1: %\documentstyle[12pt,aasms4,psfig]{article}

2: \documentclass[12pt,preprint,psfig]{aastex}

3: \newcommand{\etal}{{\it et al.} }

4: \newcommand{\rxte}{{\it RXTE} }

5: \newcommand{\rxtep}{{\it RXTE}}

6: \newcommand{\asca}{{\it ASCA} }

7: \newcommand{\ascap}{{\it ASCA}}

8: \newcommand{\exosat}{{\it EXOSAT} }

9: \newcommand{\ginga}{{\it Ginga} }

10: \newcommand{\bbxrt}{{\it BBXRT} }

11: \newcommand{\rosat}{{\it ROSAT} }

12: \newcommand{\xmm}{{\it XMM-Newton} }

13: \newcommand{\chandra}{{\it Chandra} }

14: \newcommand{\chandrap}{{\it Chandra}}

15: \newcommand{\rrg}{$r/r_{g}$ }

16: \newcommand{\rrgp}{$r/r_{g}$}

17: \newcommand{\olya}{O~{\sc viii}~Ly$\alpha$ }

18: \newcommand{\olyap}{O~{\sc viii}~Ly$\alpha$}

19: \newcommand{\feka}{{Fe~K$\alpha$} }

20: \newcommand{\fekap}{{Fe~K$\alpha$}}

21: \newcommand{\fexxv}{Fe~{\sc xxv} }

22: \newcommand{\fexxvp}{Fe~{\sc xxv}}

23: \newcommand{\fexxvi}{Fe~{\sc xxvi} }

24: \newcommand{\fexxvip}{Fe~{\sc xxvi}}

25: \newcommand{\feklya}{{Fe~{\sc xxvi }~Ly$\alpha$} }

26: \newcommand{\feklyb}{{Fe~{\sc xxvi}~Ly$\beta$} }

27: \newcommand{\feklybp}{{Fe~{\sc xxvi}~Ly$\beta$}}

28: \newcommand{\feklyap}{{Fe~{\sc xxvi}~Ly$\alpha$}}

29: \newcommand{\resonetwo}{{$1s^{2}-1s2p$} }

30: \newcommand{\resonetwop}{{$1s^{2}-1s2p$}}

31: \newcommand{\bsax}{{\it BeppoSAX} }

32: \newcommand{\bsaxp}{{\it BeppoSAX}}

33: \newcommand{\fighistlc}{{Fig.~1} }

34: \newcommand{\fighistlcp}{{Fig.~1}}

35: \newcommand{\figgvsg}{{Fig.~2a} }

36: \newcommand{\figgvsgp}{{Fig.~2a}}

37: \newcommand{\figivsi}{{Fig.~2b} }

38: \newcommand{\figivsip}{{Fig.~2b}}

39: \newcommand{\figxteltcrv}{{Fig.~3} }

40: \newcommand{\figxteltcrvp}{{Fig.~3}}

41: \newcommand{\figgvsf}{{Fig.~4} }

42: \newcommand{\figgvsfp}{{Fig.~4}}

43: \newcommand{\figevsl}{{Fig.~5} }

44: \newcommand{\figevslp}{{Fig.~5}}

45: \newcommand{\figivsl}{{Fig.~6} }

46: \newcommand{\figivslp}{{Fig.~6}}

47: \newcommand{\figewvsl}{{Fig.~8} }

48: \newcommand{\figewvslp}{{Fig.~8}}

49: \newcommand{\figresultsvst}{{Fig.~7} }

50: \newcommand{\figresultsvstp}{{Fig.~7}}

51: \newcommand{\figgrpplrat}{{Fig.~9} }

52: \newcommand{\figgrpplratp}{{Fig.~9}}

53: \newcommand{\figgrprat}{{Fig.~10} }

54: \newcommand{\figgrpratp}{{Fig.~10}}

55: \newcommand{\figivsecont}{{Fig.~11} }

56: \newcommand{\figivsecontp}{{Fig.~11}}

57: \newcommand{\figibvsincont}{{Fig.~12} }

58: \newcommand{\figibvsincontp}{{Fig.~12}}

59: \newcommand{\figidvsrout}{{Fig.~13} }

60: \newcommand{\figidvsroutp}{{Fig.~13}}

61: \newcommand{\figktvstaucont}{{Fig.~14} }

62: \newcommand{\figktvstaucontp}{{Fig.~14}}

63: \newcommand{\figfeatewvssig}{{Fig.~14} }

64: \newcommand{\figfeatewvssigp}{{Fig.~14}}

65: \newcommand{\tableobslog}{Table~1 }

66: \newcommand{\tableobslogp}{Table~1}

67: \newcommand{\tableresultslog}{Table~2 }

68: \newcommand{\tableresultslogp}{Table~2}

69: \newcommand{\tablegrpresultslog}{Table~3 }

70: \newcommand{\tablegrpresultslogp}{Table~3}

71: \newcommand{\src}{NGC~2992 }

72: \newcommand{\srcp}{NGC~2992}

73: \newcommand{\sigfeka}{$\sigma_{\rm Fe \ K\alpha}$ }

74: \newcommand{\sigfekap}{$\sigma_{\rm Fe \ K\alpha}$}

75: \newcommand{\efeka}{$E_{\rm Fe \ K\alpha}$ }

76: \newcommand{\efekap}{$E_{\rm Fe \ K\alpha}$}

77: \newcommand{\ifeka}{$I_{\rm Fe \ K\alpha}$ }

78: \newcommand{\ifekap}{$I_{\rm Fe \ K\alpha}$}

79: 

80: \begin{document}

81: \title{Monitoring the Violent Activity from the Inner Accretion Disk of the Seyfert 1.9 Galaxy NGC~2992 with RXTE}

82: \author{Kendrah D. Murphy\altaffilmark{1}, Tahir Yaqoob\altaffilmark{1,2}, \& Yuichi

83: Terashima\altaffilmark{3} }

84: \altaffiltext{1}{Department of Physics and Astronomy,

85: Johns Hopkins University, Baltimore, MD 21218}

86: \altaffiltext{2}{Astrophysics Science Division,

87: NASA/Goddard Space Flight Center, Greenbelt, MD 20771}

88: \altaffiltext{3}{Department of Physics, Ehime University, Bunkyo-cho, Matsuyama, Ehime 790-8577,

89: Japan}

90: 

91: \begin{abstract} 

92: 

93: We present the results of a one year monitoring campaign of the Seyfert 1.9 galaxy NGC~2992

94: with {\it RXTE}.  Historically, the source has been shown to vary dramatically in 2--10 keV

95: flux over timescales of years and was thought to be slowly transitioning between periods of

96: quiescence and active accretion.  Our results show that in one year the source continuum flux

97: covered almost the entire historical range, making it unlikely that the low-luminosity states

98: correspond to the accretion mechanism switching off.  During flaring episodes we found that a

99: highly redshifted Fe~K line appears, implying that the violent activity is occurring in the

100: inner accretion disk, within $\sim100$ gravitational radii of the central black hole.  We also

101: found that the spectral index of the X-ray continuum remained approximately constant

102: during the large amplitude variability.  These observations make NGC~2992 well-suited for

103: future multi-waveband monitoring, as a test-bed for constraining accretion models.

104: 

105: \end{abstract}

106: 

107: \keywords{galaxies: active - galaxies: Seyfert - line: profiles - X-ray: galaxies - X-rays: individual

108: (NGC~2992)}

109: 

110: \section{INTRODUCTION}

111: \label{intro}

112: 

113: NGC~2992 is a relatively nearby ($z=0.00771$) Seyfert 1.9 galaxy.  Observations of  NGC~2992

114: have been made by every X-ray mission since its discovery by $HEAO-1$ in 1977 (Piccinotti

115: \etal 1982). \fighistlc shows the historical light curve (2--10 keV flux) for X-ray

116: observations over the past $\sim30$ years (Piccinotti \etal 1982, Mushotzky 1982, Turner \&

117: Pounds 1989, Turner \etal 1991, Nandra \& Pounds 1994, Weaver \etal 1996, Gilli \etal 2000,

118: Yaqoob \etal 2007).  In 1994, \asca observed NGC~2992 to have a very low X-ray continuum flux

119: compared with previous observations (Weaver \etal 1996), down by a factor of $\sim20$ since

120: it was first detected.  The \asca data revealed a very prominent, narrow \feka emission line,

121: believed to have originated in matter distant from the supermassive black hole (possibly in

122: the putative obscuring torus).  Later, X-ray observations with \bsax (Gilli \etal 2000)

123: showed a revival to a high-flux state.  This was accompanied by complex variability in the

124: \feka line profile, which appeared to have possible contributions from radiation originating

125: in both the accretion disk and distant matter.  Gilli \etal (2000) suggested that the

126: variation in continuum flux was evidence of quenching and revival of accretion on a timescale

127: of years.  In fact, they believed that \bsax witnessed a revival of the AGN between its

128: two observations, which were separated by about a year in the period 1997--1998.  Not shown

129: in \fighistlc is a third observation made by \bsax during its last week of operation in

130: April 2002, from which only PDS data were taken (see Beckmann \etal 2007). 

131: Hard X-ray data taken by {\it INTEGRAL} (in 2005 May) and {\it Swift} (in 2005--2006) are also

132: presented by Beckmann \etal (2007). 

133: 

134: We obtained observing time for a one-year monitoring campaign of NGC~2992 with the Rossi X-ray

135: Timing Explorer ({\it RXTE}) beginning in 2005.  NGC~2992 had never been observed with \rxte

136: prior to this campaign.  The goal was to test the response of the \feka line to the variation

137: in the X-ray continuum.  Spectral analysis of these data revealed an unexpected result: the

138: variation in flux for the twenty-four \rxte observations covered nearly the entire dynamical

139: range in flux in the historical data in less than a year.  Previously it was thought that the

140: variation occurred over decades, but the \rxte data showed flares in the X-ray continuum flux

141: on timescales of days.  Additionally, the \feka emission line (that is present in the majority

142: of the observations and centered at $\sim6.4$ keV in most cases) was redshifted and broadened

143: during high-flux periods, implying that we may have been witnessing short-term flaring

144: activity (on the order of days to weeks) from the inner regions of the accretion disk.

145: 

146: In \S\ref{obs}, we describe the \rxte data and our data reduction process.  We describe the

147: results of our spectral analysis of the data in \S\ref{fitting}, including a discussion of

148: the variability of the \feka emission line and the physical implications.  In

149: \S\ref{conclusions} we summarize the conclusions of our analysis.  An Appendix is included to

150: discuss a calibration feature found in the data in the 8--9 keV range.

151: 

152: \section{OBSERVATIONS \& DATA REDUCTION}

153: \label{obs}

154: 

155: NGC~2992 was observed between 2005 March 4 and 2006 January 28 with the \rxte Proportional

156: Counter Array (PCA; Jahoda \etal 2006). We obtained data from a total of twenty-four

157: observations. The observation log is given in \tableobslogp; hereafter we will refer to each

158: observation as obs~1 to 24 as listed in \tableobslogp.  It can be seen that the interval

159: between observations ranged from $\sim 3$ to 33 days.   

160: 

161: The PCA is an array of five Proportional Counter Units (PCUs), each with a net geometric

162: collecting area of $\sim1600$ cm$^{2}$ (Jahoda \etal 2006).  Each PCU consists of 3 layers of a

163: mixture of Xenon (90\%) and Methane (10\%) gas and a $1^{\circ}$ collimator (FWHM).  For

164: energies less than 10 keV, $\sim$ 90\% of the cosmic photons and 50\% of the internal

165: instrumental background are detected in the top layer (layer 1) of the Xenon/Methane detector. 

166: Therefore, in order to maximize the signal-to-noise ratio, only data from layer 1 of the PCUs

167: were used.

168: 

169: Fewer of the PCUs were in operation later in the mission, so we did not obtain data from all

170: five units.  For two observations (obs~2 and obs~3), PCUs 0, 1, and 2 were operational;

171: however, only  PCU~0 and PCU~2 were operational for {\it all} twenty-four of the NGC~2992

172: observations.  In order to perform a uniform analysis of all of the observations, we

173: discarded the PCU~1 data for obs~2 and obs~3, and only used data from PCU~0 and PCU~2.

174: 

175: In the spring of 2000 (May 12), the propane layer in PCU~0 lost pressure and thus the

176: calibration and background subtraction model were adversely affected (Jahoda \etal 2006). 

