Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://hea-www.harvard.edu/~pgreen/Papers/flarestar01.pdf Äàòà èçìåíåíèÿ: Mon Sep 19 19:13:08 2005 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 01:42:55 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: annular

Mon. Not. R. Astron. Soc. 323, 577 ± 583 (2001)

Extensive serendipitous X-ray coverage of a flare star with ROSAT
J. D. Silverman,w K. A. Eriksen, P. J. Green and S. H. Saar
Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138, USA

Accepted 2000 October 18. Received 2000 October 9; in original form 1999 November 17

ABSTR A CT

We report the serendipitous discovery of a flare star observed with the ROSAT X-ray observatory. From optical spectra, which show strong and variable emission lines of the hydrogen Balmer series and neutral helium, we classify this object as a M3.0Ve star, and estimate a distance of 52 pc from published photometry. Owing to the close proximity of the star (13.6 arcmin) to the calibration source and RS CVn binary AR Lacertae, long-term X-ray coverage is available in the ROSAT archive (,50 h spanning 6.5 yr). Two large flare events occurred early in the mission (1990 June ± July), and the end of a third flare was detected in 1996 June. One flare, observed with the Position Sensitive Proportional Counter (PSPC), had a peak luminosity LX 1:1 á 1030 erg s21 ; an e-folding rise time of 2.2 h and a decay time of 7 h. This decay time is one of the longest detected on a dMe star, providing evidence for the possibility of additional heating during the decay phase. A large High Resolution Imager (HRI) flare (peak LX 2:9 á 1030 erg s21 is also studied. The `background' X-ray emission is also variable ± evidence for low-level flaring or microflaring. We find that >59 per cent of the HRI counts and >68 per cent of the PSPC counts are caused by flares. At least 41 per cent of the HRI exposure time and 47 per cent of the PSPC are affected by detectable flare enhancement. Key words: stars: flare ± stars: late-type ± X-rays: stars.

1

I NTR O DUC TION

X-ray emission from late-type M dwarfs has been studied extensively to investigate the structure and emission mechanisms of stellar coronae. Coronal heating to X-ray emitting temperatures is attributed to either impulsive flares or quiescent energy release in magnetic structures. The corona of these stars are thought to be similar to the Sun, but often with luminosities that are orders of magnitude higher. The high magnetic activity of these flare stars is also seen in their optical spectra. Continuum enhancement and strong emission lines of the H Balmer series, Ca ii and neutral He are often evident (Montes et al. 1999). An EXOSAT study (Pallavicini, Tagliaferri & Stella 1990) showed that flares have a wide range of energies and time-scales. Most outbursts can be described as either impulsive (decay time ,1 h) or long-decay flares (decay time .1 h) and have thermal X-ray spectra with temperatures similar to solar X-ray flares. The impulsive stellar flares have time-scales similar to the compact solar flares. The long-duration flares have greater total energy and are more similar to two-ribbon flare events. From ROSAT observations, coronal emission from dMe stars has been shown to have two distinct spectral components, a lowtemperature component attributed to quiescent active regions and

w

E-mail: jds@head-cfa.harvard.edu

a variable, high-temperature component owing to compact flaring regions (Giampapa et al. 1996). Schmitt (1994) has shown conclusively the existence of longduration flares on M stars using the ROSAT all-sky survey. In some flaring stars, the long decay can be attributed to continual heating of the flaring region (Schmitt & Favata 1999; Ottmann & Schmitt 1996). New flare models have been developed (Reale & Micela 1998) in which the additional heating determines the characteristics of the decay. Using this model, an analysis of long-duration flares on AD Leo (Favata, Micela & Reale 2000b) and EV Lac (Favata et al. 2000a) have shown the emitting regions to be compact with length-scales of less than the stellar radius and similar in size to solar flares, thereby providing evidence that long-duration flares are produced in high-pressure structures. During an analysis of observations with the ROSAT High Resolution Imager (HRI) of the RS CVn binary AR Lac, we noticed an X-ray source within the field of view to be highly variable and undetected in many fields. The first X-ray detection of this source was with the Einstein observatory (hence the catalogue name, 2E 2206.614517; Harris et al. 1993). Coaddition of six IPC observations spanning 26.6 ks yield a 4s detection with 115 net counts. The source is undetected in a 1.5 ks HRI observation. No flaring activity is evident during these observations. We analysed 39 observations from the ROSAT public data archive (16 PSPC; 23 HRI) which included this source position within the field of view. The total observing time was <50 h. The

