
590: observations. The larger distance (solid line) is consistent with
591: the average of the sub-millimeter observations on days 234 and 356 but
592: not the radio data. At 1.5 kpc the model is in better agreement with
593: the radio observations but overestimates the sub-millimeter data. The
594: disagreement may be due to a number of factors in the model including
595: being optically thin or having the wrong temperature. In addition,
596: the radio images clearly showed an ellipsoidal shell whereas our
597: models are spherical.
598:
599:
600: \begin{figure}
601: \plotone{f4.eps}
602: \caption{Model for Day 300 as compared to the observed radio and
603: sub-millimeter data in the literature. The radio data are taken from
604: Hjellming (1996) [filled circles] and Eyres et al. (1996) [day 270 and
605: 315 = triangles]. The sub-millimeter data are from Ivison et
606: al. (1993) [day 234 = asterisks and day 356 = diamonds]. The dotted
607: line represents the model luminosity scaled to a distance
608: of 1.5 kpc while the solid line is scaled to 3 kpc.
609: \label{fig3}}
610: \end{figure}
611:
612: After finding a fit to the Day 300 data we then proceeded to use our
613: two-component model to fit days 400 and 500. The individual emission
614: line fits for these days are shown in Table 3 and the model parameters
615: are given in Table 4. These later dates are not fit as well as Day
616: 300 as there is not as much data at other wavelengths to constrain the
617: fits. There are some lines in Table 3 that have much large $\chi^2$
618: than most of the others. Some of these, such as $[$\ion{Ne}{4}$]$]
619: 1602\AA\ and $[$\ion{Ne}{4}$]$ 4721\AA, are probably a result of the
620: line being blended making an accurate measurement of the flux
621: difficult. Other lines, such as \ion{N}{4} 1719\AA\ and
622: $[$\ion{N}{2}$]$ 5755\AA, are weak lines which, again, increases the
623: error in the measurement of the flux. We may have underestimated the
624: errors for these lines in our calculation of the $\chi^2$.
625:
626: If we compare the results from all three dates, we notice some trends.
627: The effective temperature of the continuum source increases slightly
628: with time. This is as expected as the ejected shell expands and
629: reveals more of the underlying WD surface. The luminosity is roughly
630: constant, as is the covering factors of the two components. This is
631: in agreement with the findings of Balman et al. (1998). Their X-ray
632: observations show the effective temperature peaking at day 511 and a
633: constant bolometric luminosity from day 255-511. The nova then turned
634: off shortly after this at roughly day 550. The abundances values for
635: all three dates typically agree within a factor of 2.5. All the
636: abundances except for carbon are enhanced relative to solar values.
637:
638: Based on the parameters determined from our three dates we can
639: calculate the hydrogen ejected mass predicted by our two-component
640: models. In order estimate the ejected mass we take the shell defined
641: by the inner and outer radii and divide it into 1000 nested shells.
642: The density of the innermost shell is set to the starting density of
643: the model and is then progressed based on the density law of the
644: model. The filling factor is applied in the same manner. The
645: resulting mass is then multiplied by the covering factor. Using this
646: method, we find an ejected mass of $2.1\pm0.2\times10^{-4}M_{\odot}$,
647: $2.8\pm0.3\times10^{-4}M_{\odot}$, and $2.1\pm0.2\times10^{-4}M_{\odot}$ for days
648: 300, 400 and 500, respectively. The ejected mass is roughly constant
649: over all dates, which is as expected. Shore et al. (1993) finds that
650: the ejected mass can be calculated as $10^{-4} Y^{-1/2} M_{\odot}$,
651: where Y is the average enhancement factor for the helium abundance.
652: If we use this equation and an average helium abundance from our three
653: models, we find $M_{ej}=1.9\times10^{-4}M_{\odot}$, which agrees with
654: our model values. Our masses agree fairly well with
655: this calculation. Other groups have found ejected mass estimates for
656: Cyg 92 (Shore et al. 1993, Krautter et al. 1996, Woodward et al. 1997)
657: in the same range that we find here.
658:
659:
660: \clearpage
661: \begin{table}
662: \caption{Abundance Comparison}
663: \vspace{1mm}
664: \begin{tabular}{@{}lccccc}
665: \hline
666: & Average from&&&\\
667: Parameter&this paper \tablenotemark{a}&A96 & Moro-Mart\'{i}n& Hayward&Paresce\\
668: \hline
669: He&1.2$\pm$0.2&4.4 &4.5&4.5 \tablenotemark{b} &$\sim$2\\
670: C&0.7$\pm$0.2& - & 70.6 \tablenotemark{b}&12 \tablenotemark{b}&\\
671: N&44.9$\pm$11&282 &50.0&50 \tablenotemark{b}&\\
672: O&12.8$\pm$7&110 &80.0&25 \tablenotemark{b}&2-4\\
673: Ne&41.5$\pm$17&250 &250.0&50&15-27\\
674: Mg&4.6$\pm$3&- &129.4 \tablenotemark{b}&5&\\
675: Al&$>$ 1.0\tablenotemark{c}&- &127.5 \tablenotemark{b}&5 \tablenotemark{b}&\\
676: Si&-&- &146.6 \tablenotemark{b}&6 \tablenotemark{b}&\\
677: S&-&- &1.0&5 \tablenotemark{b}&\\
678: Fe&4.9$\pm$4&16&8.0&4\tablenotemark{b}&\\
679: \tablenotetext{a}{Errors given are 1$\sigma$}
680: \tablenotetext{b}{No lines of this element are present in their spectra}
681: \tablenotetext{c}{This is a rough estimate based on the IR line ratios}
682: \end{tabular}
683: \end{table}
684: \clearpage
685:
686: \section{Comparison to other results}
687:
688: The first extensive analysis of the optical and UV data was done by
689: A96 using an older version of the Cloudy code. As mentioned earlier,
690: A96 made an error in applying the reddening corrections to their fluxes.
