1997MNRAS.286..369M Mon. Not. R. Astron. Soc. 286, 369-383 (1997) An EUV-selected sample of DA white dwarfs from the ROSAT All-Sky Survey - I. Optically derived stellar parameters M. C. Marsh/*t M. A. Barstow/* D. A. Buckley,2 M. R. Burleigh/* J. B. Holberg,3:j: D. Koester,4 D. Q'Donoghue,5 A. J. Penny6 t and A. E. Sansom7t lDepartment of Physics and Astronomy, University of Leicester, University Road, Leicester LE1 7RH South African Astronomical Observatory, Cape Town, South Africa 3Lunar and Planetary Laboratory, University of Arizona, Gould-Simpson Building, Tucson, AZ 85721, USA 41nstitut jUr Astronomie und Astrophysik, Universitiit Kie~ D-24098, Kie~ Germany 5Department of Astronomy, University of Cape Town, Rondebosch 7700, South Africa 6Astrophysics Division, Rutherford Appleton Laboratory, Didcot, Oxfordshire OXll OQX 7 Centre for Astrophysics, University of Central Lancashire, Preston PR1 2HE 2 1996 November 7. Received 1996 November 5; in original form 1996 May 30 ABSTRACT One of the most important results of the ROSAT All-Sky X-ray and EUY Surveys has been the detection of a large number of white dwarfs, allowing a detailed study of the general properties of DA white dwarf atmospheres to be carried out. However, this work relies on a knowledge of the effective temperatures, surface gravities and visual magnitudes. We present analyses of optical data from our follow-up programme, including the values of T eff , logg, mv and stellar mass obtained. We also list the PSPC and WFC count rates for each star. The distribution of masses and surface gravities derived from the optical work also provides important information about the sample of white dwarfs and gives some indication of possible selection biases. Comparing the ROSAT sample of 89 stars with the cooler optically selected sample of Bergeron, Saffer & Liebert reveals a statistically significant excess (~3 times the number found by them) of hot, massive DAs, which may represent a population of coalesced binary white dwarfs. As the optical and EUY samples do not cover the same range of white dwarf temperatures, the highmass excess may partly arise from differences in the cooling rates of 'normal' (~O.6-M0) and massive (> 1.0-M0) stars. Consequently, this feature is not necessarily the result of selection on the basis of EUY flux, and it might also be present in an optically selected sample covering an appropriate temperature range. Key words: stars: atmospheres - stars: luminosity function, mass function - white dwarfs. 1 INTRODUCTION Among the most significant problems in the study of hot white dwarf evolution has been the existence of two distinct groups having either H- or He-dominated atmospheres and the possible relationships between these groups and their *Guest Observer with the SAAO l.9-m telescope. tGuest Observer with the l.O-m Jacobus Kapteyn Telescope of the ING. :j:Guest Observer with the Steward Observatory 2.3-m telescope. proposed progenitors, the diverse types of central stars of planetary nebulae (CPN). While the very hottest H-rich DA white dwarfs outnumber He-rich DOs by a factor of 7 (Fleming, Liebert & Green 1986), the relative number of Hand He-rich CPN is only about 3:1. In addition, there is an apparent absence of He-rich stars in the temperature range 30000-45000 K, the so-called DO-DB gap, suggesting that H- and He-dominated groups are not entirely distinct. Several competing processes can affect the composition of a white dwarf atmosphere. He and heavier elements tend to sink out of the photosphere under the influence of gravity, ©1997 RAS © Royal Astronomical Society • Provided by the NASA Astrophysics Data System Downloaded from http://mnras.oxfordjournals.org/ by guest on October 6, 2014 Accepted 1997MNRAS.286..369M 370 M C. Marsh et al. results with the white dwarf evolutionary sequences of Wood (1992), they were able to estimate the stellar masses and, consequently, obtain the mass distribution of the sample. This spectroscopic technique provides a more detailed picure than earlier photometric studies. However, statistical work of this nature often has a number of limitations. The BSL sample of 129 DA white dwarfs is the most extensive study reported in detail, for which the measured mean mass was 0.56 Mo. However, the sample spanned only the temperature range from ~ 13 000 to 40000 K, and the number of white dwarfs with temperatures above ~25 000 K, the point at which X-ray/EUV emission becomes significant, made up no more than 25 per cent of the total. Consequently, the BSL sample does not overlap significantly with the stars in the ROSAT survey; indeed, only eight of their stars have been detected as X-ray/EUV . sources, and their results are not necessarily representative ..of the hotter, younger population we have observed. More recently, some results have been presented extending study of the mass distribution up to higher temperatures. Liebert & Bergeron (1995) have studied a sample of 'white dwarfs with temperatures above 15 000 K, drawn from the Palomar Green survey and including 200 white dwarfs. They obtain a slightly higher mean mass of 0.582 when carrying out the analysis with the earlier evolutionary models, which did not include an outer hydrogen layer. . However, using new models with thick hydrogen layers (10- 4 Mo; Wood 1995), which are more in keeping with theoretical predictions, yields a still higher value of 0.609 (note: wid). these models the BSL sample mean increases to 0.590). Finley (1995) reports a study of 177 DA white dwarfs with temperatures above 24000 K, obtaining mean masses of 0.601 and 0.63 Mo (Finley, private communication) with the zero and thick hydrogen layer mass models respectively, higher than either BSL or Liebert & Bergeron (1995). However, this sample is rather more heterogeneous than that drawn from the PG survey, containing a significant fraction of EUV-selected objects besides those discovered by optical means. , . All these re~rled mass distributions are dominated by stars selected by optical observations. As Liebert & Bergeron (1995) point out, these are strongly biased against the highest· mass stars with the smallest radii. Selecting stars on the basis' of their EUVemission may diminish this effect to some extent. Radiative levitation calculations (e.g., most recently, Chayer et al. 1994, 1995) predict that photospheric abundances of heavy elements should depend strongly on gravity, higher gravity stars having a lower heavy-element content. Consequently, those white dwarfs with the greatest mass should also be the most luminous in the EUV. How much this can counteract the effect of smaller radius will, however, depend on the stellar temperature, since the radiative levitation mechanism becomes efficient only above 40000 K In addition, other, different, selection effects may operate on the EUV sample, for example, interstellar absorption and effective temperature, to which the observed EUV flux is very sensitive. We present here a detailed comparison of the EUV-selected sample of white dwarfs with those drawn from the optical catalogues. A larger number of low-gravity objects are found in the ROSAT WFC survey, which is consistent with the bias to high white dwarf temperatures. However, more import© 1997 RAS, MNRAS 286, 369-383 © Royal Astronomical Society • Provided by the NASA Astrophysics Data System Downloaded from http://mnras.oxfordjournals.org/ by guest on October 6, 2014 but this can be counteracted by radiation pressure. Convective mixing, accretion or mass loss via a weak wind may also playa Significant role in determining the heavy-element content of white dwarf atmospheres. Shipman (1976) made the now universally accepted proposition that the X-ray/EUV flux from DA white dwarfs was, in fact, from their hot photospheres. He pointed out that, since the emission· would emerge from the· hotter, deeper layers of the stellar atmosphere, X-ray/EUV observations should provide a sensitive probe of its structure and composition. This important idea has underpinned all subsequent X-ray/EUV studies of white dwarfs. In particular, it was possible to demonstrate, from an analysis of 30 catalogued DA white dwarfs detected in the ROSAT All-Sky Xray and EUV Surveys, that most white dwarfs hotter than ~ 40000 K contain significant quantities of elements heavier than either H or He, while below this temperature the stars can be represented by more or less pure H atmospheres (Barstow et al. 1993). Subsequent studies, increasing the numbers of white dwarfs in the sample (e.g. Marsh et al. 1995; Wolff, Jordan & Koester 1996), have confirmed these results. The data provided by ROSAT have led to a clear breakthrough in our understanding of the general properties of the atmospheres of the hot DA white dwarf population. However, the full exploitation of this resource