Lang-term Photometrie Campaign at ESO and the New Eelipsing P Cygni Star R 81 in the LMC 2 3 4 B. WOLF 1, C. STERKEN , 0. STAHL and J. MANFROIo

1Landessternwarte Heide/berg, 2Vrije Universiteit Brusse/s, 3ESO, 4 Universite de Liege

1. A Programme for Long-term ticipants who work in a tight team with approach is a very efficient and useful Photometry of Variable Stars the principal investigators. From the be­ one. Besides the direct yield of data, the ginning we aimed at obtaining several project also offers to young scientists At the ESO workshop The Most Mas­ months of observing time each observ­ (e. g. Ph. O. students) the possibility to sive Stars in November 1981 (Eds. ing season, and this for a time span of at get into contact with the fields of re­ O'Odorico, Baade and Kjär), C. Sterken least one decade. search of the other participants, wh ich presented results of his observations of A project of that size, including the substantially contributes to their obser­ HO 160529 during a time span of more interpretation of measurements ob­ vational experience. than 8 years. HO 160529 is an A2 hyper­ tained at different telescopes and by giant wh ich was found to be variable by different observers cannot work without 2. R 81, the Counterpart of P Cyg Wolf et al. (1974) in an irregular way with astriet agreement on observing proce­ in the LMC an amplitude of several hundredths of a dures and reduction techniques. Espe­ magnitude. Sterken continued the cially for what concerns the data reduc­ P Cygni is one of the most outstand­ monitoring of this star at every possible tion, high requirements are needed so ing stars of the Galaxy. It had a remark­ occasion in the period 1973-1981 and that the final data may be of the highest able outburst in 1600 and is particularly found evidence for the presence of obtainable accuracy. It was decided distinguished by its unusual spectrum. periodic light variation of about 100 that all reductions would be done cen­ The so-called P Cygni profiles in the days. From the phase diagram he con­ trally (and not by the individual obser­ visual range (wh ich have been disco­ cluded that in spite of the intensive ob­ vers) at the University of Liege by J. vered already in the 19th century) are servations by one person, there still re­ Manfroid. The reduction algorithm is a named after this prototype and indicate main long gaps in the phase diagram. If generalized method developed by Man­ that this star is losing mass at an ex­ sampling of observations happens in froid and Heck (1983) and is charac­ cessive rate. P Cygni belongs to the such an irregular way (gaps caused by terized by the use of practically every luminous (Mbol = -9 to -11) blue vari­ irregularly allocated short observing measurement of any non-variable star. ables (LBV's) which have been recog­ runs), long-term orbital effects may be In doing so, all measurements are trans­ nized during the past few years as key­ hidden by or ascribed to the super­ formed into a standard system and the objects for the understanding of the giant's irregular intrinsic fluctuations. It observations obtained during different evolution of the very massive (initial was therefore suggested that the obser­ seasons and with different instruments masses ~ 50 MCi)) stars. According to vations of e. g. variable supergiants be can be re-evaluated regularly so that the current evolutionary models (cf. e. g. done by a larger team of observers at a homogeneity of the data is assured. Humphreys 1986, Chiosi and Maeder dedicated small instrument. Such an Because of the availability of the suit­ 1986) these very massive stars do not approach would also be very useful for able photometers on the ESO 50 cm become red supergiants; instead a criti­ all differential photometry on a basis of and the Oanish 50 cm telescopes. and cal luminosity boundary exists (calIed one measurement per night or less: for the usefulness for astrophysical in­ Humphreys Davidson limit) beyond supergiants, Be stars, Ap stars ... terpretation of the data. we decided to which the most luminous stars do not Ouring the discussions following the carry out all measurements in the evolve. The LBV's are located close to talk, several colleagues expressed their Strämgren uvby system. Since we deal this boundary and are supposed to be interest in the idea, and by the end of the with variable stars, preference was gi­ the immediate progenitors of the Wolf­ Workshop, the project Long-term ven to differential measurements with Rayet stars. Photometry of Variables was born. In two comparison stars (although we have The physical properties of P Cygni February 1982 a meeting was held at also carried out absolute measure­ have been scrutinized during the past the University of Brussels, and the pro­ ments). few years at a wide wavelength range ject finally started with an application for In practice the participants may ex­ from the UV with IUE to the IR with observing time submitted in April 1982. pect to receive their data not later than IRAS! A drawback of P Cygni is that The participants (originally about 15) three to six weeks after termination of both its interstellar reddening and its merged the objects for wh ich photome­ the observing run. Since 1982 the distance are difficult to determine (see try on a long-time basis was needed into number of participants has doubled, e. g. Lamers, de Groot and Cassatella one object list of about 70 entries, and and this is also so for the number of 1983). For this reason we searched the then the file was split in seven sections stars. At least 30 stars, for which zoo of the luminous blue supergiants of according to the nature of the variable enough data were obtained, have been the Large Magellanic Cloud (LMC). As stars (see Sterken, 1983). Later on, an omitted from the object list. 26 observ­ outlined e. g. in our recent atlas of high eighth section (Peculiar Late Type Stars) ing runs with a duration of about one dispersion spectra of luminous LMC was added. For each section a principal month length each have already been stars (Stahl et al. 1985) the P Cygni type investigator and a co-investigator were allotted by ESO and a huge amount of stars form in fact a populous group appointed. These scientists decide ab­ data are already available. The results among the peculiar emission line stars. out which objects need monitoring, and have already led to more that 20 pub­ Already in 1981 Wolf et al. presented they carry out and/or supervise the lished scientific papers. a detailed spectroscopic and photome­ analysis of the data of the stars be/ong­ The nature of some discoveries (Iike trie study of R 81 of the LMC. R 81 ing to their sections. We agreed that the e. g. the binary nature of R 81 wh ich is turned out to be a particularly close observers would be volunteering par- described below) clearly proofs that this counterpart of the galactic star P Cygni.

