INFORMATION TO L'SERS
This reproduction was made from a copy of a document sent to us for microfilming. While the most advanced technology has been used to photograph and reproduce this document, the quality of the reproduction is heavily dependent upon the quality of the material submitted.
The following explanation of techniques is provided to help clarify markings or notations which may appear on this reproduction.
1. The sign or “target” for pages apparently lacking from the document photographed is “Missing Page(s)”. If it was possible to obtain the missing page(s) or section, they are spliced into the film along with adjacent pages. This may have necessitated cutting through an image and duplicating adjacent pages to assure complete continuity.
2. When an image on the film is obliterated with a round black mark, it is an indication of either blurred copy because of movement during exposure, duplicate copy, or copyrighted materials that should not have been filmed. For blurred pages, a good image of the page can be found in the adjacent frame. If copyrighted materials were deleted, a target note will appear listing the pages in the adjacent frame.
3. When a map, drawing or chart, etc., is part of the material being photographed, a definite method of “sectioning” the material has been followed. It is customary to begin filming at the upper left hand comer of a large sheet and to continue from left to right in equal sections with small overlaps. If necessary, sectioning is continued again—beginning below the first row and continuing on until complete.
4. For illustrations that cannot be satisfactorily reproduced by xerographic means, photographic prints can be purchased at additional cost and inserted into your xerographic copy. These prints are available upon request from the Dissertations Customer Services Department.
5. Some pages in any document may have indistinct print. In all cases the best available copy has been filmed.
U n iv erse Microfilms International 300 N. Zeeb Road Ann Arbor, Ml 48106 8518929
Crenshaw, Daniel Michael
AN ANALYSIS OF THE BROAD EMISSION LINE PROFILES OF SEYFERT 1 GALAXIES
The Ohio State University PH.D. 1985
University Microfilms International 300 N. Zeeb Road, Ann Arbor, Ml 48106
Copyright 1985 by Crenshaw, Daniel Michael All Rights Reserved An Analysis of the Broad Emission Line Profiles of Seyfert 1 Galaxies
DISSEBTATION
Presented in Partial Fulfillment of the Requireaents for
the Degree Doctor of Philosophy in the Graduate
School of The Ohio State University
By
Daniel Michael Crenshaw* B.S.
The Ohio S tate U niversity
1985
Reading Coaaittee: Approved By
Dr. Eugene R. C a p rio tti
Dr. David G. Lawrie
Dr. Bradley H. Peterson adley Hi Ml Peterson* Pet To Jo - i i - ACKIOULIbGEBENTS This dissertation would not have been possible without the guidance of ay adviser. Professor Bradley fl. Peterson. 1 would like to thank Professor Peterson for his original suggestion of the nain topic of this dissertation, and for his valuable coaaents and advice on various aspects of this work. I would like to thank Professor Eugene B. C ap rio tti, for his helpful suggestions and encourageaent throughout the course of ay research on this topic. 1 aa also indebt ed to Professor David G. Lawrie, for his suggestions and coaaents concerning earlier drafts of this aanuscript. Dr. Paul L. Byard is responsible for the development of the Ohio State University iaage dissector scanner, which proved to be an ideal detector for this study. I an alsc grateful to Dr. Arthur A. Hoag and the staff of Lowell Observatory, for their hospitality during ay visits to Flagstaff. Special thanks go to Dr. Karie A. Beyers, for her assistance with the observations and contributions to the IDS reduction software needed for the analysis of the data. - i i i - Finally, I would like to express ay gratitude to ay wife, Jo, for her constant lowe and support. I aa p a rtic u larly proud of her own acadeaic success, which she attained despite aany sacrifices aade on ay behalf. - i» - m i March 16, 1957. . . Born - Rock Hill, South Carolina 1979 ...... B. S., Georgia State University, Atlanta, Georgia 1979 - 1982 . . . . Teaching Associate, Department of Astronomy, The Ohio State University Columbus, Ohio 1982 - 1983 . . . . Research Assistant, Department of Astronomy, The Ohio Sta te Uni ve rs i t y Columbus, Ohio 1983, Lovell Observatory Summer Fellov, Lovell Observatory, Flagstaff, A rizcna 1983 - 1989 . Perkins Research Assistant, Department of Astronomy, The Ohio State University, Columbus, Ohio 19 89 - 1985 . . . . Teaching Associate, Department of Astronomy, The Ohio State University Columbus, Ohio PUBLICATIONS "Redshifts of 16 Harkarian Galaxies", D. H. Crenshav, B. H. Peterson, C. B. Foltz, and P. L. Byard, pub, Astron. Soc. Pacific, 94, 16-18, (1982). "The Variability of the Spectrum of Akn 120", B. If. Peterson, C. B. Foltz, H. S. Hiller, R. M. Vagner, D. H- Crenshav, K. A. Meyers, and P. L. Byard, Astron, J.. 88, 926-933 (1983). "Variability of the Emission-Line Spectra and Optical Continua of Seyfert 1 Galaxies. II.", B. H. Peterson, C. B. Foltz, D. M. Crenshav, K. A. Meyers, and P. L. Byard, Astronhvs. J .. 279, 529-540 (1984). "the Effects of Stellar Absorption Features on the Broad- Line Profiles of Seyfert 1 Galaxies", D. *1* Crenshav and B. B. Peterson, Astrophvs. J.. 291, 677 (1985). "Variability of the Eaission-Line Spectra and Optical Continua of Seyfert 1 Galaxies. III. Results for a Hoaogeneous Saaple", B. H. Peterson, D. W. Crenshav, and K. A. Beyers, As+roohys J. . 298, in press (1985). ABSIBACTS "The Effects of Stellar Absorption Features on the Broad- Line Profiles of Seyfert 1 Galaxies", D. H. Crenshav, -BnHx-ix-ix-S*, 16, 659 (1986). "The Broad Emission Line P ro files of Seyfert 1 Galaxies", D. D. Crenshav, Bull. A. A. S.. 16, 98B (1986). FIELDS OF STUDY Hajor Field: Astronosy Studies in Seyfert 1 Galaxies and QSOs. Professors Bradley B • Peterson and Eugene R. Capriotti Studies in Eaission-Line Galaxies. Professor Bradley N. Peterson Studies in Early- and Late-Type Stars. Professors Arne E. Slettebak and Phillip C. Keenan - v i - TIBLE OF COITEVTS Page Dedication ...... i i Acknowledge neats ...... i i i V i t a ...... v List of Figores ...... ix List of Tables ...... x i Chapter I. introduction ...... 1 XI. Observations ...... 10 2.1 Instrumentation ...... 10 2.2 Observatioual Parameters ...... 14 2.3 Data Seduction ...... 17 III. Contanination of the Profiles by S t e l la r Absorption Features ...... 20 3.1 Introduction ...... 20 3.2 Synthetic Profiles ...... 21 3.3 Stellar Fractions ...... • 30 3.4 Removal of the Stellar Contanination • 36 3.5 Discussion 42 - v i i - IT. Contaainatioa of the Profiles by Baission F ea tu res ...... 44 4. I introduction ...... 44 4.2 continuua Banges ...... 47 4.3 Heaoval of the Narrow Lines ...... 48 4.4 Reaoval of the Shelf of Baission froa H 6 ...... 53 4.5 Discussion ...... 62 T. The Decontaeiaated Profiles and Profile Katios 64 5.1 Resolution Corrections ...... 64 5.2 Analysis of the Profiles ...... 65 5.3 Analysis of the Profile Ratios .... 71 TI. Interpretation of the Profile Ratios ...... 78 6.1 Relation to Profile Hidths ...... 73 6.2 Relation to Luainosity ...... S3 6.3 Suaaary ...... 30 hPPSIOIXBS A. Broad-Line Profiles ...... 94 B. P r o f ile B a tio s ...... 107 Bibliography ...... 115 - w iii - LIST OP FI60KSS Figare Page 1. fiesolution as a function of channel p o sitio n ...... 13 2. Synthetic profiles ...... 25 3. Measured properties of the synthetic p ro file s ...... 27 4. Asyanetry of the synthetic H6 profiles .... 23 5. Portions of the spectra of M32 and three Seyfert 1 galaxies ...... 31 6. IDS scan of HGC 3516: observed (upper) and corrected ...... 38 7. IDS scan of NGC 4593: observed (upper) and corrected ...... 39 8. IDS scan of Mrk 590: observed (upper) and corrected ...... 40 9. IBS scan of Mrk 335 (observed) ...... 41 10. Mrk 509 deolend assuning shelf is entirely F e l l ...... 55 11. Mrk 110 deblend assuaing shelf is entirely F e l l ...... 56 12. Mrk 509 deblend assuaing shelf is Fe II and broad [0 III ] ...... 57 13. Mrk 110 deblend assuaing shelf is Fe II and broad £0 I II ] ...... 58 14. H8/Ha p ro file r a t i c for NGC 3516 ...... 73 15. H6/Ha p ro file r a tio for Mrk 509 ...... 74 - ix - 16. HB/flo pro file r a tio for Hrk 335 ...... 75 17. HB/Ha profile ratios for two Seyfert 1 galaxies ...... 80 18. The FWHH of HB plotted against the range in He/Ha...... 01 19. The nonthereal luminosity plotted aqainst the range in Hb/H 0 ...... 36 20. Broad eaission line profiles of Seyfert 1 galaxies ...... 94 21. Profile ratios of Seyfert 1 galaxies ..... 95 - x - LIST Of TkBLBS T able Page 1. Journal of Observations ...... 16 2. stellar Fractions ...... 33 3. Narrow-Line Intensities ([0 ill] X5Q07 = 1.0) . . 52 4. Properties of tiie Extended Shelf of Eaission . . 61 5. Broad-Line Intensities ([0 III] X5007 = 1.0) • • 6b 6. Broad-Line widths and hsyaaetries ...... 70 C hapter I IITBODUCTIOK Seyfert galaxies have extremely bright nuclei that shov signs of strong activity. Although Seyfert galaxies were first identified in the early 1940s (Seyfert 1943), their importance was not recognized until decades later when their similarity to guasars, the brightest known objects in the universe, became apparent. In 1963, Haarten Schmidt found that the spectra of guasars are highly redshifted. He realized that if guasars are at the distances implied by their redshifts, their luminosities in some cases must exceed those of the brightest known galaxies by several orders of magnitude. However the rapid variability of some guasars suggests that the continuum source, which is responsible for most of the emitted radiation, can be smaller than a few light-days in size. Even though much progress has been made, the basic process by which so much ra d ia tio n i s em itted from such a sm all volume of space is still not understood. Soon after Schmidt's, discovery it mas realized that the nuclei of Seyfert galaxies are very similar to many qua sars. Ia tact# Seyfert nuclei and siailar objects# includ ing guasars# are new grouped together into one class of objects known as active galactic nuclei (AGN). Known Sey fert galaxies are such closer to us than known guasars# and therefore they are aucb brighter in apparent aagnitude on average. Thus the iaportance of Seyfert galaxies is that they provide us with an opportunity to investigate th5 basic physical processes occurring in AGN with even aoder- ate aperture telescopes. On the basis of spectroscopy# Seyfert galaxies can be divided into types 1 and 2 (Khachikian and Heedaan 1974). Seyfert 1 galaxies have optical spectra that consist of four distinct components: 1. A nontheraal continuum nornally c o n trib u te s most of the total light in saall apertures and is well fit by a power-law# Fy « u”a (where a is typically in the range 1.0 to 1.5)# at wavelengths between 4000 A and 8000 A (in the rest fraae of the galaxy). 2. A contribution fron the host galaxy# consisting prima rily of a stellar continuua and stellar absorption features# may be significant if the nontheraal contin- uui luainosity is rather low. 3. Broad emission lines with typical full widths at zero intensity of order 10* kn are present. The broad- line region (BLB) is thought to contain clouds of eaitting gas traveling at high speeds and is probably 0. 1 pc or less in size. The lack of strong to 111] aa** 959, 5007 e aissio n froa th is region and the observed strength of C ill] 11909 eaission in the ultraviolet suggests that the electron density in a cloud is soaewhere in the range 10* to 10io ca~3 on average. h. Narrow eaission lines with typical full widths at zero intensity of order 103 ka s_l are present. The narrow-line region (NLR) is also thought to contain clouds of gas traveling at soaewhat lower speeds and is likely to be 102 to 103 pc in size. NLR clouds are characterized by auch lower densities in the range 103 to 10* cm-3 . The main difference between the two types of Seyferts is that Seyfert 2 galaxies lack the broad eaission lines dis cussed above, although other differences exist as as well (Ueedaan 1977; Osterbrock 1978a). This dissertation is a study of the optical spectra of Seyfert 1 galaxies; in particular we will concentrate on the broad eaission lines and their profiles. One of the most basic unanswered guestions concerning the BLR is: Hhat is the velocity distribution and eaissivity of the BLR gas as a function of position relative to the nontheraal continuua source? Specifically, is the BLR gas undergoing gravitational infall, radiatively-driven outflow, outflow due to an explosive event, turbulence, rotation, or some other type of notion? An answer to these questions will provide clues to the nature of the nontheraal continuum source, since it probably dictates the dynanics of the BLF. Although we do not know the answers to the above ques tions, we do know that photoionizing radiation from the nontheraal continuua source provides the energy neccesary to produce the eaission lines in both the BLB and RLE. Early photoiocization models were fairly successful in explaining the observed broad-line ratios until the discov ery by Baldwin (1977) of the **La/HB problem". Baldwin finds rrom a composite spectrum of low- and high- redshift quasars that La/HB = 3, a factor of 10 lower than predicted by models. Subsequent studies (Hu, Boggess, and Gull 1933) find that for Seyfert 1 galaxies La/HB = 5 on average. Recent photoionization aodels (Kvan and Krolik 1931; Weisheit, Shields, and Tarter 1961) are much more success ful in matching the observed La/HB ratios and the other strong optical and ultraviolet relative line intensities. These new models consider a BLB cloud which is very opti cally thick to ionizing radiation at the Lyman limit. As in the earlier aodels, the continuua radiation produces an HII zone in which acst of the cooling is done by the ultra v io le t lin e s La, c IV A1549, C I I I ] Al 909, and 0 VI XIOJh. 