q 2001 RAS


578

J. D. Silverman et al.
We measured count rates using the iraf/pros data analysis software and corrected for vignetting caused by the large range of off-axis angles 2 , u , 47 arcmin: The observations were subdivided into multiple time bins to achieve a higher temporal resolution while preserving a minimal 2s detection for each bin. An annular region was centred on the source to correct for the background count rate for most cases. For detections near the edge of the field, a nearby circular background region was chosen. A log of the ROSAT observations is given in Table 1 which includes the exposure time and off-axis angle. Spectral fitting of the PSPC data was done with the xspec software package. Source and background counts were extracted using the xselect task from the ftools package, and the ancillary response files were constructed with pcarf to account for off-axis vignetting. We ignored the lowest 11 spectral energy bins to avoid scattered solar extreme ultraviolet (EUV) contamination. The highest 56 spectral channels were also omitted because of poor statistics ± a result of a marked decrease in the effective area of the instrument at these energies. The energy bin distribution oversamples the intrinsic spectral resolution of the PSPC, so we grouped the bins by a factor of 5 to improve the statistics. Multiple optical spectra were taken by Perry Berlind with the Tillinghast 60 arcsec telescope and FAST spectrograph (Fabricant et al. 1998) at the Fred Lawrence Whipple Observatory on Mount Hopkins. A slit width of 3 arcsec, a 300 lines mm21 grating, and the Loral charge-coupled device (CCD) with 15 mm pixels é provided a resolution of <5A. A 3.5-min exposure was taken on ut 1998 May 16 and two 10-min exposures were acquired on ut 1998 May 30 and ut 1998 June 24. An observation of Feige 34 was used for extinction correction and flux calibration. Standard bias subtraction, flat-fielding, the extraction of one-dimensional spectra and wavelength calibration were performed using iraf.

observations span two flares with almost complete light curves, the tail end of a third flare, and show variability in the low-level X-ray emission. The flare detected with the PSPC has a long decay time, possibly providing evidence for significant continual heating during the flare decay. Thus, a further in-depth study of flares and quiescent emission from 2E 2206.614517 could provide a useful test of current models.

2

O BSER VA TIO N S AND D A T A REDU CTION

We detect the X-ray source 2E 2206.614517 in 16 observations with the ROSAT Position Sensitive Proportional Counter (PSPC) and 23 observations with the HRI, within the instrument bandpass of 0:1±2:4 keV: Extensive and continuous X-ray observations (<17 h) were made with the PSPC between 1990 June 18±22 (during the ROSAT in-orbit calibration period), 1991 December 30±31; and 1993 May 29 and June 02. The HRI calibration observations of AR Lac began 13 d after the completion of the PSPC observations. 14 HRI pointings between 1990 July 2±8; include the source 2E 2206.614517. Eight additional HRI observations are available in the archive over the period of 1992 June through 1996 November for a total observing time of 33 h.
Table 1. ROSAT X-ray observations. Seq. name Obs. date MJD Exp. time (s) 27 1 1 1 1 13 3 1 2 2 2 1 2 2 13 4 20 1 1 1 2 1 1 1 1 2 2 2 1 1 1 2 5 1 19 13 4 21 3 843.3 945.2 919.7 986.5 826.9 932.7 330.2 885.0 340.6 132.7 542.1 838.5 255.3 011.2 062.9 702.8 667.5 419.7 972.8 687.1 208.0 933.6 987.0 700.6 664.4 264.7 501.1 124.4 660.4 619.7 718.8 437.5 515.6 919.7 142.9 664.0 488.0 157.0 218.4 Off-axis angle (arcmin) 16.5 29.3 4.4 19.2 37.2 28.3 47.1 10.5 20.8 27.4 9.0 39.3 34.7 45.5 16.3 16.3 13.6 17.7 10.1 13.5 15.2 8.7 14.2 16.2 8.8 4.0 9.2 17.8 10.0 1.6 10.8 13.6 13.5 13.5 11.7 13.5 13.5 13.6 13.4

rp100588 rp110586 rp110591 rp110599 rp110589 rp110592 rp110601 rp110590 rp110602 rp110595 rp110598 rp110596 rp110597 rp110600 rp1600991 rp4002781 rh100247 rh110249 rh110251 rh110252 rh110253 rh110254 rh110255 rh110259 rh110260 rh110261 rh110262 rh110267 rh110268 rh110269 rh110270 rh141876 rh202061 rh202062 rh202063 rh202159 rh202160 rh202161 rh180166