10 bound dust envelope since Stahl et al. R 81 I UE, LWR (1987) identified R 81 as an IRAS point source. The dust formation could be a 1982, Feb. 17 !SI (Y) consequence of their slow and dense N ~ Q) N Q) u:> N le N N u:> winds. le le N ~ le le le ~ ~ .... One of the most paradoxical results

~ ~ ~ ~ ~ ~ ::: ::: ::: ::: ::: ::: ::: derived from the IUE spectrum of P .. Cl Cl Cl Cygni (Cassatella et al. 1979) is that this ..,>-.. u.. u.. u.. u.. I I I I hot star with its unusual P Cygni profiles (l) c in the visual range is right the other way ..,CD distinguished by a lack of P Cygni pro­ C ..... files in the ultraviolet (which are other­ CD > wise so typical for hot stars in this .., range). Just this behaviour is found in 0 the case of R 81 as weil. Still more in­ CD L triguing is the fact that e. g. the Fe II lines of both stars agree even in details. The (Y) lf) complex profiles show a multiple com­ ....Cl ­+ + ponent structure (for R 81 see Fig.1) ascribed to the ejection of discrete shells (cf. Lamers et al. 1985 and Stahl et al. 1987). 0. In addition to the major historical out­ 2612. 2622. A[J.J 2632. bursts of P Cygni in the 17th century, Figure 1: A seetion of the IUE high-dispersion spectrogram around 2600 A. This wavelength irregular photometrie variations of 0.1 to range is dominated by Fell absorption lines. Like in P Cyg, different components are present. 0.2 magnitude have been reported dur­ The velocities of these components are indicated for different lines. ing this century by many authors. No strict periodicity has yet been found (van Gent and Lamers 1986). Likewise in the Both stars are (of course) characterized of the Hertzsprung-Russell Diagram case of R 81, which until 1980 was re­ by strong P Cygni profiles of the Balmer (HRD) close to the Humphreys-David­ peatedly observed during various ran­ lines and by very similar stellar parame­ son limit. The winds of both stars are domly distributed photometrie observ­ 1 ters: Teff = 20,000 K, Mbol = -10, and R slow (vmax = 250 to 300 km 5- ) and very ing runs, variations of about this am­ 5 I = 70 R0 for R 81 and Teff = 19,300 K, massive (M = 3 . 10- M0 yr- for R 81 plitude are quoted. Wolf et al. (1981) 5 1 Mbol = -9.9, and R = 76 R0 for P Cygni and M = 1.5 . 10- M0 yr- for P Cyg). regarded these variations as to be "of (for the data for P Cyg see Lamers et al. Like P Cyg (Waters and Wesselius 1986) irregular nature rather than of a periodic 1983). They are located in the same part R 81 is surrounded by a cool loosely one". Shortly after this the programme

