\/ Oh/Ir Stars in the Galaxy

$\/ Oh/Ir Stars in the Galaxy$

\/ OH/IR STARS IN THE GALAXY

B. BAUD OH/IR STARS IN THE GALAXY OH/IR STARS IN THE GALAXY

PROEFSCHRIFT TER VERKRIJGING VAN DE GRAAD VAN DOCTOR IN DE WISKUNDE EN NATUURWETENSCHAPPEN AAN DE RIJKSUNIVERSITEIT TE LEIDEN, OP GEZAG VAN DE RECTOR MAGNIFICUS DR. D.J. KUENEN, HOOGLERAAR IN DE FACULTEIT DER WISKUNDE EN NATUURWETENSCHAPPEN, VOLGENS BESLUIT VAN HET COLLEGE VAN DEKANEN TE VERDEDIGEN OP WOENSDAG 12 APRIL 1978 TE KLOKKE 15.15 UUR

DOOR

BOUDEWIJN BAUD GEBOREN TE DJAKARTA IN 1948

STERREWACHT LEIDEN

1978 elve/laborvlncit - lelden PROMOTOR: DR. HJ. HABING Aan mijn moeder Aan Alberl ine CONTENTS

Summary

CHAPTER I A SYSTEMATIC SEARCH AT 1612 MHZ FOR OH MASER SOURCES I SURVEYS NEAR THE GALACTIC CENTRE

I Introduction 2 II Equipment and survey observations 2 III Analysis and confirmatory observations 5 IV The Type II OH/IR sources 6 V Selection effects 15 V.1 Velocity coverage and spectral resolution 15 V.2 Variability 16

CHAPTER II A SYSTEMATIC SEARCH AT 1612 MHZ FOR OH MASER SOURCES II A LARGE-SCALE SURVEY BETWEEN 10 i 15° and \b\ 17 I Introduction 17 II Equipment 17 11.1 The survey 17 11.2 Confirmation observations 20 III Results 20 111.1 Analysis of the survey observations 20 111.2 Confirmation of Type II OH/IR sources 21 111.3 The new Type II OH/IR sources 21 111.4 Known Type II OH/IR sources 28 111.5 Unclassified sources 33 IV Discussion 35

CHAPTER III GALACTIC DISTRIBUTION AND EMISSION PROPERTIES OF OH/IR SOURCES 38

I Introduction 38 II Observations and basic data 39 III Distribution of OH/IR sources in velocity,longitude and latitude 43 111.1 Velocity distribution 43 111.2 Longitude distribution 47 111.3 Latitude distribution 48 IV Luminosity function 49 IV.I Spectral resolution 49 IV.2 Mean peak flux density of OH/ItC sources 50 IV.3 Luminosity distribution 51 V Comparison with the identified Type II OH/IR sources 53 VI The galactic distribution of OH/IR sources 54

VI. 1 Distribution in log S"& ,. and I 54 VI.2 The radial velocity distribution 57 VII Discussion 59 VII.1 Stellar dynamics,age of the objects 59 VII.I.I Velocity dispersion perpendicular to the plane 6Ü VII.1.2 Association with stellar objects, age determination 60 VII.2 Density Distribution 62 VII.2.1 Comparison with the gas distribution 62 VII.2.2 OH/IR sources and planetary nebulae 62 VII.2.3 Spiral arm structure 63 VII.3 Luminosities 63 VII.3.1 Influence of the UV radiation field on the occurrence of the maser emission 64 VII.3.2 Luminosity function and AV - distribution 65 VII.4 Conclusions 68

CHAPTER IV OH/IR SOURCES IN THE NUCLEAR REGION OF THE GALAXY 71

I Introduction 71 'I II Observations and selection effects 71 11.1 Observations 71 11.2 Selection effects within 2° from the Galactic Centre 73 III Distribution of OH/IR sources on the sky 74 IV Radial velocities 75 IV. 1 The sources at II < 5 76 IV.2 Difference in the distribution as a function of AV 78 V Emission properties and luminosities 79 VI Discussion 80

Samenvatting 82 Studie overzicht 84 Dankbetuiging 85 SUMMARY

Around 1970 strong radio emission lines of the OH molecule were detected in the direction of Mira stars and M supergiants. These cool stars, in a late evolutionary stage, appear to loose a lot of mass. Due to this mass loss they form an expanding circumstellar shell, in which the OH line is emitted. Because of the special physical conditions in the shell this line emission is strongly amplified. These socalled OH/1R sources are characterized by their double peaked OH emission profile at a wavelength of 18 cm and by their often strong IR infrared emission. They offer a new and unexpected opportunity to study the spatial distribution and kinematical properties of stellar objects throughout the Galaxy with radio astronomical methods, i.e. unhindered by the interstellar extinction. This extinction has always been a severe limitation for optical studies.

The first systematic radio surveys, in Sweden and Australia, revealed the presence of several new OH/IR sources. Their strong concentration to the galactic plane confirmed the suggestion that these objects were detectable at large distances from the Sun. However, none of these new sources could be identified optically; as a result, their association with Miras or H supergiants was uncertain. Also, many were found with large radial velocities tind they appeared to be very

'• I luminous at radio wavelengths as compared to the nearby, optically identified sources. This thesis describes the radio astronomical observations that have led to the discovery of 71 additional OH/lR sources; in addition, an analysis is presented of the population characteristics of a large sample of sources. In Chapter I the observations around the nuclear part of the Galaxy are presented. They were done with the Dwingeloo 25 m telescope, the Effelsberg 100 m telescope (West Germany) and the NRAO 43 m telescope (U.S.A.); 38 new sources were found. Chapter II contains a description of the large-scale survey that covers the major part of the Galaxy, visible on the northern hemisphere. These observations, done with the Dwingeloo telescope, revealed the existence of 33 new OH/IR sources. Several previously known sources were also observed. The completeness of the Dwingeloo survey is studied and an atlas of OH line profiles of nearly all discussed objects is presented. In Chapter III I analyse the distribution and the velocities of 114 previously known and new OH/IR sources, found inside the area covered by the Dwingeloo survey. The parameter AV (the velu- city separation between the two emission peaks of the 18 cm line profile) appears to be a good criterion for a population classification. Sources with large AV are young objects, with small peculiar motions; they are probably associated with M supergiants. The sources with small AV show large peculair motions similar to the Miras. Model calculations indicate that both kinematical groups have a maximum density at 5 kpc from the Galactic Centre; the density decreases steeply on either side of this maximum. This characteristic density distribution suggests, that OH/IR sources are not the precursors of planetary nebulae, as has been suggested before. The large-AV sources appear to be concentrated in spiral arms. The similarity in spatial distributions to that of the gas in the galactic plane ',ets a lower limit to the age of this gas distribution. The radio luminosity function is consistent with an increase in the number of OH/IR sources with decreasing luminosity. This indicates an increase in the mass loss rate of the star in the course of its lifetime as an OH/IR source. In Chapter IV the results of Chapter I are analysed. It is shown that the central part of the Galaxy contains very few OH/IR sources, in agreement with the conclusions in Chapter III. The radial velocity distribution of these stars within 5° longitude is similar to that of the planetary nebulae. The spatial distribution, however, is more like the gas distribution. A comparison with the radial velocity distribution of the gas clouds is hampered by various selection effects, which are discussed in the first part of this chapter. The AV-distribution of sources within 90 pc from the Centre appears to be different from the AV-distribution of sources found elsewhere in the Galaxy, possibly indicating a higher m^tal abundance. The results of the present study, in particular the discovery of the correlation between AV and population characteristics, clearly demonstrate that a further study of OH/IR "Sources can provide an important contribution to our knowledge of the dynamical properties and the evolution of the Galaxy. ... Miras consist of a continuum of kinematic families with overiapping period-frequency distrubutions.

Smak, 1966

-J CHAPTER I: A SYSTEMATIC SEARCH AT 1612 MHZ FOR OH MASER SOURCES

I Surveys near the Galactic Centre

I Introduction

Recent surveys of 1612 MHz OH emission have revealed a large number of optically unidentified Type II OH/IR roaser sources in the plane of the Galaxy (e.g. Caswell and Haynes, 1975; Johansson et al., 1977a). Such sources snow strong emission in the 1612 »Hz transition of the ground rotational state of OH and sometimes weak main line emission (Wilson and Barrett, 1972). The characteristic double-peakod emission profile with a velocity separation, Av, between the two peaks of ^ 30 km s"' allows them to be easily identified. Because of tlie high brightness temperature of the maser emission the stronger ones may be seen throughout a large part of the Galaxy. The strong concentration of Type II OH/IR sources to the galactic plane, as found in systematic surveys, suggests that many are situated at large distances. The association of some nearby maser sources with long period (Mira) variables, considered to be intermediate population or Population II objects, and Che general increase in the number of Type II OH/IR sources towards smaller longitudes, as found by .Johansson et al. (1977b), indicate that there may be a concentration of such sources in the central part of the Galaxy. We have searched for 1612 MHz emission around the Galactic Centre (G.C.) using both the Dwingeloo 25 m telescope and the 100 m telescope at Effelsberg. The Dwingeloo observations covered an area of 3° x 4° around the G.C. with low sensitivity. With the 100 m telescope we surveyed with three times better sensitivity a narrow strip between í = 358 and Í. = 14 and between b = +0?5 and b = -0?5. Upon discovery of several sources with high radial velocities we surveyed a similar area, mere estended in latitude, with an increased velocity coverage using the NRAO 43 m telescope at Green Bank. This telescupe was later used again to confirm all Type II 011/IR sources found in the three surveys and to obtain a spectrum of eacli source with improved velocity resolution. The survey observations and equipment are discussed in Section II. Analysis of the observations and a description of the confirmatory observations is given ir. Section III. Section IV presents the sources and their spectra. Selection effects in the observations are finally discussed in Section V. A detailed analysis of the results will be given in Chapter IV.

II Equipment and Survey Observations

Instrumental parameters for the three telescopes at 18 era are listed in Table 1. HPBW is the halfpower beamwidth; n» is the aperture efficiency for a point source. S/T, is the conversion factor from antenna temperature to flux density and T is the system temperature measured on cold sky. The surveys were carried out by making time-i.itegrations on a grid of positions, covering several different areas. A summary of the areas surveyed is shown in Figure i and i;.^«?d with the

The National Radio Astronomy Observatory is operated by Associated Universities, Inc., under contract with the National Science Foundation. Table t

Instrumental parameters

HPBW S/T T Telescone 'A A syst

Dwingeloo 31' 0.58 9.6 Jy/K 45K Effelsberg 7.7 0.49 0.7 60 Groen Bank , Cassegrain 18 0.49 4.0 65 Prime Eucus 18 0.54 3.6 60

'Turner (private communication) relevant observational parameters in Table 2. Column I of Table 2 designates with a capital letter the area that has bRen observed with each telescope using different observing parameters. Column 2 gives the telescone used, while the surveyed ranne in galactic longitude and latitude is listed in colu.nns 3 and 4. Areas C, il and J are not described in Table 2 because thev are somewhat irregular. Instead, thev are onlv shown in Figure '• The spacing between gridpoints,AO, is given in column 5. It is in most cases equal to the HPBW. The overall bandwidth and useful velocity coverage are in columns 6 and 7 respectively. The frequency and velocity resolution, u.sini; a uniform weighting function, are listed in columns 8 and 9, while the integration time ¡¡pent on each gridpoint and the 3a sensitivity limit are given in columns 10 and II. The Jo sensitivity limit refers to the actual value measured on most gridnoints. It is usually somewhat higher than that inferred from the system temperature on cold skv listed in Table 1, because of the noise added by the galactic background radiation around the (3.C. and bv the relatively strong ground radiatiun often entering the first sidelubes. For spectrally unresolved featu' es the sensitivity limit will be higher than the values listed in Table 2. This will be discussed in Section V.

Tab It

Survey parameters

Area Telescope 1 nt An Overall Radial Resolution t 3o Bandwidth Velocity range MHz km s 1 kHz km s—! min Jv to2ç o,n to A Dwmgeloo 359?1 ;'l -2 '+ 2" 1 0Ï3 2.5 -220 +220 II.7 2,2 40 2m 5 B i* 2.0 8..0 +2 .0 + 8.0 0?3 2.5 -220 +220 11.7 2 5 3.6 C Effelshcrg 358.0 14,.0 -0.5 + 0.5 0. 125 2.5 -220 +220 15.6 9 1 0.3 D •t 359. 5 1 ..0 + 1.0 + 1.75 0.125 2.5 -220 +220 15.6 2.9 1 0.3

K 359. 5 0..25 + 8.5 + 9.25 0.125 2.5 -220 + 220 15.6 2k 9 1 0.3 V 0.25 1..75 +9 .25 + 10.0 0.125 2.5 -220 +220 15.6 2.9 1 0.3 (J Green Bank 0.25 2.5 + 11>7 +632 15.6 2.9 5 0.8 see Figure 1 -632 -167 15.6 2.9 5 0.8 (1,-1 0.25 2.5 -547 + 547 15.6 2.9 5 0.7 All the velocities quoted in this paper are given with respect to the local standard of rest (l.s.r.). The velocity of the Sun is assumed to be 20 km s" towards RA = 18 , Dec = +20° (epoch 1900.0). We will now turn to a description of the observations with each telescooe.

ummiiioo The Dwingeloo observations of areas A and B were carried out in February and March 1975, using a closed-cycle helium cooled parametric amplifier and a single linearly polarized feed. Because the 1612 MHz emission of most Tyne II OH/IR sources is unpolarized, the use of a linearly polarized feed does not introduce a bias in the source detection. The 256-channel autocorrelation spectrometer was used in a frequency switched mode, switching over 2.5 MHz, with a dutv cycle of 30 seconds. The signal band was centered on V = 0 km s , the reference band on V = -465 km s The signal was clipped in a one-bit mode. The system was calibrated continuously against a noise tube of well-known and constant temperature. The off-line reduction procedure yielded spectra calibrated in antenna temperature. A first order baseline, fitted to both ends of each snectrum, was subtracted. Near the G.C. such a procedure did not significantly improve the nresentation of the data because of broad absorption lines in that direction due to the OH in molecular clouds.

Kb'i'KLSBUHÜ Observations with the 100 m telescope in Effelsberi, were performed with a cooled dual channel receiver system, using two orthogonally polarized feeds. The 384-channel autocorrelation snectro- meter was split into two equal sections, each receiving signal over the same bandwidth of 2.5 MHz, centered on V = 0 Ion s . The output of both receiver sections was later added and averaged. The total power mode of operation was used, with a reference of six minutes duration after every 16 spectra, taken on the position I = 4°, b = +4°. The integration time on the grid positions was 1 minute. The signal and reference spectra were generally well balanced, and a baseline fit was unnecessary, except for spectra taken near the G.C. Area D, E and F and a major part of area C were observed in March 1975; the latter area was completed in October 1976 and Anril 1977.

CRF.KN BANK The survey observations in Green Bank with the NRAO 43 m telescope were carried out in De- cember 1975. The linearly polarized measurements were made at the Cassegrain focus using an up- converter system operating in only one channel. The anerture efficiency of the Cassegrain system is not well known but is measured to be ~ 10% worse than that if the primary focus (Turner, nri- vate communication). We therefore adont S/T. = 4.0 JvK . The 384 channel autocorrelation recei- A ver was divided into two sections of 192 channels, each covering a iiotal bandwidth of 2.5 MHz but centered at different velocities. In the case of area G, close to the pLane, the velocity coverage complemented that of the Effelsberg survey, making it nossible to search for high velocity sources. Each of the two receiver sections had an overlap of 65 km sec with the Effelsberg observations. On areas H and J, the two receiver sections were arranged to cover a larger velocity range symmetrically around V = 0 km s . They had a mutual overlan of 30 km sec" . Observa- tions were made in the total power mode, and a total of five references, taken throughout the observing run at i = 4°, b = +4° for area G, and at f, = 20°, b = +20° for areas H and .1, were used. The references were averaged and the result was used in the final analysis. This was done in order to reduce as far as possible the noise in the spectra. Baseline irregularities were eliminated with a sixth order polynomial fitted to the whole baseline. In view of the narrowness of the 1612 MHz maser emission this is a permissible procedure. .'•'(';/. 1. Kxtent of the. suv— iv¡ied oï'Cas. The hatched ••Ci'L'tco indicate, the site of the HPBW of the Lclcst'Open usad, 'l'ha aquai'cis nitii'fc t-hii Wo nenoitti'itiea of the sw'vciteJ ai'cas.

13° \f 11° 10° 9° B° 7° 6° b° 4° 3° 2 1° & 359° 358

of the 1612 MHz raaser emission this is a permissible procedure.

Ill Analysis and cunt*irmatorv observations

The extent of the survev observations is summarized ¡n Figure I, where all the areas, listed in Table 2, are drawn in galactic coordinates. The corresponding capital letter (column 1 of T.ihle 2) is nut in the lower lefthand corner of each area. The blank and hatched squares on the left denote the i>> sensitivities, in Jv, of the correspondingly marked areas. Hatched circles represent the HI'HW's of the three telescopes used. The rough lv l.-slinned recién around the C.c;., consist ing of areas A, C, I), G, H and J, is the most imnort.mt iw IMV.IH-ÍI' ill the new 'l'vpe 11 OH/tH sources have been found within its boundaries. It has been surveyed u-ilh different sensitivitv yver different velocitv intervals at anv one position, in a sense th.il the sensitivity is halved at |v| • 220 km s . This mav introduce a bias against weak sources at lii^h velocities. However, almost all sources were found within the velocity range of l\'! •' 220 km s , covered bv the Uwingeloo and Effclsberg observations. Only two sources were discovered outside this velocitv range. Thev are both more than five times stronger than the i sensitiv i tv limit (0.7 ,lv) at those velocities. If the sources at high radial velocities have a gaussian luminosity distribution such as that found bv Johansson et al. (1997b) we extrapolate f rum the two strong sources seen, that 'here should exist about 30 sources outside the velocitv range of -220 to +220 km s~ that are stil! stronger than the actual sensi- livitv limit. Since no such sources were found we conclude that the two sources detected are un- usual in their high radial velocitv and that the maioritv of Tvne II OH/IK sources at small longitudes have |\'| 220 km s . This conclusion ís supported bv recent further 1612 MHz observations ariMind the C.C. with improved sensitivity and velocitv coverage. I're 1 iminarv analysis of these d.U.-i taken with the 100 m telescope has revealed no new sources with |v[ • 220 km s~'. Hence, because the velocity coverage of the observations .it all positions spans a range from ¿it least -220 km s to +220 km s , we mav consider the l.-shaned region to have been sampled completely in radial velocity extent. A spectrum for every gridpoint was plotted. In the case of the Green Bank observations we made separate plotò for each receiver section. Each individual spectrum was then inspected bv eve, searching for both emission and absorption. More than 100 independent features were found. In this paper we present only those sources with the typical characteristics of Tvpe II OH/IR emission (Wilson and Barrett, 197?). Since we had no data on the other OH lines, the identification of Tvpe II OH/IR sources was based on two criteria only: (1) candidate sources should have a profile that shows the characteristic double structure with steen outer edges and more gradual inner slones; (2) the intensity of the emission in neighbouring gridnoints around the nosition of the source should decrease as expected for a point source. Only those sources that were visible in at least two adjacent gridpoints were accented. In August 1977 the sources that satisfied the above criteria were reobserved at 1612 MHz in order to confirm their identification as Tvne II OH/IR obiects and to obtain a nrofile with better velocitv resolution. For these observations we used again the 43 m telescope at fireen Bank will] a cooled dual channel 18-cm receiver at the nrime focus. The stability of the receiver allowed a total power observing mode, with a reference snectrum taken daily at '- = 40 , b - +40 . The autocorrelation receiver was used in the parallel mode, with both receivers centered on. the same velocity. The orthogonally polarized signals from the two receivers were added and averaged. Tho total bandwidth was 625 kHz, resulting in a frequency resolution, after Hanning weighting, of 6.5 kHz or 1.2 km s . With this resolution most emission features of Tune II OH/IR sources are expected to be spectrally resolved (see Johansson et al., 1977a). In most cases the telescone was pointed at the nositions determined bv internolation of intensities between neighbouring gridpoints of the survey observations. The accuracy of the positions thus obtained are usually better than 2.'4. An offset of 2.'4 from the centre of the 43 m telescope beam at 18 cm corresoonds to an attenuation of v, 5%. For most sources the measured flux density may thus be underestimated bv at most a few percent.

In some cases an improved position was needed before an accurate line nrofile could be obtained. This was required when several emission lines were seen at one r.osition and identification of Tvpe II OH/IR emission was difficult. In such case we observed a grid of 4 nositions around the suspected source and determined the position of each individual emission feature. The features with coinciding positions within the error limits were considered to belong to the same source. A spectrum using a total bandwidth of 625 mHz was subsequently taken at the derived nosition. A few weak sources, detected with the 100 m telescone and suspected to be of Tvne II OH/IK, could not be confirmed with the 43 m telescone, in most cases because of a strong baseline curvature due to the presence of strong and broad OH absorntion bv molecular clouds near the Galac- tic Centre. The observations with the 43 m telescope in regions of strong absorntion gradients were generally more confused than the observations with the 100 m telescope, because of the larger beam of the former telescope.

IV The Type II OH/IR sources

In total we have found 43 sources which we identify with Tvpe II OH/IR objects, five of these were previously known. All sources are listed in Table 3 in order of increasing galactic longitude. The first column gives the name derived from the galactic coordinates, which are lis- Table 3

Type II OH/IR Sources near the Galactic Centre

Name e b Ríl (1950) DEC V S

It 0 o h tn s o • (km a"') (Jy)

OH 358. 1+0. 1 358.0610.02 +0.13 + 0.04 17 37 12 -30 29 32 - 40 - 11 1.1 1.3 OH 359. 1 + 1.2 359. 1 '-0.3 + 1.2 + 0.3 17 35 37 -29 02 26 -145 -127 1 .8 1.6 OH 359.4-1.3 359.3710.02 -1. 25 +0.04 17 45 50 -30 06 26 -234 -203 4.4 5.3 OH 359.5+1.3 359.5410.01 + 1.29 10.01 17 26 21 -28 37 16 + 18 + 45 1.6 0.4 OH 359.7+1. 3 359.70+0.02 + 1.25 10.06 17 36 53 -28 30 27 + 33 + 62 0.7 0.8 OH 359.8+2. 6 359.7510.02 +2.63 + 0.02 17 31 44 -27 43 13 + 37 + 60 4.7 5.3 OH 0.0-0. 1 0.03+0.01 -0. 13 10.01 17 43 01 -28 57 34 + 99 + 127 0.8 1.9 + OH 0.1+5. 1 0.1 +5. 1 17 23 18 -26 03 00 -151 -135 2.6 1.1 OH 0.1+0, 1 0.1410.01 +0.06 + 0.01 17 42 33 -28 45 58 + 146 + 174 2.0 3.1 OH 0.3-0. 0 0.2610.01 -0.01 10.01 17 43 06 -28 42 03 + 58 + 93 1 .5 4.4 OH 0.3-0. 2 0.33+0.005 -0. 19310. 005 17 43 59 -28 44 13 -357 -328 4.9 3.4* OH 0.4+0. 1 0 38+0. 04 +0.13 ±0.04 17 42 51 -28 31 31 + 125 + 155 1.0 0.9 OH 0.5-0. 1 0.48+0.01 -0. 13 ±0.02 17 44 06 -28 34 34 + 130 + 154 0.9 1.4 OH 0.5-0.8 0.5 ±0.2 -0. 8 + 0.2 17 46 34 -28 52 55 - 64 - 30 1.1 0 8 OH 1.1+0.4 1 09+0 03 +0. 38 10.03 17 43 34 -27 47 22 -143 -108 0.8 1 3 OH 1 1-0 8 1 10+0 03 -0 80 ±0. 03 17 48 10 -28 23 39 - 9 + 30 25.4 30 5 OH 2+1 3 1 21+0 01 +1 26 + 0 01 17 40 29 -27 13 27 + 31 + 58 5 6 5 3 OH 1 3+1 0 1 33+0 01 +1 01 +0.01 17 41 43 -27 15 15 - 27 + 3 4 2 4 2 OH 5-0 0 I 4610 01 -0 01 ±0 01 17 45 57 -27 40 38 -141 -114 3 0 4 1 OH 7-0 0 1 74±0 01 -0 01 + 0 01 17 46 57 -27 29 04 + 105 +134 •0 8 4 2 OH 2 2-1 7 2 17+0 01 -1 65 +0 01 17 53 58 -27 54 31 - 91 - 55 4 2 6 7 OH 2 5+0 3 2 50+0 06 +0 25 ±0 06 17 47 22 -26 39 11 - 86 - 71 0 4 0 5 OH 2 6-0 5 2 59+0 01 -0 45 10 01 17 50 ¡6 -26 56 10 - 27 + 19 6 1 7 2** OH 3 3-0 3 3 25+0 06 -0 28 ±0 02 17 51 08 -26 16 56 - 96 - 69 1 6 1 7 OH 3 3+0 0 3 2510 06 +0 00 + 0 06 17 50 03 -26 08 20 - 57 - 25 1 2 1 5 OH 3 9+0.0 3.9410.02 +0.00 + 0 03 17 51 37 -25 32 45 - 31 + 1 1 2 1.1 OH 4.0-0.5 4.00+0.06 -0.50 10 06 17 53 40 -25 44 5t - 13 + 10 0.8 0.4 OH 4 4+0 0 4.40+0.03 +0.01 ±0 03 17 52 37 -25 08 42 - 18 + 16 3 2 1.2 OH 4.5-0.4 4.5410.01 -0.38 +0 02 17 54 25 -25 13 19 -144 -100 1.3 1.9 OH 5.0+1 .5 5.0 +1 .5 17 48 30 -23 50 00 + 107 + 135 5 4 ++ OH 5.9-0.4 5.90±0.01 -0.38 +0.01 17 57 26 -24 02 49 - 21 + 13 34.7 1.5 OH 6.0+0.3 6.00±0.02 +0.25 ±0.02 17 55 16 -23 38 37 + 94 + 106 1.8 1.3 OH 6.5-0 . 1 6.50+0.04 -0.13 ±0.04 17 57 48 -23 24 08 + 93 + 114 0.7 1.2 OH 3.0+1 .4 8.0 +1.4 17 55 12 -21 20 08 - 34 - 9 4.2 6.6+ OH 9.0-0.1 9.00+0.04 -0.13 ±0.04 18 03 10 -21 13 56 - 66 - 33 0.7 I.3 •OH 9.6+0.4 9.63±0.02 +0.38 ±0 .02 18 02 36 -20 25 57 - 77 - 46 1.3 2.8 OH 9.9-0.1 9.88+0.06 -0.13 ±0.06 18 05 01 -20 27 58 + 79 + 112 1.9 0.8 OH 10.0-0.1 10.04±0.02 -0.10 ±0.02 18 05 14 -20 18 43 + 23 + 80 0.7 2.7 OH 10.4+0.0 10.38±0.04 +0.00 ±0.04 18 05 35 -19 57 59 + 43 + 79 0.6 0.8 OH 11.1+0.0 11.13±0.06 +0.00 ±0.06 18 07 08 -19 18 43 - 35 - 8 0.9 1.1 OH 11.3+0.0 11.32+0.02 +0.00 ±0.02 18 07 32 -19 08 46 + 64 + 98 2.1 2.9 OH 11.4-0.1 11.38+0.06 -0.13 ±0.06 18 08 08 -19 09 25 + 87 + 123 0.5 0.4 OH 11.5+0. 1 II.54±0.01 +0.10 ±0.01 18 07 37 -18 54 19 + 20 + 64 4.2 25.0

Previously known, position quoted from Kerr and Bowers (1974). Previously known, all narameters quoted from Kerr and Bowers (1974). Previously known, all parameters quoted from Baud et al. (1975). **Previously known as OH 2.7-0.3 (Chaisson and Dickinson, 1972). -63 -53 -U3 -33 -23 -13 -J 7 17 o -193 -173 -163 -153 -113 -133 -123 -113 -ID3 VELOCITT KM/S VELOCITY KM/S i IK. .' on m. 1*0.1 FÍR. 1 OH 359.UI.2

' ¡

-857 -2U7 -237 -227 -217 -207 -197 -187 -177 13 23 33 «3 S3 63 73 VELOCITT KM/S VELOCITT KM/S Fx«. ~ Oil 359.4-1.3 Flg. 5 OH 359.5*1.3

Eg.

0 20 30 40 50 SQ 70 80 90 VELOCITY KM/5

FiC. 6 ÜH 359.7*1.3 Fig. 7 OH J59.H*2.6

A h AA/ ^\r^ A * .

90 100 HO 120 130 o -182 -172 -162 -152 -112 -132 -122 -112 -102 VELOCITY KM/S VELOCITY KM/5 Ftg. 9 OH 0.1 -5.1 o 120 130 110 ISO 150 170 190 190 200 19 59 59 79 89 99 109 119 VELOCITY KM/S VELOCITT KM/S Fig. 10 OH 0.1*0.1 . II OH 0.1-0.0

100 110 120 130 WO ISO 160 170 -BZ -72 -62 -SS -H2 -32 -22 -12 -2 VELOCITY KM/S VELOCITY KM/S Fig. li CH 0.4*0.1 Fig- n OH 0.5-0.1

100 110 120 130 WO 150 160 170 150 o -163 -1S3 -1H3 -133 -123 -113 -103 -93 -B3 VELOCITY KM/S VELOCITY KM/S Fie. IS OH 1.1*0.4

i i ^ i 1——1 —1 1——1 UR E

6 -

CE Z ' II E Z • iv CE f» 1 1 1 ' • 1 1 1 -25 -15 -5 S 15 25 35 «5 55 VELOCITY KH/S FiK. ib OH I.I-0.8 10

-49 -19 -29 -19 -9 I II 21 31 165 -135 -115 -US 155 -US -IBS -99 VELOCITT KH/S VELOCITY KM/5

FIR. lã Oil I, )• .8 0

X r = - L. J x; A "s cc ó z • /v l J i i i i i i ""Y 80 90 100 Hü 120 130 -IDS -9S -S9 -79 -Ga -59 -49 -39 -29 VELOCITY KM/S VELOCITY KM/5 Fie. :t UK 2.2-1.7

117 -107 -97 -Í7 -77 -67 -57 -H7 -37 -41 -31 -21 -11 -1 9 19 29 39

FIB. IJ Uli Z.5»0. lie .'i UH 2.6-0.5

ëi

X IC o ÜJ ™ 1- o £s CC D z Z a flNTE

•5t -Hl -31 -ai -II -l -38-2» -IB -o a ia sa 32 VfcLOC.'TT KM/S VELOCITT KM/S

-38 -28 -18 -8 2 la 22 32 «2 -160 -ISO -110 -130 -ISO -110 -100 -90 -80 VELOCITT KM/S VELOCITT KM/S

-33 -33 -13 -3 7 17 27 37 VELOCITT KM/5

SO 60 70 80 9D 100 110 120 130 60 70 80 90 100 110 120 130 1U0 VELDCITT KM/S VELBCITT KM/S ¡2

-U3 -39 -29 -19 -9 I II 21 •77 -67 -57 -H7 -57 -27 -17 -7 VELOCITY KM/S VELOCITY KM/S ig. 13 Uli B.O • Fie. 11 OH 9.0-0,1

LU o rr T

A 0.2 0 .

HPERflT U „ } \\ A A . £ a íiyiA JA« / v^yw^'v- 2 0 . ENN R — ¿ 's -107 -97 -87 -77 -67 -S7 -U7 -37 -27 GO 70 80 90 100 HO 120 130 WO VELOCITT KM/S VELOCITT KM/S Flg. 36 OH 9.9-0.1

40 50 60 70 60 90 SO 60 70 60 VELOCITT KH/S VELOCITT KH/S FtR. 3' (1H IQ.O-O.I Fig. 3& OH IO.ú+0.0

-GO -SO -HO -30 -20 -10 0 10 20 40 50 60 70 SO 90 100 110 120 VELOCITY KM/5 VELOCITY KM/S Fig. 39 OH II.U0.0 11.1*0.0 Daihea line represents th« low-velocity cropanent, that i» :alned fron the tir.al prof lit (drawn) in a Banner describe in the text, 13

Ui e CC 9 - f " K IC O JQ ¡I ; 0.2 0 0.8 NTENN R TE »

's i i 1 1 1 ia 7a aa «9 toa na iza isa ma 15 25 as 15 ss es 75 es VELDCITT KH/S Ftg. 4i QK 11.5*0.1

Figures S to 42. 1612 Wffa profiles of the Type II OH/IR sources presented in Table 3. hinltiply antennatemperatures by S.G to convert to Jy. 1 Jy = 10~26 N m~2 Hs'1 . ted on columns 2 and 3 together with their errors. The corresponding right ascension and declination aopear in columns 4 and 5. The radial velocities of the low- and high-velocitv emission neaks relative to the l.s.r. are given in columns 6 and 7. Columns 8 and 9 list the peak flux densities (epoch 1977.6) of both emission peaks assuming a value of S/T, = 3.6 Jv.K for the ¿3 ra prime focus 18 cm system. Figures 2 to 42 show the 1512 MHz spectra of the Type II OH/IR sources, all taken with the 43 m telescope in August 1977. All sources listed in Table 3 are included excent for OH 5.0+1.5 and OH 0.3-0.2, which were not observed with the 43 m telescone. These spectra appear in Kerr and Bowers (1974) and Baud et al. (1975) respectively. Each spectrum has a velocity coverage of 92 km s , and a resolution of 1.?. km s . The large velocity coverage is necessary in order to trace the spectral baseline, which was not always at nominal zero due to either 1612 MHz absorntion within the bandpass or a strong continuum background that is absent in the reference snectrum. For everv source the outnut from the two receiver sections was analyzed separately in order to determine the linear polarization of the emission. In most cases the sources were practicallv unpolarized. Some sources are discussed below. OH 358.1+0.1 ^.ig. 2). The emission at -26 km s~ and -18 km s is probablv real, as well as the absorption at -4 km s . This spectrum is an example of the confusion of features at 1612 MHz near the Galactic Centre. Observations with a smaller beam are necessarv in order to snatiallv separate the individual emission and absorption features.

OH 359.5+1.3 and OH S59.7+1.3 (Figs. 5 and 6). These OH/IR sources are separated bv less than one HPBW of the 43 m telescooe and are visible in both Figs. 5 and 6. OH 359.8+2.6 (Fig. 7). The low velocitv component of this source does not show the tvnical steep outer edge. It is considered to be a Tvne II OH/IR source because of its point-like behaviour and the shape of the high velocity component, with a steep outer edge and a more gradual inner slope. OH 0.0-0.I (Fig. 8). Strong absorntion is present at velocities lower than +90 km s~ . The low velocity component in this spectrum is rather weak. Earlier survev observations with the 100 m telescope were less confused by the absorntion and showed the low velocity peak more clearly. OH 0.1+5.1 (Fig. 9). This source was first discovered bv Kerr and Bowers (1974). The neak flux density of both the low and high velocity components is now (1977.6) 0.6 of the flux density in 1973.5. Off 0.1+0.1 (Fig. 10). The two emission neaks are superposed on a broad plateau of emission that extends to well outside the velocity range determined by the neaks. A similar kind of profile has been found by Andersson et al. (1974) in the 1667 MHz emission of the strong source OH 26.5+0.6, although there the emission outside the two neaks is less pronounced. OH 0.3—0.0 (Fig. II). The emission of the low velocity component is superposed on the broad absorption at +54 km s . Wide band observations around the indicated position show that both • velocitv components have the same position, confirming the identification as Tvne II OH/IR source. OH 1.7-0.0 (Fig. 20). This source shows a broad plateau of omission. It cannot be ruled out that this is due to a curvature in the baseline. OH 2.8-1.7 (Fig. 21). The emission feature at -100 km s~ was discovered during the confirmatory observations. Its position is not well determined and it is not clear whether this feature' is associated with OH 2.2-1.7. OH 2.6-0.5 (Fig. 23). Chaisson and Dickinson (1972) discovered this Type II OH/IR obiect near Hoffman source 39. The peak flux density of both velocitv components determined from the present observations differs bv a factoE of 1.5 compared with the earlier observations. Chaisson 15

and Dickinson (1972) suggest that there mav be broad band 1612 MHz absorntion in the direction of OH 2.6-0.5, due to the presence of a diffuse foreground cloud. Our wide-band survey observations give an upper limit to anv such absorption of 0.4 K antenna temperature. OH S.3-0.S (Fig. 24). Both emission neaks show linear polarization of about 20%. OH S.9+0.0 and OH 4. '3+0.0 (Figs. 26 and 28). The emission between the two peaks mav be due to baseline curvature. OH 6.9-0.4 (Figs. 30a and 30b). This source is situated 1° south of the strong continuum source W28. The strong emission peak at -21 km s was discovered bv Coles and Rumsev (1970). It is weakly circularlv polarized. Because of insufficient velocity coverage they did not detect the associated emission at +13 km s . In Fig. 30a, the profile has been clipped at 4 Jy in order to show the latter emission more clearly. The contrast in shape between the irregular high velocity component and the smooth low velocity component (Fig. 30b) is remarkable. OH 8.0+1.4 (Fig. 33). The flux densities (enoch 1977.6) for both emission peaks are 25% less than those measured at epoch 1973.5 by Kerr and Bowers (1974). OH 9.6+0.4 and OH P.9-0.1 (Figa. 35 and 36). Both sources show emission everywhere between the two peaks, this may well be due to baseline curvature. OH ll.i+0.0 and OH 11.5+0.1 (Jigs. 40 and 42). These sources are separated bv 14'. The high velocity component of OH 11.5+0.1 and the low velocity component of OH 11.3+0.0 coincide in velocity and the latter is seriously confused with the emission of the former. To obtain an estimate of the low velocity component of OH 11.3+0.0, we subtracted the high velocity component of OH 11.5+0.1 from the profile taken at the position of OH 11.3+0.0, correcting for beam attenuation. The resulting profile (shown dotted in Fig. 40) represents the low velocity component of OH 11.3+0.0. Itsshaoe is clearly reminiscent of Tvpe II OH/IR emission. The ratio between the peak flux densities of both velocity components agrees well with the ratio determined from the observations with the 100 m telescope, which are not hampered bv confusion because of the smaller beam size. OH 11.4-0.1 (Fig. 41). This source lies close to the two sources discussed above. The emission at +64 km s and +98 km s belong to those strong sources and should be disregarded.

V Selection Effects

Here we mention briefly which selection effects are important in anv large scale systematic OH survey. In ChaDter IV the specific selection effects, which are particularly important for the analysis of the data presented here, will be discussed.

V. 1 Velocity ooverage and resolution.

In a large-scale spectral line survey one has to make a compromise between velocity coverage and resolution. As can be seen from Table 3 almost all sources of this survey lie well within the velocity range |v| < 220 km s .In Section III we have argued that this velocity range covers essentially all Type II OH/IR sources near the Galactic Centre. The velocity coverage of the observations therefore seems to be complete. However, we also found that objects with exceptional velocity do occur.

The velocity resolution can seriously affect the sensitivity limit for spectrally unresolved features. Inspection of the Type II OH/IR sources found by Johansson et al. (1977a) shows that many velocity components have a full width at half-intensity (FWHI) of less than 3 km s~'. The minimum FWHI they find is about 1.5 km s~ . With the velocity resolutions of 2.3 and of 2.9 km s~' 16

used in the survey phase of the present work, many such sources would have been unresolved. This results in observed peak antenna temperatures that may be a factor or two lower than the values that would have been obtained with adequate spectral resolution. Our sample of sources is therefore biased against weak sources that have narrow velocity components. For the detection of these sources the sensitivity limit of the survey may be twice the value quoted in Table 2 (see also ChaDter II).

V.!'. Variability.

Many, if not all OH/IR stars show regular variable intensity of both velocity comnonents, typically with periods of several hundred da\s and ratios of maximum to minimum 1612 MHz flux density of typically between 2 and 4 (Harvey et al. 1974). Surveying a part of the Galaxy at any given epoch may be compared to a short-time exnosure of a sky with twinkling stars. Those sources that have a time averaged intensity comoarable to the detection limit will not always be detectable depending on the nhase of their radio "light curve". We do not however expect that variability will result in anv svstematic bias in our sample, because of its stochastic character for a given sample of sources at a given time.

References

Baud, B., Habing, H.J., Matthews, H.E., Winnberg, A. 1977, in preparation. Caswell, J.L., Haynes, R. 1975, Monthly Notices Roy. Astron. Soc. 1975, J_73, 649. Chaisson, E.J., Dickinson, D.F. I972, Astrophys. Letters \2, 119. Coles, W.A., Rumsey, V.H. 1970, Astrophys. J. J[59_, 247. Harvey, P.M., Bechis, K.B., Wilson, W.J., Ball, J.A. 1974, Astrophys. J. Suppl. 27, 331. Johansson, L.E.B., Andersson, C, Goss, W.M., Kinnberg, A. 1977a, Astron. Astrophys. Suppl. ¿8, 199. Johansson, L.E.B., Andersson, C, Goss, W.M., Winnberg, A. 1977b, Astron. Astronhys. 5_4_, 323. Kerr, F.J., Bowers, P.F. 1974, Astron. Astronhvs. 36_, 225. Wilson, W.J., Barrett, A.H. 1972, Astron. Astroohvs. 17, 385.

This chapter has been submitted as a separate article to Astronomy and Astrophysics Supplement. It was written together with Drs. Habing (Leiden), Matthews and Winnberg (Bonn). 1 17

CHAPTER II: A SYSTEMATIC SEARCH AT 1612 MHZ FOR OH MASER SOURCES

II A Large-Scale Survey between 10° < I £ 150° and \b\ £ A?2

I Introduction

Several systematic 1612 MHz OH surveys of the Northern Hemisphere (Johansson et al., [977a; Bowers, 1978a) and of the Southern Hemisphere (Caswell and Haynes, 1975) have recently been completed. These surveys have revealed the existence of a large number of maser sources with emission profiles similar to the characteristic double-peaked point-source emission which is associated with several M-giant Mira variables and supergiants (Wilson and Barrett, 1972). The latitude distribution of these sources (designated as Type II OH/IR), which is strongly concentrated towards the galactic plane, suggests that they are at large distances from the Sun. None of these sources, first discovered during the above mentioned surveys, have been identified optically, but subsequent observations (Schultz et al., 1976; Evans and Beekwith, 1977; Glass, 1978) show strong IR (\ % ¿v) point-source emission at the radio positions. The similarity of the OH emission profiles, the IR photometric data (Evans and Beckwith, 1977) and the long period variability, both at 1612 MHz and at infra-red wavelengths (Nürnberg, private communication) indicate that many of these optically unidentified Type II OH/IR sources are associated with stars of similar typo to those with optical counterparts.

We have undertaken a large-scale 1612 MHz OH survey between galactic longitudes 358 and 150°, with increased latitude extent, sensitivity and sky coverage compared to previous surveys, particularly in order to enlarge the number of known Type II OH/IR sources, and to provide a broader statistical base for the study of their galactic distribution and kinematics. The observations near the Galactic Centre are reported in Chapter I. Here we are concerned with the observations at I > 10° which were carried out in a homogeneous manner with the Dwingeloo 25 m telescope (DRT). The results will be compared with those of Johansson et al. (1977a) and Bowers (1978a) whose observations overlap the present survey. The survey observations and equipment, as well as that for subsequent confirmatory observations taken with the NRAO A3 m telescope, are discussed in Section II. The results are presented in Section III, and are followed by a discussion on completeness and a comparison with the other surveys in Section IV. We shall refer to the present observations as the Dwingeloo survey, to the observations by Johansson et al. (1977a) as the On- sala survey and to the observations by Bowers (1978a) as the Green Bank survey.

II Equipment and Observations

II. 1 The SiLPVey

The survey observations were performed between June 1976 and November 1976 and between February 1977 and April 1977, with the DRT almost continuously in use. The Dwingeloo 25 m telescope and observing system has been completely modernized in 1974— 1975 and is now almost optimal for large-scale survey work. The telescope has been equipped with a new mesh surface and a quadruped feed support, specifically designed to bear the heavy load of a cryogenically cooled receiver. The HPBW is 31 arc inin at 18 cm and the aperture efficiency is

The National Radio Astronomy Observatory is operated by Associated Universities, Inc., under contract with the National Science Foundation. 18

150° 140° 130° 100° 90' 80° 70° 60° 50° 40° 30° 20° 10° Galactic longitude

Fi.j. 1. The extent of the 1613 MHs OH survey described in this papei; in galactic coordinates. The hatched avea at Lowev galactic loYigitudes is pai't of the Onsala survey (Johansson et al., 1977a) and was not observed. The approximate So sensitivities (in Jy; see also Table 3) ave indicated for the various parts of the survey. The boundaries between the sensitivity regions are shoiõn by dashed lines where neaessary.

'^ 0.6 (Sondaar, private communication), resulting in a ratio between flux density and antenna temperature of S/T = 9.3 Jy K . The root-mean-square pointing accuracy is about 1.6 arc min (Tenkink and Slottje, 1977). The single-channel closed-cycle helium-cooled 18 cm receiver had a dipole inside the cooled area and a parametric amplifier that could be tuned to any of the four OH ground state rotational transitions. This tuning could not be done remotely, so the receiver was centred at 1612 MHz for the duration of the survey and no observations were made at any of the other 18 cm transitions. Because most Type II OH/IR sources show very little, if any, linear polarization, the orientation of the dipole was kept fixed in a vertical direction. The systtm temperature on cold sky in 1976 and in 1977 was 41 and 37 K, respectively. The 256-channel autocorrelation spectrometer and its operation have been described by Bos (1976) and O'Sullivan (1975). Both the telescope and the spectrometer are controlled by a minicomputer, which also handles the transfer of the autocorrelation data to magnetic tape. With a well prepared observing program the system can run for several days continuously without human intervention. The fourier transform and subsequent calibration to antenna temperature scale were performed off-line. The observations covered regions along the galactic equator between galactic longitudes IOU and 100° and between 130° and 150°, generally between latitudes +4?2 and -h°l. In the following longitude intervals the latitude extent was smaller: -2.4 < b < +2?7 for 90 < 100u, I < 77?5 and 33° < I < 42?5; -1?5 < b < +1?8 for 42?5 < I < 67?5. Part of the area covered in the Onsala survey was omitted. The extent of the sky coverage in galactic coordinates is shown in Figure 1. The separate area at large galactic longitudes is circumpolar at Dwingeloo and was used as a time-filler, when the rest of the area was below the horizon. The total area was covered with a rectangular grid in I and b, with a gridpoint separation of 0?3 in both i and b. The actual search at any given time was done with a separation of 0% and the gridpoints in between were usually covered wiL'uin a few days. In this way it was possible to discriminate between earth-based

Jy 10"26 W m"2 Hz"1 19

Table 1

Central Velocity of the Bandpass

interval longitude central km s~'

10?0< í.< 27?5 +50 27.5< í < 42.5 +25 42.5< í < 57.5 0 57.5< i < 77.5 -25 77.5< l < 150.0 -50

narrow-hand interference, which was common and usually lasted for several hours, and astronomical sources. The observations were done in a total-power mode and a reference spectrum was taken about every six hours. The duration of the integration on the reference position (North Pole) depended on the number of survey integrations to follow (Ball, 1975). The correlator was operated in a 1-bit mode, and the integration time was 15 minutes, except for 3° < |ii| < 4?2, 130° < !t < 150°, where an integration time of 5 minutes was used. Every 30 seconds the system was calibrated against an internal noise source of well-known and constant temperature. A bandwidth of 2.5 MHz was used, corresponding to a velocity coverage of 465 km s and a resolution (employing uniform weighting) of 11.7 kHz (y 2.2 km s ). The centre of the band was set at different radial velocities, depending on the longitude interval, to provide a first- order correction for galactic rotation. Table 1 lists the longitude intervals and their corresponding central radial velocities. All velocities quoted in this paper are given with respect to the local standard of rest (l.s.r.). The velocity of tne Sun is assumed to be 20 km s towards RA = 181', Dec = +30° (epoch 1900.0). The sensitivity of the survey is a function of galactic longitude. This is due to 1) the use of different integration Limes, 2) a difference in system temperature between the first (1976) and the second (1977) ubscrv ini; period , 3) variable galactic background emission, and 4) ground radiation entering tin first siJi; lobes for observations at elevations below 20°, which is particularly important at galactic Longitudes less than 50°. Table 2 summarizes the average 3o sensitivity as a function of galactic coordinates. The uncertainties in these values are derived from the day-to-day variations in the cold-sky system temperature and the uncertainty in the measured aperture efficiency. Near the galactic equator the sensitivity could be reduced to 1J 50/5, due to strong galactic continuum sources. This occurred at only a few positions, most notably around Í = 25 in the W43-refiion. Notice that the average sensitivity at >. £ 27° is 20% lower than that at larger longitudes. The stability of the receiver system was checked by daily comparison with two calibration sources. One source was either Virgo A or Cygnus A, which were assumed to have a 1612 MHz flux density of 189.5 Jy and 1369 .Iy, respectively (Baars et al., 1977). The other calibration source, NML Cygnus, was used as a check on the performance of the auLocorrolaLor and the LO-system. No systematic temperature variations of larger than 37. were found. The frequency setting stayed constant within the accuracy of the measurement ('\. 1 kHz). 20

Table 2

Sensitivity of the Dwingeloo Survey

Observed range AS (3o) Jy 10° < i < 27T5 i *•2 0.64 ± 0.05 27.5 < I < 100 l»l < 4.2 0.53 ± 0.05 130 < I < 150 1*1 < 3 0.53 ± 0.05 130 < Í < 150 3 < \b\< 4.2 0.80 ± 0.05

Because of strong (radar?) interference at 1612 MHz about 18% of the total number of 8200 positions had to be reobserved. In those cases the interference covered a considerable part of the bandpass with strong peaks, possibly masking astronomical sources. Another 2000 spectra were also influenced by interference, but this was usually rather weak (5o - 10a), narrowband (10 - 40 kHz) and occurring mainly at the extreme edges of the bandpass, where astronomical sources are unlikely. We are confident that no Type II OH/IR sources were missed because of the interference.

II. 3 Confirmation Observations

Confirmatory 1612 MHz observations at high spectral resolution (4.9 kHz) of the newly-discovered Type II OH/IR sources were carried out with the DRT immediately after the completion of the survey. The telescope was pointed at a position determined from an interpolation of intensities in neighbouring survey-gridpoints and the integration time was chosen according to the expected strength of the source in order to reach a signal-to-noise (S/N) ratio of at least 6. Most of the confirmed sources were then reobserved in August 1977 with the NRAO 43-m telescope in order to obtain a better S/N ratio and in some cases to improve the positions. These observations have been described in Chapter I. Some sources that had been discovered earlier in the Green Bank survey were not reobserved because of lack of time. No observations were made in the OH main-line transitions.

Ill Results

III.1 Analysis of the survey observations

After calibration and removal of a linear baseline the spectra were plotted and inspected by eye. The detection limit varied between 3o and 5a, depending on the width of the feature. All spectral features, both absorption and emission, were then listed and many were found to appear in spectra taken at neighbouring positions. In two regions, one close to 1= 43° (W49) and the other to a = 25 (W43), several features were confused with ne another. The total extent of these regions was not greater than 4 square degrees, and the observations were repeated with a smaller grid separation to resolve the situation. A total of ^ 700 independent absorption and emission features were found, 130 of which were Type II OH/IR candidates, as indicated by the characteristic double peaked point-source emission. In this thesis we are concerned with the Type II OH/IR sources only. 21

771.2 Confirmation of Type II OH/IR Sources

Of the 130 candidate Type II OH/IR sources 32 had been previously detected in the Onsala survey or by Wilson and Barrett (1972) and 24 had been found in the Green Bank survey. The remaining 74 sources were reobserved at Dwingeloo as described above with better resolution: 41 of them were not confirmed, because either (a) no emission could be detected or (b) only one single emission peak was found. The remaining 33 sources show the double peaked emission with the steep outer edges and the more gradual inner slopes. In addition, in each case, the positions of both emission components were found to coincide within the uncertainties. Thus the identification was confirmed in all 33 cases.

One slight uncertainty remains; the discovery of 1665 or 1667 MHz emission stronger than the 1612 MHz emission in these sources would demand their reclnssification. At present no data on the main-line emission is to hand. However, Johansson et al. (1977a) and Bowers (1978a) have searched for main-line emission from their Type II OH/IR sources. None of these were strong main-line emitters and only a handful showed any main-line emission at all. The new sources discussed here, which have the same characteristic 1612 MHz signature, are thus unlikely to have strong main-line emission and we are confident that the designation of Type II is correct in most cases. We used the NRAO 43-m telescope in August 1977 to obtain 1612 MHz spectra for the 33 new sources with a better S/N ratio and a resolution of 6.5 kHz. In addition, 31 Type II OH/IR sources, discovered initially in the Onsala and Green Bank surveys, and a few unclassified objects were observed in the same manner. Thus a large sample of Type II OH/IR sources was obtained, observed at one epoch with a single telescope and with the same spectral resolution. This is important if it is wished to compare emission characteristics.

III. 3 The Neu 'ñjpe II OU/IR Sources

The spectra obtained with the NRAO 43-m telescope are presented in Figures 2 to 34. Table 3 lists the parameters of the sources in order of increasing galactic longitude. The right ascension and declination correspond to the position in galactic coordinates determined from the survey observations. The positional uncertainty in each case depends on the intensity of the source in question. The average positional uncertainty can be determined by comparing the accurate positions of the sources presented by Johansson et al. (1977a) and Bowers (1978b) with the positions of the same sources that we have derived from the survey observations. Ignoring the 15 arc sec positional uncertainty of the sources from Johansson et al. and taking into account the positional uncertainty of 1.25 arc min for the sources from Bowers we derive a root-mean-square uncertainty in each coordinate of ^ 2.5 arc min for the strong and ^ 3.7 arc min for the weaker sources. The radial velocity of the high- and the low-velocity emission peaks, their corresponding

peak flux densities at a resolution of 6.5 kHz, S, , and the integrated flux densities, ST, using

S/TA = 3.6 Jy K are determined from the observations at NRAO. In view of the positional uncertainties, the actual flux densities may be 10 to 20% higher. Enne of the sources require .idditio- nal comments; these are presented below. Oil lO.b+'l.E, OH IS. 1+6.0 (Figs. 2 and 6). These sources lie outside the area indicated in Table 2 and have been found when a few regions at small longitudes were incidentally observed at higher latitudes. OH IV..8-1.9 (Fig. 4). The low-velocity emission component shows two distinctly separate features, of which the inner is the more intense. This could be due to a double slit 11 structure with different expansion velocities. Note the symmetric shape of the feature at -7 km s~ . 22

Table 3

New Type II OH/IR Sources

i b ct(195O) 6(1950) S AS s \SR 6.5 rms I (epoch 1977.6) (epoch 1977.6) km Jy Jy 10-22 W m"2 LV HV LV HV LV HV 10?5 +4?5 17h49?4 -17°4I' - 70 - 43 1.4 1.8 0.1 3.5 2.6 10.9 + 1.5 18 01.1 -18 47 + 1 13 + 145 1. 1 0.4 0.05 1.3 0.5 12.8 -I.9 18 17.6 -18 49 - 7 + 32 5.5 3.6 0.2 16.4 9.6 12.8 -0.9 18 13.7 -18 15 - 67 - 45 5.1 1.4 0.1 12.7 3.6 13.1 +5.0 17 52.9 -15 07 - 82 - 53 7.7 2.3 0.1 9.8 3.6 13.4 +0.8 18 08.9 -16 56 - 20 + 4 0.9 0.7 0.07 1.4 1. 1 15.4 -0. 1 18 16.3 -15 38 - 43 - 13 1.2 2.1 0.1 2.2 3.5 16.1 + 1.4 18 12. I -14 21 + 30 + 65 0.7 1.2 0.06 0.7 1.6 17.0 -0. 1 18 19.3 -14 13 + 35 + 67 2.8 2.5 0.2 6.2 5.3 17.1 -1.2 18 23.5 -14 40 - 8 + 13 1.7 1.0 0.1 4.0 4.5 17.2 -1. 1 18 23.1 -14 29 + 159 + 184 1.1 1.7 0.1 1.8 1.5 IS.3 +0. 1 18 21.0 -12 58 + 38 + 64 0.3 0.3 0.1 0.5 0.3 18.7 + 1.6 18 16.4 -11 53 - 17 + 15 1.7 0.9 0.1 1.9 0.8 19 5 +4 0 18 09.3 -10 04 + 30 + 62 1.6 2.0 0.1 2.5 3.1 20 4 + 1 4 18 20.5 -10 30 + 21 + 56 0.6 1.0 0.07 0.6 1.1 20 6 +0 3 18 24.8 -10 54 + 71 + 111 1.4 1.8 0.2 3.0 4.7 20 8 -0 8 18 28.9 -11 13 + 32 + 63 0.4 0.9 0.1 0.5 1.8 20 8 +3 1 18 15.2 - 9 22 + 13 + 41 2.8 3.9 0.1 5.1 6.1 21 9 +0 4 18 26.8 - 9 40 + 68 + 104 1.4 0.8 0.1 3.1 1 .9 22 3 -2 5 18 38.2 -10 39 + 99 + 126 0.9 0.8 0.1 1.0 0.9 24 7 -1 7 18 39.8 - 8 11 + 79 + 105 1.3 1.3 0.1 1.3 1.6 25 5 +0 4 18 33.9 - 6 28 + 23 + 55 2.0 2.2 0.1 5.5 4.0 26 3 +0 1 1« 36.2 - 5 52 . - 39 - 13 1.6 2.5 0.2 1.9 3.3 26 4 -2 8 18 46.8 - 7 09 - 78 - 54 0.7 1. 1 0.1 1.0 0.9 35 2 -2 .6 19 02.2 + 0 46 + 37 + 65 3.9 3.6 0.1 12.1 15.0 36 .9 + 1.3 18 51.5 + 4 01 - 20 - 5 2.7 2.8 0.1 2.6 2.5 40 .1 +2 ,4 18 53.6 + 7 24 + 29 + 66 1.2 2.0 O.I 1.2 2.6 42 .8 -1 .0 19 10.7 + 8 14 - 28 - 3 0.8 0.7 0.07 1.7 1.6 43 .9 -1 .0 19 27.8 + 9 12 + 39 + 65 0.9 0.9 0.1 1.0 1.1 43.9 + 1.2 19 04.9 + 10 13 + 34 + 66 1.0 0.7 0.1 1.1 0.7 57.5 + 1.8 19 29.5 +22 29 - 87 - 62 1.3 0.9 0.1 2.6 2.9 63.9 -0.2 19 51.1 +27 01 - 10 + 20 1.4 2.2 0.1 2.7 3.3 65 .5 + 1.3 19 49.4 +29 06 - 39 - 5 3.3 3.8 0.1 5.3 7.4

LV = low velocity; HV = high velocity 23

-96 -86 -76 -66 -56 -»6 -36 -25 -16 90 100 110 120 130 ttO ISO 16D 170 VELOCITT KM/S VELOCITT KM'5 Fig. 2 OH 10.5+4.5 Fig. 3 OH 10.9+1.5 .4 0 UR E TEMPEHR T I S O 1.0 0 Í À 0.2 0 . NTEItN R F

2 0 1 1 1 1 1 1 1 t 1 -27 -17 -'( 3 13 23 33 H3 S3 d -95 -85 -7S -6S -55 -HS -35 -25 -IS VELOCITT KM/S Fig. A OH 12.8-1.9 Fig. 5 OH 12.8-0.9

= -110 -1Q0 -90 -80 -70 -60 -50 -HO -30 •uo -so -20 ia a a 20 30 no VELOCITT KH/S VELOCITY KM/S Fig. 6 OH 13.1+5.0 Fig. 7 OH 13.4+0.8

63 -5S -IS -39 -23 -19 -9 1 II

Fig. 8 OH 15.4-0.1 Fig. 9 OH 16.1+1.4 24

r— 1 n— 1 1 l —r i i " tii a iu o CC 4 E » \ * ft CE f X cc o 51 Ü4 °¡ - / O. B V D S3 \ /- A /VV^/X CE a J CE a «v y/ Z Z O Z a - - NT Í -0 .

i , 1 , 7 i i i i i i i 20 30 40 50 SO 70 BO SO -37 -27 -17 -T 3 13 23 33 43 ! VELOCITT KM/S VELOCITT KM/S Fig. 10 OH 17.0-0.1 Fig. II OH 17.1-1.2 ! iV/" SjJ w v

1 _J 1 1 ! I ! 1 L_ 129 139 1*49 1S9 169 179 183 199 209 9 19 29 39 49 59 69 79 VEUQCITT KM/5 VELQCIIT KM/S Fig. 12 OH 17.2-1.1 FÍR. 13 OH 18.3+0.1

iu a n tr I

A v TEMP I a . A \ (X d V J Vr

a 1 1 f i t 1 I t 13 23 33 43 15 25 35 45 55 65 75 85 VELOCITT KH/S VELOCITT KM/S Fig. 14 OH 18.7+1.6 Fig. 15 OH 19.5+4.0

2 ÍS 22 32 42 52 62 72 62 50 60 70 60 90 100 110 120 130 VELOCITT KM/S Fig. 16 OH 20.4+1.4 Fig. 17 OH 20.6+0.3 RNTENNR TEMPERRTURE 2 HNTENNR TEHPERRTURE -0.10 0.10 0.30 0.50 flNTENNfl TEHPERRTURE ONTENNR TEHPEHRTURE ? -coa o.oi o.n o.ai o.ai o.is -coa o.oa o.ia D.2B ni -0.06 o.o; n.m o. IB 1 1 r~

2 RNTENNfl TEHPEflHTURE HNTENNR TEHPERRTURE 2 RNTENNH TEHPEflHTURE RNTENNfl TEMPERRTURE V -0.I2 -o.oa o.PB cíe o.aa ao o.oo 0.20 o.ya o.6o ™ -o.ia -con con o.ia O.ÍO V -o.i.ao -D.ao a.on o.iio o.eo

f 26

o so 30 «o so so 70 no -si -m -si -si -u -i VELOCITY KM/S VELOCITY KM/S Fig. 26 °H 35.2-2.6 Fig. 27 OH 36.9-H.3

9 19 29 39 19 59 59 79 89 -56 -45 -36 -26 -16 -6 U 14 24 VELOCITY KM/S Fig. 28 OH 40.1+2.4 Fig. 29 OH 42.8-1.0

0 20 30 40 50 60 70 80 20 30 40 50 60 70 60 90 VELOCITY KH/S VELOCITY KM/S Fig. 30 OH 43.9-1.0 Fig. 31 OH 43.3+1.2

o -112 -102 -92 -82 -72 -62 -52 -42 -32

Fig. 33 OH 63.9-0.2 27

Figs. 2 to 34. 1612 MHz profiles of new Type II OH/TR sources discovered in the Dwingeloo survey. These spectra were taken with the NRAO 43 m telescope and have a resolution cf 6.5 kHz. The ratio of flux density to antenna temperature is 3.6 Jy K . Velocities are re- -JO -59 -IS -» VELOCITY KM/S ferred to the l.s.r. (see text). Fie. 34 OH 65.5*1.3

OH 17.0-0.1 (Fig. 10). The weak absorption at +48 km s is probably real; it can also be seen in the survey at the neighbouring positions 2. = 16?9, b = 0?3 and Í = I7?2, h = -0?6. Judging from the close velocity agreement between the Type II OH/IR source and the absorption feature they may be physically close. Main line observations with high spatial resolution are needed to investigate this source. OH 17.1-1.S (Fig. II). The emission extends over all velocities between the two emission components, in common with several other Type II OH/IR sources (Andersson et al., 1974, Paper I). The high-velocity component is rather flat-topped; in Table 3 we have chosen the +13 km s feature for consistency although it is only marginally stronger than the feature at +16 km s . OH 18.3+0.1 (Fig. 13). This is a very confused region of sky centered on W39. The survey observations show several sources, most notably OH 18.2+0.4 (Winnberg et al., I973) at +32.5 and +63.5 km s ; OH 18.8+0.4 (Turner, 1970) at -2 and +27 km s~' and the Type II(b) source OH 18.3+ 0.1, whose emission features at +11 and +17 km s do not coincide spatially (Johansson et al., 1977a). The latter source is visible in the figure. The confirmatory observations with the DRT found another very weak source (OH 18.3+0.1) at +38 and +64 km s~' at the same position. Note that the high velocity components of both OH 18.3+0.1 and OH 18.2+0.4 coincide in velocity. It is unlikely that they are the same source, because the 19 arc min difference in position between the two sources means that OH 18.2+0.4 is well below the noise of the spectrum in Figure 13. Observations with better sensitivity are needed to confirm the positional coincidence of the +38 and the +64 km s components. OH Ï0.6+0.S (Fig. 17). The feature at +118 km s~ is the low-velocity component of OH 20.7+ 0,. 1, first discovered by Johansson et al. (1977a). The confirmatory observations with the DRT show that the high-velocity component consists of two blended features, separated by about 3 km s . This can also be seen in the spectrum presented here. OH 20.tí-0.S (Fig. 18). The absorption at +47 km s~ is extended and clearly visible in the survey observations at tlip positions i = 20?5, b =-0?6 and Í = 20?8, b =-0?6. At about the same velocity Downes (1970) fmind the nearby Type 1 maser OH 20.1-n.l, indicating that this is a region of active star formation. OH 21.9+0.1 (Fig. 20). The emission feature at +96 km s~' belongs to OH 21.5+0.5, which was discovered by Winnberg et al. (1973). The high-velocity component of OH 21.9+0.4 consists of the emission features at +100 and +104 km s~'. The integrated flux density has been derived from these features.

OH ¿-ff.i+ . / (Fig. 24). The low-velocity component of OH 26.5+0.6 (Andersson e al., 1974) is visible at +14 km s 28

OH -iO. 1+2.4 (Fig. 28). The weak feature at +52 km s is also seen in the Dwingeloo confirmatory observations. Its relation to OH 40.1+2.4 is not clear. OH ¿7.5+1.8 (Fig. 32). The emission between -72 and -82 km s is probably due to instrumental baseline curvature.

III.4 Known Type II OH/IR Sources

Table 4 lists the previously known Type II OH/IR sources that were observed with the NRAO 43 m telescope in August 1977, in order of increasing galactic longitude. Columns 2 and 3 give the peak 1612 MHz flux densities of the low- and the high-velocity components; the uncertainty in these values is given in column 4. Corresponding integrated flux densities are listed in columns 5 and 6. For all sources, except OH 42.3-0.1, the peak-velocity agreed with previous authors. A 6.5 kHz resolution profile has been presented for those sources that have either been observed before with a lower spectral resolution and sensitivity or show interesting changes in profile shape since previous observations. The corresponding figure number is noted in column 7. The best known position of the source, at which the spectrum was taken, was provided by the authors referenced in column 8. The sources marked with an asterisk were independently discovered in the Dwingeloo survey.

We have reobserved as many known Type II OH/IR sources as possible for the following reasons. Firstly, the spectral resolution influences the measured peak flux densities. As is clear from the profiles presented by Johansson et al. (1977a) many emission features are just resolved with a spectral resolution of 1 kHz (y 0.2 km s ), and would certainly be unresolved when observed with a resolution of 6.5 kHz, resulting in a lower peak flux density. Before we may compare the results from the different surveys, we have to account for the differences in spectral resolution. A rough estimate of the ratio between the peak flux density at I kHz resolution, S., from Johansson et al. (1977a) or at 13 kHz resolution, S..., from Bowers (1978a) and the peak flux density at 6.5 kHz resolution, S, ,., can be obtained by comparing the results in Table 4 S with those obtained by the other authors. This yields the following results: •-* = 0.7 and h ^ ^ I „ = 1.3. These are only mean values and the actual values for individual sources may differ by up to 30Z: for example, OH 18.5+1.4 was first discovered by Bowers (1978a) to have equal S., values for both velocity components. Our data (Fig. 35) show that the peak ratio between the two components at a resolution of 6.5 kHz is 1.4. However, the ratio of the integrated flux densities does not vary, thus excluding significant independent variability of the two components. The narrowness of.the low-velocity component suggests that it Is more likely to be due to the difference in spectral resolution. Secondly, many Type II OH/IR sources are variable, with a typical period of several hundred days. In order to determine the character of the variability, its period and its amplitude, it is necessary to observe these sources regularly at different epochs. Because of the above mentioned spectral resolution problem, we have compared the integrated intensities in Table 4 with those at earlier epochs (Bowers, 1978b). The mean difference of 17% is negligible in view of the uncertainties, which are 15-20% for Bowers' values and 5-10% for our values. Twenty-eight sources show a variability of more than 30£ in one or both velocit> components. In some cases the difference in the integrated flux densities is partly caused by the appearance of broad, low intensity wings in our profiles, which have a better signal-to-ncise ratio. However, it is difficult to be very exact in this matter since baseline curvature can be important. 29

Table 4

1612 MHz Flux Densities (epoch 1977.6) of Previously Known Type II OH/IR Sources

Ref. Name °6.5 ASrms Fig.

22 2 Jy Jy 10 W m

OH 18.5+1.4+ 5.8 4.1 0.1 9.7 10.8 35 4 OH 24.3+0.7+ 0.7 4.3 0.1 1.6 6.7 36 1 OH 24.5+0.3+ 6.8 2.3 0.2 6.6 5.1 37 4 OH 24.7+0.3+ 3.7 1.9 0.1 7.8 5.4 38 4 OH 27.2+0.2+ 1.1 1.3 0.1 0.9 1.8 39 4 OH 32. 1+0.9 1.4 2.5 0.1 1.8 3.3 3 OH 34.7+0.9 1.8 2.,1 0.1 3.2 3.9 40 3 OH 36.4+0.3 2.7 1.5 0.1 3.7 3.5 41 3 OH 37.,1-0.,8+ 8.5 13.,8 0.2 16.7 18.3 3 OH 37.,7-1.,4 1.,0 0,.4 0.1 2.1 1.3 3 OH 39..7+1. 49.,8 74..1 0.3 76.8 121.9 4 OH 39..9-0..0 3..4 4..8 0.1 6.0 6.3 3 OH 42..3-0,.1 4,.9* 20,.3 0.1 17.7 32.5 42 3 OH 42,.6+0,.o' 3,.6 3.2 0.1 5.7 5.0 43 3 OH 43,.6-0,.5 1,.7 1,.3 0.1 1.3 1.5 3 OH 43,.8+0 2,.9 1.9 0.1 3.2 3.1 3 OH 44.8-2 .3 b,.3 1 1.2 0.1 7.6 9.1 44 4 OH 45.5+0.1H 5.5 4.4 0.1 6.1 5.0 2 OH 51.8-0.2H 5.8 3.9 0.1 l 1.6 9.2 45 4 OH 52.4+1 .8 2.7 2.5 0.1 2.8 3.2 4 OH 53.6-0.2 6.6 14.8 0.1 9.2 18.7 46 4 OH 55.0+0.l' .5 2.3 0.2 6.5 3.7 4 OH 65.7-0.8" 1.0 2.3 0. 1 1.7 2.1 47 4 OH 66.8-1 .3' 1.M 0.9 0.07 1.4 1.6 48 4 OH 75.3-1 .8' h , -4 5.3 0.1 8.3 8.0 49 4 OH 83.4-0.9' 12. 4 2.9 0.2 13.2 4.6 50 4 OH 85.4+0.1 1.7 1.7 0.1 2.6 1 .6 51 4 OH 104.9+2.4 41.3 37.5 0.2 76.6 57.6 52 4 OH 127.8+0.0 28.8 8.5 0.2 28.8 17.9 53 4 OH 138.0+7 _ 2 9.7 7.1 0.2 10.0 6.6 54 4 OH 141.7+3 5 . 1 1.0 0.1 4.2 1.5 55 4

Determined at +43 km s

1 - Wilson and Barren (1972) 2 - Uinnberg et al. (1973) 3 - Johan.sson et al. (1977a) 4 - Buwc-rs (1978a) 30

159 l«9 15a 189 119 199 IBS 209 219 20 90 Id 50 10 70 10 90 100 VELnciTT KH/S VELOCITY KM/S FÍ8- 35 OH 18.5+1.4 Fig. 36 OH 24.3*0.7

S 19 23 S3 Hi S3 65 73 S3

Fin- 37 OH 24.5*0.3 Fig. 3B OH 24.7*0.3

bJ a E V - -

£s cc o Ä a lit °* - Z ? CE

sa ia 70 on ^ loo no 12a 130 a 10 20 3B US SB 6B VELOCITT KH/S VELOCITY KH/5 Fig. 39 OH 27.2+0.2 Fig. 40 OH 34.7+0.9

BO 70 83 90 100 110 120 130 140 20 30 MO 50 BO 70 BO SO 100 VELOCITY KM/S VELOCITT KH/S Fig. 41 OH 36.4+D.3 Fig 42 OH 42.3-0.1 31

1 1 r——• r i 1— "1 1 1

ë§ • - 2» 1 X ïïrf tu i-» CE «¿ Z V . Jl 20 3Q

fVA .y-Vv -J l I 1 I l I ' ' -37 -27 -17 -7 3 13 23 33 13 VELOCITT KM/S Fig. 45 OH 51.8-0.2 Fig. 46 OH 53.6-0.2

„ l\ A.A n>A l\l 5°^ f -100 -90 -BO -70 -60 -SO -10 -30 -20 VELOCITT KM/S VELOCITT KM/S Fig. 47 OH 65.7-0.1 Fig. 48 011 66.B-1.3

i 1 1 l [ —I

cc =• -

2-9 0 1 HUR E

!° - 2.1 0

I EMPER f - or ca cc — z z o - • z° V / w 2 1 ( 1 1 1 1 ik | 1 1 tl "i i i U3 -33 -S3 -13 -3 7 17 27 37 Cl -76 -00 -30 -l| ! -30 -20 -1 -G 1 VELOCITT KM/S VELOCITY HM/S Fig. 49 OH 75.3-1. >H 1)11 81.4-11 'I 32

tu §" 6

MPEBB T 1 - 3 - - ENN O T E

HN T VJ tn i i i i i | | i i -52 -52 -142 '32 -22 -12 -13 -33 -23 -13 VEIOCITT KH/S VELOCITT KM/S Fig. 51 Oil 85.4+0.1 Fig. 52 Oil 104.9*2.4

1 1

-92 -8S -72 -62 -52 -tl -32 -22 -12 -55 -15 -3S -25 VELOCnr KH/S

Fig. 53 OH 127.8*0.0 Fie. 54 OH 138.0+7.2

VELOCITY KM/5 Fig. 55 OH 141.7+3.5

Figs. SS to 55. 1612 MHz profiles (epoch 1977.8) of previously knoun Type II OII/IR goiwces. The details are as for Figs. 2 to 34. Referenaes to individual objects are to be found in the text and Table 4.

Some comments on individual sources follow: OH 24.S+0.7 (IRC-10434) (Fig. 36). A second feature in the low velocity component at a velocity of +47 km s has appeared since its discovery by Wilson and Barrett (1972). OH 24.7+0.3 (Fig. 38). The emission features at +42 and +72 km s~' belong to OH 24.7+0.1. OH 27.2+0.2 (Fig. 39). Bowers (1978a) discovered this weak source with features at +73 and +112 km s ; the low-velocity component is rather faint in the published spectrum. The emission feature at +63 km s~ belongs to OH 27.3+0.2 (Wilson and .Barrett, 1972). OH 34.7+0.9 (Fig. 40). This source showed a very broad low-velocity component when it was first- detp^ted (Johansson et al., 1977a), wich emission extending between +10 and +30 km s~'. Our profile shows no emission above +18 km s~ . The shape of the high velocity component does not appear to have changed markedly. OH cò.4+0.¿ (Fig. 41). The high-velocity component of this source has increased in intensity relative to the low velocity component, since the discovery by Johansson et al. (1977a). The emission feature at +90 km s is visible also in their published spectrum. OK 42.S-0.1 (Fig. 42). This source was discovered by Johansson et al. (1977a). Johansson (private communication) has pointed out that their velocity peale is in error for the low velocity component. The correct scale (shifted by +5.6 km s ) gives excellent agreement with both our data and those of Bowers (1978a). OH 4",6+0.0 (Fig. 43). This source was first detected by Johansson et al. (1977a). Their published profile, taken with a low signal-to-noise ratio, does not show the wing of emission in the low velocity component up to +45 km s . The weak emission at +75 km s probably arises from OH 42.3-0.1. -1 OH 66.8-1.S (Fig. 48). The profile published by Bowers (1978a) shows emission at -60 km s hetween the two velocity components. The present observations, made with a better signal-to-noise ratio do not confirm this. OH 33.4—0.9 (Fig. 50). Our observations indicate a second feature in the low-velocity component at -52 km s . This is apparent as an emission wing in the profile published by Bowers (1978a). OH 104.9+8.4 (Fig. 52). Our observations clearly indicate that the low-velocity component consists of at least two features, at -40 and -36 km s OH 127.8+0.0 (Fig. 53). This Type II OH/IR source was discovered by Kerr and Bowers (1974) at epoch 1973.6 and was reobserved by Bowers (1978a) at epochs 1976.5 and 1976.8. Using the period -AV relation from Dickinson et al. (1975), Bowers suggested that the intensity variations between the first three epochs were consistent with a period of 540 days and a ratio between maximum and n.'nimum Elux of 5. The integrated flux density at 1977.6 is almost twice the value that would be expected Erom such a lightcurve. We derive a period of 640 days or 510 days from phase and amplitude of the last three epochs and the amplitude of the first epoch. The expected phase at 1973.6 extrapolated from the latter three epochs difieres by 90 . We conclude that either the phase or the amplitude (or both) of the variation of OH 127.8+0.0 has changed between epochs 1973.6 and 1977.6.

III. 5 Unclassified Goicraes

Several candidate Type II OH/IR sources, initially discovered in the Dwingeloo survey, were reobserved at 1612 MHz with the NRAO 43 m telescope. The shape of their emission profile did not justify classification as Type II OH/IR. Main-line observations should be made to determine their nature. These sources and a previously known Type II(b) source, denoted by their galactic coordinates, are listed in Table 5, along with their peak flux density and corresponding radial velocity. A 1612 MHz profile is presented for each case. They are discussed below. OH 12.6+0.2 (Fig. 56). The shape of the feature at +169 km s~' is similar to that of the low velocity component of a Type II OH/IR source. Another feature appears at a velocity of +153 km s for which no position was determined. This source could be a Type I maser with a relatively strong 1612 MHz line although its radial velocity is unusually high for such a young object. All H II regions observed near this longitude (Mezger et al., 1970) have radial velocities well below +100 km s~'. OH 16.S-S.0 (Fig. 57). The character of this profile is remarkable. It has a symmetrical double structure, but the individual velocity components are reversed with respect to those nor- mally found in the spectra of Type II OH/IR sources. They appear to be coincident within the positional uncertainties and no other emission has been detected between -180 and +280 km s . This 34

142 iso isa ÍES 174 ias 190 laa 20s 20 3(1 HO VEUOCITT KM/S VELOCITT KM/S Fig. 56 Oll 12.6*0.2 Fiß. 57 Oll 16.3-3.0

-149 -139 -129 -119 -109 -99 -99 -79 30 40 50 SO VELDCITY KM/S VELOCITT KH/S Fig. 58 0.11 17.2-0.2 Fig. 59 OH 20.2*3.0

I 1 1 1 —i —i—i— i —

tu o ÜJ o

i r Sir »_ _a: a 1 1. - a. *¿ a: ui c LA i- 2 - - D.8 D - .— A - - " 5s r Is V.J\. v cr V \V fs. •a v * r i i i t -e 4 14 21 34 44 • i i -40 -30 -20 10 0 10 VELOCITT KM/S VELOCITT KM/S Fie. 60 01! 43.2*0.0 Fig. 61 OH 56.4-0.3 Figs. SB to 61. 161" Mffs profiles (epooh 1977.6) of the unclassified sources. The details aín- as for Figs. 2 to S4. See also the text and Table 5. Table 5 Unclassified Sources b a(1950) 6(1950) AS VLSR S6.5 rms (epoch 1977.6) km s Jy Jy 1296 +092 18h 9 4 -17°56' + 169 1.3 0.07 ° - + 153 0.3 16.3 -3.0 18 28.6 -16 1 1 + 21 2.0 0. 1 + 36 4.0 17.2 -0.2 18 20.0 -14 04 -100 1.2 0.06 20.2 +3.0 18 14.4 - 9 55 + 32 0.4 0.06 43.2 +0.0 19 07.9 + 9 00 + 15 9.8 + 36 2.0 0.3 56.4 -0.3 19 35.2+ +20 33 - 3 6.8 + 0 3.7 0.1 + 4 3.5 Position taken from Bowers (19/8a). 35

suggests that they are not a coincidental arrangement of the high (+21 km s ) and the low-velocity (+36 km s~') component of two different Type II OH/IR sources. Judging from the narrow emission peaks, the overall velocity range and the very weak emission between the peaks the object is similar to a Type II OH/IR source. This could be verified by 1612 MHz observations at different epochs to determine variability, and by main-line observations. Olnon (1977) has studied various velocity fields in the circumstellar shell where the masing emission originates, and their influence on the shape of the emission profile. Whereas the typical shape, with gradual inner slopes, corresponds to an expansion velocity that is the same at all distances from the star, the profile of OH 16.3-3.0 is more indicative of decelerating expansion velocities.

Oil 17.2-0.2 (Fig. 58). The single, narrow (< 5 km s ) and sharply-peaked profile suggests that it is a maser. The large negative velocity indicates that the feature is probably associated with an old stellar object. OH P.O.2+S.O (Fig. 59). This could be a velocity component of a Type II OH/IR object. Because of the high latitude it is unlikely to be satellite line emission of a Type I maser. OH JS.S+O.O (Fig. 60). This source lies in the complicated W49 region. The emission feature at +15 km s is well known (Palmer and Zuckerman, 1967; Weaver et al., 1968). It is strongly circularly polarized (Robinson et al., 1970). The emission at +36 km s was not found by the previous authors due to insufficient velocity coverage. The emission and the absorption features are blended and the survey observations show that they vary rapidly in intensity across the sky. Observations with better angular resolution are needed to determine the position of the individual features. OH eb'.J-O.¿ (Fig. 61). This Type II(b) source was first detected by Bowers (1978a) who noted ti'at the shape of the 1612 MHz line is similar to that of the peculiar Type II OH/IR star U Ori (Reid et al., 1977), so that the emission may be associated with a similar object. The present observations show at least 4 emission features within a range of 17 km s . The profile shape has not changed significantly since the earlier observations. The integrated intensity has however -y> _?9 —9 changed from 16.31.0.5 x 10 to 21.7±0.5 x 10 W m since 1976.5. This variation indicates

that the source is probably associated with a stellar object.

IV Discussion Any spectral line survey is a sampling in 4-dimensional parameter space: two positional coordinates, radial velocity and flux density. If the sources are variable the epoch of the observa- tion will enter as a fifth parameter. For an unbiased statistical appraisal of the results, it is necessary to estimate the completeness of the survey with respect to each of the four parameters. It is clear from the range of positions and radial velocities in Table 3 that the latitude extent and the radial velocity coverage is fairly complete: i.e. veiy tew sources have been found at high latitudes and all velocities lie well within the surveyed range. We have therefore assumed that the present survey is complete in radial velocity and latitude coverages. The probability of detecting a source with a certain flux density at a given spectral resolution depends on llio sensitivity of the survey, and the ratio between the HPBW and the gridpoint separation. When this ratio is 2, according to Nyquist's sampling theorem the detection probability is equal to unity for all sources with flux densities larger than the detection limit, i.e. the survey is complete to this limit. A smaller ratio (larger relative gridpoit.t i.naration) results in an undersampling of the weak sources and a higher completeness limit. We have calculated the detection probability as a function of the ppzU flux measured with a resolution of 11.7 kHz 36

—1— i —I —1 1——1 1 Fig. 62. The detection probability for Type II OH/IR sources as a function of 1.0 _ Dwingeloo 11 7lfH2 - /^~ i i.f un»-' -^^ flvx density (in Jy) for the present 10Jy i survey and the Onsala (Johansson et al., »0.8 / 10kHz. 1977a) and Green Bank (Bowers, 1978a) 1.0 Jy r~- Green Bank surveys. The speotral resolution and £a Jbili i i o 0.6 / / / 7.9 kHz. - detection limit ave indicated iïi each / / / 2.5 Jy case. See text, Section .!', c ƒ / / o / // / 5 0.4 / / 2 / / / o J1 / 0.2 /

j¿£S i 1 - -0.2 0.0 0.2 0.4 0.6 0.8 1.0 logS

and a 5a detection limit of 1 Jy with the method described by Johansson et al. (1977b). The results are shown in Figure 62; for comparison, similar curves for the Onsala and Green Bank surveys are shown. The limit of completeness for the Dwingeloo survey is 1.8 Jy. This is a considerable improvement with respect to the Onsala survey, which has the same 3a sensitivity. The difference is due to the improved sky coverage of the present survey: the ratio between HPBH and gridpoint separation is 1.7 as compared to 1 for the Onsala survey. Because most emission features of Type II OH/IR sources have narrow velocity components with respect to the spectral resolution used in the three surveys, the probability for detecting Type II OH/IR emission increases with increasing resolution. From the comparison between peak flux densities at different spectral resolutions in the previous section we estimate that the sensitivity and correspondingly the completeness of the present survey, scaled to the spectral resolution of 10 kHz used in the Onsala survey, would be less than 10%. The improved completeness of the present survey is consistent with the fact that out of a total of 51 Type II OH/IR sources discovered in the Onsala and the Green Bank surveys in areas common to the Dwingeloo survey only three sources were not rediscovered. These are OH 24.7+0.1, OH 37.7-1.4 and OH 43.6-0.5 and are among the weaker 1612 MHz emitters. If we take into account that many of these sources are variable, with an amplitude variation of a factor of about 4, and that their positions do not in general coincide with survey grid points, this is a very satisfactory result. Part of the Onsala region (hatched area in Fig. 1) was not re-observed in the Dwingeloo survey. Extrapolating from the number of new Type II OH/IR sources that we^e discovered in the region covered by both the Onsala and the Dwingeloo surveys we conclude that we would have detected 7 more Type II OH/IR sources between I = 27° and 5, 42 had this region been included in the Dwingeloo survey.

This chapter is submitted as a separate article to Astronomy and Astrophysics Supplement Series with Drs. H.J. Habing, H.E. Matthews and A. Winnberg as co-authors. 37

References

Altenhof f, W.J., Dowries, D., Goad, L., Maxwell, A., Rinehart, R. 1970, Astron. Astrophys. Suppl. U 319. Andersson, C., Johansson, L.E.B., Goss, W.M., Winnberg, A., N-Q-Rieu 1974, Astron. Astrophys. 30, 475, Baars, J.W.M., Genzel, R., Pauliny-Toth, I.I.K., Witzel, A. 1977, Astron. Astrophys. 6J_, 99. Ball, J.A. 1976, Methods of Experimental Physics Jj?C, 46. Baud, B., Habing, H.J., Matthews, H.E., Winnberg, A. 1978, Astron. Astrophys. Suppl., in press (Paper I). Bos, A. 1976, Netherlands Foundation for Radio Astronomy, Internal Technical Report 142. Bowers, P.F. 1978a, Astron. Astrophys. Suppl., 21. 127. Bowers, P.F. 1978b, Astron. Astrophys.,in press. Caswell, J.L., Haynes, R.F. 1975, Monthly Notices Roy. Astron. Soe. J_73_, 649. Dickinson, D.F., Kollberg, E., Yngvesson, S. 1975, Astrophys. J. J_99, 131. Evans, N.J. II, Beckwith, S. 1977, Astrophys. J. 2T7, 729. Feast, M.W. 1963, Monthly Notices Roy. Astron. Soc. J_25_, 367. Johansson, L.E.B., Andersson, C, Goss, W.M., Winnberg, A. 1977a, Astron. Astrophys. Suppl. 2!3, 199. Johansson, L.E.B., Andersson, C, Goss, W.M., Winnberg, A. 1977b, Astron. Astrophys. _54, 323. Kerr, F.J., Bowers, P.F. 1974, Astron. Astrophys. _3jl» 225- Mezger, P.G., Wilson, T.L., Gardner, F.F., Milne, D.K. 1970, Astron. Astrophys. 4_, 96. Olnon, F.M. 1977, Thesis, Leiden University. O'Sullivan, J.D. 1975, Netherlands Foundation for Radio Astronomy, Internal Technical Report 138. Palmer, P., Zuckerman, B. 1976, Astrophys. J. J_48, 727. Reid, M.J., Muhleman, D.O., Moran, J.M., Johnston, K.J., jehwartz, P.R. 1977, Astrophys. J. 214, 60. Kobinson, B.J., Goss, W.U., Manchester, R.N. 1970, Australian J. Phys. 21, 363. Schultz, G.V., Kreysa, E., Sherwood, W.A. 1976, Astron. Astrophys. J>0, 171. Tenkink, J., Slottje, C. 1477, Netherlands Foundation for Radio Astronomy, Note 239. Weaver, H., Dieter, N. II. , Williams, D.R.W. 1968, Astrophys. J. Suppl. _[6, 219. Wilson, W.J., Barrett, A.H. 1972, Astron. Astrophys. J_7, 385. Winnberg, A., Goss, W.M., Höglund, B., Johansson, L.E.B. 1973, Astrophys. Letters 13, 125. 38

CHAPTER III: GALACTIC DISTRIBUTION AND EMISSION PROPERTIES OF OH/IR SOURCES

Introduction

Large scale 1612 MHz OH surveys have revealed the existence of many Type II OH/IR sources (Caswell and Haynes, 1975; Johansson et al., 1977a; Bowers, 1978a). These sources are classified according to their characteristic double-peaked emission line profile with steep edges at the outside and more gradual slopes inside. The maser emission is strongest at 1612 MHz although in some cases weak mean-line emission occurs. Wilson and Barrett (1972) discovered several sources associated with nearby Mira-variables, M-type supergiants and IR point-sources from the IRC survey. The double-peaked emission, with a velocity separation AV between the two peaks of 20 to 50 km s , originates in the expanding circumstellar shell around these stars and indicates a considerable stellar mass-loss. This agrees with the fact that the objects show heavy intrinsic reddening (Hyland et al., 1972). The majority of known Type II OH/IR sources cannot be identified optically but probably all show IR point-source emission (Schultz et al., 1976; Glass, 1978), with the photometric and spectral characteristics of a stellar source embedded inside a cool dust shell (Evans and Beckwith, 1977). The stl iig concentration of the unidentified sources towards the galactic plane and to the inner parts of the Galaxy (Caswell, 1974; Johansson et al., 1977b, hereafter designated JAGW; Bowers, 1978b) suggests that they are situated at large distances from the Sun. Hence it appears possible to study the spatial distribution and kinematics of a class of stellar objects throughout the Galaxy by radio astronomical means, i.e. unhindered by interstellar extinction. Neverthe- less, the existing analyses of the known Type II OH/IR sources (JAGW; Bowers, 1978b) remained limited by the relatively small number of known sources. A major question that has so far remained unresolved concerns the kinematics of the unidentified Type II OH/IR sources: although the sources appear to follow galactic rotation, several show radial velocities that deviate from circular motions by amounts up to 100 km s (Kerr and Bowers, 1974). Therefore the kind of stellar objects with which the sources are associated remains uncertain. A second standing problem is the suggestion from previous analyses that the 1612 MHz luminosities of the unidentified sources are much larger than the luminosities of the nearby, identified objects, that are otherwise similar in their OH emission properties.

In this chapter we shall analyse the large-scale galactic distribution and kinematics of a sample of Type II OH/IR sources consisting of all known sources at 10° < I < 150°, \b\ ^ 4°. This region of sky covers the major part of the Galaxy visible in the northern hemisphere and it has been extensively studied in various systematic surveys. Our data base consists of the results of the 1612 MHz Dwingeloo survey, presented in Chapter II, supplemented with the results of the Onsala survey (Johansson et al., 1977a) and the Green Bank survey (Bowers, 1978a). It includes a total of 114 Type II OH/IR sources. We will not make a distinction between identified and unidentified sources, because, as we will show, their differences can be interpreted in terms of the observed extremes of a continuous variation in emission properties. Section II summarizes the Dwingeloo survey and presents the final source list, including all 114 sources. In section III we analyse the velocity distribution of the Type II OH/IR sources and their angular distribution on the sky. In section IV a luminosity function is derived; a comparison between the far and nearby sources is made in section •J. The results of sections III and IV are used in section VI to derive a model of the galactic distribution of Type II OH/IR sources. The final results are discussed in section VII. Throughout this chapter we will refer'to the objects as "OH/IR sources", implying that they 39

are all of Type II and thus strongest at 1612 MHz. We use a cylindrical coordinate system (R, 9, z) centered at the Galactic Centre. The Sun is at (10 kpc, 0, 0).

II Observations and basic data

The Dwingeloo survey covered an area of sky, more or less symmetrical with respect to the galactic equator, between 4 = 10° and 100° and between i = 130° and 150°. The latitude extent varies for different longitude intervals, as is shown in Chapter II, Fig. 1; it avoids a region 27° < I < 42° on both sides of the equator, because this was covered previously by the Onsala and Creen Bank surveys. The Dwingeloo region was systematically covered using a grid point separation of 0?3. The half-power beamwidth of the Dwingeloo Radio Telescope is 0?5. The 30 sensitivity was 0.64 ± 0.05 Jy at S. < 27° and 0.53 ± 0.05 Jy at í > 42°. The total velocity coverage was % 465 km s~ . We rediscovered 48 out of the 51 OH/IR sources, known from the Onsala and Green Bank surveys. One source from the Galactic Centre survey (Chapter I) was also found. In addition 38 new sources were discovered. Most of the new sources are weak ones, discovered inside the extent of the On- sala and Green Bank surveys. As discussed in Chapter II their discovery is due to the fact that the Dwingeloo survey remains complete to lower flux density levels than the previous surveys. The completeness of a systematic survey can be expressed in terms of the probability P(S) of detecting a source with peak flux density S. The completeness limit (the minimum value of S for which P(S) ^ 0.99) of the Dwingeloo survey was calculated to be 1.8 Jy at S. < 27° and 1.5 Jy at Í > 42D, which may be compared to 3.5 Jy and 9 Jy for the Onsala and Green Bank surveys respectively (see Fig. 63 in Chapter II). The new sources discovered in the Dwingeloo survey and a few others, found in the Onsala and Green Bank surveys, were subsequently reobserved with the NRA0 43 m telescope with higher spectral resolution (6.5 kHz) and a better signal-to-noise ratio. None of the new OH/IR sources have obvious optical counterparts on the Palomar Sky Survey. For an analysis of the galactic distribution and of the emission characteristics we will use those OH/IR sources that have been discovered in the Dwingeloo survey between S, = 10 and 150 and \b\ < 4?2. But because of the improved completeness of this survey we also include all other OH/IR sources found in the Onsala and Green Bank surveys outside the extent of the Dwingeloo survey. In addition a few identified OH/IR sources, discovered by Wilson and Barrett (1972) and rediscovered in the Dwingeloo survey, have been included. We will show that they are similar to the unidentified OH/IR sources. Because Bowers (1978a) found no OH/IR sources at I > 150°, the total sample of 114 sources includes almost all known OH/IR objects at 10° < i < 180° within % 4° from the galactic equator. They are listed in Table 1 in order of increasing longitude. Their radial velocity V, defined as the mean radial velocity between the two emission peaks and measured with respect to the l.s.r., and the velocity separation AV are denoted in columns 2 and 3. In columns 4, 5, 6 and 7 the peak flux density S, ^, measured with a resolution of 6.5 kHz, and the integrated flux S of the low- (LV) and high-velocity (HV) components, respectively, are listed. All these parameters have been

obtained from the authors referenced in column 8. The values for S, _ re derived in a manner discussed in section IV. 1. D. J

The National Radio Astronomy Observatory is operated by Associated Universities, Inc., under contract with the National Science Foundation. 40

Table 1

The Type II OH/IR Sources at l Ï 10

Name AV a6.5 Ref. - km s-1 km s-1 Jy 10 W m LV+ HV HV OH 10.5+4.5 - 57 27 1.4 1.8 3.5 2.6 1 OH 10.9+1.5 + 129 32 1.1 0.4 1.3 0.5 1 OH 11.5+U. 1 + 42 44 4.2 25.0 18.2 111.4 2 OH 12.3-0.2 + 37 27 3.9 6.5 5.3 7.1 4,5 OH 12.8-1.9 + 13 39 5.5 3.6 16.4 9.6 1 OH 12.8+0.9 + 27 22 6.5 6.5 13.4 13.1 4,5 OH 12.8-0.9 - 56 22 5.1 1.4 12.7 3.6 I Oil 13. 1+5.0 - 68 29 7.7 Z.3 9.8 3.6 1 OH 13.4+0.8 - 8 24 0.9 0.7 1.4 1.1 1 OH 15.4-0. 1 - 28 30 1.2 2.1 2.2 3.5 1 OH 15..4+1.9 + 12 29 3.9 3.9 4.6 3.9 A,5 OH 15..7+0.,8 - 1 28 28.6 16.9 45.8 24.4 4,5 OH 16.. 1+1,.4 + 48 35 0.7 1.2 0.7 1.6 1 OH 16..1-0..3 + 23 41 11.7 15.6 41.8 41.4 4,5 OH 17..0-0..1 + 51 32 2.8 2.5 6.2 5.3 1 OH 17..1-1,.2 + 3 21 1.7 1.0 4.0 4.5 I OH 17 .2-1,.1 + 172 25 1.1 1.7 1.8 1.5 1 OH 17 .4-0,.3 + 30 33 5.2 3.9 13.7 12.2 4,5 OH 17.7-2,.0 + 61 24 75.4 85.5 115.1 124.3 4,5 OH 18.2+0.5 + 126 24 1.7 1.9 5.3 2.7 3,5 OH 18.3+0.1 + 51 26 0.3 0.3 0.5 0.3 1 OH 18.3+0.4 + 48 31 5.3 5.6 7.9 9. 1 3,5 OH 18.5+1 .4 + 176 20 5.8 4.1 9.7 10.8 1 OH 18.7+1 .6 - 1 32 1.7 0.9 1.9 0.8 1 OH 18.8+0.4 + 13 28.5 9.8 14.0 27.5 17.1 3,5 OH 19 .5+4.0 + 46 32 1.6 2.0 2.5 3. 1 1 OH 20 .2-0.1 + 27 32.5 4.6 4.2 5.2 4.7 3,5 OH 20 .3-1 .5 - 12 25 0.8 3.4 1.4 3.8 4,5 OH 20 .4-0.3 + 42 34.5 2.2 4.0 3.6 6.1 3,5 OH 20 .4+1 .4 + 39 35 0.6 1.0 0.6 1.1 I OH 20 .6+0.3 + 91 40 1.4 1.8 3.0 4.7 I OH 20 .7+0.1 +136.5 36.5 8.9 4.2 I 1.0 8.3 3,5 OH 20 .8-0.8 + 48 31 0.4 0.9 0.5 1.8 1 OH 20 .8+3.1 + 27 28 2.8 3.9 5.1 6.1 I 7.8 OH 21 .5+0.5 + 116 37 5.6 13.9 11.7 3,5 0.8 OH 21 .9+0.4 + 86 36 1.4 3.1 1.9 1 4.2 OH 22 .1-0.6 +116.5 26 3.6 3.4 5.4 3,5 0.8 OH 22 .3-2.5 + 113 27 0.9 1.0 0.9 1 2.0 OH 23 .1-0.3 + 34 28.5 3.1 17.0 13.0 3,5 41

Table 1 The Type II OH/IR Sources at i Ï 10°, continued

Name S S Ref V AV 6.5 I 1 22 km s km s Jy • o" W m~2 LV+ HV LV HV OH 23.7+1.2 + 0.5 28.5 3.6 3.6 5.1 4.7 3,5 OH 23.8-1.1 + 48.5 28 1.0 4.7 1.8 2.3 3,5 OH 23.8+0.2 + 107 21 1. 1 5.6 2.3 8.8 3,5 OH 24.3+0.7 + 59 31 0. 7 4.3 1.6 6.9 1 OH 24.5+0.3 - 73 27 6. 8 2.3 6.6 5.1 1 OH 24.7-1.7 + 92 26 1.3 1.3 1.3 1.6 1 OH 24.7-0. 1 + 124 36 4.6 0. 5 8.2 0.9 3 OH 24.7+0.1 + 57 31 2.6 2.6 3.3 3.6 4,5 OH 24.7+0.3 + 41 38 3.7 1.9 7.8 5.4 1 OH 25.1-0.3 + 142.5 25 6. 9 4. 1 6.7 3.7 3,5 OH 25.5-0.3 + 36.5 34.5 3.6 3.4 5.0 6.9 3,5 OH 25.5+0.4 + 39 32 2.0 2, 2 5.5 4.Q 1 OH 26.2-0.6 + 71.5 44 16.8 9. 1 36.8 14.9 3,5 1 OH 26.3+0.1 - 26 26 1.6 2,5 1.9 3.3 1 OH 26.4-2.8 - 66 24 0.7 1. 1 1.0 0.9 1 OH 26.4-1.9 + 27 24 6. 2 9. 8 5.0 9.5 3,5 < OH 26.5+0.6 + 27 28 140 315 132 410 3 5 OH 27.0-0.4 + 102 29 2.2 4 2 4.7 4.3 3 5 OH 27.2+0.2 + 93 39 1 1 1 3 0.9 . 1.8 1 OH 27.3+0.2 + 50 25.5 21.0 7 0 28.0 14.0 3 5 OH 27.5-0.9 + 106 5 27 3 2 3 1 4.9 6.3 3 5 ÓH 27.8-1.5 + 85 30 1 1 1 8 2.0 3.8 3 5 OH 28.5-0.0 + 107 5 26.5 ' 5 1 4 9 8.3 9.6 3 5 OH 28.7-0.6 + 46 5 34.5 5 7 4 2 6.7 6.6 3 5 OH 29.4-0.8 + 125 5 23.5 5 0 2 8 14.7 10.4 3 5 OH 30.1-0.6 + 99 40.5 39 6 30 5 70.5 48.5 3 5 OH 30.1-0.2 + 50.5 34.5 9 4 6 7 13.5 13.0 3 ,5 OH 30.7+0.4 + 6b.5 35.5 11 2 7 7 12.4 10.4 3 ,5 OH 31.0-0.2 + 126 28.5 4 7 6 4 10.2 12.9 3 ,5 OH 31.0+0.0 + 33.5 13 13 3 38 5 14.8 47.0 3 ,5 OH 31.5-0.I + 35 29.5 2 6 1 1 2.2 1.0 3 ,5 OH 31.7-0.8 + 79 25.5 2 8 4.6 1.9 5.3 3,5 OH 32.0-0.5 + 75.5 41 6.7 3.2 21.8 7.1 3,5 OH 32.1+0.9 + 137.5 24.5 1.4 2.5 1.8 3.3 1 OH 32.8-0.3 + 61 30.5 20 .4 13.5 67.2 36.1 3 OH 33.4-0.0 + 61 31 1.3 3.9 3.7 4.5 4,5 OH 34.7+0.9 + 28 29 1.8 2.1 3.2 3.9 1 OH 34.9+0.8 + 68.5 30.5 0 .7 1.1 2.4 1.4 3,5 OH 35.2-2.6 + 51 28 3.9 3.6 12.1 15.0 1 OH 35.6-0.3 + 78 28.5 12 .6 9.8 20.2 13.8 3,5 OH 36.4+0.3 + 102.5 36.5 2.7 1.5 3.7 3.5 1 Table 1 The Type II OH/IR Sources at I i 10°, continued

Name V AV S6.5 sI Ref. 22 km s km s Jy .o" W m~2 LV+ HV LV HV OH 36.9+1.3 - 12 15 2.7 2.8 2.6 2.5 1 OH 37.1-0.8 + 87 28.5 8.5 13.8 16.7 18.3 1 OH 37.7-1.4 + 111 21 1.0 0.4 2.1 1.3 1 OH 39.6+0.9 + 19.5 32 0.9 2.4 1.2 5.5 3,5 OH 39.7+1.5 + 20 33 49.8 74.1 76.8 121.9 1 OH 39.9-0.0 +148.5 39.5 3.4 4.8 6.0 6.3 1 OH 40.1+2,4 + 48 37 1.2 2.0 1.2 2.6 1 OH 42.3-0.1 + 56.5 38.5 4.9 20.3 17.7 32.5 1 OH 42.6+0.0 + 53 35.5 3.6 3.2 5.7 5.0 1 OH 42.8-1.0 - 16 25 0.8 0.7 1.7 1.6 1 OH 43.6-0.5 + 71 23 1.7 1.3 1.3 1.5 1 OH 43.8+0.5 + 9 23.5 2.9 1.9 3.2 3.1 1 OH 43.9-1.0 + 52 26 0.9 0.9 1.0 1.1 1 OH 43.9+1.2 + 50 32 1.0 0.7 1.1 0.7 1 OH 44.8-2.3 - 72 33 6.3 11.2 7.6 9.1 1 OH 45.5+0.1 + 35 34.5 5.5 4.4 6.1 5.0 1 OH 51.8-0.2 + 2 38 5.8 3.9 1 1.6 9.2 1 OH 52.4+1.8 + 15 28 2.7 2.5 2.8 3.2 1 OH 53.6-0.2 + 10 27 6.6 14.8 9.2 18.7 1 OH 55.0+0.7 + 28 27 5.5 2.3 6.5 3.7 1 OH 57.5+1.8 - 75 25 1.3 0.9 2.6 2.9 1 OH 63.9-0.2 + 5 30 1.4 2.2 2.7 3.3 1 OH 65.5+1.3 - 22 34 3.3 3.9 5.3 7.4 1 OH 65.7-0.8 - 59 21 1.0 2.3 1.7 2.1 1 OH 66.8-1.3 - 65 26 1.0 0.9 1.4 1.6 1 OH 75.3-1.8 - 3 23 6.4 5.3 8.3 8.0 1 OH 77.9+0.2 - 39 21 9.0 5.9 5.8 5.8 5 OH 80.8-1.9 - I 48 420.0 160.0 1200.0 400.0 7 OH 83.4-0.9 - 39 35 12.4 2.9 13.2 4.6 1 OH 85.4+0.1 - 23 27 1.7 1.7 2.6 1.6 1 OH 104.9+2.4 - 26 29 41.3 37.5 76.6 57.6 1 OH 127.8+0.0 - 55 22 28.8 8.5 28.8 17.9 1 OH 138.0+7.2 - 38 19 9.7 7.1 10.0 6.6 1 • OH 141.7+3.5 - 57 24 4.1 1.0 4.2 1.5 1 LV = low-velocity; HV = high-velocity 1 Chapter II 2 Baud et al. 1978a 3 Johansson et al. 1977a 4 Eowers, 1978a 5 Bowers, 1978b 6 Winnberg et al. 1973 7 Harvey et al. 1974 1 1 1 1——i r 1 1 1 1 1 1 1 1 1 1 1 5o

4° Dwingeloo survey limit 3o * ?° 1 ,

0o - *• *. -1° * o i * • • -2 - • 1 o -3 • -4o i i i i i i i i i i i i i i i 100° 90° 80° 70° 60° 50° Utf 30° 20° 10°

Fig. 1. Distribution of OH/IR sources in Table 1 in galaatia coordinates.

Ill Distribution of OH/IR sources in velocity, longitude and latitude

The positions of all OH/IR sources at I < 100° have been plotted in Figure 1 in galactic coordinates. We like to argue that this diagram gives a rather complete picture of the distribution of OH/IR stars on the sky. First, consider the completeness in latitude. The area of sky, covered by the Dwingeloo survey, has been indicated (cf. Paper II); the majority of the sources are situated well within this area. Only a few sources have been discovered at \b\ > 2°. Even at those longitudes, where the survey extends to \b\ ^ 2°, the latitude coverage is adequate. If, at i < 27 and at 77 < i < 90 , we compare the number of sources at \b\ < 2° with those at \b\ > 2°, we estimate that at 27° < 8. < 77° less than 3 OH/IR sources have remained undetected outside the latitude extent of the survey. Such a small number does not distort the apparent distribution in Figure I. The two sources near i = 10° at b > +4° were found incidentally in the Dwingeloo survey; OH 44.8-2..! w-js found by Bowers (1978a). Second, consider the longitude distribution. In Figure I the sources concentrate strongly towards lower longitudes at i - 90°. Only three sources were found in the longitude and latitude range 100° < I < 234°, \b\ • 2?S (Bowers, 1978a), whereas no additional OH/IR sources were found in the more sensitive Dwingeloo survey between Í = 130° and 150°, |¿>| < 4?2. Therefore, the lower density of sources at t > 90° is definitely real.

In section III.2 and III.3 we will discuss the longitude and latitude distributions in more detail. But here we conclude that our larger sample qualitatively confirms the conclusion reached previously by others(CaswelI, 1974; JAGW; Bowers, 1978b): most OH/IR sources are concentrated at galactocentric distances R • R^ and the strong concentration ti' the galactic equator indicates that they are located at large distances from the Sun.

///./ Vrlosity distribution

The 1612 MHz spectrum of an OH/IR source is characterized by the LV and HV emission components with radial velocities VL and VH> respectively. Several convincing arguments, summarized by Nürnberg (1977), indicate that the two relevant physical velocities are the radial velocity of the central star V = J(VL + VH) and the velocity separation AV = VR - V, , which corresponds i i i i i i i i i i i i T n N — -

26 -

22 -

18 -

14 : -

1 10 -

6 -

2 - 1 1 1 1 1 1 1 1 1 1 1 1 ¡ 1 11 14 17 20 23 26 29 32 35 36 41 44 47 50 ' I AVIkm s')

Fig. 2. Distribution of the velocity separation At' for all sources in Table 1.

to twice the expansion velocity of the circumstellar shell. First we will consider the quantity AV. The distribution of AV of all sources in Table 1 is presented in Fig. 2. The number of sources increases steeply for AV 5 20 km s , reaches a maximum around AV = 29 km s and decreases more gradually to larger values of AV. The shape is similar to the distribution presented by JAGW for their much smaller sample. We will discuss this shape further in section VII. Next we turn to a discussion of the distribution of V. Here the basic question to consider is: do the OH/IR sources constitutea kinematically homogeneous sample? This question has been asked previously by Kerr and Bowers (1974), JAGW and Bowers (1978b). Until now the sample of OH/IR sources has not been large enough to discuss this problem adequately. In this section we show that the kinematic properties of the OH/IR sources are not uniform and that, in fact, they are probably a mixture of at least two galactic populations that can be classified according to the value of the velocity separation AV. Observationally, the velocity separation is very well determined. Theoretically, it is a sensible candidate for a population in- dicator because for those OH/IR sources that are optically identified with Long Period Variables (LPV), AV appears to correlate with the period P of the variation (e.g. Dickinson et al., 1975) and for the Mira's the period P, in turn, is correlated with kineuatic properties (Feast, 1963). Therefore we have divided our sample into two groups consisting of approximately equal numbers of sources: sources with large ( AV > 29 km s~') and small velocity separation (AV < 29 km s"1).

The radial velocities of both groups may be compared with the velocity distribution of the CG-emission. Those with large AV have been plotted on a diagram of the CO distribution at b = 0° (Fig. 3a), kindly provided by Dr. Burton. Most sources lie on the main ridge of the CO emission. Only a few sources are found at forbidden velocities above maximum radial velocity or at, Í < 30°, with negative radial velocities,and their deviation from this velocity range is generally very small. 1 —i—i—i i 1— —i-f— i —i i 1 1 1 r— ;—i—i—i—i—i— , 1 ! ( 1 1 1— a , b

•

55° + \ • \ " 1 rs 1 ; 60° •V V • •ai 55° • 'i A, • -V • VI

50° i "*^ V,n'l

. <- 45° r. +

40°

• • •

: , í:=. >»lí

• • • 25 -

15 + -• v : 10 8k ÍA -100 -50 0 50 100 150 200 -lOO -50 0 50 100 150 200 Vlkms') Vlkms'l

.-.. Diat-1'íb.it ¿on •'_•' '•'• ¡'ti';! "iv'ii"'1;/ I' i;' ,':',/!H su,it' •• i: (;; .i point i¡»i of Lonji titile, supei1-

ti i'n ' /:. • ,'ci.i:-ity .;•'.;• "••.••.•.' '. •;' ' /.', .'' . •••i',1.1;:', •>; .,•: /• -" ;'", '•"•ocided by Hi'. Hurten. Tlu

•' 'it poal'i '>' '..' <'i't.i.: '• ;'•'•,•,'.•.! •'; ••;, .-.:•'"i .."i >>::' ¡' .'•'.', '';/ ,:!'••:,}<',! by ,•! vi;ilai' motion

• -n, /.i",11. ,„•! .'.'i!",'« ,i •:;.•,'.: s •.. " =• ...•.••"•;. \[ •- -iji ííi .i"'. The correlation between both distribution indicates that the OH/IR sources with large AV have only very small deviations from the bulk motion of the interstellar medium (£ 10 km s ). The velocity differences at any given longitude interval mainly reflect the differences in distance along the line of sight and not large random motions. The coincidence with the main ridge of CO emission furthermore suggests that the radial density distribution of the OH/IR sources with large AV is similar to that of the CO-abundance, which has a maximum at R ^ 5 kpc and decreases rather steeply on both sides (Burton, 1976). In fact the lack of sources around Í '^ 30 with V < +40 km s and at 8. •£ 20° with V a, 63 km s indeed indicates relatively low densities at R > 7 kpc and at R < 4 kpc, which is consistent with the CO distribution. In Fig, 3b the radial velocities of the sources with small AV have been plotted on the same CO-diagram. Their overall distribution is also consistent with differei.tial galactic rotation, but many sources are found with radial velocities outside the range covered by the CO emission; in fact the radial velocities of these OH/IR sources seem to avoid the main ridge of CO emission and many sources have radial velocities well outside the maximum and minimum values allowed by circular motion. Those with large positive radial velocities are probably located near the subcentral point and their large deviations from the maximum radial velocity indicate a large velocity dispersion OQ in the direction of galactic rotation superposed on the galactic rotation. In principle the large negative velocities at S. £ 30 could be explained by postulating-that these sources are situated on the other side of the Galaxy, at R > R . But this contradicts our conclusion reached earlier, that all sources are essentially at R < E_. Therefore it seems reasonable to assume that these large negative velocities reveal a comparably large velocity dispersion with respect to galactic rotation in the radial direction, O„.

We conclude that the OH/IR sources in Table 1 do not represent a kinematically homogeneous < _| sample. Instead, the criterion AV > 29 km s separates the sources schematically into two groups with different properties: a large AV corresponds closely to the kinematics of extreme Population I objects, such as molecular clouds, H II regions and supergiants, while the sources with small AV possess much larger velocity dispersions and are presumably of older age. This dichotomy, however, does not imply that the relation between the velocity separation and velocity dispersion is unique - it may have only a statistical significance. In fact, when we split all the sources into three groups of approximately equal numbers (AV < 26 km s , 26 < AV £ 31 km s , AV > 31 km s ) it follows that the sources with values of AV around 29 km s are a mixture of both groups (see Fig. 4b), whereas about half of the sources with AV < 26 km s have radial velocities well outside the range allowed by differential galactic rotation (Fig. 4a). Inside the allowed velocity range the distribution of the small AV sources in (£, V) space is fairly homogeneous between Í = 15 and 45 and does not resemble the ridge structure shown by the sources with large AV. The r.m.s. deviation from galactic rotation velocity of the small AV sources in Fig. 4a may be estimated from the sources outside the allowed radial velocity range. From the sources above the maximum allowable velocity we derive a lower limit o„= 30 ± 5 km s ; the sources with negative velocities at I < 30° suggest a somewhat higher value a.. = 40 Í 8 km s~ . The ratio o„/o. = 0.75 is somewhat higher than the ratio of ^ 0.6 for stars in the solar neighbourhood (Delhaye, 1965).

The sources with AV > 31 km s (Fig. 4c) appear to follow galactic rotation quite closely. Only three sources have radial velocities just above the maximum value allowed by circular motion and no sources have been found at i < 30 with negative radial velocities. 47

90°

2631km s" •

80' ! \ 70° \ \ • \ 60° \

50Q

40o-

30°

20'

10' -40 «¿0 *120 -40 •40 +120 -40 440 +120 V (km s1)

Fig. 4. Radial velocity distribution of OH/IR sources for three intervals of AK, containing approximately equal number of sources. Thin drawn lines indicate the model velocity distribution derived in section VI.5 for (a) a„ - ¿6 km s and (b) o„ - 10 km s . Contour units ave in relative numbers 1, 2, '!, 8 and Iß. The dashed line corresponds to maximum radial velocities (Burton, 137-1).

III.2 Longitude distribution

The observed longitude distribution of all OH/IR sources between I = 10° and 90° is drawn in Fig. 5 (top). The general shape is similar to that found by Bowers (1978b, his Fig. 2): a steep increase in the number of sources at I > 10°, a steep decrease for 30° < 8, < 45° and a more gradual decrease at I > 45°. Both JAGW and Bowers (1978b) found a maximum around 8. = 30° i 3°. In our results, obtained from a larger sample", the maximum has shifted to lower longitudes, around I 'k, 23 . The cut-off at £% 15° is quite sharp with very few sources below 15°. This cut-off is much larger than the statistical variation in the number of sources per longitude interval and should therefore be considered as real. It should be noted that for the OH/IR sources with flux densities above the completeness limit of the Dwingeloo survey the longitude distribution has the same shape. We are therefore quite confident that the observed longitude distribution is not significantly influenced by the longitude variation of the completeness limit mentioned in section II.

The longitude distributions of the sources with large and small velocity separation (Fig. 5, bottom) are not significantly different when we take into account the larger uncertainty in the individual distributions due to the small numbers. The similarity in the longitude distribution suggests that both groups have a similar large-scale spatial distribution. Notice that both kinematic groups show the decrease at i Í 20°. There is no obvious indication for an enhanced i i 1 1 1 r——i 1 1 1 1 1 1 1 1 r 22

18 all sources.

H 1 1 1 1 1-

AV < 29 km sJ — AV X29 km s1

J=± 90 80 70 60 30 20 10

b'L:. ¿. Loiijitiule dialribntion between f, - 10° and 90° for all OH/1R sources (Lop1 ami J'ov the soioveo uith larje und small áV sepáratela (bottom). The i;irnc is the distribution pvediated from a density model j'ov k, = +7, k, - —! and R - 1.5 fcn* J ' z max as discussed in section VI. 1. density of OH/tR sources in the directions tangential to spiral arm features such as the Scutum arm (.1 % 30°), the Sagittarius arm (Í. % 50°) and the Cygnus region (H % 80°).

III.3 Latitude distribution

The latitude distribution in Fig. 6a illustrates the strong concentration of OH/IR sources towards the galactic equator: 65!? of all sources are found within Io from the galactic equator. The symmetry, with half of the total number of sources on either side of Llie plane, is rem-".rkablo Previous authors (JAGM; Bowers, 1978b) found a larger number of sources just below the ,ilane. According to Bowers (1978b) this indicates a mean value = -50 pc between R ï 4.5 and 5.5 kpc. Because the overall latitude distribution of our larger sample is highly symmetric, even within 0.5 from the galactic plane, the previously found asymmetry may have been fortuitous. The latitude distributions of the two groups of small and large AV separately are shown in Figs. 6b and 6c. They are significantly different: the sources with large AV are more strongly concentrated towards the galactic equator than the sources with small AV. The mean absolute latitude JFf = 0?5 for the sources with AV > 29 km s~' and Jb[ = 1?I for those with AV < 29 km s~'. Because it is evident from the similarity in the longitude distributions, that the sources in both groups fill approximately the same volume of space, this difference of a factor of 2 in the latitude distribution reflects a similar difference in the z-distribútion. 49

a ' b -

all sources iV >/ 29 km S-'

- 12

N c -

AV <29 km s"1 . 12

1 • 4

.3i° -2° -1° .1° .2,° 1 i , , 3° 4°

?'?\7. f', a) Litiiiïlc distribution of all Uil/Ih' soui\-es and b, i-J the absolute latitude distributions o'" lhe sourues itiih lavae and small- Al'.

IV Luminosity function

A detailed knowledge of the luminosity function of the OH/IR sources is required for a statistical distance determination and thus for a derivation of their spatial distribution. How- ever, to avoid future confusion we first have to discuss the influence of the bandw'd.h of the observations on the flux densities.

¡V.I l'}'Ci¡t¡':!. i'CBul.i' i,-"i

The spectral resolution used in the observations is of a considerable influence on the observed shape of the usually narrow line profile of each velocity component, and thus on the value of the peak flux density S. As is shown in Chapter II, S decreases on the average by a factor of about 0.7 going from a resolution of I kHz, used by Johansson et al. (1977a), to 6.5 kHz (used by us) and increases by a factor 1.} from 13 kHz, used by Bowers (1978a), to our 6.5 kHz. The influence of the spectral resolution is further illustrated by discussing the distribution of the equivalent widths, E = S /S , where S is flux density integrated over frequency and hence determined independently of the instrumental bandwidth, and S_ is the observed peak flux density for 13 each component, as measured with a bandwidth of B kHz. .;-. Fig. 7 the distribution of E, , is shown for the LV and HV components of all OH/IR sources presented in Chapter II. The distribution is similar for the high and the low velocity component. This is to be expected from the similarity of the detailed shape of both components when averaged over a large number of sources (JAGW). The distribution in Fig. 7 compares very well with a similar distribution for E., from these authors.

Their E^distribution peaks at 10 kHz, whereas the E& ^distribution peaks between 10 and 14 kHz. This difference is a results of the difference in spectral resolution and can be expressed in a difference in the mean value for a bandwidth B: for the sample shown in Fig. 7, = 18 kHz; from Fig. 4 in JAGW we derive = 13 kHz. The ratio / = 0.7 agrees with the ratio =0.7 derived in Chapter II for a much smaller but independent sample of sources. 50

i 1 1 1 1 t i i i i i

1 16 _ LV HV ! - U J -

10 - -

i i 6 - -

i —! —| •> i | |

1 II i i i i 1 i i i 12 20 28 36 IkHz) "6.5

Fig. 7. Distribution of equivalent width, E„ „ of the LV and HV components of those sources, that were observed with a resolution of 6.5 kHz.

In order to be able to compare the flux densities from different authors, we have to take into account the differences in spectral resolution. Hence, for those OH/IR sources in Table I that were not observed by us, we multiplied peak flux densities S. (from JAGW) with 0.7 and S., (from Bowers, 1978a) with 1.3. Notice that this conversion has only statistical significance and cannot be applied if one considers the sources individually.

IV. 2 Mean peak flux density of OH/IR sources

To determine the luminosity distribution of OH/IR sources it is important to define first the mean peak flux density S, of a source. This is necessary because each source has two emis-

sion peaks that differ in intensity. JAGW showed that the parameters S_, ST and E_ of the HV and the LV components for each source are statistically equal. This is also true in the present sample, even when one divides the sources between those with AV > 29 km s~' and those with AV < 29 km s~'. Although the integrated flux S^. can be determined more accurately, we have chosen to use the

observed peak flux density S& to derive the intrinsic luminosities of the OH/IR stars. Because of the statistical equivalence of HV and LV components we have combined both peak flux densities

of each source into the harmonic mean value S, _ S /{S, C(LV).S, ..(HV)} . There are several rea- O.J 0.3 O. J sons for these two choices: 1) A mean value of the intensity of both components agrees with the fact that they are statistically equivalent, both in strength and in shape. 2) Using a measure determined by both components is consistent with the criterium for the detection of an OH/IR source from the survey observations (see Chapter II): the occurrence of a double peaked emission profile. The use of a harmonic mean seems justified because it is the strength of the weakest component, that determines the identification of a double-peaked profile. 3) Because the equivalent widths of the emission components are comparable to the spectral 51

i i i ' ^ III! 40 • ñ - 32 \ - 24 \ - / \ - 16 - / \

8 -

0 -0.6 -0.2 0.2 0.6 1.0 U 1.8 2.2 2.6

log sS5

Fig. 8. Distribution of the harmonio mean of the peak flux density of both components, S r,for all OH/IR sources. The aupve is the distribution pre- lioted from Lhe density model for a - 1.65, as dismissed in Section VI. 1. resolution used in the survey observations (^ 10 kHz), the emission peaks appear only in a few spectral channels. Hence the detection probability depends strongly on the peak flux density, and not so much on the integrated flux density. We will return to the influence of the detection probability in section IV.3. 4) S is dependent on the velocity width of the individual components. Its value is determined by saturation effects OVLT the velocity profile, and through this on the detailed structure of the circumstellnr shell in which the OH masers occur. Physically the peak flux density appears to be a better measure of the total strength of the source. Although we have used the hai'monu mean, we should stress that none of our conclusions differ significantly when using a arithmetic mean of S(LV) and S(HV). Fig. 8 shows the distribution of log S of all sources listed in Table 1, in logarithmic intervals of 0.4. The increase in the number o£ sources for log S, _ decreasing from +2.4 to +0.4 is determined by a sampling of a larger volume ol space at larger distances and by the shape of

the luminosity function. The maximumat log Sft r'"+0.4 agrees satisfactorily with the completeness

limit of Sc of the Dwingeloo survey (log Sc '<• 0.3 at 6.5 kHz); the decrease at log S, , £ +0.2 is a result of the steep drop-off in the detection probability (si.-e Fig. 3 of Chapter II).

ƒ'.'..' Liiminor.' • j ,!'u! I'ib1: '•;'.

We define the luminosity L of an OH/IR source as the value of S, . which a source would 3 h.ivu at a distance of 1 kpc: L6>g = r~r~.SS 6 5> «here r is the distance along the line of sight in

kpc. To determine the luminosity distribution N(L& 5)dL6 5 we have to obtain dis,:ai ces to specific sources. This may be done by assuming that those OH/IR sources, that have a radial velocity close to or exceeding V^^ (the maximum radial velocity allowed by differential rotation), are at the 52

1 1 1 1 1 1 ' V ' ' ' a \ b 1.6 _ —.y—j - \ - - - \ 1 \ J812 1 \ z 1 \ g i\ ! \ 0.8 5 «.i - ¡V! \ - \ \ 0.Ä _ V - i nn \ 1.46 1.86 2.2t, 6 2.66 3.06 1.65 2.05 2A5 2.85 3.25 3.65

Fig. S. Observed luminosity distribution of the OII/IR sauvaeo (drawn) at the ünL- acntral point at a) Í < S7 and bi Z > S? . in eaeh aase the distribuiion, ¿owcc-Lcä ' I for inoompl.eteT.ess of the sample (dashed}, was approximated by eye with a straight line with the denoted slope. The arrow indicates the luminosity completeness limit, averaged over the corresponding longitude interval.

subcentral point where the line of sight is closest to the Galactic Centre. Ue select-tíd 23 sources with large radial velocity from our Table 1 and calculated the distance r to the subcentral point, assuming R = 10 kpc. In selecting appropriate sources a compromise must be made between the accuracy of the distance determination from the maximum velocity criterium and the few number of sources at V > V .We use V > V for sources with AV '-' 29 km _. " max max "" m s and V V -10 km s for sources with ¿V > 29 km s ; the difference is a first-order max — attempt to corree- for the effect of the deviation from circular motion. Because for I 27° (On- sala survey) and I < 27 (Dwingeloo survey) the completeness of the sample is different, the sources were divided accordingly into two groups. The resulting observed luminosity distributions are shown in Fig. 9a, b on a log-log scale. Notice that the range in the derived luminosities is about a factor of 25. This is much larger than the range of intensity variations with time, expected from Harvey et al. (I974), who found variations of typically a factor of 3 in amplitude. Therefore a real range in the distribution of the time averaged luminosities must exist. The maximum in both distributions occurs at the corresponding luminosity completeness limit L = S x o c c (average distance to the subcentral point)", suggesting that the shape at log L < log L is determined by the incompleteness of the sample. The distribution can be corrected for such incomplete- neáy by dividing tho number of sources found in each luminosity interval by the corresponding detection probability P(log L), averaged over the interval. The values for P(log L) are derived from Fig. 63 in Chapter II. The corrected distributions (dashed) can be approximated by a straight line, resulting in a luminosity distribution

!itL6.5) dL6.5 dL 6.5

where a has a value between I.I and 1.5. The uncertainty in a is Í 0.5, which is mainly determined 53

by the statistical errors due to the small number of sources. It should be noted that the shape of the above luminosity distribution, i.e. the value of a, is independent of the spectral resolution. In view of the small number of sources with L, _ > L and in view of the rough determination of the stellar distance, we will consider briefly the question whether the above derived luminosity distribution is anywhere realistic. The slopes in Fig. 9a, h are mainly determined hy the corrected number of sources detected below L . Although the value of this correction is uncertain, it seems impossible to avoid the conclusion that the actual numbers of sources below L must be larger than the observed numbers, because of the steep fall-of in detection probability. Because the observed number of sources above and below L are approximately equal, this must imply an actual increase in the number of sources with decreasing luminosity, at least over the presently discussed range of luminosities. The above luminosity distribution deviates strongly from the results of JAGW and Bowers (1978b). They obtained a luminosity distribution which is of symmetrical shape, with a maximum at L % 3J0 Jy in both cases. The most important difference between tlieir methods and ours is the fact that apparently they did not take into account the incompleteness of the sample below L . As follows from the discussion above, this neglect results in an observed luminosity distribution, which is more or less symmetrical with a maximum at a luminosity corresponding to the completeness limit. In fact, if we had not corrected for incompleteness we would have derived from our sample a mean value of about 400 Jy for the observed luminosities; this lower mean value is a direct consequence of the lower completeness limit of the Dwingeloo survey. The luminosity distribution given above has been derived by combining the sources with AV 2 29 km s and AV •>' 29 km s . In Fig. 10 we have plotted the luminosities of the individual sources as a function of longitude for each of the two AV-groups. The curve represents the completeness limit of the Dwingeloo survey (f. • 27 ) and of the Onsala survey C '• 27 ) for sources at the subcentral point. There are no apparent differences in the luminosities of both groups independently. Henccwe shall assume that the luminosity distribution is independent of AV. Fur- thermore there is no indication for a variation in the distribution as a function of longitude. We conclude that the luminosity distribution N(L, ), derived for sources at the subcentral point, is a reasonable representation oí the luminosity function iML) of all OH/IK sources in the Calaxy: ML) - iT'.

•V Comparison with the iúV'nt i I i. .1 Type II OH/IR sources

Many OH/IR sources at high latitudes have been identified with nearby Mira variables and a few with M-type supergiants (for reviews, see Caswell, 1974; Winnberg, 1977). Those identified with Mira variables are either cif Type 1 (strongest at the mainlines) or of Type II (strongest at 1612 MHz). All OH emitting stars are reddened objects, indicalinp the presence of a cool circumstellar dustshell (Hyland et al., 1972); the reddening is larger for stars, identified with Type II 011/ 1R sources. Wilson and Barrett (1972) extended this relation to thicker dustshells: they found strong double peaked 1612 MHz Oil emission and only very little or no 1665 MHz or 1667 MHz emission from nearby, very reddened (I-K < 6) sources from the IRC survey, that could not be identified optically, llrwi-vn , the II! and OH emission properties are quite similar to thai of the optically identified Type II OH/IR sources, indicating that the embedded star is probably also a Mire variable or ;i swpergiant. Turning nm to the unidentified Type II OH/IR sources situated at large distan es, JAWG found that on the average they are about a factor of 100 more luminous at 1612 MHz than the iden- 54

l i i i 3.4 * Fig. 10. Distribution of luminosities of the OH/IR sources at the subcentral point plotted against 3.0 - - longitude; (+): sources with Al' 2 29 kn s , (*): sources with àV < 29 hn s~ . The curve represents 2 2.6 - ++ • **• - the luminosity completeness limit L aJ which decrea- • •

f 22 i1 ses with longitude due to the corresponding smaller • * distance to the subcentral point. Tlie discontinuity 1.8 + shows the difference in completeness between tha 1 | 1 i I | 1.4 Dwingeloo survey (I < '¿7°) and the Onsala unn'eu 40° 30° 20" 10° ! (I > 27°).

tified sources in the solar neighbourhood. Nevertheless the shape of their 1612 MHz emission profile is quite similar. This similarity and the subsequent discovery of IR point source emission (Schultz et al., 1976; Evans and Beckwitli, 1977; Glass, 1978) from many unidentified OH/IR sources detected in systematic 1612 MHz surveys suggests that the unidentified and the identified OH/IR sources belong to the same class of objects, and that the former have much thicker shells. This assumption however leads to the question why we do not find high-luminosities (L £ 500 Jy) among the optically identified sources in our immediate neighbourhood. The answer to this question will lie in the luminosity function i)i(L): if iML) is a steeply declining function of L (such as found in section IV.3) there may not be enough Mira variables in our neighbourhood to yield even a few high-luminosity sources. The occurrence of high luminosity sources at large distances in the inner parts of the Galaxy is then simply a reflection of the murh higher OH/IR source-density there. This explanation agrees with the galactic density distribution of OH/IR sources derived in the next section.

VI The Galactic distribution of OH/IR sources

In this section we derive a model of the galactic density distribution of OH/IR sources. This model explains the observed distribution of the sources on the sky and in radial velocity. Further- more it predicts the observed flux density distribution. In deriving the model we will assume that the density distribution has axial symmetry around the rotation axis of the Galaxy, and hence is independent of 0. The coordinates are defined as follows: R is the galactocentric distance, r is the distance along the line of sight and z is the distance perpendicular to the galactic plane. R and z are related by:

2 2 2 2 R = r cos b + R - r R cos b cos I o s z = r sin b

VI.1 Distribution in log S„ c and I. v. 0 The number of sources found in each interval of S. and S depends on the density of sources along the line of sight, their luminosity function and the sensitivity of the observations: b 2 S(J., log SB) ál dlog NodÄdlogSB ƒ» P(log SB) f(R, z) i|)'(log L) r dr áb (1) 55 where N is a normalization factor which includes the normalization of the axially symmetric den- o sity distribution f(R, z) and the luminosity function ijj'(log L); f(R, z) is assumed to be symmetrical in z; b = 4°; P(log Sß) is the detection probability for a source with a peak flux Sß, using a spectral resolution B kHz. For the Dwingeloo and Onsala survey, B % 11 kHz.

Our model calculations consist of deriving N(S,, log Sß) from the integration of equation (1) for a model density distribution function f and a model luminosity function i))'(log L). Integration of N(2., log S„) over I and log Sß then yields the predicted flux density distribution Ng( log Sß) dlog S- and the predicted longitude distribution N^ODdl respectively. We will now specify the functions f, ip' and P. The density distribution f should be consistent with an increasing density for R < R@ up to a certain radius R , and a decrease beyond (see section III. I). We will assume a symmetrical max z-distribution of Gaussian form and a power-law radial density distribution:

f(R, z) = Rkl'2 exp {~(z2/2ol)} (2)

where the power of R should be ki if R < R and k2 if R > R and a is the dispersion in the nuix max z z-distribution, which was taken to be 0.11 kpc (Bowers, 1978b). Because no OH/IR sources have been found close to the plane at I > 150° (Bowers, 1978a) we can safely cut off the integration over r, when R > 12 kpc. The value of the maximum distance, r , ranges between 22 kpc at £ = 10 and 2 kpc at t= 150°. For the detection probability P(log S ) we have used the functions as given in Fig. 63 in Chapter II. We took into account that P(log S ) varies as a function of longitude because 1) there exists a difference in the sensitivity in the Dwingeloo survey above and below I = 27 (3o = 0.53 Jy and 0.64 Jy respectively); 2} che region between i = 27 and 42 along the plane, which was avoided by us, has been covered by the Onsala survey with a different shape of the function P(log S_). For the logarithmic luminosity function we use -a+1 );1 (log L) = L (3)

Trial calculations, using values of a derived in section IV.3, showed that N(£, log S„) was is completely dominated by the nearby, very weak (L « 1 Jy) sources. This is due to the adopted shape of ijKD a L and results in an N. (H)-distribution that hardly varies with longitude, which is in obvious contradiction with the observations. It indicates that there must be an effective cut-off at a minimum luminosity L . for which i|)(L) = 0 at L < L , .We have adopted L . =1 Jy, min min min which corresponds to the low-luminosity end of the luminosity distribution of those optically identified Type II OH/IR sources for which distances are determined (Hyland et al., 1972). It also corresponds to the detection limit of the Dwingeloo and Onsala survey. We also applied a cut- off at the high-luminosity end of iJj(L) at a luminosity L = 4000 Jy corresponding to the highest luminosities derived in section IV.3. However calculations without the latter cut-off produced similar results, because the sources close to the completeness limit contribute most to the observed numbers at any distance along the line of sight.

For each interval of di, = 5° from Í = 10° to 150° and dlog Sß = 0.1 from log S = -0.15 to 3.0, equation (1) was numerically integrated in steps of dr = 0.33 kpc and Ab = 0?2; then N(£, log Sg) dl dlog Sg was normalized to the total number of observed sources. The values of kl 2' Rmax and a were adJusted to match the observed distribution in I and log S, _. We now dis-

cuss briefly how the free parameters a, kx 2 and R were determined; then we present the final solution and discuss the accuracy by which these values could be obtained. 56

The shape of the flux density distribution Ngdog Sß) turned out to be rather insensitive to variations in the parameters kt 2 and Rmax because Ng(log Sß) is mainly determined by the sources between fi.= » 2Q° and 40°. In this direction the sources near the subcentral point contribute most to the observed numbers. As a result distance variations play a relatively little role in the shape of N_(log S ) and the calculation was mainly sensitive to the parameter a. Hence the value of a could be obtained with reasonable certainty from fitting the calculated flux density distribution N„(log S„) to the observed distribution in Fig. 8. (The spectral resolution B = II kHz for o a the survey observations. On the other hand the observed flux densities are measured with a resolution of 6.5 kHz. The conversion from a resolution of II kHz to 6.5 kHz corresponds to a small horizontal shift of the calculated distribution to the right with respect to the observed distribution). The value of a was determined at 1.65 Í 0.1, which is well within the range 1.3 Í 0.5 obtained in section IV.3. A steeper luminosity function (larger a) would result in only very few sources above the completeness limit, while a lower value of a would mean many sources at the high luminosity end of the distribution, both contradicting the observations. The predicted N„(loe S, _) distribution with a = 1.65 and the values of k, ,, R , derived below is drawn in S o, 5 '»£ max Fig. 8. Using the above derived value of a we calculated N„(H) for a range of k - and R -values: x ' i¿ max kj : 0 -> +10

k2 : 0 H- - 9 R : 4 •+ 6 kpc

with dk 1 and dR =0.5 kpc. max v In Fig. 5 the predicted longitude distribution is drawn for the model with R max 4.5 kpc• ,- ki =

+7 and k2 = -4. This model gives the best fit to the main features, that characterize the observed longitude distributions: 1) the steep rise at 1 > 10°; 2) the maximum around I = 25°; 3) the steep decrease from 25 to 45 and 4) the gradual decrease at I > 45 . No other combination of values could equally well predict these four features at the same time, although some combinations would give a better fit to one or two of these features. In this sense the above solution may be called unique. Because sources with small and large AV have different latitude distributions the calculations were also done with c = 0.06 kpc and 0 =0.18 kpc,

using the above derived values for R , k, and k2. No significant differences were found in the predicted longitude distribution. Bowers (1978b), using a mean luminosity of ij 800 Jy, found a qualitatively similar density distribution for the OH/IR sources in his survey, with a maximum at R '^, 5.5 kpc and a sharp decrease on either side. We have tried to reproduce his results with a mean luminosity, varying from 100 to 800 Jy, and the above function P(log S ). No reasonable fit to the observed longitude and flux density distribution could be obtained, indicating that the luminosities of OH/IR sources cannot be represented by a mean value.

R The uncertainties in the derived values for m kj and k2 can be estimated by comparing the distributions predicted from other combinations of values. Such a comparison produces the following very conservative constraints: 4.5 < R < 5 kpc; +7 < k, < +9, in order to account - max - r - i -

for the steep rise at i > 10 . The value for k2 could be somewhat lower, between -3 and -4. A smaller or larger value respectively results in a much steeper or less steep decrease at 9. > 25° and in both cases only n moderate increase at I > 10°.

Errors in P(log Sß) could in principle contribute considerably to the uncertainty in the derived parameters. Nevertheless we do not think that it plays an important role here because, as 57

we have noted before, the longitude distribution of sources with S& ,. above the completeness limit of the Dwingeloo survey is remarkably similar to the longitude distribution of all sources. The assumption of axial symmetry introduces another uncertainty, which is inherent to our model. We will show below, that spiral arm structure may be important for the sources with large AV.

VI," The vadiat velocity distribution

In this section we will discuss the observed distribution of OH/IR sources in longitude and radial velocity by comparing it with a theoretical radial velocity distribution, expected from the model density distribution derived in section VI.1, and the assumption of differential galactic rotation. This is required to check the consistency of the density model and it will also lead to a better estimate of the velocity dispersion with respect to galactic rotation for each of the two kinematic groups. In constructing the theoretical distribution we will ignore the latitude distribution; only the velocity n in the radial direction, and 0 in the direction of galactic rotation are considered (0 = 250 km s ). The expected total number of sources in each longitude and radial velocity interval is represented by

£, V) dV H di. dV ƒ ' r.f(R).(r).F(r, V) dr o (A)

M is a normalization constant, which includes the normalization of the quantities f, 0 and F. f(R) is the density distribution in the plane (=» Equation 2 with z = 0); cj>(r) is the fraction of luminosity distribution that can be observed at a distance r, and it is given by

,+3.0 1 if dog D.PQog SB) dlog (5)

where the integration interval corresponds to the observed range in flux densities (L = r~S). In equation 4 the product <|>.f equals the total number of sources that can be detected at a distance r. F(r, V) is the local radial velocity distribution; we adopt a circular distiibution, superposed on the mean radial velocity V , assuming that the two velocity dispersions a., and a„ are equal:

F(r,V) = h exp {-h2(V-V )2} , with h2 = -\ = -\ (6) 2Qi 24 and sin I {0 (R) -2. - G } V m R a (7)

where 0m(R) is the mean rotational velocity at a distance R from the Galactic Centre. The assump-

tion that the velocity dispersions a^ and oQ are equal simplifies the calculations significantly.

It is in reasonable agreement with the observed ratio aQ/on = 0.75 (section III.l) and also with 22 the value for expected from the theoretical relation (oQ/o„) -B/(A-B) where A and B are Oort's constants (Oort, 1965). In the solar neighbourhood, with A = 15 km s~' kpc~' and B -10

km s kpc , we obtain (o0/on)R 0.63. In the inner part of the Galaxy around 5 kpc, where the majority of observed OH/IR sources are situated, Oort's constants have different values. Using the rotation curve by Burton (1974), we find A = 16.5 km s~' kpc"1, B = -31 km s~' kpc"1 and ^°0^On*5 kpc = °-81» resulting in an expected velocity distribution at R = 5 kpc that is nearly symmetrical.

According to Oort (1965) the mean rotational velocity 0m(R) is related 'to the circular velo-

city 0c(R) in the following way 58

(8) en(R) - e

R and to (see section VI. 1). The density gradient 3 ^o|j R is equal to k, at R < max k2 at R > R^ The quantity on the left hand side of Eq. 8 is called the asymmetrical drift. In the solar neigh-

e bourhood it has a negative value because ~ -° R is usually between -2 and -4 (e.g. Table 5,

Oort, 1965). In our case at R < R , kt = +7; the resulting asymmetrical drift is positive, which corresponds to a mean rotation velocity larger than the circular velocity. Physically this is easy to understand: when the stellar density decreases further inward, Chen, at any point, there are more stars in the inner half of their epicycle than in the outer half. Since epicyclic motion is retrograde the average velocity is larger than the local circular velocity, A small correction for the observed asymmetry in the velocity dispersions is applied in Eq. 8. The rotation curve 0 (R) at R > 5 kpc is from Burton (1974), and at R < 5 kpc from Simonson and Mader (1973).

For intervals of dV = 5 km s and dil = 5° Eq. 4 has been integrated numerically using the values of the parameters ct, k2 and R derived in section VI.1. The velocity dispersion a^ is a free parameter in the calculations. The predicted distribution N(it, V) is drawn in Fig. 4a for o„ = 35 km s and in Fig. 4c for a„ = 10 km s . These values for the velocity dispersion for both kinematic groups are in reasonable agreement with the observations. The calculations provide a good tit for the sources with V < 26 km s , both outside and inside the radial velocity range allowed by circular rotation, although within the allowed range the observed number of sources is too small to trace out the predicted structure in detail. The agreement between the observations and the calculations indicates a rather smooth density distribution along the line

of sight. Comparison with the calculated velocity distribution over a wide range of on-values -1 shows that the sources with small ¿V have 30 < a„ < 40 km s n -1In view of the small difference between a„ ando™ discussed earlier, we estimate a_ ^ 30 km s -I The calculated radial velocity distribution for the sources with AV > 31 km s gives a good fit to the total range in radial velocities; this suggests very low velocity dispersions with respect to circular motion (a_ £ 10 km s ). Nevtrtheless, at all longitudes inside the allowed radial velocity range the maximum of the observed radial velocity distribution appears to occur at a velocity which is systematically 20 to 40 kra s lower than that predicted by the model. The cause of this difference could be either a true density distribution or a true velocity distribution that is different from what has been assumed. The following possibilities have been considered: (i) A different axially symmetric density distribution, corresponding to relatively higher densities closer to R = R^. We can get a reasonable agreement between the predicted and the observed velocity distributions for the sources at lower radial velocities when we take k = -3, and R 7 kpc; but then the resulting longitude distribution of sources disagrees strongly with the observed longitude distribution. Also the presence of sources with V £ 80 km s at !. £ 30° cannot be explained by such a density distribution. (ii) More likely is the presence of axial asymmetries in the density distribution, such as spiral structure. Although the longitude distribution of the sources with large AV does not suggest (nor exclude) such a possibility, the good agreement between the velocity distribution of the CO, which is certainly concentrated in spiral arms, and the sources' with large AV (Fig. 2a) points strongly to such an explanation. The calculations of the model H I velocity distribution by Burton (1971) who included spiral arm structure in the large-scale density distribution, indicate ridge-like structures in the Ä./V diagram at lower velocities. Figure 11 shows the qualitative agreement 59

10 -100 -50 0 +50 +100 •150 V (km s"1)

Fij. ¡1. Radial vclofitij distribution of the OH/IK sources with AK - 29 km s -i , plotted on the nhitjl ¡I 1 velocity dis'-i'ibul .\."¡ by Bin-ton (lO'.'l). The grey ridije-like s tinta Uwes correspond to spiral arm stt'iictiwe in ihr moi ! density distribution. Dark areas near maximum radial velocity arc the spiral aims, seen l-anjcit ially.

between Burton's model H I velocity field and the observed radial velocity distribution of OH/IK sources with large AV. This suggests that spiral arm structure prolinhly is present in tha large- scale distribution of OH/IR sources with AV > 29 km s~'. As a result the concentration of sources at lower radial velocities corresponds to a somewhat smaller mean distance Trum the Sun.

VII Discussion

In this section the main results of the previous analysis are discussed in terms of the population properties and galactic distribution of OH/IR sources.

VII.1 Stellar dynamics, ages of t'l • J'Jec-ts

The most important result of this chapter is the discovery of the continuous variation with expansion velocity AV of kinematic properties among the OH/IR sources, i.e. with increasing AV 60

there is a decrease in the velocity dispersion with respect to galactic rotation and a decrease in the latitude extent. Schematically there exist two groups of sources. Those with large AV (> 29 km s ) follow quite closely the radial velocit-y distribution of the CO-emission, indicating that they are young objects with very low velocity dispersions (a„ ^ a„ >\, 10 km s ). Their radial velocity distribution suggests a density distribution which is probably concentrated along spiral arm features. On the other hand the sources with small ñV (< 29 km s ) have a larger velocity dispersion with respect to circular motion, as is apparent from the many sources with large radial velocities outside the range allowed by galactic rotation. For these objects o^ % 30 km s and a„ ^ 35 km s~ . Their (4, V)-diagram does not show any evidence for concentration to spiral arms, but instead, the sources are smoothly distributed in the galactic disk. It should be stressed that the relation between AV and velocity dispersion is net a very sharp one and has only statistical significance.

VII.1,1 Velocity dispersion perpendicular to the plane

We will show, that for the sources with AV < 29 km s the ratio of the velocity dispersions in the z- and R-direction, a„/(J„, is significantly less than 1. This fact is well known for the stars in the solar neighbourhood, but it appears to be also true at R = 5 kpc. The velocity dispersion c„ can be estimated from the distribution of the sources in the z-direction. We'do this in two ways, each based on a different assumption. Although each estimate has its uncertainties the overall result should be reliable. (i) Consider the sources with small AV at 20° < 2, < 50°. They are mainly situated around R = 5 kpc (c.f. section VI.1). Their mean distance is about 8 kpc. In the case of the sources with large AV, which appear to be concentrated in spiral arms, the mean distance is between 4 and 8 kpc. From the mean latitude TE] = 1° (AV < 29 km s~') and TFf = 0?6 (AV > 29 km s~') we derive TzT = 140 pc S J. ™* _ S and 40 < TzT < 80 pc respectively. Since TzT is proportional to a,, we find 1.3 ^ a, /a_ < 1.9. i _l L ¿,s ¿,i We adopt a„ . = 8 km s , which is a representative value for the young stars in the solar neighbourhood (Delhaye, 1965). This velocity dispersion reflects the kinematic conditions during stellar birth and one does not expect thos-3 conditions to vary drastically over the Galaxy. This value of a , is also consistent with the value o. , i{ 10 km s_ i . Hence we derive 10 < a, < 15 km s i ¿,1 11,1 Z,s for the sources with small AV at R = 5 kpc. (ii) From |z| = 140 pc we can derive CJ„ once we know the force perpendicular to the plane, K . Using various standard mass distribution models for the Galaxy (e.g. Schmidt, 1965) we derive o = 10 to 15 km s at R = 5 kpc. -1 -I Since for the sources with AV < 29 km s D. 1 35 tan s , we find a„la„ %. 0.43. This is simi- 11 Z 11 lar to the average ratio derived for the late-type stars in the solar neighbourhood (Delhaye, 1965). It shows, that over a large range of distances between R = 5 and 10 kpc those dynamical conditions that are responsible for the difference in velocity dispersion within and perpendicular to the galactic plane, do not vary significantly. This is in agreement with the theoretical predictions of Martinet and Mayer (1975), who showed that for various galactic mass-mndeLs the ratio o /o„ has to be constant between R = 5 and 15 kpc.

VII.1.8. Association uith stellar objects, age-deteimination

The strong concentration to the galactic plane of both groups of OH/IR sources indicate that they range between extreme population I (AV > 29 km s~') or young disk population (AV < 29 km s~') objects (Blaauw, 1965). This result is consistent with the conclusion derived from the optical 61

and infrared observations of nearby OH/IR sources, namely that most of them are associated with Mira variables and a few with M-type supergiants. The kinematic properties of the optically known Miras in the solar neighbourhood have been studied extensively by Feast (1963). He found a correlation between the period P of the light

variation, and the velocity dispersion (o"n, aQ, a^). The dispersions range from (94, 91, 61 km s~') for the stars with the shortest periods (140 < P < 200 days) to (38, 48, 21 km s ) for the stars with the longest periods (P > 400 days; cf. Smak, 1966, Table 3). Furthermore, for the OH/ IR sources which are optically identified with Miras there appears to be a positive correlation between AV and P (Dickinson et al., 1975). These two pieces of evidence indicate that many of the distant, optically unidentifed OH/IR sources are also associated with Mira variables. From the measured velocity dispersions (35, 30, 15 km s ) of the OH/IR sources with small AV, we conclude on kinematic grounds that they are associated with stellar objects that form an extension of the Mira populations to longer periods (P > 500 d.). This indirect evidence is in good agreement with the direct results of Dickinson et al. (1975); for a limited sample of OH emitting Mir?, variables they found that all sources with AV > 20 km s (which is about the minimum value of the presently discussed sample) have periods larger than 500 days.

What is the age of these OH/IR sources with small AV? For the stars in the solar neighbourhood there is a well-known relation between (i) the observed velocity dispersion with respect to circular motion and (ii) the stellar age i as derived from stellar evolution theory (e.g. Eggen, 1/3 1970; Wielen, 1974); the velocity dispersion appears to increase secularly as r For a velo-

city dispersion o = 35 km s and aR = 30 km s as derived in section VI.2 for the sources with small AV, the data presented by Wielen (his Fig. 4) lead to i = (3 to 5) x 10 years. However, Wielen's relation between the velocity dispersion and \ is derived at R = R , while most of our sources are at R = 5 kpc. The above derived age is probably an upper limit. We will assume that the timescale of increase of the velocity dispersion depends on the frequency of encounters between the stars and massive gas clouds (Spitzer and Schwarzschild, 1953), and thus on the period of galactic rotation. Because of the difference of a factor of two between the period of rotation at 10 kpc and at 5 kpc, a more reasonable age estimate of the OH/IR sources with small AV is probably (I to 2) x 109 years.

The OH/IR sources with l.iri;e AV are mainly associated with very young objects as is apparent from the strong conceiitrat UMI L,> the galactic plane and the low velocity dispersion. This implies that they are probably similar t.- those M-type supergiants (T % 10 years) in the solar neighbourhood, which are known OH/IR sources (sucli as VY CMa and NML Cyg; the low age of VY CMa, suggested by Herbig, has been confirmed recently by Lada and Reid (1978)). This is consistent with the fact that these OH emitting supergiants generally have values of AV ,;', 30 km s (Caswell, 1974).

Kinematically, sounus with AV around 29 km s are probably a mixture of Miras and M-type supergiants. This is in qu.i I i Uil ivo agreement with a similar mixture for the optically identified OH/IR sources as is suggested by Bowers and Kerr (1978), from ;..' analysis of the 1612 MHz OH emission profiles of M-type supergiants. These authors propose that all sources with AV • 40 km s an- associated with M-type supergi.ints, while at lower values of AV an increasing portion are Miras.

Finally we suggest that ultimately the argument of the ages should be turned around: once tlii' age of tlio OH/IR sources can be determined directly fiom infrared spectrosconv, then one can derive empirii-.il I v the relation between age and velocity dispersion elsewhere in the Galaxy. 62

VII.2 Density distribution

The density distribution of the OH/IR sources is concentrated towards the inner region of the Galaxy. The sources with small AV (Miras) show a rather smooth increase in density = R to a maximum at R = 4.5 kpc. Such a steep density increase inwards has also been found for the older disk-population stars in the solar neighbourhood (cf. Oort, 1965, Table 5). The sources with large AV (M-type supergiants) exhibit the same kind of large-scale behaviour, although their radial velocities indicate a more patchy distribution consistent with a concentration to spiral arms. Very few sources of either type are located outside the solar circle. Both groups of OH/IR sources show a very sharp cut-off in density inside R = 4.5 kpc. The observational evidence for this verj low density of OH/IR sources at R < 4.5 kpc can be summarized as follows: i) a sharp decrease in the total number of sources detected at % < 20° (Fig. 5) and ii) the conspicuous lack of sources with V > +60 km s at i < 20° (Fig. 3). We emphasize that this low density at R < 4.5 kpc cannot be due to observational selection. For example the average distance to the "empty" inner part is approximately equal to the distance of the subcentral point at 8. ^ 20°. Many of the subcentral sources are well above the completeness limit of the Dwingeloo survey and would even have been detected on the other side of the Galaxy at a distance from the Sun of 20 kpc! In addition, our much more sensitive observations at 358° < S, < 14°, discussed in the next chapter, are consistent with a very low density of OH/IR sources in the inner part of the Galaxy.

VII. 2.1 Comparison with the gas distribution

The low density of OH/IR sources in the inner part of the Galaxy and the maximum at R % 5 kpc has been found before by Bowers (1978b). He noted that the radial density distribution of the OH/IR sources is quite similar to the distribution of such species as CO, ionized hydrogen and supernova remnants. These typical Population I components all have a maximum abundance around R = 5 kpc, indicating a large rate of star formation at that radius. Burton and Gordon (1978) show that the inner boundary of the CO distribution is more sharply defined than the outer boundary. The same effect is also apparent in the distribution of OH/IR sources derived by us and it is a further indication of the close agreement between the large-scale distribution of the extreme Po- pulation I objects and the OH/IR sources. That the sources with large AV exhibit the same kind of overall density distribution is not surprising in view of the fact that their kinematics indicate young objects, associated mainly with M-type supergiants of T« 10 years old. But the similarity of the density distribution of the Q sources with small AV, that are £ 10 years old, and the extreme Population I objects is somewhat unexpected. We interprete this to mean that the large-scale CO distribution in the disk and hence q the region of active star formation has not changed significantly during the last 10 years, or 4 to 8 rotations of the Galaxy. Therefore the "hole" in the gas- and the OH/IR distribution at R < 4.5 kpc and the maximum density at R = 5 kpc appear to be rather stable phenomena.

VII.S.S OH/IR sources and planetary nebulae

It has been suggested that the OH/IR Miras are the precursors of planetary nebulae (Elitzur et al., I976). These objects show a strong concentration towards the central region of the Galaxy, indicative of disk Population II objects (Minkowski, 1965). Although a detailed study of the galactic distribution of planetary nebulae is hampered by foreground absorption in the plane, there 63

is no clear indication for a maximum density of planetaries at R = 5 kpc. The obvious differences in density distribution of the planetaries show that even the small AV Type II OH/IR emitters that are assocaited with Miras of longer period (P £ 500 d), are probably much younger objects than the parent stars of most planetaries.

I'll. 2. 5 Spiral arm s true ture

The indication of spiral-arm concentration of the OH/IR sources with large AV is particularly interesting since it offers a new possibility to study the large-scale spiral structure of the Galaxy. Optical studies have used the luminous supergiants as tracers for spiral arms (e.g. Humphreys, 1970, 1972), but these studies have been necessarily confined to a region within 3 kpc from the Sun. Radio investigations have revealed spiral-like structure to much larger distances throughout the Galaxy, but they give information only on the gaseous components. An extra compli- cation factor in the analysis of the H I and CO data is the so-called velocity-crowding, which makes it difficult to separate density fluctuations from streaming motions in the gas (e.g. Burton, 197A). The present results show that it may well be possible to study the stellar component of the spiral structure in the Galaxy over large distances using the OH/IR sources with large ¿V as tracers.

VII. S Luminosities

The 1612 MHz luminosity distribution, corrected for the incompleteness of the observations, is consistent with a luminosity function of OH/IR sources that increases with decreasing luminosity. The luminosities of the sources near the subcentral point cover a range of a factor of % 25; the high-luminosity end at L . = 2000 Jy is indicative of a physical limit in the luminosities of these maser sources, while the cut-off at the low luminosities is determined by the sensitivity limit of the observations. Nevertheless it seems reasonable to assume that the luminosity function extends further to lower luminosities. This is also suggested by the lower luminosities of the nearby, optically identified OH/IR sources. Before we turn to a discussion of the physical implications of the derived luminosity function it is necessary to comment on the large difference between our result and that of JAGW, who determined this function for the first time. JAGW considered the LV and HV components as separate sources. By statistically removing the kinematic distance ambiguity with a model of the galactic density distribution, they derived a luminosity distribution of logarithmic-gaussian shape. Hence if)'(log L) has a gaussian form with an expected luminosity L = 830 Jy and a dispersion of a factor of 3, suggesting that all OH/IR sources have luminosities within a factor of 3 of the mean value. This shape is fundamentally different from our function i|)(L) * L~ " . There are several reasons why we think that our broad luminosity function is a better approximation, (i) As discussed in section IV.3, JAGW have ignored the incompleteness of their sample, (ii) The distribution of the flux density ratio between the LV and HV velocity components has itself a logarithmic-gaussian shape with an average value of 1, and a dispersion of a factor of 2.5. Any luminosity distribution based upon the flux density of the emission components separately is convolved with this distribution of the LV and HV flux density ratio and will necessarily be of symmetrical shape. Because the observed luminosity range is of the order of the range in ratio between the LV and HV flux densities this implies that the true luminosity distribution is masked by this convolution, (iii) JAGW found that 80% of their sources had the highest probability of being situated at the far-kinematic distance. This is a highly unlikely result given the observational limits and the 64

smooth longitude distribution. If anything, one expects that Ear more sources would be detected on the near-kinematic side. This means that JAGH probably overestimated their luminosities strong-

ly. Bowers (1978b) has assumed a delta function luminosity distribution, corresponding to the mean luminosity of all sources near the subcentral point. He explains the total range of a factor of 10 in the luminosities as due to the uncertainty in the distances and to the intensity variations of the individual OH/IR sources. As in the case of JAGW he appears to have ignored the incompleteness of his sample. The nearby identified Type II OH/IR sources resemble the distant sources in almost all emission properties such as line shape, AV-distribution, distributions of the flux density ratio between the LV and HV components and variability, indicating that they are the same kind of objects. However their luminosities are considerably lower. This furthermore suggests a very broad luminosity function, as derived by us.

VII.A. I Inj'lucncii! oj' the ¡IV fujiution field on the ocauweiiae of the maotn' cmiaaion

The OH molecules are probably created through dissocation of H,0, which is an abundant molecule in the circumstellar shell of the OH/IR stars (Goldreich and Scoville, 1976). These authors have suggested that photodissociation of the H^O molecules by interstellar UV (A ^ 1650 A) photons probably is important in producing enough OH in the outer parts of the circumstellar shell where the Type II OH masur operates.This suggestion implies that the maser turns on only in the vicinity of sources of strong UV radiation, i.e. 0 and B stars. If the interstellar UV radiation field is indeed an important parameter, the observed space distribution of OH/IR sources is influenced strongly by the distribution of young hot stars whi^.i are concentrated in the galactic plane. As has been noted before by Habing (1977) this so-called Goldreich-Scoville mechanism could have important consequences for our analysis. We think, however, that it does not play an important role in the observed space distribution for the following reasons. (i) According to Guldreich and Scoville (1976) the interstellar UV photons are not sufficiently available to produce enough OH. Their calculations demonstrate that collisional dissociation of

H?0 by the dust grains in the circumstellar shell could be an equally important source for OH production. (ii) The calculations were made for the interstellar UV field in the solar neighbourhood, i.e. in the galactic plane. The strength of the UV field decreases rapidly outside the plane (Zimmer- man, 1965) and the source of UV photons, the OB stars, have an average z-distancc of about 50 pc -1 (Blaauw, 1965). Nevertheless the observed distribution of OH/IR sources with AV < 29 km . has |z| = IAO pc and thus disagrees with the UV field. This argument could be undermined in the following way. According to Goldreich and Scoville (1976), the OH molecules will recombine very slowly at the temperatures (T < 500 K) in the outer envelopes of the circumstellar shell. Hence the OH abundance stays constant ever, after the OH production has stopped. One could then argue that the sources at |z| N 50 pc are those objects whose OH abundance is created during the previous crossing of the galactic plane. In that case, however, the conditions for masing should not vary significantly in the time-scale for oscillation with respect to the plane. This oscillation period is about 4 x IO' yr; it is much larger than the time-scale of thp masing phenomenon, which is determined by the time-scale of mass-loss.

5 6 1 With a mass-loss rate of I0" to 10~ M@ yr" (Hyland, 1974) and a typical mass of about 1 Il_ for the Miras (Smak, 1966) we derivp an upper limit of 10 yr for this time-scale, although more 65

reasonable values probably lie between 10 and 10 yr (Wood and Cahn, 1977). We therefore conclude that the distribution of interstellar UV photons does not influence our results on the distribution of OH/IR sources.

VII, ¿'.2 Luminosity function and ¡\V~dÍ3tvibution

A broad OH luminosity function i)i(L) naturally explains the apparent dichotomy between the distant, luminous and the nearby, weaker OH/IR sources as a difference in the overall density of OH/IR sources. At large distances around K = 5 Icpc the stellar density is much higher and only the high-luminosity tail of i|)(L) is observed, with a maximum luminosity at ^max ^ 2000 Jy. The weaker (L ii 1 Jy) sources at that distance are below the sensitivity limit of the survey and can only be observed in the solar neighbourhood. The sources with large and small AV show a similarly large luminosity range and there appears to be no obvious correlation between AV and L for the sources with high radial velocities, which are situated near the subcentral point. In other words, these sources have a AV-distribution similar to that of all sources presented in Fig. 2, implying that the shape of the AV-distribution is not influenced by observational selection effects. There are several questions in relation to L and AV, that should be discussed. Firstly, how can such a large range in OH-luminosities of a factor of 1000 or more arise? Secondly, what determines the characteristic shape of the AV-distribution (Fig. 2) with its steep rise at AV > 20 km s , a maximum around AV = 29 km s and a more gentle decrease at larger values of AV? Thirdly, how can we reconcile our results with the following facts concerning the nearby, optically identified OH/IR sources: (i) they have a broader AV-distribution, that is less peaked (e.g. Bowers, 1378b) and (ii) the most luminous sources in the solar neighbourhood have generally large values of AV, i.e. they are among the supergiants (Hyland, 1974), suggesting that AV is indeed correlated with L. First, we will discuss the main stellar parameters that determine the maser luminosity L(1612) and AV and then we will attempt to explain, at least in a very crude approximation, both the luminosity function il)(L) and the distribution of sources as a function of the expansion velocity, AV/2. We summarize a few relevant observational facts from the literature. A distinction wis made between Type I and Type II OH/IR sources. We will refer to the 1612 MHz OH luminosity as (1) Mira variables exhibit a correlation between the period P and the kinematical properties: statistically, longer periods correspond to a lower velocity dispersion and hence to younger and more massive objects. Feast (1963) estimates that (lira's with P < 400 d have HÏI II and those

with P = 400 to 500 d have M ^ 2.5 Mg. Uur observations show that the relation between P and velocity dispersion can be extended when AV is substituted for P; increasing AV corresponds to decreasing velocity dispersions with respect to circular motion. Consequently b.V is a measuve of Lhe stellar mass in a sense that AI' increases with M. This is consistent with the fact that most optically known OH emitting supergiants, which are very massive objects, have generally large values of AV (Bowers and Kerr, 1978). Kinematically, the Type II OH/IR sources with low AV (20 -•• AV < 26 km s ) correspond to Mira's with P £ 500 d and have M £ 2.5 M . Sources with ex- -1 s trame values of AV (Í 40 km s ), associated with H-type supergiants, have masses between 20 and

30 H0 (Stothers and Leung, 1971). The relation between AV and M is also consistent with the difference in properties between Type I and Type II OH/IR sources. The former sources are strongest in the mainlines and have 5 < AV < 20 km s~' (e.g. Bowers, and Kerr, 1977, Fix and Weisberg, 1978). 66

They are optically identified with the kineraatically older and less massive (1 M^) Mira's of 300 < P < 450 d. Because AV and velocity dispersion are related in a statistical sense, we emphasize that the relation between M and AV is only true for a sample of sources and should not be applied in individual cases. (2) Infrared observations (e.g. Hyland, 1974) show, that the difference between Type I and Type II OH/IR sources can also be expressed in terms of a difference in the parameters that characterize the circumstellar shell in which the maser occurs: Type I OH/IR sources have usually less pronounced IR colours between I and 10 um JT-J— is small , whereas the photospheric temperatures of their central stars are similar to those of the Type II sources. This indicates that the latter objects have a larger shell opacity than the Type I Miras. The shell opacity is determined by the total mass of dust in the shell, i.e. by the mass loss rate Í1. We conclude, that the [R colours «i'e a muaauipe of the mass Losa fate M. (3) Hyland (1974) has shown for the nearby, optically identified OH/IR sources that L increases with increasing IR colours. This relation can be extended to more extreme colours by considering the six distant unidentified Type II OH/IR sources with large AV studied by Evans and Beckwith (1977) at wavelengths between 2 and 20 [i. Assuming that these sources are situated at the near kinematic distance (arguments for the assumption have been given by these authors) we find that L«,. increases with the colour index between 3.4 and 12.5 \i. Also, for these six sources the opacity in the

10 um absorption feature appears to increase with LnH- These pieces of evidence and the conclusion reached in (2) indicate that, observationally, L is alosehj related to M. In order to ob- Un

tain more specific relation between Ln„ and M, we now turn to the question of the physical parameters, that determine the maser output of Type II OH/IR sources. (4) The correlated variation between the OH and the IR flux (Harvey et al., 1974) provides strong evidence, that the 1612 MHz OH maser is radiatively pumped. In their model of the 1612 MHz maser Elitzur et al. (1976) argue on theoretical grounds, that the pumping is done by 35 um photons provided by the reradation of the stellar light by cool circumstellar dust. In addition, for each source studied by Harvey et al. (1974) the fractional time variation of the 1612 MHz flux density never exceeds the fractional time variation of the 10 lim emission. The model calculations by Merrill and Jones (1977) indicate that the flux density around 10 um is a rather good measure- far the flux density at 35 um. This suggests that the maser is saturated, that is, the stimulated transition rate at 1612 MHz equals tie 35 vim pump rate. Hence the maser output depends only on the local pump rate, which is a function of the OH density, n_„, and the 35 pm radiation density,

U(35 um): LQH °= nQH . U(35 um). Because all Miras appear to have the same luminosity of 10 L_ and because the circumstellar dust shells all have a temperature of about 500 K (Hyland et al., 1972; Evans and Beckwith, 1977), U(35 um) is determined by the density of the dustgrains in the

shell and thus by M; in addition nQH <* ÍI. VLBI observations indicate that the total masing volume does not vary by more than a factor of 10 from source to source (Moran et al., 1977). As a result we suggest on physical grounds that

L0H « «•

This relation can be checked quantitatively by comparing the observed IR colours with the 1612 MHz OH luminosities(°= M ). Assuming that all stars have approximately the same luminosity, the near infrared flux density S(I2 um) is a good measure of the IR-colour; consequently S(12 um) <* M. Er-.ause S(I2 \m) is a good measure of S(35 um) (Merrill and Jones, 1977), we expect S(1612)/S(12 um) •* M. From Hyland et al. (1972) we find that for the nearby, optically identified 67

Type II OH/IR sources the ratio S(1612)/S(12 urn) varies from 0.01 to 0.06. The six distant sources, studied by Evans and Beckwith (1977), have S(I6I2)/S(12 Mm) = 0.21 to 0.58. This implies that the average massloss rate of the unidentified sources is about 10 times higher than the average mass loss rate of the optically identified sources, and that their corresponding 1612 MHz luminosities are expected to differ by a factor of 100. The latter value agrees remarkably well with the actually observed differences in maser luminosity of about a factor of 100 between the nearby and distant Type II OH/IR sources (Johansson et al., 1977b). Also the maximum range in S(1612)/S(12

um) of 30 to 60 gives a satisfactory explanation for the large observed LQH-range of mere than a factor of 1000. Although we cannot exclude the possibility that differences in tho stellar luminosity from source to source also contribute to the large range in 1612 MHz luminosities, the above quantitative consistency suggests that the stellar luminosity plays a relatively minor role in the width of the 1612 MHz luminosity function.

In summary, we conclude that AV is determined by the stellar mass, prior to the mass-loss phase, and that L is proportional to H". The width and the shape of the luminosity function

i}i(LnH) suggest that M can vary up to a factor of 30 from star to star and that the number of stars decreases rapidly with increasing M. This appears to be true for a variety of stellar masses, because no correlation has been found between AV(= M) and LQUÍ™ M). The most simple explanation for ! such large differences in M for stars with the same mass is a variation of H with time for each i star individually: decreasing numbers of increasing luminosity then corresponds to an acceleration of M in the course of the evolution of the Mira or the M-type supergiant. This implies that all stellar objects associated with Type II OH/IR sources start their 1612 MHz OH maser at a minimum value of M, required to invert the 1612 MHz transition; M increases subsequently faster and faster up to a maximum value M , corresponding to a maximum 1612 MHz luminosity L OH,max M . A minimum mass loss rate for Type II OH/IR sources, [I . , is about 10 M yr (Hyland max •" min a J 1974). The corresponding maximum value should then be 3 to 6 x 10 M yr , which is in good -5-1 9 agreement with the value of M = 1.5 to 4.5 x 10 M yr , which Forrest et al. (1978) deduced from their IR data for the very luminous Type II OH/IR maser source, OH 26.5+0.6. The minimum value tl . is probably governed by the minimum 011 density, necessary to invert the 1612 MHz transit ion. El itzur et al. (1976) show that at n_„ K, 1 cm the 1612 MHz transition OH is inverted and saturated. A ruximum value M could imply two things: (i) M is a physical max _ B max v 3 maximuis quenchem fodr bthy e collisionmass loss witnitho .mhv;lrnged no nstar molecules with s higheat Mr M M exist, becaus, or (iie o)f ththe e 161correspondingl2 MHz inversioy n large densities in the cirrumstH i.ir shell. The first possibility appears attractive because at mass loss rates larger than u few times 10 M yr the presently discussed stars,with a few solar masses, cannot live very long O. 10 yr). Nevertheless, quenching may also be important. Elitzur et al. (1976) noted in tliuir model calculations that the 1612 MHz inversion is quenched when the hydrogen density is artificially raised by a factor of 20. However, it is not clear whether their result is rfluvant here, because they do not seem to have increased n,,,, accordinly. OH At this stage we can only speculate on the question of tin- shape of the observei! AV-distribution. It may be explained qualitatively in a simple way by assuming that the fraction of its lifetime as an OH source that a star spends at the high-luminosity end of the OH luminosity function, and therefore the fraction of time that it is detectable at large distances, increases with increasing W and hence with increasing M. Sources with small AV (M "\', 2.5 M ) spend a long time at a low m.Tser luminosity and then rise rapidly to high luminosities: At(L "OH,lüw'<1"V':(I0H,hiehJ'» [or small AV. On Liu- other hand, the OH luminosity of the large AV sources (M '^ 20 ;1 ) rises more 68

At L for steadily:At(L0H low)/ ( 0H hi h> 'i ' large AV. As a result, the time these sources spend at high luminosities is comparable to the total OH maser time. The steep increase in the number- distribution at AV > 20 km s would then correspondió a comparably steep transition of At(L ,, , )/At(L_,, , . , ) from very large values at AV = 20 km s~ to about 1 at AV = 29 km s . uH,low UrUhign , The decrease at AV > 29 km s reElects the Salpeter mass distribution, with fewer stars of large

masses. The decrease in At(LQH low)/At(LQH hifih) f°r decreasing AV implies that the number of Type II OH/IR sources with low AV increases relative to the number of sources with large AV as one goes to lower luminosities; this is in qualitative agreement with the relatively larger number of nearby, optically identified sources that have values of AV below 29 km s . If the timescale of variation in L (1612) depends onAV,as is suggested by the AV-distribution, the luminosity function i|)(L) must necessarily be different in shape for sources with large and small AV, in contradiction with our earlier assumption that <¡>(L) is independent of AV. This assumption, however, is based on the observed luminosity distribution of only 23 sources near the subcentral point. These sources represent the high-luminosity tail of distribution, while the difference in I|J(L) as a function of AV is expected to be largest at much lower luminosities. The observations, therefore, cannot exclude a AV-dependence of i))(L) at low luminosities. A difference in the rate of change of L as a function of AV implies a corresponding diffe- • I¡ rence in the rate of change of M with M: for massive stars the mass loss rate steadily increases with time, while stars of lower mass first appear to lose mass at a rather constant rate and then suddenly M rises steeply in the final stages of mass loss. The physical reality of such a difference in the rate of change of M as a function of stellar mass needs further study.

VII.4 Conclusions

In this chapter we have presented an analysis of the galactic distribution, the kinematics and the emission properties of 114 optically unidentified Type II OH/IR sources, which have been found in various systematic 1612 MHz OH surveys at I > 10 and |¿| ^ 4 . From the results of the analysis we conclude that: (1) The velocity separation between the two emission peaks, AV, ii correlated with the kinematic properties and the z-distribution of the sources. The sources may be divided schematically into two kinematic groups: sources with AV < 29 km s are characterized by a velocity dispersion a = 35 km s , an average distance to the plane |z| = 140 pc and a ratio o /a £ 0.43. The sources with AV > 29 km s~ have a^ £ 10 km s~' and |"z"| = 40 to 80 pc. (2) The range in kinematic properties is reminiscent of the Mira variables and the H-type supergiants with which nearby, optically identified Type II OH/IR sources are associated. The sources -1 9 with AV < 29 km s are probably Miras with periods longer than 500 days and ages of about 10 yr. As AV increases an increasing portion is associated with M-type supergiants of about 10 yr old. (3) The galactic density distribution of Type II OH/IR sources is similar to the distribution of Population I objects: the density increases at R < R , reaches a maximum around R = 4.5 kpc and then decreases steeply inside. This is true for both kinematic groups, indicating that this i:ha- q racteristic distribution has been a stable phenomenon during the last 10 yr. (4) The radial velocity distribution suggests that the Type II OH/IR sources with large AV are concentrated along spiral arms. (5) The 1612 MHz flux density distribution of Type II OH/IR sources is consistent with a luminosity function, I(J(L), that increases with decreasing 1612 MHz luminosity L. Such a luminosity function explains the large differences in L between the nearby, optically identified and the 69

distant, unidentified sources. • 2 ' (b) The combined results of the radio and infrared observations suggest that L "- M where M is the mass loss rate of the star. The shape of i|((L) can be explained by an acceleration of M in the course of the evolution of each star; the ratio between maximum and minimum values of L implies a variation in mass loss of '^ 30 during the evolution of a star. (7) The interstellar UV-radiation field cannot play an important role in the production of OH in the circumstellar shell, in contradiction to theoretical predictions.

References

Baud, B., Habing, H.J., Matthews, H.E., Winnberg, A. 1978a, Chapter II. Blaauw, A. !965, Stars and Stellar Sys., V, p.435, The University of Chicago Press. Bowers, P.F. 1978a, Astron. Astrophys. Suppl. J^U 127. Bowers, P.F. 1978b, Astron. Astrophys., in press. . Rowers, P.F., Kerr, F.J. 1977, Astron. Astrophys. _5_7, 115. f Bowers, P.F., Kerr, F.J. 1978, preprint. j Burton, H.B. 1971, Astron. Astrophys. J¿, 76. Burton, U.B. 1974, Galactic and Extra-Calactic Radio Astronomy. Eds. G.L. Verschuur, K.I. Kellerman, Springer Verlag, Heidelberg, p. 82. Burton, W.B. 1976, Ann. Rev. Astron. Astrophys. Jj4, 275. Burton, W.B., Gordon, H.A. 1978, Astron. Astrophys. 6J3, 7. Caswell, J.L., Haynes, R.F. 1975, Monthly Notices Roy. Astron. Soc. £7_3>bl> '3- Caswell, J.L. 1974, Galactic Radio Astronomy, Eds. F.J. Kerr and S.C. Simonson III, Reidel Publ. Comp. Dordrecht, p. 243. Delhaye, J. 1965, Stars and Stellar Sys. V, p. 61, The University of Chicago Press. Dickinson, D.F., KollbcTB, F.., Yngvesson, S. 1975, Astropl.ys. J. _T99, 131. Eggen, O.J. 1970, Vistas in Astronomy, Ed. A. Beer, Oxford: Pergamon Press, V2_, p. 367. Elitzur, H. , Goldreich, P., Scoville, N. 19/6, Astrophys. J. 2O5_, 384. Evans, N.J. II, Beckwith, S. 1977, Astrophys. J. 2_T7_, 729. Feast, M.K. 1963, Monthly Noliii'S Roy. Astron. Soc. \25^, 367. Fix, J.Ü. , Weisberg, J.M. 1

Jones, T.N., Merrill, K.M., Astrophys. J. 209, 509.

Kerr, F.J.; Bowers, P.F. I974, Astron. Astrophys. 36, 225. Lada, C.J., Reid, M.J. 1978, Astrophys. J. 2J_9, 95. Martinet, L., Mayer, F. 1975, Astron. Astrophys. 44^, 45. Minkowski, R. 1965, Stars and Stellar Sys. V, p. 321, The University of Chicago Press. Moran, J.M., Ball, J.A., Yen, J.L., Schwartz, P.R., Johnston, K.J., Knowles, S.H. 1977, Astrophys. J. 211, 160. Oort, J.H. 1965, Stars and Stellar Sys. V, p. 455, The University of Chicago Press. Schmidt, M. 1965, Stars and Stellar Sys. V, p. 513, The University of Chicago Press. Schultz, G.V., Kreysa, E., Sherwood, W.A., Astron. Astrophys. 5_0, 171. Simonson, S.C., Mader, G.L. 1973, Astron. Astrophys. 2]_, 337. Smak, J.I. 1966, Ann. Rev. Astron. Astrophys. 27, 337. Spitzer, J.I. 1966, Ann. Rev. Astron. Astrophys. 4_, 19. Stothers, R., Leung, K.C., Astron. Astrophys. JJ5, 290. Wielen, R., Highlights of Astronomy, Ed. G. Contopoulos, Reidel Publ. Comp., Dordrecht, 3, 395. Wilson, W.J., Barrett, A.H. 1972, Astron. Astrophys. J^, 385. Winnberg, A., Goss, U.M. Högland, B., Johansson, L.E.B. 1973, Astrophys. Letters JJ3> 125. Winnberg, A. 1976, unpublished contribution to the I.A.U. General Assembly XVI, Grenoble. Wood, P.R., Cahn, J.H., Astrophys. J. 211, 499. Zimmerman, H. 1965, Astr. Nachr. 288, 95. 71

CHAPTER IV: OH/IR SOURCES IN THE NUCLEAR REGION OF THE GALAXY

I Introduction

In Chapter III we have analyzed the galactic distribution and kinematics of Type II OH/IR sources at 1 ^ 10°. We showed that the OH/IR sources may be divided schematically into two kinematic groups, that can be distinguished by the velocity separation AV between the two emission peaks. The OH/IR sources with AV £ 29 km s have kinematic properties similar to extreme Popu- lation I objects: they have a low velocity dispersion with respect to circular motion (£ 10 km s"1). The sources with AV "< 29 km s~' generally show larger deviations from galactic rotation with an r.m.s. value of 30 to 35 km s , indicating that they are older objects probably belonging to a young disk population. The kinematic difference between the two groups is also reflected in their distribution perpendicular to the galactic plane: for the sources with small AV (< 29 km s~') Jzf = 140 pc, for sources with large AV (> 29 km s~') |z| = 60 pc. This range in kinematic properties is reminiscent of the Mira variables and of the M-type supergiants with which the nearby Type II OH/IR sources have been identified (Wilson and Barrett, 1972; Hyland et al., 1972). Hence we concluded in Chapter III that the OH/IR sources, found in systematic 1612 MHz OH surveys at large distances from the Sun are indeed associated with stellar objects in a late stage of evolution. The sources with small AV are probably Hira's with periods larger than 500 days, those with large AV are M-type supergiants. Around AV = 29 km s they consist of a mixture of both types.

In this paper we will study OH/IR sources in the nuclear region of the Galaxy, using the first results of a 1612 MHz OH sky survey at longitudes closer to the Galactic Centre (G.C.) (Chapter I). The distance R to the G.C. is assumed to be 10 kpc.

II Observations and selection effects

In this section we briefly summarize the observations and discuss the influence of selection effects within 2° from the G.C.

II. 1 Observations

The Effelsberg 100 m telescope was used to cover a strip along the galactic equator at 358 < I < 14°, \b\ < 0?5, and some selected regions around S. = 0° at b = +1° and at b - +9° (see Chapter I, Fig. 1). The observations reached a 3a sensitivity of ^ 0.3 Jy at a resolution of 15.6 kHz. A low-sensitivity (3a = 3.6 Jy, 11.7 kHz) search was made with the Dwingeloo 25 m telescope around the G.C. and in an area above the galactic equator. The strip along the equator was then extended, both in latitude and in radial velocity coverage, using the NRAO 43 m telescope and reaching a 3o sensitivity of 0.7 Jy at 15.6 kHz. This is comparable to the sensitivity of the large Dwingeloo survey discussed in Chapters II and III. In total we found 43 Type II OH/IR sour-, ces. Five of these had been detected previously. The new sources were confirmed with the NRAO 43 m telescope in August 1977 and for 42 sources a spectrum with high signal-to-noise ratio, has been obtained, observing with a resolution of 6.5 kHz.

The National Radio Astronomy Observatory is operated by Associated Universities Inc., under contract with the National Science Foundation. 72

Table 1

OH/IR Sources Near the Galactic Centre

AV Name V S6 .5 SI -22 -2 km s"1 km s Jy 10 Wm (epoch 1977.6) (epoch 1977.6) LV* HV LV HV OH 358.1+0. 1 - 26 29 l.l 1.3 1.4 0.8 OH 359.1+1.2 -136 18 1.8 1.6 1.9 1.8 OH 359.4-1. 3 -219 31 4.4 5.3 8.1 7.9 OH 359.5+1.3 + 32 27 1.6 0.4 1.8 0.9 OH 359.7+1.3 + 48 29 0.7 0.8 0.9 1.0 OH 359.8+2.6 + 49 23 4.7 5.3 10.2 9.9 OH 0.0-0. 1 + 113 28 0.8 1.9 0.5 4.2 OH 0.1+5. 1 -143 16 2.6 l.l 2.3 l.l OH 0.I+Q. 1 + 160 28 2.0 3.1 1 .1 II.9 OH 0.3-0.0 + 76 35 1.5 4.4 .3 6.5 OH 0.3-0. 2+ -343 29 4.9 3.5 í .0 7.8 OH 0.4+0. 1 + 140 30 1.0 0.9 .1 1.2 OH 0.5-0 1 + 142 24 0.9 1.4 .3 1.4 OH 0.5-0.8 - 47 34 l.l 0.8 .5 1.9 OH 1.1+0 4 -126 35 0.8 1.3 2.3 1.8 OH 1.1-0 8 + 11 39 25.4 30.5 59.9 60.8 OH 1.2+1 3 + 45 27 5.6 5.3 6.0 5.5 OH 1.3+1 0 - 12 30 4.2 4.2 8.4 5.4 OH 1.5+0 0 -128 27 3.0 4.1 2.7 3.5 OH 1.7-0 0 + 120 29 0.8 4.2 4.7 8.5 OH 2.2-1 7 - 73 36 4.2 6.7 5.1 7.0 OH 2.5+0 3 - 79 15 0.4 0.5 0.3 0.4 OH 2.6-0 5 - 4 46 6.1 7.2 11.2 14.3 OH 3.3-0 3 - 83 27 1.6 1.7 2.0 1.6 OH 3.3+0 0 - 41 32 1.2 1.5 1.3 2.9 OH 3.9+0 0 - 15 32 1.2 l.l 2.7 2.6 OH 4.0-0 .5 - 2 23 0.8 0.4 2.3 0.9 OH 4.4+0 .0 - 1 34 3.2 1.2 5.4 2.1 OH 4.5-0 4 -122 44 1.3 1.9 4.2 3.7 OH 5.0+1 .5++ + 121 28 5 4 3.2 2.5 OH 5.9-0 .4 - 4 34 34.7 1.5 37.9 3.8 OH 6.0+0.3 + 100 12 1.8 1.3 1.8 1.2 OH 6.5-0 .1 + 104 21 0.7 1.2 1.3 2.1 OH 8.0+1 .4 - 22 25 4.2 6.6 7.4 8.2 OH 9.0-0 .1 - 50 33 0.7 1.3 0.8 1.6 OH 9.6+0.4 - 64 31 1.3 2.8 1.8 4.9 OH 9.9-0 .1 + 96 33 1.9 0.8 4.0 1.8 OH 10.0-0.1 + 52 57 0.7 2.7 1.5 5. 1 OH 10.4+0.0 + 61 36 0.6 0.8 0.5 2.2 OH 11.1+0.0 - 22 27 0.9 l.l 0.8 0.7 OH 11.3+0.0 + 81 34 2.1 2.9 3.1 3.7 OH 11.4-0.1 + 105 36 0.5 0.4 0.6 0.9 OH 11.5+0.1 + 42 44 4.2 25.0 18.2 111.4

LV = low velocity; HV = high velocity

+Baud et al. (1965)

^Bowers (1978) 73

i r \ i i ii i i iii

• • I : n. -1 r »_ „ „.

-2Q 14s 12° 10 358° Ï

Fií¡. 1. Distribution of Oll/IR sonríes in ijalaotia aoovdinales. The dashed and the aonbinuous Lineo indicate the aveas surveyed with a So sensitivity of 0.3 Jy and 0.7 ,ly respectively.

In Table 1 all OH/IR stars found in this survey are listed in order of increasing galactic longitude. Columns 2 and 3 give the radial velocity V with respect to l.s.r., taken as the velocity half-way between the two emission peaks, and the velocity separation AV between the peaks. The peak flux density S, , at a resolution of 6.5 kHz and the integrated intensity of the low- and the high velocity components S. are listed in columns 4, 5, 6 and 7. The integrated flux densities have been derived from the 1612 MHz emission profiles presented in Chapter I. In those ca-

ses, where the emission extends over all velocities between the two peaks, Sj(LV) and ST(HV) were determined over all velocities below and above the source velocity respectively. We will refer to the large-scale 1612 MHz survey at I > 10° (Chapter II) as the Dwingeloo survey.

II.2 Selection effects uithin i. 2° fvom Lhe Galaotia Centre The distribution of OH/IR sources is affected by several selection effects. (i) The radial velocity distribution of the OH/IR sources at I £ 3?5 and \b\ £ 0?5 is strongly influenced by broad absorption due to the well-known molecular cloud complexes in the central part of the Galaxy. Our observations, as well as those of Cohen and Few (1976) in the 1665 and 1667 MHz OH line near the G.C., show strong absorption profiles between S, = 2° and 358?5. The absorption is strongly varying as a function of position and of radial velocity, causing steep gradients of the frequency baseline of the spectrum. These steep gradients make the discovery of the characteristic OH/IR emission profiles very difficult, so that a bias results against sources with velocities, that coincide with the absorption lines. This is particularly severe at 359° < i < Io, where very deep absorption is present at practically all velocities between -200 and + 100 km s~'. (ii) A related, selection effect is the possibility that the emission from OH/IR sources behind the molecular clouds is attenuated, if these sources coincide in radial velocity with that of the absorbing clouds. This may be important even in those directions, where optically thick molecular clouds have not yet been found, because of the absence of strong background continuum sources. For example Cohen and Few (1976) have found an optical depth of about 2 in the mainline OH absorption, even at 2 away from the G.C.. Because our 1612 MHz observations indicate similar values 74

Fig. 2. Longitude distribution of souroes found at \b\ ~ 0°.5; (a) all (bi àV * 29 km s~2 (dashed) and Al' < 29 hn s {continuous)

within 3 from the G.C.Jioreground c'.ouds may provide a considerable attenuation of possible background sources, (iii) Within an angular distance of about 0?5 from the G.G. the strong background continuum emission increases the sensitivity limit significantly.

Ill Distribution of OH/IR sources on the sky

Fig. I shows in galactic coordinates the distribution of all OH/IR sources in Table 1. The shape of the surveyed area is adapted from Fig. 1 in Chapter I. Within the continuous lines the 3a sensitivity is 0.7 Jy. Within the dashed line the 3a sensitivity is 0.3 Jy over a velocity -1 -1 range from -220 km s to +220 km s and 0.7 Jy between +220 < |v| < +630 km s . Only one source was found in the range |v| > 220 km s~' (Baud et al., 1975). The three sources at positive latitudes outside the surveyed regions were discovered by Kerr and Bowers (1974). The concentration to the galactic equator is enhanced by the increased sensitivity in the strip at \b\ < 0?5. In retrospect we think that the sensitivity of the survey at latitudes \b\ > 0?5 was insufficient. Hence in our analysis we will concentrate on the results obtained in the strip along the equator. The latitude distribution of sources found in this strip is symmetrical with respect to the galactic equator. The number of sources decreases only by a factor of two between |fc| =0° and 0.5. With a mean latitude \b\ £ 0?5, assuming that many sources are situated near the G.C. at a distance of 10 kpc (arguments for this assumption will be given in the next section), this corresponds to a mean distance to the plane |z| £ 90 pc. This value is consistent with the mean z- distance of both kinematic groups at larger longitudes (Chapr.T III). The longitude distribution of all sources in the strip along the galactic plane at \b\ < 0?5 (inside the dashed line in Fig. 1) is shown in I'ig-re 2a.- We comment on two important properties of Che longitude distribution: (i) the small numcur of sources found and the absence of a strong nuoliw bulge; (ii) the asymmetry around i = 0o. (i) In Chapter III we showed that the small number of sources discovered at Ü. < 20° corresponds to a steep decline in density of OH/IR sources 75

Fig. 3, Radial üeloatty distribution of OH/IR sources as a function of longitude. Curve I indicates maximum radial velocity allowed by pure galactic rotatloyi, according to a mass model by Sanders and Lowinger (197"). Curve II corresponda to the minimum radial velocity for sources at H > A' .

AV < 29 km s-1 AV *29 km s"1

12" l between R = 4.5 and 2.5 kpc. Although the present survey reveals three times more sources per square degree at |b| < 0?5 than the Dwingeloo survey at I < 20°, \b\ < 0?5, this difference can be entirely ascribed to the improved sensitivity of the present survey. Adopting the shape of the luminosity function derived in Chapter III and confirmed here (see section V) we find that the source density near the G.C. at R < 2.5 kpc is as low as the density at 4.5 < R < 2.5 kpc. There is only a weak tendency for an increase around S, = 0°. Though the increase may be underestimated by the selection effects, that play a role at í ^ 2°, it is clear from the longitude distribution that there is no obvious evidence for a strong nuclear bulge component in the density distribution of OH/IR sources, such as for instance is found for the planetary nebulae (c.f. Fig. 2 in Minkow- ski, 1965). (ii) The maximum near I = 0 is due to the clustering of sources within 0?5 from the G.G. corresponding to a radial distance of 90 pc. The asymmetry is rather remarkable: 9 out of 10 sources were found at positive longitudes and only one source was found on the other side, about 2 away from the G.C. This peculiar distribution cannot be explained by the above mentioned selection effects, because the absorption profiles at i < 0° are rather weak. The significance of this asymmetry can be better assessed when a more extended region on the southemhemisphere side of the G.C. is observed with the same sensitivity. Nevertheless it should be noted that the distribution of CO molecular clouds (Bania, 1977) and of the continuum emission (Downes and Maxwell, 1966; Pauls et al., 1976) show the same asymmetry.

In Fig. 2b the longitude distribution of the sources with AV < 29 lem s~' and AV > 29 km s~' is shown separately. The shape of both distributions is quite uncertain be< ->use of the small number of sources; nevertheless there is a weak indication for a relative increase of the number of sources with small AV towards lower longitudes.

IV Radial velocities

An analysis of the radial velocities of the OH/IR sources is required to locate the OH/ ÍR sources along the line of sight. For a descri,.ion of the overall characteristics of the radial 76

velocity distribution we divide the velocity diagram in Fig, 3 schematically in two halves at 8, = 5°. At 8. > 5° the velocities of most sources agree with galactic rutation. Because the density of OH/IR sources outside the solar circle is negligeable {Bowers, 1978), it is unlikely that the few objects found at negative radial velocities are situated on the other side of the Galaxy at distances larger than 20 kpc. Instead, negative velocities probably reflect deviations from circular motion as found for the OH/IR sources at larger longitudes (Johansson et al., 1977; Bowers, 1978; Chapter III). At i¿ < 5° there is a strong increase in the number of sources with large negative velocities. The velocity distribution is centered at 0 km s and it has a total width of about 360 km s at I = 0°, indicating that the majority of sources at 8. < 5 is actually associated with the nuclear region of the Galaxy. The apparent lack of sources with large positive radial velocities V > +120 km s at I > 2° is consistent with a low density of OH/IR sources in the central region of the Galaxy. The symmetry around V = 0 km s , suggests that the bulk of kinetic energy of the OH/IR sources near the G.C. is in random motions, and not in galactic rotation or systematic expansion as has been fouuJ in H I and molecular clouds. Notice that this is equally true for the sources with large and small AV, The importance of the selection effects, mentioned in section II, is illustrated when comparing the radial velocities of the OH/IR sources found in the strip along the galactic plane (|¿)| < 0.5) with the map of the velocif.y distribution of apparent optical depth of the 1667 MHz OH absorption by Cohen and Few (1976). Between JL = 358° and 2°, 9 out of 10 sources lie outside or on the edge of the 0H-absorption velocity range.

IV.1 The sources at I < 5

It is interesting to compare the velocities of the OH/IR sources with the velocities of both very old and very young objects in the nuclear region. Fig. 4 is a diagram of the radial velocities of planetary nebulae (PN) at |£>| < 5°, derived from Perek and Kohoutek (1967). These Popu- lation II or old disk-population objects have been studied extensively in the optical regime (e.g. Minkowski, 1965). Their longitude distribution increases strongly towards the G.C. and they show a moderate concentration towards the galactic equator, although this concentration is certainly underestimated because of the interstellar extinction close to the equator. Regarding the high velocities the i-V diagram of the OH/XR sources is quite similar to that of the planetary nebulae near the G.C.. At I í 5 many PN's are found with large negative radial velocities indicating that the nuclear bulge component of the PN's distribution manifests itself kinematically at about the same distance from the G.C. as that of the OH/IR sources. There is very little evidence for galactic rotation of th" PN's in this longitude range. Only the few PN's, found with high positive radial velocities at i > 2 , which have no counterparts at large negative radial velocities, could possibly be a sign of such a rotational component in the distribution of radial velocities. In fact, these few high-velocity PN's represent the main difference between both velocity distributions. In Fig. 5 the distribution of |v| at I < 5° of both the OH/IR sources (continuous) and the PN's (dashed) is drawn. The distribution of the OH/IR sources has been normalized to the total number of PN's. There is a remarkable agreement between both velocity distributions, indicating that in the Galactic Centre region the OH/IR sources and the planetary nebulae have more or less the same kinematic properties.

Let us now compare the radial velocities of the OH/IR sources at \b\ < 0?5 with the velocity distribution of extreme Population I objects, i.e. with the molecular clouds in the centre of the 77

T I 1 T

'in E

g ••o CE

1 1 1 1 1 14° 12° 10° 8o 6o 4° 2o 0° 358° 120 240 IVllkms-1)

Fig. 4. Radial velocity distribution of planetary Fig. 5. \v\-distribution of OH/IR nebulae known at 14° ^ l¿ 3S8°, \b\ < 5°. Curve I souroes relative to that of the and II are the same as in Fig. 3. planetary nebulae.

Galaxy. Because OH absorption measurements only reveal the existence of the clouds in front of the continuum sources, we use for such a comparison the CO observations by Bania (1977) taken along the galactic equator. Fig. 6 is adapted from his paper; the contours indicate the velocity distribution of CO emission at b = 0°. The emission at 358° < i <2° between -200 and +200 km s~' originates from the massive molecular clouds that lie within 300 pc from the G.C. (Bania, I977). Large non-circular motions are found at I = 0° around -125 and +165 km s~'. This has been interpreted as the emission from a rotating and expanding molecular ring around the G.C. (Scoville, 1972). The part of the ring at I = 0° around -125 km s~' can also be traced out in the OH absorption data from Cohen and Few (1976) and it lies on the near-side of the G.C., in front of the continuum sources. The CO emission at V £ +130 km s~' originates from the part of the ring on the far side of the G.C.. It has no counterpart in the OH absorption, because it is situated behind the strong continuum (e.g. Oort, 1977).

All OH/IR sources found in the strip are plotted in the diagram. At negative radial velocities they avoid the CO emission; this is to be expected from the same tendency with respect to the OH absorption, as previously discussed. Around i = 0° at nositive velocities the situation is somewhat different. Although the OH/IR sources here have been found outside the OH absorption velocity range, they coincide with the high velocity CO-emission that belongs to the far-side of the expanding molecular ring. If the attenuation of OH emission by foreground molecular clouds mentioned in section II indeed plays an important role, this implies that these OH/IR sources are situated in front of the far-side of this ring.

Because of the small number of OH/IR sources and the influence of the above mentioned selection effects it is not clear to what extent the detailed spatial distribution of OH/IR sources 78

sr- 8°

5°

- -343 km O' 359' 358° -300 -250 -200 -150 50 100 150 '200 250 300

Fig. 6. Radial velocity distribution of OH/IR sources at \b\ t 0°.5, superposed on the CO distribution taken from Bania (7977). Dots and crosses as in Fig. S.

near the G.C. is similar to that of the CO. The large negative velocities at S. < 0°, that are present in the OH/IR sources, but not in the CO emission, indicate that, at least kinetnatically, OH/IR sources are different. Nevertheless, bath the OH/IR source density and the CO abundance is very low in the inner part of the Galaxy and they both increase only within about 2° (Ï35O pc) from the G.C. This correspondence in the large-scale distribution is consistent with the general agreeme c in the space distribution between the OH/IR sources and the CO abundance elsewhere in the Galaxy, which was found in Chapter III.

IV.2 Difference in the distribution as a function of àV.

There is some indication for a difference in the spatial distribution of sources with AV < 29 km s with respect to those with AV > 29 km s~ . The relative increase in the number of sources with small AV towards S. = 0° (Fig. 2b), suggests a weak concentration towards the nuclear region of the Galaxy, which is absent for the sources with large AV. The difference in the spatial distribution is emphasized when we consider the radial velocities of the same sample of sources (Fig. 6). Near V = 0 km s , where confusion with nearby foreground OH/IR sources is most likely, five out of the six sources have a large AV. Hence the nuclear bulge contribution of the sources with large AV to the observed numbers at S. < 5° is, if anything, overestimated. The AV-distribution of the present sample of sources (Fig. 7a) is more or less symmetrical with a maximum around 30 km s~' and it extends between 10 and 50 km s~'. The overall distribution is similar to the AV-distribution of the sources found elsewhere in the Galaxy (Chapter III), suggesting that statistically they are the same mixture of objects. In Chapter III we found a more asymmetric distribution, with a very steep decrease below and a more gentle decrease above AV = 29 km s . The symmetry in Fig. 7 indicates a relatively larger number of sources with small AV. The relative decrease in AV with decreasing I is shown quantitatively in Figs. 7b, c and d 79

N —i i l l i l i i i i i i 4 "a " b n all sources 6 14 2 - 1 1 L, "'" - 12 n r- h n - 4 10 - c 2°v< ?< 6° " • n - 2 8 -• r—' 1 1 6 - — .d -6 358°v

2 . r '—i H -2 1 r , , hn 1 1 1 1 ) i 10 20 30 40 50 60 10 20 30 40 50 60

Fij. ?. &V-distribution of all OH/IR oouvaes la) und of the aoutvco found at \h\ ',[ 0°5 for three 'oiijitudc intervals, 'joniain in

for the sources found in the narrow strip along the galactic plane. The maximum of the distribution shifts from AV ^ 32 km s~' at I > 2° to AV ^ 27 km s~' in the inner 2° around the G.C.. Also the distribution around S. = 0 seems to be narrower: no sources have been found with AV ;• 40 km s~' or < 20 km s~'. We infer that the density distribution of the OH/IR sources with large and small AV near the nuclear region are marginally different, with a weak indication for a lower mean value of AV and somewhat narrower ranf.e in the distribution of AV within a distance of about 350 pc from the G.C., compared to the AV-distribution of sources elsewhere in the Galaxy.

V Emission properties and luminosities

The emission properties of the OH/IR sources have been discussed extensively in Chapter III. It was shown there that the 1612 MHz luminosities could be represented by a simple luminosity function iKD « L " , where a = 1.65 ± 0.1. This luminosity function qualitatively explains the high luminosities of the distant sources as well as the lower luminosities of the nearby, optically identified OH/IR stars.

Since most sources within 5° from the G.C. are at a distance of 10 t 0.9 kpc (when we assume

RQ = 10 kpc), their 1612 MHz luminosities are known to an accuracy of about 18% and hence one can derive their luminosity function in a direct way. We take those sources found in the strip along the plane at Ï < 5°, excluding the sources within 0?5 from the G.C., because here the sensitivity of the survey is affected by the intense and strongly varying background continuum emis-

sion. Fig. 8 (full drawn) shows a histogram of the observed peak flux densities S, g. As in Chap- ter III the harmonic mean of the peak flux density of both velocity components is used (S, _ = "b.5 ^LV^ ' S6.5 (Nv))- To take into account the incompleteness of the survey we divide the

observed number of sources in each interval of log S& 5 by the detection probability P(log S, 5) (c.f. Chapter III). This leads to a corrected flux density distribution (dashed in Fig. 8). 80

i i 1.6

1.4 Fig. 8. Flux density distribution of the OH/IR 1.2 sources at I < 5°, \b\ ¿ 0°5, excluding the sour-

1.0 - ~P(log%s)- 1.0 ces within 0°.i from the G.C. (draw>i). The dashed line is the distribution corrected for the incom- = 0.8 z A pleteness of the survey. The detection probability o 0.6 0.6 õ P has been calculated with a So deteotion limit 5' of 0.5 Jy. S„ r has been defined in the text. 0.4 0.4 ^ / o 0.2 0.2 g!

0.0 0.0 1? -0.4 0.0 04 0.8 1.2

This corrected distribution can be approximated by a straight line with a slope of -1.6 t 0.5, where the uncertainty corresponds to the formal statistical error due to the small numbers of sources. This slope agrees remarkably well with the value of a determined in Chapter III for the sources found at larger longitudes. The agreement emphasizes, that the OH/IR sources in the nuclear bulge and those in the galactic disk are quite similar objects in their emission properties. Apart from a possible difference in the relative numbers of sources with small and large AV there is no indication in the present data that the OH/IR sources in the nuclear region of the Galaxy are physically different from those found elsewhere. The distribution of equivalent width of the individual velocity components, defined as the ratio between S and S, - and the distribution of the ratio between the LV and HV flux densities are similar to the same distributions of the sources at larger longitudes, first discussed by Johansson et al. (1977) and subsequently confirmed in Chapter III.

VI Discussion

The first results from the 1612 MHz survey of the nuclear region of the Galaxy can be summarized in the following points. a) OH/IR sources have been detected in the region near the Galactic Centre. Their physical association with the nucleus is apparent from the radial velocity distribution at t < 5°, which has a total width of about 360 km s , centered on V = 0 km s~ . b) The density of OH/IR sources in the inner region of the Galaxy is very low; it is consistent with the steep density decrease at R < 4.5 kpc found in Chapter III. Within 0?5 rrom the Centre (90 kpc) the density appears to increase again. Because the detection of OH/IR sources in the galactic plane at i < 2 is strongly hampered by various selection effects, the actual density increase of OH/IR sources within 90 pc from the G.C. is probably much stronger. c) There is some indication for a small enhancement in relative number of OH/IR sources with AV < 29 km s~' in the inner 350 pc. d) A comparison between the luminosity distribution and the statistical properties of the 1612 MHz OH emission profiles of the presently discussed sample and that of the sample discussed in 81

Chapter III indicates that, with respect to their emission characteristics, the OH/IR sources in the Galactic Centre region are physically not significantly different from the OH/IR sources found elsewhere in the Galaxy. The ahsence of a strong density increase for OH/IR sources towards the nuclear bulge, which is present in the distribution of planetary nebulae, confirms the conclusion in Chapter III, that the stellar objects associated with the OH/IR maser sources are much younger objects. Neverthe- less, the velocity distribution of the OH/IR sources at S. < 5° indicates that kinematically they behave like the PN's inside the nuclear bulge; they show large random motions and there is no obvious evidence for net rotation. By inference one would expect also a more extended z-distribution of the OH/IR sources at S. < 5°, as compared to the z-distribution at I > 5°. Observations at higher latitudes with the same sensitivity as in the strip, (\b\ < 0?5) are presently underway. These observations should provide important information on K , the force perpendicular to the galactic plane in the nuclear region.

An interesting problem is posed by the OH/IR sources at I < 5° and AV > 31 km s . In Chap- ter III we concluded that such sources are generally associated with M-type supergiants and have ages of about 10 yr. However, such an age is much too short for these stars to have acquired large random velocities up to 100 km s , as is apparent from the sources with large negative radial velocities (OH 359.4-1.3, OH 1.1+0.4, OH 2.2-1.7, OH 4.5-0.4). One therefore concludes that these objects must be older, probably belonging to the population of Mira variables. However, we showed in Chapter III that the [liras generally have AV < 29 km s . Hence the OH/IR sources in the nuclear bulge appear to be intrinsically different in their AV distribution from the similar sources at S. > 5 : they may have larger AV and longer periods than is expected from their age (i.e. their mass). In this connection it is interesting to note the suggestion by Lloyd-Evans (1976). He studied long-period variables near the G.C. and found a relative increase in the number of Miras with larger periods (P > 400 days) as one approaches the Centre. He explained this increase as an effect of higher metal abundance. Because AV and P are positively correlated (Dickinson et al., 1975), the increased AV could also be caused by a higher metal abundance.

References

Bania, T.M. 1977, Astrophys. J. 216, 381. Baud, B., Habing, H.J., Matthews, H.E., O'Sullivan, J.D., Winnberg, A. 1975, Nature 25£, 406. Bowers, P.F. 1978, Astron. Astrophys., in press. Cohen, R.J., Few, R.W. 1976, Monthly Notices Roy. Astron. Soc. J_76, 495. Dickinson, D.F., Kollberg, E., Yngvesson, S. 1975, Astrophys. J. J^9_, 131. Downes, D., Maxwell, A. 1966, Astrophys. J. ^46_, 653. Hyland, A.R., Becklin, E.E., Frogel, J.A., Neugebauer, G. 1972, Astron. Astrophys. |6^ 204. Johansson, L.E.B., Andersson, C., Goss, W.M., Winnberg, A. 1977, Astron. Astrophys. 54,323. Kerr, F.J., Bowers, P.F. 1974, Astron. Astrophys. 36^, 225. Lloyd-Evans, T. 1976, Monthly Notices Roy. Astron. Soc. 174, 169. Minkowski, R. 1965, Stars and Stellar Systems V, 321. Oort, J.H. 1977, Ann. Rev. Astron. Astrophys. j_5, 295. Pauls, T., Downes, D., Mezger, P.G., Churchwell, E. 1976, Astron. Astrophys. 46^, 407. Perek, L., Kohoutek, L., 1967, Catalogue of Galactic Planetary Nebulae, Prague. Sanders, R.H., Lowinger, T. 1972, Astron. J. ]T_, 292. Scoville, N.Z. 1972, Astrophys. J. Letters 175, L127. Wilson, W.J., Barrett, A.H., Astron. Astrophys. 17, 385. 82

SAMENVATTING

In het begin van de zeventiger jaren ontdekte men zeer sterke OH emissie lijnen van het OH molekuul op radiogolflengten, afkomstig van Mira sterren en M superreuzen. Deze zeer koele, variabele sterren, die zich in een laat stadium van hun ontwikkeling bevinden, blijken zeer veel massa te verliezen en vormen zodoende een expanderende circumstellaire gasschil, waarin de OH lijn straling wordt uitgezonden. Als gevolg van de speciale fysische omstandigheden in de schil wordt deze lijn straling bovendien nog krachtig versterkt. Deze zogenaamde OH/IR bronnen, met hun karakteristieke dubbele OH lijn profiel op een golflengte van 18 cm en hun vaak sterke infrarood emissie, boden een nieuwe en onverwachte mogelijkheid tot de bestudering van de ruimtelijke verdeling en kinematische eigenschappen van stellaire objecten door de i;ehele Melkweg met behulp van radio sterrenkundige methoden en dus ongehinderd door verduistering van interstellair stof. Deze verduistering is altijd een ernstige beperking geweest in optische studies. De eerste systematische radio waarnemingen, gedaan in Zweden en Australië, toonden de aan- wezigheid aan van een aantal nieuwe OH/IR bronnen. Hun sterke koncentratie naar het Melkwegvlak bevestigde het vermoeden, dat deze objecten tot op zeer grote afstanden van de Zon waarneembaar waren. Geen van deze bronnen kon echter optisch geïdentificeerd worden, zodat een associatie met Mira's of M superreuzen onzeker bleef. De onzekerheid over de aard van deze ongeidentifeerde OH/IR bronnen werd bovendien benadrukt door hun soms grote radiele snelheden en het feit dat de OH straling veel sterker bleek te zijn dan dat van de nabije, optisch geidentificeerde bronnen. Het doel van het onderzoek, dat in dit proefschrift staat beschreven, is een uitbreiding v. - hef aantal bekende OH/IR bronnen d.m.v. radio astronomische waarnemingen en een analyse van de ' .— ?

jnnen • t»-ii ->n in hun populatie eigenschappen. noüf stuk I beschrijft de waarnemingen van het centrale deel van de Melkweg, uitgevoerd met d Liwinyulcn ?3 ¡n t>. lescoop, de Effelsberg 100 m telescoop (West Duitsland) en de NRAO 43 m telescoop fVereuigde Staten). Deze waarnemingen leverden 38 nieuwe OH/IR bronnen op. .inofaFtuk II behandelt de grote survey die vrijwel het gehele Melkwegvlak, zichtbaar op het no^rdelxjk halfrond, bestrijkt. Deze waarnemingen, uitgevoerd met de Dwingeloo telescoop leverden 3 . nieuwe bronnen op. Ook werden enkele reeds bekende bronnen waargenomen. Dit hoofdstuk geeft ook een beschrijving van de kompleetheid van de Dwingeloo survey. Een atlas van OH Lijn profielen van vrijwel alle besproken Objekten wordt gerepresenteerd. In hoofdstuk III wordt een analyse gegeven van de ruimtelijke verdeling en de snelheden van 114 reeds bekende en nieuwe OH/IR bronnen, die gevonden zijn in het gebied van de Dwingeloo survey. Het blijkt dat de parameter AV (het snelheids verschil tussen de emissie pieken) een goed kriterium is voor een populatie klassifikatie. Bronnen met grote AV zijn jonge objekten met kleine peculiaire snelheden; zij zijn waarschijnlijk geassocieerd met M superreuzen. De bronnen met kleine LI hebben veel grotere peculiaire snelheden en lijken in dat opzicht sttiK op de Mira sterren. Modelberekeningen tonen aan, dat beide groepen sterren een maximum in de dichtheidsverdeling op 5 kpc van het Melkweg Centrum vertonen en een sterke afval in de dichtheid aan weerszijden van dit maximum. De karakteristieke dichtheidsverdeling van OH/IR bronnen toont aan dat deze sterren geen voorlopers kunnen zijn van de planetaire nevels, in tegenstelling tot suggesties van andere auteurs. De grote AV bronnen lijken in spiraalarmen te zijn geconcentreerd. De overeenkomst in ruimtelijke struktuur met de gasvcrdeling maakt een datering van deze gasverdeling mogelijk. De radio lichtkracht funktic van OH/IR bronnen komt overeen met een toenemend aantal objekten bij afnemende helderheid. Dit wijst op een toename in het massa verlies van deze sterren in 83

de loop van hun ontwikkeling als OH/IR bron. In Hoofdstuk IV worden de resultaten van Hoofdstuk I geanalyseerd. Het centrale deel van de Melkweg bevat relatief weinig OH/IR sterren, in overeenstemming met de konklusies over hun dichtheidsverdeling in Hoofdstuk III. De radiële snelheidsverdeling van deze sterren binnen 5 lengte lijkt sterk op die van de planetaire nevels. De ruimtelijke verdeling, echter, lijkt meer op die van het gas. Een vergelijking met de snelheidsverdeling van het gas blijft moeilijk vanwege ver- schillende selektie effekten, welke in het begin van dit hoofdstuk besproken worden. De AV-verdeling van de sterren binnen 90 pc van het Melkweg Centrum lijkt af te wijken van de AV-verdeling van sterren elders in de Melkweg. Dit zou kunnen wijzen op een hogere metaal abundantie. De resultaten van het onderzoek, met name de gevonden korrelatie tussen AV en populatie eigenschappen, tonen duidelijk aan, dat een verdere studie van de OH/IR bronnen een waardevolle bijdrage kan leveren aan de kennis over de dynamische eigenschappen van de evolutie van de Melk- weg. 84

STUDIE OVERZICHT

Na het behalen van het einddiploma Gymnasium ß in 1967, verbleef ik gedurende een jaar in de Vereningde Staten en studeerde aldaar aan het Bowdoin College te Brunswick, Haine. In september 1968 begon ik met mijn studie aan de Rijksuniversiteit te Leiden en behaalde het kandidaatsexa- men sterrenkunde en natuurkunde met bijvak wiskunde in februari 1971. Voor mijn doktoraal oplei- ding volgde ik de colleges van de hoogleraren H.C, van de Hulst, R.G. Conway, P. Mazur, alsmede van Dr. H.W. Capel, Dr. J. Tinbergen en Dr. C. Zwaan. Tevens volgde ik een interakademiaal kolle- ge. In maart 1974 behaalde ik het doktoraal examen sterrenkunde met bijvak mechanica. Hierna werd ik aangesteld als tijdelijk wetenschappelijk medewerker aan de Leidse Sterrewacht en bezocht gedurende 4 maanden het Harvard College Observatory, te Boston. In augustus 1974 kwam ik in dienst van Z.W.O. en begon met het onderzoek dat in dit proefschrift beschreven is. In december 1975 en augustus 1977 verrichtte ik waarnemingen met de N.R.A.O. 43 m telescoop te Green Bank (V.S.). Tevens bezocht ik verscheidene malen het Max Planck Institut für Radioastronomie te Bonn en deed ik waarnemingen met de Effelsberg 100 m telescoop in samenwerking met Dr. H.J. Habing, Dr. H.E. Matthews en Dr. A. Uinnberg. In september 1976 bezocht ik het IAU congres XVI te Grenoble; ook nam ik deel aan het IAU Colloquium 42 te Bamberg. Sinds 1 januari 1978 ben ik weer in dienst van de Rijksuniversiteit te Leiden,als doktoraal assistent verbonden aan de Leidse Sterrewacht.

DANKBETUIGING

Dit proefschrift is tot stand gekomen mede dank zij de hulp en de inspanning van velen. De prettige samenwerking net de medewerkers van de S.R.Z.M, gedurende de afgelopen vier jaren heb ik zeer gewaardeerd. In het bijzonder dank ik Jean Casse, Jaap Visser, Lud Sondaar on Albert Koeling voor hun voortdurende inspanningen aan de zeer koele 18 cm ontvanger; John O'Sullivan voor zijn onmisbare liulp bij het testen van de apparatuur en het oplossen van de talrijke problemen, die bij een langdurig waarneemprogramma onvermijdelijk zijn; Cees Slottje voor zijn inspanningen aan het verbeteren van de Uwingeloo telescoop en voor een rustige Zon; en Hans Tenkink, Ilouite Kramer en Harm Meijer, zonder wiens dagelijkse supervisie geen van du ruim 20000 spoktra zou zijn geregistreerd. Ook dank ik NRAO voor de genoten gastvrijheid; Tom Cram was zeer behulpzaam bij de waarnemingen met de 43 m telescoop.

Veel dank ben ik verschuldigd aan mijn mede-auteurs Anders Winnberg en Henry Matthews voor de vruchtbare samenwerking. De stimulerende discussies met hen en ook met Elly Dekker waren onmis- baar in de voortgang van het onderzoek. Dankzij de goede sfeer op de Leidse Sterrewacht heb ik altijd met veel plezier kunnen werken. Hiervoor dank ik allen hartelijk. Tenslotte datik ik mijn promotor voor zijn enthousiasme en voortdurende inzet bij het begeleiden van dit onderzoek. Z.U.O. en het Kerkhoven Bosscha Fonds ben ik erkentelijk voor hun financiële steun.

Typewerk : L.R. de Leeuw

Grafische vormgeving: J.J. Ober, W. Rijnsburger

Fotografie : W.J. Brokaar, L.A. Zuyderduin

Korrekties : F. van Huut Stellingen behorende bij het proefschrift van B. Baud 1. Er bestaan goede, indirekte aanwijzingen dat de ruimtelijke verdeling van Type I OH/IR sterren (Miras met perioden korter dan 4QC dagen, geassocieerd met OH hoofdlijn emissie) sterk is geconcentreerd naar het centrale deel van de Melkweg, Onderzoek naar de ruimtelijke verdeling van deze sterren zou een nauwkeurige datering van het ontstaan van het maximum in de storvor- mingsaktiviteit op 5 kpc van het Melkweg Centrum mogelijk maken.

2. Het aantal Mira variabelen met perioden langer dan 400 dagen wordt sterk onderschat door verscheidene waarneem selektie effekten. Het is dan ook onjuist om aan de hand van de waargenomen periode verdeling zonder meer konklusiea te verbinden betreffende het tijdsverloop van massa- verlies van Miras. K,iíalis el al., 1977, Astrophys. J. 2_[6, 526. 'i

3. Hat is zinloos de geïntegreerde fluxdichtheid van een OH/IR bron te vermelden zonder de presen- rqtie van het betreffende OH lijnproflel cf zonder een opgave van hst snclheids interval, waar- over is geïntegreerd.

De in dit proefschrift gevonden relatie tussen enerzijds de kinematische eigenschappen van OH/IR sterren °n anderzijds hun snelheidsverschil tussen de twee OH emissie pieken, AV, toont aan dat AV een betere maat is voor de stellaire leeftijd dan de periode van helderheids varia- tie in de tijd, zoals gevonden door Feaat. Feast, 1963. Monthly Notices Roy. Astron. Soe. 125, 367.

5. Een uitbreiding van liet onderzoek naar de verschillen in ruimtelijke verdeling tussen type B en type C planetaire nevels is van het grootste belang voor een beter inzicht in de relatie tussen Mira sterren en planetaire nevels. Greig, 1972, Astron. Astrophys. _18_, 70. C.uckworth, 1974, Astron. J. 79, 1384. Het verschil bij galaktische lengte kleiner dan 20° tussen de waargenomen snelheidsverdeling van OH/IR sterren met AV $ 26 km s en de snelheidsverdeling, welke op grond van de model berekeningen voorspeld wordt, duidt ofwel op een niet-axiaal symmetrische dichtheidsverdeling ofwel op een lagere rotatie snelheid in de binnenste delen van de Melkweg dan doorgaans wordt aangenomen. Met behulp van^afstandsbepalingen van de individuele sterren kan een onderscheid tussen beide mogelijkheden gemaakt worden. Fi". 4, Hoofdstuk III van dit proefschrift.

7. Zonder de beschikbaarheid van infrarood waarneemfaciliteiten vanaf de grond is de deelname aan de IRAS satelliet voor de Nederlandse sterrenkunde een weinip, zinvolle inspanning.

8. Kleine Leleskopen, mits voorzien van goede randapparatuur, kunnen evenzeer een waardevolle bijdrage aan het astronomisch onderzoek leveren als de enkele zeer grote teleskopea. De alge- mene erkenning van de juistheid van deze bewering betekent echter tevens haar ontkrachting.

9. Maar al te vaak worden in politieke diskussies de begrippen wetenschap en technologie met elkaar verward.

10. De departementale splitsing van Kunsten en Wetenschappen betekent een miskenning van de aard van een groot deel van het wetenschappelijk werk. Han, /•, iÜ«r/„,,„,„, ,/,,/.,„„ ,/r ..,:/,.- ,.. ,'r „•/,, m.<¿i: m.t,/.• ¿,/:,,,/. i ,/r..,,,,,,,,/, ,.,/„,•; ;•,;/„r,..,/,.., <„/,•„,„„,.