177: Since then, the calibration and background estimation have improved, so we retained the PCU~0

178: data, but performed initial analyses on PCU~0 and PCU~2 separately, in order to check for

179: inconsistencies in the two detectors.  We found the variation in the results from spectral

180: fits to the data from the two detectors to be statistically insignificant.  As we will

181: describe in \S\ref{fitting}, we analyzed the spectra from the combined PCU~0 plus PCU~2 data

182: in addition to the individual PCU data in order to obtain better statistics on the important

183: model  parameters. 

184: 

185: The data were selected to exclude time intervals when the Earth's elevation angle was less than

186: $10^{\circ}$, during passage of the satellite through the South Atlantic Anomaly (SAA), and

187: during times of high electron contamination (ie, the housekeeping parameter `ELECTRON2' was

188: selected to be greater than 0.1).  The background was estimated with the program PCABACKEST,

189: using the mission-long `Faint' model file pca\_bkgd\_cmfaintl7\_eMv20031123.mdl (Jahoda \etal

190: 2006).  Spectral response matrices were made for each of the observations for both PCU~0 and

191: PCU~2 using pcarmf version 10.1 and the channel-to-energy calibration file

192: pca\_e2c\_e05v03.fits.

193: 

194: We extracted spectra and light curves for PCU~0 and PCU~2 for each of the twenty-four

195: observations.  In addition, we co-added the PCU~0 and PCU~2 data to make twenty-four

196: combined (PCU~0~+~PCU~2) spectra.  In obs~7 and obs~15, we found flares in the PCU~0 light

197: curves, so we applied further data selection with time cuts to remove those periods from

198: the PCU~0 data.  These selection criteria resulted in net exposure times for the

199: PCU~0~+~PCU~2 data in the range of $\sim1.5$ to $\sim6.5$ ks, as shown in \tableobslogp. 

200: The grand total integration time for the twenty-four observations, summed over

201: PCU~0~+~PCU~2, was 133.088 ks.  The mean, full-band, background-subtracted count rates for

202: the combined PCU~0~+~PCU~2 data varied by over an order  of magnitude, ranging from

203: $\sim0.8$ to $\sim10.6$ cts/s (see \tableobslogp).  

204: 

205: \section{SPECTRAL FITTING RESULTS}

206: \label{fitting}

207:  

208: We modeled the PCA spectra using XSPEC version 11.3.2 (Arnaud 1996).  Preliminary

209: examination of the spectra revealed poor background subtraction above $\sim15$ keV.  Since

210: the PCA data below $\sim3$ keV are not well calibrated (Jahoda \etal 2006), we performed

211: spectral fitting in the 3--15 keV range.  We used $\chi^{2}$ as the fit statistic.  In all

212: of the model fitting the Galactic column was fixed density at $N_{\rm H} = 5.26 \times

213: 10^{20} \rm \ cm^{-2}$ (Dickey \& Lockman 1990).  Although the 3--15 keV energy range is

214: not sensitive to Galactic absorption, we included it for consistency with models in the

215: literature that were fitted to lower energy data.  All model parameters will be referred to

216: in the source frame.  

217: 

218: \subsection{Individual Observation Fits} 

219: \label{indfits} 

220: 

221: We fitted a simple power-law  model to each of the spectra for the twenty-four observations,

222: initially for the PCU~0 and PCU~2 data separately and subsequently for the combined PCU

223: data.  This model consisted of two free parameters, namely the photon index ($\Gamma$) and

224: the power-law normalization.  We found, in most cases, that the plots of the data/model

225: ratios of the twenty-four combined PCU spectra showed residuals that may correspond to the

226: \feka emission line known to be present in the source (e.g. Weaver \etal 1996, Gilli \etal

227: 2000).  Each case was confirmed by examining the separate PCU~0 and PCU~2 fits.

228: 

229: We therefore proceeded to fit a power-law plus a Gaussian emission-line component to the

230: three sets of twenty-four spectra.  The model consisted of five free parameters: the

231: power-law normalization, $\Gamma$, the centroid energy of the line (\efekap) in keV, the

232: intrinsic line width (\sigfekap) in keV, and the intensity of the line (\ifekap) in $\rm

233: photons \ cm^{-2} \ s^{-1}$.  Although for eight of the twenty-four fits the probability of

234: obtaining values of $\chi^{2}$ as high as those measured from the data by chance was less

235: than 10\%, we note that our data do not include compensation for systematic errors so the

236: goodness of fits cannot be assessed by consideration of the values of $\chi^{2}$ alone.  We

237: will return to this question when discussing fits to grouped data sets (\S\ref{groupfit}). 

238: In cases where the line energy could not be constrained, the fit was repeated with the Fe

239: K$\alpha$ energy fixed at the neutral Fe value of 6.4 keV.  In obs~22--24, where the width of

240: the line could not be constrained (the line was too weak), we fixed the intrinsic Gaussian

241: line width at \sigfeka = 0.05 keV in order to obtain statistical errors on the line

242: intensity.  This width is much less than the PCA energy resolution ($\Delta E \sim 1.02$ keV

243: FWHM at 6 keV, Jahoda \etal 2006).  For seventeen of the twenty-one remaining data sets, the

244: decrease in $\chi^{2}$ ranged from 6.3 to 22.2 for the addition of three free parameters for

245: the line, compared to the continuum-only fits.  This corresponds to detections of the line at

246: a confidence level greater than 90\% ($\Delta\chi^{2}=6.251$ for three parameters).  Three of

247: the remaining four observations (obs~2, 11, and 12) had detections of the line at greater

248: than 68\% confidence ($\Delta\chi^{2}=3.506$ for three parameters) and one (obs~15) had a

249: detection at only marginally less than 68\% confidence ($\Delta\chi^{2}=3.4$).

250: 

251: We compared the values of $\Gamma$, \efekap, \sigfekap, and \ifeka that were obtained from

252: spectral fits to the PCU~0 data with those obtained from the PCU~2 data and found no

253: evidence of systematic differences.  For example, \figgvsg and \figivsi show plots of

254: $\Gamma$(PCU~0) vs. $\Gamma$(PCU~2) and \ifeka (PCU~0) vs. \ifeka (PCU~2) respectively.  Also,

255: the ratios of the data to the model for PCU~0 and PCU~2 showed no systematic anomalies within

256: the statistical errors.  Therefore, hereafter we refer only to the results obtained from the

257: combined PCU~0~+~PCU~2 spectral fits.  

258: 

259: The power-law plus Gaussian fit results for the combined PCU data are shown in

260: \tableresultslogp.  Statistical errors for each parameter in \tableresultslog are 68\%

261: confidence for $n$ interesting parameters, where $n$ = 2 or 3 (corresponding to

262: $\Delta\chi^{2}$ = 2.279 and 3.506 respectively), depending on the number of

263: parameters that were fixed in order to obtain stable fits during the error analysis.  The

264: power-law normalization is not included as an interesting parameter.  Quoting the 68\%

265: confidence ($1\sigma$) errors facilitates statistical analysis of the results.  We also quote

266: (in brackets, below the best-fit values) 90\% confidence ranges for one interesting parameter

267: ($\Delta\chi^{2}$ = 2.706), for comparison with other results in the literature.  Parameters

268: that were fixed are labeled with `$f$' in \tableresultslogp.

269: 

270: For twenty-one of the data sets, statistical errors were found for \sigfeka by fixing the line

271: energy at the best-fit value and statistical errors were found for \efekap, \ifekap, and

272: $\Gamma$ by fixing the line width at the best-fit value.  Therefore, for these cases, 68\%

273: confidence errors were found using $\Delta\chi^{2} = 3.506$ ($n=3$) for all interesting

274: parameters.  The fits became highly unstable if we allowed both \sigfeka and \efeka to be free

275: during the error analysis.  A stable fit for the line was not obtained for obs~22, 23, and 24

276: since the line was not detected with sufficient statistical significance in these

277: observations.  In these cases, it was necessary to fix both the line energy at 6.4 keV and the

278: line width at 0.05 keV to obtain statistical errors ($n=2$) for the remaining free

279: parameters (\ifeka and $\Gamma$).   

280: 

281: \subsection{Light Curve}

282: \label{ltcrv}

283: 

284: In order to facilitate comparison with the literature, we calculated the 2--10 keV continuum

285: flux and luminosity values by extrapolating the 3--15 keV models down to 2 keV. The 2--10

286: keV fluxes and luminosities are presented in \tableresultslogp.  All fluxes\footnote[1]{Note:

287: \rxte absolute fluxes are systematically higher than \ascap, \bsaxp, \chandrap, and \xmm by

288: $\sim10-20$\% due to the particular spectrum and normalization adopted for the Crab Nebula for

289: calibration of the PCA (Jahoda \etal 2006).} are observed-frame values, not corrected for

290: absorption.  All luminosities are rest-frame values, corrected for absorption.   

291: 

292: The $\sim1$ year light curve (\figxteltcrv) ranges from $\sim0.8 \times 10^{-11} \rm \ to

293: \sim8.9 \times 10^{-11} \ ergs \ cm^{-2} \ s^{-1}$  in 2--10 keV flux.  The 2--10 keV

294: luminosity ranges from $0.11 \times 10^{43}$ to $\sim1.17 \times 10^{43}$ ergs $\rm s^{-1}$

295: (assuming $H_{o}=70$ km s$^{-1}$ Mpc$^{-1}$, $\Lambda=0.73$, $\Omega=1$), while the fraction

296: of the Eddington luminosity ($L/L_{\rm edd}$) ranges from $\sim1.5 \times 10^{-3}$ to $\sim1.8

297: \times 10^{-2}$, assuming a black hole mass of $5.2 \times 10^{7} M_{\odot}$ (Woo \& Urry

298: 2002).  Historically, $L/L_{\rm edd}$ has ranged from $\sim0.8 \times 10^{-3}$ to $\sim 1.9

299: \times 10^{-2}$.  The first \rxte observation, which had the highest flux, was followed by two

300: additional high-flux peaks (obs~4 and obs~8).  The flux then dropped to $\sim1\times 10^{-11}

301: \rm \ ergs \ cm^{-2} \ s^{-1}$ during obs~9 to obs~12.  This was followed by a slight rise in

302: flux for obs~13 and obs~14.  The flux measurements in the remaining observations stayed in the

303: $1-2 \times 10^{-11} \ \rm ergs \ cm^{-2} \ s^{-1}$ range, with the last three appearing to

304: make an upward turn. 

305: 

306: We quantified the variability in the \rxte light curve by calculating the fractional

307: variability amplitude, $F_{\rm var}$, comparing this measure of variability with results

308: obtained from \rxte monitoring of other AGN (e.g. see Vaughan \etal 2003; Markowitz \etal 2003;

309: Markowitz \& Edelson 2004). The quantity $F_{\rm var}$ is simply the square root of the excess

310: variance, that is also used to characterize AGN variability, taking into account Poisson noise

311: (e.g. see Turner \etal 1999). Markowitz \& Edelson (2004) calculated $F_{\rm var}$ uniformly

312: for a sample of AGN, based on 2--12~keV \rxte PCA count rates, and several monitoring

313: durations, one of which was 216 days. Using the same energy band for NGC~2992, we obtained 

314: $F_{\rm var} = (91.5 \pm 0.5)\%, (84.0 \pm 0.5)\%$, and  $(41.6 \pm 0.8)\%$ for all 24

315: observations, the first 16 observations (207 days duration), and the last 14 observations (211

316: days duration), respectively. Our observations of NGC~2992 were spaced roughly two weeks apart

317: except for two pairs of observations for which each pair had a separation of only three days.

318: The count rates from these pairs of observations were averaged in order to obtain roughly equal

319: spacing in time intervals before calculating $F_{\rm var}$. Markowitz \& Edelson (2004) found

320: that the majority of the AGN in their sample had 216-day values of $F_{\rm var}$ lying the

321: range $\sim (1-40)\%$, and the two sources with the highest values of $F_{\rm var}$ of $\sim

322: 60\%$ were NGC~3227 and NGC~4051 (both of these have  X-ray luminosities that are similar to

323: the luminosity of NGC~2992). Thus the high value of $F_{\rm var}$ obtained for the first part

324: of the \rxte NGC~2992 campaign (and indeed for the whole campaign) is exceptionally high.

325: Moreover, the value of $F_{\rm var}$ obtained for the second part of the campaign (in which

326: NGC~2992 was much ``quieter'') is still at the high end of the range that was obtained by

327: Markowitz \& Edelson (2004) for their sample of AGN.