would not have been possible without prior knowledge of the temperatures, gravities and visual magnitudes of the stars detected. These parameters, derived for the most part from optical observations, serve to specify the EUV/X-ray flux for a given model composition. In principle, if the composition of the white dwarf atmosphere is known, the entire spectrum can be modelled from the temperature and gravity. The visual magnitude then acts as a convenient normalization factor defining the ratio between stellar distance and radius. Thus the remaining free parameters, related to the photospheric and intervening interstellar opacity, can be determined from the data. . When ROSAT was launched, this necessary information was available for only a relativ:ely small number of already catalogued white dwarfs, and this was the main factor limiting the size of the sample in the first analysis of the survey. Approximately 176 white dwarfs Were ultimately detected by the X-ray telescope (Fleming·et aL 1996), and 120 were seen in the EUV. CQnseq~ent1y, the 'original survey sample can potentially be increased by a factor of 4, although in practice not all ROSAT detections are bright enough to provide useful information. These additional objects include a number of known white dwarfs not previously reported as X-ray sources, and not well studied at optical wavelengths, together with a substantial number of newly discovered stars. This latter group account for ~ 50 per cent of the total number of white dwarfs in the ROSAT catalogues. A determination of effective temperature and surface gravity of a white dwarf can be obtained from the profiles of the hydrogen Balmer lines in the optical spectrum, comparing observations with the calculations of theoretical model atmospheres. This technique was pioneered by Holberg et al. (1985) and Kidder (1991). Bergeron, Saffer & Liebert (1992, hereafter BSL), have used it to carry out a systematic study of the DA white dwarf population. Combining these 1997MNRAS.286..369M A UV-selected sample of DA white dwarfs antly, we find an excess of high-mass objects in the ROSAT group. This might arise from the preferential selection of those white dwarfs with the highest EUY luminosity, with high gravity counteracting the effect of radiative levitation, lowering the abundances of heavy elements and, as a consequence, the photospheric opacity in the EUY. Alternatively, the excess may be a product of the bias to higher white dwarf temperatures. As Finley (1995) points out, there is a sharp decrease in the cooling rate of massive (> 1 Mo) white dwarfs in the range 40000-50000 K, which would increase the numbers of such stars in this interval, compared to the 'normal' (lower mass) population. 2 THE ROSAT SKY SURVEY AND OPTICAL IDENTIFICATION PROGRAMME 3 OPTICAL SPECTROSCOPY 3.1 Observations As hot DA white dwarfs are thermal sources of radiation with spectra peaking in the EUY and soft X-ray, their spectral shape and hence their observed count rates in the ROSAT bandpasses are very sensitive to T eff . It is therefore important that Teff is well determined, in order to fully utilize the ROSAT survey data. This work demands relatively high signal-to-noise spectra (SIN;:::: 50) of moderate resolution (~8 A), and the data from the optical identification programme were, in general, unable to fulfil this requirement. Consequently, a programme of spectroscopic observations was undertaken in both northern and southern hemispheres to obtain data for the newly discovered group of white dwarfs and those catalogued stars where suitable spectra did not exist. Southern hemisphere observations were made in 1993 October and 1994 March usirig the 1.9-m Radcliffe telescope of the South African Astronomical Observatory (SAAO), with the Unit spectrograph and Reticon photon counting system (RPCS). The RPCS has two arrays, one which accumulates energy from the source while the other records the sky background through an adjacent aperture. The grating was blazed to cover a wavelength range of 3600-5100 A and had a reciprocal dispersion of 75 A mm -1, yielding a resolution of ;::;2.5 A (FWHM). Flat-field spectra were obtained at the start and end of each night, and wavelength calibration was provided by a CuAr arc lamp which was observed at the beginning and the end of each spectrum acquisition. Several spectrophotometric standards such as LDS 678A were observed in order to convert the observed counts to flux. Stars in the northern hemisphere were observed with the Steward Observatory 2.3-m telescope on Kitt Peak. This instrument is equipped with a Boller & Chivens spectrograph and a UY-flooded 800 x 1200 Loral CCD detector. The grating was also blazed to cover the spectral range occupied by the Balmer lines, and gave a resolution of ;::; 8 A (FWHM). The spectra were then flat-fielded and converted to flux using observations of standard stars. A detailed description of thee reduction procedures can be found in Kidder (1991). All data were acquired at the parallactic angle using 1.25- or 2.5-arcsec slits, depending upon the seeing conditions. Fig. 1 shows a representative sample of 18 flux-calibrated spectra obtained in both northern and southern hemisphere programmes. 3.2 Determination of temperature and gravity Comparison of previously published DA temperatures derived from Balmer line fitting reveals that systematic differences in the values determined may occur as a result of using different model spectra. This problem was first noted by Bergeron, Wesemael & Fontaine (1991), who identified the cause as arising from the treatment of the high-lying levels of hydrogen in the model atmosphere calculations. For example, work included in Kidder (1991) used an extended grid of line-blanketed homogeneous LTE models computed by F. Wesemael (Wesemael et al. 1980), whereas Finley, Koester & Basri (1997) employed fully line-blanketed homogeneous LTE models developed by Koester (see Koester 1991 and Finley et a1. 1997). A comparison of these results shows that, while the temperatures obtained by Kidder (1991) are similar to those of Finley et aI., the surface gravities of the former sample are on average 0.08 dex higher (Finley, private communication). The more recent Koester models do take into account higher excitation levels of the H atoms, giving more reliable results, and these are © 1997 RAS, MNRAS 286, 369-383 © Royal Astronomical Society • Provided by the NASA Astrophysics Data System Downloaded from http://mnras.oxfordjournals.org/ by guest on October 6, 2014 The ROSAT satellite, launched in 1990 June, carries two 00aligned imaging instruments, an X-ray telescope (XRT) (Pfefferman et al. 1986) and the Wide Field Camera (WFC) (Sims et al. 1990), covering the EUY waveband. The prime objective of the mission was to carry out the first survey of the entire sky in both the EUY and X-ray during the first 6 months of operation. The EUY survey was conducted in two bands, the S1 filter covering the range 60-140A (90-200 eV), and the S2 filter for 112-200 A (60-110 eV). A complete catalogue of X-ray sources has not yet appeared, but the white dwarf fraction has been published by Fleming et al. (1996), who list the 'C band' (0.1-0.28 ke V) count rates for each detection. Two catalogues of EUY sources have been produced from analysis of the WFC data. The Bright Source Catalogue (BSC) (Pounds et a1. 1993) contains 119 identified white dwarfs, while a second catalogue (2RE) with lower detection thresholds yields only six more (Pye et a1. 1995), although the total number of sources is increased from 383 to 479. Around 35 2RE EUY sources have yet to be identified optically but, as the vast majority (;::;75 per cent) of the identified new sources are active stars, we may expect only a few further white dwarf discoveries. Hence the EUY-selected sample of white dwarfs is drawn predominantly from the BSC. Once EUY sources had been detected in the initial survey analyses, the next step was to identify the optical counterpart, initially through cross-correlation of source positions with astronomical catalogues. When no catalogued counterpart could be found, the EUY source fields were included in a programme of optical spectroscopic follow-up observations (Mason et a1. 1995). From this work ;::;50 new white dwarfs were found. The complete sample is listed in Table 1, ordered by BSC name and including the most commonly used name for those stars already catalogued. Also indicated are a number of objects found in astronomical catalogues but which were not previously known to be white dwarfs (e.g., REJ0237 -12=PHL 1400). 