I -I--r~ 1 R8l, y band IOllg-lcrm phololllcl.ry or variables lSl'" N'" I ~ o o 0 '80 0 .-< 0 ~o § l ~@ I 0 .-< 000 8 0() 0 0:: '" ~ (§J0 00 o 0 I o 8 0 o 0 0 o 0 0 '" - o '"M 0 0 0 0 I - o 0 0 0 B 8 0

'" 0 lf1'"

N 0 0 rlJj 0 I .-< @~~ ~ ~~.l w ~ ~ 0 fI .. '" 0 "'" I L -l- _1_ L ----l- ---l-- -L- -.L ---l- -L I). ('OO S. 10 6.200 6. 700 JD244- (1000 days) Figure 2: The differential light curve in the y band for the observations of R 81 obtained within the programme of long-term photometry of variables. The upper part shows the observations in the sense C 1-R 81 and the 10 wer part shows on the same scale (but shifted) the differences of the two comparison stars. The y magnitudes of the comparison stars C 1 (= HO 34144) and C2 (= HO 34651) are 9.32 and 8.37, respectively. The scatter in the differences y (C 1-C2) of the two comparison stars is 0.007 which documents the quality of the photometry.

11 tions are shown in Figure 2. This data set was analysed with aperiod search­ ing algorithm and we finally found that the period of the variations was about es> +'. es> 74.6 days. With the period al ready es> o+.~ tn , .. +.;t:G. known with some precision, the obser­ I ~+ t +0 vations of Appenzeller (1972) could also -8 ~ ++ o. be tied in. They then define the period >- es> 0 es> '" particularly weil since they extend the C\J , time base by more than ten additional o o years (which is about 50 cycles). Taking + x + x .. • + all data together, we obtain aperiod of es> ...es> 74.59 ± 0.01 days. The phase diagram , of the y-band data wh ich was con­ o o structed by using this period is shown in Figure 3. This phase diagram clearly im­ plies that R 81 is an eclipsing binary. The light curve is very distorted which N u suggests that both components are in I close contact. -8 Obviously, the detection of a binary P >- Cyg star is very important since we might hope to derive the parameters of the system, specifically the mass. This is a very important undertaking, since Phase there are no direct mass determinations Figure 3: The phase diagram of the data shown in Figure 2. In addition the V data of of a P Cyg star available. So far, we Appenzeller (1972) have been included. The different symbols denote observations performed obtained two CASPEC spectra and one during different observing seasons (0 for 1982/1983,0 for 1983/1984.6 for 1985/1986, + for CES spectrum of R 81. A portion of the 1985/1986 and x for 1986/1987. Appenzeller's earlier observations (JD2441274 - 1292) are CASPEC spectra is shown in Figure 4. It denoted by.. The period of the eclipses is particularly weil defined due to the earlier is clear from this figure that most of the observations of Appenzeller. The eclipse with a depth of 0.4 mag can clearly be seen. The lines have a P Cyg profile or are blue­ shape of the light curve is particularly distinguished by the absence of a pronounced shifted absorption lines originating in secondary minimum and by a pre-eclipse dip of 0.15 mag at phase 0.80. the stellar wind of R 81. They are thus not very useful to determine the orbital velocity of the star. From the very few of long-term photometry of variables addition, we found that the minima re­ photospheric lines in the spectra we described above was initiated. Un­ peated every 300 days. At that point we could so far not obtain a reliable radial doubtedly, a detailed knowledge of the realized that the variations of R 81 might velocity curve. In addition, no secondary time dependence of brightness varia­ be at least partly periodic and we de­ spectrum has been found so far. tions is important for the derivation of a cided to give this star a high priority in Luckily, some information can be ob­ physical model for P Cyg stars. Since "P the further observations since never be­ tained from the photometry: The ab­ Cyg of the Galaxy" has obviously never fore had strictly periodic variations been sence of a pronounced secondary been monitored during several years by found in a classical P Cyg star. Until now eclipse probably means that most of the high-precision photometry at sites with we have collected observations in 240 visual brightness is supplied by one comparable excellent weather condi­ different nights distributed over aperiod component. The depth of the eclipses tions as at La Silla, we included "P Cyg of more than four years. The observa- suggests that the eclipsing body has a of the LMC" (i. e. R 81) in the long-term photometry programme at ESO.