328: 

329: We can compare the timescales probed by our \rxte campaign with the expected break frequency,

330: $\nu_{b}$, given the black-hole mass, based on a correlation observed from studying samples of

331: Seyfert galaxies and other black-hole systems (e.g. Papadakis 2004 finds $\nu_{b} \sim

332: 15/[M_{\rm BH}/M_{\odot}]$~Hz; see also McHardy \etal 2007 and references therein).  This break

333: frequency marks the division where the power spectrum flattens from $\sim \nu^{-2}$ to $\sim

334: \nu^{-1}$ towards lower frequencies. For NGC~2992 the predicted timescale corresponding to the

335: break frequency is $\sim 40$ days. On the other hand, if we use the relation between the break

336: timescale and luminosity that McHardy \etal (2007) empirically from studying a sample of AGN,

337: we obatined $\sim 1-2$ days. Clearly, more frequent sampling than that achieved in our

338: monitoring of  NGC~2992 is required in order to derive reliable, direct, quantitative

339: constraints on the  power-spectrum amplitude and break frequency (or timescale).

340: 

341: \subsection{Continuum \& Fe~K Emission Line}

342: \label{cont}

343: 

344: The photon index, $\Gamma$, ranged from $\sim1.61 \rm \ to \sim2.25$.  A plot of $\Gamma$

345: versus 2--10 keV flux is given in \figgvsfp.  It is interesting to note that, despite the

346: significant variation in flux, $\Gamma$ is consistent with a constant value within the

347: statistical errors.  The weighted average of $\Gamma$ is $1.732 \pm 0.022$.   This behavior

348: constrasts with some other Seyfert galaxies in which $\Gamma$ increases with continuum

349: luminosity. Utilizing the 68\% confidence statistical errors, we show in \figevsl the best-fit

350: line energies plotted against the 2--10 keV luminosity from each observation, except for the

351: observations in which \efeka had to be fixed (represented by open circles on the plot).  With

352: the exception of observations that were made when NGC~2992 was in a high-flux state, the line

353: energies are consistent with 6.4 keV, confirming the presence of an \feka emission line from

354: cool matter in the source.  In the high-flux spectra, the centroid energy of the line was found

355: to be $\sim5.6$ keV.  As will be discussed later, this may correspond to a broad component in

356: the \feka line complex due to redshifted emission resulting from enhanced illumination of the

357: line-emitting material in close proximity to the central black hole.

358: 

359: The line width, \sigfekap, was a free parameter in twenty-one of the twenty-four combined

360: PCU~0 plus PCU~2 fits.  For these observations, the best-fit value for \sigfeka ranged from

361: 0.0 to 1.1 keV, often with large statistical errors.  A plot of \feka line intensity

362: (\ifeka) vs. 2--10 keV luminosity is given in \figivslp.   Although the 2--10 keV

363: luminosity varies by a factor of $\sim11$, we see from \figivsl that the line intensity

364: variability amplitude appears to be less than that of the continuum.  While the 68\%

365: confidence errors of the line intensity covered a range of $0.0 \ \rm to \ 46.2 \times

366: 10^{-5} \ \rm photons \ cm^{-2} \ s^{-1}$, the weighted mean of \ifeka was $\sim6.6 \pm 1.0

367: \times 10^{-5} \ \rm photons \ cm^{-2} \ s^{-1}$.  The results imply a constant line

368: intensity or significant time delays between the response of the line emission to continuum

369: variations.  For comparison of the emission line and continuum variability, we show in

370: \figresultsvst the results of $L/L_{\rm Eedd}$, $\Gamma$, \efekap, \ifeka for the

371: twenty-four observations versus time.  As shown in \figewvslp, there is no clear trend in

372: the variation of the \feka line EW with 2--10 keV continuum luminosity.  Although there is

373: larger variation at lower luminosities, the uncertainty is also greater.  At higher

374: luminosities, the EW appears to tend towards values less than 500 eV.

375:  

376: We can compare our measured values of the \feka line intensity with the theoretical

377: expectation in the limit of a Thomson-thin spherically-symmetric distribution of gas

378: surrounding a central X-ray source.  In the limit of low redshift, the theoretical value of

379: \ifeka is 

380: \begin{eqnarray}

381: \label{eq:ifeka}

382: \nonumber I_{\rm Fe~K\alpha} & = & 7.44\times10^{-5}

383: \ h_{70}^{2}

384: \left(\frac{L_{2-10}}{10^{43} \ \rm ergs \ s^{-1}}\right) 

385: \left(\frac{<L>}{L}\right)

386: \left(\frac{N_{\rm H}}{10^{23} \ \rm cm^{-2}}\right)

387: \left(\frac{\Delta\Omega}{4\pi}\right)

388: \left(\frac{\omega_{\rm K}}{0.34}\right)

389: \left(\frac{A_{\rm Fe}}{4.68\times10^{-5}}\right)

390: \\ & & \left(\frac{3.4}{\Gamma+1.646}\right)

391: (7.11^{1.7-\Gamma})

392: \left(\frac{2-\Gamma}{0.3}\right)

393: \left(\frac{0.764}{10^{2-\Gamma}-2^{2-\Gamma}}\right)

394: \left(\frac{z}{0.00771}\right)^{-2}

395: \rm photons \ cm^{-2} \ s^{-1} \hspace{1.5cm} 

396: \end{eqnarray}

397: (see for example Yaqoob \etal 2001), where $L_{2-10}$ is the 2--10 keV luminosity  and

398: $(\Delta\Omega/4\pi)$ is the fraction of the sky covered by the line-emitting region, as

399: seen by the source.  Here $L$ is the measured luminosity and $<L>$ is representative of the

400: historically averaged luminosity on timescales greater than the light-crossing time of the

401: line-emitting region.  The factor $h_{70}$ is defined as $\left(\frac{H_{0}}{70 \ \rm km \

402: s^{-1} \ Mpc^{-1}}\right)$, where $H_{0}$ is the Hubble constant.  $N_{\rm H}$ is the

403: column density of the shell.  The value of $A_{\rm Fe}$ is the iron abundance in the emitting

404: material (relative to hydrogen) and $\omega_{\rm K}$ is the fluorescence yield.  All of the

405: factors in parentheses involving the photon index in equation \ref{eq:ifeka} evaluate to

406: $\sim1$ for $\Gamma=1.7$, the approximate mean value found for NGC~2992 from the \rxte

407: observations. Note that equation~(1) does not represent any assumptions

408: about NGC~2992. It is simply a theoretical limit with which results from NGC~2992

409: can be compared. From this comparison one can deduce constraints on the

410: physical parameters that must apply to NGC~2992.

411: 

412: Overlaid on \figivsl are four theoretical \ifeka versus luminosity curves corresponding to

413: values of $\left(\frac{<L>}{L}\right)\left(\frac{N_{\rm H}}{10^{23} \ \rm cm^{-2}}\right)$

414: equal to $0.1,\ 1,\ 5$ and $10$.  The curves were calculated using equation \ref{eq:ifeka} with

415: $\frac{\Delta\Omega}{4\pi}=1$ and $\Gamma=1.732$, the weighted mean value of the photon

416: index.  We took $A_{\rm Fe}$ to be $4.68\times10^{-5}$, the solar value given in Anders \&

417: Grevesse (1989), and we used the value $\omega_{\rm K}=0.34$ as given in, for example, Palmeri

418: \etal (2003).  The typical line-of-sight $N_{\rm H}$ that has been measured by Suzaku and

419: other missions is $\sim10^{22} \ \rm cm^{-2}$.  For this value, $\frac{<L>}{L}=1$ corresponds

420: to the shallowest theoretical line shown on \figivsl (labeled with `0.1'), which is not a good

421: representation of these data.  Better agreement is given by a larger value of 

422: $\left(\frac{<L>}{L}\right)\left(\frac{N_{\rm H}}{10^{23} \ \rm cm^{-2}}\right)$, implying

423: that the line-of-sight $N_{\rm H}$ is not representative of the entire system and/or the

424: observed variation in the line intensity is affected by time delays following continuum

425: variations.  Values of $\left(\frac{<L>}{L}\right)\left(\frac{N_{\rm H}}{10^{23} \ \rm

426: cm^{-2}}\right) \sim 5-10$ are more likely. Weaver \etal (1996) deduced time delays 

427: of the order of years between the X-ray continuum and distant-matter Fe~K emission line.

428: 

429: For the other extreme, a Compton-thick reprocessor, the Fe~K line equivalent width (EW)

430: depends on the reprocessor geometry and its orientation with respect to the observer. However,

431: we can identify two very general scenarios regardless of the details of the geometry. The two

432: cases correspond to whether or not the structure's orientation and/or geometry is such that it

433: intercepts the line-of-sight between the X-ray continuum source (that illuminates the

434: reprocessor to produce the Fe~K emission line) and the observer. If the X-ray continuum is

435: obscured, then the Fe~K line EW can be in the range of hundreds to thousands of eV, but if it

436: is not (in which case the Fe~K line is observed in `reflection' - i.e. from the same surface

437: that is illuminated by the continuum) the EW is typically not more than $\sim 200$~eV. 

438: Examples of Monte Carlo simulations of such scenarios can be found in Ghisellini \etal (1994)

439: and references therein. What is apparent from these calculations is that for a given geometry

440: and orientation, the EW of the Fe~K line, as a function of the value of the highest column

441: density through the reprocessor, reaches an asymptotic value once the structure becomes

442: Compton-thick. Therefore, for the purpose of simple estimates and comparison with the

443: Compton-thin case, one can parameterize all of the Compton-thick models by the asymptotic

444: value of the Fe~K line EW {\it for the time-steady, static limit for a constant X-ray

445: continuum}. We can then express the intensity of the Fe~K line in terms of the 2--10~keV

446: luminosity of a steady-state illuminating X-ray continuum and directly compare that with the

447: Compton-thin scenario. For the Compton-thick case we get

448: \begin{eqnarray}

449: \label{eq:ithick}

450: \nonumber I_{\rm Fe~K\alpha} & = & 8.00\times10^{-5}

451: \ h_{70}^{2}

452: \left(\frac{L_{2-10}}{10^{43} \ \rm ergs \ s^{-1}}\right)

453: \left(\frac{<L>}{L}\right)

454: \left(\frac{z}{0.00771}\right)^{-2}

455: \left(\frac{2-\Gamma}{0.3}\right)

456:  \\ & &

457: \left(\frac{0.764}{10^{2-\Gamma}-2^{2-\Gamma}}\right)

458: \left(\frac{6.4 \ \rm keV}{E_{\rm 0}}\right)^{\Gamma}

459: \left(6.4\right)^{1.7-\Gamma}

460: \left(\frac{\rm EW}{100 \ \rm eV}\right)

461: \rm photons \ cm^{-2} \ s^{-1}

462: \end{eqnarray}

463: where $E_{\rm 0}$ is the rest frame energy of the Fe~K line photons in keV.  All of the

464: factors in parentheses involving the photon index in equation \ref{eq:ithick} evaluate to

465: $\sim 1$ for $\Gamma=1.7$.

466: 

467: Lines of \feka line intensity versus 2--10~keV luminosity are overlaid on \figivsl for several

468: values of the asymptotic EW.  This asymptotic EW parameter bundles the unknown information

469: about the geometry and orientation of the reprocessor into a single number and is degenerate

470: with the factor $\frac{<L>}{L}$ that represents the ratio between the the historical

471: luminosity, averaged on timescales longer than the reprocessor light-crossing time, and the

472: observed continuum luminosity.  We used the weighted mean value of $\Gamma=1.732$ and assumed

473: that the line arises from neutral Fe. Using equation \ref{eq:ithick}, we calculated

474: theoretical curves for $\left(\frac{<L>}{L}\right)$EW=100, 250, 500, 750, and 1000 eV.  It can

475: be seen in \figivsl that the \rxte data points cover the range of theoretical values of the EW

476: from 100 to 1000 eV for $\frac{<L>}{L}=1$, and therefore the data do not rule out either

477: extreme of reflection or transmission through a Compton-thick reprocessor.  However, the case

478: of transmission (EW=1000 eV) appears to be unlikely since this would require the historically

479: averaged luminosity ($<L>$) to be lower than than the \rxte luminosities in order to fit the

480: data.  Since the \rxte data covers nearly the entire historical range in luminosity, we would

481: expect $<L>$ to be larger than the lowest \rxte luminosity values.