5 The improvement to earlier nodels is due to the realization that continuum X-rays penetrate further into the cloud past the HI1 zone to create a partially ionized zone (PIZ) in which approximately 10 - 20% of the hydrogen is ionized. Due to the large La optical depths in the PIZ, a signifi cant amount of hydrogen is maintained in excited states. Host of the cooling in this zone is done by collisional e x cita tio n of the Balmer lin e s and Hgll X2798, and by Bal- mer continuum emission. The HO flux in particular is enhanced by a factor of 5 tc 10 and tnus the La/HB problem is resolved. Our understanding of the kinematics of the BLR is much less satisfactory; we still do not know if gravity, radia tion pressure, or wind pressure is the dominant force in the BLB. Although detailed kinematic models exist and var ious arguments have teen made for and against these aodels, we apparently do not have enough information at present to prove i f any given mcdel i s the c o rre c t one. For example, the functional form of individual line profiles, which is logarithmic in many cases, is not sufficient to distinguish among many aodels. C a p rio tti, Foltz, and Byard (1980) show that several cases of spherically symmetric radial motion can result in logarithmic profiles. Also, van Groningen (1983) claims that many of the observed profiles are simi lar to profiles produced by a turbulent rotating disk. Studies of the tine-delayed response of the broad-lines to changes in the nontheraal continuua flux aay eventually help us understand the BLB kineaatics (Blandford and McKee 1982; Capriotti, Foltz, and Peterson 1982). This technigue should be effective for Seyferts in which the light-travel tine across the BLB is of the sane order of magnitude as the tine scale for significant continuua changes to occur (Peterson, Crenshaw, and Meyers 1985). k large anount or high-quality data aust be accunulated however before we can determine if variability studies are the key to understand ing the BLB kineaatics. If the physical conditions and the cloud velocities botn change in a systematic fashion across the BLR, then eais sion lines such as Ho, h8, and He I *5876 should have d i f ferent profiles. Thus another possible way to probe the BLB kineaatics, as discussed by Shuder (1902,1984), is to coapare the profiles of different eaission lines. Shuder does this by dividing one profile by another, point by point, to obtain line ratios as a function of radial veloc ity. Since the "profile ratios" for individual obgects art rather noisy, he averages then by normalizing the velocity scale in each object to the full width at zero intensity of Hq in that galaxy. Shuder finds that the average HB/Ha ratio increases by a factor of 2.2 froa the core to the wings, whereas the average He I/Ha ratio increases by a 7 factor of 5 froa the core to the wings. fiecent photo ioni zation aodel results froa Kwan (1984) demonstrate that both these ratios increase with the ionizing flux incident on a BLB cloud; thus the ionizing flux apparently increases with velocity. These results place an iaportant restriction on kinenatic aodels: clouds closer to the continuua source aust have higher velocities on average. Our first objective in this dissertation is to test the results of Shuder by deteraining if the intrinsic HS/Ha ratio really does increase with radial velocity in aost Seyfert 1 galaxies. For exaaple, if the intrinsic profiles are identical, then any contanination effect which aak.es the observed profile have a weaker core or broader wings than the observed Ho profile will cause the observed H8/Ha ratio to increase with radial velocity. Therefore we con sider three effects not considered by Shuder which, if not corrected for, weaken the core or strengthen the wings of the profile relative to the Ha profile: 1. He will deaonstrate that contanination of the broad enissiou lines by stellar absorption features lowers the peak of the He profile in Seyferts with low nucle ar luminosities. 2. A shelf of eaission is often seen in the red wing of the HB profile which is not intrinsic to H8. Although Shuder removes the Fe II that contributes to the 3 shelf, an additional component (probably broad [0 111]) exists as well. 3. The velocity resolution for H$ is worse than that for Ha* The observed HB profile therefore has a weaker core than i t would have i f i t s reso lu tio n were the same as that of Ha, Even after correcting for the above effects, wc find that HB/Ha increases dramatically with radial velocity in basic agreement with Shuder's results. This suggests addi tional notivation for this study. Photoionization models have reached the stage of so p h isticatio n where they can now be conbined with different kinematic aodels to produce not only single profiles, but profile ratios as well. Thus we will provide high signal-to-noise profile ratios so that they can be compared in the future with re s u lts from these aodels in the hope that they will provide additional con straints on the kinematics of the BLB. He will also inves tigate the dependence of the profile ratios on the luminos ity of the nontheraal continuum and the width of the broad eaission lines. Finally we will use the photoionization model results of Kwan (1964) to put restrictions on the size and geometry of the BLfi. In Chapter II we give the details of the observations and data reduction procedures. In Chapters III and IV we discuss the contanination of the broad-line profiles by stellar absorption features and eaission features froa the BLB and NLBv and give the procedures we use to correct the profiles* He present the decontaainated profiles and pro file ratios in Chapter V and Appendixes A and B. Finally, we discuss our interpretation of the profile ratios in Chapter VI. chapter XI OBSMVATIONS 2x1 U£Ifi£B£mXIS£ He have observed twelve Seyfert 1 galaxies in order to study the profiles of the broad eaission lines. The obser vations were aade with the Ohio State University inage dis sector scanner (IDS) on the Perkins 1.8-a reflector of the Ohio Hesleyan and Ohio State Universities at Lowell Observ atory. The IDS is a dual-bean spectroscopic scanner that allows simultaneous observations of an object and the sky. The linear response of the IDS over a larqe dynasic range wakes i t a useful tool for studying the spectra of Sey- fe rts . The design of the IDS is very siailar to that of the Lick Observatory scanner (Bcbinson and Haapler 1972). kr optical layout and detailed description of the spectrograph is given by Byard et al. (1981). The aost iwportant compo nent of the IDS is a chain of three 40 aa S-20 extended-red electrostatic Varo iaage tubes, which increase the gain by a factor of ^ 10* and provide temporary storage of the pho ton events. The output phosphor of the last iaage tune is - 10 - 11 scanned by an ITT iaage dissector* Since each photon event is counted aore than once and the nuaber of counts per pho ton varies, the counts are not distributed according to Poisson statistics and the signal-to-noise is aore diffi cult to deteraine than with a photon counting device. An IDS spectrua is stored in a file on disk in the fora of counts per channel. Each spectrum consists of 2046 channels, and the number of counts in each channel can range froa 0 to 32,767. Usually an observation or a par ticular object will generate a number of files which can be combined at a later date. Since this is a profile study, it is important to know tne properties of the instruaental profile, whicn is approximated well by the profile of an emission line fron an iron neon hollow cathode. The instrumental profile is fit well by a Gaussian froa its peak to the point at which the eaission is ^3X of the peak; in the far wings however the profile rises above a Gaussian. The resolution can be characterized by the full width at half aaziaua (FWHH) of the instrumental profile. The resolution over a red scan (defined in the next section) as a function of channel position is shown in Figure 1. As can be seen in Figure 1, the resolution is 5-6 channels in the aiddle half of the scan and degrades rapidly at the ends of the scan. To get optimum resolution therefore, it is neccesary to choose the 12 wavelength coverage so that the broad-line profiles of interest are not near the extreae ends of the scan. Resolution (Channels) 10 4 7 6 8 9 9 00 90 0 0 0 2 1900 1000 0 0 9 0 eouin s fnto o canl position. channel of function a as Resolution oiin (Channels) Position 13 14 1 * 1 fifiSlfimiSliLfiiiHIUS All of the IDS scans were taken with a 600 lines «r> grat ing blazed at 5500 A in first order. Circular apertures of projected diaaeter 5** or 7" were used to reduce the effects of ataospheric refraction and to get an estieate oi abso lute fluxes. The resolution in the Biddle of the scans for th is grating is ^7 k with the 5" apertures and ^9 k with the 7" apertures. A blue scan centered near 5200 A and a red scan centered near 6800 A were obtained to provide cov erage of each Seyfert from 4000 A to 8000 A and to insure that the profiles of 8a and H8 were rear the center of the scan. The region of overlap for the blue and red scans spans 600 - 800 A and contains the He I A5876 profile. The integration tine per individual scan was United to 300 - 600 s to winimize the effects of variable sky bright ness, inage tube afterglow, and changes in the extinction. The total integration tine in the blue or red required to obtain high signal-to^noise in the continuun of each Sey fert ranged fron 1 to 3 hours. Since the broad emission lines of aost Seyfert 1 galaxies are variable over times- cales of Bonths or years (Peterson et al. 1982, 1984, 1985) the blue and red observations were aade within days of each other. The only criteria for inclusion of a Seyfert 1 galaxy in this study are that (1) it is bright enough to yield high 15 signal-to-noise data and (2) its Fe II emission is not so strong that it severely ccntasinates the other broad-line profiles. A journal of our observations is presented in Table 1. 16 Table 1 JgnERSi-ai.fiiigsrgflilfifls Seyfert UT Spectral Ex posure ipert Galaxy Date fiange Tiae Diaae NGC 3227 1983 Apr 9 3800-6200 4800s 5" 1983 Apr 11 5400-7800 4800s 5" NGC 351o 1983 Apr 10 3800-6200 10800s 7" 1983 Apr 11 5400-7800 24C0s 7 » . NGC 4051 1984 Feb 6 4000-6400 6600s 7" 1984 Feb 7 5600-8200 6000s 7" NGC 4593 1983 Apr 9 3800-6200 9600s 7" 1983 Apr 14 5400-7800 9600s 7" NGC 5548 1984 Feb 4 4000-6400 6000s 7" 1984 Feb 5 5700-8100 6000s 7" NGC 7469 1983 Aug 12 4000-6400 4800s 5" 1983 Aug 13 5600-8000 6000s 5" Hrk 79 1984 Jan 2 4100-6500 10800s 7" 1984 Jan 5 5800-8200 b600s 7" Hrk 110 1984 Jan 2,4 4100-6500 12600s 7" 1984 Jan 5,7 5800-8200 6000s 7" Hrk 335 1983 Sept 11 4000-6400 7200s 5" 1983 Aug 31 5600-8000 6000s 5" Nrk 509 1983 Sept 14 3900-6500 6000s 7** 1983 Sept 16 5700-8300 4800s 7" Hrk 590 1983 Oct 13 4000-6400 8400s 7” 1983 Oct 12 5600-8400 10200s 7" 3C 120 1983 Dec 5 3900-6300 9000s 7" 1983 Dec 7 5700-8100 8400s 7" 2*2 £*Xi_SI2i2£XI£i The reduction of our ZDS data involves converting the observed counts as a function of channel to flux units (ergs s~' ca_* Hz-*) as a function of wavelength (A). He reduced all of oor data on the PDP 11/34 conputer of the Department of Astronoay at Ohio State University using the Interactive Reduction System discussed by Jenkner (1980). The following procedures are performed for each 3OO-6O0S object scan from a given aperture: 1. Subtract the night-sky scan (taken immediately before or after the object was observed) froa the object scan. To a good approximation, the iaage tube after* glow from the object is also reaoved by this subtrac tio n . 2. Divide by the exposure tine to get the count rate per channel for the object. 3. Divide by a flat field scan (froa a guartz-halogen lamp) to renove channel-to-channel sensitivity varia tions of the iaage tubes. 4. Apply a dispersion-curve fit (deterained froa a scan of an iron neon hollow cathode) to the scan to convert channel position to wavelength. 5. correct for atmospheric extinction using the aean e x tin ctio n curve deterained by Tug# Hhite, and Lock wood (1977) for Lovell Observatory. 1 0 6. Remove the telluric A and B absorption bands in the red scan created by the atnosphere by using the proce dure described by Cota, Wagner, and Mewsoa (1985). The above steps are repeated for each object scan. All of the scans for a particular aperture are then averaged and the following steps performed: 1. Shift the zero-point of the disperson curve to correct for any s n a il {always < 5 A) s h if t of the object scars relative to the iron neon scan in the disperson direc tion. The zero point shifts are deterained by compar ing the laboratory wavelengths of the atmospheric eaission lines [0 1] 1X5577.3,6300.3 to their observed wave lengths. 2. Resauple the data to linearize the wavelength scale. 3. Multiply by a calibration array to convert counts to flux units. The calibration array is deterained by observing standard stars, perforaing the above reduc tion procedures on thea, and comparing the resulting spectra to the fluxes deterained by Stone (1977) for these stars. The east and west aperture data are reduced separately and are independent observations of the sane object; they are compared and averaged to obtain the final reduced spectrun. In addition to perforning the reduction just given, wc also modify the reduced spectra in several ways. First we scale the red spectrue by a aultiplicative constant to the sase relative flux as the blue spectrua by coaparing the region of overlap. Then ve correct the spectra for redden ing in our galaxy with y values froa the Baps of Bur- stein and Heiles (1982). Finally ve transfora the spectra to the rest fraae of the galaxy by using the redshift deterained froa the upper third of the [0 III] A5007 pro f i l e . Chapter H I CDITiEIUTIOE OP T » PBOFILBS BI STELLAS ABSOBFTIOV FEATERES 2 * 1 i n z m u o£uqs The observed broad-line profiles of aany Seyfert 1 galaxies show considerable structure that is not attributable to the effects of blended eaission lines (Ueedaan 1977; Osterbrock 1978b). Several ideas nave been put forth to explain the observed structure: 1. The distribution of gas in the BLR is inhoaogeneous. C a p rio tti, F oltz, and Byard (1981) have constructed discrete-cloud aodels and they find that the observed structure can be produced by the statistical cluaping of 10* - 10* clouds in radial velocity space. 2. There are excitation inhoaogeneitits in the BLE. Thc- coabination of a variable continuum source and light travel tiae effects in the BLB can produce profiles with tiae-dependent structure (Capriotti, Foltz, and Peterson 1982). 3. The broad eaission lines are produced in several dis tinct kineaatic regions. Foltz, Uilkes, and Peterson - 20 - 21 (1983) note that there are two large buaps in the 88 profile of Akn 120 which reaain stationary in radial velocity space, but vary in strength over tiae. They speculate that in addition to the region that produces aost of the H8 eaission, there nay be additional regions (possibly jets) which produce the two buups. 4. The intrinsic BLR profiles are contaainated by stellar absorption features froa the host galaxy. This possi bility has received little attention in the past, tut it nay be iaportant if the contribution of the host galaxy to the observed Seyfert 1 spectrua is signifi cant . In this chapter we will evaluate the iaportance of contaai- nation of the broad-line profiles by stellar absorption features. In particular, we are interested in deteraining whether stellar contaaination can introduce significant structure into the profiles. £*2 SIIIflOTC-UMIMS He have coaputed synthetic spectra to investigate the Ban ner in which stellar absorption features affect the broad- line profiles of Seyfert 1 galaxies. The advantage of this approach over a purely observational study is that we can begin with a saooth, syaaetric profile, free of any other blended eaission lines, and deteraine how its appearance and measured properties change with increasing stellar con tamination. To illustrate the most important changes that occur in the profiles, we will describe the results for a particular set of components: 1. A featureless power-law spectrum, F = (F « X) u ' u ' in particular, represents the nonstellar continuum of a Seyfert 1 galaxy. 2. A spectrum of Gaussian broad emission lines Hy, He I X 5876) gives the lines that are most studied in the optical. The profiles are characterized by parameters that are fairly typical of Seyfert 1 broad emission lines (Osternrock 1977, 1978c). The velocity width of each profile is 5000 km s~* (FUHtf) , the equivalent width of HB relative to the continuum is 100 A, and the fluxes of Hot, Hy, and He I X 5876 re la tive to H are 5.0, 0.4, and 0.2 respectively. 3. A high sigaal-to-noise spectrum of the dwarf ell ipti- cal galaxy H32 is taken to represent the host galaxy spectrum. There is a certain amount of evidence that Seyfert 1 host galaxies at redshifts <0.02 are pre dominately early- or intermediate-type spirals (de Vaucouleurs 1975; Simkin, Su, and Schwarz 1900; Balick and Heckman 1982). If this is the case, then the stellar light from a low-redshift Seyfert should be dominated by a spiral bulge in apertures with project 23 ed diaaeters <7H- Since dwarf ellipticals and spiral bulges have siailar late-type stellar populations (Pritchet 1977; Keel 1983)# we will assuae in this paper that the K32 spectrun is representative of the host galaxy spectrua of a low-redshift Seyfert. We will characterize the aaount of stellar contaaination present in a given spectrua in two different ways: F (X4861 ) rU«861) - Fjj(x4M1) (D the ratio of stellar continuun flax to nonstellar continuua flux a t 4 861 A, and F (i486 1 ) e(x486,) = Fs( 4861 ) +FnU4861) (2) the ratio of stellar continuua flux to total continuun flux at 4861 A. The paraneter r is a convenient one to use in this investigation, whereas e, the "stellar fraction", is the paraaeter noraally discussed by observers. We generated the synthetic spectra by scaling the H32 spectrua and adding it to the sua of the other coaponents. We chose e(i4861) = 0*8 (r=4) as an upper liait to the aaount of stellar contaaination present, because the larg e st value of e (i4861} for our observed S eyferts (presented later) was 0.7 ± 0.1. The synthetic profiles are therefore characterized by values of r(A4861) ranging froa 0 to 4. The synthetic protiles in Figure 2 shot# the structure that is introduced by stellar contaaination. He can iden tify virtually all the absorption features in the profiles with stellar lines observed in late-type stars. The broad Balaer lines are affected at line center by absorption from stellar Balaer lines in the M32 spectrua. Since these cen tral absorption features are not normally seen observed profiles, we believe that they aay be filled in by Balmer eaission froa the narrow-line regions coaaon to Seyferts. Two extremely strong absorption features are the G band (^Xi|300) in the Hy p ro file s and the Da D doublet (%Xb8S3J in the He 1 *5876 p ro file s . Host of the other s te lla r absorption features in the profiles are due to Fe I and Ca Z. Stellar contaaination affects not only the appearance of the profiles, but the aeasureaent of their gross properties (e.g.,FHIift and flux) as well. He w ill demonstrate the effects of stellar contaainaticn on the aeasured properties of HP, since it the broad eaission line aost frequently studied. He deterained continuun points for the line «tas- ureaents by selecting a wavelength region on either side of HB that was reasonably free of observed Seyfert 1 eaission lines and strong stellar absorption lines. He then defined the continuua underneath the profile by placing a straight line between the two continuua regions. He kept the saae lifl3£S_i: Rtlotlv* Flu* Rtlotiv* Fliw sp e cified by r(x4861), the ra tio of s te lla r to to r lla te s of tio ra the r(x4861), by cified e sp froa each p ro file . The s t e l l a r contaaination is is contaaination r a l l e t s subtracted The been have . file continoa ro p each stellar n o n froa and r a l t lr otnu fu a 46 A and A, 4861 graph. each at in top to flux bcttca froa continuua increases llar ste n o n Velocity (km **') (km Velocity re la tiv e f l a t un its (ergs s - » ca-* Hz~l 1 as as 1 Hz~l ca-* in » - s plotted (ergs are s its file un ro p The t a l f s. e file tiv ro la p re Synthetic function of ra d ia l velocity velocity l ia d ra of function eoiy k '1) k (km Velocity O 0 12000 12000 t f • t • f ■ 4 ■ r • 3 • r ■ 3 ■ f > 2 > r ■ 0 ■ r t ■ r 4 • ■ 2 « * £ C 5 0 s t » M -12000 -<2000 Ha (ha (ha *«iy km*-1) * m (k V*t«ci1y eoiy k» 1 two continuua regions for the neasurenents of the synthetic H6 profile at each new value of r(1486 1). Since the actual placement of the continuun in an observed Seyfert t spec- trun is very subjective, our results are only neant to be an example of the types of effects that can occur. In Figures 3 and 4, ve give the neasured values of HB properties as functions of the stellar contaaination. Fig ure 3a shovs that the neasured flux, relative to the flux of the broad enission line at zero contanination, decreases linearly with r(14861) , as night be expected fron inspec tion of Figure 2. However, th is e f f e c t introduces an error of only *7% at the largest value of r (14861). Figure 3b shows that the neasured velocity width (FWHf!) of H6 increases with the stellar contaaination, because the peak of the profile is lowered by stellar H0 absorption. Figure 3c shows a very interesting result: the wavelength of the peak of H6 shifts to the blue with increasing stellar con taaination, due to the presence of Fe I 14888 absorption slightly redward of stellar HP absorption. Even at snail values of r (14861) this effect is iaportant, and could bias studies of the differences in redshift between the broad lines and narrow lines of a Seyfert 1 galaxy. Again, the presence of a narrow H0 conponent would diaioish this effect, since it would at least partially fill in the cen tral ausorption feature due to stellar HP. 27 * ( k 4 H ll 0 0 0.9 0 6 I 00 — 0 * 9 b- 090 0.0 1.0 2.0 3.0 4.0 rd .4 1 1 !) • tX4MI> 0.0 0.9 0 6 6 0 0 0 & 1 9900 £ 9 0 0 0 - 0.0 1.0 2.0 9 0 4 0 f U 4 M I ) «(X4M II 0 0 0.9 06 00 10 20 9 0 4 0 f I X 4 B 6 I ) £ i3 J I £ e - i Measured properties of the sy n th etic HP pro files. The aaount of stellar contaaination is specified hy either r(A4861) or e(X4861). (a) The ratio of neasured flux to flux at zero con- taaination. (b) The neasured velocity width (FWHH) in ha s - 1. (c) The neasured wavelength of the peak of H$ eaission in ft. 28 « (X486I) 0.0 0.5 0.8 0.2 o V CL 00 . 0 .1 X II 0.0 0.0 1.0 2.0 3.0 4.0 r (X4861) < { X486I) a 8 0.0 CD < - 0.1 0.0 1.0 2.0 3.0 4.0 r (X 4 8 6 I) figure it: Asynaetry of the synthetic He profiles. (a) Asynaetry neasured relative to H6 line peak, (b) Asyaaetry neasured relative to A1I861. 29 Figure b shows how s t e l l a r contaaination a ffe c ts >eas* □reaents of the asyaaetry of Hp, which we define as Fr * Fb *u0) - rVrr <3> r b where F is the flux of H$ redward of A , and F, i s the r o b flux of H0 hlueward of aq. The profile at zero stellar contanination is syaaetric by definition, and is therefore ch aracterized by the value A(Ab861) - 0. As shown in Fig ure b, if the asyaaetry of the contaminated profile is determined relative to the original line center ( \ Q = <1861 A), the HB profile is aeasured to be asymmetric to the blue, because there is aore stellar absorption in the red half of the profile. If the asyaaetry is deterained rela tive to the line peak, the profile is aeasured to be rather strongly asymmetric to the red, because the line peak is blueward of <1861 A. He have computed synthetic spectra using different val ues for the Gaussian profile paramters (FHtifl, equivalent width of HB relative to the nonstellar continuum). He have also generated synthetic spectra with logarthmic profiles, since aany observed profiles seen to have logarithmic shapes (Bluaenthal and Mathews 1975; Capriotti, Foltz, and Byard 1960). As expected, our results our qualitatively the same as those just described, although the magnitude of 30 the effects on measured line properties changes. Hith a better understanding of the effects that stellar contamina tion might have on a broad-line profile, ve can now deter mine whether these effects are important in observed Sey fert 1 spectra. 3*3 £22I44£-m£X£OiS In order to evauluate the importance of the contamination effects discussed in the previous section, we must know the fraction of light in an observed spectrum that is stellar. This can be done by determining how much a normal galaxy spectrum must be diluted by a nonstellar continuua to give the observed equivalent widths of the stellar absorption features* Xn Figure 5, the strongest stellar features in the spectrum of H32 are also visible in the spectra of the Seyfert 1 galaxies NGC 4593, NGC 3516, and Hrk 590. With the exception or hg I b (^\5176) and Na D (^x5893), the featu res labeled are due mostly to Fe 1 and Ca 1 (Fay, Stein, and Barren 1974; P ritchet and van den Bergh 1977). He used these features to determine the stellar fraction in an observed Seyfert spectrua because the wavelength region from 5100 A to 6000 A contains strong stellar absorption features and no broad Balmer emission. Be did not use the Ha D feature, however, because in some of the spectra it could be partially due to interstellar absorption in the host galaxy. figure 5i Portions of the spectra of f!32 and three Seyfert Seyfert three f!32 and of spectra the of Portions 5i figure Relative F 0 0 0 6 0 0 5 5 0 0 0 5 spectra. t r e f sd o eemn eX81 i u osre Sey observed ti n e our id in res featu e(X4861) determine absorption The to used AU861. t a flux i wt te xeto o H D S9, were AS893, D Ha of exception the with , d fie glxe. h seta r sae t te same tue to scaled are spectra The galaxies. 1 5176 5268 5329 5401 X(A) 5592 5707 5782 5893 NGC4593 MRK 590 NGC3516 M 32 31 He did not Measure the equivalent widths of the absorp- tion features directly, since the presence of nuaerous emission lin e s makes the placement of the continuum extremely difficult. Instead, we scaled the 032 spectrua by a small constant and subtracted it from the Seyfert spectrum, and repeated this procedure until the absorption features disappeared. The values of e There are several possible sources of error in e(AU86 1). If the H32 spectrua and the Seyfert spectrum are not aligned in wavelength space or if they are at different resolution, subtraction of the scaled n32 spectrum will result in unusual residual features at the wavelengths of the absorption features, in all cases, however, the errors in wavelength calibrations (determined from dispersion curve fits to comparison lines) are less than 1 A, and the differences in resolution are less than 1 A (FWHM) in the middle of the spectra. The effects of these small errors are further minimized by considering the behavior of the entire absorption line when the scaled 932 spectrua is sub tracted, and not just the behavior at line center. 33 Table 2 Seyfert Aperture e {A 540 0) Galaxy 2 D im e te r t (A4861) 10" apertures NGC 3227 0. 004 5" 0.3 0.59 NGC 351b 0.009 7" 0.7 NGC 4051 0.002 7” 0.4 0.56 NGC 45 93 0.008 7H 0.6 NGC 5548 0.017 7" 0.3 0.33 NGC 74b9 0.016 5" 0.4 0.48, 0.36b firk 79 0.022 7" 0.1 Hrk 110 0.035 7" <0. 1 Hrk 335 0.026 5" <0. 1 0.17 Hrk 509 0.034 7" <0.1 0. 15 Hrk 590 0.026 7" 0.7 3C 120 0. 033 7" <0.1 a) Balkan and Filippenko 1963. b) Besults for two different dates. The dominant source of error is our inability to recog nize the exact value of e(A4861) at which the absorption features disappear, since the absorption features are superposed on a complicated set of eaission lines and blends (especially Fe II). Since a dip between two eais sion lines can be aistakenly identified as an absorption feature, we coapared the residual spectra with those of high-luainosity Seyferts,especially those with strong Fe II (Phillips 1976, 1978), when determining e(X4861). He esti mate that the uncertainty in e(X4861) due to this aeasure- ■ent error is ±0.1. Errors in e(X486 1) nay also be introduced if the host galaxy spectrua is substantially different froa the spec trum of H32, which has an integrated sp e c tra l type of G3 (Huaason, Hay all, and Sandage 1956). In obvious extreme example is the host galaxy of 3C 48, which has an integrat ed spectral type of about A7, and is exceedingly luminous (Boroson and Oke 1982). flalkan and Filippenko (1983) and Goodrich and Osterbrock (1983) deaonstrate that significant errors in the determination of the stellar fraction of a Seyfert spectrua can result froa an incorrect choice of a normal galaxy. However, we believe that to f i r s t order the H32 spectrua is a good choice for the normal galaxy spec- trua, since all the strong stellar features can be aade to vanish froa the Seyfert spectra. Hore accurate deteraina- 35 tion of the host galaxy spectrua is clearly the next level of sophistication and should be atteapted in the future. It should also be noted that inaccurate subtraction of the host galaxy spectrua aay result if the velocity disper sion of the host galaxy differs substantilly froa the velocity dispersion of H32. However, differences in veloc ity dispersions, if less than 300 ka s- 1, lead to errors smaller than those discussed for resolution differences. The contribution of the stellar flux to the total flux in a given aperture has been aeasured for a large number of Seyfert 2 galax ies (Koski 1978; Shuder 198 1) and Seyfert 1.8 and 1.9 galaxies (Csterbrocx 1981; Goodrich and Oster- brock 1983). Fewer aeasureaents have been made for Seyfert 1 galaxies, because of the difficulties introduced by the presence of strong broad emission lines. The stellar frac tio n s aeasured by osterbrcck (1978c) froa the Ca I I K absorption line cannot be compared to ours, because the host galaxy contributes auch less to the continuum flux at the wavelength of Ca I I K (3934 A) than i t does at ,4 66 1 A. Recently, Halkan and Filippenko (1983) determined stel lar and nonstellar fluxes of nine Seyfert 1 galaxies from high resolution (1.6 A) spectra and direct iaages. The stellar fractions that ve calculated froa these fluxes for each of the six galaxies that we have in coaaon are given in the last coluan of Table 2. Considering the variable 3b nature of these objects and the fact that our observations were aade one to three years later, our values are remara- bly consistent with theirs. In fact, our value is approxi- ■ately equal to or slightly saaller than theirs for each galaxy, which is to be expected since their measureaerts are at a slightly longer wavelength (5400 A) and are quoted for 10" apertures. He also agree with Halkan and Filippen- ko that aost of the Na D absorption in the spectrua of NGC 3227 is due to interstellar gas associated with taat gal axy, since the absorption is still extreaely strong in our spectrua after it has been corrected for stellar contamina tion. 3** fiM2TAl-S£_THl_5$51Ui-CQmfi£l*XISI Can subtraction of the scaled H32 spectrua successfully reaove stellar absorption features froa the observed broad- line profiles? He can answer this question oy considering spectra of the Seyferts with the greatest aaount or stellar contaaination. Figures 6-8 give the observed and corrected spectra of these Seyferts. A scan of Hrk 335, a Seyfert with no detectable stellar absorption features, is shown in Figure 9 for comparison purposes. It is interesting to note that the corrected spectra are very similar in arpear- ance to the spectra of Hrk 335 and other high-luminosity Seyfert 1 galaxies; in particular they all have nearly flat 37 continua. This gifas as additional confidence that our values of c (A4861) ace approximately correct. Although the stru ctu re in the (A4861J and Hr (A4340) profiles does not disappear completely, the corrected pro files are certainly smoother and more symmetric than the observed profiles. The He 1 A5876 profiles, which are affected by strong molecular TiO as well as Na D absorp tion, are not corrected well enough to be very useful. W-: could not detect any stellar absorption features in our Ho profiles, which is to be expected since Ha is by far the strongest emission line in the optical. qt 6 IS cn f G 31: bevd(pe) n cor- and (upper) observed MGC 3516: of scan IDS 6: jqvt? F U- Relative * rce fr cnaiain l r. The er). v (lo r a contamination l l e t s for rected ~ 0 pcr ae lte i r atve lx t (ergs its n u flux e tiv la re in plotted are spectra ~ c-* z1) s fnto o aeegh (A), of wavelength function a as ) Hz-1 * cm- s~1 0 0 0 6 0 0 0 5 0 0 0 4 NGC 3516 X(2> 38 At^_: D sa o MC 53 osre (pe) n cor o c and (upper) observed MGC 4593: of scan IDS ZAgtt^g_7: Relative F O 0 0 0 4 ect or r a l l e t s r fo d te c re pecta ae oted a i Fiue 6. igure F in as d tte lo p are tra c e sp NGC4593 0 0 0 5 X( A) contamination over)- The - ) r e v lo ( 0 0 0 6 39 i : D sa o Nk 9: bevd upr ad r o c and (upper) observed 590: Nek of scan IDS 8: Fi Relative 0 ect or ani i over). The . ) r e v lo ( n tio a in n ta n o c r a l l e t s r fo d te c re pecta ae oted as i Fiue 6. igure F in s a d tte lo p are tra c e sp MRK 590 5000 X(£> 60004000 40 gr : D sa o Mk 3 ( er . o a e f r a l l e t s Ho ). d e rv se b (o 335 Mrk of scan IDS 9: igure F Relative t t i i i i l I L I I l I i i i ■ i t i i 00 00 6000 5000 4000 ures wr det ed i can. h setu spectrum The . 6. n a sc ure ig s F i h t in in as d d te c tte te lo e p d were s i s e r tu 1 ------MRK 335 1 ------1 ------1 ------1 ------X 3*5 MSCflSSIQi Osterbrock and Shuder (19 82) have shown that the broad emission line profiles of Seyfert 1 galaxies should be cor* rected for contamination by other broad and narrow emission lines. He have demonstrated that soie profiles must also be corrected for stellar absorption contamination, which introduces structure into the observed profiles. He find that we can remove most of the small-scale features in our H 6 aad Ur profiles by subtracting a suitably scaled H32 spectrum . The effects of stellar contamination can be reduced by using apertures that are smaller than the ones we have used. Smaller apertures will reduce the structure in the broad-line profiles, but they may also introduce errors in line measurements caused by atmospheric seeing (Peterson and Collins 1983), guiding errors, and atmospheric refrac tion. Also, reducing the projected aperture area oy a cer tain factor will not reduce the stellar flux by the same fraction, because the aperture is still centered on the briqhtest part of tae host galaxy. The s t e l l a r contam ination of a S e y fe rt 1 spectrum would increase, of course, if the Seyfert were placed at a higher redshift, because more of the host galaxy would be included in the aperture. However, the opposite effect is seen in Table 2: there is a general trend toward smaller stellar 43 fractions at large redshifts. Hie reason for tJiis is that we observed Seyferts that are apparently very bright, and therefore have not included low-luainosity Seyferts at high redshif ts. Finally, we note that stellar contawination can aifect the results of spectral variability studies, because the changing nonstellac continuuw results in different degrees of stellar contamination at different tines. If the observed continuuw is assumed to entirely nonstellar, vuec in reality there is a large stellar contribution, then fractional changes in the ncnstellar continuum are serious ly underestimated. Fractional changes in the HB flux are only slightly in error, however, because the effects of stellar contamination on the measurement of the HB flux arc small. Another interesting effect is that structure in an HB profile due to stellar features is enhanced (relative to the rest of the profile) when the HB flux is small and diminished when the HB flux is large. An observer unaware of this effect would assume that the change in structure that accompanies a change in flux represents a real change in the emission line profile. C hapter IT COiTAHIiATIOi OF TBE PHOFILES BI EfllSSIOK FEITOBES 4 * 1 III£Q£UCUOf The study of the optical spectra of Seyfert t galaxies is ■ade difficult by the fact that the broad lines of interest are often blended with other broad and narrow eaission lines. The narrow-line contamination is easily identified and can be resowed successfully from the broad-line pro files in aost cases (Ostercrock and Shuder 1982; Cohen 1983), but the broad-line contamination is more difficult to identify and remove. An extended shelf of emission seen in the red wing of many H8 profiles is probably broad-line contamination* since it is not seen in the profiles of Ha o r Hy. osterbrock and Shuder (1982) attribute the shelf to Fe II AA4924, 5018 emission* which is in the same wavel ength region as the shelf and can be identified in Seyfert 1 galaxies with relatively narrow permitted lines (Phillips 1976, 1977, 1976). De fiobertis (1985) gives a procedure to remove the shelf from the U8 profile which assumes that the shelf is entirely Fe II emission. - <14 - 45 Recent evidence indicates that eaission from a species other than Fe XI contributes to the shelf. Heyers and Peterson (198S) find that the strength of the shelf is only very weakly correlated with the strength of the Fe II blends near 4570 A and 5250 A. They also find that a strong shelf is often present even when Fe II eaission is extremely wea*. These results are confirmed by van Gronin gen (1984), who finds that the removal of Fe II AA4924, 50 18 in an amount consistent with the strength of Fe II A5169 (which is to the red of the shelf) leaves strong residual eaission in most cases. The possibility that broad [0 III] AA4959, 5007 eaission exists has been discussed in the past (Osterbrock 1378b ; Shields 1978), and it seems that broad (0 111] is a good candidate for the excess shelf eaission. The identifica tion of broad £0 III] would be extremely useful for two reasons: (1) it would allow better removal of the shelf that contaminates the H profile and (2) it would provide an estimate of the densities in the broad-line region (BLR) . The evidence that there is another species contributing to the shelf of emission is based on the observed strength of the shelf; proof that this species is broad £0 III] eaission however must cone from the observed shape of the shelf. One way to investigate the shape of the shelf is to 46 assune that the intrinsic HP profile is known. Heyers and Peterson <1985) nse this aethod for the Seyfert 1 galaxy Akn 120 by folding the blue, unblended half of the Hr pro file about various wavelengths until it approxiaately Batches the red half between 30% and 70% of the strength at the peak of the line. They then subtract the folded syuue- tric profile fron the H£/shelf blend, and use the syanetric profile as a tenplate to aodel the shelf of eaission in Akn 120. They find that a conbination of broad £0 111] and Fe II eaission is a auch better natch to the observed shelf than Fe II eaission alone. Foltz, Rilkes, and Paterson (1983) reach the sane conclusion for Akn 120 by using the HY profile as a tenplate. He will use another netbod to study the shelf which is based on the aethods that Osterbrock and Shuder (1982) and i)e Robertis (1985) use to reaove Fe II. This netnod assuaes that the profiles of the contributors to the shelf are known, and that subtraction of these contributors in the correct aaount will result in a fairly saooth HP pro file. For this investigation we prefer this aethod, because we are interested in studying the asynaetries of the profiles and the differences aaoung thea and therefore cannot assune that the HB profile is syaaetric or giver by another profile. 47 5*3 £OUJiS0£_ilI£££ lie identified the continuua ranges for a Seyfert 1 galaxy by coabining its blue and red spectra and fitting all 0 1 the regions that appeared to be relatively free of eaission lines with a 2nd degree polynomial. The regions that yielded the best fit fron 4000 A to 7000 A were selected as continuua ranges. The centers of the continuua ranges cho sen are near the wavelengths 4200 A, 5100 A, 5650 A, 6200 A, and 6850 A, and the ranges are nornally 30 - 40 A v id e . lie resaapled our spectra to a bin size of half a resolu tion eleaent (^4.5 A) for the purpose of estimating the signal-to-noise of the continuua. lie determined a lower liait to the signal-to-noise frca the variance of a linear fit to an individual continuum region. The signal-to-noise may be higher t har. the value deterained because weak, uni- aentitied eaission lines aay be present in the continuua regions. The signal-to-noise of the continuua in a resae* pled spectrum aust be greater than or egual to 70 to be included in our profile analysis. Three Seyfert 1 galaxies are excluded froa further anal ysis in this investigation. The continuua of ffrk 590 does not satisfy the signal-to-noise requirements, and ve suspect that auch of the "noise*1 results froa incomplete subtraction of the host galaxy spectrum (Chapter III). 48 NGC 32 27 is excluded as a result of the difficulties encountered in removing the narrow-line contaaination: the narrow-line profiles of different species are very dissimi lar and deserve further study at higher resolution. Final ly, MGC 4051 is excluded because its broad-line profiles are very narrow relative to those of the other Seyferts considered here, and are seriously altered by convolution with the instrumental profile. 4*3 The technigue for deblending narrow eaission lines froa the broad-line profiles involves using the narrow [0 111] A5007 line as a tenplate, reproducing it at another wavelength, and scaling it by a constant in strength and width. The tenplate is scaled and subtracted froa the overall blend by trial-and-error until the subtraction leaves a saooth broad-line profile. The teaplate aust be scaled in width because the velocity resolution is better at longer wavel engths. For exaaple, the [0 111] A5007 teaplate is always scaled in width by a factor less than one to natch the narrow-lines that contaninate the broad Ha profile. This nethod is inherently iaperfect, because an observed profile is a convolution of the intrinsic profile (which is not Gaussian) and the instruaental profile (which is Gaussian to a good approxiaation). The contribution of the instru- 49 rental profile to the width of the observed profile decreases with wavelength; thus profiles at different wav elengths cannot be matched exactly by sieply scaling their widths. This procedure still works rather well however at this resolution. The following iterative procedure resoves the emission line contamination froa the broad-line profiles of inter e s t; 1. Dsing the narrow [0 III] 15007 profile as a template, remove the narrow lines blended with the broad Ha pro f i l e . 2. Using the decontaminated Ha profile as a template, remove the shelf of eaission in the red wing of the He profile (discussed in detail in the next section) • 3. Remove [O I I I ] A4959 w ith th e £0 I I I ] A5007 tem plate to recover the blue wing of the intrinsic [0 IIIJA5007 profile. The scale factor in this case is known from the ratio of the transition probabilities of the two lin e s . 4. Be peat the above procedure with the improved [0 III] X50 07 template. In practice, only two passes through the procedure are needed. It is then possible to use the improved [0 III] X5007 and broad Ha templates to remove emission- line contamination from other broad-line profiles. 50 Table 3 gives the relative narrow-line intensities derived from the deblending process. The intensifies of [O III] 14959, [O I] 16364, and [M II] 16548 can be deter mined easily from the ratio of their transition probabili ties to those of [O III] 15007, (O I] 16300, and [N II] 16 56 4 respectively (Osterbrock 1974). The major source of error in the intensity of a line is our inability to recognize the exact value that results in the "smcoth- estM broad-line profile after subtraction of the line. He estimate that the internal percentage error is ^10K for the lines of [N II] and [S II] and ^20]l for the weaker lines of [0 III] (14363) and (o I]. The narrow Balner Lines ar ? more difficult to remove frca from their broad-line count erparts, and their intensities are likely to have internal percentage errors of at least 20%. Comparison of our meas urements with those of Osterbrock (1982) and Cohen (1983) for the galaxies we have in common indicates percentage differences of less than 20 - 25% in the intensities. The narrow Ha/H8 ratio is always greater than the case B recoabination value of ^2.8, which is to be expected since recent photoionization calulations indicate that this value should be greater than 2.8 in the narrow-line region (Fer- land and osterbrock 1985). also, reddening by dust in the narrow-line region would result in an even higher Ha/HB value. This gives us confidence that our removal of the narrow components of the Balmer lines is at least approxi mately correct. The broad lines with widths (FHHH) less than 3000 km s~1 are likely to have weak narrow components that are undetectable at this resolution. 52 Table 3 Sa£E2jizlAflS-lJiifi£sitifi£_JlS-III-L219<22_z_li21 S e y f e rt 10 I I I ] He [ 0 I ] Ha tN I I ] IS I I ] [S I I Galaxy A4 363 A4861 A6300 A6 563 A6584 A6716 A673 NGC 3516 0.03 0.06 0.09 0. 23 0.30 0. 13 0.12 NGC 4593 0.07 0.06 0.09 0.50 0.48 0. 12 0.12 NGC 554 8 0.08 0. 13 0 .0 5 0. 60 0.21 0.08 0.08 NGC 7469 o . o i a 0.31 0.09 1.30 0.60 0.16 0. 16 Mrk 79 o.oia 0.08 0.06 0. 28 0.30 0. 14 0 .1 0 Mrk 110 0.09 ------0.08 ------0.09 0. 10 0.09 Mrk 335 o.o i a ------0.09 0.06 0.04 Mrk 509 0 . 02 3 ------0. 15 0.06 0 .0 7 3C 120 0 .0 2 a ------0.03 ------0.12 0.04 0.03 a) fielative uncertainties are a.5OS. 53 fUii lHfi!ll.fiUU.5JISUJ2UUS5I9IJeMJ£ To investigate the shape of the shelf of eaission in the ted wing of HB Be consider two cases. In the first case, we assuae t h a t th e s h e lf i s e n t i r e l y Fe I I AA4924, 5010 eaission. In the second case, we assune that the flux of Fe II AA4924, 5018 is known froa the Fe II blend at 5250 &, and the r e s t of th e s h e lf i s oroad (o I I I ] AA49S9, 5007 in the expected 1:3 ratio. In both cases we use the deconta- ninated Ha profile as a teaplate, since the Fe II profiles are siailar to the Balaer profiles in Seyfert 1 galaxies with relatively narrow peraitted lines (Phillips 1977,197b). He also take the ratio of Fe II A4924 to Fe II 5016 to be 1.0, because Phillips (1970) finds that the intensities of the two lines are approxiaately equal in two Seyfert 1 galaxies where the lines could be isolated. Also, theoretical considerations indicate that the ratio of these two lines is unity to within 15% in BLB clouds (van Groningen 1984). The procedure that we follow in both cas es is to subtract as aucn of the shelf as possible without allowing the residual H6 profile to drop below the adopted continuua level. He give the results for the first case in Figures 10 and 11 for Hrk 509 and Mrk 110 respectively. These figures show the original spectrua around HB, the profiles of the Fe II lines subtracted, and the residual H0 profile. The 514 sharp dip at ^(1924 a in the residual profiles of both Sey ferts is strong evidence that Fe II is not the only con tributor to the shelf. In other words, the reaoval of Fe II A5018 in an anount sufficient to elininate the red end of the shelf results in too auch subtraction of Fe II 14924. In the second case, we assune that the ratio of Fe II 14924, 5018 to the Fe II blend at 5250 A is 0.19, since Phillips (197b) finds this intensity ratio is 0.19 ± 0.03 for four Seyfert 1 galaxies with relatively narrow peraitted lines. He believe that this assuaption is valid for Seyferts with brcader peraitted lines, because the relative intensities of individual Fe II lines are apparently very siailar aaong all Seyferts (Phillips 197b}. He assune that the rest of the shelf is broad [0 III] eais sion with the sane profiles as Ha. In Figures 12 and 13 we give the original HB profile, the coabined profiles of the Fe II and broad [0 III] lines subtracted, and the residual HB profile for Hrk 509 and Hrk 110. The residual 11$ pro file in this case is saooth, suggesting that the assuaption that broad [o III] contributes to the shelf in these tec Seyferts aay indeed be valid. To discuss the results for our entire collection of Sey ferts, we aust divide then into two groups. For Seyfert 1 galaxies whose HB widths (PWHH) are less than or approxi- gr 1: r 59 eln asmn s f i ly e tir n e s i lf e sh assuming deblend 509 Hrk 10: igure F RELATIVE FLUX si 4800 ract r given. are d te c a tr ad he F II profil ee sub HB Here l t a a u h id t s re s le i the f o r p , n tru I c e I sp Fe l e a th in ig and r o The , e l i f o r p . I I Fe WAVELENGTH 5100 55 ue 1 Hk 1 dbed suang shel is entrely e tir n e s i lf e h s g ain assu debleud 110 Hrk 11: gure i F RELATIVE FLUX (B e II. Te nal spectua, h resi HB l a u id s e r the , a tru c e p s l a in g i r o The . I I re ract e given. re a d te c a tr ad h F II profil hat ee sub Mere t a th s le i f o r p I I Fe the and , e l i f o r p 4900 WAVELENGTH 005100 5000 5b F ig u re 12; Hrk 509 deblend a ssu a in g s h e lf i s Fe II and and II Fe s i lf e h s g in a ssu a deblend 509 Hrk 12; re u ig F RELATIVE FLUX G3 0 ee ract r gi iv g are d te c a tr b u s were t a h t s e l i f o r p ] I I I £0 r e s i d u a l HB p r o f i l e , and th e Fe I I and broad broad and I I Fe e th and , e l i f o r p HB l a u d i s e r broad £0 H I ]. The o r i g i n a l s p e c tr u a , the the , a u tr c e p s l a n i g i r o The en. ]. I H £0 broad 4900 WAVELENGTH 5000 5100 57 F ig u re . J3 s Mrk 110 deblend assuming s h e lf i s Fe II and and II Fe s i lf e h s assuming deblend 110 Mrk s . J3 re u ig F RELATIVE FLUX ca bod 0 III]. Te gi s r the , n tru c e sp l a in ig r o The . ] I I I [0 broad ~ 0 ] es t wr subtact r giv are d te c tra b u s were t a th s le i f o r p ] I I I £0 r e s id u a l H6 p r o f i l e , and th e Fe I I and broad broad and I I Fe e th and , e l i f o r p en. H6 l a u id s e r 4900 WAVELENGTH 5000 5100 58 59 ■ately equal to 3000 ka s~1 (Ark 110, Mrk 335, flrk 509, and 3C120), the assuaption that the shelf is entirely Fe £1 results in a residual HB profile with a sharp artificial dip at 4924'1, vhereas the assuaption that the Fe II enis- sion is known and the rest of the shelf is broad [0 III] results in a saoother profile. For Seyfert 1 galaxies whose HB widths (FUHfl) are greater than 3000 ka s~» (ngc 3516, NGC 4593, NGC 5546, NGC 7469, and flrk. 79], the two cases are virtually indistinguishable: both assuaptions result in rather snooth profiles because the lines are so broad that the shape of the shelf is rather insensitive to the exact central wavelength of the contributors. Table 4 gives a suaaary of our results for the case where broad [0 III] contributes to the shelf of eaission. It is evident froa a coaparison of the first two coluans that broad [O III] can contribute several tiaes the flux that Fe II contributes to the shelf in soae cases. He nust note however that this is not a coaplete representative sanple of Seyfert 1 galaxies, since Seyferts with extreaely strong Fe II were not observed. He can estinate the average electron nuaber density in the BLH clouds using the saae arguaent given by Heyers and Peterson (1985). In the narrow-line region, the £0 I I I ] XX4959,50Q7/HB r a t i o i s a p p ro x in a te ly 16 and the density is less than 10*•• (Koski 1978), the critical den 60 sity for collisional deexcitation of the [0 111] lines. Assuming that the [0 III] eaission is suppressed relative to a low-density value of 16 by collisional deexcitation, the density can be calculated. The [0 Iil]/He ratio, aver aged over all of the Seyferts in Table h, is 0.10 ± 0.03, ifhich corresponds to a density of •v-IO* ci*3. This value is only a rough estimate, since a proper determination of the dependance of the [O lll]/H0 ratio on the physical condi tions in the BLB lust await detailed photoionization calcu l a t i o n s . Table 4 S e y fert IMIa IflJIU L 2 .U I Galaxy Ha hs MGC 3516 0.010 0.02 8 0.102 MGC 4593 0.047 0.012 0.0J3 MGC 5548 0.008 0.02 0 0. 124 NGC 7469 0.029 0.020 0.086 Mrk 79 0.011 0.024 0.089 Mrk 110 0.007 0.024 0. 1 08 Mrk 33 5 0.015 0.028 0.1 12 Mrk 509 0.006 0.032 0.124 3C 120 0.011 0.024 0. 101 a) Fe II x*4924, *018 b) [0 III] XX4959. 5007 62 £«£ PISCflSSIQl The evidence that Fe II is not the only species that con tributes to the shelf of eaission has been largely based or the observed strength of the shelf. He have shown that the shape of the shelf is also inconsistent with the idea that Fe II eaission alone is responsible for the shelf. If we relax the assnaption that the intensities of Fe II X4924 and Fe II X5Q18 are equal, we find that we can obtain a smooth residual HB profile only if this ratio varies from t to 5 aaong the Seyferts in our collection. However, as discussed in the last secticn, there is both observational and theoretical evidence that the ratio of these two lines is always close to unity. The assuaption that broad [0 III] contributes to the shelf leads to a smooth, but not perfect HB profile. In particular, a broad residual eaission feature is often seen in the neighborhood of 5050 A after subtraction of the shelf (Figures 12 and 13). There are three possible rea sons for this excess eaission: 1. Broad [0 III] is not a significant contributor to the s h e lf . 2. The [o III] profile has such stronger wings than the Ha template. 3. There is weak broad eaission in this range froa yet another species. Van Groningen (1984) gives evidence 63 that Si II XX5041,5056 eaission nay contribute up to 20X of the shelf, based on the observed intensity of Si II lines in the ultraviolet. He prefer the last possibility given above since the cen tral Havelength of the residual eaission is near 50 50 A. Peterson et al. (1985) note that there is indeed a distinct feature present at 5050 A in spectra of Akn 120. Through out the rest of this investigation we will assune that broad [0 III] is present, and that we can reaove the (o III] and Fe II contaaination well enough to study tne intrinsic 8$ profile. Finally, we note that the density derived froa the broad £0 III ]/Hp ratio (^10* ca—*) is at the low end of those used for hLR phototionization node Is (10s - 10*° c r J) . Furtheraore, Peterson et al. (1985) suggest that the densi ty of the Hp-eaitting region in Akn 120 is 10*° - 1011 ca*3. It is likely that there is a wide range of densities in the BLR and that 10s ca~s represent soae sort of aver age. If this is the case, then the clouds eaitting the aajority of the broad £0 III] flux aay be of even lover density than that quoted. Therefore an atteapt should be aade in the future to extend the present photoionization calculations to both lower and higher densities. Chapter V VIE DECOITAHIBATED PEOPILES All PIOPILE BATIOS IJ§Qitgxj2i_£QfiiKX20i£ He will study the broad eaission line profiles of Hp, Ha, and He 1 X5876 in this chapter, since these are the only profiles suitable for further analysis, itteapts to decon- taninate the Hy and He II \A666 profiles reveal that they are too severely ccntaainated by Fe II eaission to be use ful in aost cases. The He I A5876 profile will only be considered for those Seyferts characterized by stellar fractions less than or equal to 0.1, since stellar contaai- nation severely alters this profile and cannot bo reaoved satisfactorily (Chapter III). The resolution near the center of our spectra is ^9 k (FHHH), which corresponds to a velocity resolution of 411 ha s~* at the position of Ba and 555 km s-1 at the posi tion of Hg. The narrowest permitted lines in our collec tion have widths of 1500 ka s~* (FHHH) ; convolution with the instrumental profile aay have a significant effect on their observed profiles. In order to compare the profiles of lines at different wavelengths therefore the profiles should all be corrected to the same velocity resolution. - 6 underneath a decontaminated profile using the continuum regions given in Chapter IT. we then convert the profiles to units of relative flux per unit radial velocity interval to facilitate the comparison of profiles at different wav- elengtns. Since the instrumental profile is Gaussian to a good approximation (Chapter II), a profile can be degraded to lower resolution by simply convolving it with a Gaussian profile of the appropriate width. The resolution of the Ha profile is therefore degraded to the resolution of the H 6 profile by convolution with a Gaussian of width 37 3 km s-*. He I A5876 is much closer to the end of the spectral scan than Ha or 116, so its resolution is 'Ml A instead of ^9 A (Chapter II). This corresponds to a velocity resolution of 562 km s-1, which is very close to the velocity resolution of H8, and therefore no resolution correction is needed. £*2 4iilJSI§J8E^fl5_EesZUSS In Appendix A, we give the profiles corrected for stellar contamination, emission-line contamination, and different velocity resolutions. For comparison with the profiles of Osterbrock and Shuder (1982), we also give profiles cor rected for stellar contamination and different velocity resolutions, but not for emission-line contamination. No attem pt has been made to remove He II X4686 from the H 6 66 p ro file or [ Fe VII] XXS721, 6087 and £ M I I ] X5755 from the He X X5876 profile. These contaminants are too far from line center to have an effect on the analysis that follows. The profiles are given in units of relative flux per unit velocity interval, the peak of the broad component is normalized to a value of 10, and the continuum level is at zero. Furthermore, we assume that the broad lines are at the sane redshift as the narrow £0 III] X5007 line. The measured position of the peak of the broad-line profiles is always within 200 km s ~ 1 of zero velocity, and the average displacement of both the Ha and HP peak is 90 km s - 1 to the red of zero velocity. The offset of the peak from zero velocity is always small compared to the width of the line in all cases, so choosing the position of the peak to rep resent zero velocity would not change any of our measur- nents significantly. Our profiles can be compared with those of others taken at earlier epochs for the galaxies we have in common. De Bobertis (1985) gives H8 p ro file s for MGC 3516, NGC 5548, NGC 7469, Hrk 79, flrk 335, and Hrk 509. Osterbrock and Shuder (1982) give Ha, H8, and He I X5876 p ro file s for flrk 79 and Hrk 335. A comparison of our p ro file s with those from these earlier studies indicates that no drastic changes have occured in any of the profiles. 67 Table 5 presents the intensities of the broad emission lines relative to the narrow [0 111] X5007 flux. The rela tive intensity of Hy has not been corrected for Fe II con tamination, but the other lines have been corrected for all of the contamination effects disscused in the previous chapters. The major source of uncertainty in the intensi ties is the placement of the continuum, and we estimate that the relative percentage error in a line intensity is 10 - 151. The f u ll widths at half maximum (FHHH) of Ha and HP, in units of Km s-*, are given in Table 6. The HP profile is broader than the Ha profile in every Seyfert 1 galaxy; the reason for this difference will be discussed in detail in the next section. Table 6 also gives the asymmetry of the Ho and HP profiles, which we define as (**) where HHHH(R) and HHHH(B) are the half widths at h alf maxi mum redward and blueward of line center respectively. He can estimate the errors in the asymmetries by choos ing different continuum placements and by measuring the east and west data separately. He find that the errors in the asymmetries are less than 0.1 in all cases, and that the asymmetry of Ha is the same as the asymmetry of Hp to 68 Table 5 fir^azkias-lBieasi£lss-JLa.mJ-A£flflZ-;.IxOi Seyfert hy H6 He 1 Ha Galaxy A 4340 A 4661 A5876 A 6563 NGC 35 16 1. 16 2.63 0.41 9.60 NGC 4593 2.06 2.87 0.82 7.91 HGC 5548 0.64 1.86 0.31 11.55 NGC 7469 0.85 1.32 0.21 5.67 !lrk 79 1.30 2.34 0.79 3.70 flrk 110 0.74 1. 71 0.29 7.70 Hrk 335 2. 13 3.93 0.73 15. 72 flrk 509 1.04 2.49 0.71 9.6 1 3C 120 0.54 1. 25 0.40 5.23 within the errors of Measurement. if we conservatively designate all profiles with values > 0 .1 as being asymme tric, then five Seyferts have syaaetric profiles* two Sey- ferts have profiles that are asynnetric to the blue, and two Seyferts have profiles that are asynnetric to the red. He are in agreenent with Os ter brock, and Shuder (1982) and De Hobertis (1985), who find that nost broad-line profiles are sy aaetric, and that there is roughly an egual number of blue and red asynnetries. 70 Table 6 BEg3d-j,iBe_liaiAs_and_igiBjgirlgJg Se yfert FNHH FUHH ksya As yn Galaxy (HB) (Ha) (HB) (Ha) NGC 3516 4760 3650 -0- 14 -0.23 NGC 4593 4900 3720 ♦ 0-07 - 0.01 NGC 554 8 4940 5010 - 0.01 - 0 . 03 NGC 7469 3460 2450 + 0.29 ♦ 0.24 Hrk 79 5140 4000 -0.06 -0 .0 8 nr k 110 1970 1820 ♦ 0.28 + 0. 24 Hrk 335 17 20 1490 - 0.12 - 0 . 10 Hrk 509 3020 2570 -0.09 -0 .0 5 3C 120 2070 2010 - 0.02 ♦ 0.01 71 5*3 iM*U5i5_ez_m-KgEiM_amfis An efficient nay to investigate the variation of physical conditions across the BLR is to analyze the "profile rat io s". A profile ratio is foraed by dividing one profile by another, point by point. He present the profile ratios of HB/Ha and He I X5876/HB for individual Seyfert 1 galaxies in Appendix B. The ratios are terminated at radial veloci tie s of - ttOOO and +4000 k.a s~i for the reasons given later in this section. For eight oat of a total of nine Sey ferts, the HB/Ha ratio increases by a factor of ^2 from 0 to ±4000 ka s-»; the HB/Ha ratic for HGC 5548 shoes only a aarginal increase. In other words, the HB profile has a greater FHHH than Ha, as deaonstrated in Table 6. The He I/H6 ratio given for five Seyferts, although noisier than the HB/Ha ratio, also increases with the absolute val ue of the radial velocity. Our results confira the discovery by Shuder [1982, 1934) that the HB/Ha and He I/HB ratios increase with radial velocity in aost of the Seyfert 1 galaxies he studied. As pointed out by Shuder [1962), reddening by dust in the intercloud mcdiun cannot account for the variation in the line ratios with radial velccity. In this case one would infer that the reddening decreases with radial velocity froa the HB/Ha ratio, but that the reddening increases with radial velocity fron the He I/HB ratio, which is a contra diction. 72 Although correction for contamination effects does not alter the fact that the HB/Ha and He I/HB ratios increase with radial velocity, it does allow for a lore accurate determination of the intrinsic ratios. We will demonstrate that contamination effects can significantly change the profile ratios hy considering the cases in which these effects make the largest contribution. Figures 14-16 pres ent the corrected ratios as histogram plots, and the ratios fo r which one sp e c ific correction is not made as smooth plots. In all cases, the contamination effects tend to either lower the ratio in the core or raise the ratio in the wings, as discussed in Chapter 1. Figure 14 gives the HB/Ha ratio for HGC 3516, the Sey fert with the greatest stellar fraction at 4861 A (=0.7). The stellar contamination lowers the ratio in the core hy ^15% and has little overall effect on the ratio in the wings. In Figure 15, we see the effects of the shelf of eaission in the red wing of HB on the HB/Ha ratio of Hrk 509, which has the strongest shelf relative to HB. Thi shelf of emission begins to have an effect at 2 0 0 0 ka s-», and ra ise s the ra tio by ^20% at 4000 km s-». Bedward of 4000 km s~», the effects of the shelf on the HB/Ha ratio become very strong, and the errors involved in the removal of the shelf become very large. In Figure 16 we see that dividing the Ha profile at its original resolution by the 73 s> m c a 0 2000 4000 VELOCITY li 0.2 0.3 0.4 0.5 0.6 p lo t is the r a tio th at has not been corrected corrected been not has at th tio a r the is t lo p p lo t i s the corrected r a tio and the the sBooth sBooth the the and tio a r corrected the s i t lo p for th e presence of the shelf of eaissio n in in n eaissio of shelf the H of of wing presence red the e th for 6 /Ha p ro file ra tio for Hrk 509. The h ist eg ist ram h The 509. Hrk for tio ra file ro p /Ha VELOCITY 0 0 . 4000 74 £± s 3 JSje_J 0.2 0.3 0.4 0.5 6 H/a ofl ato fr r 35 Te histogran The 335. Hrk for tio ra file ro p Hp/Ha : s te ato ta hs o be cretd for corrected been plot not saooth has the and that tio a tio r ra corrected the the is s i plot eoiy eouin differences. resolution velocity VELOCITY 0 6000 75 76 HB profile results in a HB/Ha ratio that is lower in the core by 'x.lOX compared to the corrected HB/Ha ra tio . He find that each of the three effects we hare consid ered can alter the observed fig/Ha profile ratios signifi cantly. The coibination of these effects systematically lead to a larger apparent increase in Hp/Hot from the core to the wings. Finally, we note that to accurately coopare theoretical profile ratios from models with the observed ratios here, the theoretical profiles must be convolved with an instrumental profile of appropriate width. He terminate the profile ratios at -4000 and +40 UC It re s~1 for several reasons. For the Seyferts with relatively narrow permitted lines, the profiles begin to approach the continuum between 4000 and 6000 km s- *, and the profile ratios become very noisy. For the Seyferts with relatively broad perm itted lin e s , the He £1 14686 profile* which can not be removed* co n trib u tes sig n ific a n tly to the emission in the blue wing of the HB profile at about -S000 ka s~1 (relative to HB line center)• Also* the intrinsic HB pro file begins to drop below the shelf contribution between e4000 and *6000 km . if the strength of the shelf is in error by 20%, then the HB intensity is in error by 20% at the point where at equals the shelf intensity. Redward of this point, the error in the intensity of HB increases rap idly with wavelength. Thus in order to compare the pro 77 files of different Seyferts in a consistent nanner, we »ust terninate then at *4000 and +4000 ka s-». Since the individual profile ratios are still sonewhat noisy, we developed a procedure to deteraine the range of ratio values for an individual Seyfert froa 0 to ±4000 ka s-i. He deternined the niniaua ratio value by averaging over the values at radial velocities within 200 ka s~l of zero. He deternined the saxiaus values by fitting a line to the profile ratios between ±2000 and ±4000 kn s->, and finding the values of the line at *4000 and *4000 kn s- *. Host of the profile ratios are fairly syaaetric, so we averaged the two ra tio values at -4000 and +4000 ka s~* to obtain a final aaxiaua value. Chapter VI IBTEBPBETATIOB OP THE FBOFILE B1TIOS IBM IouS-EISZIldLlimS Shuder {1982, 1984) averages his profile ratios by normal izing the velocity scale ir each object to the full width at zero intensity (FHZI) of Ha in that galaxy. In other words, he iaplicitly assumes that the radial velocity that corresponds to a certain value of HB/Ha scales with the width of the Ha profile. He will test this assumption with our high signal-to-noise data, which allow us to consider the properties of profile ratios for individual Seyferts. In Figure 17 we see an unexpected result: the HB/Ha pro f i l e ra tio s of Hrk 79 and flrk 335, plotted on the sane velocity and intensity scales, are similar, in spite of the fact that the width (FUHH) of the H8 profile in Hrk 79 is three tines greater than the width of the HB profile in Hrk 335. It we had scaled the profiles of Hrk 335 to the sane width as the profiles of Hrk 79, the HB/Ha ra tio of Hrk 335 would have appeared to rise nuch more slowly with rad ial velocity than the HB/Ha ra tio of Hrk 79. - 78 - 79 Figure 18 gives another way to demonstrate this point for all nine Seyfert 1 galaxies. Here we plot the FWHM of HB versus the range in HB/Ha values from the niniaua value a t zero kn s~1 to the average of the aaxiaua values at -4000 and +4000 ka s- *. If the radial velocity that corre sponds to a certain value of HB/Ha scales vith the width of tne p ro file s , then we would expect to see the range ir HB/Ha decrease with the the FtfiiH of HB* since we are sam pling such less of the total velocity range in the brcader profiles. However* Figure 18 denonstrates that there is no trend in the range of Hp/Ha with profile width. Thus Shu- der's aethod of normalizing the velocity scale to the Y4Z1 of Ha is an inappropriate way to coapare the profile ratios of different Seyferts. The range in HB/Ha values over this fixed velocity in te rv a l is fairly s i a i l a r among most S eyferts, and i s cot a function of the profile width. This is actually evidence that the BLR is spherically syaaetric. Consider, for exam ple, a BLR that is in the fora of a rotating disk that is initially edge-on to the observer's line of sight. If the plane of the disk is then tilted vith respect to the line- of-sight, the Ha and HB profiles decrease in width by the saae factor* as does the radial velocity that corresponds to a certain HB/Ha ratio. He observe however that the radial velocity that corresponds to a certain HB/Ha ra tio iue 7 H/J pr ie r i or w Syet galax 1 Seyfert two r fo s tio ra file ro p Hg/fJa 17; Figure 0.2 0.3 0.4 0.5 0.6 of hrk 79 and the ssooth plot i s the p rofile rofile p the s i plot ssooth the and 79 hrk of ato o ak 335. ark of tio ra e. h hitgrs ot h pr ie r i tio ra file ro p the s i t lo p ras istcg h The ies. VELOCITY 0 80 i e 8 Te HH f pote agai t h rne ir, range the st in a g a tted plo H 6 of FHHH The 18: re u Fio FWHM (H^) 0 0 0 3 2000 0 0 0 5 0 0 0 4 0.0 H 8/H a. 0.1 NGC 5548 0.2 C 120 3C R 110MRK MRK 335 R 5 9 MRK 50 G 7469 NGC H*/Ha G 3516 NGC R 79 MRK . 0.4 0.3 G 4593 9 5 4 NGC 0.5 0.6 82 is roughly constant aaong Seyferts with different profile widths. Thus for a collection of rotating disks, the disks at greater inclinations aust have HB/Ha ratios that increase such aore slowly with actual rotational velocity to natch the observations, which is of course absurd. Therefore the BIB is spherically syaaetric, since this arguaent is valid for any non-spherical geonetry that is utilized to explain the wide range in observed profile widths as an aspect effect. Arguments have been put forth in favor of spherical sym metry based on the incompatibility of profiles fron rotat ing disk, nodels with observed profiles (Shields 1978; C a p rio tti, F oltz, and Byard 1980; Hathews 1982a). However, recent models incorporating turbulent rotating disks have been aore successful in matching observed profiles (van Groningen 1983). There are several arguments for spherical symmetry th at do not depend on the profile shapes: 1. There is no evidence for a correlation between the axial ratio of the disk of the Seyfert host galaxy and the width of the broad lines (Keel 1980; Siakin, Su, and Schwarz I960). However, there are reasons to believe that the inclination of the BLB disk does not have to be the sane as the inclination of the host galaxy disk (Tohline and Osterbrock 1982). 83 2. The distribution of the FtfZI of Seyfert broad lines is inconsistent with the expected distribution due to a collection o£ rotating disks (Osterbrock 1977). In particular, there are no Seyfert 1 galaxies with very narrow permitted lines* This arguaent is not valid if the rotating disks are very turbulent (Osterbrock 1978c, 1979). 3. X-ray observations of the active galactic nuclei Cen A (dushotzky et al. 1978) and NGC 4151 (Holt et al. 1980) show that the ratio of the 7.1 keV Fe K absorp tion edge to the associated 6.4 keV flourescent lire is consistent with a spherical distribution of BLR Material. If the BLB were a face-on disk, the eais sion feature would be such nore prominent and if the BLR were an edge-on disk the absorption feature would be nuch More prominent. Although there are objections to the first two argunents given, the third argument and our own results support the view that the BLR is indeed spherically syaaetric. 5*2 l£imgE.ig_iiiaiig2lzl Shuder (1982, 1984) claias that the Balaer line profiles are aore unifora in higher luainosity Seyfert 1 galaxies and QSOs. He arrives at this result by separating the Sey fert 1 galaxies and QSOs into three luainosity classes and au averaging the profile ratios in each class. He finds that the average H$/Ha profile ratio for the low luainosity objects increases by a greater factor froa the core to the wings than the average ratio for the internediate luainosi ty objects, which increases by a greater factor than the average ratio for the high luainosity objects. He will test this result by deteraining the nonihtrmal continuum luainosity at 4861 A froa the observed flux, the stellar fraction, and the redshift of each galaxy. In deteraining the luainosities we assuae that Hq = 75 ha s-» Hpc~1 and that qQ = 1. He can estimate the errors ia the relative luminosities by comparing the fluxes of the red and blue scans of a given Seyfert, which were taken on dif ferent nights. On average, the flux of the red scan dif fers from that of the blue scan by 20%, and the maximum difference is 40%. These e rro rs are not very sig n ific a n t for this analysis, since the luainosity ranges by a factor of ^100 in our co llec tio n of Seyferts. He present in Figure 19 a plot of the log of the nonth- ermal continuum luminosity at 4861 A versus the range in the 116/Ha ratio. Although there is some variation in the range of HB/Ha from one Seyfert to the next, there is no correlation with luainosity. Thus we do not agree with Shuder*s claim that the Balner profiles become more uniform vith increasing luminosity. 85 He do not know the reason for the discrepancy between our re s u lts and the re s u lts of Shuder. His range in lumi nosities coincides with ours, except that his objects extend to higher luminosities by about an order of magni tude. However, there are problems with Shuder*s analysis of the data: 1. He normalizes the velocity scale for the profile rat ios in each object tu tne full width at zero intensity of the Ha profile in that object. He have shown that th is method i s inappropriate. 2. He includes the ratios in the far wings in his analy sis. He have shown that there are large systematic errors in the ratios at radial velocities greater than 4000 - 6000 km s~*. Thus we believe th a t Shuder*s claim th at the Balmer pro files are more uniform in higher luainosity objects is suspect, even though his sample contains a larger range in luminosity than ours. There have been claims that the broad-line profiles of different eaission lines in high-redshift, high-luminosity QSCs are the same tc within the observational uncertainties (Baldwin and Netzer 1978; fiichstoue, Hatnatunga, and Schaeffer 1980). Studies of QSOs at higher resolution and signal-to-noise, however, conclude that there are substan tial profile differences among different lines. In partic- Figure 19; The nonthernal lu a in o sity p lo tte d aqainsx. the the aqainsx. d tte lo p sity o in a lu nonthernal The 19; Figure log [ L n(\4 8 6 I) ] 27.5 28.5 29.0 28.0 0.0 ag i d@/Ha in range 0.1 G 5548 4 5 5 NGC 0.2 C 120 3C R 110 MRK G 7469 6 4 7 NGC R 335 MRK */Ha / H* 3516 NGC MRK 509 R 79 MRK 0.3 G 4593 NGC 0.4 0.5 0.6 6 8 87 ular, the La profile noraally has stronger wings and a sharper peak than the C IV A1549 profile (Kilkes and Cars well 1982; Wilkes 1984). In order to proceed farther with our analysis, we must use the results froa recent photoionization aodels. The aost sophisticated treataent of the variation in line rat ios with the physical conditions in the BIB is by Kvar* (1984). Kwan finds that for a given nontheraal con tin u um energy distribution, the HB/Ha ratio is principally a func tion o f the ionizing flux incident on a BLR cloud, although it also depends weakly on the nuaber density and column density. The He I/H8 ratio is a strong function of the ionizing flux and is alaost independent of the other param eters. In all of our Seyferts, with the possinle exception of NGC 5548, these ratios increase with radial velocity; thus the BLR clouds closer to the nontheraal continuum source aust have higher velocities on average. The range in HB/Ha froa 0 to ±4 000 ka s~*, averaqed over all of our Seyferts, is 0.22 to 0.42. If this ratio con tinues to increase with radial velocity past ±4000 ka s~*, then the aaxiaua value of HB/Ha in the BLR aay be aucii greater than 0.42, since soae of our Ha p ro file s have wings th at extend out to a t le a s t ±10,000 ka s~*. Kwan's c alcu lated HB/Ha ratios as a function of the ionizing flux range froa 0.1 to 0.37. It is apparent that these calculations 89 aust be extended to higher fluxes before coabined photoion ization and kineaatic aodels can be used to Batch the observed HP/Ha profile ratios. If we assuae that the observed increase in HP/Ha with radial velocity is entirely due to an increase in the ion izing flux, ve can investigate the size of BLR. be aust keep in aind, however, that our results are photoionization aodel dependent. In the past, it was realized froa pho toionization aodels that the ionization paraaeter and the electron density are fairly siailar anong the broad line regions of kGN. The ionizing flux, vhich is proportional to the product of the ionization paraaeter and the density, is therefore siailar aaong the broad line regions, and the size of the BLR aust scale with the square root of the luainosity. He can elaborate on this idea by again consid ering figure 19, where we see that the range in H8/Ha, and therefore the range in ionizing fluxes, is siailar aaong aost Seyfert 1 galaxies. Thus, to a rough approxiaation, the radial extent of that part of the BLR characterized by velocities froa 0 to ±UOOO ka s~1 scales with the square root of the nontheraal continuua luainosity. The fluxes that correspond to our average aininua and aaxiaua values of H B/tta can be estiaated froa Kuan's graph. The ratio of the aaxiaua to ainiaua flux is * 25, which aeans that the outer radius of the BLR is at least^ 5 tiaes 89 greater than the inner radius. This estimate is an extreae lover linit to the relative extent of the BLR, because the ainiaua value of HB/Ha at the radial velocity of 0 ha s“ 1 receives contributions fron clouds traveling at space velocities perpendicular to our line-of sight. Also, as discussed previously, the HB/Ha ratio for clouds traveling at velocities greater than 4000 ha s~* say be auch higher than the aaxiaua ue guote. In any case, it is clear that the BLR is not a thin spherical shell. If HB/Ha is a good indicator of the ionizing flux on a BLR cloud, as the aodels of Kvan (1984) suggest, then the size of the BLR determined froa HB/Ha should at least be consistent with the size deterained froa aore direct aetn- ods. Cherepashchux and Lyuti (1973) find froa variability studies that the BLR of NGC 3516 has a size of 15-30 light days. He can calculate a aodel dependent lover liait to the size of the BLR in NGC 3516 froa the HB/tia aininun val ue of 0.23. Froa the graph of Kvan (1984), ve obtain the flux corresponding to this value. He then use the noniber- aal continuua energy distribution assuaed for Kuan’s aodels (Kvan and Krolik 1981) and our lu ainosity at 4861 A to deteraine a lover liait to the size of the BLR in NGC 3516. He find that the size of the BLR in NGC 3516 is <0.025 pc (V30 light days), which is consistent with the aeasuraents of Cherepashchuk and Lyutyi (1973). 90 $*3 sflflfim He obtained high signal-to-noise spectra of Seyfert 1 gal* axies to test Shuder's claia that HB/Ha and He I A5876/H8 increase with radial velocity (Shuder 1982, 1964). He con sidered three effects that sight coctasinate the profiles in such a way as to weaken the core or strengthen the sings of the HB profile relative to the Ha profile: 1. He find that stellar absorption features fron the host galaxy of a low-luminosity Seyfert can significantly lower the peat of the Hb profile. ilso, stellar con tamination can introduce considerable structure and asymmetries into the profiles of HB, Hy, and He I A5876. 2. Emission froa lines other than HB strengthen the red wing of the observed HB profile considerably. He give evidence that Pe 11 is not the only species that con tributes to the shelf of emission, and that at least one more species (probably broad (O III]) contributes as well. 3. He find that Ha is at a better velocity resolution than HB in our spectra, which leads to a wearer core for the observed UB profile. Each of the effects that we considered can alter the profile ratios in the core or the wings by 10*20% at most. Correction for these effects cannot change the observed 91 increase in Hs/Ha end He I/ttp with radial velocity. The change in these line ratios with radial velocity indicates that both the physical conditions and the cloud velocity change in a systematic fashion across the BLR. Froa the broad [0 IIiyHB ratio, we determine that the electron density in the BLR is of order 10* cm- *. However, there is prooably a wide range of densities in the BLR and this value probably represents some scrt of average. also find that the HB/Ha profile ratio extends to higher values than th o se c a lc u la te d by Kuan (1984) a s a fu n ctio n of the ionizing flux incident on a BLR cloud. Therefore, existing photionization calculations should be extended to a greater range of densities and ionizing fluxes. From a comparison of profile ratios for individual Sey fert 1 galaxies, we can put restrictions on BLR kinematic aodels. The BLR clouds are in a spherical distribution around the nontheraal continuum source, since there is no observed correlation of the range in HB/Ha with profile width. Zf HB/Ha is primarily a function of the ionizing flux, as the aodels of Kvan (1984) suggest, then clouds closer to the continuum source must have higher velocities on average. Finally# the large range in HB/Ha and He l/HB for a given Seyfert 1 galaxy suggests that the BLR is not a thin spherical shell. 92 Obviously, sore than one type of kinenatic model can satisfy the above requirements. For example, in the recent models of Mathews (1982b), the clouds undergo radiatively driven outflow. Clouds that fora close to the continuum source are accelerated to high velocities and travel short distances, whereas clouds that form farther from the con tinuum source are accelerated to shorter velocities and travel greater distances. However, in the models of Car roll and Kwan (1985), the clouds undergo gravitational motion in orbits that are essentially parabolic. In these models, the velocity also decreases with increasing dis tance from the nontheraal ccntinuua source. Further work on profile ratios of Seyfert 1 galaxies and QSOs is definitely needed. On the observational side, pro files of Ha# HB# and He 1 A5876 should be ontained for a wider range of luminosities and profile widths to further investigate the results we have given. Also, hign-guality profiles of emission lines in the ultraviolet are needed, since it is apparent from recent photoionization aodels (Kwan 1984) that ultraviolet line ratios are needed to evaluate such physical conditions in the B1H as number den sity and column density. Finally, in order to get more information out or tne observed profile ratios, photoionization models should be combined with various kinematic models. From the photoion 93 ization aodels It is possible to determine the dependence of line eaissivities on the physical conditions. This inforaation can be used in a given kineaatic aodel to determine profiles for different lines. The calculated profiles and profile ratios froa various kineaatic aodels can then be coapared to those observed to deteraine if anv given kineaatic aodel is favored. Ipptidiz A. BKOAD-LIVE PBOFILES Figure 20: Broad emission line profiles of Seyfert 1 gal- “ “ ” axies. The lover plot has been corrected for all contamination effects. The upper plot has been corrected lor stellar contavination acd velocity resolution differences# but net for emission-line contamination. - 9U O' RELATIVE RELATIVE FLUX KJ B S B a a M VELOCITY RELATIVE RELATIVE FLUX KJ B B 00 B □0 K) < n —i -< o 20: (continued) RELATIVE RELATIVE FLUX NJ 9 00 * s s 00 VELOCITY U1 RELATIVE RELATIVE FLUX IV> oo s < —i o n figure 20: (continued) 0 1 00 X0 OI OI X0 RELATIVE RELATIVE FLUX CD VELOCITY tn oo RELATIVE RELATIVE FLUX CD IS < —I -< m n o n £l9tt£S-.2fi : (continued) to CD RELATIVE RELATIVE FLUX s B B B S B B m VELOCITY NGC 7 4 6 9 10 0 T 6 RELATIVE RELATIVE FLUX Figure 20: (continued) VO RELATIVE RELATIVE FLUX CD 00 S B S VELOCITY RELATIVE RELATIVE FLUX to s B to ts b —t —t -< < m r“ o n Figure 20: {continued) 100 m RELATIVE RELATIVE FLUX a a a a VELOCITY 00 O RELATIVE RELATIVE FLUX a a a a a a a a a a a K) < o m n -I -< Figure 20: (cuutinued) 10 1 RELATIVE RELATIVE FLUX B to B “ B 9 B tsJ s a> s VELOCITY RELATIVE RELATIVE FLUX b V a) s -i < -< O n figure 20; {continued) RELATIVE FLUX RELRTIVE FLUX 4 6 8 10 fO61 S a> S6) s s EOCITY VELO SG> s CD PO s& 102 : 0 i _ § £ u s i f RELATIVE FLUX RELATIVE FLUX 00 oo to rsi CO X5876 7 8 5 X ! E H EI 6 7 8 5 X I HE R 509 9 0 5 MRK R 335 5 3 3 MRK (continued) -8000 -4000 VELOCITY VELOCITY 0 4000 8000 6000 12000 10 3 '01 RELATIVE FLUX S a FU OD (9 CD KJ VELOCITY MRK 5 0 9 RELATIVE FLUX Figure 20: (continued) 501 M RELATIVE RELATIVE FLUX 9 ro ts 9 9 00 9 a> VELOCITY RELRTIVE FLUX RELRTIVE 9 tO 9 00 a KJ 9 -t < -< o n m r~ ! (continued) gue 0 (continued) 20: uye ig F RELATIVE FLUX 04 CD CD CD EI 6 7 8 5 X I HE 3C 120 120 3C -4000 VELOCITY 0 6000 106 Ipp«ldix B PBOFZLB B illO S Profile ratios of Seyfert I galaxies. The ~ Mg/Ha profile ratio is given for nine Seyfert 1 galaxies and the He I A&876/H0 profile ratio is given for five Seyfert 1 galaxies. 108 ts"CT NGC 3 5 ) 6 LD H 0 /H « S m s (M (B -6000 0 2000 4000 VELOCITY to s' N G C 4 5 9 3 H /3/H o s cn s ot s 0 4000 VELOCITY Figure 21; (continued) B EL2 0J3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 40 4000 -4000 (continued) VELOCITY L CITY ELO V G 7469 9 6 4 7 NGC G 5548 8 4 5 5 NGC Ha /H 0 H JHa /J/H H 0 0 2000 2000 4000 109 110 MRK 7 9 in H 0 /H o s m s CN -6000 VELOCITY to MRK 110 in H /J/H a s m -2000 0 2000 60004000 VELOCITY Figure 21: {continued) 0.2 0.3 0.4 0.5 0.6 . 0-2 0-3 0.4 0.5 0.6 -4000 (continued) -2000 VELOCITY L CITY ELO V R 509 9 0 5 MRK Hj9/Ha 2000 4000 4000 111 112 to 3 C 120 m H 0 /H a s s (N •6000 -4000 -2000 0 2000 4000 6000 VELOCITY 10 s MRK 7 9 in HE I/H j3 B S fN 0 2000 4000 6000 VELOCITY Figure 21: (continued) dl£U: idJl££-U X 60 -00 20 0 00 00 6000 4000 2000 0 -2000 -4000 -6000 0-2 !LJ 0 ^ 0^5 0.1 0.2 0.3 0.4 0.S c i d) ed u tio D (co -2000 VELOCITY VELOCITY R 335 5 3 3 MRK H/3 /H 1 E H R 110 MRK HE HE 1 1/H0 4000 113 11** (0 MRK 5 0 9 ID H E I /H /9 S (M 0 2000 4000 VELOCITY ID 3C 1 2 0 HE I / H 0 in s (S CN -2000 VELOCITY Figure 21s (continued) BIBLIOGBAPSY Baldwin, J. A. 1977, H.H.B.i.S,. 178, 67P. Baldw in, J . A ., and N etzer, H. 197B, Ap . J.. 226, 1. Balick, B., and Beckaan, T- 0. 1982, Ann. Hew. &str. Ap.. 20, 431. B landford, B. D., and McKee, C« F. 1982, A p. J. . 255, 419. Bluaentnal, G. B., and Mathews, H. G. 1975, Ap . p, » 198, 517. Boroson, T. A., and Oke, J. B. 1982, Nature. 296, 397. Burstein, D., and Heiles, C. 1982, A. . 87, 1165. Byard, P. L., Foltz, C. B., Jenkner, H., and Peterson, B. H. 1981, Pah. A. S. P.. 93, 147. Capriotti, E., Foltz, C., and Byard, P. 1980, Aw J.# 241, 903. ______. 1981, 245, 5 9 6 . Capriotti, E. B., Foltz, C. B., and Peterson, B. N. 1982, l E i - J t • 261, 35. Carroll, T. J ., and Kwan, J. 1985, Aw J., 288, 73. Cherepashchuk, A. M., and Lyutyi, V. H. 1973, ftstrophys. Letters. 13, 165. Cohen, K- D. 1983, Ap . J.. 273, 489. Cota, S., Wagner, It. M., and Newsoa, G. B. 1985, in preparat ion. De Bobertis, H. 1985, 289, 67. de Vaucouleurs, G. 1975, Ap . J . Su ppI .. 29, 193. - 115 - 116 Fay# T. D. # S tein # H. 1.# and Barren# H. H. 1975# 0 6# H2m Ferland# G. J* and OsterbrocJt# D. E. 1985# Ap , J . . 289 # 105. F oltz# c. B. # ffilkes# B. J ,# and P eterson# B. H. 1983# A. J . . 88# 1702. Goodrich# B. H-# and Osterbrock# D. E. 1983#J. . 269# 416. Holt# S. S-# Moshotzky# B. P.# Becker# B. H-# Boldt# E• A.# Serleaitsos# P. J.# Szyakoviak# A. E. # and whit*? n. E. 1980# 241# L13. Huaason# H. L.# flayall# N. 0.# and Sandage# A. B. 1956# I x - J n 6 1* 97. Jenkner# H. 1980# Astron. laage Processing Circ. No. 6,. 30. Keel# V . C. 1980# 8S# 150- Keel# If. C. 1983, I P . J . . 269# 466. Khachikian# E. Y. # and Ifeedaan# 0. U. 1974# Ap^J*, 192# 581. Koski, A. T. 1978# Ap. J . - 223# 56. Kean# J . 1984# Apt . J.. 283# 70. Kean# J .# and K rolik# J. B. 1981# Ap . ,_{?t . 250# 478. Balkan# H. A.# and Filippenko# A. V. 1983# JU?A_ J i , 275# 477. Ha thews# 8. G- 1982a# Ap . J . . 258# 425- Hathews# B. G. 1982b# Ap. J.. 252# 39. Beyers# K. A.# and P eterson# B. FI. 198 5# Pnfr. A .5 . P,# 97# in p re s s . Hnshotzky# B. F. # Serleaitsos# P. J.# Becker# B. H.# Boldt# E. A.# and Bolt# S. S. 1978# Ap# J # . 220# 790. Osterbrock# D. E. 1974# AstroPbYSicsof Gaseons_Bebulae (San Francisco; Freeaan) • 1 J 7 ______. 1977, |pf. , 215, 733. ______. 1978a, Phvs. ScriPta, 17, 137. ______. 1978b, Phys. ScriPta. 17, 285. ______• 1978c, , 75, 540. ______. 1979, 84, 901. ______. 1981, l£x_dx, 249, 462. Osterbrock, 0. E., and Shuder, J. H. 1982, |f t Jf Supp }.. 49, 149. Peterson, B. S., and Collins, G. H., II. 1983, 270, 71. Peterson, B. H., Crenshaw, £. H., and Beyers, K. A. 1985, |pt J,. 298, in press. Peterson, B. B., Foltz, C. B., Byard, P. L., and Wagner, R. H. 1982, Ap . J. SUPPl. 49, 469. Peterson, B. H., Foltz, C. B., Crenshaw, D. B., Beyers, K. A., and Byard, P. L. 1984, |p. Jt. 279, 529. P e te rso n , B. 11., Beyers, K. A., Capriotti, E. R. , Foltz, C. B., Wilkes, B. J., and Hiller, H. R- 1985, |p. Jj. 292, 164. P h i l l i p s , H. H. 1976, Ap. -J.. 208, 37. - 1977, A P. J.. 215, 746. • 1978, , 38, 187. Pritchet, C. J. 1977, Ap. J. Suppl. . 35, 397. Pritchet, c. J., and van den Bergh, S. 1977, Ap . Sjjppl., 34, 10 1. Bichstone, 0. O., Ratnatunga, K., and Schaetfer, J. 1980, 240# 1- Robinson, E. L. and Wanpler, E. J. 1972, Pub. A.S.P.. 84, 161 . Seyfert, C. K. 1943, Ap . J.. 97, 23. 118 Shields, G. 1. 1978, Pittsburgh Conference on B1 Lac fitiS S lS * ®d *• ■* Wolfe, (Pittsburgh: University o£ P ittsb u rg h ) , p. 257* Shuder, J. H. 1981, Ap. J. 244, 12. . 1982. I p . Jt. 259, 48. . 1984, A P. J.. 280, 491. Siskin, S. H. , So, B. J ., and Schvarz, (1. P. 1980, ££x J., 2 3 7 , 404. Stone, 8. P. S. 1977, iE s -d s.# 218, 767. Tohline, J. £., and Osterbrcck, 0. E. | p. 252, L49. Tug, H., White, N. H. and Lockwood, G. B., 1977, AgtJOb._AstrQPhjsa. 61, 679.