371 1997MNRAS.286..369M 372 M, C. Marsh et ai, Table 1. Epoch-corrected count rates from the ROSAT All-Sky Survey, Alternative Name new ID GD 2 MCT GD 659 GD 683 GD 984 PG 0136 + 251 GD 1401 GD 421 LB 1628 Feige 24 PHL 1400 LB 1663 GD 50 V471 TAU MCT G191 B2B new ID new ID GD 257 new ID GD 71 new ID new ID new ID SIRIUS B GD 80 new ID new ID PG 0824 + 289 new ID new ID new ID PG 0904 +511 new ID new ID PG 1026 +454 new ID EG 070 GD 123 new ID PG 1041 +580 new ID PG 1057 + 719 TON 61 PG 1123 + 189 PG 1125 - 026 PG 1145 + 188 PG 1234 +482 EG 187 HZ 43 new ID GD 336 new ID new ID CD - 3810980 KUV 433-3 new ID PG 1725 +587 new ID KUV 18004+6 new ID BPM 93487 PSPC" 356.0 2520.0 3790.0 3004.0 87.0 1745.0 632.0 32.0 158.0 50.0 14.0 329.0 230.0 2086.0 1110.0 106.0 40.0 971.0 190.0 725.0 373.0 2800.0 4.0 132.0 9.0 1506.0 19730.0 682.0 3536.0 455.0 1833.0 714.0 32.0 2154.0 81.0 61.0 4744.0 286.0 288.0 196.0 9068.0 242.0 294.0 1634.0 223.0 96.0 1192.0 508.0 436.0 50.0 88.0 2050.0 7640.0 37170.0 45.0 153.0 153.0 225.0 969.0 468.0 157.0 1387.0 24.0 1248.0 812.0 1691.0 1423.0 Survey Count Rates (c ks- 1 ) 1"." Sl 1". S2 47.0 37.3 22.5 6.8 121.0 148.9 12.4 50.4 190.0 625.7 35.9 1008.2 30.0 765.5 30.1 2079.1 19.0 43.5 12.1 134.2 88.0 259.4 14.7 656.3 63.0 52.2 6.8 50.6 13.5 24.7 11.0 5.6 20.0 23.3 4.1 49.4 16.0 24.8 4.5 135.1 758.0 4.9 2.3 39.0 55.6 10.2 121.6 68.9 76.0 6.8 210.4 71.0 291.7 16.2 500.7 250.0 340.0 20.0 1040.0 27.0 120.5 10.2 2567.3 62.2 22.0 8.3 2824.1 168.3 12.3 282.1 64.0 34.0 55.7 7.2 205.7 100,4 8.4 13.0 94.4 23.0 41.0 4.2 15.4 113.0 679.0 21.9 2281.3 7.4 3.2 34.0 27.0 10.7 4.3 43.6 6.4 3.2 428.0 66.0 142.0 9.0 255.0 290.0 3200.2 41.8 8187.4 55.0 106.7 8.6 279.8 1127.0 539.3 15.9 867.8 109.5 45.0 9.9 369.9 80:0 297.9 15.3 987.6 53.0 46.2 7.7 52.8 15.0 20.2 5.9 64.2 238.2 16.8 448.2 102.0 23.0 26.9 5.6 37.4 19.0 35.3 47.6 528:0 19.9 116.0 594.4 28.3 27.0 5.9 77.0 30.0 58.6 8.4 137.7 22.0 31.4 4.5 88.7 151.0 1136.5 25.8 2689.8 27.0 35.5 118.0 7.1 57.2 6.7 212.8 26.0 69.0 226.6 11.2 477.9 32.0 49.2 6.7 121.6 16.0 43.1 7.7 128.3 74.0 120.3 19.9 325.0 42.0 90.1 11.7 88.8 183.2 13.0 990.5 33.0 22.7 4.8 42.2 15.0 41.6 100.8 7.1 18.0 30.0 515.6 20.1 1890.2 163.0 1509.5 37.6 4749.8 412.0 12157.6 92.9 36982.6 10.0 10.6 4.7 15.4 22.6 4.5 20.4 25.0 26.0 17.0 3.4 38.7 17.3 7.1 24.0 23.2 54.0 199.5 14.3 605.8 29.0 51.7 7.1 51.4 17.3 36.0 1.0 33.3 28.0 47.7 5.7 14.0 10.0 11.4 1.1 11.8 106.2 18.9 306.0 267.6 237.0 51.1 2.3 21.5 104.5 3.4 112.0 112.0 260.7 15.0 697.6 75.0 1". 9.7 40.3 44.6 15.7 23.6 7.5 7.5 5.0 9.7 30.1 12.9 11.8 19.1 29.0 42.5 40.8 14:2 14.3 9.2 6.2 37.9 5.1 6.2 15.6 15.0 68.7 15.7 23.7 19.0 31.6 10.6 7.8 22.4 8.5 8.5 24.1 8.8 12.2 10.7 37.4 14.3 12.8 18.2 9.6 11.7 20.1 13.2 34.2 7.8 10.0 40.8 54.1 148.4 6.6 6.4 6.4 9.1 22.3 7.1 8.1 1.1 26.8 2.2 4.3 23.9 Mean Julian Date 48179.7 48178.8 48195.9 48223.8 48226.8 48178.8 48182.6 48178.8 48111.1 48178.8 48183.3 48182.6 48179.7 48108.6 48110.6 48115.5 48125.0 48118.6 48122.1 48129.9 48129.5 48129.5 48133.9 48141.8 48146.2 48141.8 48146.2 48149.2 48183.3 4;8159.5 '48159.5 48162.4 . 48215.2 . :4817'3.3 48174.5 48163.4 48204.3 48208.3 48218.9 48175.4 48174.0 48210.8 48175.4 48175.4 48173.0 48165.8 48164.4 48204.3 48210.8 48219.4 48215.7 48211.3 48165.8 48165.8 48208.8 48179.6 48180.6 48113.0 48116.7 48111.6 48114.0 48183.3 48183.3 48133.5 48183.3 48183.3 48146.2 © 1997 RAS, MNRAS 286, 369-383 © Royal Astronomical Society • Provided by the NASA Astrophysics Data System Downloaded from http://mnras.oxfordjournals.org/ by guest on October 6, 2014 WFC Name REJ0003 +43 REJ0007 + 33 REJ0029 - 63 REJ0053 - 32 REJ0108 - 35 REJ0134 -16 REJ0138 +25 REJ0148 - 25 REJ0151 + 67 REJ0230 - 47 REJ0235 + 03" REJ0237 -12 REJ0322 - 53 REJ0348 - 00 REJ0350 + 17 6 REJ0457 - 28 REJ0505 + 52 REJ0512 - 41 REJ0521-1O REJ0550 + 00 REJ0550 - 24 REJQ552 + 15 REJ0558 - 37 REJ0605 - 48 REJ0623 - 37 REJ0632 - 05' REJ0645 -16 REJ0654 - 02 REJ0715 -70 REJ0720 - 31" REJ0723 - 27 REJ0827 + 28 REJ0831- 53 REJ0841 + 03 REJ0902 - 04 REJ0907 + 50 REJ1016 - 05" REJ1019 -14 REJ1024 - 30 REJ1029 +45 REJ1032 + 53 REJ1033 -11 REJ1036 +46 REJ1043 +49 REJI044 + 57 REJ 1058 - 38 REJ1100 + 71 REJ1112 +24 REJ1126 + 18 REJ1128 - 02 REJ1148 + 18 REJI236 +47 REJ1257 + 22 REJI316 +29 REJ1340 +60 REJ1431 + 37 REJ1529 +48 REJ1614 - 08 REJ1623 - 39 REJ1638 + 35 REJ1650 +40 REJ1726 +58 REJ1738 +66 REJ1746 -70 REJ1800 +68 REJ1820 +58 REJ1847 + 01 1997MNRAS.286..369M A UV-selected sample of DA white dwarfs 373 Table 1 - continued WFC Name new ID new ID AOO 2000 - 56 new ID L 210 -114 MCT new ID GD 391 GD 394 new ID new ID MCT new ID new ID new ID PHL 0396 GD 246 new ID MCT PHL 580 PSPC" Survey Count Rates (c ks- 1 ) 1"." S2 Sl 1". 316.0 46.0 404.0 82.0 88.0 27.0 12040.0 430.0 935.0 34.0 200.0 55.0 382.0 39.0 442.0 30.0 9.0 71.0 65.0 1282.0 464.0 48.0 158.0 33.0 1650.0 84.0 6900.0 175.0 223.0 21.0 220.0 38.0 13.0 87.0 20.0 6467.0 114.0 1049.0 82.0 55.0 22.0 25.0 136.0 36.3 19.6 22.8 1274.8 61.8 26.1 61.2 22.3 23.1 418.1 57.1 32.5 166.5 1146.3 18.9 75.3 15.6 39.4 1613.6 261.3 40.6 36.9 8.5 4.6 10.8 45.6 6.7 12.0 9.8 5.6 5.8 13.2 7.8 6.7 13.4 29.0 4.7 12.4 6.7 16.2 35.9 20.7 18.5 9.5 48.8 23.9 41.9 2626.6 33.2 32.5 181.0 18.1 27.2 1411.3 98.2 85.5 181.3 2673.7 75.0 133.8 1104.8 70.9 4711.3 656.8 364.1 64.2 1". 11.4 5.4 12.6 72.3 . 6.4 1.0.~ 18·.9 7.6 23.4 lO.7 10.7 12.8 43.7 11.0 12.8 30.9 14.2 57.8 29.3 27.3 12.2 Mean Julian Date 48142.3 48192.7 48153.4 48153.4 48176.4 48155.2 . 48157.7 48172.0 48195.9 48220.4 48173.5 48174.5 48172.5 48171.3 48210.8 48176.4 48172.5 48197.4 48219.4 48192.7 48197.4 48215.2 Notes *Count S-l in the 0.1-0.28 keY band. 'DAD stars. 'SI rate is PV/Cal data. bRates are after subtraction of K-star component and removal of eclipses (see Barstow et al. 1992). CWFC rates derived from 2RE catalogue (Pye et al. 1995). dpSPC rate is after subtraction of a dMe companion, and should be treated as an upper limit (see Barstow et al. 1995a). '-' in the error column indicates that the count rate is an upper limit. used in the current work. Consequently, as part of our programme, we have reanalysed 48 spectra for which Teff and logg have already been published (e.g. Kidder 1991; Barstow et al. 1993) to ensure that we have an internally consistent set of values for the subsequent detailed study of this white dwarf sample. DA white dwarfs are defined as having no detectable helium in their optical spectrum. However, a potentially important effect that may influence the results of Balmer line fitting is discussed by Bergeron et al. (1994). They have shown that for DA stars, homogeneous helium abundances of He!H ~ 10-5 -10- 4 , which yield no detectable He II absorption at either 4686 or 1640 A, can significantly reduce the derived effective temperature compared to that for a pure hydrogen atmosphere. For example, a 52000 K DA with a homogeneous helium abundance of He!H=10- 4 could be interpreted as having Teff = 60 000 K if it were assumed to be pure H. This effect is even stronger at higher temperatures, giving a discrepancy as high as 25 per cent around 80 000 K. If these levels of He were really present in the hot DAs, the net effect would be a non-linear compression of the temperature scale towards the hot end. Nevertheless, comparisons between individual objects would still be valid unless the abundances of helium varied considerably at a given T eff . However, conversely, Napiwotzki (1995) has demonstrated that the effect trace helium has on the Balmer lines in the work of Bergeron et al. (1994) may be exaggerated due to the assumption of LTE. Also, observations of hot white dwarfs with EUVE (e.g. Barstow, Holberg & Koester 1994a,b, 1995) have shown evidence for a significant He abundance in only one DA, the exceptionally massive star GD50 (Vennes, Bowyer & Dupuis 1996). This suggests that helium has a minimal influence on the hot DA temperature measurements and that, on this evidence, pure H models should be appropriate for determining the DA temperature scale. On the other hand, both EUVE and IUE spectra (e.g. Vennes et al. 1992; Holberg et al. 1993; Dupuis et al. 1995; Barstowet al. 1996) do show evidence for significant abundances of elements heavier than helium in a number of the very hot DAs. These elements may well have a similar effect to helium on the Balmer lines. This is being investigated, but the evidence is conflicting. Driezler & Werner (1993) found that with non-LTE model atmospheres, the inclusion of line blanketing from Fe-group elements had a very small effect on the hydrogen line profiles. Alternatively, Lanz & Hubeny (1995) found that the inclusion of heavy elements in their non-LTE models did have a significant effect on the profiles. Koester (1996) finds that heavy-element blanketing in LTE models could lower the measured effective temperature of the hot DA G191-B2B by ~5000 K, while a recent non-LTE study of this star, including an increased level of Fe line blanketing, shows that the possible effect on the accuracy of temperature determination could be as large as © 1997 RAS, MNRAS 286, 369-383 © Royal Astronomical Society • Provided by the NASA Astrophysics Data System Downloaded from http://mnras.oxfordjournals.org/ by guest on October 6, 2014 REJ1847 - 22 REJ1943 + 50 REJ2004 - 56 REJ2009 - 60 REJ2013 + 40%" REJ2018 - 57 REJ2024 - 42 REJ2024 + 20 REJ2029 + 39 REJ2112 + 50 REJ2127 - 22 REJ2154 - 30 REJ2156 -41 REJ2156 - 54 REJ2207 +25 REJ2210 - 30 REJ2214 -49 REJ2244 - 32 REJ2312 + 10 REJ2324 - 54 REJ2334 - 47 REJ2353 - 24 Alternative Name 1997MNRAS.286..369M 374 M. C. Marsh et al. 20 , - - - - - , - - - - - - , - - - - - , - - - - - - , - - - - - , - - - - - - , - - - - - - , - - - - - , 18 REJ0029 16 --.....