482: 

483: The narrow \feka line found by Suzaku had an intensity of $2.49^{+0.71}_{-0.40}

484: \times10^{-5} \ \rm photons \ cm^{-2} \ s^{-1}$ (Yaqoob \etal 2007), which is generally

485: smaller than the values obtained by \rxte.  The excess measured by \rxte could be due to part

486: of the flux from an underlying broad line, which cannot be decoupled from the narrow line due

487: to the poor PCA resolution.

488: 

489: \subsection{Grouped Data} 

490: \label{groupfit} 

491: 

492: In order to better constrain the key model parameters and investigate variability in more

493: detail, we combined groups of the twenty-four spectra into six data sets.  The groups are

494: identified on \figxteltcrvp.  Group 1 (filled circles) consists of the three high-flux

495: observations that were completed in the earlier part of 2005 (obs~1, 4, 8).   The

496: observations in an intermediate-flux state were combined in group 2 (stars; obs~2, 3, 6, 7,

497: 13, 14).   The remaining, low-flux observations were broken up chronologically to create

498: group 3 (squares; obs~5,  9, 10, 11, 12), group 4 (triangles; obs~15, 16, 17), group 5

499: (diamonds; obs~18, 19, 20, 21), and  group 6 (open circles; obs~22, 23, 24).  As before, we

500: fitted a simple power-law to the $3-15$ keV data.  Plots of the ratios of the model to the

501: grouped data sets are given in \figgrpplratp.  We used different y-axis scales on these

502: plots in order to clearly display the spectral features.  We found, again, clear residuals

503: at $\sim6.4$ keV for each of groups 2, 3, 4, and 5.  In group 1 the residuals appeared to

504: peak at an energy lower than 6.4 keV, while in group 6 no line-like residuals were

505: evident.  We fitted the data again, adding a Gaussian component to model \feka line

506: emission.  \tablegrpresultslog gives the results of these fits and the plots of the ratios

507: of the data to this model are shown in \figgrpratp.  The results were consistent with those

508: from the individual observations.  The addition of a Gaussian component to the model gave a

509: better fit for groups 1 through 5, with $\chi^{2}$  decreasing by at least 19.9 for the

510: addition of three free parameters.  However, the absolute $\chi^{2}$ values remained high. 

511: Although \tablegrpresultslog shows that the probability of obtaining these values by chance

512: is very small, the $\chi^{2}$ values do not take into account systematic errors in the data

513: and therefore are not an adequate assessment of the goodness of the model fits.  The \feka

514: line was not significantly detected in group 6.  However, we were able to derive a line

515: intensity with a non-zero lower limit (at 68\% confidence) by fixing the line centroid

516: energy and width (\sigfekap) at 6.4 keV and 0.05 keV respectively (see

517: \tablegrpresultslogp).

518: 

519: The 2--10 keV flux ranged from $\sim1.1 \times 10^{-11}$ to $\sim 7.4 \times 10^{-11} \rm \

520: ergs \ cm^{-2} \ s^{-1}$.  The photon index ranged from $\sim 1.71$ to $\sim 1.96$, with a

521: weighted mean of $\sim1.733 \pm 0.022$.  The emission line energy appeared to be consistent

522: with the neutral Fe value of 6.4 keV except for group 1, the high-flux group, where the

523: centroid peaked at $\sim 5.6$ keV.  Excluding group 6, the best-fit value of \sigfeka

524: ranged from 0.00 to 0.89 and the best-fit EW of the line ranged from $\sim200$ to $\sim

525: 700$ eV.  The weighted mean value of \ifeka for all six groups was $\sim(6.6 \pm 1.0)

526: \times 10^{-5} \rm \ photons \ cm^{-2} \ s^{-1}$ and the weighted mean value for groups 2

527: through 5, where the 6.4 keV Fe~K line dominated, was $\sim(7.5 \pm 1.2) \times 10^{-5}

528: \rm \ photons \ cm^{-2} \ s^{-1}$.

529: 

530: We directly compared the results from the grouped data fits with the results from the

531: individual-observation fits by over-plotting the group values on the figures discussed in

532: \S\ref{indfits}.  The grouped data points are marked with crosses in \figgvsfp, \figevslp,

533: \figivslp, and \figewvslp.  Like $\Gamma$, the EW of the \feka line did not appear to vary

534: significantly between the groups (corresponding to a weighted mean value of $310\pm48$ eV

535: for groups 1 through 5), although the EW values are consistent with the trend of being

536: smaller for higher continuum luminosities.  However, within the 68\% confidence errors, the

537: intensity of the \feka line appears to be higher during the high-luminosity observations,

538: as shown in \figivslp.  In addition, \figevsl shows the distinct difference in the energy

539: of the \feka line between the high-luminosity grouped data and the lower-luminosity grouped

540: data.  We investigate this further in \S\ref{highlow}.  

541: 

542: In \figgrprat we see that there sometimes appears to be a statistically significant `dip'

543: in the spectrum between $\sim 8$ and 9 keV. Even though the residuals in this region are

544: sometimes as large as those in the Fe K band (when compared to a simple power-law), we

545: believe that the 8--9 keV dip is an artifact of the background subtraction model and we

546: are confident that the Fe~K emission-line measurements are reliable because they are in

547: large part consistent with our knowledge from observations and measurements with other

548: missions. Our reasons for believing that the 8--9 keV dip is an artifact are detailed in

549: the Appendix. Since such an artifact in the spectrum can adversely affect fitted values of

550: the model parameters when a Compton-reflection continuum is included in the model, we

551: omitted three spectral channels (in the range of 7.7--9.5 keV) in subsequent spectral

552: fitting to remove the dip.

553: 

554: \subsubsection{Compton Reflection}

555: \label{reflection}

556: 

557: The effect of Compton reflection is important to consider when modeling spectra of AGNs (e.g.

558: Reynolds \& Nowak 2003 and references therein).  Although the \rxte spectral band used here is

559: not very sensitive to a Compton-reflection continuum, we added a reflection component to our

560: model (hrefl in XSPEC; see Dov{\v c}iak, Karas, \& Yaqoob 2004) in order to ensure that our

561: results are robust.  The resulting model has six free parameters including the so-called

562: reflection fraction, $R$, in addition to the power-law normalization, $\Gamma$, \efekap,

563: \sigfekap, and \ifekap.  $R$ is the normalization of the reflection continuum relative to that

564: expected from a steady-state, X-ray illuminated, neutral disk subtending a solid angle of

565: $2\pi$ at the X-ray source.  The Compton-reflection model used here assumes a centrally

566: illuminated, neutral, infinite disk with the abundance of Fe fixed at the solar value from

567: Anders \& Grevesse (1989).  We investigated the variation in \ifeka for values of $R$ between 0

568: to 3 by creating confidence contours of \ifeka versus $R$.  We found in general that the best

569: fit value of $R$ was consistent with 0 and that it did not significantly impact the value of

570: \ifekap.  The 99\% confidence contours were open and flat over the range in $R$ considered but

571: the spectral ratios to a simple power-law model indicated we should have obtained an upper

572: limit in the searched range of $0 \le R \le 3$, or at least that the contours should have begun

573: to close. Unfortunately, what is happening is that the limited energy resolution, bandpass and

574: signal-to-noise ratio of the data results in steep intrinsic continuua with strong reflection

575: degenerate with intrinsically flatter continuua with little reflection. Thus, upper limits on

576: $R$ for these data are not necessarily physically meaningful. However, using \bsax data with

577: PDS coverage out to beyond 100~keV, Gilli \etal (2000) found that Compton reflection was in

578: fact weak in NGC~2992 (see also Beckmann \etal 2007).  Our main concern here is that the Fe~K

579: emission line intensities are not sensitive to assumptions about the Compton reflection

580: continuum and with the above considerations in mind, we did not include Compton reflection in

581: most of the spectral fits described below (and we will note those for which it is included).

582: When we do include a Compton reflection continuum in the model, it is only a single component,

583: although two components might be expected. One component might be associated with the Fe~K line

584: from an accretion disk, and the other from distant matter (if it is Compton-thick). However the

585: quality of the data and the restricted bandpass does not warrant modeling multiple

586: Compton-reflection components so the value of $R$ then represents the relative amplitude of all

587: reflection components that might be present in the data.

588: 

589: \subsubsection{Variability of the Broad Component of the Fe~K Line}

590: \label{highlow}

591: 

592: In order to explore the apparent difference between the high- and low-flux states further,

593: we combined all of the spectra from the observations which made up groups 3, 4, and 5

594: (low-flux) to compare with the group 1 (high-flux) spectrum. We did not include group 6 in

595: the low-flux spectrum since the \feka line was not significantly detected in these

596: observations.  

597: 

598: Confidence contour plots of \ifeka versus \efeka for the high- and low-flux groups are

599: shown in \figivsecontp.  The model used included Galactic absorption, a power-law

600: continuum, and a Gaussian line.  Interestingly, the contours for the two sets of data are

601: mutually exclusive at 99\% confidence.  As can be seen in \figivslp, we did not detect the

602: same order of magnitude increase in intensity of the \feka line as in the 2--10 keV

603: continuum luminosity.  We do, however, see a broadening as well as a significant shift in

604: the centroid energy of the detected \feka line from $\sim6.3$ keV in the low state to

605: $\sim5.6$ keV in the high state.  It is possible that we are seeing two different

606: components of the \feka line complex: a component (with a centroid energy near 5.6 keV),

607: originating in the accretion disk close to the black hole, that is broadened and

608: redshifted due to Doppler and gravitational effects, and a narrow component (with a

609: centroid energy closer to 6.4 keV) from more distant material.  The increase in the

610: redshifted line flux in the high state may be due to a localized flare close to the black

611: hole.  

612: 

613: We investigated the extent to which the redshifted, broad Fe~K emission component and the

614: narrow, distant-matter Fe~K line may be present in both the low- and high-flux states by

615: fitting both high- and low-flux groups with a second Gaussian added to the previous power law

616: plus Gaussian line model.  We fixed the centroid energy of one Gaussian at 6.3 keV, the

617: best-fit value for the low-flux group in the previous, single Gaussian fit and we fixed the

618: other at 5.6 keV, the corresponding best-fit value for the high-flux group.  The confidence

619: contours of the dual Gaussian fit for the two groups are shown in \figibvsincontp.  The

620: contours for the two groups are mutually exclusive only up to 90\% confidence.  The contours

621: show that it is not required that both components exist in both of the data sets (only a broad

622: line is definitively detected in the high-flux group, only a narrow line in the low-flux

623: group).  On the other hand, at 99\% confidence, both of the components may be present in both

624: the high- and the low-flux states. 

625: 

626: In order to determine the implications for the proximity of the flaring region to the black

627: hole, we fitted a disk-line component in XSPEC (see Fabian \etal~1989) to the high flux data in

628: order to model the redshifted line.  The parameters of the disk-line model are the rest-frame

629: line energy ($E_{0}$), the power law index of the emissivity ($q$, where the emissivity$\sim

630: r^{q}$), the inner and outer radii of the disk emission ($R_{\rm in}$ and $R_{\rm out}$

631: respectively) relative to the central black hole, and the inclination angle of the disk with

632: respect to the line-of-sight of the observer ($\theta_{obs}$).  We included a power-law

633: continuum component in the model as well as a Gaussian component with the line energy (\efekap)

634: fixed at 6.4 keV and \sigfeka at 0.05 keV in order to model a possible narrow component to the

635: Fe~K line complex (although none was detected).  For this fit, we also included Compton

636: reflection (as described in \S\ref{reflection}) since the parameter constraints may be

637: different when the emission line is modeled as a disk line rather than a Gaussian.  Therefore,

638: this model had five free parameters: $\Gamma$, \ifekap, the reflection fraction ($R$),

639: $\theta_{\rm obs}$, and $R_{\rm out}$.  We fixed $E_{0}=6.4$ keV in the rest frame, $R_{\rm

640: in}=6R_{\rm G}=6GM/c^{2}$, and $q=-1.5$.  This value of $q$ is consistent with the

641: value found from Suzaku data (see Yaqoob \etal 2007). 

642: The data certainly do not rule out steeper emissivity laws ($q \sim -3$ to $-2$

643: may be more typical in AGN), and indeed some theoretical

644: models predict very steep emissivity laws with $q \sim -6$. 

645: However, 

646: $q$ is degenerate with $R_{\rm out}$: steeper values of $q$ will result in

647: larger allowed values of $R_{\rm out}$ for the same data because steep

648: emissivity laws by definition give most of the line emission from the innermost

649: regions of the disk and  $R_{\rm out}$ eventually becomes irrelevant. 