18/06/90 ± 29/06/90 19/06/90 19/06/90 19/06/90 19/06/90 ± 20/06/90 20/06/90 19/06/90 ± 22/06/90 20/06/90 20/06/90 20/06/90 ± 21/06/90 21/06/90 ± 22/06/90 20/06/90 20/06/90 19/06/90 30/12/91 ± 31/12/91 29/05/93 ± 02/06/93 02/07/90 02/07/90 02/07/90 02/07/90 02/07/90 02/07/90 02/07/90 03/07/90 03/07/90 03/07/90 03/07/90 04/07/90 ± 05/07/90 04/07/90 05/07/90 05/07/90 10/06/92 10/12/95 11/12/95 12/01/96 ± 13/01/96 03/06/96 ± 04/06/96 06/06/96 09/07/96 25/11/96 ± 27/11/96

480 480 480 480 480 480 480 480 480 480 480 480 480 480 486 491 480 480 480 480 480 480 480 480 480 480 480 480 480 480 480 487 500 500 500 502 502 502 504

60.13 61.25 61.53 61.85 61.85 62.19 61.98 62.48 62.79 62.92 63.85 62.12 62.05 61.91 20.26 36.37 74.14 74.28 74.48 74.59 74.66 74.73 74.80 75.07 75.14 75.66 75.86 76.88 76.94 77.01 77.07 83.53 61.04 62.90 94.89 37.73 40.05 73.20 13.38

3 S OU RCE IDEN TIFICA TION AND O P T I C A L SPECT R A After obtaining an X-ray source position from a nearly on-axis HRI observation, we identified two candidate optical counterparts within 10 arcsec on the Digitized Sky Survey red plates. Two short observations were adequate to classify candidate 1 as an active Me star based on its Balmer emission, and candidate 2 as a normal G dwarf, with no evidence for a composite spectrum. We thus identify the Me star as the optical counterpart to 2E 2206.614517. Coordinates from the USNO-A2.0 catalogue (Monet et al. 1998) are a 22h 08m 37s5; d 145831 H 27: 9 : 0 (J2000) and a 5-arcmin finder chart is provided in Fig. 1. To facilitate an accurate spectral classification, the emissionline star was observed again on ut 1998 May 30 and ut 1998 June 24 for 10 min each. Both spectra showed strong emission lines, typical of a dMe flare star. The first long spectrum (Fig. 2) showed strong emission lines, typical of a dMe flare star in an active flaring state. The second observation (Fig. 3) showed a marked decrease in the strength of the emission lines, indicating that the flare star was either in or close to quiescence. Table 2 lists the measured emission line equivalent widths (EQs). The strength of chromospheric heating owing to the flare is evident by the increase in the hydrogen line emission by a factor of 1.6 Ha±4:1 (Hg ). The optical magnitudes from the USNO catalogue are B 16:5 and R 14:2 for this source, colours consistent with a late-type star. However, these colours may be affected by the strong
q 2001 RAS, MNRAS 323, 577 ± 583

Notes: rp PSPCC; rp1 PSPCB; rh HRI:


X-ray coverage of a flare star with ROSAT

579

Figure 1. Digital Sky Survey plate showing the two possible optical counterparts to the X-ray flare. The star labelled dMe is the likely counterpart. The field of view is 5 á 5 arcmin2 (north is up, east is left). The thin structures to the south-east are plate artefacts.