691: %The optical line
692: %fluxes listed in Austin's Table 7 are noted as dereddened. In fact,
693: %the only way to reproduce the numbers in Austin's table is to assume
694: %that these fluxes have had the reddening correction applied in the
695: %wrong sense such that they are redder than observed.
696: %These doubly reddened fluxes were then used in Austin's analysis.
697: Moro-Mart\'{i}n et al. (2001) and Hayward et al. (1996) also
698: conducted abundance analyses of Cyg 92 using Cloudy. Unfortunately,
699: both of these groups based their modeling on the results of A96 and
700: hence propagated the errors from that analysis into their work. In
701: addition, both of these studies report abundance values for elements
702: which are not represented in their spectra. It is possible to
703: predict an upper limit for a given element by increasing its
704: abundance until the model produces emission lines that should have
705: been seen but are not seen in the data. These two groups, however,
706: report such large abundances of these unseen elements that emission
707: lines would have been easily seen in the spectrum.
708:
709: A third analysis is found in the literature. Paresce et al. (1995)
710: gave rough abundances values for a few elements determined by using
711: Cloudy on a specific "knot" of material from their HST spectra. These
712: results, shown in Table 5, did not rely on the analysis of A96. They
713: state that their results are lower limits on the abundances.
714:
715: Finally we note that Shore et. al (1997) took the results from A96 and propagated them
716: forward in time to Day 1300. Using GHRS data, with a much higher S/N
717: than that of A96, they noted that the carbon abundance found by A96 was
718: too high.
719:
720: Given the error in A96 a comparison between our results and those of
721: Hayward and Moro-Mart\'{i}n would be misleading. It is not
722: surprising that our abundance results do not agree with their
723: results. The discovery of the error in A96 was one of the primary
724: motivations for our re-analysis of Cyg 92. Table 5 shows our results
725: along with those of A96, Hayward and Moro-Mart\'{i}n. In
726: general, our abundances are much lower than those of the three groups.
727: Our results are higher than those found by Paresce, however, they are
728: not in disagreement since Paresce's numbers are lower limits.
729:
730: Through our previous work we have found striking similarities between many
731: of the ONeMg novae (Shore et. al 2003, Vanlandingham et. al 1999). A complete
732: and thorough comparison between the abundances found here and other ONeMg
733: novae will be the subject of a future work.
734:
735: \section{Conclusions}
736:
737: We have applied a two-component photoionization code to three separate
738: observations of Cyg 92 and have derived the physical characteristics
739: of the ejecta on these dates. Our initial one-component model was
740: unable to reproduce the high ionization lines seen in the spectra.
741: By adding a second, low density, component to our models we were able
742: to correct this problem. We find the ejecta to be enhanced, relative
743: to solar, in He, N, O, Ne, Mg and Fe. Carbon is found to be subsolar.
744: Our models predict an ejected mass of $\sim 2\times10^{-4}M_{\odot}$
745: which is in agreement with what has been found for other ONeMg novae.
746:
747: Our results replace the earlier analysis of A96 that contained an
748: error in the reddening correction. The two other analyses in the
749: literature (Hayward et al. (1996) \& Moro-Mart\'{i}n et
750: al. (2001)) based their work on the results of A96 and so propagated
751: this error into their work. Because there is such a large parameter
752: space and many of the parameters are interdependent it is difficult to
753: determine if a solution to one set of spectra is unique. By modeling
754: three sets of observations independently, taken at different times
755: during the evolution of the nova shell, we increase the confidence in
756: our solution. If all three days arrive at the same abundance
757: solution, then we can have much more confidence that it is the true
758: solution. The fact that our models are able to fit the IR, radio,
759: sub-millimeter and X-ray observations further strengthens our
760: conclusions.
761:
762: \acknowledgments
763:
764: The authors would like to thank G. Ferland for use of his Cloudy
765: photoionization code and J. Aufdenberg and J. Jos\`e
766: for many useful discussions. We would also like to thank the referee
767: for a very thorough review. The referee's comments have greatly improved
768: the quality of the paper.