---:..-.-_~_ 14 REJO 138 REJ0623 REJll00 12 " . - - - - - - - _ REJ1148 >< ;:J ",------_-lREJ 1257 Q) :> ...... ......, ro ~---------- REJ/1~3~___ 10 Q) 0:: 8 ~-""""-_ _-1REJ1417 -!REJ1638 ,r-'~-_ _ _ 6 REJ1726 EJ1800 , . - - - -_ _ _ REJ1847 4 REJ2024. 2 REJ2112 o L -____ 3600 ~ ____- L_ _ _ _ 3800 4000 ~ _ _ _ _ _ _L -_ _ _ _ 4200 4400 ~ ____ 4600 ~ ______ 4800 ~ 5000 __ ~ 5200 Wavelength Figure 1. Representative optical spectra, from both northern and southern hemisphere progranlffies, for the sample of white dwarfs detected by the ROSAT WFC during the All-Sky Survey. about 10 per cent (Lanz et al. 1996). In this analysis, we only use pure H models and, consequently, temperature values for stars which contain significant quantities of heavy elements may have uncertainties which are much larger than the formal errors. This potential problem will only apply to stars with effective temperatures in excess of 55 000 K We have used the homogeneous and stratified grids of H + He model atmospheres provided by Detlev Koester (e.g. Koester 1991; Finley et al. 1997). These span the temperature range 20000-100000 K and logg from 7.0 to 9.0. Except for those stars known to be DAO white dwarfs, the analyses were carried out in the limit of negligible helium abundance or large hydrogen layer mass (MH ), equivalent to pure H atmospheres. For most of the stars, H p, H y, H~ and H.lines were included in the analysis. However, several of the observations obtained by Kidder (1991) spanned only Hy and H~ lines. Our spectral fitting technique is described in several earlier papers (e.g. Barstow et al. 1994c), but, as the results presented in this paper rely on it completely, we repeat the details here. The analysis was performed using the program XSPEC (Shafer et al. 1991), which adopts a X2 minimization technique to determine the model spectrum giving the best fit to the data. Models were convolved with the appropriate instrument resolution, and all the lines included were fitted simultaneously, with an independent normalization constant applied to each, ensuring that the result was independent of the global slope of the continuum and reducing the effect of any systematic errors in the flux © 1997 RAS, MNRAS 286, 369-383 © Royal Astronomical Society • Provided by the NASA Astrophysics Data System Downloaded from http://mnras.oxfordjournals.org/ by guest on October 6, 2014 ~ 1997MNRAS.286..369M A UV-selected sample of DA white dwarfs calibration of the spectrum. The SAAO RPCS, used to record the southern hemisphere spectra, is a photon-counting instrument. Consequently, the data errors can be determined simply from the Poisson statistics associated with the source and background counts in each spectral bin. However, in dealing with the CCD spectra obtained at the Steward Observatory, it is difficult to determine errors on individual points, and we take the following approach. First, an initial fit to the data is performed without any errors included. From the scatter on the residuals between the best-fitting model we estimate the average errors on the data points, which are included in a subsequent fit. In addition, for both RPCS and CCD data, we correct for any systematic deviation in the spectral slope between model 375 and data, arising from errors in the flux calibration in the locality of each line. With the CCD data, a second spectral fit results in no change to the best-fitting parameters but, with the inclusion of errors on the data points, yields a lower reduced X2 • To make sensible estimates of the uncertainty in the fitted parameters, the value of the reduced X2 should be less than ~ 2, and this is usually achieved with a single iteration of the steps outlined. Two examples of the procedure are shown in Fig. 2 including two contrasting objects, the very hot DA REJ0029 - 63, observed at SAAO, and the cooler star REJ0428 + 16 from the Steward sample. Errors on Teff and logg were determined by allowing the model parameters to vary until the (jX2 reached the value corresponding to the 10" (a) I I CJ "" ...".:: ~ x til V ::s0 CJ . ~ '0 I " '"~ y ...0Ei z 0 '"I ~ 'iii" ::s '0 '!il 10 0 10 I "... - --\j~~ !~~~----\I~J~---------------l~b"~~----- 3600 j~I II I 4000 4200 4400 4600 Wavelength (1\.) ~ jI ~ I I II 4600 5000 (b) i « i CJ '"I0 ....x '" "" '" ..." IV I .:: ::s 0 0 .... x ., ~ til CJ '0 " '" .!:! 'iii Ei... I 0 .... 0 Z 0 '"I ~ 'iii" ::s '0 '!il 10 0 10 I "... 3600 4000 4200 4400 4600 Wavelength (1\.) 4600 5000 Figure 2. Example spectral fits for two stars where all Hp-H.lines are available. (a) REJ0148 - 25: Teff=24 540 K, logg=7.84 (x; = 1.68); (b) REJ0029-63: Teff =60595K, logg=7.97 (X;=1.74). © 1997 RAS, MNRAS 286, 369-383 © Royal Astronomical Society • Provided by the NASA Astrophysics Data System Downloaded from http://mnras.oxfordjournals.org/ by guest on October 6, 2014 V i :::: '" 1997MNRAS.286..369M 376 M C. Marsh et al. Table 2. Summary of derived stellar parameters. Target 16.82 13.85 15.314 13.37 14.76 13.96 15.87 14.69 14.41 14.79 12.56 14.92 14.83 14.04 13.65 13.951 11.73 17.257 15.815 14.773 16.371 13.032 14.369 15.825 12.089 15.537 8.35 14,82 141178 14.87 14,60 14:22 14.65 14[475 13,190 16 154 14.21 14,93 16.67 16.13 14.455 13.012 14~34 161234 14,64 13:78 14:68 15.773 14.127 15.73 14.33 14.38 13.38 12.99 16.941 15.277 15.083 14.013 11.00 14.83 15.831 15.45 14.606 16.60 14.74 13.949 12.95 13.720 err. T'II (K) I". bounds log g (em s-2) 0.30 0.01 0.011 0.02 0.30 0.03 0.03 0.30 0.02 0.30 0.05 0.30 0.3 0.02 0.05 0.009 0.01 0.013 0.028 0.024 0.012 0.009 0.006 0.017 0.001 0.015 0.1 0.03 0.015 0.04 0.30 0.03 0.30 0.D28 0.020 0.2 0.05 0.30 0.30 0.03 0.020 0.009 0.05 0.D28 0.05 0.30 0.2 0.028 0.028 0.30 0.30 0.05 0.02 0.05 0.020 0.016 ·0.028 0.028 0.01 0.05 0.D28 0.05 0.028 0.30 0.04 0.028 0.05 0.D28 46205 47936 60595 34684 28580 44850 38964 24540 30120 63400 62947 31290 32860 42373 34200 58080 57340 53960 31770 45748 51870 32008 70275 33040 62280 41765 24700 32280 44300 53630 37120 51934 29330 36605 23310 33459 53827 31340 36610 34224 44980 22790 28766 47560 29016 27970 39555 38970 55640 31280 25107 55570 37880 49000 42970 34419 46230 38500 24760 36056 37850 54550 880lO 5lO50 43701 45330 28744 31920 44476 - 47850 46839 - 49086 58130 - 63800 34382 - 34998 28320 - 28835 42786 - 48296 38458 - 40708 24300 - 24765 29874 - 30338 62680 - 67280 61481 - 64336 30970 - 31710 32445 - 33135 41577 - 43473 33600 - 34800 55875 - 60170 56040 - 58670 51400 - 58435 31090 - 32160 44360 - 47405 48750 - 56400 31962 - 32287 66400 - 78780 32260 - 34370 60590 - 64140 40330 - 44420 23700 - 25700 31740 - 32840 43310 - 45445 52400 - 54525 36390 -. 37860 48457 - 56007 28970 - 29590 36180 - 37015 22560 - 24120 32882 - 34239 52629 - 55931 30895 - 31680 35780 - 38360 33889 - 34740 44210 - 45310 22570 - 23030 28349 - 29236 46550 - 48630 28750 - 29304 27680 - 28200 38747 - 40503 38620 - 39780 54770 - 56940 30840 - 31770 24764 - 25489 54540 - 56720 37209 - 38584 47000 - 51000 42030 - 44240 33582 - 35135 45540 - 47000 38160 - 38890 24333 - 24805 35479 - 36662 36800 - 38430 49616 - 58835 85400 - 90400 48410 - 53930 42700 - 44759 44600 - 46290 28485 - 29017 31540 - 32260 8.85 7.77 7.97 7.89 7.90 7.96 9.00 7.84 7.70 7.43 7.53 8.44 7.66 9.00 8.80 7.90 7.48 7.62 8.70 7.79 7.29 7.70 7.37 7.80 7.22 8.51 8.65 8.34 7.69 7.64 7.75 8.00 7.79 7.69 7.74 7.86 8.08 7.79 8.69 7.85 7.68 7.57 7.92 7.62 7.79 7.88 7.66 7.91 7.62 8.11 7.81 7.57 7.69 7.70 7.68 7.66 7.70 7.85 7.92 7.71 7.95 8.49 7.79 8.84 7.80 7.73 7.72 8.00 I". I". bounds 8.73 7.67 7.76 7.82 7.83 7.70 8.92 7.79 7.62 7.25 7.45 8.35 7.62 8.90 8.52 7.78 7.34 7.33 8.61 7.66 7.02 7.63 7.13 7.57 7.13 8.35 8.35 8.23 7.62 7.54 7.65 7.73 7.75 7.64 7.64 7.70 7.98 7.70 8.50 7.76 7.64 7.53 7.79 7.52 7.71 7.81 7.54 7.81 7.54 8.00 7.73 7.49 7.59 7.50 7.57 7.52 7.62 7.79 7.88 7.61 7.81 8.25 7.70 8.68 7.68 7.68 7.65 7.91 - 8.97 7.86 8.16 7.96 7.97 8.32 9.10 7.92 7.77 7.67 7.62 8.53 7.78 9.10 9.08 8.07 7.58 7.86 8.82 7.90 7.59 7.13 7.50 8.26 7.40 8.65 8.95 8.45 7.80 7.74 7.85 8.22 7.89 7.76 7.83 7.99 8.17 7.89 8.80 7.93 7.77 7.60 8.06 7.70 7.87 7.95 7.78 7.96 7.69 8.21 7.87 7.64 7.79 7.90 7.84 7.82 7.75 7.90 8.09 7.80 8.12 8.77 7.93 9.lO 7.92 7.81 7.79 8,09 Mass(err.) (Me) 1.16 (0.07) 0.58 (0.04) 0.70 (0.10) 0.61 (0.03) 0.59 (0.03) 0.66 (0.17) 1.22 (0.05) 0.55 (0.02) 0.51 (0.03) 0.52 (0.07) 0.54 (0.03) 0.91 (0.05) 0.50 (0.05) 1.23 (0.05) 1.12 (0.04) 0.66 (0.07) 0.51 (0.04) 0.55 (0.10) 1.07 (0.06) 0.59 (0.05) 0.44 (0.07) 0.51 (0.03) 0.52 (0.07) 0.56 (0.20) 0.46 (0.05) 0.97 (0.07) 1.03 (0.20) 0.85 (0.07) 0.54 (0.04) 0.55 (0.03) 0.55 (0.04) 0.70 (0.13) 0.54 (0.05) 0.52 (0.03) 0.50 (0.04) 0.59 (0.08) 0.65 (0.05) 0.55 (0.04) 1.06 (0.10) 0.58 (0.04) 0.54 (0.03) 0.43 (0.01) 0.61 (0.07) 0.53 (0.03) 0.54 (0.04) 0.58 (0.04) 0.52 (0.05) 0.63 (0.05) 0.55 (0.03) 0.71 (0.06) 0.54 (0.04) 0.53 (0.03) 0.53 (0.04) 0.56 (0.07) 0.54 (0.06) 0.50 (0.06) 0.55 (0.03) 0.60 (0.03) 0.60 (0.09) 0.53 (0.04) 0.64 (0.09) 0.97 (M6) 0.65 (0.21) 1.16 (0.12) 0.59 (0.05) 0.56 (0.03) 0.51 (0.03) 0.65 (0.05) © 1997 RAS, MNRAS 286, 369-383 © Royal Astronomical Society • Provided by the NASA Astrophysics Data System Downloaded from http://mnras.oxfordjournals.org/ by guest on October 6, 2014 REJ0003 + 43REJ0007 +33 REJ0029 -63 REJ0053 - 32 REJ0108 - 35' REJ0134 -16 REJ0138 +25 REJ0148 - 25' REJ0151 + 67 REJ0230 - 47REJ0235 + 03 REJ0237 - 12' REJ0322 -53 REJ0348 -00 REJ0350 + 17 REJ0457 -28 REJ0505 + 52 REJ0512 -41 REJ0521-10 REJ0550 + 00 REJ0550 -24 REJ0552 + 15 REJ0558 - 37 REJ0605 -48 REJ0623 - 37 REJ0632 - 05 REJ0645 -16 REJ0654 - 02 REJ0715 -70 REJ0720 - 31 REJ0723 - 27" REJ0827 + 28 REJ0831 - 53REJ0841 + 03 REJ0902 - 04 REJ0907 + 50 REJ1016 - 05" REJ1019 - 14' REJ 1024 - 30" REJI029 +45 REJlO32 +53 REJ1033 -11 REJlO36 +46 REJI043 +49 REJlO44 + 57 REJ1058 - 38" REJ1100 + 71 REJ1112 +24 REJ1126 + 18 REJ1128 - 02REJ1148 + 18" REJ1236 +47 REJ1257 +22 REJ1316 + 29! REJ1340 + 60 REJ1431 + 37 REJ1529 +48 REJ1614 - 08 REJ1623 - 39 REJ1638 + 35 REJ1650 +40 REJ1726 + 58 REJI738 + 66 REJ1746 - 70' REJ1800 +68 REJ1820 +58 REJ1847 + 01 REJ1847 - 22 mv 1997MNRAS.286..369M A UV-selected sample of DA white dwarfs 377 Table 2 - continued Target 14.62 15.05 13.59 14.6 13.61 14.74 16.59 13.38 13.08 14.66 14.17 15.38 14.44 14.58 14.79 11.708 15.66 13.09 15.197 13.441 15.444 I". err. 10bounds TefJ (K) 0.30 33500 0.30 44456 0.30 44200 0.60 47057 0.30 26579 0.30 28597 0.30 50564 0.01 22153 0.02 38866 0.30 48297 0.30 28741 0.30 49764 0.30 45860 24610 0.30 28268 0.30 0.007 65600 31692 0.30 0.01 57380 0.017 45860 0.007 56682 0.014 29120 33330 43381 43650 45680 26034 28230 49332 21749 38139 46455 28363 46513 44600 24490 27823 63790 31312 56190 44490 54883 28720 - 33690 - 45327 C" 44800 - 48230 - 26892 - 28988 - 51810 - 22560 - 39634 - 49934 - 29152 - 53131 - 47200 - 24690 - 28640 - 68510 - 32188 - 58670 - 47320 - 59500 - 29620 log g (cms- 2 ) 7.86 7.54 8.14 7.74 7.78 8.54 7.96 7.79 7.84 7.69 8.18 7.75 7.74 8.16 7.60 7.42 8.07 7.86 7.73 7.64 8.14 I". bounds 7.82 7.43 8.03 7.49 7.72 8.38 7.83 7.72 7.74 7.56 8.08 7.57 7.64 8.14 7.53 7.24 7.96 7.78 7.61 7.57 8.01 - 7.90 7.66 8.22 8.04 7.85 8.67 8.06 7.87 7.94 7.82 8.26 7.92 7.83 8.20 7.72 7.57 8.17 7.93 7.82 7.78 8.21 Mass(err) (Me) 0.59 (0.02) 0.49 (0.04) 0.76 (0.06) 0.57 (0.12) 0.53 (0.03) 0.97 (0.10) 0.67 (0.06) 0.52 (0.04) 0.59 (0.05) 0.55 (0.05) 0.75 (0.06) 0.58 (0.07) 0.57 (0.04) 0.73 (0.02) 0.46 (0.04) 0.52 (0.05) 0.69 (0.06) 0.64 (0.03) 0.56 (0.05) 0.56 (0.05) 0.73 (0.08) Notes. *mv determined from spectral flux. ·Vmagnitude is taken from Tweedy et al. 1993. bValues for HZ43 are from Napiwotzki et al. 1993. CMCf survey (McCook & Sion 1996, in preparation) give V = 14.87, B - V = - 0.18, U-B= -0.97 dMCf survey (McCook & Sion 1996, in preparation) give V = 15.89, B - V = - 0.30, U-B= -1.14. level (bX 2 = 1.0). The errors quoted include only statistical uncertainties and do not take into account any possible systematic effects related to the model or observed spectra. The values obtained for Teft and logg from the profile fitting for the whole sample of 89 objects are given in Table 2. The 1CT errors are shown as lower and upper bounds to each parameter. Also included her~ are the adopted values for mv for each star in the sample (see Section 4). Where several photometric observations were available, including our results and published photometry (available for 47 stars), a weighted average was taken. Objects for which mv was estimated from the optical spectrum are indicated. 4 OPTICAL PHOTOMETRY The simple H + He model atmospheres used to interpret X-ray and EUV data are specified by five main parameters. These are the effective temperature Teft , the surface gravity, the interstellar H I column (NH ), the atmospheric helium abundance (or the mass of the hydrogen layer M H , for a stratified model) and a distance/radius related normalization constant. Since only three independent data points are available from the ROSAT survey, additional information is needed to be able to model the ROSAT data. The optical V magnitude (mv) may be used as a convenient normalization constant, while Teft and logg can be derived from the optical spectra as described above. Magnitude measurements in other bands are equally useful, although it is mv that is most often available for catalogued white dwarfs. To provide optical photometry for the newly discovered white dwarfs and those known stars for which such measurements had not been published, an extensive programme of observations has been undertaken to obtain coverage in the U, B and Vbands. Since the ROSAT survey covers the entire sky, these observations required time on telescopes in both heInispheres. Data were obtained for 18 stars in the southern heInisphere and 25 stars in the north. The northern heInisphere stars were observed over a twoweek period from 1994 March 28 to April 9 with the 1.0-m Jacobus Kapteyn telescope (JKT) using an EEV7 CCD chip placed at the Cassegrain focus, and a set of Harris UBVRI filters. To increase readout speed, the chip was windowed to 400 x 400 pixel as opposed to the full 1200 x 1200 pixel. Standards were adopted from the Kitt Peak (Landolt 1983) and RGO (Argyle et al. 1988) surveys which covered a range of airmass, 1.0 <X < 2.2. In the main, these standards are redder than the white dwarf programme stars and while some blue standards could be observed, giving a complete range in colour of - 0.336 < (B - V) < 1.1 and "':"'1.245 < (U - B) < 1.1, the observations were not made under completely photometric conditions. As a result, a small systematic error was introduced into the determination ofmv. All the CCD data were reduced using standard techniques, after debiasing· and flat-fielding each frame. Standard stars at a range of airmass were observed throughout each night and used to determine atmospheric extinction. Since.the character of the atmosphere is not uniform and, at a given instant, the extinction in various parts of the © 1997 RAS, MNRAS 286, 369-383 © Royal Astronomical Society • Provided by the NASA Astrophysics Data System Downloaded from http://mnras.oxfordjournals.org/ by guest on October 6, 2014 REJ1943 + 50' REJ2004 - 56' REJ2009 - 60' REJ2013 +40 REJ2018 - 57' REJ2024 - 42*' REJ2024 + 20' REJ2029 +39 REJ2112 + 50 REJ2127 - 22' REJ2154 - 30' REJ2156 - 41'd REJ2156 - 54' REJ2207 + 25' REJ2210 - 30' REJ2214 - 49 REJ2244 - 32' REJ2312 + 10 REJ2324 - 54 REJ2334 -47 REJ2353 - 24 mv 1997MNRAS.286..369M 378 M C. Marsh et al. sky may differ, all the standard observations from different nights were grouped together to give a mean extinction curve for the whole observing run. Non-uniformity of the extinction leads to a scatter in the residuals between the data points and extinction curve. This scatter can be used as a measure of the internal uncertainty for the magnitudes ( ± 0.03 mag), and dominates the statistical errors. The derived stellar magnitudes and colour indices were transformed to the standard Johnson photometric system using the equations of Hardie (1962). In the southern hemisphere, photoelectric photometry was obtained as part of a continuing general photometry programme during 1993-1994 using the VeT photometer, with GaAs photomultiplier, on the SAAO 0.75-m telescope, based at Sutherland, South Africa. The data were reduced using similar techniques (but dealing with photomultiplier rather than CCD data), using observations of E-region standards to transform the observations to the standard Kron-Cousins system (Menzies et al. 1989). The Cousins system gives magnitudes which are equivalent to those obtained using that of Johnson for UBVbands. The combined results for both northern and southern programmes are listed in Table 3, with the respective observatories indicated. As a consistency check for the JKT pro- Table 3. UBV photometry of ROSAT DAs. V 15.314 16.444 13.912 13.827 17.257 15.867 16.198 14.768 16.371 13.043 14.369 15.825 12.089 15.537 15.690 14.178 14.475 13.190 15.747 14.306 14.455 13.028 16.955 16.234 16.781 15.773 14.127 16.809 16.941 15.934 15.272 15.216 15.779 15.083 