Figure 2. Spectrum of the optical counterpart of 2E 2206.614517 observed on 1998 May 30. The strong Balmer emission lines are indicative of a dMe star in an active flare state. q 2001 RAS, MNRAS 323, 577 ± 583


580

J. D. Silverman et al.

Figure 3. A second optical spectrum of 2E 2206.614517 was obtained on 1998 June 24. The decrease in the emission line strengths, as compared with the spectrum obtained one month earlier, could be attributed to a decrease or lack of flare activity. Table 2. Optical emission line equivalent widths. Date ID H11a l 3771 H10a l 3798 H9 l 3835 H81He i l 3889 Ca ii(K) l 3934 He 1Ca ii(H) l 3970 Hd l 4102 Hg l 4340 He i l 4471 Hb l 4861 He ia l 5876 Ha l 6563
a

1998 May 30 é EW (A) 6.7 13.6 23.0 26.6 24.1 48.6 33.8 43.4 2.3 26.2 1.4 13.0

1998 June 24 é EW (A) ± ± 7.6 9.2 22.7 19.6 9.4 10.6 ± 8.4 ± 8.0

Continuum difficult to define.

emission lines of the star. To derive an accurate spectral classification, we first removed the emission lines from the spectra by interpolating to the adjacent `continuum'. We then cross-correlated the resulting spectrum (using the iraf fxcorr task) with a sample of digital spectra of M dwarfs from the Gliese catalogue (Henry, Kirkpatrick & Simons 1994). By far the best correlation, and a very close match was with the M3.0V star Gliese 251 (cross-correlation peak height 0.99). For an M3.0V star, we find M V 11:7 ^ 0:6 from equation (1) of Henry et al. (1994; note that this error is the empirical rms value, more representative of the actual error than 1s ). Bessell (1991) tabulates V 2 R 1:1 and B 2 V 1:55 for an M3.0V star. From this we derive M R 10:6 for comparison to the USNO red magnitude, to obtain a distance of 52 pc.

4 4.1

RO S A T X- RA Y O BSER VA T I O N S Flare activity ± light curve

The PSPC light curve (Fig. 4) shows prominent X-ray flare

activity on 1990 June 19. Using the parameters for the best-fitting spectral model (Section 4.3), the peak flare luminosity is LX 1:1 á 1030 erg s21 ; a 24-fold increase over the mean quiescent luminosity LX 4:6 á 1028 erg s21 (Section 4.2). The outburst can be characterized by an e-folding rise time tr . 2:2 h and decay time td . 7h: The total rise and decay time for the flare is Dtrise . 6 h and Dtdecay . 30 h as measured from the quiescent to peak count rate. The tail end of the flare significantly departs from the exponential decay of the flare (Fig. 4). This flare is evidently a long-duration event. Most long-decay flares on dMe stars have td , 1 h (Pallavicini et al. 1990). Recently, long-duration flares have been detected on EV Lac td 10:5h; Schmitt 1994) and AD Leo td 2:2h; Favata et al. 2000b). Continual heating of the flaring region during the decay has been proposed to explain such long-decaying events. Because of the long decay time, the total energy Etot 3:5 á 1034 erg (Table 3) released is similar to the flare seen on EV Lac Etot 9 á 1033 erg; Schmitt 1994) and large compared with other dMe flare stars < 3 á 1030 ±1 á 1034 ; Pallavicini et al. 1990). These long-decay flares on dMe stars are, however, 2±3 orders of magnitude less energetic than giant X-ray flares on RS CVn stars such as Algol td 8:4h; Etot 7 á 1036 erg; Ottmann & Schmitt õ 1996), and CF Tucanae td 22 h; Etot 1:4 á 1037 erg; Kurster & Schmitt 1996). On 1990 July 04 at 21:11:35 UT, a flare with a peak luminosity, about the same magnitude as the PSPC out-burst (Fig. 5), was detected with the HRI. Since the HRI has extremely limited spectral resolution, we input the model fit to the PSPC flare and the HRI count rate into the iraf/pros task hxflux to convert counts to luminosity. The peak flare luminosity was LX 2:9 á 1030 erg s21 a factor of 54 larger than the mean quiescent luminosity LX 5:4 á 1028 erg s21 : The outburst can be characterized by an e-folding rise time tr . 15 min and decay time td . 1:2h: The total rise and decay time for the flare is Dtrise . 40 min and Dtdecay . 4:6h: The total energy released during the flare is Etot 1:6 á 1034 erg (Table 3). We also discovered the tail end of an additional flare observed
q 2001 RAS, MNRAS 323, 577 ± 583