1. 500 3. The Eclipsing Binary R 81 R 81 was already found by Appenzel­ ler (1972) to exhibit variations with the rather large amplitude of 004 mag within two weeks. We subsequently observed ...·c R 81 in several years in observing runs ·c of about two weeks each, but we were ...·> unable to find a similar event. Only varia­ ~ 500 tions of about 0.15 mag, which seemed ~ 0. 1 to be irregular, were found. So R 81 was regarded as an irregularly variable P Cyg star, although the time coverage was I too incomplete to do a meaningful 0.0001 LI , .l.--l-l , period analysis. 4380.000 4430.000 4480.000 4530.000 4580.000 A. [Äl In the first three observing seasons of Figure 4: Comparison of two CASPEC spectra taken at phase 0.97 and 0.27, respectively. the long-term programme we had ob­ Variations can c/early be seen. During eclipse, numerous Felliines could be identified and the served each year once that R 81 had Hel lines and Mg/l ;. 4481 show a P Cyg profile. The velocities of the different lines are faded far below its normal brightness. In indicated in the Figure. 12 radius which is comparable to the radius wind lines, are not irregular but phase­ Macchetto, F., Penston, M., Selvelli, P. L., of the visible star, which is known from dependent. An important point to clarify Stickland, 0.: 1979, Astron. Astrophys. 79, the brightness, temperature and dis­ is also the nature of the secondary. No 223. tance of R 81. The width of the eclipses obvious secondary minimum is present, Chiosi, C., Maeder, A.: 1986, in Ann. Rev. Astron. Astrophys., 24, 329. so the star must be visually faint. New of the star with respect to the period Humphreys, R. M.: 1986, in Luminous stars then allows us to estimate the distance spectroscopic observations of high and associations in ga!axies, lAU Symp. of both stars from each other and thus quality and good phase coverage are No. 116, eds. de Loore, Willis, Laskarides, from Kepler's third law (with the period) badly needed. p.45. the total mass of the system. We find a The more general question is, of Lamers, H.J.G.L.M., de Groot, M.J.H., Cassatella, A.: 1983, Astron. Astrophys. mass of about 35 M0 for R 81. This course, whether many P Cyg stars are result is in good agreement with the binaries. Our finding certainly does not 128,299. expectations for a luminous P Cyg star, prove this. However, the small scatter Lamers, H. J. G. L. M., Korevaar, P., CassateI­ but note that it is the first direct mass around the mean light curve shows that la, A: 1985, Astron. Astrophys. 149,29. R 81 intrinsically is at most slightly vari­ Manfroid, J. Heck, A: 1983, Astron. Astro­ estimate for such astar. We note that phys. 120,302. this mass estimate is very uncertain so able. If other P Cyg stars (wh ich are Stahl, 0., Wolf, B., de Groot, M.J.H., Leith­ far and clearly more spectroscopy is observed to be variable with relatively . erer, C.: 1985, Astron. Astrophys. Suppt. needed in order to confirm this result. large amplitude) are similar to R 81 in 61,237. Interestingly, the scatter around the this respect, then the variability ob­ Stahl, 0., Wolf, B., Zickgraf, F.-J.: 1987, As­ mean curve of Figure 3 is only of the served in these stars requires an expla­ tron. Astrophys., submitted. order of 0.05 mag. This means that also nation. Sterken, C.: 1983, The Messenger 33,10. most of the smaller variability which we van Gent, R.H., Lamers, H.J.G.L.M.: 1986, Astron. Astrophys. 158, 335. observed between the eclipses is due to References Waters, L.B.F.M., Wesselius, P.P.: 1986, the changing aspects of the observa­ Aslfon.Aslfophys. 155, 104. tions and not intrinsic to the star. This Appenzeller, 1.: 1972, Pub!. Astron. Soc. Ja­ Wolf, B., Campusano, L., Sterken, C.: 1974, result leads to the suspicion that also pan 24, 483. Astron. Astrophys. 36, 87. other variations of R 81, e. g. the spec­ Cassatella, A, Beeckmans, F., Benvenuti, P., Wolf, B., Stahl, 0., de Groot, M. J. H., Ster­ troscopic variations of certain stellar Clavel, J., Heck, A., Lamers, H.J. G. L. M., ken, C.: 1981, Astron. Astrophys. 99, 351.