14.013 15.831 14.606 13.949 13.720 11.708 15.197 13.441 15.444 loB-V 0.011 -0.352 O.OlD -0.307 0.009 -0.359 0.007 -0.258 0.013 -0.266 0.015 -0.215 0.017 -0.296 0.028 -0.298 0.012 -0.351 0.028 -0.239 0.006 -0.331 0.017 -0.214 0.016 -0.329 0.015 -0.262 0.016 -0.306 0.015 -0.351 0.028 -0.305 0.020 -0.168 0.020 -0.265 0.028 -0.033 0.020 -0.276 0.028 -0.181 0.028 -0.333 0.028 -0.504b 0.028 -0.470 b 0.028 -0.291 0.028 -0.280 0.028 -0.396 0.020 -0.289 0.028 -0.379 0.028 -0.305 0.028 -0.151 0.020 0.695 0.028 -1.476 b 0.028 -0.295 0.028 -0.251 0.028 -0.343 0.028 -0.279 0.028 -0.212 0.007 -0.341 0.017 -0.344 0.007 -0.296 0.014 -0.066 U-B 100.015 -1.148 0.021 -1.249 0.011 -1.233 0.019 -1.024 0.017 -0.966 0.015 -1.183 0.023 -1.036 0.022 -1.086 0.017 -1.059 0.022 -1.149 0.003 -1.194 0.024 -1.177 0.021 -1.204 0.020 -1.225 0.021 -1.246 0.020 -0.230 0.022 -1.175 0.016 -0.951 0.016 -1.192 0.022 -1.554 b 0.016 -1.195 0.022 -0.983 0.022 -1.189 0.022 -1.130 0.022 -1.018 0.022 -1.145 0.022 -1.187 0.022 -1.227 0.016 -1.040 0.022 -1.256 0.022 -1.117 0.022 -1.101 0.016 0.206 0.022 -0.863 0.022 -1.141 0.022 -1.269 0.022 -1.250 0:022 -1.194 0.022 -1.123 0.008 -1.205 0.021 -1.161 0.010 -1.168 0.022 -1.175 100.014 0.015 0.009 0.010 0.021 0.013 0.021 0.035 0.015 0.035 0.002 0.023 0.018 0.017 0.018 0.029 0.035 0.025 0.025 0.035 0.025 0.035 0.035 0.035 0.035 0.035 0.035 0.035 0.025 0.035 0.035 0.035 0.025 0.035 0.0315 0.035 0.035 0.035 0.035 0.006 0.020 0.009 0.023 Obs. SAAO SAAO SAAO SAAO SAAO JKT,SAAO SAAO JKT SAAO JKT SAAO SAAO SAAO SAAO SAAO SAAO (bin) JKT JKT JKT JKT JKT JKT JKT JKT JKT JKT JKT JKT JKT JKT JKT JKT JKT (bin) JKT JKT JKT JKT JKT JKT SAAO SAAO SAAO SAAO Notes "WD not included in the full sample of 89 objects. bUnreasonably large value, probable error on magnitude measurement. (bin) Red colours, possible binary. © 1997 RAS, MNRAS 286, 369-383 © Royal Astronomical Society • Provided by the NASA Astrophysics Data System Downloaded from http://mnras.oxfordjournals.org/ by guest on October 6, 2014 Target REJ0029 - 63 REJ0415 - 40 a REJ0503 - 28 a REJ0512 - ooa REJ0512 - 41 REJ0521 -10 REJ0534 - 02 a REJ0550 + 00 REJ0550 - 24 REJ0552 + 15 REJ0558 - 37 REJ0605 - 48 REJ0623 - 37 REJ0632 - 05 REJ0632 - 57 a REJ0715 -70 REJ0841 + 03 REJ0902 - 04 REJ0957 + 85 a REJ1016 - 05 REJ1032 + 53 REJ1033 -11 REJ lD43 + 44 a REJ1043 +49 REJ1059 + 51 a REJ1112 + 24 a REJ1126 + 18 REJ1128 + 17 a REJ1340 + 60 REJ1425 + 53REJ1431 + 37 REJl440 + 75REJ1501 + 30 a REJ1529 +48 REJ1614 - 08 REJ1650 +40 REJ1738 + 66 REJ1820 + 58 REJ1847 - 22 REJ2214 - 49 REJ2324 - 54 REJ2334 - 47 REJ2353 - 24 1997MNRAS.286..369M A UV-selected sample of DA white dwaifs gramme a white dwarf standard, GD 71, was observed on one night. GD 71 has mv= 13.032 (Landolt 1992) and we obtain a value of mv= 15.053, a difference of 0.02 mag, within our predicted systematic error. Furthermore, the star REJ0521 - 10 was observed at both the JKT and at the SAAO. REJ0521 - 10 has a visual magnitude of 15.815 ± 0.028 and 15.889 ± 0.018 from the JKT and SAAO work respectively, in agreement to within 20'. The value given for REJ0521 - 10 in Table 3 is a weighted average. In the sample of stars observed photometrically, it was not possible to obtain spectra for 13 objects, which were generally too faint to observe with the 2.0-m class of instruments at our disposal. Consequently, they are not included in our overall survey of the properties of the EUY-selected white dwarfs but are listed in Table 3 for completeness, as no previously published magnitudes exist. Not all the stars for which spectroscopic data were available could be included in the photometric programme. However, useful flux information can be extracted from the spectra and used to provide an estimate of mV. although the photometric accuracy of the data will not be as reliable, as fewer standard stars were used and the quality of sky conditions allowed not so high. The adopted procedure is outlined below. Stellar flux is related to a magnitude measurement by the standard relation 10g/;.(mJ = - OAmx + log/;.(O), stars observed at Steward. Interestingly, a sample of 12 stars from the SAAO showed no similar systematic error. Combining these with the corrected Steward values for V. (Fig. 3b) then enables an estimate of the statistical error from the scatter around the V. = Vp line, yielding 0' = 0.3. This is the value of the error assigned to all the magnitudes estimated from the spectral data and included in Table 2. 5 THE MASS DISTRIBUTION OF THE EUVSELECTED SAMPLE The sample of DA white dwarfs studied in this work were selected purely on the basis of detection in the EUY by the WFC during the ROSAT all-sky survey. A bias is clearly introduced in that only stars with temperatures in excess of ~25 000 K, above which the EUV emission becomes significant, are included. It is important to investigate any effects of this selection by comparing the sample with other studies, (a) 1:1 c 's, o Cl rn o ....... Cl Q) where/;.(O) is the flux for a magnitude 0.0 star in the waveband considered. Values of this constant can easily be obtained from standard tables (e.g. Zombeck 1990) for all bands of the Johnson system. While we adopt mv as the normalization constant for analysis of EUY and X-ray fluxes, the optical spectra obtained do not cover the wavelength range of the V band but do include the B band. Consequently, it is possible to determine the value m B and then estimate mv using B - V measurements obtained from those stars for which we do have photometric data. At the temperatures we are considering, the B - V colour depends primarily on Teff and depends only weakly on gravity (see Cheselka et al. 1993). As Teff has already been determined from the Balmer line profiles for all the stars of interest, it is straightforward to determine B - V from the sample of Cheselka et a!., which is larger and, therefore, statistically more reliable than ours. This procedure was carried out for 25 stars where accurate photometry was not available. Since photometric observations are available for a significant fraction of the stars for which we have spectral data, a direct test of the reliability of the mv estimates can be made by applying the technique to these objects. Fig. 3(a) shows the comparison for 27 stars observed at the Steward Observatory and indicates that the magnitudes determined from the spectra ev,.) are systematically fainter than the photometric values (Vp). This effect does not appear to be magnitude dependent. Keeping a gradient of 1, a linear fit to the data gives a constant shift in the V. axis of 0.15 compared to the Vp = V. line. This is then an indication of the systematic error which should be subtracted from all values of V. for .... ...+ p., .... (1) ~ > V (Photometric) (b) 1:1 +' , c 's, o Cl rn o .bCl Q) .... p., .... ~ > V (Photometric) Figure 3. (a)V, against Vp for 27 stars observed at the Steward Observatory, showing a systematic error in V,. (b) Magnitudes from Steward after a systematic correction of - 0.15 mag has been applied to v" plus 12 stars from the SAAO. Errors on Vp are indicated. Errors on V, for the SAAO objects are from detector counting statistics. © 1997 RAS, MNRAS 286, 369-383 © Royal Astronomical Society • Provided by the NASA Astrophysics Data System Downloaded from http://mnras.oxfordjournals.org/ by guest on October 6, 2014 4.1 Deducing mv from the spectrum 379 1997MNRAS.286..369M 380 M C. Marsh et al. 6 DISCUSSION To place the mass and gravity distributions into context it is necessary to compare them with those obtained from optically selected samples. Only then can we assess the possible biases that might arise from the fact that our sample is collected on the basis of EUV emission. There are several studies of mass distributions of large samples of DA white dwarfs available in the literature. The majority of these have been summarized by BSL, who compare them with their own findings. Most of the earlier work is dominated by objects below Teff ~ 15 000 K and also relies on photometric methods for deriving temperature and gravity. BSL show that for Teff > 15 000 K optical Balmer line fitting provides a far more accurate way of determining Teff and logg, justifying the method used in subsequent studies, including ours. Reid (1996) has used new gravitational redshift measurements to obtain the masses of 34 field (non-cluster) white dwarfs in binary systems. These masses, which are independent of spectroscopic mass determination methods, yield a mean mass of 0.583 ± 0.078 Mo. This is in very good agreement with a redetermination of the BSL spectroscopic masses using the thick H layer Wood models by Bergeron, Liebert & Fulbright (1995), who find a mean of 0.590 ± 0.134 Mo. The study of BSL remains the most accurate and complete sample with which to compare the EUV-selected 25 ~ 25 .. "': ,, , ,, '- 20 20 15 >'-' <= >- ~ 15 '-' <= Q) t':? cr ,, ::l U. --, ~ 10 10 5 5 , , 0.4 0.6 9.5 0.8 Mass (Solar Units) Figure 4. Distribution of white dwarf mass, determined from the evolutionary models of Wood (1992), for a1l89 stars in the EUVselected sample and all 129 stars in that of BSL. The BSL data are scaled by the ratio of the sample sizes for ease of comparison. log g Figure 5. Distribution of surface gravity for all 89 stars in the EUV-selected sample and all 129 stars in that of BSL. The BSL data are scaled by the ratio of the sample sizes for ease of comparison. © 1997 RAS, MNRAS 286, 369-383 © Royal Astronomical Society • Provided by the NASA Astrophysics Data System Downloaded from http://mnras.oxfordjournals.org/ by guest on October 6, 2014 which may have other biases. One way to do this is by comparing mass distributions. Once the temperature and gravity are known, the mass can be found from a theoretical mass-radius relation. Koester & SchOnbemer (1986) demonstrated that, when dealing with hot white dwarfs, the effects of temperature, which cause departures from the zero-temperature mass-radius relation (Hamada & Salpeter 1961), should be taken into account. This point is reiterated in the work of BSL. Failure to do this will lead to underestimates of the white dwarf mass, because the stellar radius is assumed to be lower than the true value. In their analysis, BSL make use of the evolutionary models of Wood (1992) for DAs with pure carbon cores surrounded by an envelope with composition log (MHe/M *) = - 4. Subsequently, this work has been extended to deal with thick (10- 4 Mo) H layers, and these new models, spanning a mass range of 0.4-1.1 Mo in steps of 0.1 M o ' have been employed in our work (Wood 1995). A number of stars appear to lie outside this range, and their masses were estimated by extrapolating the models. The complete ROSAT sample of 89 stars has a mean logg of 7.899 and mean mass of 0.644 Mo. Mass and logg distributions are shown in Figs 4 and 5 respectively, with the data of BSL, renormalized by the ratio of the sample sizes (WFC = 89, BSL = 129) to place them on the same scale, included for comparison. 1997MNRAS.286..369M A UV-selected sample of DA white dwarfs Table 4. Comparison of EUY- and optically selected white dwarf samples. This work BSL Liebert and Bergeron < logg > 17 < M/M0 > 17 7.90 7.91 7.88 0.38 0.26 0.34 0.644 0.590 0.609 0.189 0.134 0.147 N stars 89 129 200 given mass, higher temperature stars have greater radii and, therefore, lower surface gravity. For example, a 0.6-Mo star will have logg = 8.08 at 40000 K, a typical temperature for an EUV-selected white dwarf, while at 10000 K (a typical BSL temperature) logg=8.23. However, the mean surface gravity is very similar to that found by BSL (see Table 4). The reason for this is apparent from Fig. 5, as discussed above, with the excess of high-gravity objects among the ROSAT stars offsetting the lower gravity peak. The primary difference between the mass distributions of BSL and this work occurs in the high-mass tail of the two distributions. As we have shown, the excess in ROSAT stars above 0.85 Mo is significant at the 99.7 per cent level. Liebert & Bergeron (1995) discuss the biases which arise in magnitude-limited optical samples such as the PG survey. These. samples are biased against high-mass white dwarfs, because such· stars have smaller radii and thus sample a smaller spatial volume than less massive white dwarfs of the same temperature. If EUV luminosity were mass-independent, then this same bias would also apply in addition to the obvious EUV sampling bias due to the increase of ISM opacity with distance. However, in comparing the BSL sample with the EUV-selected objects, we must remember that we are not considering the same white dwarf temperature range in each case. The BSL mass distribution may not represent that of an optically selected sample covering the same temperature range as the EUV sources. Hence the apparent excess of high-mass stars seen in the EUV data can only be claimed to exist in comparison with BSL, and may not be present in a comparison with an optically selected group spanning the appropriate temperatures. Unfortunately, a suitable sample does not yet exist in the literature. Both Finley (1995) and Liebert & Bergeron (1995) make the point that there are large differences between the cooling rates of massive and 'normal' ( ~ 0.6 Mo) while dwarfs in the similar temperature ranges, which affects the number of stars we might expect to find in a given sample. Indeed, the sharp decrease in the cooling rate of massive (> 1 Mo) white dwarfs in the range 40000-50000 K might increase the numbers of such stars in this interval, compared to the 'normal' (lower mass) population. Alternatively, the excess may arise because the high-mass white dwarfs are significantly more EUV-Iuminous than lower mass white dwarfs of the same temperature. Since the photospheric composition of a white dwarf is determined by the balance between radiative levitation effects and gravity, lower abundances of heavy elements and, consequently, lower EUV opacity should be found in those stars with the highest gravities. However, since many of the high-mass white dwarfs have temperatures below ~ 40000 K, where heavy elements are ahnost certainly unimportant (see Barstow et al. 1993), this does not seem to be a viable explanation of the high-mass excess. Whatever selection effect is operating, the presence of a number of previously unknown massive white dwarfs in the EUV sample poses several interesting questions. First, has ROSAT detected a new population of high-mass objects? Secondly, could the group of stars above ~ 0.9 Mo represent a secondary mass peak produced by coalesced binary white dwarfs? Such products of binary evolution may be similar to the excess of low-mass white dwarfs below the He- © 1997 RAS, MNRAS 286, 369-383 © Royal Astronomical Society • Provided by the NASA Astrophysics Data System Downloaded from http://mnras.oxfordjournals.org/ by guest on October 6, 2014 ROSAT objects. While both Finley (1995) and Liebert & Bergeron (1995) have extended the optical work to higher temperatures than BSL, comparable to the range explored by ROSAT, these results are preliminary and are not presented in a form that allows detailed comparison. In particular, the mass determinations of Finley (1995) relied on the zero H layer mass evolutionary models. Consequently, we compare our sample in detail with that of BSL, but also note the summarized results of Liebert & Bergeron (1995). Table 4 lists the mean surface gravities, mean masses and associated standard deviations of the EUV-selected, BSL and Liebert & Bergeron samples. The BSL masses were redetermined using the thick H layer evolutionary models. The BSL mass and surface gravity distributions are compared with the EUV-selected data in Figs 4 and 5 respectively. To facilitate this analysis, the number frequencies in each mass/gravity bin of the BSL distributions have been scaled to account for the differences in the number of objects included in each sample (89 as compared with 129 objects). It can be seen that the mass distributions have very similar well-defined peaks, with more than 50 per cent of stars lying in the range 0.5-0.6 Mo' The ROSAT distribution has an interesting but modest enhancement in the number of stars with masses in the range 0.65-0.75 Mo by a factor 4, but at larger masses (~1 M o )' there seems to be a large excess of stars compared to BSL. The statistical significance of this high-mass group can be examined by considering the BSL sample to represent a parent distribution of objects and predicting the number of high-mass stars that should be seen in the EUV-selected group if these stars were drawn from the same population. Separating the BSL stars into high- and low-mass groups at 0.85 Mo (124 and five stars respectively), the probability that an individual star has high mass ( > 0.85 Mo) is found to be 0.039. After removing from the ROSAT sample those stars that are already in the BSL data, we are left with a total of 83 objects, and we predict that 3.2 ofthese should have high mass. However, 12 stars in the group fall into this category, an excess of 8.8 compared to the number expected. Therefore the significance of the high-mass excess (excess/prediction) is 2.80', corresponding to a confidence level of 99.7 per cent. We note that a survey of 18 southern hemisphere EUV-selected hot white dwarfs by Vennes et al. (1996) reveals the presence of three highmass objects. These are also included in our sample. However, with a comparatively small sample size, Vennes et al. were unable to construct a mass distribution or determine the significance of these detections. Significant differences between the gravity distributions are apparent in Fig. 5. That of the ROSAT stars peak at logg=7.75, lower than the BSL group which has a maximum at logg=7.9. Indeed the main portion of the ROSAT distribution is offset by 0.15 in logg when compared to BSL. This can be explained straightfolWardly as a result of the higher temperature of the stars in the ROSAT sample. For a 381 1997MNRAS.286..369M 382 M. C. Marsh et al. 7 CONCLUSION In this paper we have presented a detailed optical study of the white dwarfs detected by the ROSAT WFC during its sky survey and, therefore, selected on the basis of their EUV flux. The major use of the information on TefI , logg and visual brightness will be to support a subsequent analysis of the EUV and X-ray data. However, the distribution of surface gravities and masses derived from the optical work provides important information about the sample of white dwarfs and gives some indication of possible biases. The most striking result is that ROSAT detects a statistically significant excess in the number of hot, massive DAs that might be expected on the basis of the optical studies, such as that of BSL. However, since the optical and EUV samples do not cover the same range of white dwarf temperatures, the high-mass excess may arise from differences in the cooling rates of 'normal' (~0.6 Mo) and massive (> 1.0 Mo) stars. Consequently, this feature may not be a result of selection on the basis of EUV flux, and might also be present in an optically selected sample covering an appropriate temperature range. Interestingly, these high-mass stars may form a discrete population of merged binary white dwarf systems, but further studies are needed to establish whether or not this is so. ACKNOWLEDGMENTS This paper is based on observations made with the Steward Observatory 2.3-m telescope operated by the University of Arizona, the 1.9-m Radcliffe and 0.75-m telescopes of the South African Astronomical Observatory, and the 1.0-m Jacobus Kapteyn Telescope operated on the island of La Palma by the Royal Greenwich Observatory in the Spanish Observatorio del Roque de los Muchachos of the Instituto Astrofisica de Canarias. MCM, MAB, MRB, AJP and AES acknowledge the support of PPARC, UK. JBH acknowledges NASA grants NAGW5-2269 and NAGW5278. The data reduction and analysis were carried out with Starlink and NOAO IRAF software. We thank the referee, Dr D. Finley, for constructive comments on the initial version of this paper, and for providing additional information which aided the revision. Finally, we thank Matt Wood for access to his evolutionary models. REFERENCES Argyle R W., Meyer C. J., Pike C. D., Jorden P. R, 1988, A User Guide to the JKT CCD Camera, ING User Manual XVIII Barstow M. A, Schmitt, J. H. M. M., Clemens J. c., Pye J. P., Denby M., Harris A W., Pankiewicz G. S., 1992, MNRAS, 255, 369 Barstow M. A et aI., 1993, MNRAS, 264, 16 Barstow M. A, Holberg J. B., Koester D., 1994a, MNRAS, 268, L35 Barstow M. A, Holberg J. B., Koester D., 1994b, MNRAS, 270, 516 Barstow M. A et aI., 1994c, MNRAS, 271,175 Barstow M. A et aI., 1995a, MNRAS, 272, 531 Barstow M. A, Holberg J. B., Koester B., 1995b, MNRAS, 274, L31 Barstow M. A, Holberg J. B., Hubeny I., Lanz T., Bruhweiler G. C., Tweedy R W., 1996, MNRAS, 279, 1120 Bergeron P., Wesemael F., Fontaine G., 1991, ApJ, 367, 253 Bergeron P., Saffer R A, Liebert J., 1992, ApJ, 394, 228 (BSL) Bergeron P., Wesemael F., Beauchamp A, Wood M. A, Lamontagne R., Fontaine G., Liebert J., 1994, ApJ, 432, 305 Bergeron P., Liebert J., Fulbright M. S., 1995, ApJ, 444, 810 Chayer P., LeBlanc F., Fontaine G., Wesemail F., Michaud G., Vennes S., 1994, ApJ, 436, L161 Chayer P., Vennes S., Pradhan A K, Thejll P., Beauchamp A, Fontaine G., Wesemael F., 1995, ApJ, 454, 429 Cheselka M., Holberg J. B., Watkins R, Collins J., 1993, AJ, 106, 2365 Dreizler S., Werner K, 1993, A&A, 278,199 Dupuis J., Vennes S., Bowyer S., Pradhan A K, Thejll P., 1995, ApJ, 455, 574 Finley D. S., 1995, in Koester D., Werner K, eds, White Dwarfs, Lecture Notes in Physics. Springer, Heidelberg, p. 150 Finley D. S., Koester D., Basri G., 1997, ApJ, submitted Fleming T. A, Liebert J. L., Green R F., 1986, ApJ, 308, 176 Fleming T. A, Snowden S. L., Pfefferman E., Briel u., Greiner J., 1996, A&A, in press Hamada T., Salpeter E. E., 1961, ApJ, 134, 683 Hardie R H., 1962, in Hillner W., ed., Astronomical Techniques. Univ. Chicago Press, Chicago, p. 178 Holberg J. B., Wesemael F., Wegner G., Bruhweiler F. C., 1985, ApJ, 293, 294 Holberg J. B. et aI., 1993, ApJ, 416, 806 Then I., 1990, ApJ, 353, 215 Kidder K M., 1991, PhD thesis, Univ. Arizona Koester D., 1991, in Michaud G., Tutukov A, eds, Proc. IAU Symp. 145, Evolution of Stars: The Photospheric Abundance Connection. Kluwer, Dordrecht, p. 435 Koester D., 1996, in Bowyer S., Malina R F., eds, Astrophysics in the Extreme Ultraviolet. Kluwer, Dordrecht, p. 185 Koester D., Schonberner D., 1986, A&A, 154, 125 © 1997 RAS, MNRAS 286, 369-383 © Royal Astronomical Society • Provided by the NASA Astrophysics Data System Downloaded from http://mnras.oxfordjournals.org/ by guest on October 6, 2014 burning core mass composed of low-mass double degenerate systems. Indeed, one of these high-mass stars, GD 50, has already been suggested as a possible merged degenerate binary by Vennes et al. (1996) on the basis of its large mass and suggested high rotation rate. The various possible outcomes of binary white dwarf mergers are discussed by Webbink (1984) and Then (1990) and others. The merger of two He-core white dwarfs may produce sdO and sdB stars, while it is suggested the merger of an He white dwarf with a CO core white dwarf may result in a R CrB star. Yungelson et al. (1994) have considered the formation rate for double degenerate (DD) systems, predicting that approximately one in 55 systems observed is expected to be a DD with a period short enough for a merger to occur within a Hubble time. Fewer will actually have merged to yield the high-mass single objects that we find. If the high-mass stars in the EUV are products of DD mergers the fraction is rather larger than expected. However, the total numbers of such stars in the sample are small and sensitive to statistical fluctuations. Furthermore, the measurement might be affected by the biases already discussed, while the theoretical predictions are dependent on the spectrum of initial binary mass ratios and assumptions about the common-envelope phase. Hence no firm conclusion can be drawn. If it is eventually possible to quantify the magnitude of the selection effect for high-mass systems, through analysis of a suitable optical sample, folding in the effects of different cooling rates, it should be possible to constrain the theoretical models of common-envelope evolution. 1997MNRAS.286..369M A UV-selected sample ofDA white dwarfs Landolt AU., 1983, AJ, 88, 439 Landolt A u., 1992, AJ, 104, 340 Lanz T., Hubeny I., 1995, ApJ, 439, 905 Lanz T., Barstow M. A, Hubeny I., Holberg J. B., 1996, ApJ, in press Liebert J., Bergeron P., 1995, in Koester D., Werner K, eds, Lecture Notes in Physics, White Dwarfs. Springer, Heidelberg, p.12 Marsh M. C. et al., 1995, in Koester D., Werner K, eds, Lecture Notes in Physics, White Dwarfs. Springer, Heidelberg, p. 328 Mason K A et al., 1995, MNRAS, 274, 1194 Menzies J. W., Cousins A W., Banfield R M., Laing J. D., 1989, SAAO Circ., 13, 1 Napiwotzki R, 1995, in Koester D., Werner K, eds, Lecture Notes in Physics, White Dwarfs. Springer, Heidelberg, p. 132 Napiwotski R et al., 1993, A&A, 278, 478 Pfefferman E. et al., 1986, Proc. SPIE, 733, 519 Pounds K A et al., 1993, MNRAS, 260, 77 Pye J. P. et al., 1995, MNRAS, 274,1165 Reid I. N., 1996, AJ, 111, 2000 Shafer R A, Haberl F., Arnaud K A, Tennant A F., 1991, ESA 383 TM-09 Shipman H., 1976, ApJ, 206, L67 Sims M. R etal., 1990, Optical Engineering, 29, 649 Tweedy R. W., Holberg J. B., Barstow M. A, Bergeron P., Grauer A D., Liebert J., Fleming T. A, 1993, AJ, 105, 1938 Vennes S., Chayer P., Thorstensen J. R, Bowyer S., Shipman H. L., 1992, ApJ, 392, L27 Vennes S., Thejll P. A, Wickramasinsghe D. T., Bessel M. S., 1996, ApJ, 467, 782 Vennes S., Bowyer S., Dupuis J., 1996, ApJ, in press Webbink R P., 1984, ApJ, 277, 355 Wesemael P., Auer L. H., Van Horn H. M., Savedoff M. P., 1980, ApJS, 43, 159 Wolff B., Jordan S., Koester D., 1996, A&A, 307, 149 Wood M. A, 1992, ApJ, 386, 539 Wood M.·A, 1995, in Koester D., Werner K, eds, Lecture Notes in Physics, White Dwarfs. Springer, Heidelberg, p. 41 Yungelson L. R, Livio M., Tutukov A V., Saffer R A, 1994, ApJ, 420,336 Zombeck M. V., 1990, Handbook of Space Astronomy and Astrophysics. Cambridge Univ. Press, Cambridge Downloaded from http://mnras.oxfordjournals.org/ by guest on October 6, 2014 © 1997 RAS, MNRAS 286, 369-383 © Royal Astronomical Society • Provided by the NASA Astrophysics Data System
© Copyright 2024