769: S. Starrfied acknowledges partial support from NSF and NASA grants to
770: ASU; S. N. Shore acknowledges partial support from NASA grants to
771: IUSB, and INAF 2002 and INFN/Pisa.
772:
773:
774: \begin{references}
775: \reference{head1996} Austin, S.J., Wagner, R.M., Starrfield, S., Shore, S.N., Sonneborn, G.
776: \& Bertram, R. 1996, \aj, 111, 869
777: \reference{head1998} Balman, S., Krautter, J. \& \"{O}gelman, H. 1998, \apj, 499, 395
778: \reference{head1997} Chochol, D., Grygar, J., Pribulla, T., Kom$\check{z}$ik, R., Hric, L. \&
779: Elkin, V. 1997, A\&A, 318, 908
780: \reference{head1992} Collins, P. 1992, IAUC 5454
781: \reference{head1996} Eyres, S. P. S., Davis, R. J. \& Bode, M. F. 1996, MNRAS, 279, 249
782: \reference{head1998} Ferland, G.J., Korista, K.T., Verner, D.A., Ferguson, J.W., Kingdon, J.B. \&
783: Verner, E.M. 1998, \pasp, 110, 761
784: \reference{head1996} Hayward, T.L., Saizar, P., Gehrz, R.D., Benjamin, A., Mason, C.G.,
785: Houck, J.R., Miles, J.W., Gull, G.E. \& Schoenwald, J. 1996, \apj, 469, 854
786: \reference{head1996} Hjellming, R. 1996, in 'Cataclysmic variables and related objects:
787: Proceedings of the 158th colloquium of the International Astronomical Union (IAU)', 317
788: \reference{head1993} Ivison, R. J., Hughes, D. H., Lloyd, H. M., Bang, M. K. \& Bode, M. F. 1993,
789: MNRAS, 263, 431
790: \reference{head1996} Krautter, J, \"{O}gelman, H., Starrfield, S., Wichmann, R. \&
791: Pfeffermann, E. 1996, \apj, 456, 788
792: \reference{head2001} Moro-Mart\'{i}n, A., Garnavich, P.M. \& Noriega-Crespo, A. 2001, \apj, 121, 1636
793: \reference{head1989} Osterbrock, D. 1989, Astrophysics of Gaseous Nebulae and Active Galactic Nuclei (California: University Science Books)
794: \reference{head1995} Paresce, F., Livio, M., Hack, W. \& Korista, K. 1995, A\&A, 299, 823
795: \reference{head1997} Rauch, T. 1997, A\&A, 320, 237
796: \reference{head2002} Schwarz, G.J. 2002, \apj, 577, 940
797: \reference{head2001} Schwarz, G.J., Shore, S.N., Starrfield, S., Hauschildt, P.H., Della Valle, M., \& Baron, E. 2001, MNRAS, 320, 103
798: \reference{head1997} Schwarz, G.J., Starrfield, S., Shore, S.N., \& Hauschildt, P.H. 1997, MNRAS, 290, 75
799: \reference{head1979} Seaton, M.J. 1979, MNRAS, 187, 73
800: \reference{head1993} Shore, S.N., Sonneborn, G., Starrfield, S., Gonazlez-Riestra, R. \&
801: Ake, T.B. 1993, \aj, 106, 2408
802: \reference{head1994} Shore, S.N., Sonneborn, G., Starrfield, S., Gonzalez-Riestra, R. \&
803: Polidan, R.S. 1994, \apj, 421, 344
804: \reference{head1996} Shore, S.N., Starrfield, S. \& Sonneborn, G. 1996, \apj, 463, L21
805: \reference{head1997} Shore, S.N., Starrfield, S., Ake, T.B.III \& Hauschildt, P.H. 1997, \apj,
806: 490, 393
807: \reference{head2003} Shore, S.N., Schwarz, G.J., Bond, H.E., Downes, R.A., Starrfield, S., Evans, A.,
808: Gehrz, R.D., Hauschildt, P.H., Krautter, J., \& Woodward, C.E. 2003, ApJ, 125, 1507
809: \reference{head1996} Vanlandingham, K.M., Starrfield, S., Wagner, R.M., Shore, S.N., \& Sonneborn, G. 1996, MNRAS, 282, 563
810: \reference{head1997} Vanlandingham, K.M., Starrfield, S., \& Shore, S.N. 1997, MNRAS, 290, 87
811: \reference{head1999} Vanlandingham, K.M., Starrfield, S., Shore, S.N., \& Sonneborn, G. 1999, MMRAS, 308, 577
812: \reference{head1995} Woodward, C.E., Greenhouse, M.A., Gehrz, R.D., Pendleton, Y.J., Joyce, R.R.,
813: Van Buren, D., Fischer, J., Jennerjohn, N.J. \& Kaminski, C.D. 1995, \apj, 438, 921
814: \reference{head1997} Woodward, C.E., Gehrz, D., Jones, T.J., Lawrencem, G.F., \& Skrutskie, M.F. 1997, ApJ, 477, 817
815:
816:
817: \end{references}
818:
819: \end{document}
820: