OPTICAL AND ULTRAVIOLET STUDIES OF X-RAY BINARIES AND MAGNETIC STARS
GODELIEVE HAMMERSCHLAG-HENSBERGE OPTICAL AND ULTRAVIOLET STUDIES OF X-RAY BINARIES AND MAGNETIC STARS
ACADEMISCH PROEFSCHRIFT
ter verkrijging van de graad van doctor in de Wiskunde en Natuurwetenschappen, aan de Universiteit van Amsterdam, op gezag van de Rector Magnificus Dr.G. den Boef, hoogleraar in de Faculteit der Wiskunde en Natuurwetenschappen, in het openbaar te verdedigen in de aula der Universiteit (tijdelijk in de Lutherse Kerk, ingang Singel 411, hoek Spui) op woensdag 30 november 1977 des namiddags te 16.00 uur
door
GODELIEVE CECILE MARIA JOSEPHINE HAMMERSCHLAG-HENSBERGE
geboren te Boom (België) PROMOTOR : PROF. DR. E.P.J. VAN DEN HEUVEL
>CO-REFERENT : DR. C. ZWAAN CONTENTS
Samenvatting 5 Dankwoord 11 Introduction and Summary 13 Part I. X-ray Binaries 23 1. The Spectrum, Orbit and Magnetic Field of HD 153919 (2U 1700-37) i 25 2. The Expanding Atmosphere of the Of Component of the Eclipsing X-ray Binary HD 153919 33 3. A Detailed Study of the Spectrum of the Binary X-ray Source HD 153919 (3U 1700-37). I. Radial Velocity Data in the Blue Spectral Region 37 4. A Detailed Study of the Spectrum of the Binary X-ray Source HD 153919 (3U 1700-37). II. Analysis of the Radial Velocities in the Blue Spectral Region 63 5. Four Colour Photometric Observations of the X-ray Binary HD 153919 (3U 1700-37) 83 6. Ultraviolet Photometric Observations of the X-ray Binary HD 153919 (3U 1700-37) 87 7. Study of the Lightcurve of the Of Star HD 153919 93 8. Orbit, Spectrum and Ha Variations of HD 77581 (3U 0900-40) 121 9. Mass Determination for the X-ray Binary System Vela X-l 129 10. Four Colour Photometric Observations of the X-ray Binary Star HD 77581 (Vela X-l). I. .Observations 131 11. Four Colour Photometric Observations of the X-ray Binary Star HD 77581 (Vela X-l). II. Analysis of the Light Curve 133 12. Photometric Variations of Wray 977 (3U 1223-62?) 141 Part II. Magnetic and Related Stars 145 1. Detection of Crossover Effect in the Ap Star HD 98088 147 2. Photometry of Peculiar A Stars 151 3. Photometry of Silicon Stars 157 4. Ultraviolet Photometric Observations of Ap and Am Stars 165 5. Spectroscopie Studies of Open Clusters - A Search for Ap Stars 191 6. A Study of the Blue Stragglers in Praesepe, M7 and the Hyades Cluster 213 Samenvatting.
Deel 1. Röntgendubbelsterren.
Het eerste deel van dit proefschrift betreft optische waarnemingen van röntgendubbelsterren en hun interpretatie. Spectroscopische en fotometrische waarnemingen werden verricht aan een aantal vroege type sterren die geïdentificeerd zijn met röntgenbronnen, ofwel goede candidaten zijn voor een dergelijke i -tentificatie. De spectra werden opgenomen met de coudé spectrograaf van de 1.5 m telescoop van de ESO, La Silla, Chili. Van deze spectra werden densiteitsregistraties gemaakt met de geautomatiseerde Paul-Coradi microfotometer van de Utrechtse Sterrewacht. Tevens werden de posities var. de spectraallijnen op de platen gemeten met de Grant machine van de Universiteit van Groningen, ten einde radiële snelheidsvariaties te bestuderen.
In artikel 1 wordt het spectrum beschreven van de Of ster HD 153919, de optische begeleider van de röntgenbron 30 1700-37. De radiële snel- heden van de sterlijnen werden gemeten om een mogelijke identificatie met de in 3.41 dagen eclipserende röntgenbron te verifiëren. Het blijkt dat de optische ster inderdaad een radiële snelheidsvariatie vertoont met dezelfde periode. Dit resultaat werd bevestigd door onafhankelijke waarnemingen van Wolff and Morrison (1974). Tevens vrerd gevonden dat P-Cygni profielen en emissielijnen in het spectrum duiden op hoge uit- stroomsnelheden (tot 4000 kia/s) van de steratmosfeer. Het vóórkomen van P-Cygni profielen in een sterspectrum duidt erop dat de ster massa verliest in de vorm van sterrewind. In het tweede artikel van deze serie tracht ik door de waargenomen P-Cygni profielen in het spectrum van de Of ster te vergelijken met theoretische profielen, een schatting te verkrijgen van het massaverlies van deze ster. Met gebruik van de metode van Lucy (1971) (isotrope verstrooiing van het sterrelicht in de uitstromende atmosfeer) vind ik een massaverlies van ongeveer 6 2 x 1O~ MQ/jaar. In artikels 3 en 4 wordt een gedetailleerde studie beschreven van de radiële snelheden van individuele lijnen in het spectrum van HD 153919. De spectroscopische platen beschreven in artikel 1 zijn nu uitgebreid met opnamen verkregen door een aantal waarnemers gedurende een drietal latere perioden: in totaal betreffen deze studies 76 spectrogranunen met hoog oplossend vermogen. Uit de snelheden kon ik de baanparameters van de zichtbare ster afleiden , welke door de omvang en homogeniteit van het waarnemingsmateriaal nauwkeuriger zijn dan in vroegere studies mogelijk was. De baanparameters leveren informatie over de massa's van beide sterren. Daar er echter geen radiële snelheidskromme voor de röntgenster beschikbaar is, kunnen de massa's zelf niet nauwkeurig worden afgeleid, doch kan er slechts een ondergrens voor de massa van de röntgenster worden vastgesteld. Periodieke extra variaties van de snelheden van sommige spectrale lijnen wijzen erop dat de sterrewind van HD 153919 niet constant is, doch varieert tijdens de omloop van de röntgenster rond HD 153919. Dit belangrijke nieuwe resultaat, dat ik uit deze spectra kon afleiden, wijst op een variabele balmer progressie: de sterrewind blijkt het sterkst te zijn in de buurt van de Lagrange punten L. en L-. Van de fotometrische waarnemingen wordt verslag gedaan in de drie volgende artikelen. De Of ster werd waargenomen in het Strömgren uvby systeem, in het Walraven vijf kleuren systeem en, met behulp van de Astronomische Nederlandse Satelliet (ANS) in 5 golflengtegebieden in het ultraviolet. De helderheid van de ster blijkt in alle kanalen te variëren met een dubbele golf tijdens de 3.41 periode. Slechts in het ultraviolet komen significante kleurvariaties voor. In het laatste artikel hebben we een analyse gemaakt van al het aan- wezige fotometrische materiaal. Door vergelijking van de aldus verkregen gemiddelde lichtkromme met de röntgenlichtkromme van de begeleider, heb ik een faseverschil tussen beide lichtkrommen gevonden:het midden van de röntgeneclipsduur valt niet samen met de mid-occultatie van de röntgenster door de Of ster. Het eerste deel van de rontgeneclj.ps lijkt een gevolg te zijn van obscuratie door de sterrewind. We hebben tevens de gemiddelde optische lichtkromme vergeleken met synthetische lichtkrommen, berekend voor een door getijdenkrachten vervormde ster in een dubbelstersysteem. Het blijkt niet mogelijk voor deze Of ster een statische model atmosfeer op te stellen die de fotometrische en spectroscopische waarne- mingen volledig verklaart. Kennelijk moet worden rekening gehouden met dynamische effecten, met name variabiliteit in de sterkte van de sterre- wind over het oppervlak van de ster, ten gevolge van de baanbeweging van de begeleider. Dit resultaat volgt zowel uit de fotometrische waarnemingen als uit de radiële snelheidsanalyse. De volgende vier artikelen behandelen waarnemingen van de röntgen- dubbelster HD 77581 (Vela X-l = 3U 0900-40) , welke verricht ./erden in samenwerking met een aantal andere onderzoekers uit de sterrenkundige instituten van Amsterdam en Brussel (Vrije Universiteit). Artikel 8 beschrijft spectroscopische waarnemingen van de ster, waarin we zowel de lijnprofielen als radiële snelheidsvariaties onder de loep nemen. Het profiel van de lijn Ha blijkt te variëren in fase met de baanomloop van de dubbelster. Nadat met de Amerikaanse satelliet SAS-3 ontdekt werd dat de röntgenbron Vela X-l een regelmatige pulsar is (Rappaport and McClintock, 1975) met een pulsperiode van 283 seconden, kon een gedetail- leerde massabepaling worden uitgevoerd voor dit dubbelstersysteem door onze radiële snelheidsmetingen van de zichtbare ster te combineren met de radiële snelheidskromme van de pulsar, verkregen uit de aankomsttijden van de röntgenpulsen. De eerste resultaten van dit onderzoek, belangrijk voor de toestandsvergelijking in theorieën over neutronensterren, zijn vervat in artikel 9. Artikels 10 en 11 betreffen een fotometrische analyse van HD 77581 in het Strömgren uvby systeem. De waarnemingen tonen - gesuperponeerd op de periodieke dubbele golf - onregelmatige variaties in de lichtkromme van de ene omloop tot de andere. Ik vond tijdens deze studie dat de kleur- index cl (balmersprongindex) varieert in fase met de baanperiode, wat consistent is met onze uit modelberekeningen afgeleide voorspellingen voor een vervormde roterende ster in een dubbelstersysteem.
Het laatste artikel van deze reeks beschrijft fotometrische waarnemingen van de B-superreus Wray 977, een vrijwel zekere candidaat voor identificatie met de röntgenbron 3ü 1223-62. Deze bron toont geen eclips in het röntgen- gebied, hetgeen de identificatie bemoeilijkt. Onze waarnemingen maken waarschijnlijk dat Wray 977 fotometrisch varieert met een periode van ongeveer 23 dagen. Dit resultaat werd recentelijk bevestigd door het werk van Petro (1977).
Deel 2. Magnetische sterren.
Magnetische Ap en Am sterren onderscheiden zich van de gewone sterren van spectraaltype B, A en F door abnormale abundanties van een aantal elementen in hun atmosferen. De zogenaamde Ap sterren zijn sterren van spectraaltype B of A, die men naargelang hun abundantie afwijkingen nog kan onderverdelen in de volgende groepen: (1) Hg-Mn sterren (B-type), over- abundant in Hg en Mn; hebben nooit een sterk magnetisch veld (<500 gauss); (2) Si-sterren (B-, A-type), overabundant vooral in Si en ijzerpiek elementen; hebben een sterk magnetisch veld (tot ca. 35000 gauss); (3) de koelere Eu-Cr-Sr sterren (late A), overabundant in de ijzerpiek elementen en zeldzame aarden; hebben vrijwel steeds een sterk magnetisch veld (tot 20000 gauss). De Am sterren (late A, vroege F) hebben als hoofdkenmerk de zwakte van de Ca II-K lijn t.o.v. hun waterstof-spectraaltype. Ze hebben bovendien te sterke lijnen van vele metalen en nooit een sterk magnetisch veld (<500 gauss).
De Eu-Cr-Sr ster HD 98088 neemt een belangrijke plaats in onder de Ap sterren omdat het de enige dubbellijnige spectroscopische dubbelster is onder de late typa Ap sterren. Besloten werd deze dubbelster spectroscopisch nader te bestuderen omdat uit vergelijking van het spectrum van bside componenten mogelijk iets af te leiden zou zijn over het ontstaan van Ap sterren. Tijdens deze studie ontdekte ik op een hoge dispersie Zeeman plaat dat lijnen van Fe, Cr en Ti het zogenaamde "cross-over" effect (= effect dat ontstaat door de rotatie van de ster gecombineerd met de aanwezigheid van twee magnetische vlekken van verschillende polariteit op het steroppervlak) vertonen. De resultaten hiervan zijn weergegeven in artikel 1. Uit dit effect leidde ik af dat de magnetische vlekken veldsterkten hebben van ongeveer 8000 gauss, terwijl daarentegen het uit vroegere metingen afgeleide "effectieve" veld (gemiddelde van het veld geïntegreerd over het hele steroppervlak) slechts 1000 gauss sterk is.
De meeste Ap sterren tonen ook periodieke fotometrische variaties. De algemeen aanvaarde verklaring voor de lichtvariaties van magnetische sterren is een inhomogene heläerheidsdistributie over het oppervlak van de roterende ster, op één of andere wijze veroorzaakt door de aanwezigheid van magnetische vlekken op het steroppervlak. Artikels 2 en 3 beschrijven fotometrische waarnemingen van 8 Si-sterren en één Sr-Cr ster in het Strömgren uvby systeem. Voor 7 van de Si-sterren kon een periode vastgesteld worden en een lichtkromme afgeleid. De overige 2 sterren zijn zeer waarschijnlijk ook variabel, maar we beschikken over te weinig waarnemingen om een periode af te leiden. De amplitude van de lichtvariaties van alle waargenomen Si-sterren neemt af naar korte golf- lengten, in overeenstemming met waarnemingen van een "nul-gebied" (geen variatie; in het ultraviolet voor Si-sterren; aan beide zijden van dit gebied zijn de lichtkrommen in tegen-fase. In artikel 4 geven we de resultaten van onze fotometrische waarnemingen in het ultraviolet met de ANS van 79 Ap sterren en 26 Am sterren. Kleur- diagrammen tonen aan dat vooral de koele Ap sterren afwijkingen vertonen t.o.v. de normale sterren. Dit is waarschijnlijk het gevolg van "blanketing" effecten veroorzaakt door de vele sterke lijnen der overabundante metalen in het ultraviolette spectraalgebied. De sterkte van de afwijking blijkt afhankelijk te zijn van de magnetische veldsterkte: hoe sterker het magnetische veld, hoe groter de afwijking in het kleurdiagram in het algemeen. De UV-fluxen van de hete Ap sterren blijken zwakker te zijn dan die van normale sterren met dezelfde UBV kleuren. Ze stemmen veeleer overeen met de fluxen die men zou verwachten op grond van het MK-spectraaltype van de Ap ster, in overeenstemming met vroegere'bevindingen van Leckrone (1973). Bij de latere Ap sterren en de Am sterren kunnen we echter dit effect niet waarnemen.
De laatste twee artikels zijn gewijd aan de studie van Ap, Am sterren •en "blue stragglers" in sterrenhopen. In artikel 5 bestuderen we enkele jonge sterrenhopen om na te gaan of hier Ap sterren in voorkomen. Tot voor kort werd nl. algemeen aangenomen dat Ap sterren niet in de zeer jonge sterrenhopen voorkwamen (Jaschek and Jaschek, 1967). Onze studie bevestigt de onafhankelijke bevindingen van Hartoog (1976) dat ooX in jonge sterren- hopen (leeftijd enkele malen 10 jaar) reeds Ap sterren voorkomen. Hieruit kunnen zekere consequenties m.b.t. theorieën over het ontstaan van de abundantie afwijkingen worden getrokken.
De blue stragglers danken hun naam aan het feit dat hun helderheid en kleur ze plaatst aan de blauwe kant van het afbuigpunt van de hoofdreeks van de sterrenhoop in het Hertzsprung-Russell diagram. Tussen hen bevindt zich een hoger percentage (>30%) Ap en Am sterren dan onder de veldsterren (10-20%). Om iets over het ontstaan van deze blue stragglers te weten te komen, hebben we de ster HD 162374 in de sterrenhoop Messier 7 nader bestu- deerd. De resultaten van deze studie sijn gegeven in artikel 6, te zamen met die voor de 2 blue stragglers 40 Cancri in Praesepe en 68 Tauri (=Am ster) in de Hyaden, die door Conti en Stickland werden bestudeerd. Geen van deze drie sterren blijkt met zekerheid een dubbelster te zijn, hetgeen weinig waarschijnlijk maakt, dat ze de pïüdukten zijn van dubbelsterevüluLie. De lage zwaartekrachtsversnelling die uit de spectra van 40 Cancri en HD 162374 blijkt, zou erop kunnen duiden dat deze twee sterren zogenaamde "horizon- tale tak" sterren zijn, vergelijkbaar mee de sterren op de horizontale tak in het Hertzsprung-Russell diagram van bolvormige sterrenhopen. Op grond van sterevolutie berekeningen worden dergelijke sterren ook in oudere open sterrenhopen (leeftijd > 3x10 jaar) verwacht.
Vrijwel alle in dit proefschrift beschreven studies werden verricht in samenwerking met andere onderzoekers, zoals uit de vermeldingen der mede- auteurs blijkt. Het is moeilijk om in elk der gevallen precies de bijdragen van de verschillende medewerkers aan te geven. Mijn eigen bijdrage varieerde van artikel tot artikel. In een aantal gevallen werkte ik mee aan het verrichten van de waarnemingen; in vrijwel alle gevallen droeg ik in meerdere of mindere mate bij tot de interpretatie der waarnemingsgegevens. Het laatste was in het bijzonder het geval in de artikels 2, 3, •-! • 5, 6 en 12 van deel 1 en in de artikels 1 en 6 van deel 2, waarin ik vrijwel geheel de interpretatie en discussie voor mijn rekening nam. In het bijzonder dank ik R.Takens voor het gebruik van zijn programma ORBIT voor het bepalen van baanelementen uit radiële snelheden en J.van Paradijs en E.Zuiderwijk voor het gebruik van hun programma voor het berekenen van synthetische lichtkrommen. P.Conti, H.Hensberge, E.van den Heuvel, C.de Loore, J.van Paradijs, W.van Rensbergen, C.Sterken en E.Zuiderwijk ben ik erkentelijk voor het ter beschikking stellen van hun waarnemingsmateriaal voor deze studies.
Referenties
Hartoog,M.: 1976, Astrophys.J. 205, 807 JaschekjC, Jaschek,M.: 1967, in The Magnetic and Related Stars, ed. R.C. Cameron (Baltimore, Mono Book Corp.) p.287 Leckrone,D.S.: 1973, Astrophys.J. 185, 577 Lucy,L.B.: 1971, Astrophys.J. 163, 95 Petro,L. : 1977, to be published Rappaport,S., McClintock,J. : 1975, I.A.U.Circ.No.2794 Wolff,S.C., Morrison,N. : 1974, Astrophys.J. 187, 69
10 In de eerste plaats ben ik veel dank verschuldigd aan mijn promotor Prof. Dr. E. van den Heuvel die het mij mogelijk maakte dit onderzoek te verrichten en die mij steeds gesteund heeft door zijn interesse in mijn werk en de vele behulpzame discussies die we gevoerd hebben. Hem en Prof. C. de Jager ben ik zeer erkentelijk, dat zij het mij mogelijk hebben gemaakt dit onderzoek aan de Sterrenwacht te Utrecht te beginnen. Met genoegen denk ik terug aan de prettige werksfeer in de sterren- kundige instituten van Brussel, Utrecht en Amsterdam. De Utrechtse Sterrenwacht, en in het bijzonder de heer G. van Gelder, dank ik voor de hulp bij het doormeten van spectra met de microfotoraeter. Het sterrenkundig instituut van de universiteit van Groningen dank ik voor het beschikbaar stellen van haar Grant machine voor het meten van lijnposities in de sterspectra.
11 Introduction and Summary
Part 1. X-ray Binaries. X-ray astronomy of objects outside the solar system came into existence in the late 1950's and the 1960's. During this period a number of X-ray dedectors carried by balloons and rockets surveyed the sky for discrete X-ray sources. The first such source to be detected was Sco X-l in 1962. By 1967, approximately 30 discrete sources had been catalogued, among which supernova remnants such as the Crab nebula and Cas A, explosive radiogalaxies such as Vir A and Cen A, and star-like sources such as Sco X-l, Cyg X-l and Cyg X-2. Several of the star-like sources were identified with optical "stars" but the nature of these objects remained unclear. In December 1970, the first X-ray satellite with the name "Uhuru" was launched. The fourth and final Uhuru catalogue (Forman et al., 1977) lists 339 sources; of about 80 sources optical identifications have been made. The X-ray sky as seen by the uhuru satellite is reproduced in figure 1 and shows that most sources are concentrated towards the galactic plane and particularly into the direction of the galactic center.
galactic north pole
HERCULES X-1 / (3U 1653+351
galactic south pole Fig.1. The X-ray sky as seen by UHURU, plotted in galactic coordinates. The galactic center is at the center of the coordinate system. Some of the most prominent X-ray binaries are indicated.
13 One of the most outstanding discoveries made with the UHURU satellite was that a number of the star-like X-ray sources are double stars. These X-ray binaries are the subject of this thesis. For most of these sources periodic variations in X-ray brightness are found of a type characteristic for eclipsing binaries. The precise periodicity of the X-ray eclipses enables often the identification of the optical companion of the source: this star shows variations in brightness and/or radial velocity with the same period. With Kepler's laws, one can for a number of them derive binary parameters such as the stellar masses. The accompanying table lists some of the characteristics of the binary X-ray sources known to date. The fact that in several systems the X-ray source is a short-period pulsar, and that in all systems the X-ray emission fluctuates on time- scales of fractions of a second indicates that the X-ray sources are very compact objects, i.e. neutron stars, or in some cases perhaps black holes. The X-ray production in these binary systems is believed to be due to the accretion of material lost from the normal star (in most cases an early-type supergiant) to the compact X-ray star. The study of X-ray binaries is expected to produce useful quantitative information about neutron stars and on the physical processes by which the X-rays in these systems are generated. The first part of this thesis describes optical observations of X-ray binaries and their interpretation. We observed photometrically and spectroscopically a number of early-type stars which are identified as companions of X-ray sources. The spectra were obtained with the coude spectrograph attached to the 1.5 m telescope of the European Southern Observatory, La Silla, Chile. Registrations of the spectra were made with the Faul-Coradi microphotometer of the Observatory at Utrecht. To study radial velocity variations, the positions of the spectral lines were measured with the Grant comparator of the University of Groningen. In paper 1 we describe the spectrum of the Of star HD 153919, the optical counterpart of the X-ray source 3U 1700-37. We measured the radial velocities of the stellar lines in order to verify the suggested identification with the in 3.41 days eclipsing X-ray source. The optical star indeed showed radial velocity variations with the same period. This result was confirmed by independent observations by Wolff and Morrison (1974) i
14 Table 1. X-ray binaries.
Spectral Binary Pulsar Distance M . M X-ray Optical opt Source Companion Type "V Period Period (kpc) ( l=optical (sec) 1 2 Her X-1 HZ Her A9-F0 13%-14?7 ifv ' 1.24 2-6 2.18±0.11 1.30±0.14 1 Sco X-1 V818 Sco F8 12?4-13?6 ofs 1.2-2.4 -1 0.7-2 1 Cyg X-2 late A,F 14% ofg 0.6-2.1 -2 ' "" X Cyg X-3 >22m 41?82 10 1 2 Cen X-3 KRZ Star O6lIIe 13?4 2?! ' 4.84 6-9 -20 0.6-1.8 Vela X-1 HD 77581 B0.5Ib 6% 8%11'2 282.9 2 21.2±2.6 1.61+0.27 1 2 4U1700-37 HD 153919 O6.5f 6?6 3?4 ' 1.6 -30 -1.5 1 2 Cyg X-1 HDE 226868 O9.5Iab 8?9 s^e ' 2.5 -30 -9 SMC X-1 Sk 160 B0.5Ia 13?3 3d9l,2 0.71 55-75 15.1+0.3 0.95+0.12 LMC X-4 Sk-Philip cand. O8V-III 13% l^1 50 -20 S2 4U1223-62 Wray 977 B1.5Ia 10% -23d'1 697 2 4U0352+30 X Per O9.5pe 6?7 835 0.35 -20 <2? x0.9? P-Cygni profiles and emission lines in the spectrum give evidence for very high outflow velocities (up to 4000 km/s) of the stellar atmosphere. The presence of P-Cygni profiles in a stellar spectrum indicates that the star loses mass by a stellar wind. In the second paper we compare the observed P-Cygni profiles in the spectrum of HD 153919 with theoretical profiles, in order to obtain an estimate of the mass loss of this star. Using the method of Lucy (1971) (isotropic scattering of stellar light in the outflowing atmosphere) we find a mass loss rate of 2 x 10 Mffl/yr. Papers 3 and 4 describe a detailed study of radial velocities of individual stellar lines of HD 153919. The spectra used in the first paper were complemented with plates obtained by several observers during three subsequent periods. In total, we studied 76 high resolution spectrograms. From the radial velocities we deduced the orbital parameters of the optical star. This could be done more accurately than in previous studies due to the size, resolution and homogeneity of the available plate material. The orbital parameters give information about the masses of both stars. However, as no radial velocity curve is present for the X-ray source, the individual masses could not be derived very accurately. Only a lower limit for the mass of the X-ray star could be derived. Periodic extra variations of the velocity of some of the spectral lines indicate that the velocity of the stellar wind of HD 153919 varies as a function of the position on the stellar surface relative to the companion star. This new result was confirmed by our detection of a variable balmer progression; the stellar wind appears to be strongest near the Lagrangian points L and L . The next three papers describe our photometric observations. We observed the Of star (i) in the Stromgren uvby system, (ii) in the Walraven five colour system and (iii) with the Astronomical Netherlands Satellite (ANS) in five medium band wavelength regions in the ultraviolet. The brightness of the star varies in all channels with a double wave during the 3.41 period. This well-known type of brightness variation of optical companions of X-ray sources is due to gravity darkening effects, caused by the presence of the X-ray star. Significant colour variations are found to be present in the ultraviolet. In paper 7 we analyse all available photometric 16 material. In comparing the mean observed light curve of the Of star with the X-ray light curve of its companion, a systematic phase difference between both curves was found, i.e. the mid X-ray eclipse time does not coincide with the mid occultation time of the X-ray star by the Of star. The first part of the X-ray eclipse seems to be due to obscuration of the X-ray source by the stellar wind of its companion; the wind from the stellar surface following the X-ray source seems to have a higher density than elsewhere. We compared the optical light curve with synthetic light curves, calculated for a tidally deformed star in a binary system. It appears impossible to obtain a static model atmosphere which fully explains the photometric and spectroscopic observations of HD 153919. Apparently, one should take into account dynamical effects, in particular the periodic variability of the strength of the stellar wind as a result of the orbital motion of the X-ray companion. Papers 8-11 describe observations of the X-ray binary HD 77581 (Vela X-l = 3U 0900-40). The observations were carried out in collaboration with other workers of the astronomical institutes of the University of Amsterdam and of the Vrije Universiteit Brussels. In paper 8 we study line profiles and radial velocities of HD 77581. The Ha profile appears to vary in phase with the orbital motion. After the discovery of X-ray pulses with a period of 283 seconds (Rappaport and McClintock, 1975) with the American satellite SAS-3, a detailed analysis of the masses of both components of this binary system could be made by combining our radial velocity measurements of the optical star with the radial velocity curve obtained from time-delay observations of the X-ray pulses. The first results of this analysis, which are of importance for comparison with theoretical models of neutron stars, are given in paper 9. Papers 10 and 11 contain a photometric analysis of HD 77581 in the Stromgren uvby system. The observations show irregular variations in the light curve from one orbit to the next, superimposed on the periodic double wave variation. A phase dependent variation of the colour index cl (balmer jump index) was detected, consistent with our predictions from model calculations of a tidally deformed rotating star in a binary system. 17 The last paper of this series describes our photometric observations of the B-supergiant Wray 977, which on the basis of positional coincidence is a practically certain candidate for identification with the X-ray binary pulsar 3U 1223-62. This X-ray source shows no eclipses, which renders the definitive identification difficult. Our observations indicate a photometric variation of Wray 977 with a period of 23 ± 1 days. This periodicity was recently confirmed by observations by Petro (1977). Part 2. The Magnetic and Related Stars. "Peculiar and metallic-line A stars" are stars of spectral types B, A and F vith abundance anomalies of certain elements in their atmospheres. Both groups are slow rotators; in star clusters they are found near the turn-off points of the main-sequence. They make up some 10 to 20 per cent of the mciin-sequence stars in their spectral region. The peculiar stars (Ap) can be divided into the following groups, according to their surface abundance anomalies: (1) Hg-Mn stars (B-type); overabundant in the elements Hg and Mn; have never a strong magnetic field (<500 gauss); (2) Si-stars (B- and early A-type); overabundant in Si and iron peak elements; have strong magnetic fields (up to c. 35000 gauss); (3) the Eu-Cr-Sr stars (A- and early F-type); overabundant in the iron peak elements and rare earths; have almost always a strong magnetic field (up to 20000 gauss). The metallic line (Am) stars (A- and early F-type) have a too weak Ca II K line for their hydrogen spectral type. In addition, their metallic lines are too strong and one never observes a strong magnetic field (<500 gauss). The magnetic peculiar stars show periodic variations in light, colour, spectrum and magnetic field strength. The metallic-line stars and Hg-Mn stars show no such variations. The commonly accepted model for explaining the variations of the magnetic stars is that of the "oblique rotator" (Stibbs, 1950; Deutsch, 1954). In this model the star is a rigid sphere with a magnetic dipole field of which the axis is inclined with respect to the rotation axis and to the line of sight of the observer. At an arbitrary inclination of the line of sight one will observe that the magnetic field varies periodically due to the rotation of the star: in the most common case one will see alternately both magnetic polarities pass through 18 the line of sight. The Eu-Cr-Sr star HD 98088 is important among the Ap stars as it is the only double-lined spectroscopic binary among the late-type Ap stars. This might unable one, in principle, to determine whether or not both components show similar anomalies; if so this would have important implications for the understanding of the causes of the anomalies. It was therefore decided to study this binary in more detail. During this study it was found that on a high dispersion Zeeman plate, a number of lines of Fe, Cr and Ti show the so called "crossover effect": this is an effect caused by the stellar rotation in combination with the existence of two large magnetic patches of different polarity on the stellar surface. The detection of this effect enables one to determine the actual surface field strength in the magnetic patches. The results are given in paper 1. The magnetic patches appear to have a mean field strength of about 8000 gauss,- the earlier determined value of ths "effective" field strength of the star (i.e. the field strength integrated over the entire stellar surface) was only 1000 gauss. Most magnetic Ap stars show periodic photometric variations. A commonly accepted explanation for these light variations is an inhomogeneous brightness distribution over the surface of the rotating star, in some way due to the magnetic patches on the stellar surface. Paper 2 and 3 describe photometric observations of 8 Si-stars and one Sr-Cr star in the Stromgren uvby system. For seven of the Si-stars a period was determined and a light curve was plotted. The other two stars are probably also variable, but we have too few observations to derive a period. The amplitude of the light variations of all observed Si-stars decreases towards shorter wavelengths, in accordance with observations of a null wavelength region (no variation) for Si-stars in the ultraviolet; light curves at both sides of this null wavelength region are in anti-phase. In paper 4 we present the results or our ultraviolet photometric observations with ANS of 79 Ap stars and 26 Am stars. In colour-colour diagrams the positions of especially the cooler Ap stars differ considerably from those of the normal A stars. This is probably due to blanketing effects produced by the many strong ultraviolet absorption lines of the overabundant metals. The strength of the anomaly seems to depend on the magnetic field 19 strength: the stronger the magnetic field, the larger the anomaly in the colour-colour diagram. Therefore, with a proper calibration the colour anomaly could be usable for estimating the strength of the surface magnetic field of the star. The hot Ap stars appear to be flux deficient in the ultraviolet compared to normal stars with the same UBV colours. Their ultraviolet fluxes agree with those which one would expect on the basis of the MK spectral type of these stars. This is in agreement with earlier findings by Leckrone (1973). This effect disappears, however, for the later type Ap stars and for the Am stars. The last two papers are devoted to the study of Ap and Am stars and blue stragglers in open clusters. In paper 5 we study some young clusters to examine whether they contain Ap stars. Until recently, it was assumed that Ap stars were not found in very young clusters (Jaschek and Jaschek, 1967). Our study confirms the independent finding by Hartoog (1976) that there are also young clusters (with a lifetime of a few times 10 year) that contain Ap stars. This has certain consequences for theories on the origin of the abundance anomalies in these stars. Blue "stragglers" in star clusters are stars which are found at the blue side of the turn-off point of the cluster main-sequence in the Hertzsprung- Russell diagram. Among the blue stragglers in clusters there is a higher percentage (>30%) of Ap and Am stars than among field stars (10-20%). To possibly obtain more information on the origin of these blue stragglers -notably to test the hypothesis that they might be evolved close binaries-- we made a spectroscopic study of the star HD 162374 in the cluster M7. The results of this study are given in paper 6, together with the results for the blue stragglers 40 Cancri in Praesepe and 68 Tauri (an Am star) in the Hyades, which were studied by Conti and Stickland. None of these three stars shows evidence for being a spectroscopic double star , a fact which makes it rather unlikely that they are the products of close binary evolution. The low gravity, which is found for 40 Cancri and HD 162374, suggests that these two stars might be "horizontal branch" stars, i.e. population I analogues of stars on the horizontal branch in the Hertzsprung-Russell diagram of globular clusters. On the basis of stellar evolution calculations one expects such stars to be prese?vt also in older open clusters (age > 3 x 10 yr). 20 References Deutsch,A.J. : 1954, Trans. I.A.U. 8_, 801 Forman,W., Jones,C, Cominsky,L., Julien,P., Murray,S., Peters,G., Tananbaum,H., Giacconi,R. : 1977, Astrophys.J.Suppl.Series, preprint Hartoog,M. : 1976, Astrophys.J. 205, 807 Jaschek,C, Jaschek,M. : 1967, in The Magnetic and Related Stars, ed. R.C.Cameron (Baltimore, Mono Book Corp.) p.287 Leckrone,D.S. : 1973, Astrophys.J. 185, 577 Lucy,L.B. : 1971, Astrophys.J. 163, 95 Petro,L. : 1977, to be published Rappaport,S., McClintock,J. : 1975, I.A.U.Circ.No.2794 Stibbs,D.W.N. : 1950, Monthly Not. Roy. Astron. Soc. 110, 395 Wolff,S.C., Morrison,N. : 1974, Astrophys.J. 187, 69 21 PART I. X-RAY BINARIES 23 Aolliipli.vs ll>. 6<> 7SH97J) The Spectrum, Orbit and Magnetic Field of HD 153919 (2 U 1700-37) G. Hensberge, E. P. J. van den Heuvel* and M. H. Paes de Barros* Astronomical Institute. Utrecht Received August 3. 1973 Summary. A radial velocity study of the O 6.5 f star around Cm A4650 (and probably also around Hen HD 153919 (optical candidate of 2 U 1700-37) shows X 4686) and indicates an outflow velocity of the order of this star to be a 3d4t2 spectroscopic binary with a 4000 km/s. Problems with the determinations of masses mass function of 0.0036 M0± 20%. A recurrent dis- of early-type supergiants in X-ray binaries are discussed turbance of the radial velocity curve around phase 0.30 and a lower mass limit of between 0.7 MQ and is noticed. From 10 Zeeman spectra no magnetic field 1.5 MQ is obtained for 2 U 1700-37. of strength larger than the probable error (3500 Gauss) is found. The shallow emission features of 100 to 180 A wide around Hf and Hr as reported by Walker are Key words: X-ray sources - Of stars - magnetic fields confirmed. A similar feature of over 120 A wide is found - emission lines 1. Introduction HD 153919 is the optical candidate for the 3*412- Eight of the other spectra were taken in the yellow red period eclipsing X-ray source 2 U 1700-37 (Jones et al, (emulsion 103 aF), covering the region between 4500 A 1973). It is a 6T6Of star. A first description of the and 6600 A (dispersion 19 A/mm). spectrum was given by Walker (1973). Photometric Ths blue spectra (emulsion IlaO, dispersion 12 A/mm), observations by Jones and Liller (1973) suggest a period cover the wavelength region between 3400 and 4900 A. of variation similar to the X-ray period. In order to The blue as well as the yellow-red spectra contain the examine the radial velocity variations of HD 153919 important strong emission lines He n 4686, N m 4634. one of us (E.v.d.H.) obtained 25 high dispe'rsion coude 4641-42 and the blend C m 4647, 4650, 4651. spectra with the 152-cm telescope of the ESO, La Silla, In the Sections 4 and 5 we give a brief description of Chile, between March 31 and April 10, 1973. Table 1 the magnetic measurements and of the most prominent lists the spectra. emission features. A discussion of the possible mass A study of the radial velocities of the hydrogen absorp- of the X-ray source and the uncertainties involved in tion lines H8, Hr and H6 revealed a variation with a the determinations of masses of binary X-ray sources 3d4l periodicity, an amplitude of about 23 km/s and the is given in Section 6. same phase as the X-ray variations (van den Heuvel, 1973), thus confirming the identification of HD 153919 2. Identification and Radial Velocities with 2 U 1700-37. This result, as well as the observation AH our spectra were traced with the digitized Faul- that the radial velocity of the He n 4686 emission line Coradi comparator/microphotometer of the Utrecht varies with about the same amplitude and phase as the Observatory and - after Fourier-noise filtering - plot- absorption line velocity, were confirmed by a study of ted in intensity. As the spectrum fairly closely resembles 17 high dispersion coude spectra by Wolff and Mor- that of the O 5 f star f Puppis, we used for the identifi- rison (1973). A detailed analysis of the radial velocity cation the detailed linelist of this star given by Baschek variations based on a combination of our data with those and Scholz (1971). Table 2 lists all lines recorded in the of Wolff and Morrison is given in Sections 2 and 3. spectrum. From the ratio of the strengths of He i 4471 In order to study the magnetic field of about and Hen 4541 (Conti and Alschuler, 1971) Wolff and 10000 Gauss as reported by Kemp (1973) ten of the blue Morrison classified the spectrum as O 6.5 f; our spectra spectra were obtained with the Zeeman analyser on confirm this classification. An outstanding feature is the loan to the ESO from AURA. These spectra are in- large intensity of the C in X 4650 emission blend which, dicated with a Z in Table I. with an equivalent width of over 2.5 A is exceptionally * Also ut the Astrophysical Institute, Vrije Universiteit, Brussels. strong for an Of star (cf. Walker, 1973; Conti, 1973a, b). •* On leave from Mackenzie University, SSo Paulo, Brasil. Diffuse interstellar bands are present near the wavc- 25 G. Hensberge el ul. liiblc I. Ik Irii. rjdml velocities. The blue plates (12 A/mm) arc listed in the upper part of the table, the red plates (19 A mini in the tower p;nl I'laie- J.D. Phase H8 + Ht + H4 H, Hen Hen Hei He II He II N ill N in Cm number 244IOOO + 4026 4200 4471 4542 4686c 4634c 464 le 4650c G 3S29Z 772.790 .68 - 91 -104 7 -55 -47 G 3830 Z 772.848 .70 -103 -125 -76 - 62 - 74 9 -84 -53 - 97 G 3836Z 773.735 .96 -115 -126 -69 - 92 - 54 4 -39 -49 - S7 G 3839 773.895 .01 - 59 -- 78 -49 - 52 - 39 10 -46 -54 -122 G 3844Z 774.753 .26 - 74 - 98 - 56 36 10 -15 G 3853Z 775.725 .54 - 82 -127 - 78 - 158 - 69 23 -14 -26 G3863 776.812 .86 - 93 -131 -81 - 77 16 -20 -16 Ci3874 778.824 .45 - 60 - 96 - 80 - 38 34 43 -26 G 3887 779.764 .73 -100 -110 -62 - 85 - 62 - 5 -73 -57 G 389OZ 779.878 .76 -122 -128 - 68 14 -41 -51 G3898Z 780.744 .01 - 81 -123 -15 - 72 - 93 26 -43 -30 G 3900 Z 780.857 .05 - 85 -123 -58 - 50 -105 - 64 36 - 8 -28 Ci 3912 781.744 .30 - 63 - 95 -40 - 38 - 55 22 90 -30 - 32 (i 39I4Z 781.830 .33 - 59 -112 -64 - 62 - 37 28 _ 9 - 8 G 3925 782.670 .58 - 89 -111 -55 - 50 -113 - 55 20 -44 -29 - 31 G 3927Z 782.774 .61 - 87 -125 - 9 - 97 - 68 -107 19 -47 -49 - 99 Ci 3931 782.917 .65 -101 -124 -50 - 54 10 -25 -52 H,(e) Cm Civ Civ Hen Hen N in N in Cm 5696e 5802 5812 4542 4686c 4634c 4641c 4650c G 3838 773.849 .99 23 -56 -139 - 95 - 67 15 -54 -56 - 145 G 3847 774.908 .30 118 -55 - 74 - 44 60 G 3855 775.879 .59 50 -57 -115 - 98 27 -45 -37 G 3864 776.869 .87 87 -70 -135 -117 - 55 29 -48 -77 G 3888 779.788 .73 130 -50 -122 -114 - 85 2 -47 -71 G390I 780.906 .06 115 24 -26 -60 G3913 781.777 .31 82 -21 -110 - 80 - 49 12 _ 2 -23 G 3926 782.706 .59 84 -29 - 99 -107 14 -10 -38 Table 2. Line-identification list Table 2 (continued) Identification Identification 3751). 15 H,2 4503.7e Unidentified 3770.63 H,, 4541.61 He u 2 3797.90 H,o ( + 3796.33 HeII 5) 4634.14e N in 2 3835.39 H, ( + 3833.80 He II4) 464l.25e N in 2 (blend of/. 4640.64 and 4641.90) 3889.05 H8 ( + 3887.44 He II4) 4649.70e C in 1 (blend of A. 4647.40,4650.16.4651.351 3923.48 He II4 4659e (weak, possibly C IV 4659 + 4660) 3933.66 Ca II K. interstellar 4685.74e Heni 3968.47 Ca II H interstellar 4861.33 Hc( +4859.32 He II2) 3970.07 H,( +3968.43 He II3) 5411.52 He II 2 4025.60 He II3 1 5696.0e Cm 2 4026.2 He I 13/ 5780.5 Diffuse interstellar 4097.31 N in 1 5801.51 Civ 1 4101.74 HJ + 4100.04 He II3) 5812.14 Civ I 4103.37 N ill 1 5875.64 He 1 11 (blend of X 5875.62. 5875.65. 5875.991 4ll6.IOe Si iv 1 5889.95 Na I D, interstellar 4199.86 He 11 3 5895.92 Na i D, interstellar 4340.47 H, (+4338.67 Hen 3) 6283.9 Diffuse interstellar 4430.6 Diffuse interstellar band 6310.8 He II 7 4471.48 He I 14 6527.1 Hen 4485.7e Unidentified 6562.82e H,( + 656O.IOHeu2) lengths k 4430, 5780 and 6284 A. The central depth of were measured independently. The resulting radial the 4430 band is between 8 and 10 per cent. All plates velocities are listed in Table I. For Zeeman spectra the were measured with the digitized Grant comparator of mean of the radial velocities of upper and lower spectra is the ESO, Santiago, Chile. On some weak plates not all given. The radial velocities of the strong interstellar lines lines could be measured. On Zeeman plates, both spectra of Can and Nai are - 12.4+0.4 and -9.5±0.3 km/s. 26 Ill) 151'II'M: t' 17110 l.'i ( H".|nviiwl\ I In- ladial velocities of stellar lines differ (cl. Ilulchings. IX>SI. pivsuiiiahK due in instabilities svsiciiKitKiiiiv fiuiii line to iiiK-. the emission iines giving in !he atmospheric outflow. larger velocities by amounts up to l50km/s than the A best fit sine curve is drawn through the points in absorption lines (ii. WolIT ;ind Morrison). Also differ- Fig. la. It shows that within the observational uncer- ent absorption lines differ in radial velocity in a systema- tainties a circular orbit gives a good 111. The radial tic way. The well known Balmer progression (increase in velocity amplitude thus derived is K, = 22 ± 1.5 (p.c.) radial velocity towards the lower series numbers) is km/s. The conjunction in the radial velocity orbit occurs clearly present. These systematic differences make it exactly at phase 0.00. [This is remarkable because of impossible to define a unique radial velocity for the the uncertainty of 0^23 in the moment of the mid- star. The much higher radial velocities for the emission occultation of the X-ray source predicted for April 1.37. lines than for the absorption lines can be explained by 1973 from the orbital elements given by Jones cl 27 G. Hensberge el al. Hen 4541.4^00 "xx ^. X/ -100 60 -H 1 1- -80 0.1 0.2 03 0.4 0.5 0.6 0.7 0.8 0.9 10 PHASE Kig. !. The radial velocities of absorption and emission lines plotted in a 3?4I2 period. Crosses indicate our measurements, dots arc IIK measure- ments by Wolff and Morrison (1973) that the blue wings are displaced by 40 to 60 km/s. In (2) the velocity disturbance is a kind of wake due to the view of the above mentioned phase shift of 0.06, phase passage of the X-ray source through the outflowing 0.30 for the emission lines corresponds to phase 0.24. gas. Assuming Tga5 = reff w 4 x 10* K (cf. Conti, I973c», Around this phase the X-ray source is approaching us the sound velocity in the outflowing matter (ionized at its highest velocity which strongly suggests that this hydrogen) is about 30 km/s. Hence, the motion of the object is the cause of the disturbance. However, the X-ray source is highly supersonic. As the wake is ex- orbital velocity of the X-ray source is expected to be pected to propagate with the sound velocity, the ob- several hundreds of km/s (as the mass ratio of the system served velocity disturbance might be compatible with is likely to be larger than 10, cf. Section 6). For this an explanation of this type. Photometric observations reason it seems very unlikely that the velocity distur- may possibly provide further clues for the explanation bance is associated with the X-ray source itself. We of the observed velocity disturbance. suggest two possible explanations: (1) the presence of the X-ray source heats the out- flowing matter in such a way that the outflow velocities 4. Magnetic Field are locally enhanced; a cloud of gas which follows the The wavelength differences between lines in the right- X-ray source in its orbit partly shields the outflowing and left-hand polarized spectra on the Zeeman plates matter behind and aside of the X-ray source from this are as often negative as positive on the same plate. heating, and causes the velocity enhancement to occur The z-values for most of the lines are unknown. mainly in the direction of the orbital motion. The Adopting z = 1 (the value for a classical harmonical secondary dip in the X-ray light curve around phase oscillator) one finds magnetic fields between zero and 0.5 (Jones et al, 1973) seems indeed suggestive for the ±2000 Gauss with a probable error of 3500 Gauss. presence of a cloud near the X-ray source. As this holds for ten Zeeman plates taken on eight 28 HD 153919 (2 U 1700-371 73 HD 153919 t G 3914 Fig 2 Very broad emission features in the vicinity of Cm X 4650 and Hell X 4686 indicating outflow velocities of about 4000 km s different nights, our observations do not confirm the is the first Of star in which velocities of this order have presence of a magnetic field with a strength of about been found from the optical spectrum. 10000 Gauss as reported by Kemp (1973). Similar On some plates there is a weak broad emission line at a conclusions were reached by Angei et ai. (i973) and by wavelength of about 4660 A (see Fig. 2). A possible Wolff and Morrison (see note). identification is C iv 4659 + 4660. It might be that this line and not Cm 4650 is the center of the above mentioned broad emission wings. In that case the out- 5. Emission Features flow velocities will be even larger ihan the above mentioned values. Table 2 indicates for which lines emission is observed. A number of the prominent emission features were described by Walker and by Wolff and Morrison. Some Helium important features not yet considered in detail will be described here. He i X 5876 has a variable P-Cygni type profile. The ab- sorption and emission components have widths of over 10 A each and equivalent widths of over 2 A. He I 4471 also has a P-Cygni type profile but its Hydrogen and Carbon ill and IV emission is very weak. H., has a variable profile with a halfwidth at the conti- He ii X 4686 has a fairly symmetric profile with a half- nuum of mostly over 21 A, indicating outflow velocities width at its base of about 5.7 A, indicating an expansion of over 900 km/s. The central intensity is about 2.0. H^, velocity of about 370 km/s. Sometimes the profile has a has a shallow P-Cygni type profile. Hv shows weak and flat top. The behaviour of the radial velocity of this variab'c emission with a width of 5 10 A at its red line was described in the foregoing section. side. To the immediate red of H4 there is a clear emis- sion feature about 5 A wide. Somewhat further to the red is Si IV 4116 also in emission and variable in 6. The Mass Function and the Masses of Non-pulsating strength. As was first noticed by Walker, underlying Hy X-ray Sources and on some plates also Ha there is a very broad shallow emission feature in the continuum, over 100 A From the orbital elements derived in Section 3 a mass wide. A similar feature, probably centered around the function follows of Cm /. 4650 blend (Conti, 1973b) has a total width of 3 f{M,, Mx) = M* sin i/(Mx = 0.0036 Mo . (1) about 120 A and a central intensity of sometimes as much as 1.2. The plot in Fig. 2 clearly shows this The probable error in this value, due to the p.e. of feature. The figure shows that similar wings may also Ky is ±20%. Here Mx and Ms denote the mass of the be present around Hen 4686. As these very broad X-ray source and of the Of-star, respectively, and / is features are also present during the X-ray eclipse, they the orbital inclination. This mass function sets a lower are not connected with the X-ray source. The outflow limit to the mass of the X-ray source by (cf. Bolton, velocities corresponding to the widths of these very 1972) broad emission wings range from 3000 to 4000 km/s. 3 > 0.154 Mi*' . Such velocities are of the same order as the outflow (2) velocity of 3400 Lm/s deduced by Stecher (1968) from In order to determine the mass Ms, the distance should the profile of the C iv 1550 line in the rocket ultraviolet be known. The large strength of the X 4430 interstellar spectrum of f Puppis. To our knowledge, however, this band (cf. Section 2) together with an equivalent width 29 ?4 (i. McnsbciiiL1 (7 i//. I .ihle i ! .nvi-j ir.i-v. iiinu.s of iinn-pulsating binary X-ray .sources. The A/, values arc tliL1 smallest OIKS possible compatible wiih speur;tl tv aiui lumuio>M\ cl.iis of (he primaries, in ease thai mass loss has taken place. Values in the case of no mass loss are giu'n in parentheses Optical Binitry Spectral Mass function Primary A/, (A? 1 A/.I.W 1 source candidate period type fiSDiM... mass (i 51' 1 Idaysl AJ,(AC 1 HD 226868 5.6 O9.7Iab 0.(82 2 10(10) 22.6(5.5) --.VOI6.4) 2 1:1)90(1 41) HD 775X1 8.93 B 0.51b 0.016 > 8(24) 2 1.0(2.1) ~.r 1.2(2 5) : i; now 37 HD 1539)9 3.412 O6.5f 0.0036 > 10130) of 0.5! A of the interstellar Ca n K-line indicate, for a In Table 3 we compare the lower mass-limit of star in this direction a distance of about 2 kpc (cf. 2 U 1700-37 with those ol the two other non-pulsating Munch. 1968). Such a distance is supported (Bessell galactic X-ray binaries. H D 226868 (Cyg X-1) is classified et «/.. 1973) by the reddening E(fi- F) = 0.58, which is O 9.7 lab (Walborn. 1973) and HD 77581 (2 U 0900 40) larger than that of the Of members HD 152408 and has spectral type B0 lb(Hiltnerc( til., 1972). Comparison HD 151804 of Sco OB I, which have a distance of about with evolutionary tracks show that such spectral types 2 kpc and are situated in the same direction (cf. Craw- correspond with masses of about 30 Mo and 24 W,.. ford, 1971). After applying a reddening correction respectively (cf. van den Heuvel, 1968). However, as AV = SE(B~V) one then finds M,.g-6?6, in good argued above, if possible mass loss is taken into account, agreement with the value of -6™3 given by Conti and the same spectral types and luminosity classes might Alschuler (1971) for O6.5f stars. Bolometric correc- correspond to masses of only 10 MQ and 8 M5; respec- tions for Of stars are poorly known but probably ex- tively (it is assumed here that these supergiants have ceed 3m (cf. Underhill, 1966). Even with the moderate normal luminosities and are not the peculiar low-lumi- m assumption of B.C. = 2 one finds Mbol = - 8T6 which nosity objects discussed by Trimble et al., 1973). The 5 yields a bolometric luminosity of 2 x 10 L@. If the star lower mass limits in Table 3 were computed for i =90 has not lost mass, comparison with Stothers's (1966) and / = 59 (the latter value corresponding to the value 3 3 evolutionary track for a 30 MQ star shows that of (sin i)" for a random distribution of inclinations). M > 30 Ma. Such a mass is in the same range as the and using the radial velocity amplitudes of HD 226868 masses of the O 7 f and O 8 f members of the eclipsing and HD 77581 given by Bolton (1972) and Hiltner et at. binaries UWCMa and HD47)29 which are 24Mo (1972) respectively. and 64 A-/,,,, respectively (Stothers, 1972). The table shows that Cyg X-l is the only system in However, since HD 153919 is a close binary, mass loss which the X-ray star is certainly very massive possibly may be a problem. The luminosity and radius of a more massive than the upper mass limit for neutron massive post-main sequence star are practically uniquely stars. determined by the mass of its helium core, regardless of the mass of the hydrogen-rich outer layers (cf. Acknowledgements. We are indebted to the ESO Stuff and Directorate Giannone et «/., 1968). For instance, Kippenhahn's for the use of observing and reduction facilities in Chile, to h. N. (1969) calculations of the evolution of a close binary Walker and P. S. Conti for stimulating correspondence and to with components of 25 M and 15 M show that after S. C. WolfTand N. Morrison for a preprint on HD I539t9. o Q Part of this research was supported by a grant from the Kundacao the 25 Mo star has lost its 16.46 MQ) envelope to its de Amparo a Pesquisa do Estado dc SSo Paulo (HAPliSP) to companion, the remaining 8.54 Mo star has about the M. H. Pacs de Barros. same effective temperature and luminosity as a 25 Mo early-type supergiant. The 8.54 MQ star practically is the helium core of the 25 Mo star surrounded by a very extended hydrogen-rich envelope with a mass of only References 0.4 M,,. Spectroscopically it would be very hard to Angel.J. R. P.. McOraw.J.T.. Stockman.H.S. IW. /l.vm./i/iyv ./. distinguish this star from a 25 Mo early-type supergiant. (in press) The helium core of a 30 MQ post-main sequence star Baschek.B., Scholz,M. 1971, Aaron. & Asirophys. 15. 2X5 Bessell,M.S., Peterson.B. A., WickramasinghcD.T. Vidal.N.V. has a mass of about 10 Mo. Hence, the mass of 1973, to be published HD 153919 might be as low as only 10Mo. With Bolton, C. T. 1972, Nature 240, 124 V/4= |() ,V/0 one obtains from Eq. (2) a lower mass Conli, P.S. 1973a, Asirophys. J. 179. 161 limit of().7 M(! for the X-ray component. For M = 30 ;Vf0 Conti.P.S. 1973b. personal communication Conti.P.S. 1973c. Astrophys. J. 179, 181 a lower mass limit of 1.5 MQ is obtained. Such masses = Conti.P.S., Alschuler, W. R. |97l, Astrophvs. J. 170. 325 are consistent with a neutron star, although for i<90 Crawford,D.L., Barnes.J.V., Hill.G., Pcrry.C.L. |97|. Asmm. ./. a black hole is also possible. 76,1048 30 HD |539I9(2(J 1700-371 Ciiiinnonc. I'.. Kohl. K.. Weigert.A. 1968. ?.. Aslrophys. 68, 107 Walborn.N.R. 1973, Aslrophys. J. 179. L 123 v;in den lluuvul.li. l'..l. 1968, Bull. Antrim, lust. Nelh. 19, 309 Walker.E.N. 1973, Monthly Notices Roy. Astron. Soc. 162. 15 P van den lleuvcl.F:. P.J. |973, Int. Astron. Union Circ. Nr. 2526 Wolff,S.C.. Morrison,N. 1973, Aslrophys. J. (in press) Ililmur.W. A.. Werner.J.. Osmer. P. 1972. Aslrophys. J. 175, L 19 Mulchings..!. K. 1968. Monthly Notices Roy. Antrim. Sue. 141, 219 .lonexC. l.iller.W. 1973. Int. Antrim. Union Circ. Nr. 2503 G. Hcnsbcrge Junes. C. Korman.W.. Tananbaum.H., Schrcier,E., Gursky.H., Kcl- E. P. J. van den Heuvcl logg.l-.. Giaceoni.R. 1973. Aslrophys. J. 181, L 43 M. H. Pacs de Barros Kcmp.J.C. 1973. Int. Astron. Union Circ. Nr. 2512 Stcrrekundig Instituul Kippenhuhn.R. 1969, Astron. & Aslrophys. 3, 83 Sterrewacht "Sonnenborgh" Munch.G. 1968, in Stars and Stellar Systems. Vol. 7, Ed. B. M. Servaas Bolwerk 13 MiUdlehurst and L. H. Allcr. Univ. of Chicago Press, Chicago. Utrecht. Nederland p. 365 Steelier. T. P. |96«, Wolf-Rayet Stars. Ed. K. B. Gebbie and R. N. Thomas. National Bureau of Standards, Washington, p. 67fT. Slot hers. K. 1966. Aslrophys. J. 143, 91 Note added in Proof. Recently Kemp (Aslrophys. J. 185. I. 21. 19731 Siothers.R. 1972. Astrophys. J. I7S, 431 found that the circular polarisation in HB shows rapid fluctuations on Trimble.V.. Rose.W.K... Weber.J. 1973. Monthly Notices Roy. a time scale of Jess than 10 m. giving smaller average values over Asiron. Soc. 162. IP periods of one to three hours. Since ihe exposure times of our L'ndcrhill.A.B. 1966. The Early Type Stars, Reidel Publ. Comp. Zccman plates always exceeded one hour, the small magnetic field Dordrecht values recorded from our plates may be due to this effect. 31 Astrim & .Vtroph)-.. 36. 295 29K (1974) The Expanding Atmosphere of the Of Component of the Eclipsing X-ray Binary HD 153919 G. Hensberge AMronomiciii Insiiuile, University of Amsterdam Rceciied Jul> :J. 1974 Summary. Mass loss rates from the Of star HD 153919 the observations). The relation between the mass loss are estimated by comparing the observed P-Cygni rate obtained and the one needed to produce the profiles with those theoretically computed using the observed X-ray intensity is discussed, and the agreement method described by Lucy (1971). The observed profiles is found to be satisfactory. arc best reproduced by adopting a rate of mass loss of h about 2x10 Mo/yr, and assuming an outward Key words: X-ray sources --• Of stars mass loss increasing expansion velocity (which is compatible with P-Cygni profiles 1. Introduction A description of the optical spectrum of the O6.5f star Due to the high expansion velocity of the layers of the HD 153919, the optical counterpart of the X-ray source atmosphere of HD 153919 where the observed P-Cygni 311 1700-37. is given by Hensberge et at. (1973) and Wolff profiles are formed (cf. Section 3), we neglect collisions and Morrison 11974). In this paper, we analyse the and assume coherent scattering as the principal agent P-Cygni type profiles of the lines Hei-i 5876, /.447I of line formation (cf. Rottenberg, 1952). and 11/)'. which arise in the expanding atmosphere of In his computations Lucy assumes that the expansion the Of star. We compute theoretical line profiles and velocity is a function increasing with distance from the estimate the rate at which mass is lost from the stellar star and that the outflow velocity is large compared to atmosphere by comparing the computed profiles with the thermal velocity of the scattering ions (the so-called the observed ones. The mass loss rate found allows us to "narrow-line limit"). The latter approximation is check theories which interpret the X-ray emission as a certainly not valid for the deeper photospheric layers consequence of mass transport from a nearby compan- where the expansion velocity is small: these layers ion (e.g. Lamb et «/., 1973: Davidson and Ostriker, contribute strongly to the line profile close to the line 1973). center. We should thus be aware that in this In Section 2 we give a general description of the method approximation we cannot expect to predict the profile used to compute the theoretical profiles. The velocity near the line center correctly. distribution in thestellar envelope is discussed in Section For the expansion velocity v(R) we adopted the profile 3: a short description of the observed profiles is also described in Section 3. The rate of mass loss is taken as given, and the physical parameters used in the a second free parameter. The density g{R) is then deter- computations are described. The results found are mined by the equation of mass conservation. The presented and discussed in Sections 4 and 5. radiation field of the underlying stellar photosphere forms the third parameter which enters in the compu- 2. The Method tation of the ionization balance. The method used for the theoretical computation of P- (. ygni profiles is the one described by Lucy (1971). This 3. Computations author solved the equation of transfer in spherical The observed P-Cygni type profiles in the spectrum of symmetric expanding atmospheres for lines formed by HD 153919 in the wavelength region /L/. 3300-6500 are isotropic scattering, which is coherent in the comoving H /? and He IX 4471 and /. 5876. The observed profiles of frame of the fluid. The number of resonance absorption these lines during the X-ray eclipse are indicated in Figs. lines arising in the ultraviolet spectral region of early 1, 2 and 3 by the full lines. This phase was chosen to be type stars seems indeed to be sufficient to produce an sure that the X-ray source does not disturb the observed expanding envelope by radiation pressure (Lucy and profiles as inspection of the profiles on tracings of the Solomon, 1970). different plates gave evidence for a stronger absorption 33 296 G. Hensnerge 09- 0.9- -10 5 10 Kig. 1: H /(-profile or HD 153919. The observed profile is indicated by Fig. 3. Her5876 profile of HD 153919. The observed profile is the full line: the crosses represent the theoretical profile indicated by the full line: the crosses represent the theorelical profile for the model with T,,, = 35000 K, log;; = 3.3 and velocity field for the model T,,, = 35000 K, log.« = 3.3 with velocily field 025 b o i(R)=---35O(/(.R,- l) ikm;s] and mass loss rate 2x ]0~ Mo/yr t'(RI = 4IO(/?/R,- l) -"[km7s] and mass loss rate 2x 10 "A/, yr. The dashed line indicates the profile that was observed on some plates and which was not fitted by any theoretical profile reasonable fits for the Her lines with different values of u0 for each profile (see Figs. 2 and 3). The populations of the ionization stages of hydrogen and helium in the layers above the photosphere are controlled by radiative processes, rather than collisional ones. This can easily be checked from the expressions for the ratio of the ionization rates for these two processes (cf. Bohm, 1960). Therefore, we used the ioni- -5 0 5 10 zation equilibrium equation given by Eq. (14) of Lucy l-ig. 2. He 14471 profile of HD 153919. The observed profile is and Solomon (1970). The radiative recombination co- indicated by ihc full line; the crosses and the dots represent the efficients «„ to state n, used in this equation, were taken theoretical profile for the model Tc,, = 35000 K, logg = 3.3 with velocity field r(R) = 27 34 The Expanding Atmosphere of HD 153919 297 with Tcrr = 40(KX) K, log TR(Hi) =31 100 K (32300 K), [Another profile observed at this phase cannot be fitted with any combination of the parameters, as the total where the values for the model Tett = 40000 K, logd = 4.0 are given in parenthesis. It should be noted width of the absorption part of the profile is too small to fit it with any reasonable model of outflow.] here that a reduction of TR in the ionization equation implies a reduction of the rate of mass loss The theoretical profiles were compared with the observ- that is required to obtain the same theoretical line ed ones by shifting them along the wavelength axis. profile. This way of fitting had to be applied because of the poor The photospheric radius of the star was adopted to be accuracy with which the stellar heliocentric velocity is 20 R (cf. Wolff and Morrison, 1974). known for thb star, due to the expansion of the o atmosphere. For the three profiles we applied the same velocity shift, as the observed absorption cores show the 4. Comparison with the Observations same radial velocity (Hensberge er al., 1973). From Figs. 1, 2 and 3 can be seen that the use of the same For H/i the best fit of the theoretical profiles with the velocity shift gives indeed a good fit with the observed profile was obtained for a rate of mass loss of observations for all three lines. 2x 10~°Me/yr and the velocity parameters given in Section 3. [For the model Teff = 40000 K, log g = 4.0 a 6 S. Discussion of the Results rate of mass loss of 2.3 x 10~ Mo/yr was required to 6 give (he same profile.] The observed profile together The high mass loss rate of about 2 x 10 ~ Mo/yr which with the best fit theoretical one obtained from the we need to explain the observed profiles does not indicated model parameters is plotted in Fig. 1. The agree with rates theoretically predicted from radiation variation with R of some of the important model pressure for "steady-state" atmospheres, by Lucy and parameters (velocity, velocity gradient, density, number Solomon (1970) [for this type of star ~2x 10"* Mo/.vr]. of scattering ions), is given in Table 1. We tried However, the rate does agree with the estimate obtained mostly to fit the far blue and red wings of the by Hutchings (1974) for this star from comparison with profile, as the fit of the line center is probably not other Of stars. It is also compatible with observations accurate (see Section 2). of other early type stars. Morton (1967) found mass 6 loss rates of ~ 1 x 10" Mo/yr for OB-supergiants from profiles of ultraviolet resonance lines. These stars do not Table 1. Parameters, variable with radius, used in the theoretical show P-Cygni type profiles in the visible spectral region 6 H/(-profile (plotted in Fig. 1) for a mass loss rate of 2x 10~ Mo/yr (except for H a). The star P-Cygni itself, however, has also this type of line in the visible spectral region, and LOG[l(R)] LOG [dv/dR] LOG[e(R)] LOGKJ 4 has the much higher mass loss rate of about 10~ Mo/yr 1.001 6.794 -2.954 -12.085 -7.983 (Pottasch, 1970). 6 1.1 7.294 -4.454 -12.667 -8.351 A mass loss rate of 2x 10~ Mo/yr is in the correct 1.2 7. .169 -4.680 -12.818 -8.386 range to explain the observed X-ray intensity of the 1..1 7.41.1 -4.812 -12931 -8.407 companion. We note that the shape of the X-ray intensity 1.4 7.445 -4.906 -13.027 -8.422 1.5 7.469 -4.978 -13.111 -8.435 curve before the eclipse as well as the long duration of 2.0 7.544 - 5.204 -13.436 -8.481 the eclipse itself (Jones et al., 1973) indicates that .1.0 7.619 -5.430 -13.864 -8.539 absorption of X-ray radiation by a strong stellar wind 5.0 7.695 - 5.656 -14.382 -8.606 takes place. For an X-ray luminosity of 101|i-10J8 erg/s, 10.0 7.78.1 - 5.920 -15.073 -8.691 9 an accretion rate of around 10 ~ Mo/yr is needed for a 35 298 G. Hensberge compact companion of a few solar masses (Davidson References and Ostriker. 1973). To obtain this accretion rate as a Aucr.L.H.. Mihalas.l). 1972, Astroplm. J. Suppl. 24,193 consequence of mass loss from the Of star at a rate of Bcihm. K.H. l%(i. in Stars tiiul Stellar Systems. Vol. h. Id. .1.1 about 2 a 3x 10 "Mo/yr, one needs—according to Grccnstcin. Univ. of Chicago Press. Chicago, p. 8XIT. Davidson and Ostriker—a value of s = vl,J2vl,(a) of Burgess.A.. Seaton, M.J. l%0. Monthly Notices Roy. Astroit. Sm. 0.8 for >i = a/R = 1.5 a 2.0, where a is the separation 121.471 t Conti, P. S. 1973. Astroph vs. J. 179. 181 of the two stars and vH. and i;esc are the outflow velocity Davidson,K., Ostriker.J.P. 1973, Astrophys. J. 179. 585 of the wind and the escape velocity from the Of star, Heap.S.R. 1971, Astron. & Astrophys. IS. 77 respectively. The applied 7-value follows from the system Hensberge,G., van den Heuvel,E.P.J.. Paes dc Barros.M.H. 1973. parameters (Wolff and Morrison, 1974). For £ = 0.8 and Astron. & Astrophys. 29 69 Hutchings,J. B. 1968, Monthly Notices Roy. Aslron. Sot: 141, 219 i\ = 2.0 we calculate a wind velocity vw(2Rt) = 420 km/s Hutchings.J.B. 1974, preprint at the height at which the X-ray component is moving Jones,C, Forman.W.. Tananbaum,H.. Schreier.I-.. (iursky.H.. through the Of star atmosphere. Kellogg, E., Giacconi.R. 1973, Astrophvs. .1.181. L 43 This velocity agrees quite well with the value at that Lamb.F.K., Pethick.C.J.. Pines.D. l973.'/l.srn>/i/i.i.v. J. 184. 271 Lucy, L. B. 1971. Astrophys. J. 163.95 layer derived from the velocity profile which we used in Lucy, L. B., Solomon, P. M. 1970, Astrophys. J. 159. 879 our calculations. Mihalas.D. 1972. NCAR Technical Note. Nat. Center for Atmos- The high expansion velocities deduced from the P-Cygni pheric Research. Boulder. No. NCAR-TN/STR-76 profiles of non-resonance lines in the visible spectral Morton.D.C. 1967, Asirophys. J. ISO. 535 region of HD 153919 make it highly desirable to Pottasch.S.R. 1970. in Interstellai Gas Dynamics, Int. Aslron. Union Symp. 39, Ed. H. J. Habing, D. Reidel Publ. Coinp.. investigate the behaviour of resonance lines in the Dordrecht, p. 272 ultraviolet spectral region. RottenbergJ.A. 1952, Monthly Notices Roy. Astron. Sw. 112. 125 Wolff.S.C. Morrison. N. D. 1974, Astrophys. J. 187. 69 Acknowledgements. I thank Dr. H. J. Lamers for many helpful discussions during the course of this study and for critically reading O. Hensberge the manuscript. I am indebted to Dr. E. P. J. van den Heuvel for Sterrenkundig Instituut obtaining the spectrograms of HD 153919 and for stimulating Roetersstraa! 15 discussions about the subject. Amsterdam-C, Nederland 36 A detailed Study of the Spectrum of the Binary X-ray Source HD 153919 (3U 1700-37). I. Radial Velocity Data in the Blue Spectral Region. G. Hammerschlag-Hensberge , C. de Loore and E.P.J. van den Heuvel1'2 ' Astronomical Institute, University of Amsterdam, Roetersstraat 15, 1004 Amsterdam, the Netherlands 2 Astrophysical Institute, Vrije Universiteit, Pleinlaan 2, B-1050 Brussel, Belgium running title: the orbit of HD 153919/3U 1700-37 publication in Astronomy and Astrophysics Supplement Series cased on observations collected at the European Southern Observatory, La Silla, Chile. 37 -2- Summary. Radial velocity measurements of selected lines in the spectrum of the Of star HD 153919 (3U 1700-37) are presented, together with the resulting radial velocity curves of this binary. Key words: X-ray binaries - Of stars - radial velocities 38 -3- This paper contains the basic data of a radial velocity study (Hammerschlag-Hensberge 1977; paper II) which uses 76 spectrograms taken in the blue spectral region (XX 3700 - 4900 8) with the 152 cm coude spectrographic telescope of the European Southern Observatory. All spectrograms are exposures on baked IlaO plates. The observations were performed during four observing runs between April 1973 and May 1976 : the plates of April 1973 and May 1974 were obtained by E. van den Heuvel, those of May 1975 by G. Hammerschlag-Hensberge and those of May 1976 by C. de Loore. All plates were measured twice by G.H.-H. for spectral line positions with the Grant comparator of the Kapteyn Astronomical Laboratory of the University of Groningen : once in direct and once in reversed direction. Each measurement is the average of two settings of the line profiles. The spectral line positions have been converted into wavelengths using a cubic dispersion relation based on the position of 25 Fe-arc comparison lines. The accuracy of the position determination of these Fe comparison lines is typically 1 JJ - For the stellar lines we used the recent list with laboratory wavelengths compiled by Conti et al. (1977) for O-type stars. The plates were measured in direct and reversed direction because the lines in the spectrum of this Of star are shallow and extremely broad. Therefore, a systematic velocity shift can occur between the measurements of the two directions, in placing the direct and time-reversed profile onto each other. One does not expect such a shift for the much narrower comparison lines. Indeed, also the sharp interstellar Ca-lines do not show any measurable systematic shift between the measurements of the two directions : 0.07 ± 0.12 km/s (la). The mean velocity of all stellar lines, however, shows a systematic shift between measurements in the direct and reversed direction of 3.21 ± 0.74 km/s (mean over all plates). The helium line velocities tend to give a smaller difference than the hydrogen lines which is in agreement with the fact that the hydrogen lines are more diffuse and more difficult to measure. Then, we adopted the mean of the two measurements as the velocity of the line. 39 -4- Systematic differences can occur from plate to plate, e.g. due to the non-uniform illumination of the slit of the spectrograph by the stellar image. With the used projected slit width of 10 p one expects differences up to 4 p, which corresponds to 4 and 6 km/s for the 12 A/mm and 20 A/mm plates, respectively. Differences up to this order are found in the measured velocities of the interstellar Ca- lines. These differences from plate to plate are larger than the differences between the two measurements of one plate which give an idea of the expected error (the setting error of the Grant machine is about In). We corrected the velocities of all plates with an amount Av = v - v , where v = -14.65 ± 0.30 km/s is the mean velocity of the interstellar Ca K line for all plates. Before making this correction we checked first that the Ca K line has no circumstellar component with a radial velocity variation within the binary period. We used for this check the program ORBIT, written by R.J.Takens (for a description, see van Paradijs et al. 1977). We did not use the interstellar Ca H line for this purpose because its velocity may be influenced by the presence of He. Table 1 lists the heliocentric radial velocities of the individual spectral lines corrected for these systematic differences. The plates have reciprocal dispersions of 12 A/mm (G-platenumbers in Table 1) and 20 S/mm (F-piatenumbers in Table 1). No systematic differences could be found for the 20 A/mm plates with respect to the 12 S/mm plates. So, they were all included in the orbital solution. The phases in the table are calculated according to the following ephemeris : phase = (JD - 2442231.13)/3.4111 (see paper II). Figs, la-ln show the radial velocity variations of each measured spectral line. The velocities are plotted against phase, calculated according to the above mentioned ephemeris. The dashed line in the figures corresponds to the orbital solution for the mean velocities of all lines together. For a discussion we refer to paper II. Figc. 2a and 2b show the radial velocity variations for the mean of the absorption lines of He II and the mean of H - Hg, respectively. Again, we plotted the orbital solution for the mean of all lines as a dashed curve in the figures. 40 -5- References Conti, P.S., Leep, E.M., Lorre, J.J.: 1977, Astrophys.J. 2NU, 759 Hammerschlag-Hensberge, G.: 1977, Astron.Astrophys. (paper II) van Paradijs, J., Zuiderwijk, E.J., Takens, R. J., Hanunerschlag- Hensberge, G., van den Heuvel, E.P.J., de Loore, C. : 1977, Astron. Astrophys. Suppl. Series (in press) 41 -6- Table 1. Heliocentric Radial Velocities for Individual Spectral Lines. 42 -7- i co '\j j ifi n (^ ^ c\i'U > J 3 a n ,0 >U *1 J ") J1 .D O \ j •« iO u~\ —t *-t I 1 « 1 •• • • * • il • I t * * I I I I I I •* I I I I -< -• I • * I I I I I I I I I f I I I < 111)111111 • -P. (M ^» - IA n »H • 1*111*11 I I t • I I • <"» I I t I * t+l *-4 SO O"* I J(P£3 ^ r\j O^ T* J" vO \O -r rs, in w o> 1 f t 1 1 < 1 j O> o (s. -o N. ^ r 1 1 1 1 1 1 I I I I I I I I • I I I til 11 1 1 1 1 1 t 1 a in —1 . j m -z> c=> .-- „-« r\j u U O i\/ i -4 n-l -H * 1 r 1 III I I I 1 1 I I I I I I t I I I I I I I I N- M H -O W 1^ -H v£3 X- L". f\l *y* ^* C3 CT^ 3^ <»D *X* ^T^ *"^ 1 1 1 1 1 1 1 1 1 *-( I I ^ | I I 1 | I -1 1 t i 1 1 1 1 1 1 1 1 1 1 -1 1 » 1 II II •T «0 CJ N. I t I I I I I I I I I I I I I I 1 I I I < •H I I I I I I I I ' I I 1 S CT is. rO .p TO \ 5 * J I ^ I I <1 I I t I I -H ( I I I I ~4 I I I I -( I I I I I I I I J X cr> p.i ^ fvj > r») »-• e^ -j f\ N. O •H I I <-* » I I I I • I I I I I I I < I t t . f- -H -1 LT> J- -T « 3D o t I I <~4 I I I I I I I I -H I t I *••' in 1 'A o f t r » r> ** r-» "r> -1 -» o • I "j f J ^ r -.1 'J •- r:j oe 3> ^ in ip in IT. i in in ir- in f 'O i) is N N N- O .D '5 J O '3 O 'J O 43 -8- .» o in a* t\j 9> *4 CO lTt j a I\J o to a > N. N. -* *-• i « —* « i " • I I I • (till It f 1 «O»O»H.r) OCT !T*(Pfs.f** O tf\ CD C\J 'NJ «*> a < I I t <0 en in .9 7 • O O H H I ^ • I H H H T^«4 H I I I I I I I I I t I I I I I I I I I I C0 I I I » I I I t I I I I I t t I H •*> in of* TH C* T1 1 7?" ' J-f\J N 3 f*l CO -» in CM«S i r *-< I i i • i i l 1 1 1 i ir\ a in 5 - 6 3 1 - 7 tn ^ oo m 9 7 1 - 5 * M 09 - IO * 0" vO 1^ J» \D -8 6 f • t 1 1 I 1 1 -3 5 1 • • 1 1 • • l 1 ^* 1 -iQ d 1 •108 . 1 —» K * in O f3 O O O •» J « y Q^MO^NfOj >«oo9 a WJ "J M M « J O «? <3 ' -9- Caprions of the Figures. Fig. la. Radial velocity variation of H.„, plotted against phase for a period of 3.4111 days. Phase zero corresponds to JD 2442231.13. The dashed curve drawn through the points represents the best-fit orbital solution for the mean velocities of all lines together. Fig. lb. The same as Fig. la, for H. . Fig. lc. The same as Fig. la, for H,Q. Fig. Id. The same as Fig. la, for Hg. Fig. le. The same as Fig. la, for Ho. o Fig. If. The same as Fig. la, for Hg. Fig. lg. The same as Fig. ia, for H . Fig. lh. The same as Fig. la, for H_. p Fig. li. The same as Fig. la, for He II X 3923. Fig. lj. The same as Fig. la, for He II X 4200. Fig. Ik. The same as Fig. la, for He II X 4542. Fig. 11. The same as ^ig. la, for He I X 4026. Fig. lm. The same as Fig. la, for He I X 4471. Fig. In. The same as Fig. la, for He II X 4686. Fig. 2a. Radial velocity variation of the absorption lines of He II. Each point represents the average radial velocity for one plate. The error bars denote the mean error of the radial velocity determination, according to an expression given in paper II. The dashed curve drawn through the points represents the best-fit solution for the mean of all lines together. Fig. 2b. The same as Fig. 2a, for the Balmer lines H.„ - Ho. 1Z o 45 o o o o o o o OOOO OOr-*C\jOO CM ro -3- LOCDf^C0 i i i I I I I I I I - X X / X 1 x X CM \ xx X \ X V x x X X \ ' X X x X - X N \ X x X X X\ X x k x X X \ X \ CO X X \ • 1 o 1 X X ( X xx / X X X / X X CD X X ' X o X X / >f X X X X *'' • x x y s xx w•* o / X X X I CM \ X X \ X o \ x X X \ X x N s X X X s Xx o X - \ X x X o X x x X X X \ X CM X X \ o 1 I 1 X X / III 1 1 1 1 Ill o o o o o o o o o o o o CM CO CM CO CO [^ 00 •—1 1 1 1 1 1 1 1 1 1 1 1 A1I3013A 46 o o o o o 00000000-*<\JCO-* OO NT LO CD C*"* GO Cfi *— «—< *—• «—• •—' I I I I I I I I I I I I X* / i X V X v V Nx x X XX \X x \ * x I x oo x i « I X x / X x x / CO X v / x / X X x' X CM x X • Vxx o X \ X XX • X o \x X \ % • o I _J I I o o o o o o o o o O o o o Csl oo CO ID CO en ^—1 1 I I 1 i 1 i 1 1 1 1 1 A1I3013A 47 O Q O o O 0000000*->CM CM on 1 1 i ^•lotor*-. coco«-««-«'-« «—* i i i i i i i i i x x X* N N \x x x X X \ X x \ * \x x x X X CO xx o x x X ID • x/. X x x • xx X' X / X X X? CM x \ o x* x * X N o x x xN • X o \ X x X V x x CM xx o I o o o o o o o o o o o CM CO •"* CO 00 en Cvl 1 1 1 1 1 1 1 1 1 48 A1I3013A o o o o OOOOOCDOOO — CM CMCO«*iocDi>oocn^-« 1 1 1 1 1 1 I 1 1 1 I.I 1 1 1 1 1 X X */' X X XV CsJ > • X X 1 x X X X X \ X N s X V * xx x - xx^-x X x \ x ^x x X X \ CO X \ X I o I X 1 X X o X • o • x \ x V X X x x \ CSJ X o *\ I . • • 1 X X o o o o c^ o o o o o •o CM *—• CM CO in CO o CO •—I ^-< 1 1 1 ,1 1 1 • 1 1 1 1 1 All 3013A 49 o o o o o o o o o o o cn cn r- CO 1 i 1 1 1 i i I I X x( xx X X CsJ x XN X \ X V X X s X \ X vX XX XX X \ x X X X X XX \ \ CO X x\ • \ X 1 A m X X *ix X X x x CJO / • K X/ 4 x / X X / * * X X X X s X X X xl X CsJ X A X • X I * x\ ? o X X N • X NX^ XX xx o V o • X \X * X \ X X X XX \ CM X ) <\ • o 1 1 X 1 1 1 1 1 1 o o o o o o o o o o o o — CM co CM CO \n co r- oo a> »- I I i i i i i i i i i 50 A1IG013A o o o Q ^ o o o o o o o o CM CO CO ^* LO (D r- CO I I I I i I I T r -i 1 1—T—T 1 1 r X X X / 1 X N C\J x x\xx X X \ \ X X A s s V xx s #xx X x\ \ XX X X \ x X X \ X X x \ CO. X X \ • I X| X X XXX I*t x x CO X ' X X Of r\£ / XX X / > • s X X ^^ xx ^' X /x X x" x ' 1 X x V C\J x v • X 5 X 1 f I 1 ( 1 I.I 1 f ' 1 f 1 f • i • i i i i o o o o o o o o o o o o o CM CO CO uv CD 1—1 «—< ,—1 «—t 1 1 1 1 . 1 1 1 1 1 1 1 1 A1I3Q13A 51 o o o o o o o O o o o CO CO to CD CoO CO f—1 .—1 n—( 1 1 1 1 1 1 1 I . 1 1 t I X / XX X f x \ (Si X \xx X X X \ X x s X ^Kxxx X - x X X \ X X XNx X X X x \ CO X \ x \ o 1 x X X x ** x CO X x /jicx x y X / jot. X X X X » X X ** o X / X X / XX X 1 C\J x x X X o X If •>< X X N X ^.xxx • - - X X * \ X X o X \ X X X Sc X x x ^ X CsJ \ X - o 1 I 1 1 x X • III o o o o o o o o o o o O *-* CM CO m CO 03 1 1 1 1 I 52 A1I3013A O o o o o o o o o O O CM CD CO CM •sf CD *—i CoM CM CM CM I I CO .r-< ^ 1 1 1 1 1 i -i r -T 1I 1 Ir I T—i 1 1 1 1 1——i 1 r r x k x CM X r x X " \ x X x \x x \x \ X X X \ \ X CO o X X 'x *^, X X I CD x y x o X /< CM o \ x x V o x X x x- o CM o I _J I I I o o o o O O o o O o CoM o CO 00 o CM CO CD CO • *—1 »—« CN! CM CM CM 1 1 1 1 1 i 1 1 1 1 1 1 A1I3013A 53 o O r i c •> c •> o O o o o CD a) t ) CO CO ••—« -—> C\J 1 1 1 1 1 ! 1 1 i 1 X XI X X |X \x X x X * xx\ • x \ \ x \* x X • x \ x XX *\ X X X • \ \ CO 1 o 1 X X 1 X / X / X X CD * / X /f o X x X / X X / X X x/ / / X x x / X / X 1 X X/ X X r CNJ X X X V o x x xx\ X \ <\* x X o - x\x XX o x^ X xy X X • CsJ X | o 1 I 1 1 X X 1 - f 1 1 1 1 II 1 1 1 1 1 1 1 1 f 1 1 o o o o o o O o O O o CM •«* CD CO o o 54 A1I3013A o o o O CM CD 00 o o CM -«fr CD CO *=f CM O 1 1 1 I 1 till 1 1 XX/X X X (NJ - X X xfxX x \ X x xxx\ x - X x \ X X ' x\ x X ~\Xx** X X \ 03 X M * 4 1 X 1t X X ^ X X 1 x f , CD X X x/ 3 X m O X x x %"? x / X * / x / X /*xx* X f • / x xx'Xx x X CSJ X X x o \ x x X * \ X x xxx\ x • x \ x \ x X X o X XxX xx ^ CM • - o M I 1 X - / X - 1 1 1 1 1 { • 1 1 1 1 1 1 1 1 1 t I 1 I 1 ^ 1 o o o o o o • o o o o X X X ( X - X* x X x \ X X V X x. X N N X *^\ • X \ y \ X ** X \ X \x xx X x x GO x \ I ! * 1 x X X x /x* • X X / * X X X X / x - ' X< X * X j^ • x X x- • / X X X X ** CSJ X* < x \x X o N X y X • X X *S x x< o - X it X • X o \ X \ x xx X x \f (NJ \ x o \ I '1 X / x . 1 ' f Illltlllll- o o o o o o -« csi O I i I 56 A1I3013A o o o o o o OOOOOO^CMCO"* LO CD C^ CO CO •—' •*-• •—' •—• *-" LO i i 1 1 1 1 1 1 1 1 1 1- X X / X X IX ! x\ X X \ ;- v X X x t V N x X X X t N 1 N X X ^ X X XX \ X i x x i X t X x 1 X \ x x X [ CO r I o i 1 X X r X x jx x x ' X UJ 1 X v 3< CD • CO X o CE X x A \ * X - X*. X y X y < s i X A CE 4 o r X y V X X X / X X \x x\ Cvl X X x\ o X \ x X ft N X X X N X X XX o > x X : x . - X \ o X \ X \ *x x X \ x CM \ X \ X O \ I ! x - / x 1 1 1 1 o o o o o o o o O O o o o »—« CNJ CO •«* LO CD CO CO ,—1 *-• »—4 LO I I 1 1 1 1 1 1 1 1 A1I0013A 57 o o o o i_) o O O Q C-J CD CO o CM CM ^t CD CO .—1 —I .—i CM CM ! 1 1 1 ! 1 1 1 1 i 1 1 '•' 1——I 1 1 1 1 r 1 [ 1 .i 1— 1 1 —r—i— I I X C\J * \ X \ x x \ xxx o x X \x x \ x \ \ X \ X X CO o v X X / CD « X ' xx x ' / x / x x X o XXX CM x o X ** \ X X \ xxx o X \x x \ x o < \ X X CM • o I I I I I t I I I O O o o o CoD CoO o CoM CoM o CD CO o »—t T—i CoM CM 1 1 1 I 1 i i 1 1 1 1 AiJ3013A 53 o o o o o o o c —' CM CO CO r- co in CO C\ r- O t 1 1 1 r X / x XX I X ; \ X X X \*x X* X *x x x \x x X - x\ N < \ x X \ XX \ CO X X * 1 1 X X X x x / *x* f X X / XX X r / x • ^x X y y y X T5" - X xx xx o y > t \ / X X i X xx 1 X X CM x \ x*x • X X \ NX * xx x\ x x X x o J 1 X X / 1 p I t 1 t f * * 1 1 1 1 1 1 1 I 1 I 1 1 o o o o o o o O o o O C\J ro CO r- CO CO CM •—I o 1 1 i A1I3013A 59 O O o o o o o o O O C\J oo O CM TJ- CD CO CM Cvl O I I i ! I I —i 1 1 1 1 1 1 1 r— T CO o CO -A O 7 (fa a (NJ O o V o =&&i (Si 0 dp I o I I I •_ I I I I I | I I | I I I I o o o o o o O o o o CM •<* CD CO o O CM CO CO »—1 •—1 «—1 CM CM O 1 1 1 1 1 1 1 1 1 A1I3013A O en O o o o o o o o o CM CO CM CO I I IX) CD I I I i Ii X—e— © -©—* \ -e— N \ ©—Q \ n - • -n X ^^ — CO ^~ • * VJ ...i ...... i —La -~- / n CD © • e— ©^- 9 n p Q—?*- - ', n CM n ® e iix— g o V| Q -©— • N o O ^ 1 o n V CM • - o I 1 / O 1 1 1 1 1 1 1 1 1 1 1 1 o o o o o o o o o o o CM CO LO en o o .-• CM CO I I I I I I I CD „_ ,_, A1I3013A I I I 61 A Detailed Study of the Spectrum of the Binary X-ray Source HD 153919 (3U 1700-37). II. iAnalysis of the Radial Velocities in the Blue Spectral Region.x G. Hamraerschlag-Hensberge Astronomical Institute, University of Amsterdam, Roetersstraat 15, 1004 Amsterdam, the Netherlands running title: the orbit of HD 153919/3U 1700-37 publication in Astronomy and Astrophysics Main Journal *based on observations collected at the European Southern Observatory, La Silla, Chile. 63 -2- Summary• The results of an analysis of radial velocities of 76 spectrograms in the blue spectral region are presented for the 06f star HD 153919, the optical counterpart of the X-ray source 3U 1700-37. From these new data, obtained over three years, we were able to derive a new value for the orbital period : P = 3?4111 ± 0^0002. The results for the orbital analysis of all lines together yield : e = 0.16 ± 0.08, K = 19.04 ± 1.27 km/s, u = 5° ± 22°. This yields a mass function f(M) = 0.0023 ± 0.0005 M . All quoted errors are la. The orbital elements were also determined for different ions separately. Apart from velocity variations due to the orbital motion, the velocities of some spectral lines show a strong extra phase dependence; this is especially clear for H6 and Hy. At phase 0.5 - and less clearly also at phase 0.0- the velocity of these lines is more negative than expected. A plausible reason for this observed phase dependent balmer progression is a longitudinal variation in the density and/or velocity profile of the strong stellar wind of the Of star. Part of the observed eccentricity may be spurious and due to this effect. No evidence is found for large changes in the stellar wind on time scales longer than the binary period. Key words : X-ray binaries - Of stars - radial velocities - orbital elements 64 1. Introduction HD 153919, an extreme Of star of 6.6 magnitude has been identified with the binary X-ray source 30 1700-37 (Jones et al., 1973; Jones and Liller, 1973; van den Heuvel, 1973). The optical spectrum has already been studied in some detail (Hensberge et al., 1973; Hutchings et al., 1973; Hutchings, 1974; Wolff and Morrison, 1974; Conti and Cowley, 1975). As it is generally known that an Of star has an extended atmosphere which disturbs the velocities of the spectral lines, we added to our plates obtained in April, 1973 (cf. Hensberge et al., 1973) new spectrograms obtained in 1974, 1975 and 1976 to get a better mean velocity curve. In total, we analysed 76 spectrograms in the blue spectral range, all obtained with the 152 cm coud£ spectrographic telescope of the European Southern Observatory. During the observing run of 1976, each night several plates of HD 153919 were taken to study possible correlated night-to-night variations, which may influence the radial velocity curve. The X-ray source 3U 1700-37 shows no regular X-ray pulses, unlike many other binary X-ray sources, but its intensity varies irregularly on time scales from minutes to hours (Jones et al., 1973; Mason et al., 1976). Possible reasons for the absence of regular pulses may be : (i) alignment of the rotation axis and the magnetic axis of the neutron star; (ii) coincidence of the rotation axis of the neutron star with the line of sight; (iii) the compact object is a black hole. The lack of an X-ray velocity curve in this system makes it impossible to give an accurate determination of the masses of both components. We can only give an accurate mass function for the system. Also an analysis of the light variations of the Of star do not help us to get a better idea of the mass ranges (cf. van Paradijs et al., 1977). In the next section we give some information about the reduction methods and accuracy estimates of the observations. In section 3 we describe the analysis of the orbital elements of the binary system and, finally, in section 4 we discuss some non-orbital effects which may have influenced the measured radial velocities. 65 -4- 2. Reduction of the Observations - Accuracy Estimates The radial velocity data of our 76 spectrograms are given by Hammerschlag-Hensberge et al. (1977; paper I) in tabular form. Paper I also contains a description of the measuring method together with the corrections applied for systematic velocity shifts between the different plates. No results of other authors were included in our analysis, as we wished to obtain a set of observations in which the systematic errors are as small as possible (use of same spectrograph, same measuring method, etc.). The spectral line positions were measured with the Grant comparator of the Kapteyn astronomical Laboratory of the University of Groningen. All plates were measured once in direct and once in reversed direction. All plates were corrected for systematic velocity shifts by assuming a constant velocity for the interstellar Ca K line (see paper I). The orbital solutions of radial velocities which were corrected for these systematic 2 errors have a reduced square sum (£(0-C) ) which is smaller than those for radial velocities without that correction, although both solutions are within la of each other. A few plates were measured twice in the same direction in order to obtain information on the accuracy of the individual measurements. The standard deviation o of the distribution of the differences 6 per plate between the two measurements of one line varies between 19.2 and 25.0 km/s. A typical value thus found is - 2 h 0 = ( 2a /N) =21.1 km/s, where N is the total number of plates which were measured twice in the same direction. This corresponds to a typical accuracy in the measurement of a single line position of W2 * a = 14.9 km/s. On the other hand, we can compare the two measurements of each plate in direct and reversed directions, after applying the correction for the systematic shift between the measurements in both directions (see paper I). The standard deviation a. of the distribution of the differences over ail plates between the two measurements of one line ranges from 11.8 to 28.2 km/s for the 66 different, lines (this value is smaller for the He II lines and larger for the lines with P-Cygni type profiles). The corresponding accuracy in the measurement of a single line position is given . • 2 h by h/2 x (Sa./n) = 13,2 km/a, where n is the number of stellar lines per plate. Another method to obtain information on the accuracy of the radial velocity determination from the measurement of a single line is to compare the result for a particular line with the average radial velocity of the plate as derived from all lines. The standard deviation of this distribution gives a good estimate of the accuracy of a single velocity measurement. For the different lines these standard deviations range from 11.5 to 22.3 km/s. These values are somewhat larger than those obtained in the previous estimate. In the latter case, not only the accuracy of a line measurement on one plate is involved but also variations from plate to plate are included. Therefore, we will use in our further analysis this last accuracy determination of the radial velocity of a single line. Accuracy estimates for different groups of lines are given in Table 1. For each spectrogram we determined the mean radial velocity for the higher Balmer lines (H12 - H , which we assumed to be more or less undisturbed by the progression due to the stellar wind), for the He II absorption lines and for all lines together. Before calculating the mean radial velocity of a group of stellar lines, we corrected each line for its deviation from the average velocity of all lines. This procedure is necessary because, due to the outstreaming atmosphere, each line has a velocity different from the stellar velocity. The line velocity depends on the place in the atmosphere where the line is formed : lines formed in deeper layers will have a less negative velocity than lines which are formed further outwards in the atmosphere. This procedure also eliminates possible systematic velocity shifts which could be introduced for some lines by the use of the laboratory wavelength list for O-stars (Conti et al., 1977). In this way the mean radial velocities are not affected by systematic effects, such as whether 67 ••(-•- or not a particular line has been included in the determination of the plate mean. The accuracy a of a mean radial velocity per plate was estimated from the expression : _ s + ps//n + p'Vn s where s is the standard deviation of the distribution of individual plate radial velocities and p is the mean accuracy of an individual 2 !< radial velocity measurement per line : p = (Ip./N) (see earlier this section), the summation extending over all lines in the group over which the average is taken. The values of p are given in Table 1. We checked whether the accuracy estimate p is affected by the fact that each stellar line has a velocity which deviates from the average stellar velocity (see preceeding paragraph). We calculated the values of p again after correcting each line velocity for its deviation from the average velocity of all lines. The values of p given in Table 1 changed less than 1% after applying this iterative procedure and, as a consequence, the values of a used in our analysis will be unaffected. The used expression for a gives a more realistic estimate of the accuracy of the mean radial velocity than just setting a equal to s : in case of a large number of lines the accuracy is determined by the standard deviation of the distribution of individual radial velocities, whereas for a small number of lines the errors in the individual line measurements dominate. In this way a reasonable bottom value p//n is introduced and we avoid possible unreasonable small errors in the case that only a small number of lines is available. Each average velocity was then weighted with a factor g = I/a . 3. The Orbital Elements For the determination of the orbital elements of each group of lines, we used the computer program "ORBIT" that searches for the best possible solutions from data with obsts national scatter and produces realistic error estimates. A description of the program is given by van Paradijs et al. (1977). The orbital elements, viz. 68 -7- the period P, the eccentricity e, the periastron position u and the radial velocity amplitude K, were derived for each set of lines and are given in Table 2 with la errors (68% confidence limit). As zero point of phase we used the mid X-ray eclipse time (JD 2442231.13) given by Mason et al. (1976). As turned out lateron, it followed from the orbital solution that this mid-eclipse time occurs 0.06 in phase earlier than the "real" mid X-ray occultation by the primary. The first part of the disappearance of the X-rays is probably due to absorption by the stellar wind, rather than to occultation by the primary star. This phase lag of real mid-occultation time is discussed in more detail by van Paradijs et al. (1977), who make an analysis of the light curve of HD 153919 which shows the same phase disagreement as our radial velocity analysis. The period derived from our observations is somewhat smaller than the ones determined earlier by other authors (e.g. Hutchings, 1974; van Genderen, 1977; Mason et al., 1976), but is in full agreement with the value obtained from photometric observations made during the same time interval (van Paradijs et al., 1977). Fig. 1 shows the orbital phase variation of the average velocity of all lines together with the best-fit solution. The variation of the average velocity of the He II lines and of the higher Balmer lines are shown in Figs. 2a,a of paper I. In that paper we have also plotted the variations of the velocities of the individual lines, and we drew through tnese points the velocity curve of the orbital solution for all lines. It is striking from Fig. 1 in this paper that around phase 0.75 all velocities lay below the average radial velocity curve. This nay correspond to the observation of extra absorption components in the P-Cygni profiles of several lines at the same phase (Hensberge, 1974; Conti and Cowley, 1975). From the orbital elements obtained from the average velocities of all lines, we derive a .sini = (8.81 ± 0.59) x 1010 cm (la opt error, single parameter confidence limit) and a mass function f(M) = 0.0023 + 0.0005 M0. As a radial velocity curve for the X-ray source is not available, a full determination of the masses for this system is not possible. Also a light curve analysis does not provide us with a unique mass ratio and inclination of the orbit (cf. van Paradijs et al., 1977)'. 69 -8- As the mass function for this system is very small, a high mass for the secondary will only be obtained for small va.'^_ ->f the inclination (even for M = 60 M and i = 60 , M is as small opt 0 x as 2.4 M ). Following evolutionary calculations which take mass loss into account for this star (de Loore, 1977) one would expect a mass for the Of star of 25 - 30 M . This corresponds roughly to a mass ratio of 0.05 (cf. van Paradijs et al., 1977) and the mass of the X-ray source is in that case smaller than 1.5 M . A recent study by Ostriker (1977) of X-ray "color-color" diagrams for galactic X-ray sources places 3U 1700-37 in the same region of this diagram as the X-ray binaries with known pulsating neutron stars as companion. This region is well separated from the region . to which Cyg X-l and Cir X-l belong, as well as from that of the low-mass X-ray binaries. This seems to suggest that 3U 1700-37 is most probably also a neutron star. In the next section we discuss possible non-orbital effects which may have influenced the orbital parameters. 4. Non-orbital Contributions to the Observed Radial Velocities a/_Effects_due_to_ stellar wind In early-type Of stars a strong stellar wind has been observed, M cf producing mass loss rates up to 10 - 10 Q/yr ( - Hutchings, 1976; tamers and Morton, 1977). Some (or even all) spectral lines are formed in the outstreaming atmosphere, which influences the measured radial velocity. The well known Balmer progression is a result of the same mechanism : the velocity of the Balmer lines becomes more negative towards the lower series members. For a spherically symmetric atmosphere no differences in the velocity stratification will exist in different directions. Houever, in the case that the wind parameters (density and velocity profile) change across the stellar surface -as may be the case for a deformed star in a binary system- the Balmer progression may vary with orbital phase. Therefore, we studied the Balmer progression as a function of the orbital phase. In Fig. 2 we have plotted the Balmer progression of the Of star for phases 0.25 and 0.50. 70 Furthermore, we have plotted for Hg, H., H and H. their average deviation from the mean velocity of ail lines, against phase. For an unvariable stellar wind one would expect this velocity deviation to be a constant. However, around phases 0.5 and 0.0, the velocity deviation becomes larger. Note that the velocity deviation for H is strongest at phase 0.5, while for Hg it is strongest around phase 0.0. For H. the two maxima in the curve have about the same amplitude. H. shows even more irregular variations : part P of this irregularity may be due to the fact that the P-Cygni profile of this line is also affected by an accretion wake around phase 0.75. The finding of this variable Balmer progression is supported by the optical study of Moffat and Dachs (1977) who found that the continuous Balmer emission increases at phases which correspond to our phases 0.54 and 0.03. To make an estimate of the possible effect on the orbital parameters, we corrected the velocities of H, and H for their deviation from a constant value of (v - v . ), according to the dashed lines in Fig. 3. It is also no ,Hy clear from Figs. .'., 1-g and 1-h of paper I that the velocities of H., H and H. around phase 0.5 are systematically too small compared to the mean velocity curve drawn through the points. The orbital elements for H. and H with and without this correction are given in Table 3. One observes clearly a decrease of the eccentricity of the orbit. Also the velocity amplitude K -and consequently the masses - are affected. At present it is not possible to correct all lines for this effect, due to the scatter in the observations of single lines (cf. Fig. 1 of Paper I). Of course, part of this scatter may result from the progression effect itself, and as long as we do not fully understand the physical background of the effect of the stellar wind on the stellar line parameters, it remains almost impossible to make the right corrections. As not all lines show this phase dependency, the eccentricity of the orbit determined from the velocities of all lines can be partly due to non-sphericity of the stellar wind. Also the values of the periastron position u in Table 2 which cluster around 0 , might be a result of velocities which are systematically too low around phase 0.5. If one assumes a Roche geometry for HD 153919 the stronger 71 -10- stellar wind may be a result of the reduced surface gravity at L. and L... In this way the mass loss process becomes in fact a mixture of Roche-lobe overflow and stellar wind. Such a phase- variable stellar wind has already been suggested to exist by Conti (1977) who proposed that both a stellar wind and an overflow of material from the Roche or tidal lobes is occuring. Table 4 shows the mean velocities per line for the three observing seasons with a good phase coverage. No variations in the stellar wind parameters on longer time scales are observed. These variations were suggested by Conti and Cowley (1975) to be present for H . b. During the 1976 observing run, mostly more than three plates were taken per night to look for possible correlated systematic variations, as observed for HD 77581, the optical counterpart of Vela X-l (van Paradijs et al., 1977). There is some evidence for this kind of variations, but the effect is not so clear as observed for HD 77581; this might be due to the larger intrinsic scatter. Especially the observations at JD 2442908 show a decrease of 30 km/s in the mean velocity during the night. We have looked at the velocity of the interstellar Ca lines on the same plates, but these show a random scatter. These observations indicate that in calculations of model atmospheres for this star dynamical effects have to be taken into account (such as the use of phase dependent parameters). A similar conclusion was reached from our photometric analysis of the light curve of this star (van Paradijs et al., 1977). 72 -11- Acknowledgments. I am indebted to Prof. E. van den Heuvel for stimulating discussions during the course of this investigation. I thank Drs. R.J. Takens for many helpful discussions about the use of his computer program and for critically reading the manuscript. 73 -12- Conti, P.S. ; ID77, preprint Conti, P.S., Covley, A.P. : 1975, Astrophys.J. 200, 133 Conti, P.S., Leep, E.M., Lorre, J.J. : 1977, Astrophys.J. 2\A_, 759 van Genderen, A.M. : 1977, Astron.Astrophys. 5_£, 683 Hammerschlag-Hensberge, G., de Loore, C., van den Heuvel, E.P.J. : 1977, Astron.Astrophys.Suppl.Series (paper I) Hensberge, G. : 1974, Astron.Astrophys. 36_, 295 Hensbfc.. 'e, G., van den Heuvel, E.P.J., Paes de Barros, M.H. : 1973, Astron.Astrophys. 29_, 69 van den Heuvel, E.P.J. : 1973, Int.Astron.Union Circ.Nr.2526 Hutchings, J.B. : 1974, Astrophys.J. 192, 677 Hutchings, J.B. : 1976, Astrophys.J. 203, 438 Hutchings, J.B., Thackeray, A.D., Webster, B.L., Andrews, P.J. : 1973, Monthly Notices Roy.Astron.Soc. 163, 13P Jones, C., Forman, W., Tananbaum, H., Schreier, E., Gursky, H., Kellogg, E., Giacconi, R. : 1973, Astrophys.J. 181, L43 Jones, C, Liller, W. : 1973, Astrophys.J. 184_, L65 Lamers, H.J.G.L.M., Morton, D.C. : 1977, Astrophys.J.Suppl.Series 3_4 de Loore, C. : 1977, private communication Mason, K.O., Branduardi, G., Sanford, P. : 1976, Astrophys.J. 203, L29 Moffat, A.F.J., Dachs, J. : 1977, Astron.Astrophys. j>8_, L5 Ostriker, J.P. : 1977, preprint van Paradijs, J.A., Hammerschlag-Hensberge, G., Zuiderwijk, E.J. : 1977, Astron.Astrophys.Suppl.Series, to be published van Paradijs, J.A., Zuiderwijk, E.J., Takens, R.J., Hammerschlag- Hensberge, G., van den Heuvel, E.P.J., de Loore, t : 1977, Astron.Astrophys.Suppl.Series, in press Wolff, S.C., Morrison, N.D. : 1974, Astrophys.J. 187, i 74 -13- Table 1. Adopted accuracy p of a radial velocity measurement of a single spectral line. spectral line P (tan/s) all lines 16.83 16.83 H12 " H8 He II abs 17.08 75 -14- Table 2. orbital parameters for HD 153919 (errors given are la; phase zero is mid X-ray eclipse time). all lines He II abs H12 H8 period (days) 3.4111+0.0002 3.4114+0.0006 3.4109+0.0003 eccentricity 0.16 +_ 0.08 0.22 + 0.17 0.25 +_ 0.10 K (km/s) 19.04 +_ 1.27 19.77 +_ 2.91 17.56 + 1.52 0) 5° + 22° 327° + 41° 350° + 19° periastron 0.27 + 0.06 0.16 +_ 0.14 0.26 + 0.07 phase Table 3. Orbital parameters for H. and H , with and without the correction for the variable stellar wind parameters. The period has been fixed on the value 3.4111 days. H H6 Y without with without with correction correction correction correction eccentricity 0.39 0.13 0.36 0.06 K (km/s) 26.41 19.18 21.89 19.59 (0 39° 43° 27° 342° 76 -15- Table 4. Mean velocities with la error bars of the different stellar lines of HD 15391') for different observing seasons. 1973 1974 1976 - 81.14 ± 4.18 - 86.08 ± 6.54 - 76.29 ± 3.93 H12 - 90.08 + 5.83 - 78.49 ± 5.15 - 82.11 + 3.74 Hll - 76.63 ± 6.54 - 69.38 + 6.45 - 70.41 ± 3.42 H10 H - 76.15 ± 4.75 - 72.66 ± 7.14 - 69.14 ± 2.94 9 t - 90.58 ± 5.10 - 80.95 ± 5.21 - 88.34 ± 2.55 H8 - 89.17 ± 5.25 - 73.21 ± 4.85 - 82.45 ± 3.20 ! H6 H -110.19 ± 4.47 -102.68 ± 6.71 -111.70 ± 2.63 ! Y He -149.67 ± 6.59 -150.06 + 9.07 -150.76 ± 4.57 He II 3923 - 80.08 ± 7.96 - 91.77 ± 7.59 - 77.88 ± 5.51 He II 4200 - 66.49 ± 3.97 - 69.84 + 8.42 - 57.35 ± 3.99 1 I He II 4542 - 62.16 ± 3.93 - 65.46 + 5.81 - 63.37 ± 2.92 He I 4026 - 97.26 ± 4.90 - 93.79 ± 5.77 - 92.63 ± 3.39 He I 4471 - 99.57 ± 5.44 -111.48 ± 11 .04 - 93.77 ± 4.72 He II 4686 17.55 + 4.06 19.04 ± 6.35 24.11 ± 2.82 77 -16- Captions of the Figures. Fig. 1. Variation of the average heliocentric radial velocity for all lines of HD 153919 with X-ray phase. Phase zero corresponds to mid X-ray eclipse time. Each point represents the average radial velocity for one plate. The error bars denote the mean errors calculated according to a formula given in the text. The curve drawn through the points represents the best-fit solution to the data. Fig. 2. Balmer progression for HD 153919. The black dots indicate for each line the radial velocity difference (averaged over all plates) from the mean value at phase 0.25 ± 0.10; the open circles denote the radial velocity difference at phase 0.50 ± 0.10. The error bars are la-confidence limits. Fig. 3. The radial velocity deviations from the mean for some of the Balmer lines, plotted against X-ray phase. The error bars are la-confidence limits. 78 o o o o «—1 oCM CO o o o o o o o o 1—) •** CO in CD CO *—• r—1 1 t 1 1 1 1 1 1 1 1 1 1 CD • O to o o o • o CSJ • o I I I I I I I I I o o O o o o o o o o o o CO CO -sf- LO CD r~ co a> «—1 I I I I i i i i I I A1I3013A 79 08 O OO en ro O O o o o 1 i 1 1 1 t 1. h-•-il o-—1 - S-O-l MM 00 - 1- - t - KH 1 1 o O O 00 CM s I 00 O JZ a. oo o Q. O O O O CM CM U3 81 Av.ron. Asiivplr.v 54, 54.V 54*. 11977) ASTRONOMY AND ASTROPHYSICS Four Colour Photometric Observations of the X-ray Binary HD 153919 (3U 1700-37) G. Hammerschlag-Hensberge and E. J. Zuiderwijk Astronomical Institute, University of Amsterdam. Roctersstraat 15, Amsterdam-C, The Netherlands and European Southern Observatory, La Silla, Chile Received May 24, revised July 28,1976 Summary. Siromgren uvby photometry of the O 6.5f star binary system we refer to Bahcall (1975)and to Hutchings HD 153919 is presented. The light variations arc- (1975) and references therein. consistent with ;i double wave variation in phase with In this paper we present extended uvby photometry the X-ray binary period of 3.412 days. The secondary of HD 153919. The observations are consistent with van minimum coincides with the X-ray eclipse. Our ob- Genderen's results. A marginal colour change around servations show evidence for a marginal colour change phase 0.2 is present. with binary phase. Key words: X-ray binaries — uvby photometry 2. Observations and Reductions The observations were made with a photometer for simultaneous measurements in the Slro.ngren uvby system with the Danish 50 cm telescope at the ESO. La Silla, Chile. The photometer is used in combination 1. Introduction with a photon counting system and is described in detail HD 153919. an O6.5f star, is the optical counterpart of by Gronbech et al. (1976). HD 153919 was observed the X-ray eclipsing binary system 3U 1700-37. The during the periods February 16-March 14 (by EJZ and X-ray luminosity varies with a period of 3.412 days and GH-H), May 27-June 8 (EJZ) and July 25-August 22, has a relatively long eclipse duration of 1.1 days (Jones 1975 (EJZ). Table 1 lists the 168 magnitude differences et al., 1973). Recently, observations with the Copernicus which were obtained for HD 153919 with respect to the satellite showed a shorter eclipse duration than measured 2 comparison stars HR6327 (SI) and HD 153767 (S2). by VHURU lasting only 0.88 days (Mason et al., 1976); [HR 6327 is an eclipsing variable (Bolton and Herbst, this indicates that the eclipse duration is variable, 1976); however, all but one of our observations were probably due to changes in the amount of absorbing done outside its eclipses and the star did not vary by material in the extended envelope of the Of star. Identi- more than 0™005 with respect to S2; peculiarly enough fication with the Of star HD 153919 was proposed by also the one observation that should have occurred Jones et al. (1973) and confirmed by optical observations during eclipse—according to the ephemeris of Bolton (van den Heuvel, 1973; Penny etal., 1973). Spectroscopic and Herbst—did not indicate a variation of the star by observations show that the star belongs to a binary more than 0T005, although, according to the ephemeris system with a period of 3.41 days (cf. van den Heuvel, it should have been 0T15 fainter at that time; this seems 1973; Hcnsberge et al, 1973; Hutchings et al, 1973; to indicate that the ephemeris is not fully correct. We Wolff and Morrison, 1974). UBV photometry of also do not find evidence for secondary eclipses.] Each HD 153919 shows a double wave variation in phase reduced tabulated observation is the result of the follow- with the binary period; the light curve has two minima ing series of integrations: SI—P—S2—P—S2—P-Sl. of about equal depth and of different shape (Penny et al, where P denotes HD 153919 and each individual stellar 1973). Observations by van Genderen (1973, 1976) in observation has an integration time of 20 s. The typical the Walraven five colour system show two unequal standard deviation of the measurements tabulated in minima—the secondary minimum occuring at X-ray Table 1 is 0T005. All observations of magnitude dif- eclipse time—and a pronounced asymmetry of the light ferences are given in the instrumental system, as it is curve. For a detailed review of the properties of the generally known that a transformation to the standard system is very inaccurate for highly reddened early type Send oj/prinl m/ut\\7.t w: G. Hammerschlag-riensberge stars [cf. E(B- K)=0.58 for HD 153919]. G. Hamrncrschlag-Ucnsbcrgeand \i. J. /uidcrwijk: Hiolomclry of HD liWHUI1 1.20 1.40 cog V -0.20 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 ** 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 "jfc* r««k '-0.20 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 PHflSE Fig. I. «. is, b and y variations of HD 153919, in the instrumental system of the Danish 50cm telescope. The differential magnitudes are plotted against binary phase (P = 3?412) 3. Discussion of the Results he did not find evidence for such a change. Our ob- servations are consistent with this result. The light curve Figure 1 shows the uvby light curves of HD 153919, has in all four colours an amplitude of about 0T07 at plotted against phase in the binary period of 3.412 days. primary minimum and of (T04 at secondary minimum Phase zero corresponds to JD 2442459.680+nx 3.412 (= X-ray eclipse time). The scatter in the light curve is as days ( = mid X-ray eclipse time), van Genderen (1976) large as 0T02 and must be intrinsic to the source. loo ked for a possible period change using all photometric Probably irregular mass streaming in the expanding observations of this star ranging from 1967 to 1975, but atmosphere of the Of star causes most of the scatter. In 84 i 11 ;!::.-.ii-.'::...:i:., -Hc;'siic.i;- .i;ij K I Zuiderwi.ik- Phoiomciry of HD 15"5919 13U i700-37) 545 Table J. Journal of observation ^Vl,r ^ '??: ~.'a'.\ ~.'l',l ~.'i*l 3.3 .ft 11 -.ill -..,;• -..,,_ r m .61 J -,Dta -,ui? ~.T. TL.7 .•>.:) -.0C -.CM -.70 -.DBS -.0 96 it .«,' -.Cb7 -.HO -.7U oJ7.S'»J -.D5S -.Oflfl -.TQJ : : :Sfl : ff -.70- .63b -. 71? . • .. b . 5 ',n :.;!" -:^ -:^' .:« .irt .085 .a-,-. -.053 -.0*2 -.E.95 '«! .'tfc .•>.'. -.051 -.J'lt. -.70 .-•. •>••- •^0^ . ,?i -.cr -.-"if -.b* .'*» .-. ;> -.0-.h -.3.- -.TiJ b-i.bi*. .biS -.CE.7 -,D'7 X X "x ^ O > '-0.20 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 '-0.20 0.00 O.tO 0.40 0.60 0.»0 1.00 1.20 1.40 PHflSE Fig. 2. Coluur indiees of HD 153919 againsl binary phase. Phase zero corresponds to X-ray eclipse time 85 C. H.iniinc.T-.chl.iy-lkn-ihcr.s. .mci I: .1. /uidcr.vijk: Pholomctry of HI) 15.1919(31 RXI-;.~i ;H.VOI'I!LH", .c v. iiii lt:c results ol'win GenJereii (I97X 1976). References we I'iiul ;in asuninviric light curve. The rising branch lluliciill.J.N : 1**75. I I'CIIII'CS .it the Vaieiuia School cm ihr I'IUMI . allcTllicpnnun mini muni is .sleeper than the descend ing Jllil Aslroplnsusol Nonlhill Stars.Hid lll.uk link's. V.ilh ll.ill.m.l branch precciiini; H. I'ublislitni! (Ump.inv. Anistcid.ini 1 *J7r> Ku.UiiU'.T.. llcrhM.W.-. l')7ft. Isdon. ./ HI. VW l-'igin" 2 shows the I'OIOUI index h — y and the colour (iinnheeh. I).. Olsen.l M. Slriimgre-n.B.: l')7fi. Annul, \-u-.ipli\- indices ni\ and el plottotl againsl hinary phase. There (in press) seems lo be an indicaiion k>r colour change around vjn Ucndcren.A.M.: 197.1. /«/. Bull. \ar. Shirs. No. S56 van Genderen.A.M.: 1476. preprint phase 0.2: especially the ml index shows the effect (see Hensbcrge.G.. van den Hcuvel.E.P.J.. I'aes de Burros. M.I I.: l'^\ Fig. 2). This effect might be due to a variation in the Annm. Aslmphys. 29. 69 emission lines He it 4686. C ill 4650 and N in 4634-4641 van den Heuvel.F. P.J.: 1973. I.A.U. Circular No. 2493 which lie in the center of the Stromgren Milter. How- Itutchings.J.B.: 1975. in Workshop Papers fora Sviiiposium .in X-iav Binaries. NASA preprint X-66O-75-2N5. F.ds. Y. Kondo and I Holdt. ever. Kr/emiiiski (1975) has observed the star with an p. 26? interference filler centered on the He n 4686 emission Httlchinos.J.B.. Thackeray. A. D- Websler. B. L... Andrews. ['.I 19' '•. •ine and found no variation within the orbital period. Monthly Notices Rny. Atfron. Soc. 163. UP F'.irther narrow band photometry is desirable to confirm Jones.C. Forman.W.. Tananbaum.H.. Sclircier.l-.. Ciuiskv.ll our resulis Kellogii.H. Giacconi.R : 1973. Astmplivs../. 181. 1.43 Kr^emiriski. W.: 1975. in Workshop Papers for a Symposium on A -r.;' Binaries. NASA preprint X-66O-75-2X5. Ms. Y. Kondo and I•'. Hold' \ckiun\ifdifi.im-ius. h. J. Zuidcrwijk acknowledges support by Ihe Mason.K.O.. Branduardi.ti.. Sanford. P.: 1976. Aslmphys. J 2011 29 NiMhcrlands Organisation tor the Advancement of Pure Research Penny.A.J.. Olowin.R. P.. Penfold.J.I:.. Warren.P.R.: 1973. \l,mtu!\ I/..W.O.I. We are indebted to Prof. Dr. If. P. J. van den Heuvel for Notices ROY. Asinm. Soc. 163. 7P Min.uktiint: discussions aboul the subject. Wolff.S.C.. Morrison.N.D.: 1974. Aslrophvs. J 187. 69 86 ASTRONOMY A-.lr.ui. Aslr.,|-,iux 5(.. 4J3- AND ASTROPHYSICS Ultraviolet Photometric Observations of the X-ray Binary HD 153919 (3 U1700-37) G. Hamtnerschlag-Hensberge' and C. C. Wu2 ' Astronomical Ins*' me, (Jnivcrsil> of Amsterdam. Roetersstraat 15. ^msicrdam-C. The Netherlands J K.iptcMi Astronomical Institute, Department of Space Research, University of Gromngen. P.O.Box 800. Groningen. The Netherlands Received October 27. 1976 Summary. The O6.5f star HD 153919, the optical counter- we give an estimate of the interstellar reddening and we part of the X-ray source 3U1700-37, was observed with discuss the ultraviolet flux distribution. The results are ihc 5 channel ultraviolet photometer on board the ANS. discussed and summarized in Section 5. The star shows a double wave light variation in all channels. The double wave variation is in phase with the optical variations in the visual spectral region, and 2. Observations is explained by the distorted shape of the Of star due The Groningen experiment consists of a 22 cm aperture to the presence of the X-ray companion. Significant Cassegrain telescope and a five channel ultraviolet colour variations are present below 2500 A. photometer with central wavelengths at 1550.1800.2200. At phase 0.8 an absorption dip is visible in the 2500 and 3300 A and bandwidth's of 150. 150. 200. 150 UV 'f.'ihi curves. This dip is confirmed by our recent and 100 A. respectively. The response functions are urhy observations which show similar dips at phases almost rectangular. The 1550-channel has the additional 0.4 and 0.8. capability that it can be narrowed to a width of 50 A The mean ultraviolet flux distribution is normal for centered at the Civ /. 1550 line, which is of special an O6 star. The derived reddening correction is interest for early type X-ray binaries (see Section 3c). The instrument is described in more detail by van Duinen etal. (1975). Keyword:,: X-ray binaries — photometry — Of stars The star was observed during March 9-10, 1975 and September 10-12, 1975. The observed count rates have been corrected for the instrumental sensitivity change (Aalders, 19*">). Table 1 lists the magnitudes in the 5 1. Introduction channels. Those magnitudes have been calibrated ac- The6T6 O6.5fstarHD 153919 is a member of the X-raj cording to the preliminary absolute calibration of the binary 3U 1700-37 (cf. Jones eta)., 1973; Jones and Liller, instrument given by Wesselius( 1975) and using the Vega y 2 1 1973). The optical star has been studied intensively both calibration of 3.64 10~ ergcm~ s~ 'A" for visual mag- spectroseopically and photometrically by several in- nitude K = 0T00 (Oke and Schild, 1970). Column 1 of vestigators. The photometric variations of the star follow Table 1 lists the times of mid-observation. The phases in a double wave. These variations are in phase with the Column 2 were calculated using a binary period of X-ra\ eclipsing light curve, with a period of 3.412 days, 3.4122 days and JD 2442231.13 as phase zero ( = mid lor a review of the properties of the system we refer to X-ray eclipse time) (cf. Mason et al.. 1976). Column 3 Hahcall (1975) and Hutchings (1975) and references gives the mode in which the star w; 87 434 G. Hamnu'r.schlaiz-Hensbcr^e and ('. C\ WH: IV Plum •metr> ol III) l.\«l'> Table 1. I :llra\ mlcl niiigulilinlcsolHD 153919 1 ,." 1 .11) HUM"' Ml Hit "'l5Mi,r '«|HHII I' ' 'mi 1 Wit i\ IvMli 2440000 I (s) (At 2480.64 0.12 normal 72 5.083 5.125 6.632 5.836 5.821 2481.24 0.30 normal 160 5.149 5.080 6.599 5.819 5.79.1 248I.X6 (1.48 normal 160 5.036 5.115 6.623 5.847 5.81 1 2666.27 0.53 normal 160 5.063 5.122 6.628 5.X45 5.824 2666.6(1 0.62 offsel 320 5.104 5.112 6.606 5.SI0 5.797 2666.68 0.65 normal 128 + 80 5.230 5.106 5.101 6.596 5 802 5.745 7 9>> i-iu.' 2666.81 0.6S olTsei 128 5.094 5.097 6.592 5.803 5.792 2666.94 0,72 offsiM 256 5.064 5.079 6.5K7 5.7K1 5.771 2667.01 0.74 offset 384 5.051 5.065 6.57K 5.780 5.758 2667.08 0.7ft normal 128 + 80 5.194 5.057 5.060 6.566 5.784 5.769 X.71 • 1)70 2667.15 0.78 normal 128 + 80 5.183 5.052 5.060 6.579 5.787 5.7X1 8.33 -1)62 2667.27 0.H2 normal 128 + 80 5.185 5.039 5.064 6.S79 5.7HX 5.777 9.21 .M1.S0 2667.42 0.86 normal 128 + 80 5.150 5.023 5.043 6.567 5.777 5.764 S.I 3 i 0.55 2667.56 0.90 offset 224 5.020 5.043 6.564 5.775 5.761) 2667.68 0.94 normal 128 + 80 5.167 5.036 5.063 6.595 5.807 5.784 8.40 + 0.72 2667.75 0.96 offset 448 5.046 5.OKI 6.606 5.X22 5.801 2667.S 1 0.98 normal 128 + 80 5.170 5.053 5.090 6.617 5.831 5.805 7.57 + 0.60 2667.88 II.IXI offset 192 5.060 5.106 6.631 5.8.V) 5.830 2667.89 0.00 offset 192 5.063 5.107 6.632 5.837 5.823 Phase zero corresponds to mid X-ray ellipse lime fcf. Mason et al.. 1976) " When narrow and wide bund al /. 1550 are observed, the first number indicates the lime of observation of the narrow hand. Ihe last one that of the wide hand. The olher channels are observed at both time intervals the narrow band. Observations were made with the (cf. Jones and Liller. 1973: Penny et al.. 1973: dc Krcita> other 4 channels during both these time intervals. Each Pacheco et al., 1974: Hammerschlag-Hensberge and listed observation is the mean of several 8 s (normal point- Zuiderwijk, 1976; van Genderen, 1976). The /H^^-light ing) or 16 s (offset pointing) integrations. The magnitudes curve ma\ be somewhat different from the light curves of the narrow band measurements were not used for of the other channels. This difference is based on onl> constructing the light curves as they are probably af- two data points. If real, this may be due to the presence fected b_\ the presence of the C iv line. of the C iv line (see Section 3c). hj Dip Around Phase 0.H 3. Results In all channels a dip is visible in the light cur\e at phase a) UV Light Curves 0.8. This dip has never been noticed before from optical To have a better impression of the character of the light observations. But an inspection of our recent urhv variations, we have plotted the observed magnitudes observations of this star (Hammerschlag-Hensberge and in Fig. I with their lir-crror bars. The errors are mean Zuiderwijk. 1976) shows that a similar dip al phases 0.4 errors calculated from the values obtained in different and 0.8 is present. Unfortunately, we do not have ultra- integrations of the source. As a check, these errors were violet observations at phase 0.4. For comparison, we compared with the errors expected from photon statistics. drew the mean light curve of the Stromgren (/-channel The\ arc all within the expected range. The light curves (centered at /. 3600 A) through the Hi,,,,,,-observations arc plotted against binary phase. The observations of in Figure I.Also the >-light curves of the other observers March 1975 arc marked with squares, whereas those of mentioned before do not exclude the presence of these September 1975 are markrd with filled circles. Un- dips. Van Genderen (1976) notes that the rising branch fortunately, there are only 31 bservations of the program following the secondary minimum is much sleeper than star in the first period, so that only the second half of the descending branch proceeding it. This can be easily the binary orbit is well represented. An object can be explained by the presence of the dips. However, the observed by the ANS when it is ±90° from the Sun in scatter in the optical light curves makes it difficult to Kcliptic longitude and with a duration of 1.8 days/cos fi, identify the dips. The setter apparently arises from the where jl is the Ecliptic latitude. This is why we could not fact that observations from different binary orbits have cover the whole orbital period of the 3U170O-37 system been plotted together. From ground based observations in a given observing season (once in every six months). it is impossible to observe the entire dip in one binar> Although only one half of the period is well represented, orbit (two consecutive nights differ 0.3 in phase), unless one can sec a double wave variation within the binary coordinated observations are done in different parts of period, which is in phase with earlier optical observations the world. Our ultraviolet observations of the dip, hew- 88 .ml C. C. Wu: UV Photometry of HD I53'M1> .00 - •4 phase- Kg. 2. Colour variations of HD 153919 with phase. Ic-error bars are given for each point tions of 1974 and 1975 (Mason et al.. 1976). The X-ray observations show also absorption dips at different phases of the orbit. Around phase 0.7-0.8 extra absorp- tion o mponents are present in the P-Cygni type line profiles of He i and Hji (cf. Hensberge, 1974; Comi and Cowley, 1975). An accretion wake following the X-ray source in its orbit has been suggested as a plausible explanation (e.g. Conti and Cowley. 1975). However. the origin of a similar dip at phase 0.4 is not so clear. .2 A , .6 .0 Eadie et al. (1975) have shown that a secondary shock phase at the side of the compact star opposite to the accretion Rg. . Utnmolel light curves of HD 153919. Observations of March axis can occur, so that material may accumulate and .ire marked with squares, observations of September 1975 with give rise to absorption also along this line of sight. On filled *circles, in-error bars are drawn for each point. The dashed curve in ihc lowest part of the figure represents the mean light curve observed the other hand, Wu (1975) suggested that the absorption in Ihc Strjimgren u-channel dips at about ±60 from mid-eclipse which are observed in many eclipsing binaries can be caused by the temporal) ever, come from one binary orbit. Inspection of our uvby- accumulation of matter at the triangular Lagrangian obscrvations made during February-August 1975 re- points (Z,4 and Ls) of the restricted three-body problem. veals that the dips were always present during that time. Although we presume that there is no continuous Roche The suggestion of extra light-absorbing material in the lobe overflow through Z., [the X-ray source would ven system is strengthened by the Copernicus X -ray observa- soon become extinguished in that case, cf. Pringle (19731 89 (i HamnKTSLhlag-l lcmhauc.mil II Wu: I V I'hi'toincU} of ll)> il) C iv A 1550 60 HD153919 As mentioned earlier, the almost simultaneous observa- . E(B-V)=.52 . E(B-V)=.58 tion of both the narrow and the wide bands at / 1550 A (difference in time is approximately 2.5 min in normal pointing) can be used to see whether spectral structure is present at the (' iv doublet. The observed count rates allow us to calculate the equivalent width of the carbon ho line by assuming: (i) only the C iv line is present in the 150 A inter\al: (ii) over this interval the copnnuum has a linear slope; (iii) the C iv line falls completely within the narrow 065 band. These conditions are probably fulfilled for this Oh star; cf. the discussion of the ultraviolet spectra of early- 2000 2500 3000 type stars by Morton et al. (1972). The calculated equiv - alcnt widths are given in the last column of Table I. l-ig. 3. I lirniolc: ilitx Jistnbiiiion for HD I5MI9 (circles). The full As Civ /. 1550 is expected to be a P-C'ygni-type line line represents the llux disiribtilion of an Ofi.5V Mar For comparison profile in this mass-losing star, the measurements gi\e t]ir !lu\ (li>[ribuuon ot /, Ccp (Oficfl is also yiven b\ plus signs in the only an estimate of the strength of the absorption com- lijulrr: ponent with respect to the emission component. The measurements do not indicate changes in equivalent width. Unfortunately, we do not have measurements and vim den Hcuvcl (1475)]. there may be occasionally around phase 0.5 where one would expect a change in the some mass concentration at llie Lagrangian points; this line strength due to ioni/ation of the stellar wind by could give rise to the extra scatter observed in the light the X-rav source (McCrav ami Hatchetl. 19751. minima at phases 0.0 and 0.5: for instance, the primary minimum observed by Penny et al. (1973) is deeper than thai observed b> van Gendcren (1976): the primary 4. Reddening and Flux Distribution minimum of our ultraviolet light curves is deeper than Since the interstellar extinction is strongly peaked at thai of our why observations (see Fig. 1). 2200 A. the observed spectrum will show a dip at that However, our ultraviolet light curves are based on wavelength if the object is reddened. To obtain the only one cycle, and consequently these curves are not dereddened spectrum, we adopted the extinction law affected by cycle to evele variations, contrary to the A,jE{B- n = K.I2. 7.X9. 9.65. 7.28 and 5.13 for /. =- 1550. urhy observations. This may explain at least part of the 1800, 2200. 2500 and 3300 A. respectively. This extinc- difference in depth. tion law is deduced from observations by the ANS and from the OAO II data folded with the ANS response function (van Duinen el al.. 1976). We compared the [ ' C nlaitl'S dereddened flux distribution of HD 153919 for different In Figure 2 we have plotted the ultraviolet colours of values of E[B-V) with those of / Cep (O6ef: HP 153919 against binary phase. Colour variation is E{B-V) = 0m5(i) and HD 4X099 IO6.5V: E{B-\) elearl;. present below 25(X)A. Colours in the visual = 01*27). From Figure 3 one can see that, adopting a spcclral region do not show variation. This is in agree- reddening correction E\B~ I ) = O"'52. the flux distribu- ment with what one expects from the flux distribution of tion of 1ID 153919 is in good agreement with those of an O-siar. The ultraviolet colours are bluest during the other stars of the same spectral type. The value of \-VUN eclipse and redder near phase 0.6. This agrees E(B-t/) = 0™58 derived from ground-based observa- with recent model calculations in the visual spectral tions (Bessell et al.. 1974) is clearly too high, if one as- region lot a tidally distorted rotating star (Zuiderwijk sumes that the star has no ultraviolet flux excess. Further- el al.. 1976: van Paradijs et al.. 1976): gravity and tem- more, a plot of log F, shows a hump at 2200 A for perature are lowest when we are looking at the point- E(B- i/) = O".158. whereas for £(B- I• ) = 01'52 the curve like shape of the star around phase 0.5-0.6, gravity and is linear between 1800 A and 2500 A as expected lor temperature are highest at the opposite side of the star. early-type stars. Earlier observations have already These observations rule out the model proposed by van proved that the early-type stars in X-ray binaries have Genderen (1976) with an X-ra> heated atmosphere, a normal ultraviolet flux distribution for their spectral where one would expect the bluest colour around phase types (Nandy et al.. 1975; Kondo et al.. 1976: Wu et at. 0.5. 1976). 90 l.iu JIIICIMIII.II! ll.nshcryc- IIIKI c. C. Wu: UV Photomclrv of HD 153919 •117 S. Concluding Remarks Bcssell,M.S., Peterson. B. A., Wickramasinghc.D.T.. Vidal.N. V.. ,>J74. Asirophys. J. 187. 355 Ultraviolet light observations of HD 153919 have re- de Boer.K.S.. Koornneef.J.: 1975, R.O.G. No. 75-63 vealed a dip at phase 0.8 of the binary orbit. Observa- Conti.P.S., Cowle; A.P.: 1975. Astrophys. J. 200, 133 tions in the Slromgren uvby system show similar dips van Duinen,R.J., A.'IdersJ.W.G.. Wesselius.P.R.. Wildeman.KJ.. Wu.C.C, Luinge.W.. Snel.D.: 1975, Aaron. Asiropliys. 39. 159 at phases 0.4 and 0.8. These dips may indicate mass van Duinen.R.J.. Wu.C.C., Kester.D.: 1976. R.O.G. No. 76-4 iiceumulation due to shock fronts in front of and behind Eadie.G., Peacock, A., Pounds, K.A.. Watson. M., Jackson.J.C Hunt. the X-ray source in its orbit around the Of star. At R.: 1975. Monthly Notices Roy. Aslwn. Soc. 172. 35P phase 0.6-0.8 additional absorption components are deFreitas Pachcco,J.A.,Steiner.J.E.,Quasl,G.R.: 1914. Aaron. Astro- found in the spectral lines of the Of star. Until now, phys. 33, 131 van Genderen.A.M.: 1976, Asiron. Astrophys. (preprint) nobody has mentioned the additional dip at phase 0.4. Hammerschlag-Hensberge.G., Zuiderwijk.E.J.: 1976. Antrim. Astro- ft would be interesting to study in more detail the line phys. (in press) profiles and the polarization at this particular phase. Hensbergc.G.: 1974, Astron. Asirophys. 36. 295 Simultaneous X-ray, photometric and spectroscopic van den Heuvel.E.P.J.: 1975. Asirophys. J. 198. LI09 observations are highly desirable to improve our under- Hutchings.J.B.: 1976, in X-Ray Binaries. Symposium held at Goddurd Space Flight Center, Greenbelt, Maryland, NASA SP-389. p. 531 standing of this binary system. Jones.C, Forman.W., Tananbaum.H.. Sdireier.E.. Gursky.H.. Kel- One should be very cautious in using the amplitude logg,E., Giacconi.R.: 1973, Astrophys. J. 181, L43 and shape of the light curves in an analysis to estimate Jones,C. Lillcr, W.: 1973, Astrophys. J. 184, L65 the mass of this system. Kondo.Y., Parsons,S.B.. Henize.K.G., Wray.J.D.. Benedict. CS. I".: 1976, in X-Ray Binaries, Symposium held at Goddard Space Flight Center, Greenbelt. MarylanJ, NASA SP-3S9. p. 551 .•U-kmmhilfiemenis. The ANS project was sponsored by the Dutch Mason,K.O., Branduardi.G.. Sanford, P.: 1976, Asirophys. J. 203, L29 Committee for Geophysics and Space Research of the Royal Nether- McCray.R., Hatchett,S.: 1975. Astrophyn. J. 199, 1% lands Acadcim of -Sciences. We acknowledge the cooperation of the Morton, D.C, Jenkins, E. B., Macy, W. W.: 1972, Astmphys. J. 177. 235 uiher members of the ANS-UV team. Drs. J. W. G. Aalders, K. S. de Nandy.K.M.D., Napier.D., Thompson.G.l.: 1975. Monthly Noiiivs Boer. R. J. van Duinen. D. Kester and P. R. Wesselius. Roy. Astron. Soc. 171, 259 We arc indebted to Prof. E. P J. van den Heuvel for his proposal to Oke.J.B., Schild.R.E.: 1970. Astrophys. J. 161, 1015 make UV observations of X-ray Binaries with the ANS and for his van Paradijs.J., Zuiderwijk.E.J.. Takens.RJ.: 1976. in preparation continuous interest during the course of this investigation. Penny.A.J., Olowin.R. P.. PenfoId.J.E.. Warren.P.R.: 1973. Monthly Notices Roy. Astron. SIK. 163. 7P Pringle.J.E.: 1973, Nature Plus. Sci. 243.90 Wesselius,P.R,: 1975, R.O.G. No. 75-1 References Wu.C.C.: 1975. Asirophys. Space Sci. 36. 407 Wu,C.C, van Duinen.R.J.. Hammcrschlag-Hensberge.G.: 1976. in A.iiders.J.W.G.: 1976. R.O.G. No. 76-2 X-Ray Binaries, Symposium held at Goddard Space Flight Center. Bahcail.J.N.: 1975. lecture notes from the Enrico Fermi Summer- Greenbelt, Maryland. NASA SP-389, p. i2-/ school on the Physics and Astrophysics of Neutron Stars and Black Zuiderwijk.E.J., Hammerschlag-Hensberge.G.. van ParadijsJ.. Ster- Holes. Varaina. llalv ken.C, Hensberge.H.: 1977, Astron. Astrophys. (in prcssl 91 Study of the Lightcurve of the Of star HD 153919 J.A. van Paradijs, G. Hammerschlag-Hensberge and E.J. Zuiderwijk Astronomical Institute, University of Amsterdam, Roetersstraat 15, 1004 Amsterdam, the Netherlands running title: lightcurve of HD 153919 accepted for publication in Astronomy and Astrophysics Supplement Series 93 -2- Suromary. Photoelectric five-color observations are presented for the 06f star HD 153919, optical counterpart of the X-ray binary 3U 1700-37. Combining these data with those obtained by other observers, we derived an orbital period of P -- 3.41117 + 0.00041 days. The light curve is variable : remarkable differences occur between the average results of different observers. The scatter of the individual observations around the mean light curve amounts to about 0.02 mag., which is much larger than expected from observational scatter alone. No definite color-index variations are found. The observed mean light curve of HD 153919 cannot be reproduced by light variations predicted from a model of a tidally distorted corotating primary. Possible reasons for this discrepancy are given. Key words: Of stars - photometry - X-ray binaries 94 1. Introduction HD 153919 is the brightest known early-type supergiant in an X-ray binary. The identification of this 6.6 mag. 06.5f star with the X-ray source was accomplished through positional coincidence and brightness and radial velocity variations in phase with the X-ray light curve (Jones et al., 1973; Penny et al., 1973; van den Heuvel, 1973; Hensberge et al., 1973; Hutchings et al., 1973; Wolff and Morrison, 1974). The system shows X-ray eclipses with a period of 3.412 days; the eclipse duration is large and variable (Jones et al., 1973; Mason et al., 1976). Outside eclipse the X-ray source is highly variable on time scales of minutes to hours. Optical light variations have been observed by Penny et al. (1973), de Freitas Pacheco et al. (1974), Hammerschlag-Hensberge and Zuiderwijk (1976), van Genderen (1976) and van Genderen and Uiterwaal (1976). These observational studies agree in giving the same overall features of the average light curve, in which double-wave ellipsoidal brightness variations are present. The minimum at phase 0.5 is deeper than the one at phase 0.0 (i.e. X-ray eclipse), which can be understood from gravity darkening together with the fact that heating of the primary by X-rays is negligible. However, the details of the light curves differ from one study to the other. This is, most probably, due to rather large erratic variations, superimposed on the average light curve. No definite color variations have been found for HD 153919. In order to obtain more information on the variation of the light curve we carried out new and extensive photometric observations of HD 153919 in the Walraven five color system. In this paper the results of these observations are presented. 2. Observations and reductions The observations were made by one of us (J.v.P.) in the VBLUW photometric system with the five channel photometer attached to the 90 cm light collector of the Leiden Southern Station near Hartebees- poortdam, South Africa. A description of the instrument and the photometric system has been given by Rijf et al. (1969) and by Lub 95 -4- and Pel (1977). HD 153919 was observed during 29 nights between August 19 and September 22, 1976, relative to the comparison star HD 153767 (V = 7?4; sp. AO). This star was also used as a comparison star by van Genderen (1976) and by Hanunerschlag-Hensberge and Zuiderwijk (1976). The non-variability of this star was tested in the .latter study. A single observation of HD 153919 consisted of 2 integrations of HD 153919 itself, one of the sky background and one of HD 153767. One integration lasted for 32 seconds. In the reduction the average value of the results for the comparison star was used, as obtained before and after the integration of HD 153919. The sky data were interpolated in a smoothed graph of the time variation of all sky readings. As both the program star and the comparison star are bright objects the treatment of the sky is not of critical importance to the results. 3. Results In table 1 the observed differences AV, A(V-B), A(B-L), A(B-U) and A(U-W) are listed, in the sense HD 153919 minus HD 153767 (in units of log(intensity), as is customary in the Walraven system). The data for the V band are displayed in fig. 1, plotted versus phase. The zero point of the photometric data was determined from a comparison of HD 153767 with standard stars of the Walraven system. The transformation to the UBV system of Johnson and Morgan was made using an expression given by Pel (1976). Also data from other observers were plotted in this figure; each set of observations is indicated by a different symbol. Due to possible small zero point errors the different data sets had to be shifted relative to each other by small amounts; the shifts were obtained by requiring that each group of observations had the same mean observed V-magnitude. The phase was calculated using the ephemeris derived below. Taking the scatter of the individual observations into account the light curves by different observers mutually agree well, with one exception : the minimum in the light curve near X-ray eclipse observed by Penny et al. (1973) is systematically deeper than in the other light curves by 0.015 mag.; i.e. this minimum is about 96 -5- twice as deep as normal. Such variations in the minima of the light curve have also been observed for HD 77581, the optical counterpart of Vela X-l (Zuiderwijk et al., 1977). From all available photometric data we derived a new value for the orbital period, by using the computer program "ORBIT", written by R.J. Takens, in the "period"-mode. This program has been described by van Paradijs et al. (1977). The value derived for the period is P = 3.41117 10.00041 days. This period is slightly shorter than obtained in earlier determinations (==3.412) , but fully agrees with the value recently determined by Hammerschlag-Hensberge (1977) from the radial velocity variations, viz.: P = 3.4111 ± 0.0002 days. These spectroscopic observations cover the same observing period as the photometry used here, viz. from 1973 till 1976. The zero point in the phase calculations was fixed by the mid X-ray eclipse time as determined from Copernicus observations (Mason et al., 1976). This yields the following ephemeris : phase = (JD - 2442231.13) / 3.41117. In fig. 2 the average light curve of HD 153919 is shown. The amplitudes of the light variations are A 0.040 ± 0.004 mag. and A. _ = 0.019 ± 0.004 mag., where we used as definition A0 5 = v(0-56) " is(V(0.31)+V(0.81)) and A = V(0.06) - )s(V(0.31)+V(0.81)) (see next section). The phase variations of the mean color indices A(V-B), A(B-L), A(B-U) and A(U-W), as derived from the present observations are shown in fig. 3. No definite variations of the color indices are observed with the possible exception of A(B-L) and A(U-W), which show a little dip of 0.005 and 0.015 magnitude, respectively near phase 0.8. This may be related to the observed additional absorption of the He I P-Cygni profiles in the spectrum at this phase (see also section 4). The average value of the color indices A(B-L), A(B-U) and A(U-W) differs from the values given by van Genderen (1976) by about 0.01 magnitude. If this difference is real either HD 153919 or the comparison star shows a long-term variation in the Balmer jump. More observations are required to confirm this long-term behaviour. 97 -6- 4. Discussion It is clear from figures 1 and 2 that the light variations of HD 153919 have the ellipsoidal character expected for a tidally distorted rotating star in an X-ray binary system. However, the whole light cur^-^ appears to be shifted in phase with respect to 'r the X-ray curve. This is especially clear for the phases of the well-defined maxima and the deepest minimum. These occur later than expected by A 98 -7- is found just before the entry into eclipse (Giacconi, 1975; Bennett et al., 1976). In this picture the real duration of the X-ray eclipse, i.e. of the occultation of the X-ray source by the primary star, is given by twice the time interval between the moments of mid-eclipse (as obtained from the optical photometric and spectroscopic data) and the observed exit from X-ray eclipse. This yields an eclipse duration of 0.47 ± 0.11 days, corresponding to an eclipse angle of 50 ± 12 . We have attempted to analyse the observed light curve, using a model for a tidally distorted rotating early-type star (Zuiderwijk et al., 1977). In the model calculations we have assumed the following ranges of system parameters : (i) a ^sini runs from 7.6 x 10 cm to 1.0 x 10 cm. This is a 2a opt range which was obtained from new radial velocity data, based on 76 blue spectrograms, obtained by Hammerschlag-Hensberge (1977). (ii) For the absolute bolometric magnitude H. , we have adopted values between -9?0 and -10?5 (corresponding to L = 1.2 x 10J erg/s and 4.9 x 10 erg/s, respectively), based on the results given by Conti and Alschuler (1971), Hutchings (1976) and Conti (1977). We have furthermore limited our calculations to cases where the average effective temperature is between 32000 and 40000 K (cf. Conti, 1973). (iii) The X-ray luminosity of 3u 1700-37 is so small relative to the luminosity of the primary that we neglected heating effects by X-rays. (iv) We have allowed the mass of the primary to have values between 8 and 70 M . This roughly corresponds to values of the mass ratio q between 0.04 and 0.07. For the calculation of the light curve we further need the mass ratio q, the inclination i of the orbital plane and the dimensionless potential parameter J2, which is a measure of the degree of filling of the Roche lobe. A detailed description of our method used for the model calculations is given by Zuiderwijk et al. (1977). As q, fi and i together uniquely determine the eclipse duration, they are not allowed to vary independently, once the observed eclipse duration is fixed. We varied q and Cl independently, and then used the fixed inclination in the light curve calculation; in this way each set of 99 -8- q and fi determines the masses of both components ( asini = a .sini (1 + 1/q) ). opt It appears that over the entire range of system parameters it is impossible to produce a theoretical light curve which represents the observed variations. As. an illustration of the difficulty we refer to fig. 4, where we have plotted the two amplitudes of the V-light curve as a function of ft, at a fixed q = 0.05, for three values of the observed eclipse duration ( 6 = 38°, 50° and 62°). Clearly a value of S2 can be found., for which the observed large amplitude A_ _ is represented. However, for this value of 12 the smaller amplitude AQ . does not fit the observations. The other way around, values offt. can be found which fit A- Q, but then A. _ does not fit. This situation occurs for all values of q between 0.04 and 0.07. Variations of the luminosity do not alter this situation. Fig. 5 shows the discrepancy between the calculated light curves and the mean observed light curve. Variations of mass ratio, luminosity or eclipse duration do not alter the shape of the light curve. For eclipse durations > 80 no solutions are possible in the used range of mass ratio's. Possible reasons for this discrepancy are : (i) Due to the strong stellar wind the surfaces which are visible in the continuum may not be at rest. In this case the shape of the star may be rather different from a Roche equipotential surface and the use of equipotential surfaces starts losing its meaning, (ii) Phase dependent variations in the stellar wind may affect the shape of the star : extension of the visible layers above the "resting" subphotospheric layers may be different in different directions. (iii) Deviations from the used LTE plane parallel model atmospheres (Kurucz et al., 1972) may not be the same in all directions. Phase dependent errors may occur due to the use of these plane parallel atmospheres and due to the deviations from isotropy (LTE). ' Until better dynamical model atmospheres for early-type stars with • expanding, atmospheres become available, it seems very difficult to use . a'light curve analysis for obtaining constraints to the masses of the . components of this system. 100 -9- Acknowledgments J. van Paradijs thanks the Leiden Observatory for granting observing time on the 90 cm light collector. The hospitallity of the staff at Hartebeespoortdam is gratefully acknowledged. E.J. Zuiderwijk acknowledges support by the Netherlands Organisation for the Advancement of Pure Research (Z.W.O.). We are grateful to R.J. Takens for the use of his computer program and for many helpful discussions about it. We thank Prof. Dr. E.P.J. van den Heuvel for critically reading the manuscript. 101 -10- References Basko,M.M., Hatchett,S., McCray,E., Sunyaev,R.A. : 1977, Astrophys.J. 215, 276 Bennett,K., Bianami,G.F., Di Gesu,V., Fiordilino,E., Hermsen,W., Kanbach,G., Lichti,G.G., Mayer-Hasselwander,H.A., Molteni,D., Paizis,C, Paul,J.A., Soroka,F., Swanenburg,B.N., Taylor,B.G., Wills,R.D. : 1976, Astron.Astrophys. 51_, 475 Conti,P.S. : 1973, Astrophys.J. 179, 181 Conti,P.S. : 1977, preprint Contl,P.S., Alschuler,W.R. : 1971, Astrophys.J. 170, 325 Conti,P.S., Cowley,A.P. : 1975, Astrophys.J. 200, 133 de Freitas Pacheco,J.A., Steiner,J.E., Quast,G.R. : 1974, Astron. Astrophys. 33_, 131 van Genderen,A.M. : 1977, Astron.Astrophys. 54_, 683 van Genderen,A.M., Uiterwaal,G.M. : 1976, Astron.Astrophys. S2_, 139 Giacconi,R. : 1975, Proc. 7th Texas Conference on Relativistic Astrophysics (Boston, Mass.) Hammerschlag-Hensberge,G. : 1977, to be published in Astron.Astrophys. Hammerschlag-Hensberge,G., Zuiderwijk,E.J. : 1977, Astron.Astrophys. 5£, 543 Hensberge,G., van den Heuvel,E.P.J., Paes de Barros,M.H. : 1973, Astron.Astrophys. 29_, 69 van den Heuvel,E.P.J. : 1973, Int.Astron.Union Circ. Nr. 2526 Hutchings,J.B. : 1976, Astrophys.J. 203, 438 Hutchings,J.B., Thackeray,A.D., Webster,B.L., Andrews,P.J. : 1973, Monthly Notices Roy.Astron.Soc. 163, 13P Jones,C., Forman,W., Tananbaum,H., Schreier,E., Gursky,H., Kellogg,E., Giacconi,R. : 1973, Astrophys.J. 181, L43 Kurucz,R.L., Peytremann,E., Avrett,E.H. : 1972, Blanketed Model Atmospheres for Early-type Stars, Smithsonian Astrophysical Observatory, preprint Lub,J., Pel,J.W. : 1977, Astron.Astrophys. S4, 137 Mason,K.O., Branduardi,G., Sanford,P. : 1976, Astrophys.J. 203, L29 102 -11- Pel,J.W. : 1976, Astron.Astrophys.Suppl.Series 2A_, 413 Penny,A.J., 01owin,R.P., Penfold,J.E., Warren,P.R. : 1973, Monthly Notices Roy.Astron.Soc. 163, 7P Rijf,R., Tinbergen,J., Walraven,Th. : 1969, Bull.Astron.Inst. Neth. 2Q_, 279 Savonije,G.J. : 1977, preprint Watson,M.G., Griffiths,R.E. : 1977, Monthly Notices Roy.Astron. Soc. 17B_, 513 Wolff,S.C., Morrison,N.D. : 1974, Astrophys.J. 187, 69 Ziolkowski,J. : 1976, 8th Texas Conference on Relativistic Astrophysics (Boston, Mass.) Zuiderwijk,E.J., Hammerschlag-Hensberge,G., van Paradijs,J., Sterken,C, Hensberge,H. : 1977, Astron.Astrophys. 54_, 167 103 -12- 'fable 1. Photometric Observations of HD 153919, relative to HD 153767, in 10log units. Heading of the table : JD2440000+ AV A(V-B) A(B-L) A(B-U) A(U-W) 104 -13- h ^C L_\ if U^ |^B __ _* l_a m ••_ _l~ l*» \O ^ i_l ^ ^ U\ U t l_k ^^ f'i _r ^U ^C LJ^ U • ^S -J ^O ^7 ^J U^ ^0 W^ •' i j U\ * ^ _• ^y ^^ ft i o C3 O DC GCJCOOOG0O O O C3 O O O O O O C! O COOOOC3OOQO C3C3C3OC300C3C3O •* «* r»' i* (* t* r r r r r r r r r r •* r r r r r r >* / r i* r r •" • r r r r • < <* •* r r r r •* • i i i i i i i i i i i i i i i i • • i i i i i i i • i i i i i i i i i i i i i i i i ' • i i • i* i < i • • i i* i i i* i i i i i i i i i i i i t i i i i • • i i i i i • i i i • »H ^*-* ^*4 c 'j CJ3 ^S1 rji ^^^ ^^ ^O -T ^J^ ^^^ ^^ PO ^^J f^ 1^- v4 f^ ^y ^^ -*?* %0 ^^ ^y ^^^ ^^\ ^H C3 ^0 U * PO ^^ ^^ ^^ ^^ ^ ^^ ^^ ^^ r^ T^ C^ 'J" ^ • ^ t ^—^ r Cu "5 a* Q ' (\J M fvJ M 1 2 y^ O ^0 ^ LJ"J ^0 U ^O u* J" *5 ^f J -J" ^\ \^ ^* ^ > b • ^^ ^^ f * ^ ^i U^ 0^ 3 ^^j ^^ ^^} 0 ^^} CZV ^^^ ^^^ 4^9 ^^^ ^^? ^Z^ ^3 C3 C^ ^ZV C^ C^ ^^ ^^ C3 ^j f^ {^ ^^ ^^) ^^7 ^^s ^3 ^^> ^f C^^ C7 ^ ^ r •* •* r •* #* / •* •* •* r / s* •* •* r r •* r r •* / •* r •* r •* r •* * •* •* «* • I i I I ( t I I | f I t I | I I I I | » I I I I I I I I I ? I t I I I • I t V i£ C.i f\t If \£ Wi t£) CT- \£. tf^^~l'3 CU i^4 ^ CM 1T> LT) WiT^I^CP^vlL^^irt iJ^OOCff'Nirf v4 J N4>(TiC*l llliill ••tlliilll i * i t i i i i i i '^0 f^ ^0 t**' ^y* ^0 f^ ^^ f^ «f PO 0* c^^ ^0 b^ Cv ^f\ v4 ^^ ^0 ^D ^^ f 1 k s 1 ( i^/r^ Lu^f If^^- ^^ \r*ir* tf^•/• I ^T^ U"u^> Ui^ tun^ u^; Vto Ub '^ tMfV^ tfit wM^ I^T^ Ub - 1b^^ uu^^ LP tf*b * b^ ^/ 4^ tb ^^ ^^ u^ b • 4f b ^^ b*^ b^ ^D b ^^ ^ n f; « o n I«I M n » r_>_ wnwMhwmnw______. ._ ._n . fj W W M f) W P> W W f; f> M f) PO f? W M (O (f, ' a; CJ K T« in o n K o o o o ^. v^ooOt> ooe-><->O£j'i:-»e->c--o ouooooccoo onDoooooou c_noc_>ooc»rjr ^> r*f) •*> ^> ^> ^^ ^o •'O ^i r^ ^o f*y ^o ^o ^> *o wt fj ^^ ro f^ ^i ^> f*j ^*J fn T*J fo P'J ro ^' T*JI f^ **> ^i r*^ PO f f > ^> ^J f P/ ^J fn f> pi p 105 -14- j in S iO r,' K fl o M i[ - J-' -3" I*'- IT. J IP. nCNSMTNM U 000« 1 • I I I I 1 I I • I I I I I I I I I I r r r • i* r r i° •* • • • i • < l\' u- CT> «-. U O> M iD If- O r i* *» i » > i i § • i i i i i * i • i • i i i > i i i i i 11 i i i i * i i < ^y 'o ^^ Q \J' ([j ^3 ^^ (^ ^^ occocr* o^ «c c ex- * I I I I I I • 1 » I I I ( I • • • t I I 1 f f f I t t i i f i i » i >* i I i t i • • t i i t i tr, in i*c (r. t ru O O C3 O +4 O O OCJOC3C9CJOOOO 0000^0 a O' sir s Ij, y^ ^^ ^ * ^^ ^^ j t^ j» FO FO ^V tO tO to f*) tO tO tO C fO MTri If C fU »f. in u* ^ i^ £ u ooooopoooo iiii*iii i i iiiiiiiiit I I I I I I I I I • i * i • i • i i i * r •* • r r »* r r r •* fro PO fo F^ ^^ r^ ^^ fo f^ t^t t^^ f ^ P^ f^ ^*^ r r r r r * r r •* •* * r ** r * r r r1 rI I *t I I I I I I • t * i * i i i i t t i i «i i * i i I * i I I i i i i 11 i • i i i i i • • i i r i i* * * * C5O0O00QOW0 OOOOQOO *-» O O * • J ^_J _ £ A J ft • • xowwwtowwfo w wo w p>; fi f) to PI w r»"'rrji»>ro (M K) fi *ri - «: *J «-• * O' 106 -15- c o o mm j ooo 000000° I » I I I I I I I I I I I I I • I I I I I I I I I t I t I t I I I • t I t I I I • I • I I I I I I • • I I t • I O O rl ooo o o o o OO00C3OOCOO If* UMT. tT U> if- If***'- IM»> W W « W W K tt W K ( w«fWMKnfown p o a o p o o o p o p o o p o p o o p a • i i i i i i i • • i > ! r i* i* •* •* i i •* i* r r r r r r r •* r»" •* r •* r r r • i • i i • 11 i • l I t f I I f l l t • i i i • i • i • i i i i i i nnnnnnnionn o n us nnnnnnnnnn r» t n n n n n fj F< n n n 107 -16- 3 ooiT.f*-h-lf\0'04D &JCU&T*J\£>\f\ '»£i^)JJii.LP'f\O'i> *D%£f-*OlfStf>iD\OlTii£ iPUsUiif3iD,£)v£»vOU)ir» vt^vTo^^^^lT, *O *C*Da>r^ft-vI>vOi£>ir.jt) ****** if * t t » » » ri» » r«i i • • * > f < « i i i » i • i » • i • i i • i i i « i » kONiroKtu^ff Wmn K,(T ro U\O IA^ njicNo^iiiccric^ i^iC'-iCN^tfif-JJ *O«D«G-»-*O:»I;M«^ - c «r> cr* o 'CiOOffOOO^O OOOOOOOOOQ OOOO0C30OOO OC<^OOC)uaOlO v • u> •*'-"* u\ to u^ * U> « f\J U\ C3C=>Crc^OC3O^OO DQOOOOQOQ0 OOC9C»OaC3O^C3 QC3OC3C3^aC30C> aOCSOOOC^CJC i • • i i i • i • i • < 111 i t i i i i 11 i • < r •* t r i* i" r i* i i* r i* r >'>>>>'{>!'> <••••<<••« • i i 11 • 11 i » i • > > i i * * • r * §" * * * * * •* i* * i" * * * * * * * «* * < > i • i • i i • i i i i i • i 11 t • i * i • 11 i i i i i • • i 11 i • i i • • i i i i i i i i 0* o o o o < \ UN U\ *iJ f^ »^J OKO1 ' P'-* PO P? P7 P? P' P) P7 VO t*} PS f^ P} P> P9 9Q p*> P9 CU (V CM V^ C^ C^ C^ ^W CM CM CU ^U CM ^^ CM ffM »••*•••••• •••»••••»• •••*•••••• *•»•• >ooooo»ooo oooaoooeoS oSoooooaoa aooooooooo oooooooooo 108 -17- OOOOOOQ0OO OQbOOOpOOO -....-..'..I . . .. ,~ . ,~~. '. . . . ". ' • • I I t I .... 1 I 1114 111111 I I < I I I t I I I I I I I I I I I I • . I < I I > I I • i : ,• .• •* .*;; .* •; ,* .* r r .* •• • .*; r ;i ri. i:; i ,-• I r • • r • .- r r.-, ••. . r...... r V4 ^^ ^^ C3 ^* »^ T^ ^^ *^ tJ w> ^^ T' ^^ ^^ ^^ *~' »* ^- •*• ^^ • ^^ *^* ^"^ ^^ *^ ^^ **^ — — — ^— ;;;; ,•; .• ,• .*; ; .*.';.'; • • •: •* •• r.'.' *' * «* •' •* * •* •' •* «* •''" •' ''' •' •' •'' •' * •' j o ca o o o o d T* • ••••••••• •••••••»•• • • •••••••• •••••»* »*•• •**••••• oto u>o* wtf1 « HJ- •••• » vD *X> O ^O lO 10 ^ (O oopooooooo toooooooo o e fllltftllf » I » # 4 I I » » » I 4 I # I I I I I f * I f I I » I f t I - W j- CO CT* U3 P r f l •••••••••• ••••••»••• •*•••••••• ••••.a ^« ^ ^ P^rJW''2r'JrJ' of^(ri^ • ••••••••• **••••*•• i • i»111 »i t * iiiiiriiir 111 i r i r • r i* N- f-~ CT >D CD v % x S^nm«^£«£5 iOOo^SOOOKffl 4«OT4^NJHnO -f 4iOoMU«t<<)U\Oi^ C C'G a^aoaDCO"0 SSSSrSS^*^0^*3 ovo OOO O O O O O O 9>OO 0^ O Ov4 r •••••••• • ••*••• ••*••• «*J M M N M M ^ M M fj AiruMniAmmtni OOOC3C3CPOOOO' OOOOOOO O O t> Q OO < tort to n #>it n ro 109 -18- coo a 0 o o 000000000° ODOOODODeO OOCTOOOCaOOO t I I I • < • I III! • I 3 I I I I I I I t • I I I I I I I I • i s ! i i : ! e • • i i i i i I i i l i i i i i • i i i i i i • I I I • I I i (T*^c^crc5( OrlO 0«H€ 0000 on o a o c a i i • i i * i i I I I I I I • t I I t • I • I I I i •> i i i i t t -ro (p in f\j vt) fC 4 OS) HM ifi SO 4 If* C3 O cooCO'lJ'O'ooa Ct*MC[\O« c^ c^ en C 0000000000 0000000000 »vj fM (V- oa o Q0OOOOOC?C3O OGOGDOOOOO OOOOOOO^O^ OO^OOOOOOO OOOG0OOOO* • I I I t I I I I I I • I I I I • I I I I | | ) • I I I I I I I I I I I | t I • • • I I I I I I rl J-OQ4H W(PlT <£ u) 10 10 a) N in I I I I I I I I I | I I I | I I I I I I • I I I I I •« I | I | 1 1 1 1 • 1 1 f tfccroacff QOQDC3OOOOO « t I 4 t I I • ••III • I • I I I • I I I 1 1 r 1 1 1 1 t 1 I • I 1 1 • 1 • 1 f 1 to M CA cj iv (p fo 0*e3oe>oooooo* o*oDoooooe(h 1O r* ft t- N N M M M M M fj N N WNWWIW C7OOC7000O00 000 es.o 000 p)p>rf>nn)n 110 -19- 0000000000 OOOOOOOOpO OOOOCOOOOO OOOOOOOOOO OOOOOOO&O • ••••••••• r •• i •* •*•"•"/1* i* r r •* •* r r r •* r r / •* r •* / r r •" r •* r r r •* •* •* r r _- — _ _. — _. _ _ _..-_ _ — — _ _ - . ,-__-__ ,_. v£ jk 10 iiO if iC lA «C O U\ ^ IT' it' i • i •»• i i i i * r * * i* •* r • i* * * f * •* r * r * * i* r r •* r r r r * r r * r r * r r r r * ** WM 4T^U\N W40 w cr en ^^\ 40 ^^^ ^3 0^ ^^^ ^^^ ^/^ ^^j ^^^ ^^ ^^ ^j1^ i » i • i i i i • i iitiiiiiii i i t 11 i r11 i i»( i * i ((• * oooooooooo ooocro< 4 * J(P^tftf r*i to *>t nronnronro -T j'd'ir'U^iricto^^K r- i^. o *o ^ cr i OOOCIOOOOOO OQOOOOC9000 OOOOOOOOOD 11 • co tf^ i^ ^^ f * ^^ ^^ ^^ m o^ **4 ^i t*} ^H P^ ^J ro PO ^4 ^^ po r f r i" r r r r i" r »" i" • • • i" • f >i > » > f r •"•••' i % •••••••*••• •••••••••• UN ^O ^^^ ^D I^B y_ ^Q ^^\ ^O ^^} ^D *^* ^^* ^0 ^? ^n U^ ^^^ ^0 ^n ^f^ ^^ft ^^^ \^\ ^D ^^k ^^^ ^^ ^^i ^Q ^y ^O ^^7 ^O ^^ ^D ^D ^^^ ^O ^D ^O ^n ^^H ^D ^E t^ ^^? U^ ^O ^C * r r r »* * r * r * r •* •* * * * * * * * * * * * * * * •* r * r * * * * * • • • * »* • * • * r • r • r w w N o us T4 J o j-NenuvnoHNM tD^-iAi^tocn^of^kO tf\^D«aroK^- «WC4» J* jj •• pij u\ oo fsj kMorottiotco4-«*4 ^-oaf< f^ fO f^ P^ ^^ f^ f^ PO ^^ ^O ^O ^O P^ F** f^ PO PjO •O frt F*^ _ . . fj ^j CJ fj fj ^J p^O 0^ P*J fO PO I^J PO IO fi^ |tf) ^j ^> f^ ^J fO PQ PO ^3 fO ••? (O ^O ^J ^^ 111 f. (J> ,J J" O *4 *3 PO -T *C JO N OOOOOOOQOQ o o o o p o o p o * ^ ^^ ^^ ^^ ^^j ^^ O ^^ f^ ^^ i* r i* i" «* i* • •* % % / r i* r •* i* t* / f i i i i i i i * * i I I I I t I t I I • r r r r r •* r < > • i » i i i i i i • i i i i < • i i » » I i I I I • i i i i i • i i i 11111114*1 • i i i i i • i i i i i i i i i i i i i i t • i • » i i I i * i i i * • t .i t i t OD a aa>o Hono 1 «D(T& Wo* u CT> w MM j CMOC1 ^ 1 . if vr tf» if. * w VP in a- 4 4 IT -f 4 ^ 4 - IT- U> IT IP If IT. tO <£> WJ r*j ro to K<^ ro ro ro n rt « to n f*> «4«4HHNtM(sjnnn (MCOPJCyCJCJMCvJtNJCvJ coaooooooo ooooooooo 4 j J- J- JT ^ J" -t * oooooooaao o&oooaooo OOO OO OC90OO0OO O O O O O O O O Q O O O • ••••••••• *»"•••"•'»"••*•'# I I I I I I I I t I | I I I I I I | I I i I i i i i i i i i i i*) r r i" i* i* i* r IIIIIIIIII i i i < • i i i i i IIII'II'II • i i i • i i • • • • i i i i • • i • i i ' i i i • i i i • i 11 > i * i i IIIIIIIIII • i • i • i i i i i IIIIIIIIII OOOOO0 O O wrtf»rtww 112 -21- O0 ooOoooeiOoo I I I I » I I I I I I I I I I I I I I I • I I I I. •*» F»J "5 W ••» >•* f» »*> w> •"» Ill 111*111 * t I t I • » I I I CO ( I t l l I I I lllltllll* t l i | l I I t I I 1 I I I I t I I t » I I I t I * I ^lAcO^ON WOO 0009*0*00009 0^00009*9*00 O O IT O O Q C3 I0^ W< W H u\ m UMMA us t i o a o o < 113 -22- Captions of the Figures. Fig. 1. V-light curve of HD 153919. The zero-point shifts for the different observations are given below. The symbols denote observations by the different observers, viz. : < Penny et al. (1973), V = V,+ 0°002; © de Freitas Pacheco et al. u (1974), V = V, + 0?004; + van Genderen (1976), V = V: J J x Hammerschlag-Hensberge and Zuiderwijk (1976), V = Ay + 5.904; $ this study, V = AV + 6.193. Phase zero corresponds to JC 2442231.13 + n x 3?41117. Fig. 2. The average V-light curve of HD 153919. The points are average values in phase intervals of 0.1. The error bars denote the standard deviation per observed point. Fig. 3. Average colour indices in the Walraven system are plotted against phase. The vertical scale is in log units. The points are average-values in phase intervals of 0.1. The error bars denote the mean error of the distribution of the individual points. Fig. 4. Dependence of the light amplitude A on ft and eclipse duration for a fixed value of q = 0.05. The horizontal lines indicate the observed values of 0.0 A0.5- A. and A,, „ are shown for eclipse durations 8 = 50 ± 12°. 0. r5 0.0 e Aft = 0.0 denotes the value of ft for which the Roche lobe is filled. At the top of the figure the inclination scale corre- sponding to the curves for 8 =50 is shown. The two shaded areas show the error boxes in which the observed and calculated values of A 5 and A_ - agree. These parameter ranges do not overlap. Fig. 5. Light curve of HD 153919. Open circles denote the observed mean light curve. The full line is the calculated light curve •?hich fits the amplitude A for the following parameters : q. = 0.05, ft = 1.839 (Aft = 0.05), i = 62°, a = 2.1 x 10 cm, M . = 30.1 M . L = 1.5 x 1039 erg/s. The dashed curve fits opt © opt the amplitude AQ _ for the following parameters : q = 0.05, ft = 2.009 (Afi = 0.22), i = 68?5, a = 2.0 x 1012 cm, M _ = 39 opt 25.7 M, L = 1.5 x 10Ja erg/s. 114 r—i C\J CO LO CD CO •—• LoD LO LD LD LD LO LO LO LO LmO CoD CD • * • « • * a a a a • CD CO CO CO CO CO CO CO CO CO CO CD CNJ O o CO O 0 o a o CM * o I o CM CO ID to CD o. «—i LD LD LD LO LD LD LD LD LD CO LD a CO CO CO CO CO CO CO CO CO CO CO CO CO 3QniIN0UW 115 O CM cr> CD CD o in ID in in in in .5 5 .5 6 in in 03 ID to uo uo uo uo uo uo" uo -25- l l l T .100 .105 * J-L) ..090 -.095 _.100 A(B-U) _.36O -.365 -.370 A(U-W) -.050 — 055 I! ...060 - I J I •0.2 0.0 0.2 0.A 0.6 0.8 1.0 1.2 Phase Figure 3 117 60° 62° ° ' 68° 70( 72C .06 .05 .03 ,__ .02 .01 .0 A XI *2 .3 Figure 4 118 -2 1- in 3 o JC a. CN o > O o o p O p o o o o <3 1 o o 119 Aslron. & Atrophy*. 35, 353—360(1974) Orbit, Spectrum and Ha Variations of HD 77581 (3 U 0900-40) E. J. Zuiderwijk The Astronomical Institute at Utrecht E. P. J. van den Heuvel* luiropcan Southern Observatory, La Silla. Chile Astronomical Institute, University of Amsterdam Astrophysical Institute, Vrije Universitcit, Brussels G. Hensberge* Astronomical Institute, University of Amsterdam Received March 22, revised May 22, 1974 Summary. Radial velocities of HD 77581 (3 U 0900-40) features a distance of 2.2±0.4kpc is derived, which obtained from ESO coude spectra are presented. The yields Mv= -71M ±0?4. Arguments are presented in- revised orbital elements are P = 8?959 + 0?004, e = 0.22 dicating that supergiants in massive X-ray binaries are ±0.02 and /(M) = 0.0147 ±0.0011 MQ. unlikely to have suffered appreciable mass loss. The The profile of Ha behaves similar to that in Cygnus masses of the components of HD 77581 are estimated to X-i, i.e. consists of a wide emission component with an be ^ 30 ± 5 M0 and ^ 2.5 ± 0.3 MQ, respectively. approximately constant radial velocity and a narrower absorption component with a periodically variable velocity. Variations in the profiles of Hf and the He i Keywords: X-ray sources — spectroscopic binaries — lines correlate with the H, variations. From interstellar supergiants — spectrum variations 1. Introduction The 6T87 B 0.5 la star HD 77581 is the optical candidate cover the region between -U 4500 and 6800 A (dispersion for the 8?95 eclipsing X-ray source 3 U 0900-40 (Hiltner 19 A/mm). a aU 1972). The identification seems well established as Identifications and an analysis of the radial velocity HD 77581 shows variations in light and radial velocity variations based on a combination of our data and with about the same periodicity and phase as expected those of Hiltner et al. are given in Sections 2,3 and 4. from the X-ray variations (Vidal et al, 1973; Jones and A description of the variations of the profiles of H,, Liller, 1973). Hp and some He I lines is given in Section 5. Possible A rediscussion by Vidal et al. of the radial velocity explanations of these variations are briefly discussed in measurements by Hiltner et al. seems to indicate an Section 6. Classification, distance and masses of the com- orbit of considerable eccentricity. Nevertheless, the ponents are discussed in Sections 7 and 8. spectroscopic orbit is not well established since a gap of about 0.4 period is present in the observations by Hiltner et al. 2. Identification and Radial Velocities In order to obtain a better radial velocity coverage All spectra were measured twice (by E.J.Z.) with the over the entire period, and to examine whether the H, Grant comparator of the Kapteyn Astronomical Labor- profile might be variable in time—as is the case in many atory of the University of Groningen. The spectra were early B-type supergiants (cf. Rosendhal, 1973)—one of also traced with the digitized Faul-Coradi microphoto- us (E.v.d.H.) obtained 23 high-dispersion coude spectra meter/comparator of the Utrecht Observatory and— with the 152 cm telescope of the ESO, La Silla, Chile, after Fourier noise filtering—plotted in intensity. As between March 31 and April 10, 1973. Table 1 lists the spectrum closely resembles that of e On (B 0 la) we the spectra. Ten of the spectra were obtained in the blue used for identification the detailed linelist of this star (dispersion 12 A/mm) on IlaO emulsior. and cover the given by Lamers (1972). The following comments can wavelength region kX 3400-4900 A. The other 13 plates be added to the identifications. No Balmer lines are were taken in the yellow-red on 103aF emulsion and visible beyond HI6; Mgn4481 is weakly visible: the * On leave of absence from the Astronomical Institute at Utrecht. unidentified line at A 6290 is clearly variable and is 121 I-. .1. Zuidcrwijk a us \\J i'v vi * ii viilliJi iii»v;i.sa'il,ir bands are present near AA443'' 5782, 5797 determine the exact value of the correction to be applied. and 62X4 A. Using trial corrections between 25 and 30 km s we first rii'.- uidi.d velocities of the strongest hydrogen and determined the period by fitting the steep pints of the helium hues are listed in Table 1. These radial velocities preliminary radial velocity curves through ihc observa- were measured by fitting the direct and reverse line tions by Hiltner el nl. i'nd through our observation!.. profiles on the oscilloscope screen, below half depth. In We used periods around the X-ray period of X'.'95 ?. 0d02 particular the central parts below half depth ol'H,, H,, as given by Forman et al. (1973). It turned out that a and the higher Btilmer lir.es are practically symmetric correction of +2Skm/s to the radial velocities from 'frcm here on we will denote these lines as "the higher Hiltner et cil. together with a period of 8?96 ± 0?0l gives Biilmcr lines"). Except for H, and H^, the radial veloci- the best fit. ties of the Balmer lines on each spectrogram show a The best-fit radial velocity orbit through all the daui good internal agreement. The mean of the radial veloci- was then computed with the program "Orbit" (Wolfe !ies of H .. H,,, H8. Hl; and H ,„ arc indicated in the fifth et ui.. 1967). It yields the radial velocity curve indicated •j'urr.n of Table 1. (H,. was omitted because of blending in Fig. 1 a, which is characterized by the following v'-iih interstellar Oi II H.) The ladial velocity of H^ is orbital elements: >ivstem;uically smaller than those of H. to H,,, by about ) l l K)to 30 km/s, depending on the orbital phase (see below). / = K .'959±0'. 004. These differences are correlated with abnormalities in K = 26.0 + 0.7 km/s. the line profiles which will be considered in Section 5, K, = 3.6 + 0.6 km/s. together with the H, variations. The mean values of the radial velocities of the Hei t> = 0.223 ±0.024. lines on Ihe blue plates agree reasonably well (within 4 w= 13T4+7.8, to fc km/s) wiih those of the higher Balmer lines. How- f(M) = 0.0147 ±0.0011 M . ever, the standard deviations for the helium lines are 3 generally over twice as large as for the Balmer lines, The formal error for the period that is obtained in this which is probably due to asymmetries and variability way might be unrealistically small. in the helium line profiles (see Section 5). Dots in the Figure indicate Hiltner el al.'s observations. For this reason the mean radial velocity of the higher crosses indicate our observations. The figure shows Balmer lines is expected to give a more accurate indi- that between orbital phases 0.3 and 0.4 (taking pen- cation of the stellar radial velocity than the mean of the astron passage at .ID 2441771.08 as phase 0.0. sec belovvi hydrogen and helium velocities together. We therefore a systematic upward deviation of the radial velocities adopted the "stellar" radial velocity to be the mean exists with respect to the "mean" behaviour of the velocity of the higher Balmer lines. On the red plates radial velocity variations at the other phases. [A similar the only absorption lines suitable for measurement are deviation, however, in downward direction, is observed H(/ and three He I lines. Their velocities are listed in the near maximum radial velocity in the case of HD 153919 lower part of Table 1. The table shows that the velocity (Hensberge et al.. 1973).] As these deviations are too of He i 5875 is, similar to that of H,,, systematically large to be fitted with any Keplerian radial velocity smaller than the stellar radial velocity by several tens curve, it seems reasonable to assume that they are due of km/s. Therefore, and since the velocities of the re- to effects of gaseous streams in the system or to devia- maining two He i lines show much scatter, we have not tions in the atmospheric outflow pattern from the super- used the red plates for the determination of the orbit. giant. Therefore, in the determination of the radial The radial velocities of the interstellar lines of Ca » (K- velocity orbit we omitted the encircled points. (On our and H-line) and Na i (D,- and D2-line) are 28.8+ 1.4 only plate around this phase—G 3842—the higher and 31.0+ 1.9 km/s, respectively. Balmer lines are about 40"c> broader than normal. If all of this excess broadening were to be at the red side of the line center, this would produce a rcdshifi 3. The Radial Velocity Orbit of about 30 to 40 km/s: consequently, the broadening of the lines around this phase may easily account for The radial velocities (i.e. mean velocities of the higher the observed radial velocity deviations, which are less Balmer lines) of the ten blue spectrograms are well than about 20 km/s.) spaced over the approximately 9 days of the orbital period. The mean radial velocity from Hiltner et al.'s In Fig. 1b we plotted the H^ radial velocities as a (1972) observations differs from the mean of our obser- function of the phase. The internal error of one in- vations by about 28 km/s. Since we could not find any dividual velocity measurement is estimated to be about obvious reason for this systematic difference we are led 5 km/s. The dashed line in the figure represents the to presume that it is of instrumental origin (use of differ- radial velocity curve defined by the above given orbital ent spectrographs, etc.). In order to combine our ob- elements, shifted by about —9 km/s. As the standard 122 Orbit of HD 77581 (3 U 0900-40) Vrnc(km/sec) • runner ei ai. ; e-a223 ±0024 f zuiderwijk et al. i 1.0'f T0.JD 2W1771.078 (balmer lines) -| 00 0.1 02 OA US 06 0.7 CX9 09 10 0.1 orbital phase Fig. I. (it) Best-fit radial velocity curve obtained from our blue plates and the data of Hiltner et al. (1972). Orbital phase is taken with respect lo pcri:mtron. Predicted phase of mid X-ray eclipse is indicated, (b) Hp radial velocities from our blue and yellow-red plates. The dashed line is the radiai velocity curve from Fig. la, shifted by — 9 km/s. The full curve is shifted in phase with respect to the dashed one by 0.04 deviations in the velocity determinations of the higher 1973). Since Hf and these emission lines are formed at Balmer lines (see Table 1) and of Hfl are smaller than the larger distances from the stellar center than the higher applied shift, this radiai velocity difference may be a Balmer lines, the phase shift might be due to the same manifestation of the Balmer progression, indicating cause in both stars. that the atmosphere is flowing out (cf.Underhill, 1966). The observation of the H emission profile supports this a 4. Comparison with X-ray and Photometric Observations idea. The solid curve in the figure was obtained by applying a phase shift A epoch that the phase shift of 0.04 is similar to the phase shift of our observations the mid-eclipse time will be at observed for HD 153919 (3 U 1700-37) between the JD 2441778.02 ±0.17. radial velocities of the higher Balmer lines and those From the known values of u> and e we can calculate of the emission lines of He n and N m (Hensberge et ai. the expected difference between the moments of zero 123 •'><> E. J. Zuidcrwijk ri ul. i Hi-h vntrii-i-idial velocities. The blue plates (12 A/mm)arc listed in the upper pad ofihe table, the red plates 119 A mm) in the lower part I'lale JD+ 2441000 Orbital Mean of He t 5875 4471 4026 3820 6678 4388 4144 3%5 14 H H 14 number phase H10 ny, rij, rig, n<>. G 3X34 773.626 0.2X4 -54 - 16.6 ±5.0 0 - 4 0 -35 -24 - 26 G 3842 774.6(18 0.395 - 1.6 + 6.6 - 4 -17 - I - 9 - 15 2X i G 3848 775.582 0.503 -54 -17.6+3.3 16 -16 __ 22 - 8 - 9 G 3858 •"7(1.583 0.614 -35 - 14.2 ±4.6 - 8 T -13 _ 3 -34 . 9 G 3866 777.61? 0.729 -11 O.C ± G 3833 773.548 0.277 -50 -54 -14 -10 G 3841 774.547 0.388 -47 -48 -19 (13857 776.515 0.608 -34 -15 G3865 777.544 0.722 -13 -14 -28 -13 G 3869 778.532 0.832 - 1 16 -21 15 G 38";i 779.505 0.941 3 13 G 3880 779.567 0.948 10 38 G 3893 780.513 0.053 25 12 26 16 G 3894 780.554 0.058 27 10 24 13 G 3895 780.597 0.063 17 5 27 16 G 3908 781.576 0.172 7 - 1 17 9 G 3924 782.594 0.285 -60 -56 0 -17 (•' 1035 775.524 0.496 -41 -41 -17 -27 radial velocity and of mid X-ray eclipse. Taking lines show some excess blue absorption in the wings e = 0.22 and w = 13° we find this difference to be about (but most probably not in the central parts, which re- 0?32. Hence, from the spectroscopic orbit the mid main symmetric, except—as mentioned above—on plate X-ray eclipse is expected at JD 2441778.08. Within the G 3842). range of the observational uncertainties (±0?2) this value is in good agreement with the above mentioned predictions from the UHURU observations. H. Photometric and photoelectric observations of In Fig. 3 the observed profile of Ha is plotted as ;t HD 77581 clearly show asymmetries, as we expect from function of the true anomaly. The binary itself is the orbital eccentricity (cf. Jones and Liller, 1973; Vidal thought to be at rest, while the observer moves around et a/., 1973: Hutchings, 1974: Vidal, 1974). Recently, the system in the direction indicated by the arrows. Hutchings (1974) analysed photometric variations of The true anomaly, as a function of the orbital phase, was HD 77581. He found values of e and to close to ours. calculated from the orbital elements, using Kepler's The variations do not repeat themselves from cycle second law. Since the orbit of the binary system is to cycle, indicating that fluctuating gaseous streams may eccentric, the true anomaly is growing nonlinear with be of considerable importance. This is confirmed by our the orbital phase. So the observer's motion around the spectroscopic observations (cf. Section 3). system is not uniform. The profile of Ha consists general- ly of two emission peaks (indicated as a and c) se- parated by an absorption component (b). Table 2 lists 5. Line Profiles the peak radial velocities of these components. Between phases 0.27 and 0.50 the profile is of the P-Cygni type, H/( and He l Lines consisting of one emission component and a blue shifted Figure 2 shows profiles of Up and some of the Hei absorption component. The Ha profile observed by lines as a function of the phase. The lines are clearly Wickramasinghe et al. (1974) around phase 0.5 (one asymmetric between phases 0.25 and 0.6. The asymme- orbital period before our observations) is in good agree- tries in the af profile correlate with the observed radial ment with our profiles around that phase. It should be velocity deviations in the same phase interval, as men- noted that these authors report the presence of a second tioned in Section 3. In this phase interval the width at minor absorption component. the continuum of H^ and the He I lines is larger than Let us assume that the two emission peaks a and c normal with an excess absorption at the blue side. are both parts of the same broad emission line upon Between phases 0.2 and 0.7 also the higher Balmer which the absorption component b is superimposed. 124 Orbit of HD 775K1 (3 U 0900-40) .157 He I 6678 A He I 5876 A Plate Phase 3A 3A Number lii'. 2. Original tracings of Hei and Hj profiles as a function of the orbital phase (with respect to psriastron). Plate-numbers are given. In- tcnsiiv und wavelength scale are indicated. The plotted profiles of He 16678 and 5878 were taken from the red plates, the ones of H,, and He i 4471 fiom the blue plates. Missing entries are due to the weakness of some plates We derived the central radial velocity of this wide emis- component shifts back and forth with respect to an sion line by fitting its profile to the outer wings of underlying emission component (which, consequently, components a and c and found that this central radial also here appears to be double peaked most of the velocity is practically constant as a function of phase time). (within the observational uncertainties, which—for this broad line—amount to ±20km/s). On the other 6. The Outflow Pattern from the Supergiant hand, the radial velocity of the absorption component b shows large variations as a function of phase. Clearly the wide undisplaced emission component of This is remarkably similar to the behaviour of the H, Ha is produced rather far from the stellar surface in an profile in HDE 226868 (Cygnus X-l) as observed by extended envelope which is flowing radially outwards Brucato and Zappala (1974), where also the absorption in all directions, whereas the absorption component is 125 E. J. Zuiderwijk et al. Fig. 3. The H, profile as n function of the true anomaly (for i = 90"). Corresponding orbital phases are marked along the circle in the figure The observer moves as indicated b> the arrows. Intensities and wavelength scale are indicated at the tracing of plate G 3857. The profiles on pialo Ci 3X93. 3S94 (taken within less than 2 h before G 3895) are identical to the one of G 3895. The same holds for the profiles on plates C; 3878 and G 3X80. The profile of G 3833, taken exactly one period before G 3924, is identical to the one of G 3924. indicating that the profile variation is a periodic phenomenon. The dashed Roche lobes indicate, exaggerated, the variations in distance between the components. The observer's positions at periastron-, apastron- and X-ray mid-eclipse time are marked produced in the part of this envelope between the ob- pattern is expected to be asymmetric due to the server and the stellar surface (cf. Underhill, 1966). asymmetric gravitational field produced by the pre- Therefore, the absorption line velocity reflects the out- sence of the companion. In the case of HD 77581 one flow velocity of the envelope into the direction of the has, on top of this asymmetry a second and independent observer at the mean level where the H, absorption asymmetry introduced by the orbital eccentricity. In is formed. The understanding of the outflow pattern is Fig. 3 the large central binary is drawn to indicate the complicated by the fact that the orbit is eccentric. position angle of the observer with respect to the line Already in the case of a circular orbit, the outflow joining the components. The variation in distance be- 126 Orbit of HD 77581 1.1 U 0900-40) 35') Table 2 H, riKii.il \o! Plale number Orbital phase Red emission Absorption (b| Blue emission peak (a) peak (u) (i 3(i33 (j.2/7 115 -125 '13X41 O.jhs 110 -200 (j 3X57 0.61W 140 - 65 -210 (i 3S65 0.722 140 - 20 -125 0.94S ISO — ""5 - 1 10 Ci 3X93 0.053 90 - XO -200 Ci 3X94 0.05X 100 - 65 - 190 G 3X95 0.063 125 - 65 -200 Ci 390X 0.172 110 - 55 -175 G 3924 0.2X5 105 -100 1 1035 0.496 110 -210 iween the components during the orbital motion is in- surface where H, is formed. This "garden hose effect" dicated by the dashed (small) binaries which, exaggerat- then might explain why one observes the largest blue- ed, show the sizes of the Roche iobes as a function of shift of the absorption component within one quarter the position angle of the observer with respect to the period after the passage. Detailed model calculations system. Position angles at ap- and periastron passage for explaining the observed profile variations of H.,. H/( :ire indicated with the letters A and P. and the He I lines will be the subject of a separate in- The figure shows that the largest outflow velocities in vestigation. Ha occur between phases 0.25 and 0.50, i.e. almost im- mediately after primary light minimum, and last about a quarter of a period (plates G 3833, G3841, G3924, F 1035). The largest outflow velocities in H,, and in the 7. Classification, Distance and Mass wings of the Hei lines occur near these same phases The classification of HD 77581 as a B0.5 star is con- which seems to indicate that also the layers where these firmed from our spectra by comparing the observed lines are formed are rapidly expanding at these phases. spectral features with the criteria given by Walborn The figure seems to suggest that the outflow is not very (1971). From the equivalent width Wk = 0.54 ± 0.05 A strongly dependent on the size of the Roche lobe, as the of interstellar Ca II K one derives a distance of 2.2 Id, profiles before and after apastron passage are quite + 0.4 kpc according to the calibration by Miinch (1968) different, although the corresponding sizes of lobes are for a star in this direction in the galactic plane. Although about the same. Also the fact that the Ha variations in a distance determination from the strength of inter- Cygnus X-l show a similar phase dependence, although stellar lines might not be the most accurate, the derived [here c^O.l (Bolton, 1972) seems to suggest that the value is in excellent agreement with the distance obtain- orbital eccentricity does not have a large effect on the ed by Wickramasinghe et al. (1974) by comparing the outflow pattern. Apparently the position (and not the reddening of HD 77581 (E(B- K) = 0.76) with those of distance) of the X-ray component with respect to the stars in its vicinity. part of the stellar surface from which we observe the This distance determination confirms the luminosity outflow velocity is the main parameter on which this classification as a supergiant: the resulting value of outflow velocity depends. Mv = - 7T1 ± 0T4 corresponds indeed to a la supergiant. The simplest way to explain why the size of the Roche There is, however, a puzzling fact which does not fit this lobe does not have much influence is, that the radius of classification, viz. the fact that only 14 Balmer lines arc the supergiant is considerably smaller than the average separately visible on our plates whereas the expected radius of the lobe, and that the mass loss process is of number for a supergiant is around 21. A number of ihe stellar wind type (i.e. has nothing to do with the 14 visible Balmer lines would lead to a classification Roche lobe). Such an explanation is well in line with the B 0.5 III. Since neither the luminosity of the star nor its rather small eclipse fraction (t/P~ 0.2) as well as with Ho emission fit the B 0.5 III classification (cf. Rosen- theoretical models of the accretion in X-ray binaries dhal, 1973) we are inclined to suppose that due to some (cf. Davidson and Ostriker, 1973). The only effect of the other mechanism the highest Balmer lines have become presence of the companion seems to be to enhance the indistinguishable. A similar effect is observed in i: Ori outflow from the surface layers nearest to it. Due to the (B 0 la), where due to the intrinsic Doppler width (pro- orbital motion the region of largest outflow velocities duced by macroturbulence) the Balmer series terminates will lag behind the companion, far above the stellar around Hl6 (Lamers, 1974). 127 E. !. /uidcrw;jk i'i .-:.' ,17.. v. 30 A/,., ami, in consequence the mass of the X-ray source is likely to be larger than about 2.5 M ,. [In Using .-I:, ;i sale lower limit B.C. = 2m for a BOia star l: view of the uncertainties involved in the distance deter- (which almost certainly is UH> low, cf. Underhill. 1966) minations, bolometric corrections, etc. (cf. Section 7) we find .W,,,,! S -9"'I + 0"'4. According to Simpson's the uncertainty in this lower limit is about +0.3 M..,.] (1971) evolutionary tracks this corresponds to a mass Such a mass seems rather large for a neutron star and of S30 A/.. + 5 M . However, as argued by van den o might be more indicative for a black hole. lieuvd and Ostriker (1973) and Hensberge et at. 11 y?3) due to mass loss in a supergiant binary this value may be ihree times smaller without considerably affecting the luminosity. Although, admittedly, large References mass loss seems rather unlikely here (see next section) Bolton.C.T. 1972, Nature 240. 124 this means that the mass of HD 77581 might be as Brucato.R.J.. Zappala.R.R. 1974, Asirophys. J. Letters lin press! small as only 10 Mo ± 1.7 MQ. Davidson,K.. Oslnkcr.J. P. 1973, Asirophys. J. 179. 5X5 Foiman.W.. Jones.C. TananbaunvH.. Gursky.H.. Kellogg.h.. (iiac- coni.R. 1973. Astrophys. J. 182. L 103 Hensberge.G., van dun Heuvel.Fi.F'.J.. Paes de Barros.M.H. 1973. H. Tfie Mass of the X-ray Source, Discussion Asiron. & Astrophys. 29. 69 van den Heuvel,E. P.J.. De Lnore.C. 1973. Astron. & Asintphvs. 25. From the mass function J\M)- 0.0147 Af.3 and an 387 adopted value of the mass MB of HD 77581 one can van den Heuvcl.E.P.J.. Oslriker.J.P. 1973. Nature Phys: Sri. 245. w calculate a lower mass limit for the X-ray source (cf. Hiltner.W.A., Werner,J.. Osmer.P. 1972. Asirophys. J. 175. L. 19 Bollon, 1972: Trimble et a/., 1973). Taking A/fl = 30 Af0 Mulchings.J.B. 1974. Asirnphys. J. 188. 341 Jones.C, Liller.W. 1973. Asirophys. J. 184. L 121 we find Mx '2.2.5 Mo. If, however, MB is only 10 Mo Lamb, F. K.. Pelhick.C.J.. Pines. D. 1973, Asirophys. J. 184. 271 we find Mx § 1.2 Mo. The X-ray emitting component Lamers,H.J. 1972, Asiron. & Asirophys. Suppl. 7. 113 i;iay therefore be a neutron star as well as a black hole. Lamers. H.J. 1974. private communication Two arguments in favour of the black hole possibility Munch.G. 1968. in Stars and Stellar Systems. Vol. 7. [id. B. M. might bet Middlehurst and L. H. Aller. Univ. of Chicago Press. Chicago. |i) the absence of regular pulsations of the X-ray source: p. 365 Pringle.J.E. 1973. Nature Phys. Sci. 243. 90 (ii) the fact that a close binary containing a neutron Rosendhal.J.D. 1973. Asirophys. J. 186. 909 star or black hole will appear as an X-ray source only Shakura.N. I.. Sunyaev.R. A. 1973. Asiron. & Astrophrs. 24. 337 during the time that the supergiant component is loosing Simpson.E.E. 1971. Asirophys. J. 165. 295 mass at a modest rate in the form of a stellar wind Trimble, V., RoscW.K.. Weber.J. 1973. Monthly Nmim K»r (.V/< iO~6M /yr) (cf. Davidson and Ostriker, 1973: Astron. Soc. 162. 1 P 0 Underhill.A.B. 1966. The Early Type Stars. Reidel Publ. C'nmp. Lamb et aL 1973). Dordrecht When the photosphere of the supergiant reaches the Vidal,N.V.. Wickramasinghe.D.T.. Peterson.B.A. 1973. Asirophvs. Roche lobe, the rate of mass loss will become very J- 182, L 77 large {— 10 ~3 M /yr, cf. van den Heuvel and De Loore, Vidal.N. V. 1974. Pub). Asiron. Soi: Paciflt- (in press) 0 Walborn,N.R. 1971. Astrophys. J. Suppl. 23. 257 1973) and the X-ray source will be completely ex- Wickramasinghe.D.T.. Vidal.N.V.. Bessell.M.S.. Peterson.B A tinguished (Shakura and Sunyaev, 1973: Pringle, 1973). Perry.M.E. 1974. Asirophys. J. 188. 167 In a binary, the "stellar wind phase" occurs when the Wolfe,R.H. Jr., Horak.H.O., Storer.N.W. 1967. m'Minimi Astro- star is a supergiant on its evolutionary track some- physics, Ed. M. Hack, Gauthiers-Villars Paris, p. 251 where to the right of the main sequence before it starts Note. The papers by Hutchings and by Wickramasinghe el al. came lo to overflow its Roche lobe. Since 3 U 0900-40 is not yet the attention or the authors only after submission of the first version extinguished, HD 77581 cannot yet have reached its of the paper. lobe. However, stars with Mi 15 Mo in binaries with P< l0d reach their lobes within a few times (0*yrs after becoming a supergiant (cf. van den Heuvel and E. J. Zuiderwijk De Loore, 1973). Therefore, the stellar wind phase in Sterrenkundig Instituut 6 Zonnenburg 2 X-ray binaries, with M< 10~ Mo/yr can last for only Utrecht, The Netherlands a few times 104yrs and, consequently the mass loss from HD 77581 cannot have exceeded a few times E. P. J. van den Heuvel 2 G. Hensbergc 10" Mo. The same argument applies to all other super- giants in X-ray binaries, i.e. also to Cygnus X-l. For Sterrenkundig Instituut Roetersstraat 15 this reason, the mass of HD 77581 is likely to be close to Amsterdam-C, The Netherlands 128 {Reprinted from Nature. Vol. 259, No. S.U4, pp. 547-549, February 19, 1976) Mass determination for the Table 1 Journal of observations X-ray binary system Vela X-l Plate no. JD2440000+ phase' Vraddons-') m.e. O-C THE 6.9 mag BO.S 1b supergiant HD77581 has been identified G3834 1773.626 0.480 - 8.64 7.79 T 8.00 as the optical counterpart of the X-ray eclipsing binary system G3842 1774.608 0.590 -18.02 3.18 + 4.77 3UO900-40 (Vela X-l)", which eclipses with a period of G3848 1775.582 0.699 -22 03 3.22 - 1.76 8.95 ^ 0.02 d. The discovery of regular X-ray pulses in Vela G3858 1776.583 0.810 -11.00 3.36 -•- 9.25 1 G3866 1777.613 0.925 - 7.72 4.14 - 3.43 X-l has also been reported , with a mean pulse period of G3879 1779.535 0.139 -16.74 5.24 - 3.1 i 282.9 s, modulated because of the radial velocity variation of G3896 1780.632 0.262 -13.85 3.35 •r 2.68 the X-ray pulsar in its orbital motion. This makes Vela X-l G3909 1781.631 0.373 - 2.37 3.11 -r 2 IS G3923 1782.556 0.476 -20.48 3.51 - 4.18 the third X-ray binary system in which the orbits of both the FI684 2169.522 0.636 -32.65 3.33 - 8.88 optical and the X-ray component can be studied, and the F1694 0.747 -26.57 1 2170.522 3.46 - .1.76 following orbital parameters have been derived , ex = F1702 2171.477 0.854 -14.37 4.06 ~ 3.25 0.15-0.05, cox = 157°±24° and Kx = 268±12 km s"i. F17O9 2173.593 0.090 + 5.68 5.52 - 2.80 Analyses of the light curve of HD77581 have shown that the F1717 2174.581 0.200 + 14.16 4.58 - 1.39 F1727 2175.594 0.313 - 3.27 4.94 - 7.56 heating effect is too small to be detected*. Furthermore, the F1739 2176.559 0.421 -13.00 6.06 - 2.36 star is bright and the spectral lines are not very broad. This F175O 2177.492 0.525 -11.60 6.31 - 8.22 means that here the first relatively accurate direct mass F1765 2180.463 0.856 -12.23 7.01) : 5.23 determination of both the X-ray and the early-type supergiant G6480 2439.647 0.763 -27.27 4.80 - 4.98 G6491 2440.710 0.882 -23.62 6.72 - 8.12 components of an eclipsing X-ray binary becomes possible. G6497 2441.734 0.996 - 8.92 3.S2 - 5.24 Earlier studies of the radial velocity variation of HD77581 G65O5 2442.684 0.102 •r 12.98 4.23 - 3.03 have given contradictory results5''. For the semi-amplitude K G6511 2443.658 0.211 + 22.75 5.54 - 7.51 1 G6519 2444.729 0.330 + 3.80 3.72 + 2.08 of the orbit, values between 19 and 40 km s" and for the F3116 2560.509 0.243 + 1.18 5.44 -11.90 F3124 2566.531 0.915 -18.34 4.19 - 5.73 • Phase zero corresponds to mid-eclipse time JD2441446.54 + n > 8.966 d. predetermined expectation of where lines should be found. Lines of H I, He I, O II, N III, N 11, Si III and Si IV were present in at least half of the spectra. From the measurements of the radial velocity of the interstellar Ca II K-line, we found 1 vc, „ = 16.1 ±0.6 (m.e.) and 13.7±l.O km s" for the 12 A mm"1 and the 20 A mm"1 spectra, respectively. To increase the homogeneity of the data we reduced the measurements of the 20 A mm'' plates to the 12' A mm ~l system, by applying the correction of 2.4 km s "'. To get an impression of the internal accuracy of these measurements five plates were measured twice. The differences in the mean velocity obtained from two such measurements oi 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Phase one plate vary between 0.3 and 4.2 km s" ; the s.d.m. per line for one plate varies from 8.0-11.5 km s~l. Fig. 1 Radial velocity curve of HD77581 = Vela X-l for a In the analysis we used mean values of the radial velocity as period of 8.966 d. Phase zero corresponds to mid-eclipse time. The points denote mean values of the measurements of lines of obtained from the He I lines and from the lines of heavier ions. He I and heavier ions. Mean errors per plate are indicated by the These average radial velocities are given in Table 1. A full table length of the vertical bars. of all individual line radial velocity measurements will be published elsewhere. eccentricity from 0.00 to 0.54 have been found. According to With the computer program 'Orbit', based on a program Wallersicin', the radial velocity data of HD77581 do not allow of Wolfe et a/.a, the best fitting radial-velocity curve through a consistent solution of the orbit, because of mass transfer in the points was computed. This was done for all lines of He I the system. Here we show that a consistent solution can be and the heavier ions together. Also separate solutions foi the obtained, however, provided that the lines of hydrogen— Hel lines and the heavier ion lines were made. The radial expected to be most sensitive to gas motions in the system— velocity measurements were weighted according to w, = I la,2, are excluded from the analysis. Our analysis of the radial where 129 Table 2 Orbital elements for HD7758! V?!a X-l Optical observations Mean values for He I Heavier ions X-ray pulsar He 1 J nd heavier elemenls 8.9*5 S.966 S.96 IS ') - 7.97 i 0.82 - 8.45 i 0.64 -7.16 f 1.06 A'(kni s' ') 19.81 1-1.19 20.54±0.99 21.19 1-1.42 268 | 12 0.20 j 0.06 0.23 ±0.04 0.22 ±0.07 0.15 ] 0.05 "(deg) 10 f 17 18-tll 2±I9 157 i 24 usin(i)(km) 2.4±0.2xl0« 2.5-tO.t ;<10« 2.6±0.2xl0« 3.27 ±0.12 X107 IM jur opinion this agreement lends confidence to the orbital 22±7 and 30±10, respectively; these values, although not ver> character of the optical radial velocity variations; the non- accurate, are consistent with the presently determined value of orhitnl (gas sti earns, stellar wind fluctuations) component does 21 2dt® ±2AJte. Therefore, the early-type supergiant in this noi sftem to have much effect on the results from He I and X-ray binary system does not seem to be particularly under- heavier elements. Using aop, sin/ = 2.5 ±0.2 10* (la) as a com- massive for its spectral type (as has been suggested" lor promise value, we find the following values for the system Cygnus X-l). puiy meters E.J.Z. p;Anowledges support by the Netherlands Organ- ization for the Advancement of Pure Research (Z.W.O.). mass p (xr) = 13.1 ±1.65 3 J. A. VAN PARADOS total mass: („//•„ -t ,(f x) sin (r» = 2I.6± G. HAMMERSCHLAG-HENSBERGE E. P. J. VAN DEN HEUVEL From a detailed analysis of the optical-light variations, Avni R. J. TAKENS UP;! Fbhtall" have found that the inclination / should be >74° E. J. ZutDERWIJK to swt a consistent picture of both the observed light curve and Astronomical Institute, the duration of the X-ray eclipse. Taking 74° and 90° as lower University of Amsterdam, and uoper limits of the angle of inclination we get Roetersstraat 15, Amsterdam, ./f ---- 1.61 ±0.27^0 and.#.„ = 21.2±2.6^ . The Netherlands, and x 3 European Southern Observatory, All quoted errors in the mass parameters are 90% confidence C. DE LOORE I.'MI.'.S. This result shows that the compact component is very Astrophysical Institute, probably too heavy to be a white dwarf. Free University of Brussels, If it is a white dwarf, its evolutionary history implies that it 10 Adolph Buyllaan 105, should consist mainly of carbon and oxygen ; the upper mass Brussels, Belgium, and limit" for such white dwarfs is ~ \AMQ>. Its most probable European Southern Observatory, mass of 1.61 Jt'. is just consistent with the presently allowed 12 La Silta, Chile theoretical masses of neutron stars (-^ =£ 1.6^3). Received October 10, [975; accepted Januuy 13.1976. The mass determination of the supergiant allows a test of the 1 Hiltner. W. A., Werner, J.,andOsmer, P.. Aslrophys. J. Utt., 175, L19-22(1972>. 2 Jones. C. and Liller. W., Aslrnphys. J. Lell., 184, L121-122 (1973). theoretically computed evolution of massive stars, through a 3 Rappaport, S., and McClintock.}., 1AU Circ. No. 2794 (1975). comparison wiih theoretical evolutionary tracks. The luminosity « Rappaport, S., and McClintock, 1.. 1AU Circ. No. 2833 (1975). * Zuiderwijk, E. J., van den Heuvel, E. P. J., and Hensbcrge. G., Astr. Astrnphyt.. of HD77581 can be inferred in two ways. The spectral type and 35, 333-360 (1974). « Hutchings, J. B., Aslrophys. J., 192, 685-689 (1974). luminosity class (B0.5 Ib) provide, in principle, the absolute ' Wallerstein, G., Amophys. J., 194, 451-457 (1974). magnitude A/», bolometric correction BC and effective tempera- ' Wolfe, R. H., Jr, Horak, H. G., and Storer. N. W., in Modern Mtrophysirs 13 (edit, by Hack. M.), 251-273 (Gauthiers-Villars. Paris. 1967). ture 7"err. Using the luminosity calibration of Blaauw or 11 » Avni. Y., and Bahcall, J. N.. Aslrophys. J. Lett, (in the press). Kecnan we derive My = -5.9 ±0.4 mag. For the bolometric *"> van den Heuvel, E. P. J., in Astrophysics and Gravitation, Proc. 16th Soimy Con/. 15 1 Physics, 119-130 (Brussels University Press, 1974). correction, values between 2.4 and 2.6 mag have been given ' ". »i Hamada. T., and Salpeter. E. E., Aslrophys. J., 134, 683-698 (1961). " Cameron, A. G. W., and Canuta, V., in Astrophysics and Gravitation. Proceedings This gives Mt. , = -8.4±0.5 mag. For the orbital parameters of the 16th Sotvay Conference on Physics. 221-267 (Brussels University Press. 4 1974). given here and by Rappaport and McClintock , and assuming a " Blaauw, A., in Basic Astronomical Data (edit, by Strand. K. A.),383-42O(Univer- minimum observed eclipse angle of 34°, Avni and Bahcall' sity or Chicago Press. Chicago and London, 1963). '• Keenan, P. C, in Basic Astronomical Data (edit, by Strand, K. A.), 78-122 derive for the radius of HD77581 R = 30 R®. For the effective (University of Chicago Press, Chicago and London, 1963). temperalure, values ranging from 22,000 K (ref. 17)-29,OOO K '! Morton, D. C, and Adams. T. F.. Aslrophys. J., 151, 611-621 (1968). '<• Schleiinger, B., Atlrophys. J., 157. 533-544 (1969). (ref. 18) have been given for early B-typt supergiants. Adopting " Osmer, P., Aslrophys. J., 181, 327-348 (1973). >• Auer, L. H., and Mihalas, D., Aslrophys. J. SuppL. 24, 193-246 (1972). Tcll 25,OO0±4,000 K we find MM = -9.0±0.75 mag. '» Simpson, E. E., Astrophys. J.. 16S, 295-316 (1971). From a comparison with evolutionary tracks19'20 we then find "> Stothera, R., Astrophys. J., 175, 431-452 (1972). the following values of the evolutionary mass: Jt]J(® = 21 Trimble, V., Rose. W. K., and Weber, J., Man. Not. R. astr. Soc. 162. IP- 3P Printed rn Great Britain by Henry Ltr.H Lfl., tt ttir Dortel Preu. Dorcheuer, Dorset 130 I'J77. A.nron. Abruptly-,. Suppl. 27. 433-434. FOUR-COLOUR PHOTOMETRIC OBSERVATIONS OF THE A'-RAY BINARY STAR HD 77581 (VELA X-l) I. OBSERVATIONS E.J. ZUIDERWIJK, G. HAMMERSCHLAG-HENSBERGE. J. VAN PARADIJS Aslronomicii] Institute, University of Amsterdam, The Netherlands and European Southern Observatory. La Sillu. Chile C. STERKEN and H. Hl-NSBERGE Astrophysical Institute, Vrije Universiteit, Brussels, Belgium European Southern Observatory, La Silla, Chile Received June 1, 1976 Results of the uvby observations of HD 77581, used in the analysis of the light curves (Zuiderwijk et a\. 1976) are presented. Kiy mirth-: X-ray binaries- uvby photometry We have observed the B0.5 Ib supergiant HD 77581, the optical counterpart of the A'-ray binary system 3U 0900-40 (Vela X-l). A discussion of the data presented here is given in a companion paper in the Main Journal (Zuiderwijk etal. 1976, paper II). The observations were made with the 50 cm Danish telescope on La Silla, Chile, with a four-channel photometer designed for simultaneous measurements in the Stromgren uvby system. The photometer is used in combination with a photon counting system, and is described in detail by Gronbech et at. (1976). HD 77581 was observed on 46 nights during the periods February 16 to March 14 (by E.J.Z. and u.H.-H.), May 27 to June 8 (by E.J.Z.) and December 3 to December 21, 1975 (by C.S.). The most extensive observations were made during the first period. Table 1 lists the 429 magnitude differences which were obtained for HD 77581 with respect to the comparison star HR 3656 (B8). The table is ordered according to JD. Each tabulated observation is the result of three integrations of 20 sec. for HD 77581 (P) and two integrations of 20 sec. for the comparison star (S). The typical standard deviation of such a sequence (one observation in table 1) is 0.005 mag. All observations of magnitude differences are given in the instrumental system, as an accurate transformation of reddened supergiants to the standard uvby system is very difficult (E(B-V) =0.73 for HD 77581, see paper II). Especially the indices cl and m\ are highly affected by the reddening. Acknowledgements are given in paper II. REFERENCES Urenbech. B.. Olsen, E.H. and Strdmgrcn, B.: 1976, Astron. Astrophys. Suppl. 26,155. Zuiderwijk, E.J., Hammerschlag-Hensberge, G., Paradijs, J. van, Sterken, C. and Hensberge, H.: 1976, Assron. Astrophys. (paper II). H.J. Zuiderwijk Astronomical Institute G. Hammerschlag-Hensberge University of Amsterdam i. van Paradijs Roetersstraat 15 N - 1004 Amsterdam, The Netherlands C. Stcrken Astrophysical Institute H. Hensberge Vrije Universiieit Brussel Pleinlaan 2 B - 1050 Brussel. Belgium 131 434 E.J. Zuiderwijk ct til. Table I Journal of observations r a..- - a—,. „ 1 -_ ..... •'"" -,„ •'.., ",', -tt-j- '.« *.»'"» ''d ii.v iii; •:';•iii; i :.•;;: ;;iJ •*tr '*w :*«! ;••« •j« ;;JJ ;;JJ ;;| isi iii: iiii is: ::::r :::s :Ki •:,s :." ::!:: :;::»: :::; :::; ::;; :•.; ' -*'*! j ^'ril£»' '.I'll '•''.'. ",'.!. '.'.{'• I -!*! •« :;;'; :.!; :5i ::if ' ••*•" | , Itt'.lct '.'.1*, '.tZ .'.'.'•• W. '-•< iii:: iiiji MhMI iii; iii: iiiri Hjiij! :i| iiii ::;: ;ii MU«J -.411 *». 1k*Nli iiii iiii i:i; :ii -1**1 »»!••> : 1 1 1 i 11 i 1 ^1-.*<« il i;i ii-i >.'**! -.'(« * -•»'* j ,l».'.* .nit .«.> .-• '.it •rt.all :!!! :!;: :;•! •:'„ ;;: :iit :!" :::; ::!." ::!S sss: ; ::;;: j 1 |Jj|;] ;S; |;-' |:| ::: a ::;: :.:', :::; •:S'. ::!5 ::::;;•; iiiis: '.'" '."' -o» -'I ; :!;»| ^:|2 :•;• iii: |;i »l ::" :!" :S1 iiii: iiiii •Sib" :s "il:«l: t*»:::*; -il'::"l !::!» Is: iii; ::'!; 3! ;-:iii 1 ; "ffli'i :.t; :,*;; ;:,' .""? 'tt* ~-'.l\* S ii:: iiii iiii iiii! iiiii Sill: ii: •;;:!:: iiii iiii ::;: :ij *L_1~—iiLJIL iiiii iiiii | ::; ii:i iiii iiiii :iit :!:i 'M'i ;:!' :::;: :.-;s j .-.. .. StKj •:;; iiii ::;; :•;;; iiii: Hiis: t* -'i«» *.>« k» .111 >•» -it* ! t* -lit* -!i*» HliiH .ttr !»rt 1 :!! iiij! 3:8! III -tkf 1 li •!! • ii 1 1 11i -*t" -'.'|M ::!'.! il -.lit -.iir lll'.tii ii! iai i •It.Ul li -*in -im Mt tilt * *i -.it* -.!•* | 132 ASTRONOMY Avlrnn. A»lrophys. 54, 167- 173 (1977) AND ASTROPHYSICS Four Colour Photometric Observations of the X-ray Binary Star HD 77581 (Vela X-l). il: Analysis of the Light Curve E. J. Zuiderwijk1, G. Hammerschlag-Hensberge1, J. van Paradijs1, C. Sterken2 and H. Hensberge2 1 Astronomical Institute, University of Amsterdam, Roetersstraat 15, 1(104 Amsterdam, The Netherlands and European Southern Observatory. La Silla, Chile 2 Astrophysical Institute, Vrije Universiteit, Pleinlaan 2, 1050 Brussels, Belgium and European Southern Observatory. La Silla. Chile Received June I, 1976 Summary. Extended photometric observations of HD tions of HD 77581 in the Stromgren uvby system. 77581, the optical counterpart of the X-ray source Our observations in this intermediate band system Vela X-l, in the Stromgren four colour system are reveal light and colour changes which are in phase discussed. These observations show changes in the with the orbital period. Colour-index changes were not light curve from one orbital period to another. A regular noticed in the UBV data obtained by Jones and Lilier double-wave variation of the colour index c\ (and (1973) and Vidal (1974). possibly b — y) is observed. These observations are In the following sections we discuss these observa- consistent with light and colour variations predicted tions and give a preliminary analysis using a theoretical from a model of a tidally distorted, rotating primary, model of a tidally distorted rotating star. These light using orbital parameters derived from spectroscopic and colour variations can be understood from the observations. temperature and gravity variations on the surface of Key words: X-ray binaries — uvby photometry the distorted primary. 2. Observations 1. Introduction The observations of HD 77581 were made on 46 nights during several observing runs in 1975 with the 50 cm The BO.51b supergiant HD 77581 is the optical counter- Danish telescope at the ESO, Chile. The results are part of the X-ray binary source 3 U0 900-40 (Vela X-l). presented in tabular form in a separate paper (Zuider- X-ray observations show an eclipsing light curve with wijk et al, 1976, paper I). In Figure 1 we show the uvby a period of 8.95 days (Forman et al., 1973). The optical magnitudes of HD 77581 during the first observing identification was first achieved by photometric observa- run, as a function of JD. A double wave in phase with tions, and confirmed by spectroscopic observations the binary period of 8.966 days is clearly visible. [This (Hiltner et al., 1972; Jones and Liller, 1973; Vidal et al, orbital period has been derived from the radial velocity 1973; Zuiderwijk et al, 1974). These observations variations of HD 77581 during a time interval of 17 showed that HD 77581 is a member of a binary system years, cf. van Paradijs et al. (1976).] It is apparent that with a period of 8.96 days. Recently a regular X-ray the minima occurring at X-ray eclipse time (=JD pulse with a period of 283 s was discovered with the 2442468.7 + n x 8.966, n integer) are less pronounced SAS-3 satellite (Rappaport and McClintock, 1975; than those occurring half a period later. We also McClintock et al, 1976). The modulation of the pulse notice the disappearance of a maximum twice during arrival times by the Doppler effect, together with the our observations. A similar phenomenon has also been optical radial velocity curve of the supergiant enabled found by Jones and Liller (1973) in their UBV light for the first time an accurate mass determination lor curves, although it should be noted, that their missing both the compact object and the supergiant in the maximum occurs after X-ray eclipse (^ = 0.25), whereas system (van Paradijs et al, 1976; Rappaport et al, the two missing maxima reported here occur half a 1976). For a review of the properties of this system we period later at 133 1-...I. /..liucmiik el al.: I'luKonielry <>f HI) 775SI (Velii X-l| tfO.OO 462.00 *6*.00 466.00 46B.0D 410.OD 472.00 474.00 416.00 470-00 480.CtJ 4SZ.O0 404 "460.00 462.00 464.00 466.00 468-00 470.00 47C.00 474.00 476.DC 478.00 480-DO 46Z.00 484.00 486.00' 488-00 "460.00 412.00 464.00 466.00 468.00 470.00 47I.0D 474.00 476.00 478.00 480.00 482.DO 484-00 486.00 488.00 JO 2**2000+ K i F s M*' V x JD Z44Z000+ Hg. l.m/iilight variations of HD 77581 for the period February 16—March 14,1975, in the instrumental system of the Danish 50 cm telescope. The orJimilcs give the differential magnitudes of HD 775X1 with -cspect to (he comparison star HR 3656 134 I. J. Zuiuerwijk ol j].: Photometry of HD 77581 (VclaX-1) IS?-00 754.00 75i.00 75S-0O 710.00 112-00 164.00 711.00 7M.00 * X X ". * X X 00.1" X * * X X X ." X X -" X X o 150.00 152 • 00 154.00 154-00 151.00 710.00 712.00 164.00 It .00 761 .00 ISO.00 7S2.O0 7S4.O0 7S»-00 ISf-OO 710.00 712.00 754.00 7M.00 7H.( **7S0.O0 7S2.00 7S4.OO ISt.OO 7SC0O 7SO.O0 7ff.O0 764.00 1H.00 7M-00 JO 2442000+ Fig. 2. uuhy fight variations of HD 77581 for the observing period of December 1975 Bessell et al., 1975; Charles et al., 1975). Another The variability of the light curve from one orbital possible explanation of the effect is that, due to the period to another is also clearly demonstrated by the non-zero orbital eccentricity, the primary may try to results from our last observing run (see Fig. 2), where adjust itself to the varying instantaneous size and the minimum during X-ray eclipse (= JD 2442755.6) shape of the Lagrangian surfaces, leading to an in- is much more pronounced than during our first observ- cipient forced pulsation [the time scale on which the ing run. shape changes significantly, i.e. half the orbital period, In spite of the variability of the light curve there are is not much different from the pulsation period 2/ periods in which the light variation appears to be rather 2 undisturbed, e.g. between JD 2442470 and JD 2442480. 135 E. J. Zuidcrwijk el til.: Photometry of HD 775X1 (VckiX-ll -0-20 0-00 0-?0 0-40 D-fO 0-ID 1 -00 1 >?D 1.40 1 .60 I-B0 Z-00 -o.to o.«o o.io o.w •( 1-tO l.#O l.W I.JO t.l PHBSE Fig. 3. v-!:gh[ curve and colour indices of HD 77581 against binary phase ( Using this part of the light curve it is possible to define curve confirms the variability of the light curve men- typical values of the amplitudes /ly1 (magnitude tioned before. Attempts to analyze the erratic varia- difference between phase 0,25 and 0.50) and Ay2 (be- tions by means of fourier methods had to be abandoned tween phases 0.00 and 0.25). We find Ay, =0"?10±0™02 because of the very irregular spacing between our and ^y2=0™06±0T015. Five observations of HD data points. The <1 index shows a double wave varia- 77581 on different nights in the uvby system were tion in phase with the light curves (see Fig. 3). Maximum made in January, 1976 by H.H. with the SOcmESO values of el occur at phases 0.0 and 0.5, minimum telescope. These observations yield the following stan- values at phases 0.25 and 0.75. The total amplitude dard V and b-y values: ^ = 6T87-6?92 and b-y = of the variation is O'!'O07+0TO03. This uncertainty O'M38 + O'!1OO2. From A NS-uttraviolet observations of has been estimated from the scatter of the mean observed HD 77581 an accurate value for the reddening could cl data points around a smooth curve drawn by free be determined: £/,_,,= 0T73 (Hammerschlag-Hensberge hand. et al., 1976). This gives £6_},=O?54 (cf. Crawford, There is less clear evidence for a systematic varia- 1975) and (f>-y)0 = — 0"Vl02. These results are in tion of the colour index b-y, and ml is constant over agreement with the spectral type of HD 77581 and the orbital period. Variations of colour indices have 7;.n =25000 K, log 0 = 2.8 (Mihalas, 1972) in agreement not been detected earlier and are indicative of tem- with the determination of Teff and gravity from con- perature and gravity variations over the stellar surface tinuum scans (Wickramasinghe et aL 1974). (see next section). The B- V and U-B colour indices A study of the colour indices reveals that these do not show the effect (Jones and Liller, 1973). We remain the same, regardless whether or not a minimum conjecture that the reason why we do find it is our or maximum disappears. In Figure 3 we have plotted larger number of observations per phase point (about the variation of y, b — y, m\ and cl as a function of 45 on the average). orbital phase. Phase zero (i.e. X-ray eclipse) corresponds Figure 4 shows the mean b—y, ml and cl values to JD 2442459.70+n x 8.966. The scatter in the y-light per phase point with 2a error bars. 136 I. J. Zuii.k-rw.jk et ai. • Photometry of HD 77581 (Vela X-1) 171 b-y 1 1 i 1 1 1 045 - } T i 047 ml -0.17 - i I I i i I i i I r I I I - -0.15 Ci -060 1 J I f I I 1 I 1 I i £ i I Kg. 4. Mean colour indices of HD 77581 against binary -058 ! phase, with 2<7-error bars 0.00 0.20 0.40 0.60 080 100 1.20 1.40 phase 3. Preliminary Discussion We have attempted to analyze these data using theoreti- we distributed equidistant points (Aq>—constant), in cal calculations of light and colour curves for a tidally such a way as to make the element of solid angle distorted rotating primary star, analogous to previous sin dAdd ). Each point on the Lagrangian surface has been (1971), Hutchings (1974), Avni and Bahcall (1975) and weighted according to the surface element rf0 = Wickramasinghe and Whelan (1975). In the following r2sin8A0A 137 i . i /'ii.kT»iik L-tal.: I'hoiomeuv of HP 775X1 (\cl.i X-li 1-rom ;!i; i.:.ncr .Jala we nave constructed a simple but made the .'.-integration over the urby fillers (Craw- approximation to me limb-darkening law—accurate ford, 1975) for the KPA models before making the lo heller than aboii! I "•• as a function of ft, TM and integration over the stellar surface. This is correct as HiiNcii.'nylli. The laige.-.t dc-.'l ''.ions occur at small long as the effective wavelength of the band duo inn values of //, where they do ni'i count much because change significantly over the stellar surface (the limb- of ihe additional projection I,;, tor /< for the surface darkening law depends on /.). This, however, is the case demem. The value of /;j/i = l) at a particu'ar pair of as the Stromgren bands are quite narrow, and the values (/;.,,. logy) has been obtained by a two-dimen- effective wavelengths will not change by more man a sional 6-point interpolation in a table giving J'^=I) few tens of Angstroms between effective tcmperaiure> for tin- KPA models. 10000 and 30000 K. The product of i) and ii) yields the contribution In the present calculations we have assumed to the total monochromatic energy received by the parameters for the VelaX-1 system, as obtained from observer per unit surface, apart from the foreshortening the analysis of the radial velocity and pulse-arriv-.i; factor ji With mass and radius values appropriate to time data. For the primary mass and the mass ratio we HD 77581 (see below) we find for the surface gravity adopted 21.4JV/0 and 0.076 ±0.009. respectively. We values around Iog?4 the 6-point interpolation formula. We have made test (in agreement with our results, see below). The *'»~ calculations for different choices of grid points, and responding uncertainty in sin / and therefore in the find that the effect on the light curves is very small. scale of the system is less than 4'',,. At a fixed value The amplitudes change by less than 0?003 over a large of the luminosity this corresponds to a change in the range of system parameters. Colour variations pose effective temperature at each point of the stellar surface more problems, as their total amplitude is of the same of two percent. Such a small difference does not show order of magnitude as extrapolation inaccuracy (a few up in the light curves. This is shown by test calcula- times (TOOI). tions in which, in turn, the scale of the system was In order to check the correctness of our program assumed constant, and the effective temperature was we calculated monochromatic light curves at /. = 5500 A, scaled by varying the luminosity by +1 magnitude. forthesameparameters( 138 1 .1. Zuidcrwijk cl ill.: Photometry ofHD 775«1 (VelaX-1) 17.1 010 q. 0.067 q.O.O85 A*. 004 - Fig. 5. Dependence of light amplitude Ay on t/. O and / (see text). The arrows denote the value o\' il I I i for which the Roche lobe is filled 189 193 197 188 192 196 200 1.91 195 199 203 O'"OO5. The maximum values of cl occur always at Hammerschlag-Hcnsberge.G.. van den Heuvd.E.P.J.. Wu.C.C.: phases 0.0 and 0.5, in agreement with the observed 1976, to be published Hiltner.W.A., Wcrner.J.. Osincr.P.: 1972, Asiruphys. J. I7S. I. I') phase dependence (see Fig. 4). We may therefore Hutchings.J.B.: 1974, Aslrophvs. J. 188, 341 conclude that we qualitatively understand the observed Jones,C, Liller.W.: 1973, Astrophys. J. 184. L 121 average variation of cl from the temperature and gravity Kopal.Z.: 1959. Close Binary Systems. Ed. Chapmann and Hull. variations over the surface ot the primary star. London, p. 125 Kurucz,R.L., Peytremann.E., Avrctt.E. H.: 1972, Blanketed Mniiel AikiitmL'tliirnwtits. We are indebted to Prof. Dr. E. P. J. van den Atmospheres for Early-type Stars, Smithsonian Astrophysical Heuvel for stimulating discussions during the course of this investiga- Observatory, preprint tion. E. J. Zuiderwijk acknowledges support by the Netherlands McClintock.J.E., Rappaport.S., Joss.P.C, Bradt.H., BulT.J.. Clark. Organization for the Advancement of Pure Research (ZWO). H. Hens- G.W., Hearn.D., Lewin.W.H.G., Matilsky.T., Mayer.W.. Primini. berge acknowledges support by the National Foundation of Collective F.: 1976, preprint Fundamental Research of Belgium (FKFO) under no. 10303. Mihalas.D.: 1972, Astrophys. J. 176. 139 The calculations described in this paper have been performed van Paradijs.J.A., Hammerschlag-Hcnsbcrge.G.. van den lleuvcl. with the CDC cyber-73 of the Stichting Academisch Rekencentrum E.P.J., Takens,R.J.. Zuiderwijk.E.J., 139 4>>, .121 .12.1(1976) Letter to the Editor Photometric Variations of Wray 977 (3U 1223-62?) G. Hammerschlag-Hensberge, E. J. Zuidenvijk and E. P. J. van den Heuvel* Astronomical Institute, University of Amsterdam, the Netherlands and European Southern Observatory, La Silla. Chile H, Hensberge Astrophysical Institute, Vrije Universiteit, Brussels, Belgium and European Southern Observatory, La Silla, Chile Received December 17, 1975 Summary, uvby—photometric observations are wave with a period of 23+1 days. The A 1 supcrgiunl presented of candidate stars for the X-ray source SAO 251905 inside the error box shows no significant 3 U 1223-62. The eariy B supergiant Wray 977 shows variations, light variations with an amplitude of about 0.06 magni- tudes in p, b and y. The variations resemble a double- Key words: X-ray binaries—uvby photometry 1. Introduction The !lth-magnitude early B-supergiant Wray 977 was for HD 107944 (C) 60s. Each observation sequence was proposed by Vidal (1973a) to be the optical counter- performed as follows: C-P1-P2-P1-C-P1-P2-P1-C. part of the X-ray source 3 U 1223-62. Photometric Table 1 gives the mean values for each sequence for the variations with an amplitude of about 0!"l on a time brightness measurement of the star minus comparison scale of days were reported by van Genderen (1973), star. The errors (m.e.) in Table 1 denote the lc-error Vidal (1973b), Mauder (1974) and Bord et al. (1975). levels of the sequence. A possible periodicity of 13.5 days was mentioned by We did not tabulate el-vaiues for Wray 977 as the errors Mauder (1974) and of 9 days by Bord et al. (1975). in the u-channel were too large to enable detection of The A ! Ha emission star SAO 251905, close to Wray periodic variations. This is due to the weakness of the 977 (at a distance of 115) is also an interesting object, and star in the ultra-violet produced by its large interstellar inside the 3rd UHURU 90%-confidence error box. reddening [i.e.: E(B-V)~ 1!"8, Vidal (1973b), see be- For these reasons, we include these two stars in our low]. program of photometry of Southern X-ray Sources Considerable zero-point variations during the night performed at the European Southern Observatory, showed up at the reduction. This is probably due to the La Silla, Chile. absence of a cooling system for the photometer. For the differential magnitudes and colours, however, the errors induced by zero-point variations are less than 2. The Observations O.ra001 and can be neglected compared to the observa- The observations were made with the Stromgren tional errors (see Table 1: columns labelled m.e.). On the uvby-four channel photometer at the 50-cm reflecting other hand, these zero-point changes make an accurate Danish telescope. The photometer has a pulse counting transformation to the standard system rather difficult. system and is described by Gronbech et al. (1975). The For these reasons we only give here the colour differences stars were observed by E.J.Z. and G.H.-H. during the in the instrumental system, which enable one to ac- periods Feb. 16-March 14, May 27-June 8 and on curately establish the variations of the candidate stars. July 25, 1975. Figure 1 shows the differential magnitudes Av, Ab and We measured the differential luminosity of our two Ay for Wray 977 in the instrumental system. Figure 2 program stars with respect to the comparison star shows the same for SAO 251905. HD 107944. The integration time for one observation of In order to be able to transform some of the colours Wray 977 (P 1) was 196 s, for SAO 251905 (P 2) 98 s and towards the standard uvby system, H.H. observed Wray 977 five times during three nights in January * Also at the Astrophysical Institute, Vrije Universiteit, Brussels 1975 on the ESO 50 cm telescope with a cooled ESO Belgium. photometer. These observations yielded the following 141 Photometry of Wruy 977 l';iblc I. Journal of observations 11.11) Wuiy 977 HI) 107944 SAO 251905 HI) 107944 2442000 I •1.1' m.c. Mh >•) mo. Ami m.c. Ay m.c. /Hh y) m.e. /1ml m.c. ,1<-1 m.c. •Kill. 7-7 2.449 .006 .814 .006 -.443 .022 1.547 .013 .316 .012 .086 .020 .548 .015 4M.76 2.475 .005 .831 .006 -.462 .016 1.546 .008 .310 ,007 .095 .006 .545 .010 462.76 2.496 .007 .814 .005 .432 .014 1.546 .010 ,304 .013 .107 .023 .535 .019 46.V76 2.511 .006 .823 .004 -.472 .007 1.552 .006 -.314 .003 .094 Oil .552 .010 464.77 2.495 .004 .820 .005 —.459 .014 1.553 .004 .318 .002 .095 .008 .5.14 .011 465.7(i 2.496 .007 .825 .010 -.459 .OIK 1.550 .003 .312 .001 .101 .007 .535 .008 466.76 2.465 .004 .819 .004 -.407 .030 1.534 .007 • .305 .006 .110 .005 .530 .(X)6 467.76 2.448 .008 .813 .010 —.430 .026 1.536 .003 -.308 .003 -.100 .007 .549 .010 468.77 2.451 .008 .Sif .010 —.449 .021 1.531 .007 —.307 .009 .100 .014 .575 .025 469.78 2.484 .005 .807 .011 —.451 .023 1.537 .009 - .304 .002 .115 .004 .532 012 470.77 2.492 .005 .819 .009 ' -.485 .010 1.533 .019 -.298 .022 .117 .020 .538 .005 471.SO 2.455 .008 .849 .010 —.495 .021 1.538 .004 .312 .004 .096 .007 .546 .018 472.76 2.476 .011 .835 .013 —.483 .016 1.548 .004 -.316 .007 .090 .010 .527 .028 •173.77 2 473 .010 .835 .018 -.457 .026 1.543 .004 .303 .002 .106 .003 .533 .010 474.79 2.470 .003 .829 .005 —.485 .015 1.537 .008 --.296 .008 .123 .013 .519 .008 •(75.77 2.493 .005 .809 .010 —.434 .012 1.542 .002 .303 .005 .106 .006 .532 .011 479.79 2.446 .003 .816 .006 -.467 .009 1.537 .005 -.311 .003 --.105 .008 .526 012 (80.79 2.457 .004 .809 .008 —.447 .020 1.541 u03 --.318 .006 ,093 .008 .532 .012 481.80 2.459 .007 .827 .004 -.462 .023 1.531 .010 .306 .005 - .100 .004 .545 .007 482.79 2.492 .005 .802 .005 -.439 .009 1.539 .008 .319 .006 —.087 .007 .555 .019 484.79 2.502 .005 .804 .007 —.431 .017 1.529 .004 --.301 .005 --.113 .007 -.539 .005 485.81 2.490 .006 .816 .009 —.469 .015 1.537 .008 — .303 .011 —.107 .018 .537 .017 4S6.80 2.500 .003 .819 .009 -.461 .018 1.541 .004 -.309 .005 .101 .009 .53X .012 560.54 2.491 .003 .808 .011 —.486 .023 1.530 .014 ,310 .011 .084 .010 .562 .014 560.57 2.483 .003 .819 .017 —.437 .007 1.541 .002 - .313 .002 .100 .012 .532 .0.16 566.57 2.535 .004 .807 .009 —.437 .026 1536 .002 -,307 .(X)2 .103 .002 .516 .011) 567.59 2.539 .007 .804 .005 —.431 .012 1.530 .013 .313 .013 -.090 .020 .554 .006 569.54 2.512 .004 .825 .007 —.469 .017 1.530 .006 - -.301 .011 .112 .015 .543 .015 570.54 2.457 .004 .833 .011 —.476 .011 1.535 .003 -.305 .003 -.109 .003 - .533 .009 571.52 2.454 .003 .824 .002 —.454 .012 1.534 .003 —.298 .002 - .108 .001 - -.553 .008 572.5! 2.463 .005 .821 .005 —.478 .012 1.538 .002 -.304 .005 ,105 .006 -.552 .002 619.52 2.482 .008 .788 .010 -.423 .015 1.531 .008 -.310 .004 ,103 .006 .531 .016 Wray 977 -' HD 107944 SAO 251905-HD 107944 : •• •••• 1 : • •••• ; 326 32$ 130 ....'•''• ''•.•. ' '•• " r • " • • * i i i i i i i i i 175 ISO CIS JD 2U2000 A Fig. 2. u, v, b and y observations of SAO 251905. in the instrumental system. No significant variations seem present MO 511 S7D JO 1U2000 . rig. I. i, h and v variations of Wray 977, in the instrumental system of the Danish 50 cm teleicope. The dasbe4*ufwe»-arc tuggeMed light 1970)]. From the above values the reddening of Wray curves, drawn by free hand 977 was calculated as follows. The spectral type B 1.5 la, m as estimated by Vidal (1973b), yields (B-V)o= -0 20. magnitudes and colours: V= Kr83±(T01 (cf.: Vidal's (ft-y)0=-(r07 (Crawford and Barnes. 1970). Then value: V= 1(T84) and b-y =l™36±(T05 [ml is not E(b-y)= r36+(T05 and E(B-V)= r9±O!"2 (cf Craw- given in the standard system, due to the uncertainties ford, 1975). The latter value is in good agreement widi produced by high reddening (cf. Crawford and Barnes, the value £(B-^)=1™8 as derived by Vidal (1973b). 142 G. Hammerschlag-Hcnsberge et al. 323 3. Results and Discussion Wray 977 is a B 1.5 la star whereas HD 77581 is a B 0.5 Ib star. Typical radii for such stars—with masses Wray 977 shows definite magnitude changes on a time around 25 M —are ~45 and 30 R , respectively scale of days. Our observations in the v, b and y channels Q Q [adopting effective temperatures of 23000 and 27 500 K, suggest a periodic variation with a double wave in a respectively, cf. Lamers and Snijders (1975)]; assuming period of 21-23 days. The double wave behaviour is both stars to be almost filling their Roche lobes, and to suggested by comparison with light curves of other have companions of about 1-2 M , one finds that X-ray binaries, e.g. HD 77581= Vela X-l (Jones and o indeed binary periods of about 20 and 10 days are to Liller, 1973b), HD 153919=3 U 1700-37 (Jones and be expected, respectively. So, the spectral classification Liller. 1973a; Penny et al., 1973), HDE22686S = of Wray 977 seems in reasonable agreement with the CygX-1 (Lester et al., 1973). Both in van Genderen's binary period derived from our observations, adopting (1973) and our observations one of the two minima per the photometric curve to be double-peaked. cycle is quite outspoken, while the other one is less well defined and subject to large and irregular variability. Acknowledgements. E. J. Zuiderwijk acknowledges support by the This behaviour is very similar to that of the lightcurves Netherlands Organisation for the Advancement of Pure Research of HD 77581 (Vela X-l) and HD 153919 (3 U (Z.W.O.). H. Hensberge acknowledges support by the National Foundation of Collective Fundamental Research of Belgium (FKFO) 1700-37). under No. 10303. When we combine our observations with those of van Genderen (1973), which were made during the period July-August, 1973 when he very frequently observed References the star, a period of around 23+1 days appears to give a good fit to the observations. Remarkable is that the Bord,D.J., Mook,D.E., Petro.L., Hiltner.W.A. 1975. Workshop papers for a symposium on X-ray binaries. Ed. Y. Kondo and mean intensity of the star has changed by 0!"03 from E. Boldt, GSFC. Grecnbelt, Maryland, preprint X-66O-75-2K5. the period February—March to the period May-June, p. 363 1975. These variations are intrinsic to the star, as such a Crawford.D.L. 1975, Publ. Astron. Soc: Pacific SI, 481 variation is not observed for SAO 251905. This might Crawford,D.L., Barnes.J.V. 1970, Astron. J. 7S, 978 be due to changes in the amount of gas around the Genderen,A. M.van 1973, Inf. Bull. Var. Stars. No. 856 Granbech, B., Olsen, E. H., Stromgren, B. 1975, Astron. & Astrophys. B star; similar changes have also been found in other Suppl. Series (in press) B-supergiants (Sterken, 1976). A more extended se- Jones,C, Liller,W. 1973a. Astrophys. J. 184. L65 quence of observations, covering several periods, how- Jones,C, Liller, W. 1973b, Astrophys. J. 184, L 121 ever, is needed to confirm our results. Lamers.H.J.G.L.M., Snijders,M.A.J. 1975. Astron. <$ Astrophys. The colour difference A(b-y) as a function of phase 41,259 Lester.D.F., Nolt.I.G. RadostittJ.V. 1973, Nature Pins. Sri. was studied and no colour variation was found. The 241, 125 scatter a{A{b~y))= "012 (one measurement) is not Mauder.H. 1974, l.A.U. Circ. No. 2673 larger than that expected from photon statistics. Penny.A.J., Olowin.R.P., Penfold.J.E., Warren,P.R. 1973. Monthly As Fig. 2 shows, SAO 251905 has no significant magni- Notices Roy. Astron. Soc. 163, 7p tude or colour changes (i.e. not larger than + 0!"01 in Rappaport,S., McClintock.J. 1975, l.A.U. Circ. No. 2833 Sterken,C. 1976, Astron. & Astrophys. (in preparation) the o. b and y channels). Vidal,N.V. 1973a, I.A.U. Circ. No. 2569 Our results imply that the B 1.5 supergiant Wray 977 Vidal.N.V. 1973b, Astrophvs. J. 186,1.81 seems to be the best optical candidate for the X-ray White,N.E., Huckle.H.E., Mason.K.O., Charles.P.A., Pollard.G.. source 3 U 1223-62. More photometric observations Culhane,J. L., Sanford, P. W. 1975,1.A.U. Circ. No. 2870 and high dispersion spectroscopy are highly needed. Recently a short periodic variability in the X-ray G. Hammerschlag-Hcnsberge intensity of 3 U 1223-62 of 11.64min was reported by E. J. Zuiderwijk White et al. (1975). This makes this system quite similar E. P.}. van den Heuvel to that of Vela X-l, which also consists of an early-type Astronomical Institute supergiant (HD 77581) and a slow (283 s) pulsar (cf. University of Amsterdam Roetersstraat 15 Rappaport and McClintock, 1975) with an orbital Amsterdam, The Netherlands period of 8.96 days. The rather long binary period of Wray 977 seems in fair agreement with the fact that H. Hensberge this supergiant is of a somewhat later type than the Astrophysical Institute Vrije Universiteit Brussel companion of Vela X-l: according to Vidal's classifi- A. Buyllaan 105 cation, and confirmed by one spectrum taken by us, B-1050 Brussels, Belgium 143 PART XX. MAGNETIC AND BELATED STARS 145 \!>trcm. & AMrophys. 32, 457 - 459 (1974) Detection of Crossover Effect in the Ap Star HD 98088 r. Hensberge stronomical Institute, University of Amsterdam, The Netherlands eceived February 19, 1974 ummary. On a high dispersion Zeeman plate of the Key words: magnetic stars - crossover effect - peculiar u-Cr-Sr star HD 98088 several lines of Fe, Cr and Ti A stars low the crossover effect. The mean effective field rength in the magnetic regions where the lines arise -8400 ±450 and +7600 + 400 Gauss, respectively, tifferent ions show slightly different field strengths. itroduction he magnetic Ap star HD 98088 was found by Abt .18 [cf. Fig. 2 of Babcock (I960)] where the effective 953) to be a spectroscopic binary with a period of field "crosses over" from negative to positive polarity. '905. A detailed description of the system is given Another Lick-plate taken at phase .40 (He maximum) y Abt et al. (1968), who also detected some spectral of the magnetic cycle shows lines which are comparable les due to the secondary component. The star belongs in strength and width in the two circularly polarized i the Eu-Cr-Sr group and its effective field strength spectra, in agreement with the expectation that at this iries from +800 to -1000 Gauss with the same phase one only observes patches of the same magnetic :riod as the orbital one (Babcock, 1958). Babcock polarity. The high accuracy in X- and y-direction of so noticed that the crossover effect seems to be the Faul-Coradi microphotometer of the Utrecht >sent, while most other magnetic stars of the same Observatory (better than 0.4 u in both directions) pe, with periodically reversing magnetic polarity enabled us to determine the field strengths of both ow the effect. We investigated coude spectra (at positive and negative magnetic patches from the lA/mm) which were obtained by Conti at Lick splitting of the lines on tracings of the left-hand bservatory in 1968. In April 1973 additional plates circularly polarized spectrum. The difference in wave- ith a dispersion of 12 A/mm) at the crossover phases length between lines on the Fourier-noise filtered ;re taken by van den Heuvel with the 152-cm tracings of the left- and right-hand circularly polarized escope of the ESO, La Silla, Chile. spectra could be determined to an accuracy better than 0.002 A. The results are given in Table 1. Column 3 lists the ie Crossover Effect z-values of the considered transitions, computed in LS coupling, which for the iron peak elements seems le crossover effect arises when integrated light from a good approximation (cf. Preston, 1969); Columns 4 o different regions of the stellar surface is observed and 5 give the displacements AXC of the lines arising th a longitudinal Zeeman analyser: due to the in a patch of negative and positive magnetic polarity, ference in projected rotational velocity of the two respectively, in mA, where AXC is related to the 2 ;ions, spectra! lines which are formed in these observed shift Ak by AAC = (4500/A) AL sions. are split or broadened in one of the two Some examples of crossover effect in HD 98088 arc- cularly polarized spectra at a certain phase (cf. shown in Fig. I. The effect is especially clear for lines bcock, I960; Preston, 1967). A similar splitting or of Fe, Cr and Ti. As can be seen from Table 2, we find jadening occurs half a period later in the spectrum slightly different field strengths He for different ions, 3 opposite polarity. The originally studied Lick-plate where He is given T>y He = 52.7x 10 ZAAJZz (cf. Bab- ich shows the crossover effect corresponds to phase cock, 1958). 147 Ci. Hciwhergi: Tjbk' ' /ccman displacements for lines showing the crossover elTccI in ill") 98088 (see also text) Kin AK( -) .4A,( + ) (mA) (mA) 4082.1 Fe 1.25 425 243 41 TO. 2 Cr 2.50 173 404 4195.4 Cr i 1.40 230 173 4200.9 Fe 1.88 229 287 4238.0 Fe 1.75 282 169 4250.1 Fe 1.50 168 224 4252.6 Cri! 1.20 168 112 4254.4 Cr 1.38 168 168 4274.8 Cr 1.96 166 111 4312.9 Til 1.48 163 109 4359.6 Cr 1.25 ' 266 266 4388.4 Fe 1.67 210 158 4397.3 Cr 2.00 314 367 4422.<- Fei 1.00 259 259 4468.5 Til 1.06 203 203 4469.4 Fei 1.50 253 203 4470.9 Til 1.27 203 203 4476.0 Fci 1.00 303 303 4489.2 FCI 1.50 251 251 45(11.2 Til 0.93 120 150 Fig. I. Positive crossover effect for 3 selecled lines in the spectrum 4533.2 Til 1.40 246 246 of HD 98088 at the phase where H,. = 0. The upper and lower pans 45.34.0 Til 1.10 148 99 of the figure represent the right- and left-hand circularly pnlari/eJ 4554.0 Bai 1.17 195 146 spectrum, respectively 4583.8 Fei 1.17 145 145 4588.2 Cri 1.07 192 !92 4590.0 Tin 1.07 192 96 Discussion of the Results Table 2. Mean effective field strengths for different ions in Unfortunately no other Lick plates (except one of bad HD 98088. H,(-) and H,{+) are the strengths of the negative and positive polarized spots respectively, n = number of measured lines quality) were taken at one of the crossover phases. Therefore some additional plates were taken in 1973 Ion B w,(+) at these phases. However, the dispersion of 12 A/mm (Gauss) (Gauss) proved too low to reveal the effect. For some lines the effect of crossover broadening can be seen, but it is Fci 8 S710±1140 8420 + 960 Fell 2 7820+ 790 7820 ±790 impossible to measure the field strengths with sufficient Fe i + ii 10 9360+ 940 8310 ±770 accuracy. The fact that the effect is only clearly visible Cri 5 6300+ 940 7630+960 on microphotometer tracings of high dispersion spec- Cm 3 8470± 410 6850±9O0 trograms is probably also the reason why it has not Cr i + II 8 6930+ 620 7410 + 650 been discovered before in this star. Til 1 9260 9260 Tin 6 7850+ 610 6560±710 The rotational velocity obtained for the observed Ti t + II 7 8090± 510 7010 + 650 patches is very low compared to the equatorial velocity nil lines 26 8400 ± 450 7640 + 400 of 25krn/s {Preston, !970b). If the oblique rotator model (Stibbs, 1950; Deutsch, 1954, 1958) is adopted and if one considers an undisplaced dipole model Similar differences have been observed in other Ap for the magnetic field of this star, one obtains a value stars (Wolff, 1969; Preston, 1970a). From the 26 lines of 75° for the angle /? between the rotational and the studied the following mean effective field strengths magnetic axis (cf. Preston, 1967). This angle doesn't for the two iron peak patches were derived: change much for a displaced dipole model (cf. Land- H ( -) --= 8400 ±450 Gauss street, 1970). If the two patches are situated near the Q magnetic poles of the star one would expect a rotational H,.(+ ) = 7640 ± 400 Gauss. velocity of 24km/s instead of the observed velocity A mean rotational velocity of 6.9km/s was adopted of 7km/s. A possible explanation is that the iron (we assume here the same velocity for the two patches). peak elements, for which the crossover effect is seen, Due to the considerable width of the lines, the estimated are not abundant at the magnetic poles but are more uncertainty per line is about 3000 Gauss. AH the or less concentrated in a broad belt or patches around individually measured field strengths differ less than the magnetic dipole equator. In this picture the magnetic this value from the mean. equator should divide the belt in two almost equal 148 ApSlar HP 98038 45') just one broad belt around the equator, about 70 HD 98088 wide - as depicted in Fig. 2. Acknowledgements. 1 thank Dr. P. S. Conli and Dr. E. P. J. van den Heuvel for obtaining the speclrograms of HD 98088. [ wish also to express my gratitude to Dr. E. P. J. van den Hcuvel for many stimulating discussions about the subject and for his continuous guidance during this investigation. This research was carried oul when the author was working at secondary the Astronomical Institute of the University of Utrecht. primary References Fig. 2. Suggested model for the element distribution on the surface Abt.H.A. 1953, Publ. Astron. Sac. Pacific 65. 274 of the primary of HD 98088. The shaded area indicates the con- Abt.H.A., Conti,P.S., Deutsch.A.J., Wailerstein.G. 1968. Astro- centration of the iron peak elements in a belt about 70° wide phys.J. 153, 177 around the magnetic dipole equator. The maximum abundance of Babcock.H. W. 1958. Asirophys. J. Suppl. Ser. 3, 141 the rare earths is indicated by the hatched polar area (see also text) Babcock.H.W. 1960, in Stars and Stellar Systems, Vol. 6. Ed. J. I.. Greenstein, Univ. of Chicago Press, Chicago, p. 282 Babcock.H. W. 1962, in Stars and Stellar Systems, Vol. 2, Ed. W. A. parts of opposite magnetic polarity. This idea is Hiltner, Ui.:v. of Chicago Press. Chicago, p. 107 supported by the investigation of the Ap star a2 CVn Deutsch.A.J. IV54, Trans. Intl. Astron. Union, Cambridge Uni- versity Press, Cambridge 8. 801 (Pyper, 1969) where the iron peak elements are con- Deutsch.A.J. 1958, in Electromagnetic Phenomena in Cusmkal centrated in a belt consisting of four extended patches Physics, (I.A.U. Symposium No. 6), Ed. B. Lehnert, Cambridge at the magnetic equator, while the rare earths are most University Press, Cambridge, p. 209 abundant near one of the magnetic poles. It should LandstreeU.D. 1970, Asirophys. J. 159, 1001 be noticed that we do not observe the crossover effect Preston,G.W. 1967, in The Magnetic and Related Stars, fid. R. C. Cameron, Mono Book Corp., Baltimore, p. 3 in the lines of rare earths in HD 98088, which might Preston.G. W. 1969, Asirophys. J. 156.967 indicate that also here these elements are concentrated Preston.G.W. 1970a, Asirophys. J. 160, 1059 mostly in a patch of one magnetic polarity. Such a Preston.G.W. 1970b, in Stellar Rotation. Ed. A. Slellebak. Keidd distribution is supported by the fact that the strength Publ. Comp., Dordrecht, p. 254 Pyper,D.M. 1969. Astmphvs. J. Suppl. Ser. IS. .147 of the Eun lines exhibits a primary maximum at the Stibbs,D.W.N. 1950, Monthly Notices Roy. Astron. Sac. 110. .W5 phase where the effective field reaches its maximum Wolff,S.C. 1969, Asirophys. J. 157, 25.1 value (Abt et a/., 1968). In the star HD 98088 the centers of gravity of the belts of opposite polarity, G. Hensberge where we measure the crossover effect would be Sterrenkundig [nstituut separated in meridional direction by about 35°. The Universiteit van Amsterdam Roetersstraat 15 simplest interpretation seems that we are dealing with Amsterdam, Nederland 149 Aslnin. &. A-lniphys. 48, 383 -387 (1976) Photometry of Peculiar A Stars H. Hensberge and C. De Loore Astrophysical Institute, Vrije Universiteit Brussel, European Southern Obsi-Tvatory, La Silla, Chile E. J. Zuiderwijk and G. Hammerschlag-Hensberge Astronomical Institute. University of Amsterdam, European Southern Observatory, La Silla. Chile Received October 28, 1975 Summary. The results of uvby-photometry of 4 Ap-stars silicon stars in our sample (the two previous ones and observed during 1975 at the ESO, La Silla, are present- HD 73340) are variable in the four channels showini; ed. One of the stars, HD 81009, was also observed their largest amplitude in u. HD 81009 shows at least during 1971 in the UBV system. Periods are proposed variability in the u-channel. for the silicon stars HD63401 (P=2?41±.O2) and HD92664 {P= l?668±,004 or 0.624±,002). The three Key words: Ap-stars - photometry 1. Introduction Bidelman and MacConnell's (1973) list of southern (.45< u-b<.55) than 41 Tau and HD 215441 (»-/)>.6). peculiar stars contains a number of newly discovered they must be significantly hotter. Only HD 34452. ob- ones. Among these, we found three silicon stars which served in B and V by Rakos (1962) is as hoi as HD 92664. might be of some interest. They are bright enough to HD 81009 has been observed in UBV by De Loore in allow accurate photometry; furthermore, until now 1971. The star shows no variability with amplitudes 41 Tau and HD. 215441 were the hottest silicon stars larger than OT005 (in V, B-V and U-B) during the to have been observed in some detail in ubvy in order to 10 day interval that we observed it. However, the <•, determine the character of the light variations. Accord- value of this star in the catalogue of Gronbech et al. ing to Megessier (1971), u—b is a relevant effective (1975) was quoted with an unusually large standard temperature indicator for silicon stars. As HD 92664, deviation. Since this star has very sharp lines (van den HD 73340 and HD 63401 all have smaller u-b indices Heuvel, 1971), we suspected it to be a long period Table 1. Information about the observation runs Star Peculiarity Comparison Observing Number of nights Observer/ type stars times (1975) (number of obs.) telescope" HD 63401 Si HD 62712 Jan. 14-28 9(18) HH/ESO (HR3032) and HD 63215 Feb. 26 1(2) GH, EZ/Dan HD 73340 Si HD 74071 Jan. 14-28 8(9) HH/ESO (HR3413) March, 2-14 7(7) EZ/Dan HD81009 Sr-Cr HD 80447 May, 11-21(1971) lil) UBV CDL/Dan (HR 3724) and HD 82,428 Jan, 14-19 4(4) HH/ESO HD 81728 (in 1971) I < 0 92664 Si HD93163 Jan, 15-18 4(5) HH/ESO (HR 4185) and HD 93194 Feb, 16-March, t 12(12) OH, EZ/Dan a Abbreviations: HH = H. Hensberge; CH-C. Hammerschlag-Hensberge; EZ=E. J. Zuiderwijk: CDL=C De Loore: ESO = ESO 50cm telescope: Dan = Danish 50cm telescope. 384 H. Hensbergc el at. variable. Th»>. we decided to reobserve this star in Table 2. C"( lours of the comparison stars our 19~5 runs, which should permit us to have Star HI h-y m, observations over a much longer time interval. ', v—h, u—v) show very low scatter however, ed in both instrumental systems are intrinsically the resulting in standard deviations smaller than 0TO03. same. Thus we did not adopt a transformation from one Therefore y. b'—y, v—b and u—v were used to select instrumental system to the other. possible periods by the technique described by Lafler A complete list of the observations and the transforma- and Kinman(1965). tion formulae adopted are available from the authors upon request. 3. Reductions The extinction corrections were computed in the usual 4. Results way, defining average coefficients per night; transforma- For those cases in which the data enabled us to propose tion to the standard system was done by fitting compari- a period only a plot is given: otherwise the observations son stars and extinction stars to the uvby system of the are tabulated. The plots were made after transformation catalogue of bright southern stars (Gronbech et al., to the uvby system of the catalogue of bright southern 1975). The mean colours of the comparison stars are stars (Gronbech et al., 1975). Phase zero is chosen given in Table 2. The present observations confirm the arbitrarily as the time of the first observation. catalogue values up to 0^005 in two third of all cases. Only two of the colour indices differ by more than 0T011, the mi-index of HD 82428 ((TO 16) and the c,-index of A. HD 63401 HD 63215 (0T018). Table 3 gives the approximate magni- All observations are plotted in Figure I; phase zero tude range of the four peculiar stars. However transform- corresponds to J.D. 2442427.56. The phases are ation from y to V is rather tentative for Ap stars since computed according to a period of 2?41. Considering 152 Photometry of Ap Stars 385 Table 4. Observations of HD 73340: final results 630 -to J.D. tn b if 632 a a • a 2442400+ 634 o % a. a & D a e 27.59 5.797 5.742 5.799 6.254 636 • 28.60 5.781 5.716 5.778 6.235 i 29.59 5.813 5.744 5.815 6.286 b 30.58 5.804 5.747 5.801 6.262 620 30.84 5.788 5.731 5.785 6.248 a D G o 622 31.59 5.788 5.719 5.775 6.234 a D 32.59 5.810 5.757 5.819 6.294 624 D a a 38.66 5.792 5.735 5.792 6.258 626 D a 41.71 5.792 5.725 5.789 6.258 74.64 5.791 5.726 5.791 6.257 82B 78.60 5.802 5.743 5.804 6.264 79.60 5.788 5.720 5.781 6.24K V 80.60 5.795 5.741 5.797 6.253 625 - Off b arfbD 81.59 5.802 5.736 5.800 6.267 6.27 a o o a - 85.60 5.794 5733 5.795 6.264 a a 86.60 5.794 5.738 5.797 6.255 629 % a 631 D D „ °D & 6.33 • u accuracy can partially be due to the fact that we had to omit one of the comparison stars. It should be noticed act that if the lightcurves have really double waves and if 6.75 - D o • the number of observations is rather limited as in our 67/ case, the true period can be masked by erroneous single D O DO1 a 6.79 - o a wave periods. As we are unable to select a period, we a a a 66. D a listed our observations in Table 4. The values suggest a a phase 6.83 period of the order of a few days or less, since the star 1 changes its brightness from maximum to minimum in 0 .2 .4 .6 .8 0 .2 .4 one day. Fig. 1. Light variations of HD 63401. Phases are computed according to P = 2f41. Open squares denote ESO telescope observations, filled squares denote observations with the Danish telescope C. HD 8W09 the uncertainty induced by the method used for the Our UBV photometry of 1971 reveals that this star determination of the period and by the neglect of a trans- remained essentially constant during the 10 day observ- formation between both instrumental systems, one may ing run. The probability is very low that this is caused by rely on this period up to ±0?02. observing the star always at the same phase but while This star exhibits a rather complicated photometric one or more cycles have elapsed, since HD 81009 has character. The y and « lightcurves show secondary very sharp lines (van den Heuvel, 1971) so that periods extrema while b and v show an almost identical and of the order of one day are very unlikely. The attempt to single wave. The primary maximum occurs at the same collect observations covering a longer time interval phase for all curves, but u reaches its secondary during our 1975 observing runs was abandoned at the minimum when the brightness at longer wavelengths is end of January because bad wheather conditions obliged minimum. Minimum brightness in u corresponds in us to reduce our program. Although no ui*y-variability phase to the other minimum in y. This characteristic-- could be detected in this observing run. some remarks of i.e. that the importance of the secondary extrema can interest for further investigations can be made. First. change rather drastically with wavelength—has also been y and earlier Vmeasurements relative to HD 80447, and found for other silicon stars i.e. HD 19832=56 Ari the agreement between our standard (V), b—y and m, (Wolff and Morrison, 1974) and HD 224801 (Stepien, values and those of Olsen et al. (1975) seem to lend 1968). support to the non-variability of this star in the blue and in the visible. However c, assures that this star is at least variable at lower wavelengths: there is a disagree- B. HD 73340 ment of 0T055 between the catalogue value off, and our The Lafler-Kinman (1965) technique cannot distinguish mean c, index (see Table 3). So we feel that further between several periods, all of them showing definitely investigations should concentrate on measurements larger scatter than in the other cases. The loss of below 4000 A during an extended time interval. 153 following: (a) ihis paper; tb) Kmiuirn 11973) and reiereiiLLs lin. 1 '3 j (•-•I WolfTI 1973): (d) WtiirTiind Morrison (1975). The IM -/i)-i:il»t 5« °*n u from Lindcmiinn :md liauck (197.1) m 451 Siai Releraia 1 i HD 92664 .45 a b HD.U452 .46 b 5:39 .49 ' ° fl a HD 73340 a • 1 ID 6341)1 .51 a • D * 9 D 1 ° "* HD25K23 = 4ITau .62 b.c j HD21544I .65 b 1 HD215O3X .69 b V HD 223640 .71 b 3.C4 o _ D ^ HD I9S32 .71 b.d Q HD 124224 .75 b • m HD 32633 .77 b SV.t HD 224801 .87 b 1 2 II HD 112413s* CVn .88 b HD 30466 .92 b D ^ D f HD 133029 1.01 d -65 •• • HD 184905 1.01 b HD l«296 1.01 b 5.87 a D • II HD 74521 1.02 b a • HD 68351 1.23 b % n HD 173650 1.26 b 591 # phase 1 c .2 A .6 .8 0 .2 .4 iiji. 2. Lijhl variations of HD 92664. Phases are computed according 20 hot Ap stars are compared with the cooler nncx a is lo P= l?664. Symbols as in Fig. I obvious that the cool group presems more diversity. We can summarize the main differences in overall photometric character in the blue-visible by two state- D. HD 154 Photometry of Ap Stars 3X7 iii that case, situated well below 3500A. Therefore, References this suggestion is consistent with the observations. Bidelman.W.P., Mac Council,D..I. 197.1. Asiron. J. 78. 687 b) None of the -.ilicon stars shows the strong variability Bonsack.W.K.. Pilachowski.C. A.. Wolff.S.C. 1974. Astmpins. .). in ihc r band (relative to the u band) which occurs in a 187. 265 number of cool Ap stars [e.g. HD 24712 (Wolff and Gronbcch.B.. Olsen.F.H.. Slromgren.B. 1975. private ciiniimiiiic;i- Morrison. 1973), 49 Cam (Bonsack el a/., 1974), tion HD 71866 (Wolff and Wolff. 1971), HD9808S (Maitzcn, van den HeuvcI.E. P.J. 1971. Astrim. & Amroplirs. II. 4M Jones, T.J., Wolff. S.C. 1973. Publ. Aaron. Sue. Pacific 85. 760 1973), HD 119213 (Wolff and Morrison, 1974). HD Kodaira.K. 197.1. Asinm. & Asirophys. 26. 385 125248 (Wolff and Wolff, 1971; Maitzen and Moffat, Lafler.J., Kinman.T.D. 1965, Aslrophrs. J. Suppl. II. 216 1972) and HD 188041 (Wolff and Wolff, 1971)]. Blanket- Lcckrone,D.S. 1974, Astrophys. J. 190. 319 ing measurements by Pilachowski and Bonsack (1975) Leek rone, D.S.. Fowler.J.W.. Adclman.S.J. 1974. Astrmi. A Astm- phys. 32. 237 show that local line blanketing in the v band is not Lindcmann.E., Hauck.B. 1973. Asmm. & Astrophys. SuppL II. 11^ responsible for this strong variability in the case of Maitzen.H.M. 1973. Aslron. & Astrophys. Slippl. II. 327 HD 125248. Maitzen.H.M.. Rakosch.K.D. 1970. Am run. & Asirnphy*. 7. Hi Our conclusive remark is that ultraviolet observations Maitzen.H.M.. Moffat.A.KJ. 1972. Amron. & Asirophys. 16. 385 Megcssier.C. 1971. Asiron. & Aaroplm. 10. 332 of silicon stars add substantial new information. Molnar.M.R. 1973. Aamphys. J. 179. 527 HD63401 or HD92664 deserve further attention, Pilachowski.C.A.. Bonsack.W.K. 1975. Hubl. Astmn. Sue. I'MIIH- because they are expected to be very suitable for 87,221 testing the temperature dependence of the null region by Preston.G.W.. Slepien.K. 1968, Aslropliys. J. 151. 583 far ultraviolet photometry. Rakos, K. D. 1962. Uneell Obx. Bull. 5. 227 Stepien.K. 1968, Aslrophys. J. 154, 945 Wolff.S.C. 197.1. Astrophys. J. 186. 951 Wolff.S.C. 1975. Aslrophvs. J. 202, 127 Aiknowh'iUivnu-iUH. We are grateful to Dr. Olsen for sending us a Wolff.S.C.. Morrison.N.D. 1973. PIIM. Astrim. Sor. Padlk 85. 141 preliminary version of the uifcv-catalogue of bright southern stars Wolff.S.C., Morrison.N.D. 1975. ft/A/. Asinm. Soc. Pucific 87. 231 (Copenhagen University Observatory. 1974). WolH,S.C, Wolff.R.J. 1971. Antrim. J. 76, 422 This research has been supported in part by the National Foundation of Collective Fundamental Research of Belgium (FKFO) under H. Hensberge n 10.10.1. C. De Loore Astrophysical Institute Vrijc Universiteit Brussel A. Buyllaan 105 B-1050 Brussel, Belgium Now Aililnl in I'nmf. Recently Wolff (1975) published data of HD 81009 [comparison Mar HD 80447); she proposed a period of 69d or 34''5. E.). Zuiderwijk Her dal.i reveal that we measured the star around minimum G. Hammerschlag-Hensberge brightness. Combination of her data with the observations of Gron- Astronomical Institute hech and O'-sen at JD 2441324.77 and JD 2441368.68 (the last one University of Amsterdam also iwai minimum brightness) and with our data favorises strongly Roetersstraat 15 the possibility of the shorter period. All data are well represented with Amsterdam a period of .14"! 1+0'.'2. The Netherlands 155 ASTRONOMY AMnm. Astrophy-.. 54, 443—449 (1977) AND ASTROPHYSICS Photometry of Silicon Stars H. Hensberge1, C. De Loore1, E. J. Zuiderwijk2 and G. Hammerschlag-Hensberge2 ' Asirophysical Institute, Vrije Universiteit Brussel, Pleinlaan 2, B-1050 Brussels, Belgium European Southern Observatory, La Silla, Chile 2 Asironomical Institute, University of Amsterdam, Roetersstraat 15, Amsterdam, The Netherlands European Southern Observatory, La Silla, Chile Received July 19, 1976 Summary. Five of the brighter silicon stars in the list photometer for simultaneous measurements in the of Bidelman and MacConnell (1973) i.e. HD3580, Stromgren uvby system. The photometer is used in HD 187473, HD 206653, HD 207188 and HD 212432, combination with a photon counting system and is were observed at the ESO, La Silla, to search for described in detail by Gronbech et al. (1976). variability in the Muby-photometric system. All of them Each observation sequence contained measure- turned out to be variable in the four channels showing ments of the program star (P) and three comparison obviously their largest amplitude in u (>0T2 in the stars (Cl, C2, C3) as follows: C1-P-C2-P-C3-P-C3-P- case of HD 187473) and periods could be derived for C2-P-C1. Extinction corrections were computed in the ail of them. One of the comparison stars, i.e. HD 185183. usual way. The observations of two comparison stars also turned out to be variable. Colour indices, amplitude, were reduced to the magnitude of the third one by using ch;i racier and periodicity of the variations and a the mean difference in magnitude over the entire 12 A/mm spectrogram suggest that this star might be a observation run. The dispersion of these differential broad-lined silicon star. magnitudes around their mean was low enough to permit this procedure. The differential magnitude for Key words: Ap-stars — photometry — Si-stars the program star was obtained as the average of all measurements of two consecutive sequences. The stan- dard deviation of this differential magnitude is practi- cally always less than 0T005. Standard indices were computed by transformation I. Introduction to the uvby system of the catalogue of bright southern stars (Grenbech et al., 1975). The transformation In continuation of our search for periodicities among formulae appear to reproduce the catalogue indices peculiar A stars (Hensberge et al., 1976) we present in with errors smaller than 0T01. However, the majority this paper four-colour uvby photometry of five southern of program and comparison stars have not yet been silicon stars of the list of Bidelman and MacConnell measured in the uvby system. (1973): HD358O, HD 187473, HD206653, HD207188 and HD 212432. The stars were selected because of The period of variation is determined by a method their brightness. One of the comparison stars, HD based on the technique used by Lafler and Kinman 185183, turned out to be a variable peculiar star itself. (1965). For a sequence of trial periods, the phase 0, In §2 we give a short description of the observations of each measurement xt is calculated. Assuming that and reduction methods. The results for each program the Xj are arranged according to their phase, S is defined star separately are discussed in §3. In § 4 our conclu- N sions are summarised. I 2. Observations and Reductions The observations were made by E. Zuiderwijk with the 50 cm Danish telescope at La Silla between July 25 N is the number of observations; for xs+l the .v, value and August 23, 1975. This telescope is equipped with a has to be adopted and x is the average of the .vn's. Obviously S is a function of the guess periods since the Send off prim requests lo: H. Hensberge arrangement according to phase depends on the adopted 157 H. H^nsbcrge el al.: Photometry of Silicon Star* Table 1. Sumniiiry of results. Consecutive columns denote: Ml) number, silicon or comparison star, number of observing nights ;UKI number of differential magnitudes, visual brightness denoted by (I) since transformation from r lo |L'| is somewhat leiUalive for Ap-slars because of the complicated variability in this wavelength region , Stromgren indices b v. m\ and <'I and, for the variable stars, the period of variation HD Type «(N) [V) b-y ml fl P(days) number 35K0 Si 11(19) 6.64/6.63 -0.065 0.119 0.53/0.50 1.480 ±0.005 or 2.94 ±0.02 .15X1 comp. 7.104 0.267 0.162 0.491 4247 romp, 5.233 0.226 0.142 0.554 -- : 4622 comp. 5.586 -0.029 0.131 0.975 187473 Si 15(21) 7.36/7.23 -0.04/0.00 0.15/0.18 0.73/0.55 4.75 ±0.04 1K5183 comp. 6.75/6.71 -0.03/-0.05 0.09/0.12 0.56/0.53 -1.737 var. 187578 comp. - 7.666 0.032 0.096 0.874 190285 comp: 7.239 0.053 0.147 1.089 206653 Si 13(16) 7.24/7.18 -0.04/-0.02 0.135 0.63/0.50 1.788 + 0.005 205348 comp. 6.775 -0.043 0.112 0.764 205417 comp. 6.226 0.008 0.151 1.133 - 209468 comp. \ 7.543 0.011 0.174 1.009 207188 Si 11(15) 7.65/7.60 -0.041 0.152 0.54/0.58 2.67 + 0.01 205705 comp. 7.446 -0.033 0.101 0.699 207439 comp. - 7.559 0.159 0.189 0.769 208482 comp. 6.757 0.035 0.182 1.059 E 212432 Si 9(12) 7.52/7.49 -0.05/-0.06 0.248 0.69/0.63 4.69 + 0.05 210739 comp. 6.197 0.091 0.193 0.920 210931 comp. - 7.401 0.014 0.184 1.005 214172 comp. 7.421 0.058 0.191 0.960 — trial period. For the true period of variation, S is ex- curve never exceeds the probable error on the observa- pected to be near its minimum value for each of the tions by a considerable amount. Terms in S, arising filters. Generally, when S<0.5 for each of the filters, from phase points which differ more than Q are omitted. the correlation is good (in the case Nx20); in excellent We compute cases S*0.2. As a rule, only for a very small number of trial periods S does not exceed 1. N This technique was used by Lafler and Kinman (1965) for the case of Cepheids. For that special case the S*=- shape of the light curve is well known and the number N* I (xn-x)* of observations ranges roughly between 30 and 50. n=l When the number of observations is smaller, i.e. less where than 20, the sum S may contain a term arising from the comparison of observations with rather large phase =\ if fm+l-f.£Q differences. Moreover, when the light curve is es- =0 if f -L>Q sentially a double wave curve, it may be expected that n+l it contains none or at least very few observations be- and N* is the number of <5's differing from zero. tween succeeding extrema. In this extreme case, it is This S*, depending on Q, is assumed to be a better exactly this term which may produce the major con- indicator for the period of the Ap stars than S. If Q is tribution to S and which might lead to the rejection overestimated double-wave periods may be overlooked of the trial period (although the observations are not (in fact g=l in the original version). If Q is under- in contradiction with this -period). estimated, the test is not selective enough. This latter To avoid this difficulty we decided to add an extra fact imposes a lower limit on Q in the case of large parameter Q, the phase difference for which the in- amplitude. Q=0.1 seems to be a reasonable choice in crease in magnitude in the case of a double-wave light most cases. 158 H. Uensberge et al.: Photometry of Silicon Stars 445 Table 2. Differential magnitudes after transformation to the uvhv Table 2. system of the catalogue of bright southern stars (Granbech et al.) in - l.v Ah Av Au JD = •1.V Ah Ar In 24426M1+ 2442600+ HD 3580-HD 4622 HD 187473- HD 187578 I9K90 .150 1.115 1.064 0.563 40.510 -0.420 -0.467 -0.436 -0.722 24.S83 1.149 1.114 1.065 0.554 40.747 -0.409 -0.474 -0.435 -0.68K 34.815 1.154 1.119 1.066 0.569 41.502 -0.345 -0.415 -0.414 -0.569 37.732 .153 1.117 1.065 0.564 41.734 -0.332 -0.408 -0.398 -0.566 37.884 .155 1.124 1.071 0.571 42.691 -0.335 -0.403 -0.397 - 0.596 38.741 .146 1.110 1.061 0.551 45.551 -0.401 -0.457 -0.437 -0.677 38.XH8 .146 1.112 1.063 0.559 46.741 -0.334 -0.403 -0.407 -0.590 39.695 .153 1.116 1.063 0.552 47.703 -0.324 -0.393 -0.386 -0.572 39.883 .139 1.108 1.052 0.532 411.697 .153 1.120 1.068 0.569 HD 206653 -HD 205348 40.886 .155 1.121 1.067 0.572 19.687 0.404 0.423 0.465 0.243 41.688 .145 1.107 1.060 0.551 23.818 0.451 0.454 0.479 0.346 4I.S75 .148 1.113 1.063 0.560 24.796 0.407 0.416 0.459 0.24S 42.718 .146 1.113 1.060 0.542 35.623 0.405 0.421 0.461 0.240 42.S8SJ .143 1.109 1.054 0.532 37.758 0.418 0.430 0.467 0.271 46.702 .155 1.120 1.066 0.568 38.765 0.438 0.442 0.471 0.325 46.S77 .155 1.121 1.071 0.571 39.583 0.427 0.434 0.470 0.287 47.673 .148 1.111 1.064 0.558 39.835 0.443 0.449 0.477 0.334 47.867 .148 1.113 1.066 0.561 40.720 0.417 0.425 0.458 (1.273 41.581 0.437 0.443 0.474 0.325 HD 185183-HD 190285 41.829 0.455 0.460 0.483 0.365 19.578 -0.525 -0.618 -0.747 -1.415 42.740 0.404 0.418 0.458 0.24(1 19.795 -0.504 -0.593 -0.731 -1.396 43.569 0.449 0.453 0.477 0.354 21.573 -0.504 -0.589 -0.718 -1.383 46.807 0.425 0.439 0.468 0.297 23.583 -0.479 -0.562 -0.697 -1.364 47.558 0.449 0.448 0.475 0.345 24.554 -0.531 -0.621 -0.757 -1.437 47.812 0.425 0.435 0.467 0.297 24.770 -0.524 -0.619 -0.753 -1.419 34.608 -0.521 -0.611 -0.743 -1.427 HD 207188-HD 205705 34.792 -0.530 -0.629 -0.752 -1.434 19.827 0.153 0.142 0.187 0.068 35.711 -0.488 -0.567 -0.703 -1.363 23.850 0.179 0.170 0.210 0.127 37.709 -0.494 -0.582 -0.709 -1.378 24.687 0.189 0.180 0.222 0.1 IS 38.716 -0.525 -0.616 -0.749 -1.418 24.862 0.176 0.167 0.211 0.097 39.508 -0.503 -0.589 -0.724 -1.394 34.841 0.189 0.176 0.218 (U27 39.751 -0.530 -0.619 -0.748 -1.432 35.676 0.159 0.152 0.197 0.080 40.510 -0.507 -0.605 -0.734 -1.393 37.808 0.195 0.189 0.226 0.136 40.747 -0.49X -0.583 -0.712 -1.379 38.642 0.149 0.144 0.190 0.07(1 41.502 -0.515 -0.607 -0.737 -1.418 38.827 0.164 0.156 0.199 0.0K5 41.734 -0.518 -0.618 -0.742 -1.418 • 39.723 0.186 0.178 0.222 0.135 -0.707 42.691 -0.502 -0.570 -1.364 40.589 0.196 0.187 O.22S 0.129 45.551 -0.523 -0.621 -0.752 -1.424 40.821 0.179 0.173 0.215 0.104 46.741 -0.513 -0.608 -0.741 -1.417 41.712 0.193 0.187 0.229 0.118 47.703 -0.489 -0.579 -0.706 -1.373 42.567 0.180 0.173 0.215 0.126 42.804 0.184 0.177 0.217 0.125 HD 187473- HD 187578 HD212432-HD21O931 19.578 -0.320 -0.383 -0.382 -0.557 19.795 -0.310 -0.376 -0.382 -0.539 19.861 0.100 0.030 -0.107 -0.601 21.573 -0.425 -0.466 -0.435 -0.732 24.839 0.104 0.036 -0.100 -0.5X0 23.583 -0.330 -0.404 -0.399 -0.598 34.869 0.113 0.053 -0.085 -0.541 24.554 -0.317 -0.379 -0.381 -0.541 37.682 0.102 0.025 -0.108 -0.597 24.770 -0.315 -0.376 -0.382 -0.539 37.860 0.094 0.025 -0.111 -0.601 34.608 -0.343 -0.398 -0.390 -0.583 38.798 0.102 0.033 -0.099 -O.5S5 34.792 -0.371 -0.422 -0.409 . -0.625 39.808 0.114 0.054 -O.0K2 -0.542 35.711 -0.420 -0.461 -0.432 -0.728 40.618 0.096 0.026 -0.102 -0.576 37.709 -0.338 -0.406 -0.399 -0.602 40.846 0.096 0.026 -0.103 -0.577 38.716 -0.312 -0.377 -0.379 -0.535 41.801 0.103 0.037 -0.097 -0.563 39.508 -0.361 -0.413 -0.403 -0.620 42.596 0.093 0.024 -0.110 -0.600 39.751 -0.395 -0.442 -0.414 -0.673 42.841 0.091 0.019 -0.112 -0.615 159 H. llcnsberge e; al.: Photometry of Silicon SUI 1 1 ; Let us focus on HD358O and HD212432. hi tin- 2 A .6 .8 .0 ,J following section it is argued that the period of HD f 3580 is either P, = I'MS or P2 = 2i94. Inspection of ihc j- light curves for each of the two cases does not lead to a final choice. The second period induces a double- .v. J r wave light curve. The S-tesl prefers P,: S(/ ,) t().3^ 1 while S{P2)>0.6 in all filters. The S*-tesi uives '• S*(PI)vS*{Pi)»S(Pl). The case of HD212432 is more illustrative. Ac- cording to the S-test P, = l?65 and P2 = 2?52 are the best candidates. However S>0.6 in both cases in all filters. The plots of the observations for periods /J, or P2 are not convincing. The S*-test prefers P, to P,: V '" the u-variation, which has the largest amplitude and hence should be the best indicator of the period, gives I 1 no satisfying result (S*%0.6). However P = 4^7 happens to be an excellent candidate with S*<0.15 in each of i. •••fa the four filters. This candidate could easily be dis- ,,, ** carded in the S-test, since S(P3)*1 as a consequence of the large phase-gaps containing no observe lions in this double-wave light curve. In these S*-tssts a Q ,-:IIR of 0.1 was adopted. Therefore it may be concluded that the S-trst -••y • favours the selection of single-wave light curves. How- ever, as many Ap stars are known to possess light curves with a secondary extremum, the S*-test is more Jo, 0 .2 4 6 8 .0 .2 selective in the case of period determination for these | 1 >*aw, stars, especially when only a restricted number of Fig. 1. Photometric variations of HD3580 in the instrumental system of the Danish SO cm telescope at La Sitla, plotted with respect observations is available. to a period of l?48 and phase zero at JD2442619.890 1 1 1 .0 2 4 6 8 0 .2 .0 .2 4 6 .8 .0 .2 . • 1 1 ab "a S 'I: '• .0 .2 4 6 .8 .0 .2 1 1 1 1 • a To2 0 .2 .4 G .8 X) .2 Fig. 2. Photometric variations of HD 187473. Phas> = 4'.'75 1 , and phase zero at JD 2 442 619.578 160 H. Hensbcrge ct al.: Photometry of Silicon Stars 447 1 1 a 2 4 .6 8 0 .2 AU r '• r• c A • • 1 1 1 A V - • 1 r Ab i ** - i L • - - - j 1 1 1 Ay A ill A • • & A :y 4> A. 02 I .0 2 .4 .6 .8 JO , .2 .0 .2 .4 .6 .8 .0 .2 iphase , 1 1 Fig. 3. Photometric variations of HD 206653, plotted in a period Fig. 4. Photometric variations of HD 207188. Phases were computed P= I?788. Phase zero corresponds to JD 2 442 619.687 with P=2?67 and phase zero at JD 2442 619.827 3. Results are in phase with each other. Although the variations A summary of our results, including standard indices, are rather small, the lightcurves are well defined. The amplitude and periodicity of variations, is listed in b—y and ml indices are intrinsically constant. Table 1. The standard indices enable us to estimate However, also P=2?94 represents the observations the effective temperature of the silicon stars involved, equally well. In this case, all curves show secondary since u - b is a relevant temperature indicator (Megessier, extrema. Unfortunately, the curves with this period 1971). Stars showing the silicon anomaly generally are not as well defined as with P=l?48 because there have 1.3>«-b>0.45 corresponding to effective tem- remain three gaps of 0.2 in phase where no observa- peratures between llOOOK and 18000 K. Our program tions exist. This is, however, not a sufficient reason for stars have u -b as 0.62 (HD3580), 0.77 (HD 206653), eliminating this longer period. 0.78 (HD 207188), 0.93 (HD 187473) and 1.05 (HD According to the accuracy of the measurements— 212432). This places HD 3580 at the same temperature demonstrated by the accuracy of differential magnitudes as 41 Tau and somewhat hotter than Babcock's star between comparison stars^—we can place the following HD 215441. limits on the value of P: l?480+0?005 or 2?94+0?02. In the subsequent section we discuss the results B. HD 187473 for each of the variable stars separately. One should notice that the plots represent the variations in the Since the comparison star HD 185183 proved to be instrumental system. Phase zero is chosen arbitrarily variable, only two comparison stars were used. HD at the time of the first observation. The transformed 187473 is variable with extra-ordinary amplitude: differential magnitudes are given in Table 2. about 0?l in y and b, 0W7 in v and more than 0".'2 in u. Weak secondary extrema show up in u; they are less A. HD3580 pronounced in the other channels. Maximum brightness The observations are plotted in Figure 1 j phases are com- in y corresponds to a maximum in b—y, which shows puted with respect to a period of 1?48. All light curves no secondary extrema anymore. The variation in ml is 161 H. Ilensbergect al.: 1'hoiomctry of Silicon Stars 1 .0 .2 .4 6 .R 0 .2 I- AU I" \ r ... a « 1 1" . - -a I I cb - ... • 1 Ay - Jo2 .0 .2 .4 .6 .8 0 .2 1 Fig. S. Photometric variations of HD 212432. Phases were computed Fig. 6. HD 185183 proved to be variable in the uiby system. The with P=4?69and phase zero at JD 2 442 619.861 observations are rjlotted with respect to a period P= 1?737 and phase zeroatJD2442619.578 less clear. All curves are in phase with the period but they are scarcely defined by lack of observations /»=4?75±0?04(Fig.2). between phases 0.25 and 0.65—except two points near phase 0.45. Again, ml is constant and the b—y variation C. HD 206653 is hardly detectable. Further observations are required The observations are well represented with P= to confirm this first set. However no alternative value 1?788±0?005 (Fig. 3). All curves are well defined single of P is consistent with our observations within the waves in phase with each other; b—y varies slightly, accuracy limits set by differential magnitudes between ml remains constant. comparison stars. This star is rather at the cool end of stars with O. HD 207188 silicon anomaly. Consequently, the importance of the Figure 4 shows the results plotted in the period P=2?67. contribution of line blocking due to metals is probably Secondary cxtrema occur in the four channels, but less growing, resulting in a much larger ml index. pronounced in u; b—y and ml are constant. Obviously observations are lacking around primary minimum. F. HD 185183 The frequent observations which define the slope This star was selected as comparison star of HD 187473. between phases 0.7 and 0.0 permit to rely on the period The reductions showed that this star is obviously up to ±0?0l. variable in the four channels within a range of 0™08. The star is classified in the Henry Draper catalogue as £. HD 212432 B9. The effective temperature derived from the u-b Differential magnitudes are plotted in Figure 5. Phases value of 0.68 is about 14500 K; its mean colour indices are computed with P=4?69. The accuracy of P is and the amplitude of the variations, as well as its 7^rr ±0?05. Pronounced secondary extrema should exist are similar to those of silicon stars. We therefore 162 11. Hensberge et al.: Photometry of Silicon Stars 449 decided to search for a periodicity in the order of days. silicon stars only with those of HD 215441 (Babcock's The best correlation was found near P= 1?737. The star). It should be noticed that at least three and possibly dispersion around the mean curves displayed in Figure 6 four stars show more or less pronounced secondary is larger, however, than in other diagrams, above all extrema. in y. This can be partially due to the fact that each Acknowledgements. H. Hensberge acknowledges support by the comparison star was observed less frequently than a National Foundation of Collective r'undamental Research of Belgium program star as appears from Section 2. So one blue (FKFO) under No. 10303. 12 A/mm spectrogram was taken in May 1976 by E. J. Zuiderwijk acknowledges support by the Netherlands De Loore at the ESO. The star appears to have relatively Organisation for the Advancement of Pure Research (ZWO). strong but broad silicon lines. We therefore suggest that this might be a broad-lined silicon star. References Bidelman,W.P., MacConnell.D.J.: 1973, Astron. J. 78. 687 Granbech,B, Olsen.E.H., Stromgren.B.: 1975. private communica- 5. Conclusions tion Granbech,B., Olsen.E.H., Stromgren.B.: 1976, Astron. Astrophys.. All five silicon stars are variable. Variations in the four to be published Hensberge,H., de Loore,C Zuiderwijk,E.J., Harnmerschlag-Hens- channels are in phase; the largest amplitude occurs berge.G.: 1976, Astron. Astrophys. 48, 383 in all cases in u. One of the stars, HD 187473, shows Lafler.J., Kinman,T.D.: 1965, Astrophys. J. Suppl. !1, 216 extra-ordinary large amplitudes, comparable among Megessier.C: 1971, Astron. Astrophys. 10, 332 163 Ultraviolet Photometric Observations of Ap and Am Stars W. van Dijk , A. Kerssies , G. Hammerschlag-Hensberge and 2 P.R. Wesselius Astronomical Institute, University of Amsterdam, Roetersstraat 15, 1004 Amsterdam, the Netherlands 2 Kapteyn Astronomical Institute, Department of Space Research, University of Groningen, Postbus 800, Groningen, the Netherlands running title: UV photometry of Ap and Am stars publication in Astronomy and Astrophysics main journal 165 -2- Summary. ANS five colour ultraviolet photometric observations of 79 hot and cool Ap stars and 26 Am stars are presented. The positions of Ap and Am stars in ultraviolet colour-colour diagrams differ from those of normal main-sequence stars of the same spectral type. The deviation is largest for the cooler Ap stars. The ultraviolet flux deficiency known to exist for the hot (early type) Ap stars appears to decrease towards later spectral types and is absent for the Am stars. The light variable HR 5857 shows ultraviolet light variations of which the amplitude gradually decreases towards longer wavelengths. Key words: peculiar A stars - Am stars - photometry - ultraviolet 166 -3- 1. Introduction A-, B- and F-type stars which show anomalies in the strengths of spectral lines of various elements have been classified as peculiar (Ap) and metallic-line (Am) stars. The Ap stars are, according to their type of spectral anomaly divided into the following subclasses: (i) Si and Sr-Cr-Eu stars; these have, in general, magnetic fields of considerable strength; (ii) the non-magnetic Hg-Mn stars and (iii) the He-weak stars. The Am stars are characterized by too weak Ca II lines for their hydrogen spectral type and a slight overabundance of the iron peak elements. For an up to date review of further properties of Ap and Am stars we refer to Preston (1974). The study of peculiar A and B stars in the ultraviolet spectral region (< 3000 5) began in the 1970's with the Wisconsin Experiment Package onboard OAC-2. So far, several Ap stars have been observed photometrically in the ultraviolet and of some of them spectrum scans at higher resolution are available for certain ultraviolet wavelength bands, e.g. the ones obtained with the S-59 spectrograph onboard ESRO's TD-1A satellite (cf. Lamers et al., 1973; Faraggiana et al., 1975). For a review of ultraviolet observations of Ap stars until 1975 we refer to Leckrone (1976) and references therein. Most of the Ap stars have too blue UBV colours for their MK-spectral type. The study by Leckrone (1973) of 24 blue Ap stars (most of them of the Si or Hg-Mn type) revealed that these stars are deficient in ultraviolet flux compared to normal stars with similar UBV colours. The UV-deficiencies are smaller for the Hg-Mn stars than for the Si stars. The ultraviolet flux distributions seem to be consistent with the published MK spectral types. A plausible reason suggested for the abnormal flux distributions of Ap stars is enhanced absorption in ultraviolet lines and continua (Leckrone et al., 1974), which removes flux from the UV and redistributes it through backwarming into the visible part of the spectrum. Leckrone et al.'s numerical calculations for heavily line-blanketed model atmospheres reproduce very well the observed flux distribution of Ap stars. In this study we discuss ultraviolet observations of 79 Ap stars and 26 Am stars carried out with the Astronomical Netherlands Satellite (ANS). We included cooler Ap stars (Eu-Cr-Sr type) and Am stars in our study, in 167 -4- order to examine whether the results obtained by Leckrone (1973) for the blue Ap stars can be extended to the other types. Cameron (1966) showed that Ap stars with surface magnetic fields larger than a few thousand gauss always have an abnormally large stromgren metal index JBI, and vice versa. The ml-index is a measure of the strength of the integrated line absorption by metals in the violet band, relative to the absorption in the much less affected blue and yellow bands. The abnormally large metal indices for the peculiar stars are most probably associated with line absorption effects or could be due in some cases to the continuous absorption features near A5300 and X4200 8. The correlation between ml and magnetic field strength can be of great importance for the discovery of stars with large magnetic fields. We included in our observing list a number of Ap stars with large magnetic fields (we used the list of Babcock, 1967) and also a number with small fields, to study possible effects on the ultraviolet narrow band colour indices. We intended to examine whether a correlation exists between the behaviour of the ultraviolet colours, the magnetic field strength and -possibly- the abundance anomalies (e.g. blanketing effects). With this purpose in mind, we selected most of our Ap and Am stars from Cameron's list (1966). On recommendation of Dr. S.J. Adelman (private communication) some other interesting Ap stars were also included. In addition some normal stars of the list of Cameron were observed, to compare their flux distributions with those of the Ap and Am stars. 2. Observations and Reductions The University of Groningen experiment onboard the ANS consists of a 22 cm aperture Cassegrain telescope and a five channel ultraviolet photometer with central wavelengths at 1550, 1800, 2200, 2500 and 3300 8 and bandwidths of 150, 150, 200, 150 and 100 X, respectively. The response functions are almost rectangular. The instrument was described in detail by van Duxnen et al. (1975). Tables 1 «»d 2 list the observed peculiar A stars and Am stars, respec- tively. The observations were made during the period September 1974 to April 1976. Thfe peculiarity classes in table 1 and the visual magnitudes 168 -5- in tables 1 and 2 were in most cases taken from the catalogue of Bertaud and Floquet (1974). For a few stars which were missing in the catalogue, the values were taken from Cameron (1966). Columns 4 and 5 of table 1 list the maximum absolute values of the effective (He) and surface (Hs) magnetic field, respectively, as found in the literature. The references used are given in the last column of table 1; the values for He in table 2 were all taken from Babcock (1958). The question marks in column 4 of table 1 and in column 3 of table 2 denote stars which could be magnetic as well but have too broad lines in their spectra to measure a magnetic field. Columns 7-11 of table 1 and columns 5-9 of table 2 give the observed mean magnitudes of the five UV-channels. The observed count rates were first corrected for the instrumental sensitivity change (Aalders, 1976). Afterwards, the magnitudes have been calibrated according to the preliminary absolute calibration of the instrument given by Wesselius (1975) and using the Vega calibration of 3.64 x 10 erg cm s 8 for visual magnitude V = 0.00 (Oke and Schild, 1970). Reddening corrections have been made according to the following procedure: we made plots of log F. and corrected for possible humps at 2200 A (the interstellar extinction is strongly peaked at 2200 8), assuming the curve to be linear between 1800 and 2500 A. The latter assumption is certainly not valid for stars with spectral types later than Al. For these stars we derived the spectral types for all channels from the difference with the 3300-channel. If the 2200-channel did give a later spectral type than the three other channels, we made corrections until this deviation disappeared. The reddening corrections turned out to be zero for all observed stars, except HR 7230 and HR 2362 which have E(B-V)-values of 0.038 and 0.172, respectively. Column 12 of table 1 and column 10 of table 2 give the number of observations of each object. The stars were observed in normal pointing mode or in offset pointing mode. In the normal pointing mode, dark current measurements were taken before and after the stellar measurement. In the offset pointing mode, on the other hand, each star measurement alternates with a sky background measurement. The difference in count rates between the two modes is negligible (cf. de Boer and Koornneef, 1975). In the calculation 2 of the mean magnitude of a set of observations from one star we used I/a as weighting factor, where a is the mean error of one observation. The mean errors calculated from the values obtained in the different integrations 169 -6- of one object, were compared with the errors expected from photon statistics. Whenever the mean error of the mean magnitude was more than a factor of two larger than the expected statistical error in the five channels, and -at the same time- the mean errors of the single observations were less than a factor of two larger than the expected statistical errors, we placed the remark "var" in column 13 of table 1, indicating that the star is probably variable. In the cases that this variability was not found in all channels, we also placed a question mark in the table. Only in one case (HR 5857) we were able to plot a UV-light curve, because we had 15 observations and the star has a relatively short period of 1.3049 days (Winzer, 1974). 3. Results Examples of colour-colour diagrams are shown in Figs. 1-3. An explanation of the symbols is given in the capture of figure 1. The symbols are used in three different sizes: the largest ones denote stars with a magnetic field stronger than 1000 gauss, the middle ones denote stars with magnetic fields less than 1000 gauss and the smallest ones denote stars with too broad spectral lines to measure a magnetic field strength. The full line in the figures represents the main sequence for normal stars derived from ANS UV-measurements of a large sample of normal stars (B2 - F0) which were averaged per spectral type (Wu and Wesselius, 1976). It is easy to see that the Ap stars as well as the Am stars have ultraviolet colours (see figures 1 and 2) which deviate from those of normal stars of the same spectral type. The deviation is largest for the late type Eu-Cr-Sr stars. The Hg-Mn stars show little or no deviation. In this respect, our results in the ultraviolet are comparable with those of Cameron (1966) for the ml-index, viz.: there is a tendency for stars with large fields to have a larger deviation. To examine the flux distribution of Ap and Am stars in the ultraviolet more thoroughly, we plotted one ultraviolet colour against b-y (see figure 3) and we compared the spectral types derived from B-V colours and the MK types (deduced from the strength of the Ca II K line) with the spectral types that we derived from our ultraviolet measurements. We compared for this purpose the ultraviolet flux distribution of the Ap and Am stars with the flux distribution of normal main-sequence stars . 170 which were also observed with the ANS. The results are given in tables 3 and 4. References from which the B-V values and the MK classes were taken are given in the last column of the tables. The B-V spectral types were obtained by using the B-V vs. spectral type calibration given by Deutschmann et al. (1976). The result found earlier by Leckrone (1973) that the hot Ap stars have lower UV-fluxes than expected from the spectral types derived from their B-V colours is confirmed by our measurements, although for spectral types up to A2 the calibration by Deutschmann et al. (1976) yields types about one subclass later than the one of Johnson (1963), which was used by Leckrone. Remarkably, the UV-flux deficiency is no longer present for the later type Ap stars and Am stars. For the late-type Ap stars the B-V type and UV type become about the same as the MK type. For the Am stars the B-V class is the same as the UV class, but the MK class is earlier; this is probably due to the fact that the Am stars have a too weak Ca II K line compared to normal stars of the same spectral type and the MK classification is largely based upon this line. Indeed, the hydrogen classification of Am stars is later than the Ca II K classification. The above mentioned results are illustrated in figures 4-6. The symbols are the same as those used in the earlier figures. The straight line indicates the position where the two classifications give the same results. 4. The light curve of HR 5857 The ultraviolet instrumentation of the ANS was not very suitable for the study of light curves of variable stars. This is due to the fact that an object could only be observed with ANS for a duration of 1.8 days/cos$, where 0 is the ecliptic latitude. The observation period was centered around the time when the object was ± 90 from the Sun in ecliptic longitude. Fortunately, one of the UV variable stars for which we had more than 10 observations, happened to have a photometric period of only 1.3049 days in the visual light (Winzer, 1974; Wehlau (1962) gives a period of l?3051). We plotted our measurements in the five UV channels against the phase in period. The light curves are shown in figure 7. Phase zero corresponds to JD2442424.20 + n x 1.3049. The light curve shows a double wave; the amplitude is largest for the A1550 channel and gradually decreases towards longer wavelengths. At A.3300 the amplitude becomes almost zero. 171 -8- For seven other Ap stars ultraviolet light curves have been obtained which show the following characteristics (cf. Leckrone, 1976): the variability in the ultraviolet is inversely correlated with the variability observed at wavelengths longward of 3500 8. These inversely correlated wavelength regions of variability are separated by a "null-wavelength region" where the star does not vary at all or the shape of its light curve becomes quite irregular. A commonly accepted hypothesis for this character of the variability is that shortward of the null wavelength region the flux removed by the increasing strength of absorption lines or of continuous absorption dominates any flux increase due to backwarming; longward of the null wavelength region increasing backwanning is not compensated by diminution of flux by enhanced blocking. Unfortunately, the very short period of HR 5857 and the long time span between the visual observations from which the light curve was derived (Wehlau, 1962) and our ultraviolet observations make it impossible to say with certainty that the light curves longward of A3300 are in anti-phase with our UV light curves. A study of the amplitude of variation for the stars which have the remark "var" in table 1 shows that 10 of the 15 stars have their null- wavelength region between A2500 and A3300. The five remaining stars (HD 25267, 140728, 173650, 196178, 196502) probably have null-wavelength regions around or longward of A3300. 5. Conclusions The ultraviolet observations of the Ap and the Am stars suggest the presence oi enhanced ultraviolet line and continuum opacity sources, especially for the hot Ap stars. Ultraviolet colour-colour diagrams show that the Ap stars - and 9a0cially the coolvr ones - deviate from the position of normal stars of the same spectral type. The deviation of the Ap stars in these diagrams seems to be the result of different grades of presence of the strong spectral lines in the different channels. We need a detailed spectroscopic investigation of line etrengths in the ultraviolet channels to check this possibility. The discrepancy seems also to be correlated with the 172 -9- strength of the magnetic field. On this basis, we suggest that the following stars - which have too broad stellar lines to measure -so far- the field strength - have a strong magnetic field: HD 4778, 140728, 170973, 183806. Comparison of figures 1 and 2 suggests that the strength of the anomalies differs from one channel to the other. This is confirmed by our UV classification of the stars. Most stars - especially the late type Ap stars - show a discrepancy of more than one subclass when we classify them separately according to their fluxes in the different UV channels. For the Si and Hg-Mn stars almost no discrepancy is present between the classes derived from different channels; also in the ultraviolet colour diagrams their deviations are smaller. The strength of the flux deficiency of Ap stars in the UV seems to be correlated with temperature: only the Si stars show clearly the effect in most cases. The mechanism proposed by Leckrone (1973; 1976) to explain the UV flux deficiencies seems to be applicable only for stars with B-V < 0 which have a strong magnetic field. Of the 24 blue Ap stars studied by Leckrone (1973) (which are almost all Si stars), 16 were also observed by us and the results for these stars are in very good agreement with Leckrone1s. Possibly, higher ionization degress of elements which appear for hotter stars - such as ions of the rare earths which have many lines in the UV for Si stars (cf. Swings, 1944; Adelman and Snijders, 1975) - cause the higher opacities in the UV for the blue Ap stars. This would explain why the UV flux deficiency disappears for the cooler Ap and Am stars. 173 -10- Acknowledgments The ANS project was sponsored by the Dutch Coumittee for Geophysics and Space Research of the Royal Netherlands Academy of Sciences. We acknowledge the cooperation of the other members of the ANS-UV team, Drs. J.W.G.Aalders, K.S. de Boer, R.J. van Duinen, D. Kester and C.C. Wu. We are indebted to Prof. E.P.J. van den Heuvel for his proposal to make UV observations of Ap and Am stars with the ANS and for his continuous interest during the course of this investigation. 174 -11- References Aalders,J.W.G. : 1976, R.O.G.No.76-2 Adelman,S.u., Snijders,M.A,J. : 1974, Publ.Astron.Soc.Pacific 86_, 1018 Babcock.H.W. : 1958, Astrophys.J.Suppl.Series 2_, 141 Babcock,H.W. : 1967, in The Magnetic and Related Stars, ed. R.C.Cameron (Baltimore, Mono Book Corp.) p.551 Bertaud,Ch., Floguet,M. : 1974, Astron.Astrophys.Suppl.Series ^6_, 71 Blanco,V.M., Demers,S., Douglass,G.G., Fitzgerald,M.P. : 1968, Publ. Naval Obs. 2j_, 1 de Boer,K.S., Koornr.eef, J. : 1975, R,,O.G.No.75-63 Bonsack,W.K. : 1976, Astrophys.J. 209, 160 Borra,E.F., Landstxeet,J.D. : 1975, Publ.Astron.Soc.Pacific 87, 961 Cameron,R.C. : 1966, Ph.D.Thesis, Georgetown Observatory Monograph no. 21 Cowley,A. : 1972, Astron.J. 77/ 750 Cowley,A., Cowley,C, Jaschek,M., Jaschek,C. : 1969, Astron.J. 74_, 375 Deutschmann,W.A., Davis,R.J., Schild,R.E. : 1976, Astrophys.J.Suppl. Series 30, 1 van Duinen,R.J., Aalders,J.W.G., Wesselius,P.R., Wildeman,K.J., Wu,C.C, Luinge,W., Snel,D. : 1975, Astron.Astrophys. _39/ 159 Eggen,O.J. : 1967, in The Magnetic and Related Stars, ed. R.C.Cameron (Baltimore, Mono Book Corp.) p.141 Faraggiana,R,, van der Hucht,K.A., Burger,M. : 1975, Astrophys.Space Sci. 38_, 243 Hensberge,H., de Loore,C. : 1974, Astron.Astrophys. 3_7_, 367 van den Heuvel,E.P.J. : 1971, Astron.Astrophys. 11_, 461 Jaschek,C, Conde,H., de Sierra,A.C. : 1964, Catalogue of Stellar Spectra Classified in the Morgan-Keenan System (La Plata Observatory) Jaschek,C., Fernandez,E., Sierra,A., Gerhart,A. : 1972, Catalogue of Stars Observed Photoelectrically, La Plata Publication 3E3_, 1 Johnson,H.L. : 1963, in Basic Astronomical Data, ed. K.Aa.Strand (Chicago: University of Chicago Press) p.204 Johnson,H.L., Mitchell,R.I., Iriarte,B., Wisniewski,w.Z. : 1966, Comm. Lunar and Planet. Lab. 4_, 99 Lamers,H.J., van der Hucht,K.A., Snijders,M.A.J., Sakhibullin,N. : 1973 Astron.Astrophys. 25, 105 175 -12- Landstreet,J.D., Borra,E.F., Angel,J.R.P., Illing,R.M.E. : 1975, Astrophys.J. 201, 624 Leckrone,D.S. : 1973, Astrophys.J. 185, 577 Leckrone,D.S. : 1976, in Physics of Ap Stars, I.A.U.Colloquium no.32, eds.W.W.Weiss, H.Jenkner, H.J.Wood, Vienna, p.465 Leckrone,D.S., Fowler,J.W., Adelman,S.J. : 1974, Astron.Astrophys. _32> 237 Mendoza V,E.E. : 1974, Revista Mexicana De Astronomia y Astrofisica 1_, 175 Oke,J.B., Schild,R.E. : 1970, Astrophys.J. 161^, 1015 Preston,G.W. : 1971a, Astrophys.J. _164_, 309 Preston,G.W. : 1971b, Astrophys.J. 175, 465 Preston,G.W. : 1974, Annual Rev.Astron.Astrophys. 12_, 257 Swings,P. : 1944, Astrophys.J. 100, 132 Wehlau,W. : 1962, Publ.Astron.Soc.Pacific 74, 286 Wesselius,P.R. : 1975, R.O.G.No.75-1 Winzer,J.E. : 1974, Ph.D.Thesis, Univ. of Toronto Wolff,S.C. : 1973, Astrophys.J. _186> 951 Wolff,S.C. : 1975, Astrophys.J. 202_, 127 Wu,C.C, Wesselius,P.R. : 1976, personal communication 176 -13- «» PJ •« < WNIAOO o u a in CM ujuj LulOV)LJC )O UJU O MSUJ V)l/f C1 (/) (OUUJOUCOXUV) W UWUO I I I I I I IJI II * Illft I i II « « a: a n? 1-4 M 3 Z> (J ^ Of MQfQ! HO K n a; •-< or or«-< ft: or ar OMOO u to w ujui x o o t/>oci in x v> ui to w oo o a: M Of tf S g W W CO Us ^ tjO eo nn ITS -T K W° MM«u»IPM 1 'Ho* Of 0.'W 1 O CO « ITU IO or < a* •* 2 y\ _j U> ,) ift ^ J ig ^ tn UJ »- a, «f*> o VA *4 W «*•* 3 f> IH J" 177 49 HER 15230* EU-CR ? 6.59 &.17 6.13 6.13 6.46 fc.4 9 1 (li 6T50 165474 S(? CR-EU 900 7200 7.45 10.03 7.69 7.73 7,75 7.10 1 ;;; t i at 6902 1£6469 SI CR-EU J 6.3S 5.94 5.90 6.23 6.43 6.39 1 (u 6370 168733 SR-TI 5.35 3.72 3.83 3.98 4.30 4.60 2 6932 170397 EU SI-CR 1 5.93 6.14 5.98 5.97 6.25 5.98 4 (IS 695 9 170973 SI SS-CR 1 6.42 6.33 6.22 6.21 6.47 6.23 4 (ii 46 OR A 173524 SI-HG 7 230 177517 HG-SI , 5.90 4.51 4.61 4.97 5.16 5.46 5 (i> 19 LYR 179527 SI 5.93 5.16 5.15 5.34 5.60 5.58 5 4 CTG 1*305 6 SI <500 5.16 3.62 3.72 3.as 4.23 4.54 7 (SI 7416 1 83306 S* CR-EU 1 5.87 5.85 5.b4 5.47 5.89 5.64 3 (1) K490S SI CR-SR 1 6.61 6.16 6.00 6.09 6.45 6.35 4 (1) 14 CV5 185V2 SI? 5.36 4.33 4.44 4.67 4.95 5.19 9 7552 1*7474 CR-EU 2000 5.30 5.18 4.98 4.83 5.22 5.13 6 7575 188041 SR CR-EU 1470 4000 5.64 8.13 6.81 6.15 6.43 5.93 4 (1) i(6) 192678 CR 2000 4600 7.37 9.04 8.01 7.50 8.04 7.41 7 (1) .(6) 7786 193 722 SI ? 6.18 5.61 5.55 5.71 5.96 6.07 4 VftR? (1) 7870 196178 SI •9 5.78 3.99 4.08 4.16 4.55 4.93 11 VAR'j (1) 73 ORA 196502 SR CR-EU 700 2000 5.19 7.10 6.19 5.74 6.13 5.56 9 VAR (1) 200 311 SI HG-MN ? 7.68 6.44 6.51 6.60 6.95 6.99 4 (1) THE1 HIC 203006 CR-SR EU-H& 650 4.81 5.32 5.09 4.76 5.24 4.82 2 VAR (1) 8216 204 411 CR-EU 7 <50Q 5.2a 7.87 6.15 S.88 6.19 5.77 6 (11 • <6) 8240 20508 7 SR-CR SI-EU 6.43 6.28 6.14 6.04 6.47 6.32 3 (It KAP PSC 220825 CR SR-EU <500 4.93 5.13 4.96 4.80 5.19 4.99 4 (5) 8913 221394 SR-CR SI-EU 6.41 7.05 6.66 6.53 6.87 6.58 2 108 AQR 223640 SI SR-CR ? 5.17 3.85 3.88 3.88 4.29 4.48 I (19 REFERENCES 1 (1) = BABCOCK (1958; 19671 (6) = PRESTOH (1971A) (2) = SON SACK (1976) (7) = PRESTON (19716) (3) = BO BRA AND LANDSTREET (1975) (8) ~ VAN OEMHEUVEL ( 1971> (4) = HENS8ERGE AND DE 1LOORE (;i9741 (9) = HOLFF (19 73) (SI a LANOSTREET ET. AL . (1975) (10) = HOLFF (1975> TABLE 2. METALLIC LINE A- AND F-TYP£ STARS NAME OR HD-NUM3ER HE V Ml 550 M1800 M220 0 M2500 N3300 N HR-NUM8ER 47 AND 8374 <500 5.53 12.42 7.49 6.89 7.15 6.07 1 loosa ? 7.69 14.47 9.95 9.31 9.54 8.38 2 16956 7.82 11.95 8.81 8.71 8.97 8.23 1 1139 23281 5.58 10.88 6.70 6.39 6.67 5.97 1 60 TAU 27628 •» 5.72 11.88 7.99 7.21 7.45 6.21 4 1403 28226 <500* 5.72 11.83 7.64 7.00 7.25 6.20 2 81 TAU 28546 <500 5.48 11.24 7.03 6.57 6.82 5.93 2 1519 30210 <500 5.37 10.07 6.70 6.44 6.69 5.86 6 16 ORI 33254 50tf 5.43 11.07 7.20 6.71 6.97 5.93 4 RR LYN «§469i • 5.80 10.55 6.93 6.60 . 6.83 6.01 7 45?33 ? 7.56 13.50 9.37 8.85 9.08 8. 03 2 Ul 50186 ? 7.36 13.04 9.51 8.84 9.13 7.90 4 I 2 UMA 72037 5* LAM VIR 125337 <».52 7.56 5.12 5.09 5.37 4.90 3 5702 1364Q3 6.14 10.68 7.56 7.34 7.60 6.76 6 153286 550 7.02 13.43 9.20 8.61 8.76 7.63 12 179143 <500 6.76 13.64 9.70 8.73 8.88 7.47 1 51 SGR 181*552 <500 5.66 10.07 6.84 6.65 6.91 6.13 2 190401 6.91 13.02 9.61 8.72 8.87 7.57 1 -16- Table 3. Spectral Classification of the Ap Stars. HD B-V UV MK ref number class class class 133880 B4 Al AO d,h 196178 B6 B9 B9 b,f 223640 B6 A0 B9 b,f 22470 B7 B9 B9 b,f 25267 B7 AO AO a,h 27309 B7 AO AO b,f 25823 B8 B9 B9 b,f 32650 B8 B9 B9 b,f 56455 B8 B8 AO d,h 66255 B8 B8 AO d,h 74521 B8 A2 Al b,f 75333 B8 B9 B9 a,f 90044 B8 A2 B9 a,f 122532 B8 Al AO d,h 133029 B8 Al B9 b,f 136933 B8 Al AO b,h 173524 B8 B9 B9. b,f 183056 B8 B9 B9 b,f 200311 B8 Al B9 e,h 68351 B9 Al B9 b,f 89822 B9 AO AO a,f 94660 B9 A2 AO d,h 140728 B9 A2 B9 b,f 143807 B9 B9 AO a,f 149822 B9 A3 B9 b,f 179527 B9 Al B9 bff 185872 B9 B9 B9 c,g 205087 B9 A2 B9 b,f b,h 10783 B9.5 B9 A2 b,f 39317 B9.5 Al B9 c,f 50204 B9.5 AO B9. c,f 120198 B9.5 A2 B9 c,f 137389 B9.5 AO AO c,f 152308 B9.5 Al B9. b,f B9 170397 B9.5 A3 a,f 170973 B9.5 A3 AO c,f 175744 5 AO B9. B9 b,h B9.5 B9 177517t B9. b,h 187474 B9. A2 5 AO c,f 193722 B9.5 AO B9 a,f Al 45827t A0 AO c,h 56022 A0 A2 AO a,f 72968 A0 A3 Al 180 -17- Table 3 (continued) 77350 A0 AO B9 a,h 125248 A0 Al AO b,f 151525 AO A2 B9 c,£ 166469 AO Al AO c,h 184905 AO Al AO b,h 192678 AO A5 A4 c,h 4778 Al A3 AO b,f 38104 Al A3 AO a,f 118022 Al A3 Al a,f 173650 Al A3 B9 b,f 203006 Al A3 A2 a,h 221394 Al A3 Al b,f 2453 A2 A7 A2 b,h 30466 A2 A5 AO c,h 55719 A2 A3 A3 a,h 108945 A2 A4 A3 a,f 220825 A2 A2 AO a,f 71866 A3 A5 AO b,h 151199 A3 A3 A2 a,f 196502 A3 A5 AO b,f 204411 A3 A4 A5 b,f 9996 A5 A2 B9 b,f 188041 A'7 "" A5 A5 b,h 24712 FO A8 A7 c,h 137909 FO A7 FO a,h 165474 FO A7 A7 d,h 176232 FO A8 FO b,f 137949 F2 A7 FO b,h a = Johnson et al. (1966); b = Eggen (1967); c = Jaschek et al. (1972); d = Blanko et al. (1968); e = Deutschmann et al. (1976); f = Cowley et al. (1969); g = Cowley (1972); h = Jaschek et al. (1964); t means that a reddening correction has been applied. 181 -18- Table 4. Spectral Classification of the Am stars. -X HD B-V UV MK ref number class class class 95608 A2 A2 Al a,c 76756 A4 AS A5 a,c 109307 A4 A4 a 125337 A4 A4 A2 a,c 23281 A7 A7 A5 b,c 30210 A7 A7 A2 a,c 72037 A7 A7 A2 b,c 107131 A7 A5 a 107276 A7 A5 b 115331 A7 A7 b 184552 A7 A7 b 33254 A8 A8 A2 a,c 44691 A8 A8 A3 a,c «L 36403 A8 A8 A2 b,c 8374 F0 F0 Al b,c 27628 F0 F0 A3 a,c 28226 F0 FO A3 a,c 28546 F0 A8 A5 a,c 124953 F0 A8 a a = Johnson et al. (1966); b = Mendoza (1974); c = Cowley et al. (1969). 182 -19- Captions of the Figures m m VS ni m Fig. 1. Ultraviolet colour-colour diagram > i8nn~ 2200 " 22OO~ 33Oo' for various classes of Ap stars and for Am stars: squares denote Hg-Mn staid; crosses, Si stars; opyn circles, si-Cr stars; triangles, Eu-Cr-Sr stars; plus signs, Am stars. The largest symbols denote stars with magnetic field strengths larger than 1000 gauss, the middle sized ones those with magnetic field strengths smaller than 1000 gauss. The smallest symbols are stars with too broad spectral lines to measure a magnetic field strength. The full line in the figure represents normal stars. Fig^_2. Same as figure 1, for m1550-jn1800 vs. m1800-m2500. Fig. 3. Ultraviolet colour (m,.n.-nincn.) plotted versus b-y. Symbols are ————— lOUU iDUU the same as in figure 1. Ap stars - especially the Si stars - are flux deficient compared to normal stars of similar b-y. Fig. 4. Ultraviolet spectral classes assigned to Ap and Am stars plotted versus spectral classes estimated from their (B-V) colours. Symbols are the same as in figure 1. The classes were derived from calibrations of colour indices vs. spectral type for normal main-sequence stars. Fig. 5. Ultraviolet spectral classes assigned to Ap and Am stars plotted versus MK epectral types. Symbols are the same as in figure 1. Fig. 6. (B-V) spectral classes of Ap and Am stars plotted versus MK spectral types. Symbols are the same as in figure 1. Black dots indicate the positions where the (B-V)- and MK-spectral types give the same results for the calibration used by Leckrone (1973). Fig. 7. Ultraviolet light curves for HR 5857, plotted versus phase. Phase zero corresponds to JD2442424.20 + n x 1.3049. Error bars give la observational errors. 183 LO "* CO C\! —• •O—•CsJCO" CM 00 o CO o CM 0 o * o o I -* o • o ? en i CNJ (NJ co o I 1 i i i i 1 1 t i I in CO CM o o o i i 1 I OOOOOOOOOO 0022-008IW Fig. 1 184 o LO O LO o LO LQ O C\J CO CO LO CO • CD o CM • o o o CsJ o I Q o CNJ 1 1 O O oo (O ^—1 o SI 1 o LO LO o LO O LO o LO o o O CM CM CO CO LO 008 I -OSS Fig. 2 185 CO CO 00S2-0081W 186 Fig. 3 B-VClass x x i i • i i • 1 • 1 1 1 1 1 / / F2 - A / / / FO &A /// - A3 m / A7 • A5 A s j / i A3 s - * J A2 • • ff A o A 4 / / A1 s v • /' L r A AO V A A - A° u o° B9 / * ? ° ox <^° 0 - B8 / xx o°X x 0° ft B7 - B6 - B5 - BA X - 11.111 l 1 1 I 1 (Ill B6 B7 B8 B9 B95 AO A1 A2 A3 kU A5 A'7 A8 FO F2 UV Class Fig. 4