650: Therefore, by using $q=-1.5$ we are addressing the question of

651: what is the largest size of the region that can account for the redshifted

652: Fe~K line because more negative values of $q$ will automatically produce

653: more of the line from a smaller region for a given value of $R_{\rm out}$.

654: We found that $\theta_{obs}$ was constrained to be $<39^{\circ}$ (90\% confidence)

655: and this is not sensitive to $q$ as the same upper limit was obtained

656: when $q$ was fixed at $-3$.

657: The inclination angle constraint is consistent with $\theta_{obs}>31^{\circ}$ obtained from

658: modeling the persistent disk line emission in Suzaku data (Yaqoob \etal 2007).

659: 

660: We plotted confidence contours of the

661: intensity of the disk line versus the outer radius and found that at 99\% confidence, the line

662: originated within $\sim100$ gravitational radii of the black hole (see \figidvsroutp). 

663: Thus, in the context of our interpretation 

664: that the redshifted broad

665: component of the Fe~K emission line is due

666: to the line emission from the inner disk 

667: in the high state being temporarily enhanced during

668: continuum flares, we deduce that 

669: the enhanced region is smaller than $\sim 100$ gravitational radii. 

670: A similar event was observed in the Seyfert galaxy MCG~-6-30-15 by Iwasawa \etal

671: (1999) in which the Fe~K line centroid shifted to $\sim5$ keV and the 6.4 keV component was

672: not the prominent peak as it usually is in that source.  Although the narrow line emission may

673: have been present during the high-flux state of NGC~2992, it is possible that the continuum

674: swamped the line so that we were unable to detect it.

675: 

676: The quasi-simultaneous Suzaku observations of NGC~2992 (see \figxteltcrvp; Yaqoob \etal 2007)

677: showed evidence of a persistent, broad, disk line component with an intensity of

678: $1.9^{+0.5}_{-1.0}\times10^{-5}$ photons cm$^{-2}$ s$^{-1}$ (90\% confidence for one

679: interesting parameter).  Such a disk line likely originates from a larger region of the disk

680: than that found in the \rxte flare spectra and so does not have a highly redshifted centroid

681: energy compared to 6.4 keV because there is relatively more line emission from larger radii. 

682: In the \rxte data, we found the upper limits on the intensity of a persistent, extended

683: ($R_{\rm out}=1000 R_{\rm G}$) disk line (at 6.4 keV in the rest frame of the disk) for the

684: high- and low-flux groups to be $5.4\times10^{-5}$ and $5.8\times10^{-5}$ photons cm$^{-2}$

685: s$^{-1}$ (90\% confidence for one interesting parameter) respectively.  Thus, the continuous

686: presence of a persistent disk line with the line intensity observed in the Suzaku data, in

687: addition to the flaring disk line and the distant matter component, is not ruled out by the

688: \rxte data.

689: 

690: The flaring behavior in the \rxte data strongly supports the idea that X-rays originate in a

691: corona above the accretion disk.  If the X-ray source is in fact a corona above the disk, a

692: flare from the inner region of the disk could significantly intensify a particular part of the

693: line profile, making the whole line appear to be redshifted.  On the other hand, a flaring,

694: centrally located source would simply intensify the entire line and therefore the centroid

695: would not be redshifted. The slope of the power-law continuum ($\Gamma$) 

696: potentially holds important

697: information about the temperature ($kT$) and optical depth ($\tau$)

698: of this plasma from which the X-rays are

699: thought to originate. 

700: The spectral index, $\Gamma$, is

701: proportional to the Compton $y$ parameter which is a function of 

702: the product $(kT)\tau$.

703: Inclusion of X-ray data at higher energies than our \rxte data is needed 

704: because $kT$ could then be constrained directly by the data and then

705: $\tau$ could be derived from fitting Comptonization models (e.g. see Titarchuk 1994). 

706: Although high-energy data is available for the two \bsax observations of NGC~2992, Gilli

707: \etal (2000) did not address the question of constraining the high-energy cut-off of the

708: power-law continuum (the cut-off was arbitrarily fixed in their model fitting).  However,

709: Beckmann \etal (2007) find no evidence for a cut-off out to $\sim200$ keV, based on analyses

710: of \bsax, {\it INTEGRAL}, and {\it Swift} data, implying that the scattering corona is

711: Compton-thin.  If $\Gamma$ is truly constant during the large amplitude variation (see

712: \figgvsfp), then this would imply that the Compton $y$ parameter remains steady and any

713: accretion model must account for this.  

714: 

715: \section{CONCLUSIONS}

716: \label{conclusions}

717: \begin{itemize}

718: 

719: \item The \rxte light curve (\figxteltcrvp) shows that the flux varied by approximately a

720: factor of 10 on timescales of days to weeks. It was previously thought that this AGN went

721: through an extended period of quiescence over many years, followed by a `rebuilding' of the

722: accretion disk (see \S\ref{intro}).  However, in less than a year the \rxte data covered nearly

723: the entire dynamic range in flux seen in the historical data (\fighistlcp), suggesting

724: short-term flaring activity as opposed to long-term changes in accretion activity. 

725: Historically, when NGC~2992 was observed in a high-flux state, it exhibited properties of a

726: type 1 AGN, such as broad optical lines and rapid X-ray variability, as opposed to when it was

727: in a low-flux state during which it showed properties more consistent with a type 2 AGN.  This

728: transitioning behavior challenges the unified model and its theoretical predictions.

729: 

730: \item In most of the observations, an emission line was detected at $\sim6.4$ keV, likely

731: dominated by the \feka emission line that is known to come from distant matter.  The flux

732: variability of the Fe line was much less than the continuum variability.  Due to the poor

733: energy resolution of \rxte, the measured intensity of the dominant line component likely

734: has contributions from matter closer to the black hole (e.g., the accretion disk) as

735: well.  The best-fitting EW ranged from $\sim200$ to $\sim1200$ eV, which is consistent

736: with the bulk of the line emission at 6.4 keV having been unresponsive to continuum

737: variability.

738: 

739: \item In the three highest luminosity observations, a highly redshifted (\efeka $\sim5.6$

740: keV), broadened Fe line dominated the spectrum and the 6.4 keV component was not detected. 

741: The redshifted line may be a signature of flaring activity from the inner disk, likely coming

742: from within $100$ gravitational radii of the black hole (see \figidvsroutp), where strong

743: gravitational and Doppler effects are important.  Although a distant-matter Fe~K line

744: component is not detected during the flares, it is not ruled out.  While the high-luminosity

745: observations may be dominated by flaring in the inner disk, both the low- and high-luminosity

746: data allow for additional, persistent Fe~K line disk emission from the whole disk, as seen by

747: Suzaku (Yaqoob \etal 2007).  The \rxte data are not sensitive to this persistent component.

748: 

749: \item Although the continuum luminosity varied by a factor of $\sim 11$, the slope of the

750: power-law fit was consistent with no variability (weighted mean $\Gamma\sim1.7$).  In terms of

751: Comptonization models, this implies a roughly constant Compton $y$ parameter, a fact which must be

752: explained by any general model of accretion onto a supermassive black hole.  

753: 

754: \end{itemize}

755: 

756: Future monitoring of NGC~2992 with higher spectral resolution will be able to resolve the

757: Fe~K line complex and will improve our understanding of the accretion disk and the structure

758: in general of this source.  Observations at higher energies (which will allow us to determine

759: constraints on the plasma temperature and therefore the Compton $y$ parameter, as

760: well as better constraints on Compton reflection), combined with

761: multi-waveband monitoring will be critical to constraining black hole accretion models.  

762: 

763: The authors thank Alex Markowitz for useful discussions and help with 

764: the 3C 273 and 3C 279 data analysis and help with characterizing the

765: variability of NGC~2992. 

766: The authors acknowledge partial support from NASA grants NNG05GM34G 

767: (T.Y., K.M.), NNG0GB78A (T.Y., K.M.), and NRA-00-01-LTSA-034 (T.Y.);

768: Y.T. is supported by Grants-in-Aid for Scientific

769: Research (17740121).

770: 

771: 

772: \section*{APPENDIX}

773: \label{append}

774: 

775: Most of the \rxte data sets of NGC~2992 show a dip in both the PCU~0 and PCU~2 spectra near

776: 8--9 keV (see \S\ref{groupfit}).  The strength of this apparent absorption feature relative to

777: the continuum appears to be variable, with the center depth reaching 0--30\% below the

778: continuum (see \figgrpplratp). Although there does not appear to be a simple relation between

779: source flux and dip strength, the dip is always weak ($\sim10$\%) in the high-flux spectra and

780: is stronger in most of the low-flux spectra.  However, group 6 (see \figxteltcrvp), which was a

781: low-flux observation where the Fe~K line was not detected, does not show the dip.   

782: 

783: We believe that the dip may be an artifact of the background subtraction model in \rxtep.  The

784: main evidence for this comes from the Suzaku observations of NGC~2992, which were

785: quasi-simultaneous with the \rxte group 5 observations (see \figxteltcrvp).  Group 5 is one of

786: the groups with the strongest dip, yet the Suzaku data did not detect the feature. 

787: \figfeatewvssig shows the confidence contours of the EW of the feature (when fitted with an

788: inverse-Gaussian model) versus its intrinsic width for the \rxte group 5 data along with the

789: 99\% confidence upper limit constraints on the Suzaku data (dotted line).  Since the 99\%

790: confidence contours for the \rxte and Suzaku data do not overlap, the presence of the dip in

791: the Suzaku data is ruled out.  We found further evidence that the dip is an artifact by

792: examining some observations of 3C~273 and 3C~279 made in 2005.  We found the same feature in

793: many of the spectra of these sources, but its strength ranged from 0\% to only roughly 10\% of

794: the continuum.  Note that attempts to fit the dip feature with astrophysical models (such as

795: absorption edges) failed to account for the residuals. 

796: 

797: As far as we know, such a dip in the 8--9 keV continuum has not been reported for other

798: \rxte sources in the literature.  For example, Rothschild \etal (2006) presented \rxte data

799: for Cen~A from observations made much earlier than those of NGC~2992 but they did not detect

800: the dip feature.  Therefore, if the dip is an artifact in the \rxte data, it appears to be an

801: inadequacy of one background model for more recent data. 

802: 

803: \newpage

804: 

805: \begin{figure}[!htb]

806:  \centerline{

807: \epsscale{0.5}

808: \scalebox{0.8}{\includegraphics[angle=270]{f1.ps}}}

809:  \caption{

810:        Light curve of the historical NGC~2992 observations, including the present \rxte campaign.  

811:        The 2--10 keV flux varies by a factor of $\sim20$ during this $\sim30$ year period.

812:        }

813: \end{figure}

814: 

815: \begin{figure}[!htb]

816:  \centerline{

817: \epsscale{0.8}

818:     \plotone{f2a.eps}}

819:  \centerline{

820: \epsscale{0.8}

821:     \plotone{f2b.eps}}   

822:  \caption{

823:  	(a) PCU 0 vs PCU 2 measurements for the photon index, $\Gamma$, obtained from the power-law

824: 	plus Gaussian fits (\S\ref{indfits}).  The dotted line represents $\Gamma$(PCU 0)=$\Gamma$(PCU

825: 	2). (b) PCU 0 vs. PCU 2 measurements for the intensity of the \feka line, obtained from the

826: 	same fits.  The dotted line represents $I_{Fe~K\alpha}$(PCU 0)=$I_{Fe~K\alpha}$(PCU 2).

827: 	}

828: \end{figure}

829: 

830: \begin{figure}[!htb]

831:  \centerline{

832: \epsscale{0.5}

833:  \scalebox{0.8}{\includegraphics[angle=270]{f3.ps}}}

834:  \caption{

835:  	The 2--10 keV light curve from the 24 RXTE observations.  Groups 1--6 are identified as follows:

836: 	filled circles: group 1, stars: group 2, squares: group 3, triangles: group 4,

837: 	diamonds: group 5, open circles: group 6.  Crosses represent quasi-simultaneous

838: 	Suzaku observations.

839: 	}

840: \end{figure}

841: 

842: \begin{figure}[!htb]

843:  \centerline{

844:    \scalebox{0.8}{\includegraphics[angle=270]{f4.ps}}}

845:  \caption{

846:  	Plot of $\Gamma$ vs. 2--10 keV flux for the combined PCU 0~+~PCU 2 data (open and filled

847: 	circles).  Open circles identify those observations in which the centroid energy and

848: 	the width of the Fe~K line had to be fixed due to weak line detection (see \tableobslogp).

849: 	Measurements for groups 1--6 are given by crosses. 

850: 	The measurement for group 3+4+5 is given by an open triangle.  

851: 	The dotted line corresponds to the weighted

852: 	mean value of $\Gamma=1.732$.

853: 	}

854: \end{figure}

855: 

856: \begin{figure}[!htb]

857:  \centerline{

858:    \scalebox{0.8}{\includegraphics[angle=270]{f5.ps}}}

859:  \caption{

860:  	Centroid energy of the \feka line vs. 2--10 keV luminosity.  See Fig. 4 caption for

861: 	explanation of symbols.  The dotted line corresponds to the rest energy of the Fe~K

862: 	line from neutral Fe,

863: 	6.4 keV.

864: 	}

865: \end{figure}

866: 

867: \begin{figure}[!htb]

868:  \centerline{

869: \epsscale{0.8}

870:    \plotone{f6.eps}}

871:  \caption{

872:  	Intensity of the \feka line vs. 2--10 keV luminosity.  (a) Combined PCU~0~+~PCU~2 

873: 	measurements.  Open circles identify those observations in which the centroid energy and

874: 	the width of the Fe~K line had to be fixed due to weak line detection (see \tableobslogp).

875: 	(b) As above, for the grouped data (crosses). 

876: 	The measurement for group 3+4+5 is represented by an open triangle.  

877: 	(a, b) Theoretical curves for the Compton-thin case with

878: 	$\left(\frac{<L>}{L}\right)\left(\frac{N_{\rm H}}{10^{23} \ \rm cm^{-2}}\right)=

879: 	$ 0.1, 1, 5, and 10 (solid lines) and

880: 	for the Compton-thick case with $\left(\frac{<L>}{L}\right)$EW = 100,

881: 	250, 500, 750, 

882: 	and 1000 eV (dashed lines) are overlaid (see \S\ref{cont}).

883: 	}

884: \end{figure}

885: 

886: \begin{figure}[!htb]

887:  \centerline{

888:    \scalebox{0.8}{\includegraphics[angle=270]{f7.ps}}}

889:  \caption{

890:  	Results of PCU~0~+~PCU~2 model fits vs. time (see \S\ref{indfits}).  For the fraction of the

891: 	Eddington luminosity ($L/L_{Edd}$), a black hole mass of $5.2\times10^{7}

892: 	M_{\odot}$ was assumed.  Also shown are the energy of the \feka line in keV, the intensity of

893: 	the \feka line in $10^{-5} \ \rm photons \ cm^{-2} \ s^{-1}$, and the photon index. 

894: 	The energy of the \feka line was fixed at 6.4 keV for the last three observations 

895: 	(22, 23, \& 24) since the line was not significantly detected in these data sets.  The

896: 	dotted lines refer to 6.4 keV in the \efeka plot and the weighted mean values of \ifeka

897: 	and $\Gamma$ in their respective plots.

898: 	}

899: \end{figure}

900: 

901: \begin{figure}[!htb]

902:  \centerline{

903:    \scalebox{0.8}{\includegraphics[angle=270]{f8.ps}}}

904:  \caption{

905:  	Equivalent width of the \feka line vs. 2--10 keV luminosity.  See Fig. 4 caption for

906: 	explanation of symbols.

907: 	}

908: \end{figure}

909: 

910: \begin{figure}[!htb]

911:  \centerline{

912: \epsscale{1}

913:    \plotone{f9.eps}}

914:  \caption{

915:  	Plots of the ratios of the grouped data to a simple model.  The model includes

916: 	a power-law continuum and Galactic absorption only.  The y-axis scales are inconsistent

917: 	in order to clearly display the spectral features.

918: 	}

919: \end{figure}

920: 

921: \begin{figure}[!htb]

922:  \centerline{

923: \epsscale{1}

924:    \plotone{f10.eps}}

925:  \caption{

926:  	Plots of the ratios of the grouped data a model including \feka line emission.  

927: 	The model includes a power-law continuum and Galactic absorption plus a 

928: 	Gaussian component to model the \feka line.

929: 	}

930: \end{figure}

931: 

932: \begin{figure}[!htb]

933:   \centerline{

934:     \plotone{f11.eps}}

935:   \caption{

936:   	The 68\%, 90\%, \& 99\% confidence contours of the intensity of the \feka line versus

937: 	its centroid energy for the high-flux group and the low-flux

938: 	group (\S\ref{highlow}).

939: 	}

940: \end{figure} 

941: 

942: \begin{figure}[!htb]

943:   \centerline{

944: \epsscale{1}

945:     \plotone{f12.eps}}

946:   \caption{

947:   	Constraints on a dual Gaussian-line model, fitted to the high- and low-flux data. 

948: 	Shown are confidence contour plots (68\%, 90\%, and 99\%) of the intensity of a line 

949: 	with the centroid energy fixed at the best-fitting value found in the 

950: 	high-flux state ($E=5.6$ keV) versus the intensity of a line with the

951: 	centroid energy fixed at the best-fitting value found in the 

952: 	low-flux state ($E=6.3$ keV) for the high- and low-flux groups.

953: 	}

954: \end{figure}

955: 

956: \begin{figure}[!htb]

957: \centerline{

958: 	\scalebox{0.8}{\includegraphics[angle=270]{f13.ps}}}

959: 	\caption{

960: 		Constraints from the redshifted Fe~K line in the high-flux (group~1) spectrum.

961: 		Shown are the 68\%, 90\%, \& 99\% confidence contours of the intensity of a

962: 		relativistic Fe~K 

963: 		disk line component versus the effective outer radius of the disk line emission in 

964: 		units of gravitational radii.

965: 	       }

966: \end{figure}

967: 

968: \begin{figure}[!htb]

969:   \centerline{

970: \epsscale{1}

971:     \plotone{f14.eps}}

972:   \caption{

973:   	Confidence contours (68\%, 90\%, \& 99\%) of EW versus intrinsic width of the 

974: 	8--9 keV feature for group 5 of the \rxte data for NGC~2992.  

975: 	The 99\% confidence, upper limit contour of the feature for

976: 	quasi-simultaneous Suzaku data is shown as dotted line.

977: 	}

978: \end{figure}

979: 

980: 

981: \newpage

982: 

983: %Table 1

984: \begin{deluxetable}{ccccccc} 

985: \small

986: \tablecolumns{4} 

987: \tablewidth{0pc} 

988: \tablecaption{NGC 2992 {\it RXTE} Observation Log} 

989: \tablehead{ 

990: \colhead{Obs} & \colhead{Start\tablenotemark{a}} & & \colhead{End\tablenotemark{a}} & & \colhead{Exposure\tablenotemark{b}} & \colhead{Count Rate\tablenotemark{c}}}

991: \startdata

992: 1 & 04/03/05 & 13:27:23 & 04/03/05 & 14:20:27 & 6368 & 10.590 $\pm$ 0.064\\	  

993: 2 & 18/03/05 & 17:18:35 & 18/03/05 & 17:31:07 & 1504 & 2.287 $\pm$ 0.106\\	  

994: 3 & 21/03/05 & 08:16:11 & 21/03/05 & 08:54:51 & 4640 & 2.730 $\pm$ 0.062\\	  

995: 4 & 05/04/05 & 10:07:39 & 05/04/05 & 10:57:47 & 4928 & 7.467 $\pm$ 0.068\\	  

996: 5 & 20/04/05 & 12:12:27 & 20/04/05 & 12:57:47 & 5440 & 1.326 $\pm$ 0.055\\	  

997: 6 & 05/05/05 & 07:37:31 & 05/05/05 & 08:28:43 & 6144 & 3.646 $\pm$ 0.055\\	  

998: 7 & 08/05/05 & 06:25:15 & 08/05/05 & 07:04:59 & 4576 & 3.673 $\pm$ 0.063\\	  

999: 8 & 18/05/05 & 10:11:15 & 18/05/05 & 11:14:43 & 5216 & 8.084 $\pm$ 0.068\\	  

1000: 9 & 02/06/05 & 07:32:11 & 02/06/05 & 08:16:11 & 5280 & 1.555 $\pm$ 0.055\\	  

1001: 10 & 16/06/05 & 06:30:35 & 16/06/05 & 07:24:27 & 6464 & 1.350 $\pm$ 0.051\\	  

1002: 11 & 01/07/05 & 03:34:35 & 01/07/05 & 04:27:55 & 6400 & 0.802 $\pm$ 0.050\\	  

1003: 12 & 16/07/05 & 03:57:31 & 16/07/05 & 04:35:07 & 4512 & 1.203 $\pm$ 0.060\\	  

1004: 13 & 29/07/05 & 03:19:39 & 29/07/05 & 04:10:03 & 6048 & 3.402 $\pm$ 0.056\\	  

1005: 14 & 10/08/05 & 03:07:55 & 10/08/05 & 04:01:31 & 6432 & 2.841 $\pm$ 0.053\\	  

1006: 15 & 12/09/05 & 22:44:27 & 12/09/05 & 23:35:07 & 5600 & 1.882 $\pm$ 0.053\\	  

1007: 16 & 27/09/05 & 00:55:55 & 27/09/05 & 01:46:51 & 6112 & 1.322 $\pm$ 0.051\\	  

1008: 17 & 14/10/05 & 01:45:31 & 14/10/05 & 02:39:07 & 5600 & 0.886 $\pm$ 0.055\\	  

1009: 18 & 29/10/05 & 19:13:31 & 29/10/05 & 20:03:07 & 5632 & 1.013 $\pm$ 0.054\\	  

1010: 19 & 13/11/05 & 22:35:23 & 13/11/05 & 23:23:07 & 5728 & 1.970 $\pm$ 0.054\\	  

1011: 20 & 28/11/05 & 22:34:03 & 28/11/05 & 23:27:23 & 6400 & 1.687 $\pm$ 0.051\\	  

1012: 21 & 13/12/05 & 17:55:07 & 13/12/05 & 18:45:31 & 6048 & 1.827 $\pm$ 0.055\\	  

1013: 22 & 28/12/05 & 19:46:51 & 28/12/05 & 20:30:35 & 5248 & 1.278 $\pm$ 0.057\\	  

1014: 23 & 12/01/06 & 18:06:34 & 12/01/06 & 18:59:54 & 6400 & 0.850 $\pm$ 0.051\\	  

1015: 24 & 28/01/06 & 09:52:11 & 28/01/06 & 10:45:31 & 6368 & 1.882 $\pm$ 0.053\\	  

1016: \enddata 

1017: \tablenotetext{a}{Universal time} 

1018: \tablenotetext{b}{Exposure time is given in seconds} 

1019: \tablenotetext{c}{Count rate is given in counts s$^{-1}$}

1020: \end{deluxetable} 

1021: 

1022: %Table 2

1023: \begin{deluxetable}{rccccccccc}

1024: \rotate

1025: \tablecolumns{15}

1026: \tablewidth{0pc}

1027: \tablecaption{NGC 2992 {\it RXTE} Fit Results}

1028: \tablehead{ 

1029: \colhead{Obs} & \colhead{F$_{2-10}$\tablenotemark{a}} & \colhead{L$_{2-10}$\tablenotemark{b}} & \colhead{$\Gamma$\tablenotemark{c}} & \colhead{\efeka\tablenotemark{d}} & \colhead{\sigfeka\tablenotemark{e}} & \colhead{\ifeka\tablenotemark{f}} & \colhead{EW\tablenotemark{g}} & \colhead{$\chi^{2}$\tablenotemark{h}} & \colhead{dof\tablenotemark{i}}}

1030: \startdata

1031:  1 &  8.88 &  1.17 & $ 1.706 ^{+0.038} _{-0.039} $ & $  5.73 ^{+ 0.57} _{- 0.58} $ & $  0.83 ^{+  0.55} _{- 0.49} $ & $  18.8 ^{+ 10.4} _{-  9.6} $ & $ 170^{+  94} _{-  86} $ & 20.1 & 23 \\

1032:  & & & [1.672-1.739] & [ 5.22- 6.23] & [ 0.41-  1.30] & [  10.3-  27.8] & [  93- 251] & & \\

1033:  2 &  2.00 &  0.26 & $ 1.812 ^{+0.292} _{-0.262} $ & $  6.61 ^{+ 0.57} _{- 0.58} $ & $  0.00 ^{+  1.03} _{- 0.00} $ & $   8.9 ^{+  7.7} _{-  7.8} $ & $ 464^{+ 405} _{- 406} $ & 17.5 & 23 \\