Blue Horizontal Branch Field Stars in the Outer Galactic Halo J. SOMMER-LARSEN and P. R. CHRISTENSEN, The Niels Bohr Institute, University of Copenhagen

1. Introduction dispersion was found to be - 60 km/so fields located at the SGP (I, b) = (38°, It is a well-known hypothesis that Even if one assumes that the velocity -51°) and (I, b) = (352°, 52°). In total galaxies are surrounded by extended distribution of these objects is isotropic, the catalogues cover 54 square degrees massive envelopes of "dark" matter a mass of only M = (2.6 ± 0.8) * 10" of the sky, and they are complete to V = reaching far beyond the visible edges of M0 is inferred. This does not support 18.5. By observing spectroscopically the galaxies. For our own Galaxy this the hypothesis that the mass of the faint blue stellar objects drawn from hypothesis is supported by kinematical Galaxy increases linearly with Galac­ these catalogues, using the selection and dynamical studies of globular clus­ tocentric distance to distances criterion 0.0 :5 B-V :5 0.2, a sampie of ters in the outer Galactic halo. ~ 100 kpc. 131 bhbf star candidates have been ob­ The usefulness of globular clusters as In order to clarify further on the prop­ tained. test objects for probing the mass dis­ erties of distant halo objects we have The observations were done with the tribution in the outer part of the Galaxy is identified and studied a sampie of blue ESO 2.2-m telescope during a number limited, however, for the following horizontal branch field (bhbf) stars in the of observing runs in the period reason: The sampie is small - the outer Galactic halo (r :5 40 kpc). The 1984-1986. The observational setup number of globular clusters with Galac­ observations are described in section 2, and procedure were the same during all tocentric distances between 15 and and the results are discussed in sec­ observing runs: The detector system 40 kpc, which can be used in a kinemat­ ti on 3. consisted of a Boiler and Chivens spec­ ical analysis, is 14. Furthermore, several trograph together with the dual Reticon recent remeasurements of globular Photon Counting System (RPCS) 2. Observations and Data cluster radial velocities have showl1 (Christensen et al., 1984). The two dec­ Analysis some of these to be substantially in ker holes, each projecting down onto its error. We have carried out a search for bhbf own Reticon-array, corresponded to an 2 Lynden-Bell, Cannon and Godwin stars at large Galactocentric distances. area of 4*4 arcsec . With a slit the (1983) studied a sam pie of dwarf Part of the observations have been de­ aperture was reduced to 2 arcsec in the spheroidals situated at very large Galac­ scribed in Sommer-Larsen and Christ­ direction of dispersion. A 600 lines/mm tocentric distances (-100 kpc). They ensen (1985 and 1986). The basic mate­ grating blazed in the first order at found the objects to have quite low line rial was three stellar object catalogues, 4200 A was used yielding a reciprocal of sight velocities (relative to the Galac­ kindly provided by Drs. G. Gilmore and dispersion of -1 Aper channel. The tic restframe) - the line of sight velocity N. Reid. The catalogues cover three wavelength range covered was 13