X-ray coverage of a flare star with ROSAT

581

Figure 4. X-ray flare of 2E 2206.614517 observed with the PSPC. The flare begins on 1990 June 19 at 12:45:22 ut and lasts for 1±1:5d: The error bars are ^1s based on count statistics. The x error bars show the time-spanned for each measurement. The broken curve is an exponential function characterized by the e-folding time, td . 7h: The soft and hard bands for the hardness ratio are defined as 0:07±0:42 and 0:42±2:48 keV:

Figure 5. X-ray flare of 2E 2206.614517 observed with the HRI on 1990 July 4. The flare was first detected at 21:11:35 ut and lasted for 4.6 h. The lower plot focuses at the outburst in detail. The broken curve is an exponential function characterized by the e-folding time of the decay td . 1:2h:

Table 3. X-ray flare characteristics. Instrument PSPC HRI HRI Date 1990 June 19 1990 July 4 1996 June 3 Peak (ut) 12:45:22 21:11:35 ± Dtrise .6h .40 min ±

t

r

Dtdecay .30 h .4.6 h ,9h

t

d

LX (peak) 1.1á10 erg s 2.9á1030 erg s2 ±
30 21 1

Etot .3.5á1034 erg .1.6á1034 erg ±

.2.2 h .15 min ±

.7 h . 1.2 h ±

q 2001 RAS, MNRAS 323, 577 ± 583


582

J. D. Silverman et al.
4.3 Spectral analysis

on 1996 June 03 with a two-fold rise above the background level. Because of our incomplete light curve, we can only provide an approximation to the decay time (,9 h). The X-ray characteristics of these flares are listed in Table 3.

4.2 Count rate distribution analysis: quiescent level and flaring fraction The `quiescent' X-ray emission outside the large flares is clearly itself variable. Fig. 4 shows a factor of 3 change in the level outside of flares, and three long HRI observations between 1995 December and 1996 July show average of LX 15:4; 6.6 and 5:0 á 1028 erg s21 : To investigate this further, we performed a statistical analysis of the complete X-ray light curve to determine the minimum fraction of counts caused by flares, and the minimum fraction of time spent in a flaring state. We assumed a truly non-variable, quiescent background rate exists that can be described by a Poisson distribution. We then determined the fraction of counts caused by flares by integrating the counts left unexplained by a least-squares fit of a Poisson function to the low end of the count rate distribution (Saar & Bookbinder 1998). We averaged the results of fits using several reasonable count rate binning sizes. This analysis yields the minimum fraction of flare counts, since flares below the instrument sensitivity, and possible quiescent level changes owing to rotation and evolution of magnetic regions are all included in the derived quiescent level. We find that (at least) 68 ^ 2 per cent of the PSPC counts and 59 ^ 2 per cent of the HRI counts are caused by flares. Upper limits to the quiescent flux level and the fraction of time that the star is quiescent are also derived from the analysis. We find that the (Poisson) mean quiescent count level is 0:022 ^ 0:001 count s21 for the PSPC LX < 4:6 ^ 0:2 á 1028 erg s21 and 0:0056 ^ 0:0006 count s21 for the HRI LX < 5:4 ^ 0:6 á 1028 erg s21 : The minimum fraction of time with a detectable flare contribution to the observed flux is 47 ^ 5 per cent (PSPC) and 41 ^ 8 per cent (HRI). All of the PSPC data were taken between 1990 and 1993, while the HRI data include a significant fraction from 1995 and 1996, with an exposure-time-weighted time difference of ,1200 d. Thus it is possible that the small (,15 per cent) difference in the PSPC and HRI quiescent fluxes may be explained by time evolution of the quiet emission, owing to, for example, long-term evolution in the numbers of active regions. Unfortunately, there are relatively few quiescent counts in the 1995±96 HRI data, making a direct test of time variation inconclusive. Further data over a longer time-scale would help decide the level of the quiescent coronal variability of the star. We find that the quiescent and bolometric luminosity agree with the linear correlation between these quantities, as noted by previous studies of flare stars (Pallavicini et al. 1990; Agrawal, Rao & Sreekantan 1986). We estimate a bolometric luminosity of 6:74 á 1031 erg s21 ; using M V 11:7 and the bolometric correction of Pettersen (1983).
Table 4. X-ray spectral model fits. Model Quiescence Quiescence Flare
a

We extracted spectra from the brightest quiescent and flare PSPC images. During the longest pointing of this field, the source 2E 2206.614517 was in quiescence (rp100588) and 16.5 arcmin offaxis. The data set corresponding to a flare event with the highest number of source counts was rp110591, for which the source was nearly on-axis (4.4 arcmin). For the quiescent X-ray emission, we found that no singlecomponent thermal plasma (Raymond & Smith 1978) model could adequately fit the data. Unfortunately, the signal-to-noise ratio of our spectrum was not sufficient to uniquely constrain a two-component model. However, we found that models that did not contain a Raymond ± Smith component of kT < 1:0 keV could be rejected (i.e. x2 @ 1: These models significantly underpredict n emission from approximately 0:85±1:0 keV; corresponding to the Fe L-shell blend. However, with an appropriate `high'-temperature thermal plasma model, the remaining low-temperature emission has x2 statistics of much less than one, indicating that the models n are not well constrained. Despite the poorly constrained twocomponent model, the spectral fit suggests the existence of a coronal plasma of kT < 1 keV: A comparison of the normalizations of the two models shows that the `hot' and `cool' components contribute nearly equal flux in the PSPC band (Table 4). Because of the fewer counts in the flare observation, a larger binning factor was necessary to provide a significant signal-tonoise ratio for spectral fitting. However, we could not fit the resulting spectrum with any one- or two-component model. The heavy binning broadens the effect of uncertain calibration features, particularly the PSPC window carbon edge at 0.4 keV. To compare the quiescent and flare spectra, we must mitigate this effect. The peaks and troughs in the spectra most affected by binning all occur below about 1 keV. Using only the photons above 1 keV has the added advantage that the `hot' component dominates in this regime, so we may use a single-temperature model. We fit both spectra in the energy range 1:0±1:8 keV with a single-temperature Raymond ± Smith model, using the value of NH from the full quiescent spectrum. A direct comparison of the flare and quiescent normalization shows an increase by nearly a factor of 20 in the emission measure. In addition, though the temperatures are not very well constrained, their 1s errors just barely overlap, indicating that the increase in X-ray emission was likely to be accompanied by an increase in the plasma temperature. This is consistent with behaviour seen during flares in late-type active stars (e.g. Giampapa et al. 1996; Singh et al. 1999). The small number of counts does not allow a determination of a temperature variation during the flare decay. To investigate the spectral evolution of the flare, we calculated hardness ratios (Fig. 4). The count distribution appears to harden by about 40±50 per cent during the onset of the flare. A decrease of the hardness ratio during the decay is evident without certainty owing to limited count statistics.

NH (cm22) 5.0á10 5.0á1019 5.0á1019
19

kT (1) (keV) 0.16 0:9910:: 20 1:3210:: 20
a

Norm (1) 4.8á10 6.8á102 1.3á102
25 5 3

kT (2) (keV) 0.97 ± ±
a

Norm (2) 6:8 á 102 ± ±
5

x

2 n

RS1RS RS RS

3 2 7 3

0.66 0.33 0.66

The one-sigma error is not bounded. q 2001 RAS, MNRAS 323, 577 ± 583


X-ray coverage of a flare star with ROSAT
Singh et al. (1999) found that Fe abundances for active dwarf stars with logLX =Lbol . 23:7 are strongly subsolar, whereas the Fe abundances in the less active stars are within a factor of 2 of the solar value. Since the quiescent luminosity ratio for 2E 2206.614517 hovers near this value, low abundances may also affect our X-ray spectral fits. On the other hand, flare activity has been observed to increase the apparent elemental abundances (Ottman & Schmitt 1996). 5 CONCLUSION A C KNO WLEDGMENTS

583

In this paper, we have identified a new flare star 2E 2206.614517 using extensive ROSAT X-ray observations and optical spectra. The object has been classified as a M3.0Ve star. Three large X-ray flare events are seen with light curves characteristic of outbursts detected in other flare stars. Two flares with almost complete X-ray coverage show a variation in strength and time-scales. A flare detected with the PSPC had a peak luminosity LX 1:1 á 1030 erg s21 ; an e-folding rise time of <2.2 h and a decay time of <7 h. We interpret this long-decay time (one of the longest ever observed) as possible evidence for continual heating as similar to recent observations and analysis of AD Leo and EV Lac (Favata et al. 2000a; Schmitt 1994). An observation with the HRI detected a flare with a higher peak luminosity of LX 2:9 á 1030 erg s21 ; an e-folding rise time of <15 min, and a decay time of <1.2 h. We used a statistical analysis of the X-ray light curves to measure the quiescent X-ray level, the (minimum) fraction of counts in flares and (minimum) fraction of time that the star exhibits flare or microflare activity. The PSPC data showed a quiescent luminosity of LX 4:6 ^ 0:2 á 1028 erg s21 ; with >68 ^ 2 per cent of counts coming from flares, and a significant flare contribution to at least 47 ^ 5 per cent of the time observed. The HRI data, taken on average almost 3 yr later, show a quiescent luminosity of LX 5:4 ^ 0:4 á 1028 erg s21 ; with > 59 ^ 2 per cent of counts coming from flares, and a significant flare contribution to at least 41 ^ 8 per cent of the time observed. We obtained two optical spectra that show strong and variable emission lines of the hydrogen Balmer series and neutral helium which is further evidence for flare activity. With the large amount of X-ray coverage of the source 2E 2206.614517 in the ROSAT data archive, this object provides an excellent opportunity to facilitate in the understanding of the physical properties of these flare stars including the flare, quiescent and long-term activity. Further measurements of this object may be afforded by calibration observations of AR Lac by Chandra or XMM-Newton.

We would like to thank the reviewer, J. Schmitt, and J. Drake for their useful comments on this manuscript, and Rick Harnden and Andrea Prestwich for their advice and useful discussions. Many thanks to Perry Berlind for obtaining the optical spectra, and to Susan Tokarz for their preliminary reduction. This research has made use of data obtained through the High Energy Astrophysics Science Archive Research Center Online Service, provided by the NASA/Goddard Space Flight Center. The Digitized Sky Surveys were produced at the Space Telescope Science Institute under U.S. Government grant NAG W-2166. The images of these surveys are based on photographic data obtained using the Oschin Schmidt Telescope on Palomar Mountain and the UK Schmidt Telescope. The plates are processed into the present compressed digital form with the permission of these institutions. REFERE NCES
Agrawal P. C., Rao A. R., Sreekantan B. V., 1986, MNRAS, 219, 225 Bessell M. S., 1991, AJ, 101, 662 Fabricant D., Cheimets P., Caldwell N., Geary J., 1998, PASP, 110, 79F Favata F., Reale F., Micela G., Sciortino S., Maggio A., Matsumoto H., 2000a, A&A, 353, 987 Favata F., Micela G., Reale F., 2000b, A&A, 354, 1021 Giampapa M. S., Rosner R., Kashyap V., Fleming T. A., Schmitt J. H. M. M., Bookbinder J. A., 1996, ApJ, 463, 707 Harris D. E. et al., 1993, The Einstein Observatory catalogue of IPC X-ray sources, 7, H Henry T. J., Kirkpatrick J. D., Simons D. A., 1994, AJ, 108, 1437 õ Kurster M., Schmitt J. H. M. M., 1996, A&A, 311, 211 Monet D. G., 1998, BAAS, 193, 12003 Montes D., Saar S. H., Collier Cameron A., Unruh Y. C., 1999, MNRAS, 305, 45 Ottmann R., Schmitt J. H. M. M., 1996, A&A, 307, 813 Pallavicini R., Tagliaferri G., Stella L., 1990, A&A, 228, 403 Pettersen B. R., 1983, in Byre P. B., Rodono M., eds, Activity in Reddwarf Stars. Reidel, Dordrecht, p. 17 Raymond J. C., Smith B. W., 1977, ApJS, 35, 419 Reale F., Micela G., 1998, A&A, 334, 1028 Saar S. H., Bookbinder J. A., 1998, in Donahue R. A., Bookbinder J. A., eds, The Tenth Cambridge Workshop on Cool Stars, Stellar Systems, and the Sun (ASP CD ROM). Astron. Soc. Pac., San Francisco, p. 1560 Singh K. P., Drake S. A., Gotthelf E. V., White N. E., 1999, ApJ, 512, 874 Schmitt J. H. M. M., 1994, ApJ, 90, 735 Schmitt J. H. M. M., Favata F., 1999, Nat, 401, 44

A This paper has been typeset from a TEX/L TEX file prepared by the author.

q 2001 RAS, MNRAS 323, 577 ± 583