1034:  & & & [1.581-2.066] & [ 6.11- 7.00] & [ 0.00-  0.81] & [   1.8-  15.6] & [  94- 817] & & \\

1035:  3 &  2.40 &  0.32 & $ 1.798 ^{+0.142} _{-0.133} $ & $  6.53 ^{+ 0.57} _{- 0.65} $ & $  0.68 ^{+  0.80} _{- 0.68} $ & $  11.9 ^{+  6.9} _{-  6.9} $ & $ 517^{+ 300} _{- 300} $ & 10.6 & 23 \\

1036:  & & & [1.681-1.922] & [ 5.98- 7.02] & [ 0.00-  1.35] & [   5.9-  18.0] & [ 256- 783] & & \\

1037:  4 &  6.22 &  0.82 & $ 1.753 ^{+0.056} _{-0.058} $ & $  5.58 ^{+ 0.48} _{- 0.44} $ & $  0.68 ^{+  0.71} _{- 0.68} $ & $  20.2 ^{+  9.8} _{-  9.3} $ & $ 252^{+ 122} _{- 116} $ & 17.1 & 23 \\

1038:  & & & [1.703-1.803] & [ 5.19- 5.99] & [ 0.00-  1.30] & [  12.0-  28.8] & [ 150- 360] & & \\

1039:  5 &  1.17 &  0.15 & $ 1.814 ^{+0.246} _{-0.229} $ & $  6.32 ^{+ 0.31} _{- 0.29} $ & $  0.00 ^{+  0.51} _{- 0.00} $ & $   8.2 ^{+  4.1} _{-  4.2} $ & $ 710^{+ 355} _{- 363} $ & 24.1 & 23 \\

1040:  & & & [1.612-2.029] & [ 6.11- 6.54] & [ 0.00-  0.43] & [   4.5-  11.8] & [ 389-1022] & & \\

1041:  6 &  2.98 &  0.39 & $ 1.656 ^{+0.085} _{-0.084} $ & $  6.10 ^{+ 0.43} _{- 0.48} $ & $  0.04 ^{+  1.09} _{- 0.04} $ & $   6.4 ^{+  4.3} _{-  4.4} $ & $ 186^{+ 125} _{- 126} $ & 22.1 & 23 \\

1042:  & & & [1.582-1.731] & [ 5.70- 6.46] & [ 0.00-  0.98] & [   2.6-  10.2] & [  75- 297] & & \\

1043:  7 &  2.98 &  0.39 & $ 1.726 ^{+0.103} _{-0.101} $ & $  6.22 ^{+ 0.35} _{- 0.35} $ & $  0.33 ^{+  0.57} _{- 0.33} $ & $  11.9 ^{+  5.6} _{-  5.6} $ & $ 375^{+ 176} _{- 176} $ & 20.3 & 23 \\

1044:  & & & [1.637-1.816] & [ 5.92- 6.53] & [ 0.00-  0.80] & [   7.0-  16.8] & [ 221- 530] & & \\

1045:  8 &  6.66 &  0.88 & $ 1.688 ^{+0.062} _{-0.068} $ & $  5.42 ^{+ 0.66} _{- 0.68} $ & $  1.10 ^{+  0.70} _{- 0.64} $ & $  28.5 ^{+ 17.7} _{- 14.3} $ & $ 297^{+ 184} _{- 149} $ & 14.3 & 23 \\

1046:  & & & [1.629-1.743] & [ 4.82- 5.99] & [ 0.53-  1.71] & [  15.8-  43.8] & [ 164- 456] & & \\

1047:  9 &  1.26 &  0.17 & $ 1.758 ^{+0.252} _{-0.234} $ & $  6.16 ^{+ 0.65} _{- 0.71} $ & $  0.80 ^{+  0.89} _{- 0.80} $ & $  12.0 ^{+  7.1} _{-  7.1} $ & $ 930^{+ 550} _{- 550} $ & 13.2 & 23 \\

1048:  & & & [1.554-1.978] & [ 5.55- 6.72] & [ 0.00-  1.54] & [   5.8-  18.2] & [ 449-1410] & & \\

1049: 10 &  1.11 &  0.15 & $ 1.676 ^{+0.250} _{-0.234} $ & $  6.15 ^{+ 0.37} _{- 0.38} $ & $  0.59 ^{+  0.53} _{- 0.59} $ & $  13.3 ^{+  5.5} _{-  5.4} $ & $1189^{+ 491} _{- 482} $ & 15.7 & 23 \\

1050:  & & & [1.470-1.894] & [ 5.82- 6.47] & [ 0.00-  1.04] & [   8.6-  18.1] & [ 768-1618] & & \\

1051: 

1052: \tablebreak

1053: 

1054: 11 &  0.81 &  0.11 & $ 1.953 ^{+0.363} _{-0.319} $ & $  6.30 ^{+ 0.59} _{- 0.60} $ & $  0.00 ^{+  1.19} _{- 0.00} $ & $   4.7 ^{+  3.9} _{-  3.8} $ & $ 619^{+ 514} _{- 501} $ &  6.0 & 23 \\

1055:  & & & [1.670-2.269] & [ 5.80- 6.79] & [ 0.00-  0.98] & [   1.4-   8.1] & [ 178-1068] & & \\

1056: 12 &  0.91 &  0.12 & $ 1.612 ^{+0.350} _{-0.324} $ & $  6.10 ^{+ 0.84} _{- 0.97} $ & $  0.55 ^{+  1.01} _{- 0.55} $ & $   7.2 ^{+  6.3} _{-  6.2} $ & $ 718^{+ 628} _{- 618} $ & 15.6 & 23 \\

1057:  & & & [1.328-1.916] & [ 5.30- 6.80] & [ 0.00-  1.35] & [   1.8-  12.7] & [ 179-1267] & & \\

1058: 13 &  2.88 &  0.38 & $ 1.774 ^{+0.099} _{-0.095} $ & $  6.41 ^{+ 0.67} _{- 0.74} $ & $  0.46 ^{+  1.19} _{- 0.46} $ & $   7.1 ^{+  5.3} _{-  5.3} $ & $ 243^{+ 182} _{- 182} $ & 14.8 & 23 \\

1059:  & & & [1.691-1.861] & [ 5.80- 6.98] & [ 0.00-  1.45] & [   2.5-  11.8] & [  85- 405] & & \\

1060: 14 &  2.43 &  0.32 & $ 1.707 ^{+0.103} _{-0.099} $ & $  6.27 ^{+ 0.36} _{- 0.39} $ & $  0.00 ^{+  1.02} _{- 0.00} $ & $   6.7 ^{+  4.0} _{-  4.1} $ & $ 253^{+ 151} _{- 155} $ & 17.9 & 23 \\

1061:  & & & [1.620-1.797] & [ 5.95- 6.57] & [ 0.00-  0.89] & [   3.1-  10.2] & [ 117- 386] & & \\

1062: 15 &  1.67 &  0.22 & $ 1.815 ^{+0.164} _{-0.155} $ & $  6.81 ^{+ 0.99} _{- 2.55} $ & $  0.00 ^{+  1.94} _{- 0.00} $ & $   3.7 ^{+  3.9} _{-  3.7} $ & $ 251^{+ 265} _{- 251} $ & 17.2 & 24 \\

1063:  & & & [1.679-1.958] & [ 6.08- 7.42] & [ 0.00-  1.09] & [   0.4-   7.1] & [  27- 483] & & \\

1064: 16 &  1.14 &  0.15 & $ 1.774 ^{+0.260} _{-0.245} $ & $  6.13 ^{+ 0.88} _{- 1.06} $ & $  0.86 ^{+  1.04} _{- 0.86} $ & $   8.9 ^{+  7.1} _{-  6.6} $ & $ 764^{+ 602} _{- 568} $ & 21.4 & 24 \\

1065:  & & & [1.559-2.001] & [ 5.24- 6.89] & [ 0.00-  1.70] & [   3.0-  15.1] & [ 256-1290] & & \\

1066: 17 &  0.89 &  0.12 & $ 2.054 ^{+0.407} _{-0.349} $ & $  6.38 ^{+ 0.31} _{- 0.31} $ & $  0.07 ^{+  0.70} _{- 0.07} $ & $   8.8 ^{+  4.2} _{-  4.2} $ & $1148^{+ 557} _{- 557} $ & 22.7 & 24 \\

1067:  & & & [1.745-2.407] & [ 6.11- 6.65] & [ 0.00-  0.69] & [   5.0-  12.5] & [ 656-1641] & & \\

1068: 18 &  0.85 &  0.11 & $ 2.075 ^{+0.423} _{-0.361} $ & $  6.45 ^{+ 0.42} _{- 0.41} $ & $  0.03 ^{+  0.75} _{- 0.03} $ & $   6.3 ^{+  4.2} _{-  4.1} $ & $ 876^{+ 584} _{- 570} $ & 32.5 & 24 \\

1069:  & & & [1.754-2.442] & [ 6.10- 6.81] & [ 0.00-  0.68] & [   2.7-  10.0] & [ 375-1391] & & \\

1070: 19 &  1.72 &  0.23 & $ 1.876 ^{+0.160} _{-0.150} $ & $  6.27 ^{+ 0.39} _{- 0.47} $ & $  0.03 ^{+  1.42} _{- 0.03} $ & $   6.2 ^{+  4.1} _{-  4.2} $ & $ 352^{+ 233} _{- 238} $ & 13.5 & 24 \\

1071:  & & & [1.744-2.016] & [ 5.88- 6.62] & [ 0.00-  1.22] & [   2.5-   9.8] & [ 142- 557] & & \\

1072: 20 &  0.94 &  0.12 & $ 2.248 ^{+0.370} _{-0.315} $ & $  6.45 ^{+ 0.54} _{- 0.95} $ & $  0.40 ^{+  0.89} _{- 0.40} $ & $   6.5 ^{+  4.6} _{-  4.6} $ & $ 893^{+ 625} _{- 632} $ & 27.9 & 24 \\

1073:  & & & [1.969-2.570] & [ 5.79- 6.92] & [ 0.00-  1.14] & [   2.5-  10.5] & [ 343-1443] & & \\

1074: 

1075: \tablebreak

1076: 

1077: 21 &  1.52 &  0.20 & $ 1.834 ^{+0.212} _{-0.198} $ & $  6.09 ^{+ 0.57} _{- 0.62} $ & $  0.84 ^{+  0.71} _{- 0.84} $ & $  14.2 ^{+  7.5} _{-  7.3} $ & $ 915^{+ 483} _{- 470} $ & 14.1 & 24 \\

1078:  & & & [1.660-2.020] & [ 5.55- 6.59] & [ 0.00-  1.44] & [   7.8-  20.8] & [ 502-1340] & & \\

1079: 22 &  1.11 &  0.15 & $ 1.830 ^{+0.202} _{-0.188} $ & $  6.40f $ & $  0.05f $ & $   2.0 ^{+  3.4} _{-  2.0} $ & $ 184^{+ 305} _{- 184} $ & 22.8 & 26 \\

1080:  & & & [1.626-2.052] & [\ldots] & [\ldots] & [   0.0-   5.7] & [   0- 517] & & \\

1081: 23 &  1.28 &  0.17 & $ 1.746 ^{+0.136} _{-0.130} $ & $  6.40f $ & $  0.05f $ & $   4.6 ^{+  3.1} _{-  3.1} $ & $ 321^{+ 212} _{- 214} $ & 20.7 & 26 \\

1082:  & & & [1.605-1.894] & [\ldots] & [\ldots] & [   1.3-   8.0] & [  87- 553] & & \\

1083: 24 &  1.63 &  0.22 & $ 1.751 ^{+0.125} _{-0.119} $ & $  6.40f $ & $  0.05f $ & $   2.5 ^{+  3.2} _{-  2.5} $ & $ 148^{+ 191} _{- 148} $ & 17.2 & 26 \\

1084:  & & & [1.622-1.887] & [\ldots] & [\ldots] & [   0.0-   6.0] & [   0- 355] & & \\

1085: \enddata

1086: \tablenotetext{a}{2--10 keV flux ($10^{-11}$ photons s$^{-1}$)}

1087: \tablenotetext{b}{2--10 keV luminosity ($10^{43}$ ergs s$^{-1}$)}

1088: \tablenotetext{c}{Photon index}

1089: \tablenotetext{d}{Energy of the \feka line (keV)}

1090: \tablenotetext{e}{Intrinsic width of the \feka line (keV)}

1091: \tablenotetext{f}{Intensity of the \feka line ($10^{-5}$ photons cm$^{-2}$ s$^{-1}$)}

1092: \tablenotetext{g}{Equivalent width of the \feka line (eV)}

1093: \tablenotetext{h}{$\chi^{2}$ for the model}

1094: \tablenotetext{i}{Degrees of freedom}

1095: \end{deluxetable}

1096: 

1097: %Table 3	

1098: \begin{deluxetable}{rcccccccccc}

1099: \rotate

1100: \tablecolumns{15}

1101: \tablewidth{0pc}

1102: \tablecaption{NGC 2992 {\it RXTE} Fit Results: Grouped Data}

1103: \tablehead{ 

1104: \colhead{Group} & \colhead{F$_{2-10}$\tablenotemark{a}} & \colhead{L$_{2-10}$\tablenotemark{b}} & \colhead{$\Gamma$\tablenotemark{c}} & \colhead{\efeka\tablenotemark{d}} & \colhead{\sigfeka\tablenotemark{e}} & \colhead{\ifeka\tablenotemark{f}} & \colhead{EW\tablenotemark{g}} & \colhead{$\chi^{2}$\tablenotemark{h}} & \colhead{dof\tablenotemark{i}} & \colhead{P\tablenotemark{j}}}

1105: \startdata

1106:  1 &  7.38 &  0.97 & $ 1.713 ^{+0.029} _{-0.029} $ & $  5.58 ^{+ 0.33} _{- 0.33} $ & $  0.89 ^{+ 0.36} _{- 0.35} $ & $ 22.3 ^{+ 6.8} _{- 6.6} $ & $ 232^{+  70} _{-  68} $ & 23.6 & 23 & 57.7 \\

1107:  & & & [1.688-1.738] & [ 5.29- 5.87] & [ 0.58- 1.20] & [16.5- 28.3] & [ 172- 295] & & \\

1108:  2 &  2.70 &  0.35 & $ 1.723 ^{+0.045} _{-0.044} $ & $  6.30 ^{+ 0.20} _{- 0.20} $ & $  0.32 ^{+ 0.39} _{- 0.32} $ & $  7.7 ^{+ 2.2} _{- 2.2} $ & $ 271^{+  77} _{-  77} $ & 28.2 & 23 & 79.3 \\

1109:  & & & [1.684-1.763] & [ 6.13- 6.48] & [ 0.00- 0.66] & [ 5.7-  9.6] & [ 200- 337] & & \\

1110:  3 &  1.05 &  0.14 & $ 1.757 ^{+0.116} _{-0.112} $ & $  6.28 ^{+ 0.18} _{- 0.19} $ & $  0.26 ^{+ 0.38} _{- 0.26} $ & $  7.3 ^{+ 2.0} _{- 2.0} $ & $ 681^{+ 186} _{- 186} $ & 27.4 & 23 & 76.2 \\

1111:  & & & [1.658-1.859] & [ 6.12- 6.44] & [ 0.00- 0.59] & [ 5.5-  9.1] & [ 513- 849] & & \\

1112:  4 &  1.23 &  0.16 & $ 1.850 ^{+0.135} _{-0.127} $ & $  6.45 ^{+ 0.33} _{- 0.34} $ & $  0.40 ^{+ 0.45} _{- 0.40} $ & $  6.4 ^{+ 2.8} _{- 2.7} $ & $ 558^{+ 244} _{- 235} $ & 37.4 & 24 & 96.0 \\

1113:  & & & [1.738-1.968] & [ 6.15- 6.74] & [ 0.00- 0.79] & [ 4.0-  8.9] & [ 348- 776] & & \\

1114:  5 &  1.25 &  0.17 & $ 1.961 ^{+0.123} _{-0.117} $ & $  6.23 ^{+ 0.27} _{- 0.30} $ & $  0.56 ^{+ 0.39} _{- 0.56} $ & $  8.6 ^{+ 2.9} _{- 2.8} $ & $ 718^{+ 242} _{- 233} $ & 42.9 & 24 & 99.0 \\

1115:  & & & [1.858-2.069] & [ 5.96- 6.47] & [ 0.00- 0.91] & [ 6.1- 11.1] & [ 509- 927] & & \\

1116:  6 &  1.35 &  0.18 & $ 1.765 ^{+0.084} _{-0.081} $ & $  6.40f $ & $  0.05f $ & $  3.1 ^{+ 1.9} _{- 1.9} $ & $ 215^{+ 132} _{- 132} $ & 26.8 & 26 & 58.2 \\

1117:  & & & [1.677-1.856] & [\ldots] & [\ldots] & [ 1.0-  5.1] & [  69- 354] & & \\

1118: 3+4+5 &  1.17 &  0.15 & $ 1.861 ^{+0.070} _{-0.069} $ & $  6.32 ^{+ 0.14} _{- 0.14} $ & $  0.37 ^{+ 0.25} _{- 0.37} $ & $  7.3 ^{+ 1.4} _{- 1.4} $ & $ 642^{+ 123} _{- 123} $ & 79.4 & 23 & $>4\sigma$ \\

1119:  & & & [1.800-1.923] & [ 6.20- 6.44] & [ 0.00- 0.59] & [ 6.0-  8.5] & [ 528- 748] & & \\

1120: \enddata

1121: \tablenotetext{a}{2--10 keV flux ($10^{-11}$ photons s$^{-1}$)}

1122: \tablenotetext{b}{2--10 keV luminosity ($10^{43}$ ergs s$^{-1}$)}

1123: \tablenotetext{c}{Photon index}

1124: \tablenotetext{d}{Energy of the \feka line (keV)}

1125: \tablenotetext{e}{Intrinsic width of the \feka line (keV)}

1126: \tablenotetext{f}{Intensity of the \feka line ($10^{-5}$ photons cm$^{-2}$ s$^{-1}$)}

1127: \tablenotetext{g}{Equivalent width of the \feka line (eV)}

1128: \tablenotetext{h}{$\chi^{2}$ for the model}

1129: \tablenotetext{i}{Degrees of freedom}

1130: \tablenotetext{j}{The values of $1-\frac{P}{100}$ correspond to the probability of obtaining the

1131: observed $\chi^{2}$ or higher from statistical fluctuations alone.}

1132: \end{deluxetable}

1133: 

1134: \newpage 

1135: 

1136: \begin{thebibliography}{}

1137: \bibitem[Anders \& Grevesse 1989]{ande1989} 

1138: Anders, E., \& Grevesse, N. 1989, Geochimica et Cosmochimica Acta 53, 197

1139: 

1140: \bibitem[Arnaud 1996]{arnaud1996}

1141: Arnaud, K. A. 1996, in Astronomical Data Analysis

1142: Software and Systems V, ed. Jacoby, G., \& Barnes, J.

1143: (Astronomical Society of the Pacific), Conference Series, Vol. 101, p. 17 

1144: 

1145: \bibitem[Beckmann, Gehrels, \& Tueller]{beckmann2007}

1146: Beckmann, V., Gehrels, N., \& Tueller, J. 2007, ApJ in press, astro-ph/0704.2698

1147: 

1148: \bibitem[Dickey \& Lockman 1990]{dick1990}

1149: Dickey, J. M., \& Lockman, F. J. 1990, ARA\&A, 28, 215

1150: 

1151: \bibitem[Dov{\v c}iak, Karas, \& Yaqoob 2004]{dovc2004}

1152: Dov{\v c}iak, M., Karas, V., \& Yaqoob, T.\ 2004, ApJS, 153, 205 

1153: 

1154: \bibitem[Fabian 1989]{fabian1989}

1155: Fabian, A.C., Rees, M.J., Stella, L., \& White, N.E. 1989, MNRAS, 238, 729

1156: 

1157: \bibitem[Ghisellini, Haardt, \& Matt 1994]{ghis1994}  

1158: Ghisellini, G., Haardt, F., \& Matt, G. 1994, MNRAS, 267, 743

1159: 

1160: \bibitem[Gilli \etal 2000]{gilli2000}

1161: Gilli, R., Maiolino, R., Marconi, A., Risaliti, G., Dadina, M.,

1162: Weaver, K. A., \& Colbert, E. J. M. 2000, A\&A, 355, 485

1163: 

1164: \bibitem[Iwasawa \etal 1999]{iwas1999} 

1165: Iwasawa, K., Fabian, A. C., Young, A. J., Inoue, H., \& Matsumoto, C. 1999, MNRAS, 306, L19

1166: 

1167: \bibitem[Jahoda 2006]{jahoda2006}

1168: Jahoda, K., \etal~2006, ApJS, 163, 401

1169: 

1170: \bibitem[mchardy 2007]{mchardy2007}

1171: McHardy, I. M., Koerding, E., Knigge, C., Uttley, P., \& Fender, R. P.

1172: 2007, Nat, 444, 730

1173: 

1174: \bibitem[Markowitz 2004]{markowitz2004}

1175: Markowitz, A., \& Edelson, R. 2004, ApJ, 617, 939

1176: 

1177: \bibitem[Markowitz 2003]{markowitz2003}

1178: Markowitz, A., Edelson, R., \& Vaughan, S. 2003, ApJ, 598, 953

1179: 

1180: \bibitem[Mushotzky 1982]{mush1982} 

1181: Mushotzky, R. F. 1982, ApJ, 256, 92

1182: 

1183: \bibitem[Nandra \& Pounds 1994]{nand1994}

1184: Nandra, K., \& Pounds, K. A. 1994, MNRAS, 268, 405

1185: 

1186: \bibitem[Palmeri \etal 2003]{palmeri2003}

1187: Palmeri, P., Mendoza, C., Kallman, T. R., \& Bautista, M. A. 2003, A\&A, 410, 359

1188: 

1189: \bibitem[Papadakis]{papdak2004}

1190: Papadakis, I. E. 2004, MNRAS, 348, 207

1191:  

1192: \bibitem[Piccinotti \etal 1982]{picc1982}

1193: Piccinotti, G., Mushotzky, R. F., Boldt, E. A., Holt, S. S., Marhsall, F. E., 

1194: Serlemitsos, P. J., \& Shafer, R. A. 1982, ApJ, 253, 485

1195: 

1196: \bibitem[Reynolds \& Nowak 2003]{reyn2003} 

1197: Reynolds, C. S., \& Nowak, M. A. 2003, Pys. Rep., 377, 389

1198: 

1199: \bibitem[Rothschild \etal 2006]{roth2006}

1200: Rothschild, R. E., \etal 2006, ApJ, 641, 801

1201: 

1202: \bibitem[Titarchuk 1994]{tita1994}

1203: Titarchuk, L. 1994, ApJ, 434, 570 

1204: 

1205: \bibitem[Turner \& Pounds 1989]{turn1989}

1206: Turner, T. J., \& Pounds, K. A. 1989, MNRAS, 240, 833

1207: 

1208: \bibitem[Turner \etal 1991]{turn1991} 

1209: Turner, T. J., Weaver, K. A., Mushotzky, R. F., Holt, S. S.,

1210:  \& Madejski, G. M. 1991, ApJ, 381, 85

1211: 

1212: \bibitem[Turner et al. 1999]{turner1999}

1213: Turner, T. J., George, I. M., Nandra, K., \& Turcan, D. 1999, ApJ, 524, 667

1214: 

1215: \bibitem[Vaughan et al. 2003]{vaughan2003}

1216: Vaughan, S., Edelson, R., Warwick, R. S., \& Uttley, P. 2003,

1217: ApJ, 345, 1271

1218: 

1219: \bibitem[Weaver \etal 1996]{weav1996} 

1220: Weaver, K. A., Nousek., J., Yaqoob, T., Mushotzky, R. F., Makino, F.,

1221:  \& Otani, C. 1996, ApJ, 458, 160 

1222: 

1223: \bibitem[Woo \& Urry 2002]{wooj2002}

1224: Woo, J. H. \& Urry, C. M. 2002, ApJ, 579, 530

1225: 

1226: \bibitem[Yaqoob \etal 2001]{yaq2001} 

1227: Yaqoob, T., George, I. M., Nandra, K., Turner, T. J., 

1228: Serlemitsos, P. J., \& Mushotzky, R. F. 2001, ApJ, 546, 759

1229: 

1230: \bibitem[Yaqoob \etal 2007]{yaq2007}

1231: Yaqoob, T. \etal 2007, PASJ, in press \\

1232: 

1233: \end{thebibliography}

1234: 

1235: \end{document}

1236: