Broadband Properties of Active Galactic Nuclei Citation Edelson, Richard Allen (1987) Broadband Properties of Active Galactic Nuclei. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/gtk1-3n45. https://resolver.caltech.edu/CaltechETD:etd-09092008-113501 Abstract The broadband radio-infrared-optical-ultraviolet properties of active galactic nuclei are used to investigate the nature of the central engine and the surrounding environment. Optically selected quasars (which have α̅ IR = -1.09; S ν ∝ ν α ) and Seyfert 1 galaxies ( α̅ IR = -1.15) tend to have relatively flat infrared spectra and low reddenings, while most Seyfert 2 galaxies ( α̅ IR = -1.56) and other dusty objects have steep infrared spectra and larger reddenings. The infrared spectra of most luminous radio-quiet active galaxies turn over near ~80 µ m. It appears that the infrared spectra of most quasars and luminous Seyfert 1 galaxies are dominated by unreprocessed radiation from a synchrotron self-absorbed source of order a light day across, about the size of the hypothesized accretion disk. Seyfert 2 galaxies and other reddened objects have infrared spectra which appear to be dominated by thermal emission from warm (~50 K) dust, probably in the disk of the underlying galaxy. A broad emission feature, centered near 5 µ m, is present in many luminous quasars and Seyfert 1 galaxies. Highly polarized objects ("blazars") can be strongly variable at far-infrared wavelengths over time scales of months. There is no conclusive evidence for far-infrared variations in normal (low-polarization) quasars or Seyfert galaxies, although low-level flickering (at the ~30% peak-to-peak level) cannot be ruled out. Seyfert galaxies tend to have steep radio spectra ( α rad ≈ -0.7). The radio spectra of Seyfert 1 galaxies often flatten out near 2 cm. There is no significant difference in the mean radio luminosities of Seyfert 1 and 2 galaxies. There are of order 10 5 Seyfert galaxies/Gpc 3 , most of which have 6 cm luminosities between 10 37.0 and 10 39.4 ergs/s and 60 µ m luminosities between 10 42.2 and 10 45.0 ergs/s. The Seyfert 2 galaxy radio luminosity function cuts off sharply below 10 37.4 ergs/s. Item Type: Thesis (Dissertation (Ph.D.)) Subject Keywords: Astronomy Degree Grantor: California Institute of Technology Division: Physics, Mathematics and Astronomy Major Option: Astronomy Thesis Availability: Public (worldwide access) Research Advisor(s): Schmidt, Maarten Group: Astronomy Department Thesis Committee: Schmidt, Maarten (chair) Moffet, Alan Theodore Scoville, Nicholas Zabriskie Readhead, Anthony C. S. Blandford, Roger D. Defense Date: 12 November 1986 Funders: Funding Agency Grant Number Caltech UNSPECIFIED Record Number: CaltechETD:etd-09092008-113501 Persistent URL: https://resolver.caltech.edu/CaltechETD:etd-09092008-113501 DOI: 10.7907/gtk1-3n45 Related URLs: URL URL Type Description https://doi.org/10.1086/164479 DOI Article adapted for Chapter Two. https://doi.org/10.1086/184763 DOI Article adapted for Chapter Three. https://doi.org/10.1086/165004 DOI Article adapted for Chapter Four. https://doi.org/10.1086/165627 DOI Article adapted for Chapter Five. Default Usage Policy: No commercial reproduction, distribution, display or performance rights in this work are provided. ID Code: 3413 Collection: CaltechTHESIS Deposited By: Imported from ETD-db Deposited On: 17 Sep 2008 Last Modified: 05 Nov 2021 23:00 Thesis Files Preview PDF (Edelson_r_1987.pdf) - Final Version See Usage Policy. 18MB Repository Staff Only: item control page

BROADBAND PROPERTIES OF ACTIVE GALACTIC NUCLEI Thesis by Richard Edelson In Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy California Institute of Technology Pasadena, California 1987 (Submitted 12 November 1986) To my family and friends — ili - ACKNOWLEDGEMENTS Many people have contributed to my development as a scientist, and this thesis is a result of their efforts as well as mine. I hope that I can carry on this tradition by helping others advance science as I have been helped. I thank my thesis advisors, Maarten Schmidt and Alan Moffet for their guid- ance. I am grateful for the freedom and independence they allowed me to pursue my ideas, and the direction and help they gave to me to turn them into reality. Without my friend and collaborator Matthew Malkan, this thesis would look much different. From helping me develop these ideas to collaborating in the gathering and understanding of the data, he has left an indelible imprint. None but the sturdiest of souls could have worked so long and closely with me—he has witnessed a particularly concentrated form of my unique brand of insanity. As my mentor during my undergraduate years, Sumner Davis inspired me to develop the discipline and knowledge a scientist needs. Tony Readhead and Nick Scoville were always willing to help me with any problem, scientific or otherwise, that I encountered. Although progress has generated dislocation and angst in today’s society, these men remind me of the compassionate side of science. Many others have contributed to this work. Walter Rice and Helen Hanson spent a great deal of time helping me understand the IRAS data which formed the backbone of the second, third, fifth, and seventh chapters of this thesis. Harry Hardeback, Richard Moore, and the rest of the Owens Valley Radio Observatory staff helped me gather the radio data used in the third and fourth chapters. George Rieke helped gather the near-infrared data used in Chapter 5 and collaborated in writing the paper. Tom Wahl reduced much of the IRAS variability data used in Chapter 6. John Huchra provided me with the list of CfA Seyfert galaxies long, long before publication. Roger Blandford, Sterl Phinney, and Jeremy Mould gave me much useful advice. David Cudaback, Carl Heiles, Ken Turner, and Harold pie Weaver aided my development as an undergraduate. Marshall Cohen, Nick Scoville, Maarten Schmidt, Alan Moffet, Tony Readhead, Tom Phillips, and the institute itself provided financial support for my research. My long hours in the library were made more pleasant by the presence of Helen Knudson. I am also grateful to Kieth Shortridge, Mike Lesser, Peter Parniky, and Chris Lee for their tireless efforts to keep the computers running and helping me with programming questions, and to Lilo Hauck, Jill King, Ann Palfreymann, Marilyn Rice, Wendy Zhorne and the other Caltech and IPAC personnel for helping me through the maze of paperwork. The company of my fellow graduate students and postdocs was very important to me. I really enjoyed the company of my office mates Erich Grossman, Dean Chou, Frank Evans, and Fernando Selman. Special thanks also goes to Shri Kulkarni, Jules Halpern, Alex Filippenko, Dave Tytler, Blaise Canzian, Chuck Steidel, Chris Impey, Jim Schombert, Lynn Deutsch, Gary Hinshaw, Charley Lawrence, Greg Bothun, Lucy Ziurys, Haimin Wang, Lee Mundy, Kwok-Wai Cheung, Kevin Lind, Alain Porter, and all you others who helped and watched me grow. This moment of reflection also reminds me how lucky I am to have such great friends. When work became too much for me, their company preserved the remnants of my sanity. To all of my crazy friends in L.A., Berkeley, and the other parts of the world (especially the denizens of our house on Prospect Street), I love you. This goes especially for my parents, brother, and grandmother. Surely this is what gives life meaning. Once again, to all of you who have shared this endeavor with me, thanks. -yv- ABSTRACT The broadband radio-infrared-optical-ultraviolet properties of active galactic nuclei are used to investigate the nature of the central engine and the surrounding environment. Optically selected quasars (which have @ypz = —1.09; S, « v™) and Seyfert 1 galaxies (@7p = —1.15) tend to have relatively flat infrared spectra and low reddenings, while most Seyfert 2 galaxies (@7z = —1.56) and other dusty objects have steep infrared spectra and larger reddenings. The infrared spectra of most luminous radio-quiet active galaxies turn over near ~80 ym. It appears that the infrared spectra of most quasars and luminous Seyfert 1 galaxies are dominated by unreprocessed radiation from a synchrotron self-absorbed source of order a light day across, about the size of the hypothesized accretion disk. Seyfert 2 galaxies and other reddened objects have infrared spectra which appear to be dominated by thermal emission from warm (~50 K) dust, probably in the disk of the underlying galaxy. A broad emission feature, centered near 5 pm, is present in many luminous quasars and Seyfert 1 galaxies. Highly polarized objects (“blazars”) can be strongly variable at far-infrared wavelengths over time scales of months. There is no conclusive evidence for far- infrared variations in normal (low-polarization) quasars or Seyfert galaxies, al- though low-level flickering (at the ~30% peak-to-peak level) cannot be ruled out. Seyfert galaxies tend to have steep radio spectra (apgqg * —0.7). The radio spectra of Seyfert 1 galaxies often flatten out near 2 cm. There is no significant difference in the mean radio luminosities of Seyfert 1 and 2 galaxies. There are of order 10° Seyfert galaxies/Gpc?, most of which have 6 cm luminosities between /s. 1037-0 and 1039-4 ergs/s and 60 ym luminosities between 104%? and 1049 ergs, o Svs The Seyfert 2 galaxy radio luminosity function cuts off sharply below 10°74 ergs/s. —-vi- TABLE OF CONTENTS Acknowledgements ........... De eee Abstract 2 6 6 ee ee ee Chapter 1. Introduction ..... 2... 1. ee ee ee ee ee ee Chapter 2. Spectral Energy Distributions of Active Galactic Nuclei Between 0.land100 wm ..............2.04+.2-. Chapter 3. Far-Infrared Properties of Optically Selected Quasars ..... Chapter 4. Broadband Properties of the CfA Seyfert Galaxies: I. Radio Properties .. 1... 1 eee ee ee te ee Chapter 5. Broadband Properties of the CfA Seyfert Galaxies: IJ. Infrared-Millimeter Properties ..........+.4.-.--. Chapter 6. Far-Infrared Variability in Active Galactic Nuclei . . 2... Chapter 7. Future Work ... 2... 1 eee ee ee ee es Chapter One INTRODUCTION ~Q- Active galactic nuclei (AGNs) are the most powerful objects known to exist in the universe. They are believed to be powered by a supermassive (~ 108 Mo) black hole, accreting matter at the rate of (~ 1 Mo/year), and converting it into energy with a relatively high (~ 10%) efficiency and a very large bandwidth ( 210 octaves). Their high energy output (~ 104° ergs/sec) often exceeds that of all other processes in the galaxy combined. However, all of this energy is ultimately thought to be generated in a black-hole-accretion-disk system which is only of order a light day across. The precise method by which these enormous amounts of matter are converted in to energy is not yet known. These objects produce nonthermal radiation over a very broad bandwidth. Therefore, only by measuring the continuum (its shape, variability, and polariza- tion) is it possible to study the central regions directly. Other techniques (emission lines, mapping, etc.) are sensitive to radiation produced in much larger regions. The primary purpose of this work is to determine the contribution of the nucleus and other more extended emission regions to the broad-band continuum. Studying the broad-band characteristics of these components and their relation to: other spectral characteristics will yield information about the physical conditions and processes occurring in and around the nucleus. Besides this introduction and a concluding chapter, this thesis consists of five separate but related papers, each confined to studies in a given wavelength range. Four have been submitted to the Astrophysical Journal, and the remaining one will be submitted shortly. There is one appendix, which gives the optical spectra of the Seyert galaxies used in Chapters 4 and 5. The remainder of this chapter is devoted to a discussion of the motivation and justification for this work. This work can be convieniently broken into three sections. The importance of the infrared-optical-ultraviolet spectral properties of AGNs is discussed in the next section. The rationale behind the study of far-infrared ~3- variability is discussed in the next section, and radio spectra are dealt with in the final section. Each section consists of a discussion of previous work and the scientific justification for the research undertaken. I, INFRARED-OPTICAL-ULTRAVIOLET SPECTRAL ENERGY DISTRIBUTIONS A) Previous Work Much has already been learned from previous studies of the ultraviolet-optical- infrared continuum properties of AGNs. In a pioneering study, Rieke (1978) found that the spectra of Seyfert galaxies rose from the optical into the near-infrared (arr © —1; Sp « v™). Their spectra are much steeper than those of normal galaxies, which cut off sharply near 1.5 ym. This near-infrared excess was explained as thermal emission from hot dust (T21000 K). That study also found that the ultraviolet-excess Markarian survey was biased against the inclusion of objects with strong infrared excesses. Neugebauer et al. (1979) studied the optical-near-infrared spectra of quasars. They favored a nonthermal explanation for the infrared excess. They also found that the 2.2-3.5 zm spectral indices of many quasars were steeper than their 1.2-2.2 um spectral indices, although this discovery was not followed up for many years. Malkan and collaborators (Malkan and Sargent 1982, Malkan and Oke 1983, Malkan 1983) fit the ultraviolet through near-infrared continuua of AGNs with a multi-component physical model. They used a nonthermal power law to model the near-infrared excess. They also found that the ultraviolet excess could be explained as a combination of thermal emission from a hot accretion disk (T = 26,000 K) and Balmer continuum emission. Thus, the nature of the infrared excess has not been definitively established. This can be seen in two models of the optical-near-infrared continuum of the best ~4- studied Seyfert 1 galaxy, NGC 4151. Rieke and Lebofsky (1981) modeled it as thermal emission from dust, while Malkan and Sargent (1982) used a nonthermal power law. Both papers obtain reasonable fits, and find that the other explanation is not acceptable. B) Scientific Justification The fundamental cause of the disagreement over the nature of the infrared emis- sion lies in the fact that the ground-based data were always limited to wavelengths A < 10 pm, and often to A < 2—3 pm. This coverage is not sufficient to allow the infrared excess component to be unambiguously separated from other emission components (galactic starlight and the ultraviolet blackbody). A second problem is that those studies used arbitrarily chosen objects, instead of well-defined samples. For instance, only one Seyfert 2 galaxy (NGC 1068) was carefully studied. This greatly reduced the power of statistical arguments to discriminate between different processes in different types of objects. This study addresses these issues directly. Chapter 2 contains detailed model fits to the 0.1-100 um spectral energy distributions of a heterogeneous sample of bright AGNs. Extending the wavelength coverage to 100 ym allows clean separation of the optical-ultraviolet components from the infrared excess. It also allows the first systematic study of the near-infrared feature discovered by Neugebauer et al. (1979), as well as an analysis of the effects of reddening and a detailed study of the ultraviolet, optical, and near-infrared features studied previously. Detailed statistical analysis of the infrared (1-100 um) properties of the bright- at NINNMG ~..n ener an A the St U/D&o Guadsals alld tie qi] 5, respectively. These chapters focus on determining the nature of the infrared emission from the nucleus and the underlying galaxy. Use of these complete, optically selected samples makes powerful statistical tools available. -~5- II. FAR-INFRARED VARIABILITY A) Previous Research Highly polarized AGNs (BL Lac objects and OVV quasars) are generally strongly variable at all wavelengths from X-rays to radio (see, e.g., Bregman et al. 1985). Seyfert 1 galaxies and normal (low-polarization) quasars are less violently variable, but variability from X-ray to near-infrared wavelengths have been well documented (e.g., Halpern 1982, Cutri et al. 1985, Neugebauer et al. 1986), while Seyfert 2 galaxies are generally not variable. Most normal AGNs cannot be detected from the ground at wavelengths longer than 10 ym, so only one object (NGC 4151) has been monitored longward of 3.5 um (Rieke and Lebofsky 1981). B) Scientific Justification Detectable thermal emission from AGNs at wavelengths significantly longer than 10 wm must generally emanate from a region many light-decades across. Therefore, the detection of far-infrared variations would demonstrate that the continuum emission is nonthermal. In Chapter 6, pointed IRAS observations are used to monitor the far-infrared fluxes of blazars, normal quasars, and Seyfert 1 and 2 galaxies. III. RADIO EMISSION A) Previous Research Although the first quasars were discovered because of their intense radio emis- sion, only a small fraction of all AGNs are “radio-loud” objects. Radio surveys of optically selected AGNs have generally had poor detection rates (see, e.g., Condon et al. 1981), so the radio properties of the underlying population of AGNs are not known. The nature of the difference between radio-loud and radio-quiet AGNs has been an outstanding question since quasars were first discovered. -6- B) Scientific Justification Two different approaches to this problem are pursued in this thesis. The first involves use of the IRAS data. The spectra of most radio-quiet AGNs rise from the optical into the infrared, but (unlike radio-loud objects) this component must cut off somewhere in the far-infrared-submillimeter region. Depending on the nature of the infrared emission, determination of this turnover frequency can be used to determine source parameters. The derivation and implications of the low-frequency cutoff is discussed in Chapters 2, 3, and 5. Also, the radio properties of the CfA Seyfert galaxies are discussed directly in Chapter 4. This is the first study of optically selected (i.e., radio-quiet) AGNs for which 100% are detected at radio wavelengths, and three frequency radio spectra were determined for a majority of the objects. This allows the radio properties of optically selected Type 1 and 2 Seyfert galaxies to be properly studied with powerful statistical tools. These properties include radio spectra, variability, resolution effects, and luminosity function. -7- REFERENCES Bregman, J. et al. 1986, Ap. J., 301, 708. Condon, J. et al. 1981, Ap. J., 246, 624. Cutri, R. et al. 1985, Ap. J., 296, 423. Halpern, J. 1982, Ph.D. Thesis, Harvard University. Malkan, M. A. 1983, Ap. J., 268, 582. Malkan, M. A. and Sargent, W. L. W. 1982, Ap. J., 254, 22. Malkan, M. A. and Oke, J. B. 1983, Ap. J., 265, 92. Neugebauer, G. et al. 1979, Ap. J., 230, 79. Neugebauer, G. et al. 1987, Ap. J., in press. Rieke, G. 1978, Ap. J., 266, 550. Rieke, G. and Lebofsky, M. J. 1981, Ap. J., 250, 87. Chapter Two SPECTRAL ENERGY DISTRIBUTIONS OF ACTIVE GALACTIC NUCLEI BETWEEN 0.1 AND 100 MICRONS * Reprinted from The Astrophysical Journal (1986, v. 308, pp. 59-78) * Written in collaboration with M. Malkan THE ASTROPHYSICAL JOURNAL. 308: 59-77, 1986 September | € 198¢ The American Astronomical Society All nghis reserved Pnotedin U.S.A SPECTRAL ENERGY DISTRIBUTIONS OF ACTIVE GALACTIC NUCLEI BETWEEN 0.1 AND 100 MICRONS R. A. EDELSON Palomar Observatory, California Institute of Technology AND M. A. MALKAN Department of Astronomy. University of California. Los Angeles Received 1985 October 28 ; accepted 1986 February 20 ABSTRACT We have combined new photometry from the Infrared Astronomy Satellite with ground-based optical and near-infrared spectrophotometry and ultraviolet spectrophotometry from the International Ultraviolet Explorer to study the spectral energy distributions from 0.1 to 100 yum for 29 active galactic nuclei. After correcting for the effects of starlight and foreground and internal reddening. we fitted the spectra with a simple model. varying the minimum number of parameters needed to describe the data accurately. The infrared continua of quasars and Seyfert 1 galaxies with low reddening were well fitted by a power law with a relatively flat slope (3, = — 1.36 + 0.21). Their fitted power law appears to be nonthermal in origin, and the power-law luminosity is well correlated with the 2 keV luminosity, with an average slope from optical to X-ray frequencies of 2, —1.15 + 0.02. In 16 of the 29 AGNs the power law had a low-frequency turnover shortward of 100 um, and many of the less luminous objects probably have turnovers which are undetected because of emission from cold dust from the underlying galaxy. For most of the Seyfert 1 galaxies and quasars in our sample, the turn- over appears to be due to synchrotron self-absorption in a region similar in size to the hypothesized inner accretion disk. The infrared continua of Seyfert 2 galaxies and other reddened objects tend to be steep (Z), = — 1.58 + 0.40), and cannot be accurately described by a power law. In these dusty objects, the flux beyond 10 ym is dominated by thermal reradiation from dust in a volume similar to that of the narrow-line region, at a minimum temperature of 40-80 K, hotter than dust in normal galaxies. An additional flux component was required to fit accurately the near-infrared continua of over half the Seyfert 1 galaxies and quasars. We described it as a parabola. and found it was always centered near 58 THz (5.2 ym), with a width of about 3.3 octaves. The strength of this feature. which contains up to 40°. of the luminosity between 2.5 and 10 um, is correlated with the luminosity and Hf ‘[O mJ ratio. The ultraviolet con- tinua of all of the Seyfert 1 galaxies without large internal reddenings required the inclusion of a hot thermal component in the fits. which we described as a blackbody with fitted temperatures around 26.000 K. The mean Balmer continuum/Hz2 ratio is 2.1 + 0.6 for Seyfert 1 galaxies, which is about 50% higher than the case B prediction. Subject headings: galaxies: nuclei — galaxies: Seyfert — infrared: spectra — quasars — radiation mechanisms — spectrophotometry 1. INTRODUCTION A major obstacle to understanding energy generation in active galactic nuclei (AGNs) has been the difficulty of measur- ing the complete continuous energy spectrum, which has a comparable luminosity per octave (in vF,) at infrared, ultra- violet, and hard X-ray wavelengths. The largest gap in pre- vious spectral coverage has been between 10 um and 1 cm, where low atmospheric transmission and poor instrumental sensitivity have prevented detection of all but the very bright- est AGNSs. The far-infrared region (30-1000 um) is especially important because it can include the bulk of an AGN’ inte- grated power, and the transition from infrared to radio wave- lengths requires a dramatic spectral turnover in all but the strongest radio sources. Until recently, only a the far-infrared. The brightest Seyfert galaxy, NGC 1068, was found to have a peak flux density around 80 um (Hildebrand et al. 1977; Telesco and Harper 1980). The quasar 3C 273 has been detected at centimeter, millimeter, submillimeter, and far- 59 infrared wavelengths (Robson eral. 1983), as has 3C 345 (Harvey, Wilking, and Joy 1982). Rickard and Harvey (1984) observed NGC 1068, NGC 3227, NGC 4051, and NGC 7469 from 40 to 160 ym. The Infrared Astronomy Satellite (IRAS) recently surveyed 98% of the sky with unprecedented sensitivity in broad wave- length bands centered at 12, 25, 60, and 100 yum. It greatly increased the available data on the far-infrared continua of the brightest AGNs, making it possible to study a large sample with uniquely broad wavelength coverage. In this paper we combine new photometry from JRAS with ground-based optical and near-infrared spectrophotometry, and ultraviolet spectrophotometry from the International Ultraviolet Explorer (IUE), to study the broad-band properties of AGNs. An outstanding question is the relative importance of thermal and nonthermal emission at infrared wavelengths. There is little doubt that thermal emission from warm dust grains dominates the 10-100 ym flux (and therefore the total energy output) of NGC 1068 and most, if not all, other -~10- 60 EDELSON AND MALKAN Seyfert 2 nuclei (Rieke and Lebofsky 1979a). Some Seyfert 2 nuclei. most notably NGC 1068, must also have a smaller quantity of hotter dust which can dominate the spectrum at wavelengths as short as 3 um (e.g. Malkan and Oke 1983, hereafter MO). A related observation is that all closely studied Seyfert 2 nuclei have emission-line regions with at least a mod- erate amount of reddening not produced in the Milky Way. e.g.. with internal continuum dust optical depths t > 1 below 3000 A (see § IVd and Malkan 1986). The mechanism producing the far-infrared emission of Seyfert 1 galaxies and quasars is more controversial (Rieke and Lebofsky 19796). Some highly reddened objects such as Mrk 231 must have infrared spectra dominated by thermal emission. Like NGC 1068. Mrk 231 has a spectra dip around 10 ym which has been identified as silicate absorption (Rieke 1976). Only two Seyfert 1 galaxies. NGC 3227 and NGC 7469, have had detected emission in the 3.3 um dust feature, but, in NGC 7469 at least. much of it comes from H 1 regions in the galaxy rather than from the nucleus (Cutri et al. 1984: Cutri 1986). On the other hand. many Seyfert 1 nuclei have optical to near-infrared spectral energy distributions (SEDs) which extend smoothly to 10 wm with a relatively constant power- law slope x ~ —1.2 (F, x v?) (Malkan and Filippenko 1983. hereafter MF). Careful narrow-band photometry in the near- infrared has failed to detect silicate absorption. or any emission features (e.g.. at 3.3 or 11.6 um) which are known to come from dust grains (Rudy er al. 1982b: Roche et al. 1984). Moreover. the optical to ultraviolet colors of many Seyfert 1 nuclei. espe- cially the luminous ones. are confined to a relatively narrow, blue range which rules out continuum reddenings exceeding E,-, = 0.05-0.10 (Malkan 1984). This is consistent with upper limits derived from the absence of 2175 A dust absorption (Malkan and Sargent 1982. hereafter MS). or the absence of soft X-ray absorption (Pounds and McHardy 1985). In § II we present further evidence that most of the Seyfert | and quasar SEDs in this study are not strongly affected by dust. One of the major goals of this paper is to determine the relation between properties derived at different wavelengths. We used the Spearman rank test (see Siegel 1956). a nonpara- metric test, to search for correlations between derived param- eters when no upper limits were present in the data. and the Cox proportional hazard method when some of the data for one variable contained upper limits. A good discussion of the Cox test is given in Isobe. Feigelson, and Nelson (1986).' We describe the data and methods used to construct the broad-band SEDs in the next section. and the multicomponent fits to the SEDs in § III. We discuss the determination and implications of the far-infrared emission and low-frequency turnover in § JV. The strong. broad feature found to be cen- tered at 5 um in many SEDs is discussed in § V. The nature of the ultraviolet emission is discussed in § VI, and the relation- ship between properties determined from ihe 0.1-100 ym SEDs and those derived at other frequencies is discussed in § VH. The major results of this paper are summarized in § VII. Throughout. we adopt a value of Hyp = 75 km s7! Mpc”! and gy = 0. HL DATA Our sample consists of 29 of the brightest AGNs which have good IRAS detections. complete ground-based optical and ) The nonparametric Kolmogorov-Smirnovy (K-S) test was used to test for differences in the distribution of univariate data (Siegel 1956). Vol. 308 infrared spectrophotometry. and JUE data. These data together provide nearly continuous spectral coverage over 10 octaves of frequency. This sample is a heterogeneous collection of AGNs. Some sources selected for our study were originally identified in ultraviolet-excess surveys, which tend to miss reddened objects. On the other hand, the limited sensitivity of 7RAS will tend to single out AGNs with exceptionally strong far-infrared contin- ua. These conflicting biases undoubtedly operated on the present sample. Only three of the brightest Seyfert 2 galaxies. NGC 1068, Mrk 3, and Mrk 348, had sufficiently complete spectral coverage to be included. We required that all program objects have ultraviolet spectra. usually in both the long and short-wavelength cameras of the JUE. Fortunately, JUE has now observed most of the brighter Seyfert galaxies at least once. Data were taken from the literature (see references in Table 1) or the archives of the National Space Sciences Data Center, using standard reduction programs at the Regional Data Analysis Facility in Greenbelt. The ultraviolet data were binned in ~ 100 A bands devoid of strong emission lines. to give typical signal-to-noise ratios of 10 or more. JUE’s 10” x 20” oval aperture agrees adequately with the ~ 10" apertures used in the optical and near-infrared. especially because most of the ultraviolet flux is unresolved. The ~ 10” apertures used in the optical are a good compro- mise between the conflicting requirements of obtaining accu- rately calibrated absolute spectrophotometry of the unresolved nucleus while excluding as much of the starlight as possible. We do not use broad-band optical photometry (1e.. UBT’). since it does not allow precise separation of emission lines and Balmer continuum emission from the principal object of our study. the underlying continuum. The sources of our intermediate-resolution optical spectrophotometry are listed in Table 1. Some optical spectrophotometry for objects in this study without previously published fluxes is presented in the Appendix. As with the ultraviolet data. we excluded optical wavelengths near strong emission lines. Low resolution is acceptable in the infrared. which seems to be free of strong emission lines. Thus. for objects without narrow-band infrared spectra, we have used broad-band infra- red fluxes. at J (1.2 um). H (1.6 wm). K (2.2 pm). L (3.5 pm). N (10.6 xm). and Q (21 ym). We converted the magnitudes to absolute fluxes using the calibrations from Campins. Rieke. and Lebofsky (1985) and Rieke. Lebofsky. and Low (1985). New near-infrared photometry for some of the AGNs analyzed in this paper is presented in the Appendix. A number of objects in this study have published CVF spectra. For NGC 1068 we adopted the average of the 8-13 ym spectra from Roche er al. (1984) and Rudy er al. (1982b). Rudy er al. (19826) adjusted the absolute flux levels of the near-infrared spectra to agree with broad-band K (2.2 um) and L (3.5 um) measurements. We readjusted them to agree with the K and L fluxes appropriate for the epoch and aperture size of our multiwavelength spectra. The far-infrared JRAS photometry. at 12. 25. 60. and 100 ym. is discussed in the Appendix. The correction of the IRAS fluxes to 10” apertures using ground-based 10.6 and 21 um data is discussed in § IIIb. Observations at 100 and 160 um from Rickard and Harvey (1984) were also included for the four galaxies in common with this study. Although most of our ultraviolet optical ‘near-infrared far- infrared spectra were not simultaneously observed, they are well matched at 0.3 and 1.0 um, and are probably quite similar to spectra which would result from truly simultaneous obser- —~1l1l-— No. 1. 1986 SPECTRAL ENERGY DISTRIBUTIONS OF AGNs 61 TABLE } AGN SaMPLE Data Source Coordinate Ey. Es-y Scale Fraction Name Name Galactic? Interna]? Factor‘ Starlight? References Mrk 338 20.0... eee 0003 + 19 0.00 0.02 0.75 0.13 1-4 Mrk 348 see 0046 + 31 0.05 0.15 0.80 0.56 2, 5.6 IZwl...... 0050+ 12 0.04 0.06 0.75 0.19 1, 2.6.7 NGC 1068 . 0240-00 0.01 0.29 0.75 0.72 6. 8-10 3C &4 0316+41 0.03 0.02 1.00 0.38 4.5. 10-13 3C€ 120... 0430+ 05 0.03 0.06 0.85 0.18 46, 10, 13.14 Akn 120. 0S13-—00 0.09 0.02 0.65 0.13 1.13.15 Mrkh 3... 0609 + 71 0.15 0.10 0.55 0.90 8 Mrk6... we 0645 +74 0.09 0.22 0.95 0.78 8 Mrk 376 ........--eee ee 0710+ 45 0.09 0.10 0.45 0.33 2.5 0732+ 58 0.03 0.08 0.85 0.18 2. 3.5.16 0738 +49 0.06 0.18 0.65 0.37 2. 3.5. 14.17 0906 + 48 0.00 0.03 1.00 0.30 9 1020+ 20 0.02 0.33 0.65 0.59 2. 5.18 1103472 0.02 0.10 0.70 0.68 2, 5.17 NGC 4051 1200 + 48 0.00 0.08 0.55 0.82 2. 5, 18 NGC 4151. 1208 + 39 0.00 0.13 0.85 0.23 2, 3. 5.12, 17-19 3€ 273 ..... 1226 +02 0.04 0.05 0.90 0.04 1-3. 6. 10. 20 Mrk 23] 1254457 0.00 0.20 1.00 0.30 2.10. 11 PG occ eee 1351 +64 0.03 0.03 1.00 0.08 21, 22 Mrk 279 ..........cee ee 1351+69 0.00 0.13 0.65 0.40 2.5 NGC 5548 . 1415425 0.00 0.02 0.90 0.26 2.3.5,12 Mrk 817 ... 14344 $9 0.00 0.06 0.80 0.19 - 1 Mrk 478 ... . 1440 + 35 0.00 0.0) 08s 0.21 2,5 Mrk 841 2.00... 1501+ 10 0.00 0.10 0.75 0.12 1, 23 3C 390.3 ooo 1845 +79 0.03 0.22 1.00 0.25 16, 24 Mrk S09 . . 2041-10 0.03 0.00 0.75 0.11 1-4. 6, 10. 25 Hi Zw 136 .. . 2130409 0.04 0.06 1.10 0.10 1.2.5 NGC 7469 ............. 2300+ 28 0.03 0.07 0.70 0.34 2. 4, 10-12, 25 * Extinction. in visual magnitudes. due to line-of-sight dust in the Milky Way. estimated from 21 cm H 1 column density method of Burstein and Heiles 1982. © Extinction. in visual magnitudes. attributed to Seyfert nucleus. in order to obtain good fit to ultraviolet contin- uum. assuming a standard Milky Way reddening law. © Scale factor by which all 1R.4S far-infrared fluxes were multiplied to force them to agree with ground-based measurements which refer to the nuclear emission (within a small aperture) * Fraction of total flua at 5450 A which is produced by starlight from the underlying galaxy. REFERENCES.—{1) This paper. (2) De Bruyn and Sargent 1978. (3) Malkan and Sargent 1982. (4) Rudy er al. 1982a. (5) Rieke 1978. (6) Roche er al. 1984. (7) Snijders er al. 1982. (8) Malkan and Oke 1983. (9) Neugebauer er al. 1979 (10) Rudy et al. 1982h. (11) Aitken. Roche. and Phillips 1981. (12) McAlary er al. 1983. (13) Sargent. de Bruyn. and Readhead 1987. (14) Oke and Zimmerman 1979. (15) Glass 1981. (16) Oke and Goodrich 1981. (17) McAlary. McLaren. and Crabtree 1979. (18) Balzano and Weedman 198]. (19) Boksenberg et al. 1978. (20) Ulrich er al. 1982. (21) Neugebauer et a/. 1980. (22) Maraschi. Tanzi. and Treves 1980. (23) Rudy. LeVan. and Rodriguez-Espinosa 1982. (24) Miley et al. 1984.(25) Malkan and Filippenko 1983. vations. Fortunately. the near-infrared continua of most AGNs are known to have small amplitudes of variability (Cutri er al. 1985: Lebofsky and Rieke 1980). In the few cases where the published optical spectrophotometry of de Bruyn and Sargent (1978) did not agree well with the JUE spectra or the near- infrared data, as noted in Table 1. we replaced it with new optical spectra supplied by Sargent. de Bruyn. and Readhead (1986). Il. FITS TO THE SPECTRAL ENERGY DISTRIBUTIONS a) Underlying Galaxies Because of their steep luminosity functions, the brightest AGWNs tend to be nearby and of low luminosity. Thus. many of our program spectra include significant starlight from the underiying galaxy, which must be accurately subtracted io study the nonstellar continuum. We use the estimates of star- light fluxes from Crenshaw and Peterson (1985). MF, MO. and Neugebauer er al. (1985). For Seyfert galaxies not included in those studies, we made new estimates of starlight fluxes based on direct SIT pictures and slit spectroscopy at the Palomar 1.5 m telescope. The tech- nique used to measure the starlight is essentially the same as in MF. Our new estimates, included in column (6) of Table 1. may not be quite as accurate as those of MF, owing to the lower resolution and signal-to-noise ratios of the spectra. The spectrum of the galaxy was assumed to be the average spec- trum given in Lebofsky and Rieke (1985). It is very similar to the early-type “ standard galaxy” spectrum in MO, and should be an adequate description of the nuclear regions of most of the program Seyfert galaxies. In most cases starlight contributes less than 25° of the visual flux, and our starlight subtraction does not significantly increase the uncertainties of our fits. However. in a few of the Seyfert 1 galaxies (NGC 3227, NGC 3516, NGC 4051. and Mrk 6) and in all the Seyfert 2 galaxies (NGC 1068, Mrk 3. and Mrk 348) starlight is a major component, dominating the near-infrared spectra of these objects, which leads to larger uncertainties in the fits to the nonstellar flux in the red and near-infrared. ~12- 62 EDELSON AND MALKAN b) IRAS Scale Factors All but three of the objects studied (Akn 120. Mrk 817. and Mrk 841) were detected by JRAS at 12 um. and from the ground at 10.6 um. Eight also have ground-based measure- ments at 21] ym which can be compared with the /RAS 25 um fluxes. Since the ground-based photometry was obtained with small apertures ( ~ 10” in diameter). the measured fluxes can be lower than the JRAS fluxes (which usually include all of the infrared emission from the entire galaxy). This difference is still present after allowing for the shorter wavelengths of the ground-based observing windows. Our aim was to estimate the far-infrared emission coming from the central region of the galaxy (within 5” of the nucleus). since all our data at shorter wavelengths also refer to that region. We attempted to correct the large-aperture JRAS data by scaling them down to match the small-aperture ground- based fluxes. For each object, all four JRAS fluxes were multiplied by a constant scale factor. listed in Table !. which varied from unity (for sources with no extended fiux) to 0.45 (for the objects with the most contamination from the underlying galaxy). The IRAS scale factor was determined by taking the ratio of the 10.6 and 12 um fluxes. assuming a spectral slope x= —1. When 21 ym measurements were also available. the 21-25 ym ratios were also determined. and the scale factor was taken to be the average of these two values (which generally agreed quite well). The scale factors of the three objects which lacked 10 um data were determined by extrapolation from shorter wavelengths. and consequently have larger uncertainties. The adopted scale factors could be in error by as much as 0.10. As expected. they are correlated with the redshifts of the AGNs. since the nearest sources (subtending the largest angles) are the best resolved. One object (I] Zw 136) had a scale factor of 1.1, probably due to uncertainty in the 10.6 or the 12 zm measurement or to mild variability. In the worst cases (NGC 4051. Mrk 376, and the Seyfert 2 galaxies) we may have over- estimated the far-infrared flux arising in the nucleus. since the sources are likely to be more compact at 10-12 ym than at longer wavelengths. The bulk of their observed 60 and 100 pm flux could actually arise outside the nucleus. in the galactic disk. NGC 6814 had such an extreme infrared scale factor (only 0.30). and so much starlight. that we were unable to include it in our sample. The far-infrared flux from the galactic disk was independent- ly estimated from published large-aperture B-magnitudes. using standard values of 23, ,, for Sab galaxies taken from de Jong et al. (1984). In most cases this was a negligible fraction of the flux at 60 and 100 um. However. for three objects (NGC 3516. NGC 4151, and NGC 5548) this estimate is over 25% of the 60 and 100 um flux. Our failure to detect the low- frequency turnovers (see § IV) for these and a few other low- luminosity AGNs (such as NGC 3227 and NGC 4051) may be due to contaminating extranuclear emission from cold dust in the disk of the surrounding galaxy. This is the simplest expla- nation of why such a high proportion (two-thirds) of the AGNs with small scale factors (<0.70) had no detectable turnover frequency. Rickard and Harvey (1984) observed four of the sources with extremely small scale factors in this sample (NGC 1068. NGC 3227, NGC 4051. and NGC 7469). They used a 40” beam at 40. 50, 100. and 160 ym. We have included their 100 and 160 um data. with no scale factor correction applied because their Vol. 308 aperture was much smaller than that used by /RAS. Their small-aperture fluxes agree well with the scaled JR.4S 100 um data. providing independent evidence that our scale factors (derived at 12 and 25 ym) adequately correct for the contami- nation from the extended disk at 60 and 100 ym in most cases. The uncertainty in these scale factors makes it virtually impossible to search for 10 ym variability by comparing RAS data with previous ground-based measurements. There 1s no evidence for strong (e.g.. Am > 0.5 mag) variability at 10 um. because no /RAS scale factor was found to be significantly larger than unity. c) Reddening The fits were all reddened assuming a standard extinction law (Seaton 1979), by amounts estimated from the 2] cm H1 column densities in the Milky Way (Burstein and Heiles 1982). In 17 cases. additional internal nuclear reddening (at the AGN’s redshift) of Eg_, 20.07 had to be invoked. Our approach has been conservative—we never included any more than the minimum internal reddening necessary to (1) be con- sistent with observed 2175 A absorption dips. such as those seen in 3C 84. 3C 120. Mrk 9, Mrk 79, Mrk 279, and NGC 7469. or (2) produce enough downward curvature at ultraviolet wavelengths to match the steep drop in fluxes seen in NGC 1068. NGC 3227. 1 Zw 1. and Mrk 376. If the strength of the 2175 A feature varies as much as it does along lines of sight in the Milky Way, any individual reddening esti- mate based on that feature could be wrong by as much as a factor of 2. The internal reddening E,-, = 0.07 is the smallest value which we believe can be detected unambiguously. given IUE’s poor sensitivity at 2100 A. and the strong Fe 1 lines at wavelengths longer than 2200 A. The true internal continuum reddenings are probably at least as large as the values listed in Table 2. and in some cases could be significantly larger. To assess the reality of these internal continuum reddenings. we have compared our estimates with those obtained by other techniques. Simple emission-line-free narrow-band optical ultraviolet color ratios, either fs220 Sy-70 OF fa220 fiaso- Were suggested by Malkan (1984) to measure the redness of the continuum. In luminous objects these ratios are clustered around lower bounds of 1.5 and 1.7, respectively. Malkan (1984) interpreted these ratios as unreddened quasar colors. and suggested that they could be used to set strict upper limits on continuum reddening. (They do not require the presence of the 2175 A absorption feature in the reddening law.) We esti- mated these two intrinsic optical-‘ultraviolet colors. and took their average. corrected for (1) Milky Way reddening. (2) our estimates of internal reddening. and (3) starlight flux at 4220 A (assumed to be 0.38 times the starlight flux at 5450 A). for each of the 21 objects with E,_,. < 0.5. The larger. more uncertain reddenings of the other eight AGNs introduce greater scatter in their ultraviolet colors. We then checked these ratios for our sample of AGNs for consistency with Maikan’s resuits. The average optical ultraviolet flux ratio is 1.5, with an individual scatter of + 0.2. The small scatter of this ratio around the value previously found for more luminous objects indicates that most of our internal continuum reddening estimates in Table ! are not in error by more than 0.1 in E,_,-. independent of the strength of the 2175 A dust feature. The simplest hypothesis. that the internal reddening is described by a standard inter- stellar law. which includes a bump around 2175 A. is consis- tent with all the spectra. except possibly those of NGC 1068 (see MO) and NGC 3227. —~13- No. 1. 1986 The continuum reddenings can also be compared with esti- mates of emission-line reddenings. from the Pax line (Lacy et al. 1982: McAlary ez al. 1986) the He 1! recombination lines (Malkan 1986), and the forbidden lines of [O 1] and [Sn] (Malkan 1983a). In 14 of 18 cases. our continuum reddenings are within the ranges of reddening found by these emission-line techniques. The emission-line reddenings of Mrk 23), Mrk 348, and 3C 120 are somewhat larger than our estimated continuum reddening. while the weak Paz line in Mrk 376 would imply no reddening (compared with our estimate of E,-y = 0.18). Our estimates of foreground plus internal reddening are well correlated with the observed Ha'Hf and Lyx‘Hf line ratios. The correlation between our estimated continuum reddening and the observed Lyx/H~x ratio, signifi- cant at the 99.5". level. is shown in Figure 1. The fitted regres- sion line is consistent with an average intrinsic Lyz,/Hz ratio of 3.2, subjected to the same extinction as the continuum. for a normal reddening law. Thus it is evident that our estimated SPECTRAL ENERGY DISTRIBUTIONS OF AGNs 63 continuum reddenings are also applicable to the emission lines, for most AGNs. Fortunately, the internal reddenings in two-thirds of our program Seyfert ] nuclei do no exceed Eg_, = 0.10. Thus errors in dereddening the SEDs of these objects are expected to be small. The larger dereddening uncertainties expected in the remaining AGNs still principally affect our estimates of the ultraviolet blackbody parameters. Our measures of the other fitting parameters are insensitive to the reddening corrections adopted. d) Model-fitting Results We fitted all the SEDs with an empirical model. with 7 free parameters, the minimum number necessary to reproduce al] the data accurately. The 4 components are (1) a power law. with the strength F,, (the flux at 5450 A). slope a1, and low- frequency exponential cutoff to be fitted: (2) optically thin Balmer continuum with a flux at 3646 A (F;,,) to be fitted: FYT T 7 4 Loe 4 ae a L a ®. 8e 4 ve ] a. 1.0F a .. ‘S ° q Ha ft a : aL 4 tC) a ! \ . O.5r 7 a 4 L 8 LS, z b ‘ a ® 6 4 -1.0+ . . 4 3.4m * ; s \ 25 um r x . 1 L . s 4 a. J “5+ 8 *S B a N L a . 4 L ‘ es 4 Ly ! \ : 0 0.2 0.4 Eg-y Fic. 1—Correlations of continuum reddening with Lyz/H2 and spectral slope from 3 to 25 ym. In the upper correlation, total continuum reddening (internal plus foreground) is plotted: in the lower correlation. internal reddening is the dependent variable. The first correlation indicates that continuum and emission-line reddenings are the same on average. The second shows that reddened, dusty active galactic nuclei have far-infrared fluxes which are significantly contaminated by thermal reradiation. 64 EDELSON AND MALKAN (3) an ultraviolet blackbody with temperature 7,,, and strength Fy, to be fitted: and (4) a S um excess. fitted by a parabola centered at 5.2 ym. with a width of 3.3 octaves FWHM. and strength F,. to be determined. The contribution to the continuum from numerous blended Fe ni lines was also included. with the overall line flux levels scaled to the observed Hx flux using the model of Grandi (1981). These emission-line strengths were set by other sources. and were not allowed to vary in the fits. Errors in the proper Fen strengths would affect primarily our estimates of the Balmer continuum. The three components other than the power law are quite localized. and only contribute significantly to the spectrum over limited frequency ranges of an octave or two. Thus. each component is accurately determined by the data. independent of the description of the other components (which dominate at other wavelengths). The correlation matrices calculated by our nonlinear least-squares fitting program confirm that each of the fitting parameters is relatively independent of the others. The work of MF and others has already indicated that a component of approximately power-law shape dominates the total energy output of AGNs. at least from 0.6 to 10 um. As expected. it is a major component in our fits to the Seyfert 1 SEDs. MS and others have already shown that the blending of higher Balmer series emission lines. merging into strong Balmer continuum emission. produces a small jump from 4000 to 3600 A. They also found that a much larger component. with an approximately thermal shape. is required to explain the strong flux in the blue and ultraviolet. often referred to as the “ultraviolet bump.” It has been adequately fitted by a single Planck function. with a typical temperature of ~ 26.000 K. and includes a sizable fraction of the total energy output of many AGNs. Thus we needed all three components to fit most of our SEDs. The components described above were sufficient to give a fairly good fit to most of the SEDs. but the fits to many objects improved dramatically when we added a bump centered at ~5 um. Also. the SEDs of many objects were found to cut off in the far-infrared. This is the first time that these features have been systematically studied for a large sample. and the results. discussed in §& IV and V. constitute some of the major new findings of this investigation. The results of the fits are stated in Table 2. The object's name is in column (1). the slope of the power law in column (2). the ratio of hot blackbody to power law at 5450 A in column (3). the blackbody temperature in column (4). the ratio of the Balmer continuum to Hz in column (5). and the logarithm of the integrated luminosity between 0.1 and 100 ym in ergs s~? in column (6). Figure 2 shows the data plotted in log (vF,) versus log v, corrected to zero redshift. and fits (represented by solid lines). reddened by the amounts listed in Table 1. Horizontal SEDs in these plots would have equal powers in each octave. Figure 3 is a plot of the data for three objects 3C 273. Mrk 335. and NGC 4151. along with fits. decomposed into the individual components. IV. INFRARED EMISSION a) Determination of the Turnover Frequency The /R.AS data show that the power-law slopes found in the optical and near-infrared (x ~ —1.2: see MF) continue well into the far-infrared. However. the broad-band SEDs of most AGNs generally peak near 80 pm. We find that 16 of the 29 —-14- Vol. 308 objects studied (55° ) have detectable turnovers shortward of 100 um. Most of the remaining objects have 1.4 mm limits which constrain the turnover frequency to occur shortward of ~ 300 pm. We determined the turnover frequency (v,) by fitting a pa- tabola to the 25. 60. and 100 um points in (log F,. log v)-space. and taking the point at which x =0 as the turnover. For objects in which x = 0 lies outside this range. the turnover frequency is considered to be undetected. For the three objects with upper limits at 100 wm (Mrk 335, Mrk 841, and 3C 390.3). the 12. 25. and 60 um points were used. The calcu- lated values of v, are in column (2) of Table 3. This method gave more consistent results than the exponen- tial cutoff used in the fits to the SEDs. Because of the exponen- tial cutoff term, that method tended to assign objects with steep infrared slopes higher turnover frequencies than flatter objects which had x = 0 at the same frequency. The parabolic fits to the far-infrared data are also preferable because they are based solely on JRAS data. so any uncertainties in the JRAS scale factor will not affect the result. During the fitting process (see § III) we tried to model the turnover with a second-order term in the power law (F, x v?" lee’) This gave unacceptable results for almost all objects. The fits tended to be too high in the near-infrared and too low in the optical-ultraviolet. when the turnover itself was well fitted. This is because the functional form used causes the slope of the power law to change too gradually. This is evidence that the power law has a constant slope from at least the optical through the mid-infrared, and it cuts off sharply in the far- infrared. b) Thermal or Nonthermal? In this paper we seek new ways to distinguish between thermal and nonthermal emission in the infrared. We find strong correlations between internal reddening. derived from the many independent estimates in § IIIc. and the slope of the infrared continuum. A steep infrared continuum is indicative of dust emission. while a fiat spectrum indicates that nonthermal emission dominates (de Grijp er al. 1985: Neugebauer er al. 1985). A further confirmation of this trend is that our model, which has an infrared power-law component but no thermal dust emission component. generally fitted objects with flat infrared spectra and low reddening much better than objects with steep spectra and high reddening. These correlations are quite strong. For example, the half of the sample with internal reddenings E,_, < 0.10 has 2,, = — 1.36 + 0.05 (error in the mean), while the half with internal reddenings E,_, 2 0.10 has 25; = — 1.58 + 0.10. The correla- tion can also be seen without power-law fitting. Figure 1 shows the correlation of internal reddening with x34". This correla- tion is significant at the 99% level. In general. we find that most objects with 2324" > —1.2 have infrared spectra domi- nated by nonthermal emission, while most with 2344" < —1.2 have strong dust contamination. The model-independent parameter x33" is a reliable indicator of dust content. c) Nonthermal Emission The good power-law fits to the SEDs of relatively dust-free Seyfert 1 galaxies with 234%" > —1.2. such as Il Zw 136. NGC 3516. 3C 120. 3C 273, Mrk 9, Mrk 79, Mrk 335. and Mrk 509. suggest that their infrared emission is primarily syn- chrotron radiation (see Fig. 1 in Neugebauer. Soifer. and Miley 1985). Under this assumption. the low-frequency turnover can ~15- No. 1. 1986 SPECTRAL ENERGY DISTRIBUTIONS OF AGNs 65 TABLE 2 Move. Fittinc PARAMETERS F,> Tov? log u Source ap? (mJy) Fob /Fpi‘ (K) Bac/Hat (ergs s~*) Mrk 335 1.34 5.5 0.60 25600 3.0 44.80 Mrk 348 -1.75 1.4 <0.14 vee see 43.69 1Zwil ~-1.53 4.6 0.37 19000 0.6 45.55 NGC 1068 —2.19 36 <0.05 vee 1.3 44.86 3C 84 —1.53 9.9 0.12 20000 18 44.83 3C 120 —1.33 4.4 0.16 29000 1.9 44.83 Akn 120 - 1.00 8.1 0.31 24000 2.1 44.89 Mrk 3 —2.58 0.2 <0.10 1.3 44.35 Mrk 6 —1.72 1.6 <0.10 see 17 44.26 Mrk 376 —1.16 3.2 0.13 19000 2.1 45.05 Mrk 9 ~1.45 2.7 0.41 27300 1.8 44.91 Mrk 79 —1.36 2.8 16 31000 2.7 44.50 0906-48 -1.39 0.5 0.71 23600 45.24 NGC 3227) -1.49 7.0 <0.10 ee tee 43.44 NGC 3516 =—1.30 5.9 0.52 22800 2.0 43.97 NGC 4051 —1.39 5.8 <0.15 see 2.4 42.81 NGC 4151 —1.37 30 0.43 24500 2.0 43.67 3C 273 -1.20 9.5 1.0 28200 1.8 46.64 Mrk 231 -1.99 6.6 <0.10 ee 0.5 45.99 1351+64 ~1.70 1.3 2.8 16300 wes 45.53 Mrk 279 1.18 2.8 11 21400 2.2 44.64 NGC 5548 =~—1.23 7.1 0.17 27200 2.8 44.44 Mrk 817 -1.55 3.5 0.68 24800 wee 44.93 Mrk 478 —1.13 2.5 0.24 29000 1.7 45.39 Mrk 841 -1.16 5.5 0.34 24300 wee 44.79 3C 390.3 —1.40 2.7 0.12 32000 2.3 45.00 Il Zw 136 -1.44 2.6 0.50 35500 7 45.29 Mrk 509 -1.09 9.6 0.23 26000 2.0 45.12 NGC 7469 —1.69 9.3 0.42 21400 3.0 44.92 ® ap: is the spectral slope of the fitted power law (S. « v*). b Fp; is the fitted flux of the power law at 5450 A, corrected for reddening. © Fop/Fpi is the ratio of fitted blackbody to power law fluxes at 5450 ¢ Tp is the temperature of fitted ultraviolet blackbody continuum component. © BaC/Ha is the ratio of Balmer continuum flux (integrated assuming an optically thin shape at 13,000 K) to the Ha flux, corrected for reddening. ' log L is the logarithm of total observed luminosity (including emission lines and starlight), integrated from 0.1 to 100 um. be interpreted as the frequency at which the source becomes self-absorbed. On the short-wavelength side. optically thin emission presumably produces the power-law slope. We assume a brightness temperature 7, = 10! K, because it is believed that most self-absorbed synchrotron sources are near this limit (Kellermann and Pauliny-Toth 1981). Cyclotron emission, with a low-frequency self-absorbed cutoff in the far- infrared, can be ruled out. A cyclotron frequency. above which little energy is emitted. as large as ~ 10'* Hz requires magnetic fields of 3 x 10° G. This is unacceptably high. indicating that the source must be radiating by the synchrotron process. which requires relativistic electrons with brightness temperatures above T, = 10!! K. Above the 10!? K limit. inverse Compton radiation would have cooled the source and caused it to emit far more X-rays than we observe.” 2 Independent equipartition calculations suggest that 7, is generally of order 10''-10'? K for these objects. The following formula allows us to estimate the size of the emitting region: [ F.\ where r, is the size of the emitting region. D is the distance to the source, F, is the flux density of the turnover at wavelength 2, and kg is Boltzmann's constant. This quantity is tabulated in column (6) of Table 3. If we assume that the central engine is powered by a black hole. we can derive a lower limit to the Schwarzschild radius and mass by assuming that all the luminosity in the power-law and ultraviolet bump between 0.1 and 100 um is ultimately powered by accretion at the Eddington limit. This is a reason- able assumption for the more powerful AGNs in our study. although the weaker ones are probably sub-Eddington. Thus. 2 —~16- ‘6B PUECLIS HIN ‘8b + 9060 Dd “RPSS JOON OT) HAV 'OLZ UW OL IN SNOV 4) Joy (9) se aes (y) “xny UL oduRYys Maa adauy ayy aty pesuOZuOY ese yeadde pynom (aperadp dad samod yenba yw) | — adoys yo mez somod v yeYD os “44 Jo WY WeTO] ay) si aRos [LINUIA “(UONEIpRs ss “Putuoppas soy suonaaisos duipnpuy Apo youyg jojoiaeayjn pue “duing pasesgui-svau “me Jamod & YIM sppour BuNIy-189q BY) MOYS Sout, ayy “Sseq 4052 [HINA Yim UmoYs Juv saxny UNAUIUOS pautty PIAssgO ‘(ste HIN PUB TP8 WIN “60S HW “BLP WIN “EL DE “VET ®Z 1) SNOV XS Jo SCS 1folseaN/jeotNdo/paseagul ayy OF SILA (VF “7 “DLA TZ Jo sonavy B smoys avg uoreuqusoros pue sya Came | oe Old (ZH) Aouanbas4 (ZH) Aouanbas4 «0! (0! 2,0 «0! pol <0! 201 2x Doce wo zx| Lg 6 NIN 1 / Ib NW S 218 WW etl ~ . oO an 2 60S NIN] OS \ Bb +9060 9d} o g | oe e = apss oon] 3 SS DON ho ab NW > oz NAV 4 ¢e2z2 2o¢ | 9€1 MZ I \ 622 NAW MM a 62 NXW * * * 4 i 4 1 66 -17- TMZ EPYROOPL ON “1S0P ON ALTE ODN = PPR OU RPE OIA “9 OIA, SNOV 24 20g (Y) SB OWES (PE O6E OE PUR CULE HIN VISE DON ISTP JON ‘OZE DE “99 + ISEL Od SNOV 242 JOy (P) SB aWILS (2) —C 4H) 7 “OLY ae | “Oy (ZH) Aduanbas4 (2H) Aduanbas4 gO! pO! ¢Ol 2,0! giOl Ol gO! 20! TT T T T T ~F t rrr \ zx| | €06e IE 92 NW [ e2re 20N A NSE DON s q ra x isov on |<” } _ a oO g P ° 3 ™~ Sib DON] <P veo | 3 J. ssi ~ i} a LY wn abe NX O21 9¢ 9 NW / pItIGe! Od i 1 i 4 ioowr 67 68 EDELSON AND MALKAN —18- Vol. 308 MKN 3 MKN 231 NGC 1068 | NGC 3227! VF, (erg/cm?/s) | T T 10% mE 10% io Frequency (Hz) Fic. 2(cont.)—4e) Same as (a). for the AGNs Mrk 3, Mrk 231. NGC 1068, and NGC 3227 this will yield a lower limit on the mass and radius of the central black hole. The mass. m,,. and the radius. ry,. are tabulated in columns (7) and (8). respectively, of Table 3. We use the following relations (Rybicki and Lightman 1979): _ bey 4nGM yc" where Ly is the luminosity in the power law and ultraviolet blackbody. a; is the Thomson cross section. and M,, is the mass of hydrogen. and Men 2Gmp,, '%, = - bh rs These calculations show that black hole masses of at least 2 x 10° to 400 x 10° Me are required for ihe cenirai objecis. The minimum radii range from 2 x 10)? to 4 x 10!* cm. cor- responding to light-travel times of minutes to hours. The syn- chrotron sources have sizes ranging from 10'* to 3 x 10'S cm. The synchrotron radii are typically 10-100 times the minimum Schwarzschild radii, suggesting that the synchrotron emission is produced in a region comparable to the inner parts of the hypothesized accretion disk. The possibility that the turnover could be due to free-free absorption from broad-line clouds can be rejected for most objects in this sample. EX OSAT measurements or upper limits of the hydrogen column density place the free-free absorption frequency of these objects in the millimeter region. Free-free absorption from either the narrow-line region or the hot inter- cloud medium hypoihesized to confine the broad-line clouds is not important in the far-infrared; their low electron densities would put the cutoff frequency at radio wavelengths. The Razin effect is unlikely to explain the turnover, since it would require densities which are too Jow and magnetic fields which are too high. d) Thermal Dust Emission Our fits to the SEDs of the three Seyfert 2 galaxies (NGC 1068, Mrk 3, and Mrk 348) and other dusty objects (3C 84, Mrk 231, NGC 3227, and NGC 7469) confirm the findings of ~19- No. 1. 1986 SPECTRAL ENERGY DISTRIBUTIONS OF AGNs 69 ich io'eE VF, (erg/cm?/s) Tom u emg as IO 7a ii4 Frequency (Hz) Fic. 3.—Detail of model fits to SEDs of 3C 273. Mrk 335. and NGC 4141. The individual components shown are the infrared power law (dashed line. which dominates the far-infrared). the near-infrared bump (dotted line: not detected in NGC 4151). the galactic starlight (dot-dash line: not detectable in 3C 273). the recombination continuum (solid line—only the Balmer continuum. and a little of the Paschen continuum are strong enough to show in the figures). and the ultraviolet black body (dot-dot-dash line). All components have been reddened by the galactic and internal reddening estimates given in Table 1. The sum of these components equals the fitted model. shown by the upper solid line. MO: No power law of any slope. with or without an exponen- tial low-frequency turnover. can reproduce their nonstellar spectra. Nor can they be fitted with the inclusion of a quadratic term to the power law. The observed SEDs show too much downward curvature to be accurately described by the power- law fits. a property of Seyfert 2 galaxies previously noted by Miley, Neugebauer, and Soifer (1985). Thus our fits tend to overpredict the 2-3.5 ym flux because the Seyfert 2 nonstellar light curves down very sharply from the mid- to the near- infrared. Also, the local slopes such as 2334" tend to disagree significantly with the fitted power-law slope. Finally, the fits, which require very steep power-law slopes to fit the mid- infrared Seyfert 2 fluxes. underpredict the optical continuum. where the nonstellar flux evidently has a flatter slope. These infrared spectra are easily understood as thermal infrared emission from dust at a range of temperatures. Although each temperature component produces a thermal spectrum with an exponentially steep high-frequency cutoff. the range of temperatures smears it into the observed steep mid-infrared spectrum. The thermal Wien cutoff causes the integrated spectrum to curve downward steeply to higher fre- quencies, quite unlike a power law. The shape of the weak nonstellar component. detectable only at wavelengths short- ward of 10 ym. is poorly determined, but could be similar to the nuclear components of unreddened Seyfert 1 galaxies and quasars (see MO). If we assume that the infrared emission is dominated by dust with an emissivity € x 1/4, the mid-infrared radiation arises from grains with a wide range of temperatures. The turnover is then taken as the peak of a thermal spectrum, and the minimum temperature of the dust is derived. If we assume that all the luminosity between 10 and 100 um is coming from this component, we can derive the minimum size and mass of the thermal emitting region. The minimum size is given by 1 [L, "4 Ti V 4x0 ° where r, is the radius of the emitting region. L, is the’ lumi- nosity between 10 and 100 ym. T, is the dust temperature. and 70 EDELSON AND MALKAN — 20 - Vol. 308 TABLE 3 DERIVED TURNOVER AND Bump PARAMETERS vy T,> ra° ma? r,* men! ron® te” Source (THz) (K (10?8 em) (Mo) (10!2 cm) (10° Mo) (10! em) (%) SL Mrk 335 8.6 105 47 26 117 6.3 1.9 44 0.131 Mrk 348 < 3.0 vee vee vee vee ee bee < 10 —0.061 1Zwil 3.7 44 300 2400 1500 35 10 25 0.068 NGC 1068 3.0 37 690 26 1000 7.3 2.2 < 10 —0.102 3C 84 4.1 50 72 170 800 6.8 2.0 < 10 —0.098 3C 120 < 3.0 see ee ee os see aoe 15 0.035 Akn 120 < 3.0 vee see see see see aoe 23 vee Mrk 3 6.3 76 41 26 230 2.3 0.7 < 10 —0.397 Mrk 6 5.7 70 34 49 230 1.8 0.5 12 -0.097 Mrk 376 < 3.0 see vee aes see eee eee 32 0.096 Mrk 9 3.3 40 200 1300 760 8.1 2.4 23 0.085 Mrk 79 < 3.0 wee see . 42 0.052 0906+48 < 3.0 . 28 0.066 NGC 3227 < 3.0 < 10 -0.173 NGC 3516 < 3.0 < 10 —0.142 NGC 4051 < 3.0 < 10 -0.157 NGC 4151 < 3.0 see ee see vee eee see < 10 —0.014 3C 273 3.7 45 2100 16000 4200 430 130 36 0.112 Mrk 231 5.1 62 600 570 3200 99 30 22 0.089 1351+64 4.0 48 340 3700 1400 34 10 < 10 ~0.063 Mrk 279 < 3.0 see see ee : < 10 ~—0.080 NGC 5548 < 3.0 see see wee see wee see 15 0.070 Mrk 817 5.0 61 96 260 600 8.4 2.5 < 10 wee Mrk 478 < 3.0 tee ee see vee ee wee 43 0.069 Mrk 841 7.3 88 67 90 210 6.1 1.8 < 10 ae 3C 390.3 8.8 110 31 120 230 10 2.9 27 0.100 Il Zw 136 5.8 71 180 540 550 20 5.7 25 0.113 Mrk 509 4.7 57 170 370 550 13 4.0 33 0.070 NGC 7469 4.0 48 98 180 1200 9.8 2.9 < 10 ~0.057 ® vy, is the frequency of far-infrared continuum turnover, where a parabolic ft to the JRAS fluxes has a =0. > Ty is the minimum temperature of dust grains (assuming graybody radiation) required to produce low-frequency turnover. © rg is the minimum radius of an optically thick sphere of dust at Tg required to produce far-infrared peak luminosity. 4 mg is the mass of dust required to produce far-infrared ‘peak luminosity. © r, is the radius of synchrotron-emitting region required to produce turnover in far-infrared spectrum by self-absorption, assuming a brightness temperature of 10}? K. ! my, is the mass of a black hole required to produce total nuclear luminosity by accretion at the Eddington limit. © vy, is the Schwarzschild radius for black hole mass mpn. b fo is the fitted fraction of 2.5 to 10 um luminosity in the near-infrared bump component. ' AL is the near-infrared excess at 3.5 um, measured logarithmically, over a linear interpolation between 2 and 10 um. o is the Stefan-Boltzmann constant. These temperatures are tabulated in column (3) of Table 3. The minimum mass is given by __DF, ™a = BIT)’ where m, is the mass of dust, D is the distance to the object, F, is the observed flux density at the turnover. x is the mass absorption coefficient. and B,(T) is the Planck function at the turnover. We use x = 250 cm? g7! (Draine and Lee 1984). These quantities are tabulated in columns (4) and (5) of Table 3. respectively. These calculations yield typical dust temperatures of 40-80 K for those objects whose far-infrared emission is likely to be dominated by dust. This is warmer than dust in normal galaxies, whose SEDs rise steeply with color temperatures ~ 30 K (de Jong er al. 1984) between 60 and 100 um. The high dust temperature suggests that much of the dust emission is probably not coming from beyond the narrow-line region. The size of the emitting region was derived under the assumption that it was optically thick: an optically thin region would be larger. We found minimum sizes of order 100 pc. This is beyond the broad-line region (which is not observed in many of the dusty objects), but well within the narrow-line region. We found typical dust masses of order 100 Af. which imply gas masses of 10*-10° Mo, for an assumed gas‘dust ratio of m,/m, = 200 in this region. These figures are not unreasonable . for the narrow-line region (Baldwin 1975). —~21- No. 1. 1986 V. FIVE MICRON EXCESS The fits in § III revealed the presence of an excess at about S$ um in about half of the objects studied. Apparently this feature has been previously observed. although not systemati- cally studied. Previous investigators (Neugebauer er al. 1979: MF: Malkan 1983h). have called it the 3 um bump. although we find that the central wavelength is closer to 5 um. The low contrast of the bump relative to the power law and the lack of uniform moderate-resolution infrared spectral coverage for most objects made it difficult to determine its precise shape. We modeled it with a parabola in log F, versus log v, given by log, mal log, F,(v) = log, Fylvo) — [ (4/2) where F,(1) is the flux due to the bump at frequency v. vo is the center frequency. and a, is the FWHM in octaves. The base 2 logarithm is referred to by log,. Besides the parabola, we tried to fit the bump with a Gaussian, which has a low-frequency tail relative to the parabola. As would be expected. the peak fre- quencies determined with the Gaussian were systematically a little higher than those determined with the parabola (by about 0.1 decade). but the widths were very similar. The fits were equally good in the two cases. We also tried to fit the SEDs by replacing the single power law plus bump model with a broken power law which steepened above some fitted break frequency (typically around 90 THz). None of the SEDs with a prominent 5 ym bump could be fitted even crudely with this scheme. The broken power law. even with a spectral steepening of Az > 1, still failed to produce a sharp enough inflection. The fits to the SEDs with the strongest bumps required Ax ~ 1.7, yet the data require a localized excess flux superposed on a power law with relatively constant slope from the visual to the far-infrared. We used the parabolic representation in our final fits. Allow- ing the center frequency to vary while the width was held constant. we found a mean center frequency of 58 THz (5.2 wm). with an individual dispersion of 11 THz. Although the bump strength relative to the underlying power law (ie.. its contrast) is highest near the L (3.5 ym) filter, the peak bump flux actually occurs at a significantly longer wavelength. This is because the bump feature.is added to a power law which increases sharply toward longer wavelengths. Likewise. holding the center frequency fixed while allowing the width to vary in the fitting process, we found a mean width of 3.3 octaves, with a dispersion of 0.3 octaves for individual objects. The dispersions in the width and central frequency were small. little larger than their individual fitting uncer- tainties. These results are consistent with a single universal shape for the 5 um bump. which varies only in overall strength. Therefore, we fixed the central frequency and width of the 5 um feature at 58 THz and 3.3 octaves. respectively, and allowed only the strength of the bump to vary in the fits. We define the bump strength (f,) to be the ratio of the lumi- nosity in the bump to the total integrated luminosity. defined between 2.5 and 10 ym. This parameter is tabulated in column (9) of Table 3. The bump contributes up to 40% of the total luminosity between 2.5 and 10 pm. We could also parameterize the bump in a model- independent fashion, using only the observed H, K, L, and N fluxes. We define the near-infrared excess, AL. as the logarith- mic excess at L over an interpolation between the average of SPECTRAL ENERGY DISTRIBUTIONS OF AGNs 71 the H and K points and the N point. This parameter is tabulat- ed in column (10) of Table 3. It is apparent that the two param- eterizations of the bump strength are very well correlated. For instance. every object except one with AL < 0 lacks a detect- able bump in the fits (i.e. f, < 10%), and no object with a detected bump has a negative value of AL. A pure power law will produce a value of AL = 0. but the presence of starlight will produce an excess at H and K. so that an object with stars plus a power law. but no 5 um bump. will actually have a deficit at L. To confirm this. we determined AL for seven star- burst galaxies (M82. NGC 253, NGC 1614, NGC 2903. NGC 5253. NGC 6764. and NGC 6946). We found a mean AL = —0.273. with a dispersion of 0.086 for individual objects. This suggests that. while most of the AGNs studied have a 5 um bump. there is a bias which makes it appear weaker and possibly undetectable in objects in which starlight is impor- tant. Using the rank correlation test described in § 1. we found that, is strongly anticorrelated with [O I1]] ‘Hf and is strong- ly correlated with luminosity. These correlations. shown in Figure 4, are significant at the 99.5% confidence level. The bump strength was not significantly correlated with any other parameters. The poor correlation observed between f, and F,,, (ultraviolet blackbody strength) suggests that the 5 um bump does not arise directly around the continuum source. The well-known relationship between luminosity and [(O 1] Hf makes it difficult to decide which parameter is most directly linked with the 5 um bump. The strong correlations among these parameters mean that the correlation of one of them with f, could give rise to an incidental correlation between f, and the other one. For the objects with clearly detected 5 um bumps and Balmer continua. the observed total luminosity in recombi- nation radiation (Balmer. Paschen. and free-free continua plus Balmer. Paschen. and Fe 1 emission lines) was 0.5-1.3 times the luminosity in the bump. With a typical reddening correc- tion to the recombination emission at shorter wavelengths. the bump is actually weaker than the emission from the broad-line Tegion. This approximate coincidence of powers suggests that there may be enough energy in the broad-line region to produce the 5 ym bump. if it is done with reasonable efficiency. Since the bump is. on average, less than half as strong as the observed blackbody flux (integrated down to 0.12 ym). and less than a quarter of the total inferred blackbody flux, it would represent a rather small perturbation on the total nuclear energy budget. Puetter and Hubbard (1985) suggest that free-free emission from very dense broad-line clouds (n, = 10'! cm~?) could produce an excess similar in strength. width and central fre- quency to that observed. Their model implies that the objects with strong bumps will have depressed Pax/Hf ratios. since the very high densities will make even the Pax optical depths iarge. Yet we found no correlation between Paz'Hx and /, for the 15 objects in our sample with published Pax ‘Hz ratios (see Fig. 4). The bump strength, f,. was also uncorrelated with the Hz’Hf and Pax Hf ratios. VI. ULTRAVIOLET COMPONENTS a) Ultraviolet Blackbody Component We found no case where the ultraviolet blackbody com- ponent was convincingly detected at a temperature markedly different from the mean of 26,000 K. The 1 o scatter of individ- ~22- 72 EDELSON AND MALKAN rv lo) iN “Nl 10*°5 ° -F Fee @ aaa I Ww 45, oe I LQ 10 rg ere © 10** ba * 10°F, E Fo T T T ty r | 10 3 [om] fe , a a HB oe S 2 ° 4 1 5, “a z Ls BN 8 e J b— ‘a = 8 sy 0.1 E a e “<3 an pt L - 0.3 + 7 i 8 - Pa # Ha ea a s, 0.1 as [a a Z L s 8 4 ee ! cl A 1 t 0.1 0.3 5 wm Excess (f;,) Fig. 4-— lations of the 5 wm bump sirengih with three other param- eters. In the top box the abscissa is total bolometric luminosity. integrated from 0.) to 100 ym. In the middle bos it is the [O in] Hf emission-line ratio. and in the lower box it is the Paz Hf emission-line ratio. The lines are proper least-squares fits to the points. The top two correlations are highly significant (at the 99.5". level). while the lower one is not. r 4-—C, c Vol. 308 ual estimates is 4000 K, a little larger than the fitting uncer- tainties. A few objects (e.g. 1 Zwil, PG 1351464. and Mrk 376) may appear to have lower temperatures because their internal (nuclear) continuum reddenings were underesti- mated. However, as noted in MS and Malkan (1983h). this thermal] component is probably broader than a single black- body. and the fitted single temperature may not have a simple direct physical interpretation. The fitted blackbody tem- perature was not correlated with any other property of the AGNs. but its range is relatively narrow compared with the fitting uncertainties. A useful measure of the blueness of the continuum. used in MS. is the ratio of fitted blackbody to power-law fluxes at 5450 A (Fyp/F i). given in column (4) of Table 2. The ratios span the same range found in MS and Malkan (1983h). The object 3C 273 is at the high end. with a ratio of 1.0. This is not unique: some QSOs (e.g.. PKS 1004+13, Malkan 1984: 4C 31.30, 4C 37.43. and 4C 34.47, Richstone and Schmidt 1980) have even stronger “blue bumps.” The average blackbody power law ratio at 5450 A is 0.3 in those AGNs where the blackbody is detected. In eight AGNs. including all three Seyfert 2 galaxies. there is no detectable blackbody com- ponent, and a typical upper limit on its strength is Fy, Fy, < 0.1. These eight objects also have some of the strongest con- tamination from starlight and show evidence of internal reddening and dust. These may have prevented us from detect- ing a weak blackbody component which might be present at the same strength as in some of the other Seyfert 1 galaxies. Our fitting procedure unambiguously detects the black body components in NGC 4151 and NGC 5548. which are among the weakest blackbodies detected. They were missed by MS because their fits did not correct for the presence of starlight flux. which noticeably affects the observed spectrum. even in the blue. The data demonstrate that the blackbody power law ratio is not tightly correlated with luminosity. MS had sug- gested such a correlation, but Malkan (19835) found that it did not extend to high luminosities. b) Balmer Continuum Although the Balmer continuum flux around the Balmer jump (~ 3650 A) is well determined. the shape of the high- frequency falloff is not. For comparison with previous esti- mates. we assumed that the Balmer continuum has an exponential falloff (i.e. it is not very optically thick) with a temperature of 13.000 K. The average Balmer continuum Hx ratio in the 19 Seyfert 1 galaxies where it is well measured is 2.1 + 0.6 (individual scatter). very similar to the values found in MS and for high-luminosity quasars by Malkan (19836). The ratios are not determined with high precision, so it may not be surprising that they do not correlate with other AGN parameters. Comparisons of the Seyfert 1 galaxies with the strongest Balmer continuum’H2 ratios (Mrk 79. Mrk 335. NGC 5548. and NGC 7469) with those with the weakest ratios (I Zw 1, Hi Zw 136, 3C 273, Mrk 9, and Mrk 478) do not eveal any other significant distinctions between these two groups. Future models of emission from the broad-line region must account for the stability of the Balmer continuum strength under widely varying conditions. Previous modeis (e.g. Kwan and Krolik 198i) predicted Balmer continuum/Hz ratios near 1.4, the value for case B recombination. They do not necessarily conflict with our observations. because they indicate that substantial Balmer continuum emission arises from regions with electron tem- — 23 - No. 1. 1986 peratures of only 8000 K. Thus our assumption of T, = 13.000 K in deriving integrated Balmer continuum fluxes may be too high. because the derived Balmer continuum flux is pro- portional to T,. On the other hand. the Balmer continuum may be optically thick. in which case it would extend to higher frequencies than in the optically thin case. The models of Puetter and Hubbard (1985) tend to overpredict Balmer con- linuum emission, except at low electron density or optical] depth. The Balmer continuum, Hz ratios for NGC 1068 and Mrk 3 are significantly lower than those of the Seyfert 1 gal- axies. and are consistent with case B recombination. which may adequately describe the hydrogen emission from the narrow-line region (MO). VH. RELATION TO OTHER FREQUENCIES a) X-Ray Luminosity MS and Malkan (1984) found a tight linear correlation between the power-law flux measured in the red and infrared and the X-ray flux at 2 keV. with an infrared-to-X-ray slope %x = —1.18 + 0.02. Thus, an extrapolation of the non- stellar flux measured at 3.5 um with a power law of slope —1.18 predicts the X-ray flux to an accuracy of better than + 70%. The present SEDs also show the same strong correlation. Figure 5 shows the correlation of fitted power-law luminosity at 5450 A with 2 keV luminosity measured by the Einstein and HEAO / satellites. The regression line in Figure 5 shows that the X-ray luminosity is linearly proportional to the power-law luminosity. The small scatter indicates that the X-ray lumi- nosity can be predicted from the power-law fits to within a factor of 2. The true scatter may be even smaller. since the X-ray luminosities used in the correlation were measured years apart from the long-wavelength observations. The same direct proportionality is present when the power-law and X-ray fluxes. rather than luminosities, are correlated. SPECTRAL ENERGY DISTRIBUTIONS OF AGNs 73 The regression line in Figure 5 implies a power-law slope from the visual to X-rays of —1.15 + 0.02 (error in the mean). Since this slope is so similar to the typical power-law slope in the optical and near-infrared, the good correlation shown in Figure 3 is still present when the power-law luminosity is mea- sured at other wavelengths. The power law-to—X-ray slope would also be close to — 1.15. This slope is slightly flatter than the — 1.18 slope between 3.5 um and 2 keV because the 3.5 ym luminosity includes a contribution from the 5 ym bump in many AGNs. The small scatter in the correlation. its linearity over a wide range of luminosities. and the similarity of the infrared optical and optical X-ray slopes support the hypothe- sis that the power law and the X-rays have a common non- thermal origin in the dust-free Seyfert 1 galaxies and quasars. The Seyfert 2 galaxies, indicated by triangles. are systemati- cally deficient in X-rays by a factor of several. at a given optical power-law luminosity. One quasar, PG 1351+640. has unusually weak X-ray emission, 10 times too weak for its optical power-law luminosity. b) Radio Emission Five of the AGNs have a strong. compact. flat-spectrum radio source (3C 84. 3C 120, 3C 273, 3C 390.3. and Mrk 231). Such radio sources are frequently associated with extremely rapid (i.e., night-to-night) optical variability and high polar- ization (3% -20%2). This is characteristic of the presence of a “blazar” continuum component which is believed to be non- thermal synchrotron emission and may be beamed along our line of sight (Malkan and Moore 1986). The quasar 3C 273 is known to vary by nearly a factor of 2 in the far-infrared (Robson er al. 1983). This highly variable red component has been identified as a “ blazar” continuum (Impey and Malkan 1987). and could also be present in the other strong. flat- spectrum radio sources. a L Y4 45 |. > 4 10 3 7 E . 4 q r y q ~_— - y 4 wn 44) 4 | L 10 - . q 1 i a ye ; ay e ~~ 43 B — E ef E > 10 E 7 3s F 2 E a 7” ; a) L 8 B 4 Teas eé . E a q L Sy 4 b 7 a 4 411 7 | lr e + 3 b 7 3 a pol J pol a poe! 4 pol _).. 10% 107? | o*4 10% L,, (erg /s) Fic. 5.—Correlation of 2 keV X-ray luminosity with fitted power-law luminosity at 5450 A. The two open circles indicate the luminosities of the two Seyfert 2 galaxies. which were not used in the regression. The Seyfert 1 with unusually weak luminosity is PG 1351 + 640, discussed in the text. —~ 24 - 74 EDELSON AND MALKAN The SEDs of radio-loud AGNs with a suspected “ blazar™ continuum do not stand out as being unusual. This is probably because the shape of the blazar continuum component is _ similar to that of the power law. which is also present in normal AGNs (see discussion in Malkan and Moore 1986). The objects 3C 84. 3C 120. and 3C 390.3 have relatively low blackbody power law ratios, which could be consistent with the idea that their infrared flux is enhanced by a (beamed?) blazar component. On the other hand. since they were selected for radio emission rather than ultraviolet excess. they could simply be more reddened than most other AGNs in our sample. VII. CONCLUSIONS We combined data from JRAS and IUE with ground-based optical and infrared spectrophotometry to drive line-free spec- tral energy distributions for 29 AGNs between 0.1 and 100 pm. Although the contaminating effects of reddening and emission from stars and dust in the underlying galaxy are significant in some spectra. they can usually be adequately corrected. We find several good indicators of nuclear dust in active galaxies. all of which are well correlated: (1) steepness of infrared spec- tral index: (2) reddening derived from the shape of the ultra- violet continuum: (3) strength of the 2200 A dust absorption feature: (4) line ratios. such as Hx Hf and Lyx Hf: and (5) line-free continuum colors, such as f4220 fy-2q and Ja220‘fiaso. We used these indicators to determine the degree to which the SEDs of our program AGNs have been altered by absorption and thermal reradiation by dust grains. We find that dust does not play a major role in most of the Seyfert 1 galaxies and quasars in this sample. The SEDs of these objects are generally well described by a simple model consisting of four spectral components. each of which domi- nates over a different wavelength region. The first. which usually contains most of the total luminosity between 0.1 and 100 um. is a power law (F, x v7). with 2,, = 1.36 + 0.21 (rms scatter for individual objects). This power law usually has a sharp low-frequency cutoff before 100 um. and the SEDs of most of the remaining objects are constrained to turn over by ~ 300 um. The simplest explanation is that the infrared lumi- nosity is dominated by synchrotron emission from a self- absorbed source which is about the size of the inner accretion disk. There is a strong correlation between the luminosity from this power law at 5450 A and the 2 keV X-ray luminosity. which suggests that the power law may extend from X-ray to far-infrared wavelengths (54 decades of frequency). The second component. which was detected in over half the SEDs, was a broad near-infrared excess which was described by a parabola centered at 5.2 um. with a FWHM of 3.3 fre- quency octaves. Although it contains less total energy than the Balmer continuum, it can account for up to 40°% of the total luminosity between 2.5 and 10 ym. Its strength is well corre- lated with total luminosity and Hf’[O 1] ratio. The far-ultraviolet is dominated by a component which we described as a blackbody at 26,000 + 4000 K (individual scatter). The fitted Balmer continuum’Hx ratio for the Seyfert 1 galaxies is 2.1 + 0.6, under the assumption that the emission is optically thin at 7, = 13,000 K. This ratio is signifi- cantly greater than the prediction of case B recombination. and shows no correlations with any other AGN properties. Seyfert 2 galaxies have a lower ratio than Seyfert 1's. consis- tent with the case B predictions. Although the power-law model fitted the relatively dust-free objects well, it failed to reproduce the SEDs of Seyfert 2 and Seyfert 1 galaxies with strong dust indicators (i.e.. NGC 3227. NGC 7469, and Mrk 231) because their fluxes beyond 10 um are significantly contaminated by thermal emission from dust. Objects with dusty nuclei had minimum dust temperatures of 40-80 K, which is warmer than dust in normal spiral galaxies. The thermal emission appears to arise from 100 Mg of dust in a volume similar to that of the narrow-line region. On the other hand. less luminous active galaxies with 100 um fluxes dominated by dust emission from the underlying galactic disk showed large differences between small-aperture ground-based 10 um and large-aperture /RAS 12 ym fluxes, and steep spec- tral indices between 60 and 100 ym. The authors would like to thank A. Moffet, A. Readhead. R. Blandford, and S. Phinney for fruitful discussions, and C. Lawrence and S. Blatt for help with the manuscript. A. Readhead, W. Sargent, and G. de Bruyn generously pro- vided spectrophotometry of three AGNs in advance of pub- lication. W. Rice of the Infrared Processing and Analysis Center aided in the understanding of the IRAS data. R. Thompson and K. Feggins provided valuable assistance at IUE’s Regional Data Analysis Facility at NASA ‘Goddard Space Flight Center. where ultraviolet spectra from the National Space Sciences Data Center were analyzed. This research was partially supported by National Aeronautics and Space Administration grants NAG 5-427 and NAS 7-918. APPENDIX NEW OBSERVATIONS We present new data gathered on the AGNSs studied in this paper. These data consist of broad-band near-infrared photometry for seven of these AGNs. and optical spectrophotometry for two. Far-infrared photometry from /RAS is also tabulated. Photometry from / RAS is presented in Table 4. For objects brighter than 0.5 Jy at 12 ym. the far-infrared fluxes were taken from the RAS Point Source Catalog (Beichman er al. 1984). Fluxes for faint objects in the 7RAS Point Source Catalog are biased because only observations in which a source was detected were used to measure the fluxes. This will tend to make objects with low signal-to-noise ratios appear brighter than they actually are. For this reason. fluxes of objects fainter than 0.5 Jy at 12 um were determined from source extractions performed on co-added survey data. Data for one source (PG 0906+ 48) were taken from pointed AOs as discussed in Neugebauer e7 al. (1984). No spectral corrections to the [RAS data were necessary, because these objects had power-law slopes near —1. At 12, 25, and 60 um, the thermal noise was about 25 mJy with a calibration error of about 10", while at 100 ym, the errors were about 75 mJy and 15%. —~95- TABLE 4 SaTELLITE Data Siz um $25 um Seo um $100 um Source (mJy)® (mJy)® (mJy)® ~ (mJy)' ~— Data? Mrk 335 340 488 413 < 323 2 Mrk 348 274 743 1290 1620 2 1Zw 539 1260 2130 2430 1 NGC 1068 38300 86800 186000 239000 1 3C 84 1030 3630 7100 7760 1 3C 120 283 673 1300 2560 2 Akn 120 292 427 703 1130 2 Mrk 3 700 2860 3880 3350 1 Mrk 6 214 634 1190 994 2 Mrk 376 241 576 841 1330 2 Mrk 9 228 524 929 1110 2 Mrk 79 297 715 1450 2280 2 0906+48 40 89 190 340 3 NGC 3227 669 1750 7840 16900 1 NGC 3516 383 929 1730 2160 2 NGC 4051 779 1420 8160 20400 1 NGC 4151 2160 4810 6550 7950 1 3C 273 539 929 2180 2800 1 Mrk 231 1820 8560 33300 30100 1 1351+64 155 481 838 943 2 Mrk 279 198 289 1080 1970 2 NGC 5548 342 764 1110 1790 2 Mrk 817 356 1230 2340 2260 2 Mrk 478 139 175 561 857 2 Mrk 841 197 453 475 < 257 2 3C 390.3 130 306 260 < 231 2 I Zw 136 136 383 490 462 2 Mrk 509 366 701 1470 1490 2 NGC 7469 1300 5500 26700 34400 1 ® Typical 12 wm statistical errors are 25 mJy with a 10% calibration uncertainty. > Typical 25 and 60 um statistical errors are 50 mJy with a 10% calibration uncertainty. © Typical 100 ym statistical errors are 75 mJy with a 15% calibration uncertainty. @ Source of IRAS flux densities: 1) JRAS Point Source Catalog; 2) co-addition of raw survey data; 3) Neugebauer et al. 1984. TABLE 5 New INFRARED MAGNITUDES Source J (1.2 um) H (1.6 wm) K (2.2mm) L (3.5 um) Aperture Date Telescope® Mrk 335 12.40 11.45 10.32 8.79 12” 1984 Aug 29 L61 I Zw i> 12.18 11.20 10.10 8.36 27 1980 Ang 20 cé60 Akn 120 11.71 10.89 9.95 see 8 1984 Feb 25 161 11.4 10.55 9.85 8.60 12 1984 Sep 14 L61 3C 273 11.63 10.84 9.70 8.36 11 1980 Jul 1 100 Mrk 817 11.70 10.98 10.24 eee 5 1981 M60 Il Zw 136 12.84 11.90 10.74 9.30 12 1984 Aug 30 Lé1 Mrk 509 11.72 10.77 9.80 8.64 12 1984 Aug 30 L161 ® The following telescopes were used to collect the infrared data: L61 = Mount Lemmon 61 inch; M60 = Mount Wilson 60 inch; C60 = Cerro Tololo Inter-American Observatory 60 inch; 100 = Las Campanas DuPont 100 inch. > Flux used in fits was corrected to 10” aperture using multiaperture data. —~ 26 - 76 EDELSON AND MALKAN Vol. 308 TABLE 6 New Optical MEASUREMENTS A. Continuum Fluxes F, (mJy) A. (A) Mkn 817 Mkn 841 4200 ee 6.4 4600 6.6 6.8 4800 7.3 6.9 5000 wee 6.6 5200 7.2 6.8 5400 7.7 7.0 5600 7.2 68 5800 7.5 7.0 6000 8.2 1A 6200 8.1 17 6400 8.4 7.6 6600 we 9.2 7000 8.0 8.4 B. Emission Line Fluxes Emission Flux (10774 ergs cm? s7*) Line Mkn 817 Mkn 841 HE wee 12 Fe Il 44570 15 14 Hy 17 26 HS 46 56 10 INT} 45007 20 34 He I A5876 6 16 Ha 190 257 [s 1 Ln 7 The co-added fluxes for 3C 390.3 disagree with some of the results of pointed AOs in Miley er al. (1984). Their data show an unusual /R.4S spectrum. with a peak at 25 pm and another at 100 um. Our data are consistent with theirs at 12. 25. and 60 ym. but we see no evidence for the unusually high 100 um flux. We find a 3 o upper limit of 230 mJy at 100 um, compared with their value of 340 + 60 mJy. Since their data were reduced very early in the mission. it is possible that confusion with infrared cirrus could have corrupted their measurements at 100 um. The near-infrared data were obtained at the Steward Observatory Mount Lemmon 6] inch (1.55 m), the Mount Wilson 60 inch (1.5 m), the Cerro Tololo 60 inch (1.5 m). and the Las Campanas Dupont 100 inch (2.54 m) telescopes (see references to Table 5). The usual J (1.2 wm). H (1.6 pm). K (2.2 um). and L (3.5 wm) filters were used. with liquid nitrogen cooled InSb photometers. Standard stars from Elias er al. (1982) were measured throughout each night. yielding photometric scales good to 0.03-0.04 mag. The photometry. along with aperture diameters, observing dates. and telescopes, are listed in Table 5. The optical spectrophotometry was done with the Palomar 60-inch (1.5 m) telescope. The SIT spectrograph was used. with a resolution of 20 A and a 10” slit. We combined spectra taken in the spring of 1984 to produce the averaged spectra. which were calibrated relative to BD +26 2606. We measured the emission-line fluxes. which are listed in Table 6B. We then removed these lines. and averaged the remaining points in 200 A wide continuum bins. These continuum fluxes are tabulated in Table 6A. We estimate calibration uncertainties of 10°%o for the continuum fluxes and 15% for the line fluxes. REFERENCES Aitken. D. K.. Roche. P. F.. and Phillips. M.M. 1981, M.N.R.A.S.. 196. 101P. Baldwin. J. 1975. Ap. J.. 201. 26 Balzano. V. A..and Weedman. D. W. 1981, Ap. J., 243, 756. Beichman. C.. et al. 1984. RAS Point Source Catalog. Boksenberg. A..et al. 1978. Narure. 278. 404 Burstein. D.. and Heiles. C. 1982. 4... 87, 1168. Campins. H.C. Rieke. G. H.. and Lebofsky. M. J. 1985, 4.J., 90. 896. Crenshaw. D. M.. and Peterson. B. M. 1985. Ap. J..291. 677. Cutn, RM. 1986. private communication Cutri, R. M.. Rudy. R. J.. Rieke. G. H., Tokunaga. A T.. and Willner, S. P. 1984. 4p. J.. 280. $21. Cutn. R. M.. Wisniewski. W. Z.. Rieke. G. H.. and Lebofsky. M. J. 1985. Ap. J.. 296.423 de Bruyn. A.G.. and Sargent, W. L. 1978, 4.J.. 83, 1257. de Grijp. RH. K.. Miley.G. K.. Lub. J.. and de Jong. T. 1985. Nature, 314, 240. de Jong. T.. Clegg. P. E.. Soifer. B. T.. Rowan-Robinson. M.. Habing. H. J. Houck. J. R.. Aumann, H. H.. and Raimond. E. 1984, Ap. J. (Letters, 278. L67. Draine. B. T..and Lee. H. M. 1984. Ap. J.. 285. 89. Elias. J. H.. Frogel. J. A.. Matthews, K.. and Neugebauer. G. 1982. A.J.. 87. 1029 Glass. 1. S. 1981, M.N.R.A.S.. 197. 1067. Grandi. §. A. 1981. Ap. J.. 251.451 Harvey. P. M., Wilhing. B. A.. and Joy. M. i982. 4p. J. (Letters), 254. L29. Hildebrand. R. H.. Whitcomb. S. E.. Winston. R.. Stiening. R. F.. Harper. D. A..and Moseley. S. H. 1977. Ap. J.. 216. 698. Impey. C. D..and Malkan. M. A. 1987, in preparation. Isobe. T.. Feigelson. E. C.. and Nelson. P. I. 1986. Ap. J.. 306, 490. . Kellermann. kK. 1. and Pauliny-Toth. I. 1. K. 1981, Ann. Rev. Astr. Ap.. 19, 373. Kwan. J..and Krolik.J.H. 1981. Ap. J... 250. 478. —~27- No. 1. 1986 Lacy. J. H..er al. 1982. 4p. J.. 256.78 Lebofsks. M. J..and Rieke. G. H. 1980, Nature, 284. 410. —. 198. private communication Mathan. M.A 1983u. 4p. J. (Letters), 264. L1. 983h, Ap. J. 268. S82 1984.1n Proc. Garching Conf. on Ultraviolet and X-Ray Emission from Actite Guiactic Nuclei (Garching: ESO), p. 121 . 1986. 4p. J.. in press Malkan. M.A.. and Filippenko. A V. 1983. Ap. J..275.477 (MF). Malkan. M. A..and Moore. R. 1986. Ap. J. 300. 216. Malhan. M. A..and Oke. J. B. 1983. Ap. J.. 265. 92(MO}. Malkan. M. A..and Sargent. W. L. W. 1982. Ap. J.. 284. 22 (MS). Maraschi. L.. Tanzi. E. G.. and Treves. A. 1980. in Proc. Second JUE European Conf. Tubingen (Garching: ESA). p. 299. McAlary. C. W.. McLaren. R. A..and Crabtree. D. R. 1979, Ap. J... 234.471. McAlary. C. W., McLaren, R. A.. McGonegal. R. J.. and Maza. J. 1983, Ap. J. Suppl’. $2. 341. McAlary. C. W.. Rieke. G. H.. Lebofsky. M. J.. and Stocke, J. T. 1986, Ap. J., 301. 105. Miley. G. K.. Neugebauer. G.. Clegg. P. E.. Harris. S.. Rowan-Robinson. M., Soifer. B. T..and Young. E. 1984. 4p. J. (Letters). 278. L79 Miley. G. K.. Neugebauer. G.. and Soifer, B. T. 1985. Ap. J. (Letters), 293. L11. Neugebauer. G.. et al. 1980, Ap. J.. 238. $02. Neugebauer. G.. Oke. J. B.. Becklin. E. E.. and Matthews, K. 1979, Ap. J., 230. 79. Neugebauer, G.. Soifer. B. T.. and Miley. G. K. 1985, Ap. J. (Letters), 295. L27. Neugebauer. G.. er al. 1984. 4p. J. (Letters). 278. L83. Neugebauer. G.. Matthews. K.. Soifer, B. T., and Elias. J. H. 1985, Ap. J., 298. g 78, Oke. J. B.. and Goodrich, R. F. 1981, Ap. J., 243, 445. SPECTRAL ENERGY DISTRIBUTIONS OF AGNs 77 Oke. J. B.. and Zimmerman. B. 1979, Ap. J. (Letters). 231.113. Pounds. K..and McHardy. I. 1985, private communication. Puetter. R.C.. and Hubbard. E. N. 1985. Ap. J., 295. 394. Richstone. D. O.. and Schmidt. M 1980. Ap J.. 235, 361. Rickard. L. J..and Harvey. P.M. 1984. A.J.. 89, 1520. Rieke. G. H. 1976, Ap. J. (Letters), 210. L5. 1978, Ap. J.. 266. 550. Rieke. G. H.. and Lebofsky. M. J. 1979a. WAP. J..227, 710. .1979h, Ann. Rev. Astr. Ap..17.477. Rieke. G.H.. Lebofsky. M. J..and Low. F. J. 1985, A.J. 90, 900 Robson. E. I. et al. 1983, Nature, 305, 194. Roche. P. F.. Aitken, D. K.. Phillips. M.M.. and Whitmore. B 1984. M.N.R.A.S.. 207. 35. Rudy. R. J. Jones. B.. LeVan, P. D.. Puetter. R. C., Smith, H. E., Willner. S. P.. and Tokunaga. A. T. 1982a. Ap. J.. 257. 570. Rudy. R. J.. LeVan, P. D.. Puetter. RR. C.. Smith, H. E.. and Willner, S. P. 19826, Ap. J.. 253. 53. Rudy. R. J., LeVan, P. D.. and Rodriguez-Espinosa. J. M. 1982. A.J.,87, 598. Rybicki. G. B.. and Lightman. A. P.1979, Radiative Processes in Astrophysics (New York: Wiley). p. 4 Sargent. W. L..de Bruyn. A 'G..and Readhead. A.S. C. 1987. in preparation. Seaton. M. J. 1979, M.N.R.A.S.. 187, 7 Siegel. S. 1956. Nonparametric Statistics for the Behavioral Sciences (New York: McGraw-Hill) Snijders, M. A.. Boksenberg. A.. Haskell. J. D.. Fosbury. R. A., and Penston. M.V. 1982. in Proc. Second IU'E European Conference (Tubingen) (Garching: ESA). p. 279. Telesco. C. M..and Harper. D. A. 1980. Ap. J.. 235. 392. Ulrich, M. H.. et al. 1982, M.N.R.A.S., 192, 56]. R. A. EDELSon: Palomar Observatory, 105-24, California Institute of Technology, Pasadena,CA 91125 M. A. MALKAN: Department of Astronomy, University of California. Los Angeles, CA 90024 — 28 - Chapter Three FAR-INFRARED PROPERTIES OF OPTICALLY SELECTED QUASARS Reprinted from The Astrophysical Journal, Letters to the Editor (1986, v. 309, pp. L69-L72) -— 29 —- Tre Astnoprysicat JouRNAL, 309.L69-L72, 1986 October 15 © 1986. The Amencen Astronommea) Somsty AD ngbts reserved Prated in USA FAR-INFRARED PROPERTIES OF OPTICALLY SELECTED QUASARS R. A. EDELSON Owens Valley Radio Observatory, California Institute of Technology . Recewed 1986 May 19; accepted 1986 July 21 ABSTRACT Pointed [RAS observations and ground-based near-infrared and radio measurements are used to study the far-infrared properties of 10 bright, optically selected quasars, nine of which were detected in at least three TRAS bands. Optically selected quasars tend to have very flat infrared spectra, with (aj342) = ~0.69 + 0.41 (individual dispersion), and (a32"™) = —1.09 + 0.16. The flat far-infrared power-law spectra, which extrapo- late smoothly to near-infrared wavelengths, and the large observed 60-100 ym luminosities, provide strong evidence that the infrared continuum emission from most quasars is nonthermal in nature. Seven of the nine quasars with far-infrared detections had low-frequency turnovers determined. Optically selected quasars tend to have higher turnover frequencies than optically selected Seyfert galaxies. The infrared emission appears to be unreprocessed radiation from a synchrotron self-absorbed region a few light hours to a few light days across, about the size of the hypothesized accretion disk Only three of the mine quasars in this complete, optically selected sample have — 1.25 < ait? < -0.5. Subject headings: infrared: spectra — quasars 1. INTRODUCTION The flight of the Infrared Astronomical Satellite (IRAS) has made it possible to study the far-infrared properties of a large number of active galactic auclei (AGNs). The far- infrared properties of AGNs are important for two reasons: (1) It is critical to determine whether the infrared radiation is primarily nonthermal or thermal dust emission; and (2) the low radio flux densities observed from most AGNs (ie., Condon et al. 1982; Edelson 1987; this Leer) generally require that the infrared continua, which rise from the optical through the infrared, cut off somewhere in the far-infrared- submillimeter region. Depending on the emission mechanism, this turnover can be used to determine source parameters which strongly constrain the physical conditions in AGNs. Progress has been made in understanding the observed far-infrared properties of AGNs. Edelson and Malkan (1986, hereafter EM) determined the spectral energy distributions (SEDs) of a heterogeneous sample of 29 AGNs between 0.1 and 100 4m Among their relevant results, they found (1) there is a strong correlation between infrared spectral index and reddening, in the sense that objects with large internal reddenings tend to have steep infrared SEDs; and (2) over half of the SEDs exhibited turnovers shortward of 100 pm. For objects with infrared SEDs which appear to be dominated by nonthermal emission (low-reddening Seyfert 1 galaxies and quasars), the best explanation is that the emission arises from a synchrotron self-absorbed source about the size of the hypothesized accretion disk. For Seyfert 2 galaxies and other dusty objects, the dust is warmer than that in normal galaxies (T, = 35-70 K), and the infrared emission appears to originate in the narrow-line region. The infrared-submillimeter properties of the complete, spectroscopically selected CfA sample of Seyfert galaxies were studied by Edelson, Malkan, and Rieke (1987, hereafter EMR). They found that Seyfert 1 galaxies tend to have significantly flatter infrared SEDs than Seyfert 2 galaxies, suggesting that the steepness of infrared SEDs provides a useful test for the presence of dust, and confirming the finding of EM that Seyfert 2 galaxies tend to have more dust than Seyfert 1 galaxies. Almost half of the CfA Seyfert galaxies detected at three or more 7RAS frequencies were found to turn over shortward of 100 pm. Miley, Neugebauer, and Soifer (1985) found that type 1 and 2 Seyfert galaxies tend to populate different regions of infrared color-color diagrams. Starburst and “normal” galax- jes tend to have steep infrared SEDs, which are generally thought to be signatures of dust emission. Evidence that the SEDs of AGNs flatten out and turn over in the far-infrared can be seen in Figure 1 of Miley, Neugebauer, and Soifer (1985) and Figures 4 and 5 of Neugebauer e al (19865). This Letter is organized as follows: The sample and data are discussed in the next section. The infrared spectral proper- ties of this sample are analyzed in § I]. The determination and implications of the far-infrared turnover are discussed in § IV. A concluding discussion and summary is given in § V. t this Letter, values of Hp = 75 km s~' Mpc™? and go = 0 are assumed. L69 D. SAMPLE AND DATA Is order to study the far-infrared characteristics of a repre- sentative sample of quasars, a complete, optically selected sample has been formed, consisting of the 10 quasars in the PG/BQS sample with B < 15.00 and My, s — 22.12 (Schmidt and Green 1983). Far-infrared studies of compiete, optically selected quasar samples have a number of advantages. Quasars are generally'so distant that there is a negligible contribution from the underlying galary. Cold dust emission from the — 30- L70 EDELSON Vol. 309 TABLE 1 flux density of IT Zw 136, for which the value of 462 + 75 IRAS Data ror Ton 1542 mJy is taken from EM. One source (Ton 1542) was not yim) Sy (my) discussed in Neugebauer er al. (19865). The JRAS flux densi- ties for this source, extracted from survey co-adds, are re- a ; 3 ported in Table 1. Nine of the bright PG/BQS quasars (90%) 60 160 + 46 were detected in three or more JRAS bands. One object 100 << 276 (0026 + 129) was not detected by /RAS, with a 3e limit of 40 mJy at 60 pm, and it will not be discussed further. Near- infrared (1.1-3.5 wm) data were generally taken from TABLE 2 Neugebauer e7 al (1986), except for I Zw 1, which was Ravio Dara measured by Rieke (1978). The data for most of these sources Source Seem (mJy) 0026+ 129 < 1.74 12a < 135 Ton 951 < 1.23 3C 273 30500 Ton 1542 PB 1732 Mkn 478 1634+ 706 Il Ze 136 Mkn 304 x 1530 AAAA Pek hededadted 0.39 RERERR a A galactic disk surrounding low-luminosity Seyfert nuclei can contaminate the low-resolution JRAS beam, causing infrared SEDs to appear to be steeper than that of the nucleus, and often masking the far-infrared turnover (EM). Neugebauer e7 al. (19865) present the results of pointed and survey observations of a heterogeneous collection of most known quasars detected with JRAS. However, the heteroge- neous nature of their sample and the relatively low detection rate (less than 25% were detected in three JRAS bands, the minimum number needed for the spectra] analysis undertaken in this Lerter) mean that the sample of detected objects may contain unknown biases. One such bias is that unusually dusty objects (such as Mrk 231) are probably overrepresented in their sample. The JRAS data for eight of the bright PG/BQS quasars are taken from and discussed extensively in Neugebauer er al. (19865). In a few cases, these data disagree with those re- ported in EM, EMR, or the IRAS Point Source Catalog (Beichman et al 1984). In these cases, the data of Neugebauer et al, (19865) are generally used. The exception is the 100 pm are of high quality, but the far-infrared SEDs of 2 sources (Ton 1542 and Ton 951) are poorly determined, and Mrk 304 lacks near-infrared data. The bright PG/BQS quasars were observed at 1.5 cm with the Owens Valley Radio Observatory 40 m telescope in 1985 June. The results of these measurements are presented in Table 2. A detailed description of the observing system can be found in Edelson (1987). Undetected objects have 3 o upper limits quoted. J. INFRARED SPECTRA Table 3 contains indices, redshifts, and luminosi- ties derived for the bright PG/BQS quasars. Source names are given in the first column and redshifts (taken from Schmidt and Green 1983) are listed in the second column. The far- infrared spectral indices a43*%, aij”, and azj*™ are given in the third through fifth columns (F, & »°). The 60-100 pm luminosities (L,,) are given im the sixth column, and the seventh column contains plotting offsets used in Figure 1. The quantity L,, is given by | a\ L,= 49 =| (F, + »F,,), Ho ' where », = 5 THz (60 pm) and », = 3 THz (100 um). The far-infrared SEDs of the nine bright PG/BQS quasars detected by RAS are presented in Figure 1. The quantity »F, is plotted as a function of frequency, so a SED witha = —1 (i.e., equal power per octave) will appear as a horizontal line. The general tendency for the objects with well-determined SEDs (i.e., the top six objects in Fig. 1) to have relatively flat, Straight SEDs is apparent. A gentle flattening toward lower TABLE 3 INFRARED SPECTRAL INDICES AND OTHER PARAMETERS bog Lie Source Fa elem = ote (erg/s) C 1Zw 1 0.061 -0.894004 +-1182004 -0842008 45.16 15.94 Ton 951 0.064 -0.162021 =1.16 + 008 0.262 0.36 44.12 13.09 3C 273 0.158 -0.91 40.04 09274003 =-0.7520.08 45.94 14.04 Ton 1542 0.064 -0.1740.24 -—132 + 0.09 0.78 + 0.36 43.95 14.10 PB 1732 (0.089 -0.21 0.08 -089240.04 -0.0140.15 44.20 20.09 Min 4768 40.077 -3.18 2008 -0864004 +1312010 4486 15.49 16344706 1.334 -103 20.13 -1.234 0.04 =-08820.14 46.95 17.80 Il Zw 136 0.061 —0.58 + 0.04 0.27 + 0.08 4443 18.62 Mkn 304 0.067 —1.06 + 0.07 -1.17 + 0.04 14542014 44.27 11.77 ~31- No. 2, 1986 q q T q QO + PB 1732 —_ + J B+ mw 136 YS 5 a n a> 7b 1634+708 ppt E 3) yy Sizes ae 4 be rs) — tw OF Min 478 ow et + n> 4f 3c 273 aan iie a ao 3 Ton 1842 ) f° 4 2 2+ ton 951 “oo A 1 > Min 306 4a - ] 1 i 1 _t 11 12 13 14 log vp) (Hz) Fic. 1.—Far-infrared SEDs of the nine bright PG/BQS quasars detected by JRAS. The independent variable is log »), where vp is the frequency in the rest frame. The dependent variable log »F, + C. where C is an arbitrary offset added for display purposes, tabulated in Table 3. Most of the data are of good quality, but the SEDs of the three quasars at the bottom (Ton 1542, Ton 951, and Mrk 304) are poorty determined. frequencies is also seen in many SEDs. In particular, it should be noted that the far-infrared SEDs generally extrapolate smoothly to the near-infrared. The high-frequency excess in the high-redshift quasar 1614+ 706 is due to the “ultraviolet bump” seen in most AGNs (see EM). The bright PG/BQS quasars generally have very flat in- frared SEDs. The mean infrared spectral indices are (ai3*™ ) = —0.69 + 0.44 (dispersion for individual objects) and (a22*™) = —1.09 + 0.16. This is flatter than the optically selected Seyfert 1 galaxies in the CfA Seyfert galaxy sample (ane) = —1.08 + 044 and (a32*™) = -115 + 0.29, EMR), or the optically selected CfA Seyfert 2 galaxies ((eo2*2) = —1.55 + 0.36 and (ai2*™) = -1.56 + 0.34, EMR). These SEDs are also significantly flatter than those of normal galaxies. De Grip et al. (1985) report that a large fraction of JRAS sources with — 1.25 < ait < —0.5 were found to be AGNs. Fons er only three of the nine bright PG/BQS quasars with 23" determined (33%) bave —1.25 < ais*® < —0.5. Thus, thet method finds only a fraction of all AGNs. INFRARED PROPERTIES OF QUASARS L71 TV. LOW-FREQUENCY TURNOVER AJ of the bright PG/BQS quasars except 3C 273 have low radio flux densities (see Table 2), which indicate that the SEDs which rise from the near-infrared through the far- infrared must cut off sharply in the far-infrared. This far- infrared cutoff has been studied before, for a heterogeneous sample of active galaxies in EM, and for the CfA Seyfert galaxies in EMR. In EM, EMR, and the present study, the turnover frequency (»,) is determined by fitting a parabola in log »-log F, space to the 25, 60, and 100 wm /RAS data (or, if the object is undetected at 100 um or has a flux density minimum at 60 pm, to the 12, 25, and 60 wm data), and taking the point at which a = 0 as the turnover, provided », = 3 THz (100 um) in the observed frame. EM contains a detailed discussion of this method. The turnover frequencies of three objects (Ton 951, Ton 1542, and PB 1732) were measured with the higher frequency RAS data. The turnover in Mrk 304 was measured by assuming that the 100 ym flux density was at the observed limit, so the value of », reported for this object is actually a lower limit. The turnover frequencies (determined in the rest ‘frame of the quasar) are tabulated in the second column of Table 4. Errors associated with the derived turnover frequen- cies can be as large as 50%. Seven of the nine bright PG/BQS quasars detected in at least three JRAS bands (78%) had turnovers detected short- yard of 100 wm. The turnover frequencies determined for the PG/BQS.quasars also tend to be higher than those found in lower “Ramin asity AGNs. Six of the nine bright PG/BQS quasars deiectéd at three or more frequencies (67%) have been », > 5 THz By comparison, only two of the 33 CfA Seyfert galaxies detected at 3° ar more JRAS bands (6%) have », > 5 THz (EMR). The noaparametric logrank (Breslow) test of the distribution of a single parameter for data with limits (Feigeison and, Nelson 1985) indicates that the PG/BQS quasars tend to have higher turnover frequencies than the CE£A Seyfert galaxies at the 99.4% confidence level. This turnover can be ysed to determine source parameters. If it is assumed that the infrared emission emanates from a synchrotron self-absorbed region with T, = 10’? K, source sizes can be derived (see EM for details). This parameter (7,,) is tabulated in the third column of Table 4. If it.is assumed that the 2 keV X-ray emission arises from inverse-Compton- boosted infrared photons, one can estimate source sizes (7,), brightmess temperatures (T,), and magnetic fields (B,). A detailed discussion of this method can be found in EMR. These parameters are tabulated in the fourth through sixth. columns of Table 4, respectively. Sources without reported TABLE 4 INFRARED TURNOVER PARAMETERS %, ns % T, 8 ’, A Source (THz) (10% cm) (10% em) (10° Ky) (G) (10% cm) (10" M.) 1Zw) 3.1 21 10 44 360 18 61 Ton 95) 10.0 0.18 0.84 47 = 1100 23 79 Ton 1842 108 0.17 0.78 4s 1200 25 85° PB 1732 78 0.32 14 $2 670 43 14 1634+ 706 9.9 33 27 41 1300-870 3000 Nl Zw 136 $6 0.33 25 47 $60 38 13 Mkn 304 5.8 0.50 22 $2 480 13 45 — 32- L72 EDELSON X-ray observations were assumed to have 2 keV flux densities of 3 pJy. This is not a very important assumption, because the source parameters depend very weakly on X-ray flux. Finally, lower limits to the Schwarzschild radius (7,) and mass (m,) of a hypothesized central black hole can be determined, by assuming that all of the 1-100 pm luminosity ultimately arises from accretion at the Eddington limit (EM). These parameters are tabulated in the seventh and eighth columns of Table 4, respectively. The Compton method yields brightness temperatures of order 7. » 5 x 10!° K. This is typical for normal synchrotron self-absorbed radio sources. Typical magnetic fields of order 10° gauss are derived. The emitting regions have derived sizes which are typically of order a light-day across. These sizes correspond to upper limits of 100-1000 Schwarzschild radii, similar to the size of the hypothesized accretion disk. V. DISCUSSION Optically selected quasars have very flat infrared SEDs. These flat SEDs are strong evidence that the infrared emission is dominated by nonthermal radiation Objects with infrared SEDs dominated by thermal dust emission (starburst galaxies, Seyfert 2 galaxies, normal galaxies, etc.) almost always have Steep infrared SEDs, while objects known to be dominated by nonthermal emission (such as BL Lacertae objects and OVV quasars) usually have flat SEDs. EM found that AGNs be- lieved to be dominated by nonthermal emission in the in- frared generally had a32?*7 > —1.25, while those thought to be dominated by thermal d dust emission usually had a3? < —1.25. AD but one of the bright PG/BQS quasars have a3?*™ > —1.25. Furthermore, EM found a strong correlation between reddening and infrared spectral slope, in the sense that objects with steep infrared SEDs tended to have high internal reddenings. Further evidence of the nonthermal nature of the infrared emission from quasars is given by the large 60-100 pm luminosities. As Neugebauer er al. (19865) point out, assum- ing that the far-infrared emission from luminous quasars is thermal emission from dust implies that the emitting regions must be a few kpc in size. The high turnover frequencies found would imply high dust temperatures, 7, 2 70 K. The sizes and masses are too large for the narrow-line region, yet the high temperatures suggest that the dust could not lie outside the narrow-line region. Finally, the fact that the bright PG/BQS quasars generally have straight or gently curving infrared SEDs, which extrapo- late smoothly to near-infrared wavelengths, is also indicative of nonthermal emission. OVV quasars and BL Lacertae ob- jects have infrared SEDs of this type. Dusty objects usually have curved infrared SEDs for which extrapolation to higher frequencies tends to underestimate the near-infrared flux den- sity, because the dust cannot get hot enough to radiate at near-infrared wavelengths (EM). For the near-infrared and far-infrared emission to join up so well, yet be caused by two different mechanisms, requires an unlikely coincidence for a large number of objects. In § IV, source sizes on the order of a light day are derived, under the assumption that the far-infrared SEDs are dominated by emission from a synchrotron self-absorbed source. As reported in EM and EMR, these sizes are of the same order as the hypothesized accretion disk. Quasars have higher values of », than type 1 (or type 2) Seyfert galaxies. There are two obvious possible explanations. The first is that both Seyfert 1 galaxies and quasars have relatively high turnover frequencies, but that emission from cold dust in the disk of the underlying galaxy masks it or shifts it to lower frequencies. This is supported by the fact that most of the nearest Seyfert galaxies in EMR and EM have undetected turnovers (ie., », < 3 THz). The other is that there may be a physical limit to the size of the synchrotron self-absorbed source. Then, the more luminous and distant objects would become optically thick at a higher frequency. This is supported by the fact that the highest turnover frequency found for an object with a well-determined infrared SED is that of PG 1634+ 706, which is the most luminous and distant object in this study, EMR, or ER. The author would like to thank A. Readhead, M. Malkan, M. Schmidt and N. Scoville for useful and interesting discus- sions. W. Rice provided invaluable help with the JRAS data. Analysis of [RAS data was supported by NASA grant NAG $-427, and centimeter-wave astronomy at Owens Valley Ra- dio Observatory is supported by NSF grant AST 85-09822. REFERENCES Beichman, C., ef af 1984, IRAS Point Source Catal ee Condon, 3. , O'Dell, & L, Puschell, J. J., and Stem, W. A. 1981, Ap. J. $6 Grp R HK. Miley, G. K, Lub, J, , aad de Jong, T. 1985, Nature, Edelson, R A 1987, Ap. J., in press. Edelson, RA, and Malkan M.A 1986, Ap. J., WB, $9 (EM). Edelson, R A, Malkan, M.A, and Rieke, G. 1987, Ap. J., in press Feigelson, E C., and Nelson, P. L 1985, Ap. J., 293. 192. Miley, G. KC, Neugebauer, G., and Soifer, B. T. 1985, Ap. J. (Letters), Neugebauer, G, , Green, R F., + Mathews, K., Schmidt, M. Soifer, B. T., J. 19864, Pad , in press. Neve pus, G.. Miley, G _K., Soifer, B. T., and Clegg, P. E 19866, Ap. Rieke, G. H. 1978, Ap. J., 266, 550. Schmidt, M, and Green, R F. 1983, Ap. J.. &®, 352. R. A EpELson: Palomar Observatory, 105-24, California Institute of Technology, Pasadena, CA 91125 —~ 33 - Chapter Four BROADBAND PROPERTIES OF THE CFA SEYFERT GALAXIES: I. RADIO PROPERTIES To appear in The Astrophysical Journal (1987, v. 313) — 34- ABSTRACT Quasi-simultaneous high-sensitivity observations of a complete, unbiased sample of 42 bright, spectroscopically selected Seyfert galaxies have been made at 1.5, 6, and 20 cm. All of the objects observed were detected above 0.3 mJy at 6 cm. Seyfert 2 galaxies tend to have steep, straight radio spectra with a » —0.7. Al- though this is often true for Seyfert 1 galaxies as well, there is a higher tendency for Seyfert 1 galaxies to have flat spectra, and it appears that about 25% of the Seyfert 1 galaxies have highly curved spectra which flatten out toward higher frequencies. This may be because of emission from a compact core or from the low-frequency side of the infrared power law. There is evidence for extended emission, probably from the disk of the underlying galaxy, in about half of the objects studied. There is no significant difference in the mean 6 cm radio luminosities of types 1 and 2 Seyfert galaxies in this well-defined sample. The radio luminosity of Seyfert galaxies is well correlated with optical luminosity, but not with the presence of nearby companion galaxies. Fits to the relation L,gq « Lé,, yields a value of s significantly steeper than s = 1. While most Seyfert galaxies have an_p * 0, the radio and optical-infrared continua apparently originate in different regions. Seyfert galaxies are generally not variable at 20 cm over time scales of ~5 years. There are of order 10° Seyfert galaxies/Gpc® with radio luminosities between 10°66 and 1039-4 erg/s. There is a sharp cutoff in the radio luminosity function of Seyfert 2 037.2 galaxies below 1 ergs/s. —~ 35 - I. INTRODUCTION The radio properties of “radio-quiet” active galactic nuclei (AGNs) are not well understood. The first quasars were discovered because of their radio emission, but only a small fraction of optically selected quasars (and Seyfert galaxies) are “radio-loud” objects (i.e., objects which have much higher flux densities at radio wavelengths than at optical wavelengths; see, e.g., Condon et al. 1981b). Some models of continuum emission (such as those involving beaming) posit that the intrinsic radio properties of radio-loud sources are the same as those of the underly- ing population of AGNs, although the observed radio properties are quite different. The radio properties of radio selected sources are well known (i.e., Kithr ez al. 1981), but these objects tend to be at the extreme radio-loud end of the radio luminosity function, so their behavior is not representative of the underlying distribution of AGNs. Relatively low detection rates in radio surveys of complete, homogeneous, optically selected samples of Seyfert galaxies and quasars have frustrated attempts to determine the radio properties of the underlying population of AGNs. Meurs and Wilson (1984), in work based partially on a survey by de Bruyn and Wilson (1976), detected 60% of a sample of Markarian Seyfert galaxies above ~4 mJy at 20 cm. These results were incorporated into a 6 cm study by Ulvestad and Wilson (1984a), which determined non-simultaneous spectral indices between 6 and 20 cm for 37% of the original sample of Markarian Seyfert galaxies. A survey of a heterogeneous collection of all active galaxies catalogued before mid-1982 and accessible to the VLA with V,aqg < 3100 km/s, detected all of the objects at 6 cm (Ulvestad and Wilson 1984b). However, that sample consists mostly of nearby, low-luminosity objects and is not a complete, unbiased, optically selected sample. Previous radio surveys of optically selected quasars have also had a low success rate. The similarities in the optical and X-ray properties of optically selected ~ 36- Seyfert 1 galaxies and quasars suggest that their radio properties could also be similar. A 6 cm survey by Condon et al. (1981b) detected 40% of a heterogeneous sample of optically selected quasars above 0.5 mJy. A quick VLA survey of the PG/BQS quasars detected a similar fraction above 1 mJy at 6 cm (Schaffer et al. 1983). A larger fraction are detected in a deep survey (Kellerman 1985), the results of which have not yet been published. The most successful published multifrequency radio survey of optically selected quasars is Condon et al. (1981a), which determined the radio spectra of 10% of the sources observed. - This paper reports the results of three-frequency observations of the CfA Seyfert galaxies (Huchra and Berg 1987)—a well-defined sample of bright, optically selected Seyfert galaxies. All of the sources observed were detected at 6 cm, and for most, 1.5-20 cm spectra have been determined as well. This permits the radio properties of a well-defined sample of Seyfert galaxies to be probed. This paper is organized as follows. In the next section, the sample is discussed, and the observations and data are discussed in § III. The radio spectral properties of optically selected Seyfert galaxies are derived and analyzed in § IV. The radio source sizes are discussed in § V, and the relation of radio luminosity to other properties such as optical luminosity and the presence of companion galaxies is investigated in § VI. The Seyfert galaxy radio luminosity function is derived and discussed in § VII, and § VIII contains a concluding discussion and summary of the major results of this paper. Throughout this paper, a value of Hp = 75 km/s/Mpc is assumed. Since the largest redshift in this sample is z = 0.06, no corrections were made for cosmological effects in the luminosity distance or for evolution. Il. THE CfA SEYFERT GALAXIES The objects observed were taken from the CfA sample of Seyfert galaxies, a collection of 25 Type 1 and 23 Type 2 Seyfert galaxies (Huchra and Berg 1987). -~37- This sample was derived from the CfA redshift survey (Huchra et al. 1983) which obtained spectra of all galaxies with Zwicky magnitudes brighter than mpg = 14.5 in a region of sky with |b| > 30°, 2 = 2.66 sr in extent. The presence of strong emission lines in these spectra was the sole selection criterion used to distinguish these objects from other galaxies in the survey. For studies of the broad-band properties of Seyfert galaxies, the CfA Seyfert galaxies have a number of advantages over other samples. They are 1-2 magnitudes brighter (and therefore easier to detect at other frequencies) than both the PG/BQS quasars (Schmidt and Green 1983) and the Markarian Seyfert galaxies (Huchra and Sargent 1973). No ultraviolet-excess selection criterion was used in the CfA survey. Selection by ultraviolet-excess has been shown to bias samples against strong infrared emitters (Rieke 1978; Edelson, Malkan, and Rieke 1987), and the effect on radio properties is unknown. The Markarian survey is known to be incomplete at both the bright and faint ends (Huchra and Sargent 1973). The results of the V/Vm test performed by Huchra and Berg (1987) indicate that the CfA Seyfert galaxies are distributed uniformly in space, as one would expect from a complete, unbiased sample. Forty-two of the 48 CfA Seyfert galaxies were observed. Three Seyfert 2 galaxies, NGC 3982, Mkn 461, and IC 4397, were not observed because they were added to the list after the radio measurements were made. Mkn 205 and Mkn 789, both laxies, and NGC 4388, a Seyfert 2, were missed because good optical coordinates were not used. Omission of these six objects from the radio survey is not expected to significantly bias the results, and they will not be discussed further. Intermediate types (i.e., Seyfert 1.5 galaxies, etc.) are classified as Seyfert 1 galaxies. — 38 - III. OBSERVATIONAL RESULTS a) Observations Quasi-simultaneous high-sensitivity measurements of the CfA Seyfert galaxies were made at observing wavelengths A 1.5 cm, 6 cm, and 20 cm. The 1.5 cm measurements were made with the Owens Valley Radio Observatory 40-m telescope, on 1983 July 7-16. The observing system was a beam-switched radiometer using a reflected-wave maser of the NRAO-JPL design (Moore and Clauss 1979), with a cooled Dicke switch. The system temperature was 45 K, with an RF bandwidth of 400 MHz centered on 20.0 GHz (1.5 cm). The FWHM beamwidth was about 1/5, and the beam separation was 7/1. Contiguous 40-second scans of each object were combined to form observational records of up to 30 minutes in duration. Each source was observed a number of times during the nine-day run, and the data were combined to yield effective total integration times of a few hours per object. Typical 30 upper limits of ~2 mJy were obtained for undetected sources. Calibration was done relative to 3C 147, 3C 286, and 3C 295, assuming 1.5 cm flux densities of 1.75 Jy, 2.76 Jy, and 1.07 Jy, respectively (Baars et al. 1977). The measurements at 6 cm and 20 cm were made at the VLA! on 1983 July 4. Fifty-megahertz bandwidths were used in the (compact) D-array, and the center frequencies were 4.89 GHz (6 cm) and 1.46 GHz (20 cm). Single snapshots were made of each object with integration times of about 10 minutes at 6 cm and of 2.5 minutes at 20 cm. Phase calibrators were observed once every hour. Flux calibration was done relative to 3C 286, which was assumed to have 6 cm and 20 cm flux densities of 14.5 Jy and 7.41 Jy respectively (Baars et al. 1977). Each source was mapped at 6 cm with an untapered synthesized beam of about 15” 1 The National Radio Astronomy Observatory is operated by Associated Univer- sities, Inc., under contract with the National Science Foundation. ~ 39 — FWHM (referred to as 6 cm,h, hereafter) and with tapered beams of about 1'5 (the size of the OVRO 40-m beam at 1.5 cm) at 6 cm (6 cm,]l, hereafter) and at 20 cm. The maps were CLEANed with the standard AIPS software. An elliptical Gaussian with its size and orientation matched to those of the synthesized beam was fitted to each map. The position and the amplitude were allowed to vary in the high-resolution 6 cm,h map, allowing the right ascension, declination, and central peak flux density to be determined. This position was used and held fixed in fits to the low-resolution 6 cm,] and 20 cm maps. Only peak low-resolution flux densities were determined from these maps. The lo noise levels for faint and undetected point sources were about 0.1 mJy in the high-resolution 6 cm,h maps, 0.3 mJy in the low-resolution 6 cm, maps, and 0.5 mJy at 20 cm. With the exception of one object (0152+06, for which good optical astrometric positions are not available), the 6 cm,h sources were required to be within 15” of published optical astrometric positions. The data of Ledden et al. (1980) indicate that the brightest random source expected within 1! of 0152+06 and 15” of the 41 remaining optical positions should be weaker than 0.05 mJy. The faintest source detected at 6 cm was brighter than 0.3 mJy, so confusion was not a problem. b) Data The observational data are presented in Table 1. The first four columns, which give the name, type (Seyfert 1 or 2), redshift, and integrated optical flux density for each object, are taken from Huchra and Berg (1986). All quoted flux densities are in millijanskies. The optical flux densities are based on Zwicky magnitudes, which represent the sum of nuclear plus galactic light. Since all objects in the sample have |b| > 30°, no correction is made for Galactic absorption. The positions in Columns 5 and 6 were determined with the VLA from the 6 cm,h maps and are generally good to 1”. The last four columns contain the flux densities and uncertainties for the 20 cm data (S20cm), the high-resolution 6 cm data (Sgem,,), the low-resolution ~40- 6 cm data (S¢gcm,j), and the 1.5 cm data (515m). The quoted uncertainty is the quadratic sum of the statistical errors and an estimated 5% uncertainty in the calibration. Upper limits are at 30. Of the 42 sources observed, all were detected at 6 cm,h, 41 (98%) were detected at 6 cm,l, 38 (90%) were detected at 20 cm, and 24 (57%) were detected at 1.5 cm. TABLE 1 Spectral indices are computed and presented in Table 2. The spectral index, Qq—p, is defined as _ logSa — logS, ~ logva — logy” Qa—b Columns 1-2 are the same as in Table 1. Columns 3-4 contain the spectral indices 6~20em (defined between 6 cm,] and 20 cm), and a1,5~gem (from 1.5 cm to 6 cm,]), respectively. Column 5 contains the quantity R, defined as R= Séem,h/ Séem,I- Column 6 contains the radio-optical spectral index (ag_p). This quantity is based on fluxes inferred from Zwicky’s integrated galactic photographic magnitudes (with mpg = 14.5 corresponding to F,=6.25 mJy at 4500 A), and 6 cm,h radio measurements. Columns 7-8 contain the logarithm of the 6 cm,h monochromatic radio luminosity (Z,,g) and monochromatic optical Zwicky galaxy luminosity (Lopt) in erg/s. They are given by ez \* Ly = 4nvSy () , N Z where vy = 4.89 GHz and 650 THz for L,,q and Lopt; respectively. TABLE 2 ~4Al-— c) Comparison with Previous Observations De Bruyn and Wilson (1976), Meurs and Wilson (1981), and Wilson and Meurs (1982) observed 14 of the objects in this sample with WSRT with a similar beam size at 21 cm from 1975 through 1981. With the exception of one object (Mkn 573), their fluxes are in excellent agreement with those reported here. This indicates that Seyfert galaxies are not generally variable at 20 cm on time scales of ~ 5 years. For Mkn 573, Wilson and Meurs (1981) find a 21 cm flux density of 21.14 1.1 mJy, while Meurs and Wilson (1984) give a flux density of 11.4 mJy. This discrepancy is not explained. A 20 cm flux density of 16.1+ 0.8 mJy for Mkn 573 is reported here. Ulvestad and Wilson (1984a,b) made very high-resolution 6 cm observations, with the VLA in the A configuration, of 13 objects studied in this paper. In general, they found flux densities which were similar to, or slightly less than, the untapered 6 cm,h values reported here. This suggests that some objects are extended at the ~0"5 level at 6 cm. The most extreme difference was for Mkn 231, which had a measured 6 cm flux density of 170 +10 mJy in Ulvestad and Wilson (1984a) and 278 + 14 mJy reported here. Furthermore, Sramek and Tovmassian (1975) report a 6 cm flux density of 137420 mJy, which differs significantly from both of these results. As the 6 cm,h and 6 cm,] fluxes are in good agreement for this object, it is probably not extended, but the fact that it is one of the few CfA Seyfert galaxies with a flat spectrum radio source suggests that it may be variable. FIGURE 1 ~42- a) Low-Frequency Spectra Figure 1 contains spectra for the 24 objects which were detected at all three frequencies. Low-frequency spectral indices (ag_20em) were determined for the 38 sources (90%) detected at 6 and 20 cm. Figure 2 is a histogram of ag—20cm. The top panel refers to Seyfert 1 galaxies, and the bottom refers to Seyfert 2 galaxies. For Seyfert 1 galaxies, the mean value of ag_20em was —0.66, with a dispersion of 0.28 for individual points. For Seyfert 2 galaxies, the mean value of ag—20em was —0.71 + 0.23. FIGURE 2 The Kolmogorov-Smirnov (K-S) test, a nonparametric rank test, was used to test for differences between two observed distributions of a parameter (Siegel 1956). The K-S test yields the parameter D, which can be used to determine if the hypothesis that two samples were derived from the same distribution can be rejected at a given confidence level. Figure 2 suggests that Seyfert 1 galaxies may have a higher tendency to have flat low-frequency radio spectra than Seyfert 2 galaxies. Five Seyfert 1 galaxies (25%) have ag_20cem > —0.5, while only 2 Seyfert 2 galaxies are this flat. However, because of the small size of this sample, no statistically significant evidence was found with the K-S test for a difference in the distribution of ag—20em for types 1 and 2 Seyfert galaxies. A larger sample would be needed to determine the differences, if any, in the distributions of ag_920em for different types of Seyfert galaxies. b) High-Frequency Spectra, Spectral Curvature High-frequency spectral indices (a1,.5-6em) were determined for the 24 sources detected at all three wavelengths. Figure 3 is a plot of ag—20em VS. 01.5—6em for these 24 objects. This quantity appears to be distributed differently for types 1 —~ 43 - and 2 Seyfert galaxies. Seyfert 1 galaxies have a mean high-frequency spectral index @15-6em = —0.58 + 0.29 (individual scatter), while for Seyfert 2 galaxies, @15-6em = —0.83 40.15. Not only is the mean value of a1.5-6cm larger for Seyfert 1 galaxies than Seyfert 2 galaxies, but the width of the distribution for Seyfert 1 galaxies is also larger. The results of the K-S test indicates that a15~6¢m is distributed differently for types 1 and 2 Seyfert galaxies at the 95% confidence level. The median value of a1 5~6¢m is —0.62 for Seyfert 1 galaxies and —0.82 for Seyfert 2 galaxies. A difference would remain even if NGC 4235, a Seyfert 1 with Q15—6em = +0.24, were excluded from the sample. FIGURE 3 Most Seyfert galaxies show little spectral curvature between 1.5 and 20 cm. The quantity |ag—20em — %1.5—6em| < 0.25 for 18 of the 24 objects with 3-point spectra. Three Seyfert 1 galaxies (NGC 4051, NGC 4235, and 2237+07), but no Seyfert 2 galaxies, have spectra with large upward curvature, (ag—20em — @1.5—6em) > 0.40. This suggests that the spectra of Seyfert 1 galaxies have a tendency to flatten out towards higher frequencies, although the sample is too small for the K-S test to find any statistically significant difference in the distribution of (ag—90em—1.5—6em) for types 1 and 2 Seyfert galaxies. Possible explanations are discussed in § IVc. Three sources (Mkn 231, a Seyfert 1 galaxy, and NGC 5252 and 0152+06, both Seyfert 2 galaxies) show strong downward curvature, with (ag_20em— @1.5—6em) < —0.35. All of these sources have flat 6-20 cm spectra, so they are probably becoming optically thick near 6 cm. c) Discussion Seyfert 2 radio spectral properties are distributed in a similar fashion to those of “normal” optically-thin synchrotron sources. They have spectral indicies clustered —~44-— around —0.7 to —0.8, with a small fraction (~ 10%) having flat radio spectra. They generally show little spectral curvature between 1.5 and 20 cm, and none of the Seyfert 2 galaxies in this sample show extreme spectral flattening toward higher frequencies. The radio spectral properties of Seyfert 1 galaxies tend to be more broadly distributed than those of Seyfert 2 galaxies, and Seyfert 1 galaxies frequently have radio characteristics not often seen in optically selected sources. Seyfert 1 galaxies have flatter high-frequency spectra than Seyfert 2 galaxies (about 20% of the Seyfert 1 galaxies have flat spectra), and about 25% of the Seyfert 1 galaxies in this sample appear to show marked spectral flattening toward high frequencies. This suggests that the radio emission from Seyfert 2 galaxies is dominated by synchrotron emission from an optically thin source. Most Seyfert 1 galaxies are also dominated by optically thin emission, but something else may be going on in the sources which flatten toward higher frequencies. Two possible explanations are: 1) Some Seyfert 1 galaxies contain a flat spectrum core not found in Seyfert 2 galaxies, which is becoming visible against the steep spectrum extended emission near 1.5 cm; or 2) In some Seyfert 1 galaxies, low-frequency emission from the nonthermal power law component, which dominates the optical-infrared continuum of Seyfert 1 galaxies and quasars (but appears to be absent or much weaker in Seyfert 2 galaxies) and which peaks near 100 ym (Edelson and Malkan 1986, Edelson, Malkan, and Rieke 1987), has a flux density which is similar to that of the extended radio emission near 1.5 cm. Millimeter-wave observations could help resolve this issue. The first scenario predicts a flat millimeter spectrum (amm * 0), and the second, an inverted spectrum (@mm > +1). V. RESOLUTION EFFECTS Untapered 6 cm maps, with a beam width of about 15’, and tapered 6 cm —~45~- maps, with a beam width of about 1/5, were made of the CfA Seyfert galaxies. This corresponds to about 4 and 25 kpc, respectively, for an object at a redshift z = 0.02 (the median redshift of the sample). Thus, the large beam typically samples the entire galaxy, including the disk, while the small beam contains only the central source. The quantity R, defined as the ratio of the flux in the central beam of the untapered map to that in the tapered map, is a good measure of the degree of resolution at 6 cm. Figure 4 contains a histogram of this quantity, which is tabulated in Column 5 of Table 2, for the 41 sources for which R was determined. Statistical fluctuations cause some sources to have values of R greater than unity. In no case is the value of R more than 1.50 above unity. FIGURE 4 Many of the Seyfert radio sources are extended at 6 cm. About half (20 of 41) of the sources have R < 0.8, and 3 sources (NGC 5033, NGC 5674, and NGC 5929) have R < 0.5. No difference was found in the distribution of R for types 1 and 2 Seyfert galaxies with the K-S test. This effect is most easily explained as emission from the disk of the underlying galaxy. The 21 cm data of Hummel (1981), corrected to 6 cm with a spectral index of —0.7, yields a mean value of S,q/Sop¢ = 0.4 for 6 cm disk emission from normal spiral galaxies. The mean value of S,gq/Sopt is 1.7 for Seyfert galaxies (see § VIc). about 0.4/1.7 = 24% of the o This suggests that, on average, the disk will contribut total 6 cm flux (i.e., R = 0.76), although the ratio should be higher for Seyfert galaxies with low radio luminosities. The observed mean value is R = 0.81 for the CfA Seyfert galaxies, in good agreement with this expectation. ~ 46 - VI. RELATION TO OTHER PROPERTIES a) Radio Luminosity of Types 1 and 2 Seyfert galaxies The relative luminosity of types 1 and 2 Seyfert radio sources is important for models which relate them in an evolutionary sequence, as well as models of the interaction between radio jets or plasmoids and broad-line-region gas (e.g., Osterbrock 1984). As all of the objects observed in this complete sample were detected at 6 cm, these data can be used for a rigorous investigation of the radio luminosities of Seyfert galaxies, as well as the relation between radio power and other properties. Figure 5 is a plot of radio power (L,gq) as a function of optical power (Lopt). Solving for the mean of the logarithm of the luminosity is equivalent to tak- ing its harmonic mean. For Seyfert 1 galaxies, this mean radio luminosity is 1037-9840.90 erg /s (RMS scatter for individual objects), while for Seyfert 2 galaxies, it is 1028-16£0.67 erg/s, The median radio luminosity of Seyfert 1 galaxies is 1038-14 erg/s, while for Seyfert 2 galaxies, it is 1038-05 erg/s. FIGURE 5 No difference was found in the distribution of L,,qg for types 1 and 2 Seyfert galaxies with the K-S test. While the mean radio luminosities are clearly not significantly different, Seyfert 1 galaxies appear to have a broader distribution of radio luminosities than Seyfert 2 galaxies. However, this trend is not statistically significant in this small sample. This result is different than that found by Meurs and Wilson (1984). Based on a higher radio detection rate for Seyfert 2 galaxies in a survey of a heterogeneous sample composed primarily of Seyfert galaxies taken from the Markarian survey, they concluded that Seyfert 2 galaxies are more powerful radio sources that Seyfert 1 ~A7- galaxies. Selection effects are the most likely cause for this difference. Relative to Seyfert 1 galaxies, the ultraviolet-excess selected Markarian sample contains half as many Seyfert 2 galaxies as the CfA sample. It primarily missed red, low- luminosity Seyfert 2 galaxies, which were included in the spectroscopic CfA survey (Huchra, Wyatt, and Davis 1982). The strong correlation between optical and radio luminosity causes the Markarian sample to be biased against the inclusion of radio- weak Seyfert 2 galaxies. Furthermore, Seyfert 1 galaxies are more optically luminous (and therefore more distant in magnitude-limited surveys), and the distribution of aoQ—R is broader for Seyfert 1 galaxies than for Seyfert 2 galaxies (see § VIc). These effects will all tend to depress the radio detection rate of Seyfert 1 galaxies relative to that of the Seyfert 2 galaxies in their sample. Figure 5 shows that the intrinsic spread in L,,q for both types of Seyfert galaxies is much larger than any difference which there may be in the mean values of Lygq. (The same is true for L,gq/Lop; see § Vic.) Differences in the radio powers of Seyfert galaxies of the same types are at least as great as global differences between different types. Theories which explain the radio luminosities of Seyfert galaxies in terms of confinement of the synchrotron source by broad-line region clouds must take this large spread in L,,q into account... b) Radio and Optical Luminagity All of the objects in this well-defined sample were detected at both radio and optical wavelengths, so it is valid to test for luminosity-luminosity. ccrrelations. Figure 5 suggests that there is a strong correlation between L,gq and Lop; for both types 1 and 2 Seyfert galaxies. The Spearman rank correlation test (see Siegel 1956), a nonparametric rank test, was used to search for correlations between parameters. This test yields t, the student’s t-statistic, which is used to determine if the hypothesis that that the two variables are uncorrelated can be rejected at a given confidence level. The results — 48 - of this test indicates that L,gqg and Lop are correlated for both types of Seyfert galaxies above the 99.9% confidence level. Proper least squares fits to the power law relation Lygq « cept were made to the data. The final regression line is the mean of two regressions, each of which was done using a different variable as the independent variable. For Seyfert 1 galaxies, s = 1.95 + 0.26, for Seyfert 2 galaxies, s = 1.66 + 0.30, and for both types of Seyfert galaxy, s = 1.8140.21. This steep slope is consistent with the results shown in Figure 7 of Meurs and Wilson (1984), although they found no correlation for Seyfert 1 galaxies, and only found a marginal correlation for Seyfert 2 galaxies, probably because of their relatively low detection rate. It is not immediately obvious why the slopes should be significantly steeper than L,gq « Lop. It may be that the strength of the radio source is related to the nonthermal optical point source, and that contamination of the optical flux density with light from the underlying galaxy steepens the L,gq — Lopt relation. c) Radio-Optical Spectral Index Figure 6 is a histogram of ag__p, the radio-optical spectral index. This quantity, given in Column 6 of Table 2, is based on the high-resolution 6 cm data (6 cm,h) and the integrated Zwicky magnitudes. FIGURE 6 Computing the mean radio-optical spectral index is equivalent to computing the harmonic mean of the radio-optical luminosity ratio. For Seyfert 1 galaxies, @o-R = 0.073, with a lo scatter of 0.118 for individual objects, and for Seyfert 2 galaxies, @_p = 0.017 + 0.109. The difference in Go_p for types 1 and 2 Seyfert galaxies is 0.056, which corresponds to Seyfert 2 galaxies being on average a factor of 1.9 brighter than ~ 49 - Seyfert 1 galaxies with the same optical brightness. The results of the K-S test indicates that the distribution of a9_R for types 1 and 2 Seyfert galaxies is different at the 99% confidence level. This is a consequence of the fact that Seyfert 1 galaxies are optically more luminous than Seyfert 2 galaxies, while their radio luminosities are similar. For all Seyfert galaxies, Gop = 0.045. which corresponds to Sygq/Sopt = 1.7. For most Seyfert galaxies, ag_g0em * —0.7. Edelson and Malkan (1986) and Edelson, Malkan, and Rieke (1987) find that the infrared spectral indicies of Seyfert 1 galaxies are clustered around ayp ~ —1.2, with most showing a sharp cutoff near 100 wm. Under the best assumption (synchrotron self-absorption), they found that the infrared emission was confined to a region of order a light day in size. Seyfert 2 galaxies and other dusty objects were found to have even steeper infrared spectra which also peaked in the far-infrared, presumably because of thermal dust emission. Extrapolation of the far-infrared peak to 6 cm would require a spectral index 2100um—6em 2 +1. No sources in this survey have such highly inverted radio spectra. Although this component may be becoming visible in some Seyfert 1 galaxies at 1.5 cm (see § IV), the 6-20 cm spectra of all sources are clearly dominated by more extended emission. d) Companions It has been suggested that activity such as radio emission from Seyfert galaxies could be triggered by interactions with other galaxies. Ulvestad and Wilson (1984a) found evidence for a correlation between radio emission from Seyfert galaxies and the presence of a companion galaxy, although Ulvestad and Wilson (1984b) did not confirm this result. This question is addressed quantitatively in this section. Two different approaches are used. In the first approach, the CfA redshift survey data (Huchra et al. 1983) were used to determine the distance from each —~5O0- of the 14 Seyfert galaxies in this sample with z < 0.015 (8 Type 1 and 6 Type 2 Seyfert galaxies) to its closest companion. The nearest companion was defined to be the galaxy with the smallest angular separation on the sky, provided that galaxy’s redshift was no more than 300 km/s different than that of the Seyfert galaxy and the companion had mpg < 14.5. No correlation was found between projected distance and radio luminosity with the Spearman rank correlation test. The second approach used a sample of 28 CfA Seyfert galaxies (15 Type 1 and 13 Type 2 Seyfert galaxies) for which 6 cm radio data and Dahari’s (1985) “interaction parameter” are available. Again, the Spearman rank test was used to test for a correlation between interaction parameter and radio luminosity, but none was found. These results indicate that the radio luminosity of Seyfert galaxies is not a strong function of the presence of companion galaxies. VII. RADIO LUMINOSITY FUNCTION a) Introduction The Seyfert galaxy radio luminosity function (RLF) is an important tool for understanding the nature of Seyfert galaxies and their relation to other objects. If possible, it would be best to determine the Seyfert galaxy RLF from a sample identified in a complete radio survey. Unfortunately, Seyfert galaxies are rather weak radio sources, and they are not found in large numbers in fully identified radio surveys, so this method cannot be used. The next best method (which is used in this paper) is to make radio measure- ments of a complete, unbiased, optically selected sample. The RLF is derived from the optical luminosity function and the bivariate radio-optical luminosity distribu- tion function. This method has the disadvantage that errors can be introduced by the spread in L,gq/Lopt. For instance, no extreme radio-loud objects such as 3C 120 are represented in the CfA Seyfert galaxy sample. Nonetheless, it is used —~51- because no other method for determining the Seyfert galaxy RLF is available. Meurs and Wilson (1984) used this method to determine the RLF of Markarian Seyfert galaxies with 21 cm observations at Westerbork. As the V/V test shows that the Markarian survey is incomplete at both the bright and faint ends (Huchra and Sargent 1973), large corrections were applied to the data. However, nothing was done about the bias of the Markarian survey against low-luminosity Seyfert 2 galaxies (see § VI). Non-Markarian Seyfert galaxies were added to the sample and data were weighted according to the derived value of V/Vm in an attempt to offset the effect of incompleteness. Both the optical luminosity function and the bivariate radio-optical luminosity distribution function were binned. No discussion was made of errors introduced by the double binning process or by the spread in the L,aq/Lopt. Further assumptions about the distribution of L,gq/Lopt were required because only 60% of the sources observed were detected. The CfA Seyfert galaxies form a complete, unbiased sample (Huchra and Berg 1987), so no large corrections for incompleteness are needed. The fact that the entire sample was detected at 6 cm means that few assumptions had to be made concerning the distribution of L,gq/Lopt. b) Derivation The derivation of the CfA Seyfert galaxy RLF is based on the V/V method described in Schmidt and Green (1983). The contribution of each object to the luminosity function is derived individually, and the final luminosity function is the sum of the individual contributions. This method has the advantage that the data are not binned until the last step, so information is not lost in intermediate steps. The maximum accessible volume (Vin) is the volume of space within which an object could be detected above both survey limits. There are two possible values for Vm, one based on the optical survey limit (Vin opt)? and the other on the radio survey limit (Vm,_4)- The accessible volume is the smaller of these two values, that —52- is Vin => min(Vin ops 5 Vingad)* This is equivalent to forming the contribution of each object to the RLF from its contribution to the optical luminosity function and the bivariate radio-optical luminosity distribution function. Once these quantities have been calculated, the RLF is derived and binned. The differential RLF, 6(L,gq), is given by 4r Ol a=1 summed over all n objects with L,gq in each bin of width AL (Huchra and Sargent 1973). In this equation, 2 is the amount of sky surveyed optically (2 = 2.66 sr), and f is the fraction of optically selected objects which were observed at radio frequencies (f = 0.92 for the Seyfert 1 galaxies, 0.83 for the Seyfert 2 galaxies, and 0.87 for the entire sample of Seyfert galaxies). . The data were binned using logarithmic bins of width AL = 10°-4 (correspond- ing to one “radio magnitude”). The RLF was determined for radio luminosities of 1087.0_1939.4 erg/s. No CfA Seyfert galaxy had radio luminosities between 10°6-6 and 1037-0 ergs/s, but an upper limit to the space density of Seyfert galaxies with L,aq Within this range was derived by using a value of Viz = 7.2x1074 Gpc?, which is appropriate for an object with L,gq = 10°58 erg/s at a survey limit of 0.35 mJy. It is difficult to estimate the errors in a RLF derived from an optically selected sample. Besides those which arise from statistical fluctuations, errors can also be introduced by the spread in the distribution of L,gq/Lop¢ and by biases or incompleteness in the optical Seyfert galaxy selection process. The optical limit is usually the significant one in a survey with sufficient sensitivity to detect all of the sources at radio wavelengths (i.e., Vin opt < Vin,gq for most objects in this sample), so the spread in L,gq/Lopt gives rise to a spread in Vin(Lyqq). ~53- The differential RLF was derived for types 1 and 2 Seyfert galaxies separately, and for both types together. It is presented in Table 3 and Figure 7. Column 1 contains the binned values of L,gqg. Columns 2-3, 4-5, and 6-7 give the space densities and number of objects in each bin for type 1’s, type 2’s, and both types together. The error bars in Figure 7 represent the lo statistical uncertainty, uncorrected for the effects discussed earlier. The crosses in Figure 7 represent the RLF of Meurs and Wilson (1984), scaled to the units used in this paper with a 6-20 cm spectral index ag_20em = —0.7. TABLE 3 FIGURE 7 c) Discussion The RLF is determined between 10°79 and 10°94 erg/s. and limits are es- tablished for L,gq between 1036-6 and 1037-9 erg/s. The space density of Seyfert galaxies with radio luminosities between 1036-6 and 10°94 ergs/s is of order 10° Gpc73. All but one of the Seyfert 2 galaxies studied fell within this luminosity range. Mkn 533 had Lygg > 1029-4 erg/s. As is apparent in Figure 5, Seyfert 1 galaxies have a broader range of radio luminosities than Seyfert 2 galaxies (see § V). Two Seyfert 1 galaxies (NGC 4051 and NGC 5273) had Lyag < 10°66 erg/s and one (Mkn 231) had Lygq > 10294 erg/s. For L,gq between 10°6-6 and 10°94 erg/s, Seyfert 2 galaxies are about 3 times more numerous than Seyfert 1 galaxies. There is a dramatic turnover in the RLF of Seyfert 2 galaxies near 1037-2 erg/s. The space density near 10°”-2 erg/s is more than one order of magnitude below that extrapolated from higher luminosities. The RLF of Seyfert 1 galaxies also appears to cut off near 1037? ergs/s, although it is not as strong as that of Seyfert 2 galaxies. ~54- Only four points from the Meurs and Wilson (1984) Seyfert galaxy RLF were of sufficiently low luminosity to overlap the CfA Seyfert galaxy RLF. These points are plotted along with the CfA Seyfert RLF in Figure 7. The RLF of the Seyfert 1 galaxies are in very good agreement. However, the type 2 Markarian Seyfert galaxy RLF is generally about a factor of two below that of the CfA Seyfert galaxies. This suggests that both the Markarian and CfA surveys found most of the Seyfert 1 galaxies, but the Markarian survey missed about half of the Seyfert 2 galaxies, probably because they are too red to satisfy the ultraviolet-excess selection criterion. The Seyfert galaxy RLF can also be used to study the relative radio luminosities of Type 1 and 2 Seyfert galaxies. If the Seyfert 1 galaxy RLF is shifted by a factor of three (i.e., Alog® = 0.48, to account for the fact that Seyfert 2 galaxies are more numerous),the two RLFs are almost identical. The similarity in the shape of the RLF of Seyfert 1 and 2 galaxies is further evidence that of the lack of significant difference in the distribution of radio luminosities for different types of Seyfert galaxies. VIII. CONCLUSIONS The results of 1.5, 6, and 20 cm observations of the CfA Seyfert galaxies are reported. The CfA Seyfert galaxies form a complete, spectroscopically selected sample which does not suffer from biases caused by use of the ultraviolet-excess selection technique. It appears that half of the low-luminosity Seyfert 2 galaxies in | the CfA Seyfert galaxy sample are too red to be included in the ultraviolet-excess Markarian survey. All of the sources observed were detected at 6 cm. Two-frequency spectra were determined for 90% of these sources, and three-frequency spectra were determined for 57%. The high radio detection rate for this well-defined sample allows the “radio-quiet” end of the Seyfert galaxy radio luminosity function to be probed. —~55- Seyfert 2 radio spectral properties are similar to those of “normal” optically-thin synchrotron sources. They have uncurved radio spectra with ag_—g em clustered around —0.7, and about 10% have flat radio spectra. Most Seyfert 1 galaxies also have steep, uncurved radio spectra, but a larger fraction (~ 20%) have flat spectra, and about 25% appear to show strong spectral flattening toward higher frequencies. This could be caused by a second component (either a flat spectrum core or low-frequency emission from the nonthermal component which peaks near 100 ym) becoming visible at 1.5 cm. Emission from the disk of the underlying galaxy is typically ~20% of the total 6 cm flux density, and can be substantially higher for low-luminosity objects. The properties of radio emission from the disks of types 1 and 2 Seyfert galaxies are similar. There is no significant difference in the radio luminosities of the types 1 and 2 Seyfert galaxies in this complete sample. The intrinsic spread in L,gq for either type of Seyfert galaxy is much larger than any differences in the mean radio luminosities. The luminosity of Seyfert galaxy radio sources is strongly correlated with the integrated optical galaxy luminosity. Fits to the power law relation Dygq « Lent yield s = 1.95 + 0.26 for Seyfert 1 galaxies, s = 1.66 + 0.30 for Seyfert 2 galaxies, and s = 1.81+0.21 for both types. It is not immediately clear why the slope is significantly steeper than s = 1. In most cases, ayp © —1.3 and ag_ogem & —0.7, while an_pR & 0, suggesting that the radio and optical-infrared power law emission arises from different regions and/or physical mechanisms. There is no correlation between radio luminosity and the presence of a nearby companion galaxy. Seyfert galaxies are generally not variable on time scales of ~5 years at 20 cm. There are of order 10° Seyfert galaxies/Gpc® with radio luminosities between 10366 and 10°94 erg/s. Seyfert 2 galaxies are about three times more numerous — 56 - than Seyfert 1 galaxies in this luminosity range, but Seyfert 1 galaxies have a broader distribution of radio luminosities. There is a strong cutoff in the RLF of Seyfert 2 galaxies near L,gq = 10°” erg/s, as the derived space density is more than one order of magnitude less than extrapolation from higher luminosities would predict. The author would like to thank A. Moffet, A. Readhead, M. Schmidt, L. Dressel, and N. Scoville for useful and interesting discussions. The anonymous referee made an important contribution to the analysis of the RLF. The assistance of R. Moore and H. Hardebeck of OVRO and J. van Gorkom of the VLA in gathering and analyzing the data is also appreciated. J. Huchra kindly released the list of CfA Seyfert galaxies well ahead of publication. This work was partially supported by NSF grants AST 85-09822 and AST 83-14134. -~57- 90 FLS tl FO Cl ¥F oz le F919 ORZ GT LO 92 £% 2 01100 ZS Z89L ODN Cl FLO st F USL ve ¥F 999 WF 0% G71 OC & =O bZ GZ Lz v1 682700 Clg WN 90 F7P 90 FLU RS0 F OSII Ol ¥F 9RZ TRpl O- Wwworez 69 0600 0cS UN 6 F Uw of FPFIL ce ¥* 199 68 FOIL G1 908 br O cz c% 09100 «6 tS 69PL ODN £0 FZ vO FD ov0 ¥F 090 Tl oF LC ze Lb «60S OR LE OL oszeoo Od LOFLEZZ et > Oo L0 FRR 600 ¥ PPT tl ¥e6 ORb GH SE sO0'Or FI OT 16 oszoo dT GC+PIOT 6T > ¢cO Fob 40 ¥ Sez zt ER Lil LOL «© GS@I 82 SI OL eccoo =o TS OK6S DON al oF Le 1% ¥ 0¢ Gt F g0Z Ze =F 0001 Zp OS Ih «SORE FS ST Or £2000 & 6266 ODN zeOCU>C«sCSG'D £60 ¥ 6Rt grr > BSS 20 01 Ozer St ol pocoo Od Tre UyIY 1 > GO ¥ 6% oro FIL 90 F998 ZC Lh Of CSI SE FT U zz100 = ORO UIWY GZ > «60 FUE wo F109 60 F001 O1h 0 6S 66LS vo FT 92 ricoo 8d 218 WIN 80 ¥*F 6% 90 96 t70 ¥ 69¢ LI ¥ 98% “le OFS «CS ZZ IE HT ct gbz00 % $299 ODN LO Fee 60 ¥ Stl G0 ¥ OF OI t& OF Lov 61 226% beh ST PT % 99100 1 &bSS ODN ce > vo FOL eco FKL vl Fw eel ce 69 COS TS CI C9 10000 oT 6L2 AN ve > £0 F0Z cV0 ¥ 6st g% > OC FS Se Tess Ct CI ce oc000 4«26T)~=— £829 ODN Leo0U}> CUO FUG 0 ¥ %y 80 FLU 08% GS-49 6LTy 6 CT OL 06000 «2 OL WAN Cl Fob wz oF ftp Lt oF wee z9 =F O01 O6Gb IC ah «BLT OF CT 16 e200 0% 99% WN ol FE 60 F Tel CeO ¥ 0791 Tl ¥ 961 ZLb Lb OBEY SE CT co 12200 2 2629 ODN gl > 0 FET v0 ¥ 191 gic > Cll beef I9'Re Ge CI fe 107000 2 6ct+scel gl F OST 0% ¥ gee 080 ¥ 0621 9¢ FOR Cle 1g of Pe TT CT oot «6 ordd0SCiCSsCéLOG DON Il ¥ cat Cl F ala bl F = Olz cl F gz Gee e Lf ITS bG ZI 16 ooo =O 12% WN Lo Fes 80 ¥ PSI 180 ¥ 08ST 61 Fest 6S% $ OC FOSS ST ZI cl TAC 99L UN Tl Fe co F8¢ 80 F ROP 80 6 BR ZL OFC PT Zt rd 14000 =T Sth ODN vo ¥ 66o C9 F OG vo ¥F OSZI 91 ¥F omc TZ Wee pO 8 Zt oct oc0oo0 )6=—Cdt SCS ODN TL FeL £0 FOIL 10 Fue bz 90 OSC Sh bb «6Z9L O ZI 002 «= zz0000SsdTSCTS0b ODN 62 > vO FOL oto ¥F OFS al FEL STW ze Bk OLE ot 16000 =I bbe UAW Ol Fes Ol ¥ 901 90 ¥ 29 Tl FSI zz OF ZL «OOwEZ EL OM a seov0otSsO9ISL ODN 9% > vo FEE 070 * 19 St FOF O1Z SS Sh Ih Zh RS OL 16 16100 2 Gh+RG01 ct > co FSP 00 F EZ 80 FOI LOC TS 9 VSI Ze OT bl 12200 «2 ~~ 29fL ODN Zl F276 6l Fost PIF Guz SS ¥F O10T 69 L 0% CL'9r 0% OF zg ec000 «= )—k@ZL ODN 99 ¥ OPEB cl ¥ 6gz VS ¥F Od oc F 169 O91 S969 Tere eo 6 ofl 20000 2% 620€ ODN 9% > 90 > WO F2c0 eb > CO LICL por 296 C9 cscoo OT CbtT IN Cl F 6cl Gt ¥ b6r LIF £62 CL ¥ 0961 OL CTO - Pee woz it gkz00 2 ~~ bPIT ODN be F 08 06 ¥ 0081 89 F OLLI 0cz F Ol «= OL LTO — MOL OZ ooc = 4000 Z—RBOT ~ ODN 90 FC vO FOL 670 ¥F cvs tl FT O18 66 O- SO Zz or £9700 06¢ YW 90 Fuze 90 ¥ 66 er0 ¥ cre TZ ¥ OSI 0% 29 SP ZT c9 a a 4 90+ZS10 gl > vO FPL 6CO F 802 Tl For Goss Z O6WZIVI ol e100 & CL¢ UY a a A A | a0 F 9VZ ¢O ge OL ZG IC ROC WT or psioo0 =a C66 UAW 61 > to ¥It L400 ¥F LZ 60 FFE L61 GZ ZI PA'LS 0S :O oL ro000 Ss 1 “ZI gl > co F8@ GUO ¥ £0% GO FL Zep L 62 GOES FO 16 6scoo iT 62+8b00 Ll > «620 FF 6U elo Fore 60 FI¥ 062 S$ 6 «OeGh ec O ot esz00.OiT sce UY L0 ¥ 9b 90 FPFIT 480 ¥ OT bl ¥ 692 OSS OF IZ sostO O 69 07200 % peo UAW (Agus) “stg Arua) (1)"°5 (Apu) (y) 95 (4pur) “70rg ‘20d ‘vu (Apu) Gg zt dA, = aUNOg Tatave VLVQ O1GVY AXVIVS) LUISATS YD — 58 - TABLE 2 CfA SEYFERT GALAXY RADIO PARAMETERS Source Type azeem aff R ag log(Lraa) log(Lop:) (erg/s) — (erg/s) Mkn 334 2 -0.71 —0.66 0.98 —0.036 38.70 43.62 Mkn 335 1 —0.29 wes 1.13 0.077 38.32 43.92 0048+ 29 1 —0.78 wee 0.72 0.105 38.39 44.17 I Zw 1 1 —0.82 we 0.89 0.070 38.98 44.54 Mkn 993 2 —0.51 wee 1.14 0.107 37.69 43.47 Mkn 573 2 —0.64 se 0.96 0.023 38.30 43.57 0152+06 2 —0.34 —0.70 0.95 —0.030 38.43 43.38 Mkn 590 1 —0.32 —0.53 0.69 0.044 38.53 43.94 NGC 1068 2 —0.78 —0.98 0.74 —0.106 39.23 43.71 NGC 1144 2 —0.90 —0.90 0.59 —0.025 39.36 44.34 Mkn 1243 1 see we wes we 37.64 43.99 NGC 3079 2 —0.68 —0.80 0.56 —0.008 38.27 43.35 NGC 3227 2 —0.88 —0.95 0.79 0.044 37.57 42.98 NGC 3362 2 —0.78 see 0.56 0.122 38.09 43.97 1058+45 2 -—1.17 wee 1.05 0.063 38.10 43.62 NGC 3516 1 -0.31 —0.49 0.61 0.141 37.64 43.64 Mkn 744 1 —0.76 we 0.77 0.075 37.62 43.21 NGC 4051 1 —1.04 —0.32 0.64 0.183 36.52 42.78 NGC 4151 1 —0.77 —0.81 1.00 0.002 38.02 43.17 NGC 4235 1 —0.41 0.24 0.81 0.124 37.41 43.31 Mkn 766 1 —0.70 —0.76 1.03 —0.014 38.38 43.43 Mkn 231 1 0.07 —0.58 0.97 —0.241 40.63 44.28 NGC 5033 1 —0.83 —0.66 0.32 0.176 37.04 43.25 1335+39 2 we we 1.20 0.115 37.78 43.62 NGC 5252 2 —0.07 —0.54 0.90 —0.068 38.91 43.62 Mkn 266 2 —0.75 —0.76 0.77 —0.093 39.37 43.94 Mkn 270 2 -0.71 we 0.95 0.029 37.57 42.88 NGC 5273 1 wee wee 0.80 0.213 36.28 42.73 Mkn 279 1 —0.92 wee 1.01 —0.013 38.81 43.86 NGC 5548 1 —0.89 —0.91 0.76 0.054 38.43 43.90 NGC 5674 2 —0.91 —0.85 0.39 0.089 38.33 44.01 Mkn 817 1 —0.53 wee 1.15 0.015 38.74 43.97 Mkn 686 2 —0.55 wee 0.60 0.130 37.38 43.31 Mkn 841 1 we wee 0.60 0.065 38.68 44.22 NGC 5929 2 —0.72 —0.76 0.50 —0.053 38.13 42.93 NGC 5940 1 —0.57 wee 0.55 0.081 38.40 44.04 1614435 1 —0.99 wee 0.51 0.129 38.02 43.95 2237+07 1 —0.94 —0.54 1.05 0.033 38.43 43.77 NGC 7469 1 —0.72 —0.83 0.93 —0.076 39.20 43.86 Mkn 530 1 —0.74 —0.73 0.98 —0.037 38.96 43.86 Mkn 533 2 —0.89 —1.07 0.89 —0.110 39.72 44.18 NGC 7682 2 —0.78 —1.02 1.02 —0.085 38.82 43.44 ~ 59 - TABLE 3 SEYFERT GALAXY Rapio LumiINosiTy FUNCTION Type 1 Type 2 Both Types log(Lraa) log(#) N log(#) N log(®) N erg/s Gpce7$mag7? Gpc~3mag7} Gpc~3mag7! 36.6-37.0 < 3.87 < 3.90 < 3.89 37.0-37.4 3.73 1 3.71 1 4.04 2 37.0-37.4 4.10 4 4.63 4 4.75 8 37.0-37.4 3.88 2 4.32 3 4.46 5 37.0-37.4 3.78 7 4.04 4 4.24 11 37.0-37.4 3.30 5 3.83 3 3.94 8 37.0-37.4 2.81 1 3.30 3 3.43 4 —~60- REFERENCES Baars, J. W. H., Genzel, R., Pauliny-Toth, I. K. K., and Witzel, A. 1977, Astr. Ap., 61, 99. de Bruyn, A. G., and Wilson, A. S. 1976, Astr. Ap., 53, 93. de Bruyn, A. G., and Wilson, A. S. 1978, Astr. Ap., 64, 433. Condon, J. J., Condon, M. A., Jauncey, D. L., Smith, M. G., Turtle, A. J., and Wright, A. E. 198la, Ap. J., 244, 5. Condon, J. J., O’Dell, S. L., Puschell, J. J., and Stein, W. A. 1981b, Ap. J., 246, 624. Dahari, O. 1985, A.J., 90, 1772. Edelson, R. A., and Malkan, M. A. Ap. J., 308, 59. Edelson, R. A., Malkan, M. A. , and Rieke, G. Ap. J., in press. Geller, M., and Huchra, J. 1983, Ap. J.Supp., 52, 61. Huchra, J, and Berg, R. 1986, Ap. J., in press. Huchra, J., Davis, M., Latham, D., and Tonry, J. 1983, Ap. J. (Supp.), 52, 89. Huchra, J. P., Wyatt, W. F., and Davis, M. 1982, Astron. J., 87, 1628. Huchra, J., and Sargent, W. L. W. 1973, Ap. J., 186, 433. Hummel, E. 1981, Astron. Ap., 93, 93. Kellerman, K. I. 1985, priv. comm. Kihr, H., Witzel, A., Pauliny-Toth, I. I. K., and Nauber, U. 1981, Astron. Ap., 45, 367. Ledden, J. E., Broderick, J. J., Condon, J. J., and Brown, R. L. 1980, Astron. J., 85, 780. Moore, C. R. and Clauss, R. C. 1979, IEEE Trans. MTT, 27, 249. Meurs, E. J. A. and Wilson, A. S. 1981, Astron. Ap. Supp., 45, 99. Meurs, E. J. A. and Wilson, A. S. 1984, Astron. Ap., 136, 204. Osterbrock, D. E. 1984, Q.J.R.A.S., 25, 1. -61- Rieke, G. H. 1978, Ap. J., 226, 550. Schaffer, D., Schmidt, M., and Green, R. F. 1983, IAU Symposium 97, Extragalactic Radio Sources, ed. D. S. Heeschen and C. M. Wade (Dordrecht:Reidel), p.367. Schmidt, M., and Green, R. F. 1983, Ap. J., 269, 352. Siegel, S. 1956, Nonparametric Statistics for the Behavioral Sciences (McGraw— Hill). Sramek, R. and Tovmassian 1975, Ap. J., 196, 339. Ulvestad, J. S., and Wilson, A. S. 1984a, Ap. J., 278, 544. Ulvestad, J. S., and Wilson, A. S. 1984b, Ap. J., 285, 439. Wilson, A. S. and Meurs, E. J. A. 1982, Astron. Ap. Supp., 50, 217. —~ 62 - FIGURE CAPTIONS FIG 1.—Three-point radio spectra (S, vs. v) of the 24 sources detected at all three radio frequencies. The Seyfert 1 galaxies are on the left and the Seyfert 2 galaxies are on the right. The two points at 6 cm are the high-resolution data (6 cm,h, open circles) and low-resolution data (6 cm,l, filled circles). The 6 cm,h points have been shifted slightly in frequency to prevent overlap with the 6 cm,] points. FIG 2.—Histogram of ag—29¢m for the 38 sources detected at both 6 and 20 cm. The top panel contains the data for Seyfert 1 galaxies, and the bottom, for Seyfert 2 galaxies. FIG 3.—Plot of ag_a0em VS. 01.5—6em for the 24 sources detected at all three frequencies. The Seyfert 1 galaxies are denoted by triangles, and the Seyfert 2 galaxies, by circles. A source with ag_a0em = @1.5—6em would lie on the dashed line. Sources with ag_o0em < 01.5—6em (i-e., which flatten toward higher frequency) lie below this line. Errors associated with these spectral indices are generally about 0.07, which is slightly larger than the symbols. FIG 4.—Histogram of R, a measure of the degree of resolution of the 6 cm radio source for the 41 sources detected with both 6 cm,h and 6 cm,] beams. The top panel contains the data for Seyfert 1 galaxies and the bottom, for Seyfert 2 galaxies. FIG 5.—Plot of radio luminosity (L,gq) as a function of optical luminosity (Lopt). Seyfert 1 galaxies are denoted by triangles, and Seyfert 2 galaxies, by circles. The dashed line is a proper least squares power law fit for Seyfert 1 galaxies and the dotted line for Seyfert 2 galaxies. FIG 6.—Histogram of ag_ p for all 42 sources observed. The top panel contains the data for Seyfert 1 galaxies and the bottom for Seyfert 2 galaxies. Mkn 231 is -~ 63 - the Seyfert 1 with a low value of ag_p. FIG 7.—The RLF of Seyfert galaxies. Space density is plotted as a function of Lyaq. Figure 7a is the RLF of Seyfert 1 galaxies, Figure 7b is that of Seyfert 2 galaxies, and Figure 7c is the RLF of all Seyfert galaxies. Error bars are lo statistical errors. Crosses represent the RLF of Meurs and Wilson (1984). Flux Density ML LLL CAA) NGC 4235 Mkn 231 NGC 3516 Min 590 NGC 4051 2237+07 Mkn 530 Mkn 766 NGC 5033 NGC 7469 NGC 4151 NGC 5548 — x ~w fo_aawener 10° fewer i 4. 10° 107° . Frequency (Hz) ~64- Flux Density MELEE LOL. Figure 1 TTT NGC 5252 0152+06 Mkn 334 NGC 3079 NGC 5929 Min 266 NGC 5674 NGC 1068 NGC 1144 NGC 3227 ° NGC 7682 Min 533 [x2 nm 10° ee eheedebet th faweert! 4 10° 10'° Frequency (Hz) -~ 65 - uID9 0 20- vo- 90- 80- OT- S@t- . I spalqo gt g adk], [ spalqo 02 T eddy, Jaquinn Jequinn Figure 2 -~ 66- Figure 3 -~67- ot O'T 8°0 9°0 v0 spalqo 61 g adfy, syalqo e2 T eddy, it | | Jaquinyn Iaquiny Figure 4 1 0*° — 68 — v reeeeery v verry SN vorrerery A, rises —_aaeree ii i L Null 1043 10*4 Figure 5 -~ 69 - © 0 oO TO spelqo 61 @ adfy, spalqo €2 T edd], | 1 | We) Jaquinn Jaequinn Figure 6 (Gpc-? mag”! ) 'o ~70- 1000 £ 10°F 1000 F 1000 F } Both types : Figure 7 —~71- Chapter Five BROADBAND PROPERTIES OF THE CFA SEYFERT GALAXIES: II. INFRARED-MILLIMETER PROPERTIES * Submitted to The Astrophysical Journal * Written in collaboration with M. Malkan and G. Rieke ~72- ABSTRACT Observations between 1.2 ym and 1.3 mm are presented for an unbiased, spectroscopically selected sample of 48 Seyfert galaxies. Most objects have complete near and far-infrared detections, but none were detected at 1.3 mm. The infrared spectra of optically selected Seyfert 2 galaxies are steep (@29-254m = —1.56), in sharp contrast to optically selected quasars, which have flat infrared spectra (@.2-25um = —1.09). This suggests that the infrared emission is predominantly thermal in Seyfert 2 galaxies and nonthermal in quasars. For optically selected Seyfert 1 galaxies, @2-25ym = —1.15, and 70% have flat infrared spectra similar to the quasars and unlike the Seyfert 2 galaxies. Thus, the infrared emission from most Seyfert 1 galaxies appears to be dominated by nonthermal radiation, although thermal dust radiation is clearly important for some others. Half of the objects detected at three or more IRAS wavelengths have far- infrared spectra which turn over shortward of 100 wm. In many of the others, emission from cold dust in the galactic disk appears to mask the turnover. For the Seyfert 2 galaxies and other dusty objects, this implies minimum dust temperatures of 35-65 K, significantly warmer than dust in normal galaxies. For the relatively dust-free Seyfert 1 galaxies, this implies that the infrared emission is dominated by unreprocessed radiation from a synchrotron self-absorbed source of order a light day in size, about the same size as the hypothesized accretion disks. The radio and optical luminosities of Seyfert 1 and Seyfert 2 galaxies are strongly correlated with infrared luminosity. There are of order 10° Seyfert galaxies/ Cpe? with L;, between 1042-2 and 1045-0 ergs/s. The infrared luminosity functions of relatively luminous host galaxies. —~73- I. INTRODUCTION Although Seyfert galaxies have been studied extensively at near-infrared wave- lengths (A < 10 wm; cf. Rieke 1978, hereafter referred to as R78), the Infrared Astronomy Satellite (IRAS) all-sky survey at 12, 25, 60, and 100 ym only recently provided the first far-infrared detections of a significant number of active galactic nuclei (AGNs). Edelson and Malkan (1986, hereafter referred to as EM) used IRAS, IUE and ground-based data to study the spectral energy distributions of a heterogeneous sample of 29 AGNs between 0.1 and 100 wm. Miley, Neugebauer, and Soifer (1985) reported the far-infrared colors of 120 active galaxies. Neugebauer et al. (1986) reported IRAS observations of 179 quasars (of which, 74 were detected in at least one IRAS wavelength), and Edelson (1986, hereafter referred to as E86) analyzed IRAS observations of the brightest PG/BQS quasars. Virtually every known Seyfert galaxy emits a large fraction of its total energy at infrared wavelengths. In some cases, this is the result of the presence of dust in or around the nucleus. The dust can absorb and thermally re-radiate a significant portion of the emission from the central source (Rieke and Lebofsky 1979). The thermal emission produces a very steep infrared spectrum which curves downward at shorter wavelengths, due to the Wein cutoff below the peak wavelength of emission from the hottest surviving dust grains. Thermal far-infrared emission is also frequently associated with dust reddening of the nuclear emission lines and nonstellar continuum. EM found correlations between steepness of the infrared spectra and internal reddening and other dust indicators. However, some active galactic nuclei produce their infrared continua by non- thermal processes. In the case of violently variable quasars and BL Lacertae objects, the rapid variability in the infrared and high polarization conclusively demonstrate this (Angel and Stockman 1980). In quasars and bright, unreddened Seyfert 1 nuclei, the case for nonthermal infrared (and optical) emission is less direct, since ~74- the continuum is neither highly polarized nor violently variable. The strongest indication is the relatively flat shape of the far-infrared-to-optical continuum, which is well-described by a power law with a2 — 1.3 (Fy «x v%, EM). Optically selected quasars have very flat infrared spectra (@.9-25ym = —1.09), which E86 identified as predominately nonthermal emission. The strong correlation found between the 3.5 ym and 2 keV X-ray flux of quasars and Seyfert 1 galaxies (but not Seyfert 2 galaxies) suggests that the same process is responsible for emission at both wavelengths, implying that the lower frequency emission from these objects is nonthermal in origin (Malkan 1984). In this paper, the infrared properties of a complete, unbiased sample of Seyfert galaxies are studied to determine the importance of different emission mechanisms. Previous AGN studies have been biased by the selection techniques used to con- struct their samples. For example, selection by ultraviolet-excess (such as that used by Markarian 1972 and Schmidt and Green 1982) has been shown to bias samples against strong infrared emitters (R78). This problem was avoided by examining the infrared-to-millimeter spectral energy distributions of the CfA Seyfert galaxies, a well-defined, unbiased, spectroscopically selected sample, to deduce the properties of the class as a whole. The CfA sample contains 26 Type 1 and 22 Type 2 Seyfert galaxies selected by optical spectroscopy (Huchra and Berg 1987). The selection of the sample is discussed in detail in Edelson (1987, hereafter referred to as Paper I) which also described their radio properties. (One object, NGC 3227, was incorrectly classified as a Seyfert 2 galaxy in Paper I.) The near-infrared, IRAS, and millimeter-wave data are discussed in the next section. The infrared spectral energy distributions are discussed in § III, and the far-infrared turnover and the relation between far-infrared and millimeter emission are discussed in § IV. The relationship between infrared properties and those derived at other wavelengths are investigated in § V, and the 60 ym infrared luminosity ~75-— function of Seyfert galaxies is derived and discussed in § VI. A summary of the major results of this paper is given in § VII. Throughout this paper, a value of Hj = 75 km/s/Mpcis assumed. As the largest redshift in this sample is z = 0.06, no corrections were made for cosmological effects in the luminosity distance or for evolution. II. DATA a) Near-Infrared Data Near-infrared photometry was obtained with standard single-element InSb de- tectors on the Steward Observatory Catalina 61-inch reflector, and with an InSb array on the Steward Observatory Kitt Peak 90-inch reflector. The usual photo- metric bands, J [1.2 um], H [1.6 wm], and K [2.2 um] were used, with absolute flux calibrations from Campins, Rieke and Lebofsky (1985). Complete near-infrared (1.2-2.2 um) photometry was obtained for 44 of the Seyfert galaxies (95%). Mea- surements were also made at L [3.5 wm] or L’ [3.6 pm] of objects which were sufficiently bright. The near-infrared database was supplemented with published measurements from the literature (R78, Rudy, LeVan and Rodriguez-Espinoza 1982, Cruz-Gonzales and Huchra 1984). Ground-based measurements of 20 objects at N [10.6 wm] were taken from the literature (R78; Deveraux, Becklin, and Scoville 1986), except for those listed at the end of Table 1. A total of 28 (58%) galaxies have 3.5 wm measurements while 24 (50%) have 10.6 ym measurements. TABLE 1 Table 1 gives the new near-infrared data for the CfA Seyfert galaxies. Source names are presented in Column 1. Flux densities and uncertainties for the J, H, and K measurements are presented in Columns 2-4, respectively. The L and N measurements are included as footnotes to Table 1. Column 5 contains the date of ~76- observations. The magnitudes listed in Table 1 refer to 9” apertures for the 61-inch measurements, and 10” apertures for the 90-inch images. Thus the fluxes refer primarily to the nucleus. In the less luminous Seyfert galaxies, these fluxes also include a substantial contribution from starlight, and there was often significant galactic light outside the aperture. Several repeated observations agreed to within about 10%. b) IRAS Data Forty-six of the 48 CfA Seyfert galaxies were observed at least once by IRAS. Two objects (NGC 3362 and Mkn 744) lay in the ~2% of the sky which IRAS did not observe. One of the 46 objects observed (NGC 7682) lay within 4’ of a 10 Jy source, and could not be detected due to strong confusion. All of the remaining 45 objects were detected with IRAS, and 32 of the 46 objects observed (70%) were detected in all four IRAS wavelengths. Far-infrared fluxes from IRAS are presented in Table 2. The source names are given in the first column. The flux densities and errors at 12, 25, 60, and 100 um, or 30 upper limits, are tabulated in Columns 2-5. For objects brighter than 0.5 Jy at 12 wm, photometry was taken from the IRAS point source catalogue (Beichman et al. 1984). Since the IRAS point source catalog fluxes ignored all observations in which a source was not detected, objects with low signal-to-noise ratios tend to appear brighter than they actually are. For this reason, all IRAS fluxes for objects fainter than 0.5 Jy at 12 zm were determined from source extractions performed on coadded survey data. No spectral corrections to the IRAS data were necessary because these objects had power law slopes near a = —1. The overall calibration uncertainty is estimated to be 10% at 12, 25, and 60 wm and 15% at 100 ym. TABLE 2 _~77- The infrared fluxes of three objects are confused. Mkn 279 appears extended in the IRAS beam because it is only 41” from a nearby spiral companion. Mkn 266 is actually a close binary galaxy (separation 9’). One of the galaxies has a Seyfert 2 nucleus, while the companion has a low ionization emission line spectrum (Fricke and Kollatschny 1984). NGC 5929 is only 90” from NGC 5930, which appears to emit most of the IRAS flux observed from this interacting system. These objects are excluded from further analysis. c) Millimeter Data The CfA Seyfert galaxies were observed at 1.3 mm with the NRAO-Kitt Peak 12-m telescope on 9-16 June 1984. The QMC-Oregon 0.3 K millimeter bolometer was used with a typical system temperature of ~325 K on the sky (double side- band). The aperture efficiency was about 15%, giving an effective noise equivalent flux density of ~5 Jy sec, The telescope had a 30” beamwidth and the pointing was accurate to ~10”. Most sources were observed for integration times of about 15 minutes, yielding typical lo errors of ~200 mJy. The actual errors were a strong function of the integration time, the time of day an object was observed, and the amount of precipitable water vapor along the line of sight. No sources were detected at 1.3 mm. The 3o upper limits are listed in Column 6 of Table 2. III. INFRARED SPECTRA a) Introduction Figure 1 contains vF, plots for the CfA Seyfert galaxies. A wide variety of spectral shapes and features are seen in Figure 1. The overall spectral energy distributions range from very steep (aS — 2, as in Mkn 533 and NGC 1068), to flat (a2 — 0.7, as in Mkn 335 and Mkn 590). The spectra of many of the less luminous AGNs have a bump in the near-infrared due to starlight which peaks near ~ 78 - ~1.5 ym. A far-infrared low-frequency turnover is detectable in about half of the infrared spectra. FIGURE 1 FIGURE 1 To quantify the following discussion of spectral shapes, spectral indices have been calculated and tabulated in Table 3. Column 1 contains the source name, and the spectral indices @2.9-25um; @12—60um» 925—602m> %60-100um, aNd a190—1300um are presented in Columns 2-6, respectively. TABLE 3 Table 4 contains parameters derived for these objects. Column 1 contains the source name, and Column 2, the Seyfert galaxy type, taken from Huchra (1987). Intermediate types (i.e., Seyfert 1.5 galaxies) are classified as Type 1. Using slit spectroscopy of the inner 10’ (Edelson and Wahl 1987), sources with strong ultraviolet excesses are denoted with a “B” in Column 3, while those with weak excesses have an “R”. The compactness parameter, R, a measure of the degree of resolution at ~11 pm, is presented in Column 4 (see §IIIc). Sources with R more than 3c below unity are labeled with a “E” in Column 5, while those with R greater than this value are labeled with a “C”. On the basis of the spectral indices 022-25 4m and a12-60ym, the infrared spectra are classified with a “Q” (quasar-like or flat) or “D” (dusty or steep) in Column 6 (see §IIIc). Estimates of the 2.2 wm flux due to nuclear nonstellar emission are presented in Column 7 (see §IIIb). On the basis of Dahari’s (1985) observations, interacting objects have an “I” in Column 8, while those with no companions have an “N” (see § Vb). ~79- TABLE 4 As discussed in EM, the major features of the observed 1.2 wm-1.3 mm spectral energy distributions of Seyfert galaxies can be explained as a mixture of four com- ponents: 1) Starlight, which is only important shortward of 3.5 um; 2) Far-infrared emission from cold dust in the disk of the underlying galaxy; 3) Emission from warm dust; and 4) A relatively flat nonthermal power law component with a sharp long-wavelength cutoff near 80 ym. A fifth component, a broad bump centered near 5 pm, is clearly detectable in some spectra, as discussed by EM. However, the lack of uniform coverage between 3.5 and 10 ym prevents a systematic study of this feature with these data. b) Starlight Starlight makes a substantial contribution to the near-infrared spectra of many objects. Continuum emission from galactic stellar populations peak near 1.5 ym and has a roughly Rayleigh-Jeans fall-off to longer wavelengths. Detailed multi-aperture measurements can be used to seperate the extended stellar and compact nuclear source components, but such data are available for only a few objects in this sample (Malkan and Filippenko 1984, McAlary and Rieke 1986). Therefore, alternate methods had to be used to estimate the contribution of starlight to the spectra in these objects. These data give little meaningful information about the stellar population itself. Instead, the goal is to remove its contribution so far as possible to obtain a better estimate of the spectrum of the nuclear source. 1e starlight-corrected nuclear flux was estimated at 2.2 ym because there is virtually complete JHK data on the sample and the ratio of nuclear to stellar flux is expected to be largest at K for these data. These estimates are listed in Table 4. For the galaxies where JHKL data were available, a least squares fit was performed using an unreddened stellar spectrum with normal colors (i.e., JI-H=0.7, H-K=0.2, -~ 80 - and K-L=0.2) and a power law with a spectral index that was allowed to vary to optimize the fit. These estimates are indicated by an “a” in Table 4. Comparison with the uncorrected K fluxes in Table 1 shows that starlight corrections can be important in many of these objects. Where only JHK data were available and the spectrum rose from H to K, it was assumed that all the detected flux might arise from the nuclear source and assigned an upper limit equal to the flux at K. These estimates are indicated by a “b” in Table 4. Where the spectrum peaks at H or is level between H and K, a least squares fit was performed with the stellar spectrum and a power law with spectral index a = —1. If the derived nuclear flux was less than 20% of the total at K, an upper limit of 20% of the K flux was assigned. These estimates are indicated by a “c”. Although our methods are not expected to be very accurate, the results are generally in rough agreement with those found by Malkan and Filippenko (1984) and McAlary and Rieke (1986). The importance of starlight can also be seen in the correlation between nuclear (12 um) luminosity and a1. 1~2.24m. The non-parametric Spearman rank correlation test (Siegel 1956) shows that there is a negative correlation between Ligum and Q11-2.2um at the 98% confidence level. As expected, an increasinigly dominant nucleus raises the 12 ym luminosity and makes the J-K color redder. c) Extended Infrared Emission Thermal emission from cool dust in the galactic disk is present to some degree in the infrared spectra of all normal spiral galaxies and can confuse the determination of the infrared spectrum of the active nucleus. This thermal emission from dust heated by starlight can, in principle, be separated from nuclear emission by its much greater spatial extent. For example, seven of the 38 sources in the IRAS point source catalog (18%) showed resolved structure on angular scales of Z 20” at 12 um. In more distant galaxies, the galactic emission may be unresolved by IRAS, but ~81l-— can be identified by comparing the IRAS 12 wm measurements with ground-based 10.6 ym measurements made with small beams (typically 6” in diameter). This comparison is simplified by the fact that Seyfert galaxies usually have little spectral structure at 10.6 wm (Aitken and Roche 1984). We define the ratio of ground-based to IRAS flux as R, the “compactness parameter”; it is listed in Table 4 for the 18 galaxies in the sample which have the requisite measurements and do not appear to suffer confusion with nearby objects. The uncertainty in R has been estimated by combining the errors in the ground-based and IRAS photometry. Two corrections must be applied to the measurements in computing R. First, the color corrections in the ground-based and IRAS photometry must be reconciled. The ground-based calibration gives equivalent monochromatic fluxes for stellar (Rayleigh-Jeans — vy?) spectra at the effective wavelength of 10.6 um (Rieke, Lebofsky, and Low 1984). No color corrections have been applied to the published Seyfert galaxy measurements nor to the new ones reported here. The IRAS system assumes color corrections appropriate to a v—! power law spectrum. To correct the ground-based 10.6 zm measurements to the same source spectrum, they must be multiplied by 1.22. This correction is a weak function of the spectral index near a = —1. The second correction allows for the differing effective wavelengths of the IRAS and ground-based photometric bands. This correction was determined by fitting a power law to the IRAS 12 and 25 ym measurements and using it to extrapolate the 12 um measurements to 10.6 ym. Sources with R less than unity by at least 3c are considered to be extended, and the remaining objects are defined as compact. These designations are indicated in Table 4 with an “E” and a “C”, respectively. Not surprisingly, all the sources that are seen to be extended in the IRAS scans are also found to be extended with this test. Four sources had classifications at low signal-to-noise: Mkn 841 and 2237+07 were classified as compact, and NGC 7469 and Mkn 530 were classified as extended. _~ 82- The classifications of these objects are not as reliable as the others. The compactness parameter is a useful indicator of the presence of extended thermal emission from dust warmed by starlight in the galactic disk. Seyfert 1 galaxies with relatively luminous nuclei appear more compact in the infrared. Figure 2 shows the correlation of 12 ym luminosity with R. The Spearman rank correlation test shows that the extended sources are less luminous, at the 97% level of significance. Less compact sources also have significantly bluer H-K colors, at the 95% confidence level. Both results are easily understood as the increasing impor- tance of galactic (extended) infrared emission in Seyfert galaxies with less luminous nuclei. Not surprisingly, the extended infrared emission is most common in those galaxies with the lowest redshifts in this magnitude-limited sample. Since these Seyfert galaxies have been more extensively observed at 10.6 ym, the frequency of strong extended infrared emission for a complete, unbiased sample of objects cannot be determined. FIGURE 2 A grey dust grain will reach a temperature of 300 K and hence emit strongly at 10 ym at a distance of roughly 2 pc from a 4x10“4 erg/s (101! Lo) energy source. At the distance of the galaxies in this sample, 2 pc typically corresponds to 0°01. Grains with realistic wavelength-dependant emissivity will reach 300 K significantly further from the energy source, but an unusual wavelength dependence would be required to account for the emission in the “E” galaxies farther than 3” (ie, 2600 pc) from the nucleus. This suggests that the “E” galaxies have an extended source of luminosity in addition to the active Seyfert nucleus. The correlation of a12-60um with compactness parameter, significant at the 99% confidence level, is shown in Figure 2. The far-infrared slopes observed in normal galaxies with high rates of current star formation tend to be steeper than ~ 83 - those of active galactic nuclei (Miley, Neugebauer, and Soifer 1985). Much of this energy is generated in the disk of the galaxy (Persson and Helou 1986), so it would account easily for the correlation of extended emission with infrared spectral index. It is suspected that many of the “D” galaxies without direct measurements of extent also generate a significant portion of their infrared emission through star formation in the disks. Conversely, it is expected that far-infrared measurements with better spatial resolution would show that many of the CfA Seyfert galaxies with “D” spectra actually have flatter nuclear spectra that more closely resemble quasars. c) Nuclear Infrared Emission While both of the remaining components, thermal dust emission and nonther- mal emission from the nucleus, are generally unresolved even with the small beams used in the near-infrared, they have very different spectral shapes. Nonthermal emission tends to produce flat, smooth spectra. The signature of warm dust is a steep, curved spectrum which can peak near 80 ym. EM found that objects with large internal reddenings and other dust indicators usually had steep infrared spectra (a & —2), which are not well-fitted by a simple power law. As can be seen in Figure 1, Seyfert 1 galaxies tend to have flat, power law spectra, while Seyfert 2 galaxies usually have steeper, curved spectra. Only one Seyfert 2 galaxy (Mkn 461, which has a 3.lo detection at 25 ym) appears to have a flat infrared spectrum. This difference can also be seen in a color-color diagram. Figure 3 is a plot of a2.2-25um VS %12-60um- Seyfert 1 galaxies are denoted by triangles, Seyfert 2 galaxies by circles, and bright PG quasars (taken from E86) by stars. An object which lies on the dotted line has a99-954m= @12-60um- For bright, optically selected quasars, E86 found that @29-954m = —1.09 + 0.16 (individual scatter), and @19-60um = —0.69 + 0.44. For Seyfert 1 galaxies, @.2-25ym = —1.15 + 0.29 (individual scatter), and @9-g0um = —1.08 + 0.44. For the Seyfert 2 galaxies, @2.2-25um = —1.56 + 0.34, and @2-¢09um = —1.55 + 0.36. The non-parametric —~ 84-— Kolmagorov-Smirnov (K-S) test for differences in the distribution of a single param- eter in different types of objects (Siegel 1956) shows that a2 9-25ym and a12-60um are distributed differently for Type 1 and 2 Seyfert galaxies at the 99% and 97% level, respectively. Seyfert 2 galaxies tend to have stronger stellar components at 2.2 um than Seyfert 1 galaxies; after removal of this component, the differences in the distribution of a2 2-25ym becomes even stronger. FIGURE 3 The dashed line in Figure 3 delineates a region with a99-95um 2 —1.37 and @12—60m > —1.25, which completely segregates Seyfert 2 galaxies from quasars. All of the optically selected quasars have a2 9-954m 2 —1.37 and a12-60um 2 —1.25, unlike any of the Seyfert 2 galaxies. Any infrared spectrum with a9 9-954m 2 —1.37 and @19-60um = —1.25 is designated as “quasar-like”, and is denoted by a “Q”’ in Table 4 and Figure 3. Any spectrum with either index steeper than these boundary values is designated as “dusty,” and is denoted by a “D”. The four objects which lie in the “quasar-like” region of Figure 3, but have a25-60um < —1.7 OF @60—100um < —2.0 are denoted with a “Q(D)”, because the flat quasar-like spectrum is apparently modified by cold dust emission near ~80 ym. The steep infrared spectra of Seyfert 2 galaxies are produced by the superpo- sition of thermal emission from dust grains at a range of temperatures (EM). The high-frequency Wien cut off causes the spectrum to curve downward steeply from the mid to near-infrared. Seventy per cent of the Seyfert 1 galaxies lie in the upper right “quasar-like” region of Figure 3, indicative of flat power law emission. This clear spectral division from the Seyfert 2 galaxies is good evidence that thermal dust emission dominates the spectra of Seyfert 2 galaxies, while nonthermal radiation is the dominant mechanism for most Seyfert 1 galaxies. Several Seyfert 1 galaxies (e.g. Mkn 231, ~ 85 - NGC 7469, etc.) occupy the “dusty” regions, suggesting that the infrared spectra of these objects are dominated by thermal emission. It is interesting to compare this classification with the observed extent of the Seyfert galaxies at 10.6 wm. Of the 7 “Q” galaxies for which compactness parameters have been measured, none are extended. Of the 10 ”D” galaxies with measured compactness parameters, only one (Mkn 231) is extended. Only one “Q(D)” galaxy (Mkn 530) has a measured compactness parameter. It is extended, but at low a signal noise ratio. This strong correlation between detectable source extension and a “D” spectrum suggests that many of the other Seyfert galaxies with “D” spectra are extende infrared sources. Unfortunately, only three remaining objects (Mkn 334, Mkn 766 and NGC 1144) have 12 wm measurements at better than 5:1 signal-to-noise ratios, so direct confirmation of this conclusion will be difficult. Seyfert galaxies tend to have strong near-infrared excesses compared with nor- mal field galaxies. The average H-K for Seyfert 1 galaxies is 0.68+0.29 (individual scatter), while for Seyfert 2 galaxies, it is 0.52+0.22. By contrast, the average H-K = 0.2+0.1 for field galaxies. IV. LOW-FREQUENCY TURNOVER The CfA Seyfert galaxies are generally strong far-infrared sources (Sg0ym2 1 Jy), with spectral energy distributions which rise from 1 to ~80 ym. However, none were detected above ~500 mJy at 1.3 mm. This means that a dramatic spectral turnover must occur in the far-infrared-submillimeter spectra of these objects. EM studied this turnover in a heterogeneous sample of AGNs, as did E86 for a sample of optically selected quasars. Over half of the objects in those studies were seen to turn over before 100 ym. In EM, E86, and the present study, the turnover frequency (vz) is determined by ~ 86- fitting a parabola in log v-log F, space to the 25, 60, and 100 um IRAS data, and taking the point at which a = 0 as the turnover, provided 4% > 3 THz (100 um) in the observed frame. EM contains a detailed discussion of this method. The turnover frequency is actually a function of just a95-6g0um and ag9-100um- At zero redshift, it is given by the formula 1 2.63 — 1.71(a25—60:m/060—100um) logi4 = 12.67 + (provided the spectrum is concave downward and logy; > 12.48). The turnover frequencies of two objects which were not detected at 100 wm (Mkn 335 and Mkn 841) were measured with higher frequency IRAS data. Table 5 contains the turnover frequencies and other parameters for the 15 unconfused objects with detected turnovers. Source names are given in Column 1, and turnover frequencies (in THz), in Column 2. TABLE 5 Of the 34 unconfused objects detected at three or more IRAS wavelengths, 15 (44%) had turnovers detected shortward of 100 wm. The other 19 unconfused sources had turnovers which could not be determined because they occurred long- ward of 100 ym. However, the 1.3 mm limits require that the spectra of these objects cut off in the far-infrared-submillimeter region. FIGURE 4 Figure 4 is a plot of a25—g0ym Versus 460—100um- A source on the dotted line would have a25—60um=60—100um- The turnover frequency is a function of the spectral indices a25_¢04m and ag0-100um- Sources with detected turnovers lie in the region above and to the right of the dashed line in Figure 4. Sources which lie significantly below the dotted line have strong excesses near 80 ym, probably due ~87- to cold dust emission from the underlying galaxy. For these objects, the far-infrared spectrum of the nucleus may turn over, but it cannot be seen because of emission from the underlying galaxy. Objects with extended far-infrared emission (i.e., those with an “E” in Column 5 of Table 4) are more likely to have undetected turnovers than compact sources. More luminous active nuclei are more likely to show detectable turnovers, and at higher frequencies: E86 found that 6 of 9 quasars (67%) had 4% > 5 THz, while only 2 of 34 of the CfA Seyfert galaxies (6%) did. This indicates that, especially in low-luminosity AGNs, the turnover can be masked by steep-spectrum emission from cold dust in the underlying galaxy. There were no trends for turnover frequency to be correlated with Seyfert galaxy type or the shape of the infrared spectrum. Forty-five percent of the Seyfert 1 galaxies had turnovers detected shortward of 100 um, as did 42% of the Seyfert 2 galaxies. Thirty-eight percent of the objects with “Q” spectra had detected turnovers, as did 50% 18 objects with “D” spectra. There may be a weak correlation between turnover frequency and ultraviolet color. Twelve of the 22 objects with blue ultraviolet colors (55%) had low-frequency turnovers detected, as did 3 of 12 red objects (25%). However, the effect was not significant with the K-S test. The turnover frequency may be used to determine source parameters, when the mechanism responsible for the infrared continuum is known. If it is nonthermal emission from a synchrotron self-absorbed source (as appears to be the case in most Seyfert 1 galaxies), then it occurs at the frequency at which the source becomes optically thick. EM showed that other explanations of the turnover in low-reddening Seyfert 1s (such as free-free absorption in the broad-line region) are unlikely to be correct. One more assumption is required to derive source parameters. If the X-ray flux ~ 88 — is compton scattered radiation, brightness temperatures (Tg), magnetic fields (B), and source sizes (rs) can be calculated (E86). These parameters are tabulated in Column 4-6 of Table 5, respectively. The typical inferred brightness temperature Tp & 3x10! K, implies a source size rs © 3x10!5 cm, and magnetic fields B ~ 300 Gauss. The Schwarzschild radius can be estimated by assuming that the source is accreting at the Eddington limit. These calculations yield infrared source sizes rs © 100 Schwarzschild radii, that is, about the size of the hypothesized accretion disk. The derived brightness temperatures and magnetic fields are also reasonable for the accretion disk region. It is interesting to note that brightness temperatures Tp 3x10! K are also found in the most compact regions of radio selected sources, using different assumptions (Readhead 1986). If the infrared continuum is dominated by thermal emission from nuclear dust (as appears to be the case for Seyfert 2 galaxies), the turnover frequency can be used to determine the minimum characteristic dust temperature (Tz), size (rg), and mass (mq). These quantities are presented in Columns 6-8 of Table 5, respectively. A dust emissivity € « 1/\ is assumed, with mass absorption coefficient of k = 60 cm2/gm at 100 pm (Draine and Lee 1984). The derived size of the thermally- emitting region (discussed further in EM) is a lower limit. It could be much larger if the source is optically thin. For most Seyfert 2 galaxies, typical minimum dust sizes and temperatures of rg +100 pc and Ty 50 K are found. This is warmer than dust in normal galaxies (de Jong et al. 1984), but similar to the temperature of dust in super-luminous infrared galaxies such as Arp 220 (Sanders et al. 1986). The sizes rg2 100 pc suggest that the warm dust resides in the narrow-line region or the galactic disk. Table 3 contains limits to a199-1300zm, based on the 1.3 mm upper limits. The source NGC 1068 has a 199~1300um2 + 2.53, which indicates that dust, and not synchrotron self-absorption is responsible for its cutoff. However, the strongest — 89 - limits which can be set for any other sources are @100-1300pm* + 1.5. While the millimeter data is not yet sufficiently sensitive to discriminate between different causes for the turnover in these objects, they constrain the low-frequency side of the cutoff to be quite sharp. V. RELATION TO OTHER PROPERTIES a) Radio and Optical Luminosity Six centimeter observations of 39 of the 45 CfA Seyfert galaxies detected at 60 um with IRAS were discussed in Paper I. All were detected at 6 cm. The high detection rates (100% at 6 cm and 4500 A, 97% at 60 ym) make it possible to search for luminosity correlations. Figure 5 is a plot of nonthermal radio luminosity as a function of infrared luminosity. Seyfert 1 galaxies are denoted by triangles, and Seyfert 2 galaxies, by circles. The dashed line is a proper least-squares fit to the relation L;, « L*,, for Seyfert 1 galaxies, and the dotted line, for Seyfert 2 galaxies. | FIGURE 5 Lip and L,gq are correlated for both types of Seyfert galaxy, although the correlation appears stronger for Seyfert 1 galaxies. The Spearman rank correlation test indicates that the two fluxes (Sgoum and Sgem) are correlated at the 99.9% confidence level, for both types of Seyfert galaxies. Assuming the power law relation L;, « LS_,, s = 0.90+0.10 for Seyfert 1 galaxies, while for Seyfert 2 galaxies, rad? s=1.06+0.14. Figure 6 is a plot of infrared luminosity as a function of integrated optical luminosity. Seyfert 1 galaxies are denoted by triangles, and Seyfert 2 galaxies, by circles. The dashed line is a proper least-squares fit to the relation L;, « cot for —90- Seyfert 1 galaxies, and the dotted line, for Seyfert 2 galaxies. FIGURE 6 A clear correlation is also seen between Lj, and Lopt, but the fitted regression line is steeper than that for L;, versus L,,g. The correlation in the optical and infrared fluxes is also somewhat weaker. For Seyfert 1 galaxies, s = 1.73+0.13 (significant at the 99.5% confidence level) and for Seyfert 2 galaxies, s = 1.7140.15 (significant at the 99% confidence level). The steep slope seen in this relation, and in the Lyaq — Lopt relation (Paper 1), is consistent with the hypothesis that the nonthermal radio, infrared, and optical emission are all correlated, but that the optical luminosity of low-luminosity Seyfert galaxies are dominated by starlight from the underlying galaxy, which is not correlated with nonthermal luminosity. b) Companions It has been suggested that interactions can trigger star formation, leading to large extended infrared luminosities. Dahari (1985) has classified many of the CfA Seyfert galaxies for the presence of close companion galaxies at nearly the same redshift. As indicated in Table 4, 16 of the galaxies in the sample have close (physically associated) neighboring galaxies, which are denote by “I”. The 15 Seyfert galaxies near which Dahari found no companions are denoted by “N”. Histograms of a19_60m for interacting and non-interacting galaxies are presented in Figure 7. The K-S test indicates that the interacting Seyfert galaxies have significantly steeper far-infrared slopes than the isolated Seyfert galaxies at the 95% confidence level. The 60 wm luminosity (L;,) is also larger for the interacting Seyfert galaxies, at the 95% confidence level. Evidently the galaxy interaction is associated with stronger emission from cool dust, which may be powered by new star formation in the body of the galaxy. ~91- FIGURE 7 VI. INFRARED LUMINOSITY FUNCTION Ideally, the Seyfert galaxy infrared luminosity function (IRLF) ought to be determined from a sample identified in a complete infrared survey such as the IRAS point source catalog. However, it is very difficult to identify a complete sample of Seyfert galaxies with the IRAS data. Completely identified portions of the IRAS survey tend to have a rather small percentage of Seyfert galaxies (Rieke and Lebofsky 1986). De Grijp et al. (1985) obtained a very high detection rate identifying Seyfert galaxies by selecting objects with —0.5 > a95-60ym 2 —1.25. However, Table 3 shows that 33 of the 40 CfA Seyfert galaxies (83%) with measured values or significant limits for a95_¢0ym lie outside this range. This method of selecting Seyfert galaxies produces a sample which is too deficient in high-luminosity blue Seyfert 1 nuclei and in dusty Seyfert 2 galaxies to allow the construction of a reliable IRLF. R78 found good evidence that the ultraviolet-excess selection criteria used in the Markarian Seyfert galaxies and PG quasars biased these samples against the inclusion of red objects. The CfA Seyfert galaxy sample, which used no ultraviolet-excess selection criterion, has a broad range of ultraviolet colors (Huchra, Wyatt, and Davis 1982; Edelson and Wahl 1987), making it ideal for studies of the relationship between ultraviolet excess and infrared spectra. Stronger ultraviolet excesses tend to be found in Seyfert 1 galaxies and AGNs with relatively little contamination from starlight. Figure 8 is a histogram of a@19-¢0ym for objects with strong and weak ultraviolet excesses. Objects with strong ultraviolet excesses tend to have flatter infrared spectra than those with weak excesses. For the blue objects, a19-60um = —1.39 + 0.42 (individual dispersion), while for the red objects, a19-¢9um = —1.51 + 0.45. ~92- The K-S test shows that a19-60ym is distributed differently for the red and blue objects at the 97% confidence level. FIGURE 8 The best alternative, which is used in this paper, is to make infrared measure- ments of a complete, unbiased, optically selected sample. The derivation of the CfA Seyfert galaxy IRLF is based on the modified V/Vm method suggested by Schmidt and Green (1982). Since this method is discussed in detail in Paper I, only a brief summary is given here. The maximum accessible volume (Vi) is the volume of space within which an object could have been detected above both survey limits. The differential IRLF, ®(L;,), is given by 4n S& 1 TAL Ving O(L;,) = summed over all n objects with Lj, in each bin of width AL (Huchra and Sargent 1973). In this equation, 2 is the amount of sky surveyed optically (Q = 2.66 sr), and f is the fraction of optically selected objects observed in the infrared (f = 0.94). The quantity L;, is the monochromatic luminosity (4nr2vF,) at 60 ym. The data were summed using logarithmic bins of width AL = 10°-4, corresponding to one magnitude. The differential IRLF derived for Seyfert 1 and 2 galaxies separately, and for both types together is presented in Table 6 and Figure 9. Column 1 contains the binned values of L;,. Columns 2-3, 4-5, and 6-7 give the space densities and number of objects in each bin for Type 1s, Type 2s, and both types together. TABLE 6 —~93- FIGURE 9 The fact that a bivariate optical-infrared distribution function was used makes the IRLF subject to incompleteness for objects with infrared spectral indices steeper than that defined by the optical and infrared survey limits. Thus, this IRLF can be systematically deficient in objects with ag7_rr < —1.3. As most Seyfert 2 galaxies, and many Seyfert 1 galaxies have infrared spectra this steep or steeper, this effect can be quite severe, and the derived IRLF should be considered a lower limit to the IRLF of all Seyfert galaxies. The IRLF of Type 2 Seyfert galaxies was determined for luminosities of 1042-6 to 104-9 ergs/s. The IRLF of Seyfert 1 galaxies and both types together were determined for luminosities of 104%? to 104° ergs/s. Two Seyfert 1 galaxies (Mkn 231, L;, = 10-73 erg/s, and NGC 5273, Lj, = 104-99 erg/s), and one Seyfert 2 galaxy (Mkn 270, L;, = 104!-% erg/s) have infrared luminosities outside of this range. Nothing can be concluded from the present data about the IRLF of Seyfert galaxies with L;, outside of 1042? to 104-9 ergs/s. Since a significant fraction of the total luminosity of Seyfert galaxies is emitted in the far-infrared, the IRLFs derived above are in effect luminosity functions for the entire active galaxies. The luminosities of Seyfert galaxies with “D” spectra are compared with that of field galaxies (Rieke and Lebofsky 1986) in Figure 10. Based on the arguments given earlier, it is expected that the bulk of their far-infrared luminosity is generated in the disk of the underlying galaxy for these objects. The Seyfert galaxy luminosities are plotted as a histogram above the horizontal line. Upper limits in the “Q” objects are plotted as a histogram below the line. These limits were set equal to the total infrared luminosity longward of 12 ym. The luminosities of the three rejected galaxies are identified with an R. The plot of the IRLF of field galaxies is normalized such that the total number of galaxies with infrared luminosity greater ~ 94 - than 6x104! erg/s (1.5x108 Lo) is equal to the total number of Seyfert galaxies in the sample (including those rejected and those with upper limits for the cold excess). The luminosities of the Seyfert galaxies tend to be significantly higher than would be expected from an unbiased sample of field galaxies. FIGURE 10 The conclusion rests primarily on eight galaxies (NGC 1068, NGC 1144, NGC 4388, Mkn 789, IC 4397, NGC 5674, NGC 7469, and Mkn 533) for which the infrared luminosities are greater than 4x1043 erg/s. All four of these galaxies with measured compactness parameters have R < 0.6, giving direct evidence that a significant amount of luminosity arises from outside the nucleus. They are also all well removed from the “quasar-like” region of Figure 3, providing further evidence that thermal dust emission is much more important than nonthermal emission in these objects. At least 7 additional objects with “D” or “Q(D)” spectra (Mkn 334, 0048+29, Mkn 231, 1335+39, NGC 5940, 1614435, and Mkn 530) have luminosities greater than 4x1043 erg/s, although for various reasons the luminosity of the underlying galaxy is less certain for these objects. If it is assumed that the bulk of their infrared luminosity is produced in their galactic disks (rather than in their Seyfert nuclei), the normalized luminosity function predicts that only 1.4 galaxies in this sample should have far-infrared luminosities larger than 4x1044 erg/s, rather than eight. This is evidence that Seyfert nuclei are more likely to be found in more luminous galaxies. VII. CONCLUSIONS For the first time, we can determine the infrared-millimeter properties of a complete, unbiased sample of Seyfert galaxies. This sample consists of the 48 Seyfert galaxies in the CfA redshift survey. For these galaxies, the following data ~95- have been obtained: IRAS measurements of 46 objects, JHK photometry of 45, L and N photometry of about half of the sample, and 1.4 mm upper limits for all. These observations allow the nuclear continuua between 1.1 wm and 1.3 mm to be studied. Seyfert galaxies have the following characteristics: There is a highly significant trend for the slope of the infrared spectrum to steepen from quasars to Seyfert 1 galaxies to Seyfert 2 galaxies. This is caused by an increasingly large ratio of thermal to nonthermal infrared emission along this sequence. No Seyfert 2 galaxy has an infrared slope as flat as any of the optically selected quasars studied by E86, all of which have a2.9-954m > —1.25 and Q12-60um > —1.37. Half of the Seyfert galaxy spectral energy distributions have sharp low-frequency turnovers around 80 pm. About two-thirds of the Seyfert 1 galaxies have flat, quasar-like infrared spec- tra, indicating that nonthermal radiation is responsible for the observed infrared emission from these objects. The turnovers in the power law spectra of unreddened Seyfert 1 nuclei are attributed to synchrotron self-absorption, and imply source sizes of ~100 Schwarzschild radii. The thermal emission which characterizes all of the Seyfert 2 spectra and one- third of the Seyfert 1 spectra appears to come from dust in the nucleus (in a volume comparable to that of the narrow-line region where it absorbs much of the nuclear ultraviolet continuum), and in the disk of the galaxy (where it is warmed by starlight). In those objects with predominantly thermal far-infrared continua, the turnover frequency gives minimum dust temperatures of ~50 K, warmer than those of normal spiral galaxies, but similar to those of “super-luminous” infrared galaxies like Arp 220. Although Seyfert 2 galaxies emit relatively more thermal flux than Seyfert 1 galaxies, their luminosity functions are not distinguishable up to 10% erg/s. There is still no explanation of why Seyfert 2 nuclei, unlike Seyfert 1 nuclei, are never found to be essentially dust-free. Seyfert nuclei appear to reside ~ 96 - in galaxies with higher than average far-infrared disk luminosities. The authors would like to thank W. Rice for invaluable help with the IRAS data, and M. J. Lebofsky and W. F. Kailey for assistance. in obtaining some of the measurements reported. J. Huchra kindly released the list of CfA Seyfert galaxies in advance of publication. Analysis of IRAS data was supported by NASA grant NAS 7-918. Millimeter-wave astronomy at Caltech is supported by NSF grant AST 84-12473. Infrared astronomy at Steward Observatory is partially supported by the National Science Foundation. This work was partially supported by NSF grant AST 83-14134. ~97- TABLE 1 CfA SEYFERT GALAXY NEAR INFRARED DATA Source Si2ym (mJy) Sium (mJy) S2.2um (mJy) Date Mkn 334 14434 0.7 208+ 10 275+ 14 15 Mar 84 0048+29 Wilt 11 1454 14 1564 1.5 13 Sep 84 Mkn 993 1284+ 06 1424 0.7 151+ 08 Mkn 573 220+ 2.0 295+ 30 263+ 2.4 26 Sep 85 0152+06 60+ 0.4 80+ 0.5 74+ 0.5 16 Sep 84 NGC 1144 1224 24 29424 3.0 315+ 36 04 Mar 85 Mkn 1243 58+ 0.6 744 0.7 79+ 08 Mar 86 NGC 3079% 31024 2.2 7134 40 919+ 5.0 14 Apr 84 NGC 3362 5.8+ 0.5 68+ 0.6 5.2+ 0.5 Mar 86 1058+45 Mkn 744® 198+ 20 25924 20 27824 2.4 15 Apr 84 NGC 3982? 92+ 0.9 124+ 1.2 14 Apr 84 NGC 4235 2114 2.00 354+ 30 299+ 24 16 Apr 84 Mkn 766 23.14 20 4264 30 5014 40 Mar 86 NGC 4388 308+ 20 48524 25 46924 2.5 04 Mar 85 NGC 5033 44.14 3.5 6164 5.0 478+ 3.0 04 Mar 85 Mkn 789 56.624 0.5 784 06 7.324 0.5 04 Mar 85 1335+39 7.724 0.5 81+ 0.6 79+ 0.66 Mar 8&6 NGC 5252 1164 10 12024 10 161+ 09 04 Mar 85 Mkn 2662 40+ 0.5 88+ 1.0 8.74 0.8 04 Mar 85 NGC 52737 14624 10 250+ 15 2314 15 04 Mar 85 Mkn 461° 1229+ 08 1744 10 134+ 09 04 Mar 85 IC 4397 69+ 0.6 744 0.5 6.324 05 Mar 86 NGC 5674 15.74 13 19524 14 16124 1.0 Mar 86 Mkn 686 111+ 08 1574 10 1582+ 1.0 15 Apr 84 NGC 5929? NGC 5940 44+ 0.3 68+ 0.5 8.7+ 0.6 Mar 8&6 1614435 3.64 0.2 68+ 0.7 6.9 + 0.7 15 Mar 84 2237+07* 1134 06 1784+ 09 19424 1.0 13 Sep 84 Mkn 533° 1014 05 15624 08 2424 2.4 NGC 7682 104 11 1494 14 9524 0.9 15 Mar 84 ® The following objects have been observed at 3.5 or 10 pm: NGC 3079 (S35 = 7346 mJy, Sio. = 210+ 20 mJy), Mkn 744 (S3.5 = 3143 mJy), NGC 3982 (Sio.6 < 44 mJy), NGC 4235 (S3,5 = 22+2 mJy), 144 mJy), Mkn 266 (S35 = 7+1 mJy), NGC 5273 (S35 = 1642 mJy) Mkn 461 (Sas < 15 mJy), NGC 5674 (S35 < 15 mJy), NGC 5929 (Sto.6 < 53 mJy), 2237407 (S35 = 16.441.7 mJy, Siog = 82412 mIy), Mkn 533 (S3.5 = 46.444.6 mJy, Sio.6 = 325+ 25 mJy). —~ 98 - TABLE 2 CfA SEYFERT GALAXY Far INFRARED AND MILLIMETER DATA Source Si2um (mJy) So5um (mJy) Séoum (mJy) Sioonm (mJy) S1.amm (mJy) Mkn 334 212+ 34 1070+ 76 4310 + 232 4660+ 494 < 699 Mkn 335 341+ 42 420+ 99 4444 49 < 323 < 852 0048+29 186+ 38 1384 41 931+ 80 1990 + 250 < 834 I Zw 1 540 + 34 1260 + 126 2130+ 298 243804 243 < 672 Mkn 993 < 93 < 102 260+ 49 1210+ 168 < 465 Mkn 573 198+ 46 806 + 73 1200+ 93 1270+ 1738 < 915 0152+06 < 104 < 144 467+ 353 1240+ 174 < 591 Mkn 590 169+ 44 2474+ 44 4385+ 61 1420+ 183 < 309 NGC 1068 38300 + 2300 88600 + 5210 186000 + 18600 239000 + 23900 < 306 NGC 1144 307 + 25 680 + 52 5400 + 288 11300 + 1150 < 495 Mkn 1243 < 103 < 160 328+ = 46 1020+ 162 < 615 NGC 3079 12504 75 20404 122 424004 5950 87600+ 8760 < 564 NGC 3227 562+ 40 1750+ 175 8030 + 784 16900 + 2370 < 738 NGC 3362 < 615 1058+45 130+ 32 198+ 34 631+ 60 1530 + 96195 < 900 NGC 3516 384+ 35 922+ 57 1840+ 107 2160+ 239 < 744 Mkn 744 < 537 NGC 3982 457+ 29 755+ 84 6730 + 683 152004 1550 < 1230 NGC 4051 780 + 78 1420+ 142 8160+ 816 204004 2860 < 456 NGC 4151 2080+ 104 46004 230 6720+ 336 8600+ 430 < 318 NGC 4235 < 121 < 174 374+ 83860 736+ 136 < 468 Mkn 766 431+ 44 1430+ 89 4010+ 242 5060+ 532 < 1431 Mkn 205 < 70 80+ 22 290+ 838 1310+ 9164 < 933 NGC 4388 1000+ 60 3370+ 214 107004 1070 173004 2440 < 1281 Mkn 231 1820+ 109 8560+ 514 333004 3330 301004 3010 < 342 NGC 5033 828+ 47 995+ 106 136004 1310 42400+4 4360 < 462 Mkn 789 =< 146 610+ 62 3720 + 221 53304 571 < 1350 1335+39 1324+ 34 168+ 38 1070+ 80 2630+ 300 < 930 NGC 5252 < 119 < 167 4254 55 750 + 130 < 816 Mkn 266 306+ 8638 976+ 70 7320 + 396 10400+ 1090 < 1041 Mkn 270 =< 85 < 78 117+ 39 < 350 < 1131 NGC 5273 13424 37 2424+ 38 991+ 87 2030+ 248 < 1185 Mkn 461 =< 109 443+ 56 456 + 106 < 1296 Mkn 279 199+ 32 289 + 27 1200+ =679)=— 1970 228 < 1548 IC 4397 1644+ 37 1124 37 1680+ 110 3130+ 345 < 735 NGC 5548 343 + 40 765 + 60 1110+ 87 1800+ 211 < 1473 NGC 5674 128+ 31 273 + 44 1630+ 104 3560+ 382 < 765 Mkn 817 357 + 32 1240+ 70 2340+ 1384 226024 242 < 1269 Mkn 686 < 82 < 80 605+ 50 1800 + 208 < 1098 Mkn 841 198+ 30 4534 45 476+ 50 < 287 < 762 NGC 5929 360 + 33 1570 + 89 9450 + 491 12000 + 1220 < 801 NGC 5940 124+ 30 1124 32 7944+ 63 1820+ 214 < 822 1614+35 < 80 < 72 48i+ 50 1790 + 205 < 1050 2237+07 1440+ 42 392+ 52 900+ #76 1270+ 39186 < 303 NGC 7469 1300+ 130 5500+ 550 26700 + 3740 3440024 6190 < 768 Mkn 530 180+ 42 191+ 56 856+ 79 2140+ 254 < 585 Mkn 533 720 + 101 1930 + 231 5470 + 766 8190+ 1450 < 903 NGC 7682 < 507 —~ 99 - TABLE 3 CfA SEYFERT GALAXY SPECTRAL INDICIES Source 2.2-25um @12~60pm 025—60pm Sob aiinm Mkn 334 —1.51 + 0.04 —1.87 + 0.09 —1.59 + 0.10 —0.15 + 0.22 > 0.70 Mkn 335 —0.72 + 0.10 —-0.12 + 0.11 0.02 + 0.29 > 0.31 0048+29 —0.90 + 0.12 —1.00 + 0.13 —2.18 + 0.31 —1.49 + 0.29 > 0.30 I Zw 1 —1.24 + 0.06 —0.85 + 0.09 —0.60 + 0.19 © —0.26 + 0.33 > 0.46 Mkn 993 > —0.80 < —0.55 < -0.91 —3.01 + 0.46 > 0.33 Mkn 573 —1.41 + 0.05 —-1.12 + 0.14 —0.45 + 0.14 —0.11 + 0.30 > 0.09 0152+06 > -1.25 < —0.88 < —1.25 —1.91 + 0.35 > 0.24 Mkn 590 —0.76 + 0.07 —0.59 + 0.18 —0.65 + 0.26 —2.32 + 0.37 > 0.55 NGC 1068 —2.11 + 0.05 -0.98 + 0.07 —0.85 + 0.13 —0.49 + 0.28 > 2.53 NGC 1144 —1.27 + 0.06 —1.78 + 0.06 —2.37 + 0.10 —1.45 + 0.21 > 1.17 Mkn 1243 >—-1.27 < —0.65 < —0.70 —2.22 + 0.42 > 0.15 NGC 3079 —1.28 + 0.03 —2.19 + 0.09 —3.47 + 0.16 —1.42 + 0.33 > 1.91 NGC 3227 —1.29 + 0.06 —1.65 + 0.07 —1.74 + 0.16 —1.46 + 0.33 > 1.16 NGC 3362 1058+45 —0.98 + 0.15 —1.32 + 0.22 —1.73 + 0.31 > 0.17 NGC 3516 —0.89 + 0.05 —0.97 + 0.07 —0.79 + 0.10 —0.31 + 0.23 > 0.38 Mkn 744 NGC 3982 —1.70 + 0.06 —1.67 + 0.07 —2.50 + 0.17 —1.59 + 0.28 > 0.94 NGC 4051 —1.27 + 0.06 —1.46 + 0.09 —2.00 + 0.16 —1.79 + 0.33 > 142 NGC 4151 —1.31 + 0.04 —0.73 + 0.04 —0.43 + 0.08 —0.48 + 0.14 > 1.25 NGC 4235 > —0.75 < -—0.63 < —0.74 —1.33 + 0.48 > 0.12 Mkn 766 —1.38 + 0.04 —1.39 + 0.07 —1.18 + 0.10 —0.46 + 0.23 > 0.45 Mkn 205 -0.75 + 0.11 < —0.82 —1.47 + 0.33 —2.95 + 0.36 > 0.09 NGC 4388 -1.77 + 0.03 —1.47 + 0.07 —1.32 + 0.13 —0.94 + 0.33 > 0.96 Mkn 231 —1.61 + 0.05 —1.81 + 0.07 —1.55 + 0.13 0.20 + 0.28 > 1.69 NGC 5033 —1.25 + 0.05 —1.74 + 0.07 —2.99 + 0.16 —2.23 + 0.28 > 1.70 Mkn 789 —1.83 + 0.05 < -—1.99 —2.07 + 0.13 —0.70 + 0.23 > 0.50 1335439 —1.26 + 0.09 —1.30 + 0.15 —2.11 + 0.24 —1.76 + 0.26 > 0.37 NGC 5252 > —0.98 < —0.73 < —0.96 —1.11 + 0.42 > —0.08 Mkn 266 -1.95 + 0.05 —1.97 + 0.08 —2.30 + 0.10 —0.69 + 0.22 > 0.85 Mkn 270 >-—0.66 < —0.03 < —0.16 —2.67 NGC 5273 —0.97 + 0.07 —1.24 + 0.16 —1.61 + 0.20 —1.40 + 0.29 > 0.17 Mkn 461 —0.90 + 0.11 < -0.81 —1.50 + 0.35 —0.06 + 0.50 > —0.47 Mkn 279 —1.05 + 0.06 —1.12 + 0.10 —1.63 + 0.13 —0.97 + 0.25 > 0.06 IC 4397 —1.31 + 0.10 —1.45 + 0.13 —2.75 + 0.25 —1.22 + 0.24 > 0.52 NGC 5548 —1.11 + 0.05 —0.73 + 0.09 —0.43 + 0.13 —0.95 + 0.27 > 0.04 NGC 5674 —1.17 + 0.07 —1.58 + 0.14 —2.04 + 0.18 —1.53 + 0.24 > 0.56 Mkn 81 —1.32 + 0.03 —1.17 + 0.06 ~—0.73 + 0.09 0.07 + 0.23 > 9.19 Mkn 686 > -—0.69 < —1.20 < —2.24 —2.13 + 0.27 > 0.16 Mkn 841 —1.13 + 0.05 —0.55 + 0.11 —0.06 + 0.17 > 1.06 NGC 5929 —2.03 + 0.06 —2.05 + 0.09 —0.47 + 0.21 > 1.01 NGC 5940 —1.06 + 0.10 —1.15 + 0.14 —2.24 + 0.30 —1.62 + 0.27 > 0.27 1614435 > —1.00 < —1.07 < —2.08 —2.57 + 0.30 > 0.17 2237+07 —1.24 + 0.05 —1.16 + 0.17 —0.95 + 0.18 —0.67 + 0.32 > 0.51 NGC 7469 —1.57 + 0.06 —1.88 + 0.11 —1.80 + 0.19 —0.50 + 0.44 > 1.41 Mkn 530 —0.84 + 0.12 —0.97 + 0.14 —1.71 + 0.31 —1.79 + 0.29 > 0.46 Mkn 533 —2.23 + 0.09 —1.26 + 0.12 -—1.19 + 0.21 —0.79 + 0.44 > 0.79 NGC 7682 — 100 - TABLE 4 CfA SEYFERT GALAXY SPECTRAL PARAMETERS Source Type UVX R Ext. SED Nuclear Flux (mJy) Int. Mkn 334 2 B D < 28> Mkn 335 1 B 0.78 + 0.18 Cc Q 718 N 0048+29 1 B Q(D) < 16> I Zw 1 1 B 0.81 + 0.10 Cc Q 613 Mkn 993 2 R < 15> N Mkn 573 2 R D 9° N 0152+06 2 R 3° Mkn 590 1 R . Q(D) 162 I NGC 1068 2 B 0.66 + 0.05 E D 302? NGC 1144 2 R D < 31> Mkn 1243 1 B < 8 NGC 3079 2 B 0.22 + 0.03 E D 52? I NGC 3227 2 R 0.70 + 0.06 E D 413 I NGC 3362 2 R < 1° N 1058+45 2 R N NGC 3516 1 R 0.84 + 0.13 Cc Q 433 I Mkn 744 1 R 18? I NGC 3982 2 R <0O.11 E D < 12> N NGC 4051 1 B 0.55 + 0.06 E D 57° N NGC 4151 1 B 0.94 + 0.06 C Q 155° N NGC 4235 1 R 7? N Mkn 766 1 B D < 50° N Mkn 205 1 B 8° NGC 4388 2 R 0.61 + 0.05 E D 178 I Mkn 231 1 R 1.23 + 0.15 C D 175° NGC 5033 1 R 0.07 + 0.03 E D 6? N Mkn 789 1 B D 3° 1335+39 2 B D 2° N NGC 5252 2 R < 16> N Mkn 266 2 R 43 I Mkn 270 2 R < 3 N NGC 5273 1 R Q 7 N Mkn 461 2 B < 3 Mkn 279 1 B 83 IC 4397 2 B D < is NGC 5548 1 B 1.10 + 0.15 C Q 48? N NGC 5674 2 R D 38 N Mkn 817 1 B Q 43° Mkn 686 2 R < 16> N Mkn 841 1 B 1.03 + 0.23 Cc Q 278 NGC 5929 2 R < 0.23 E I NGC 5940 1 B Q(D) < g 1614+35 1 R 5S 2237+07 1 B 0.85 + 0.28 C Q < 19> N NGC 7469 1 B 0.72 + 0.09 E D(D) 118® I Mkn 530 1 B 0.53 + 0.16 E Q 128 N Mkn 533 2 B 0.65 + 0.11 E D < gb I NGC 7682 2 R N - 101 - TABLE 5 DERIVED TURNOVER PARAMETERS 4 Ts Th B Ta Ta ma Source (THz) (10!2em) (10!°K) (G) (K) (10cm) (Mo) Mkn 334 4.5 3900 3.3 930 55 160 730 Mkn 335 7.9 720 4.2 1000 97 25 130 I Zw 1 3.7 8300 4.2 490 44 430 9400 Mkn 573 4.1 1500 4.5 470 50 79 520 NGC 1068 3.0 8800 1.8 2000 36 340 100 NGC 3079 3.5 3100 2.3 1500 43 120 54 NGC 3516 3.6 1100 4.3 440 44 58 190 Mkn 766 3.6 2600 3.7 620 44 130 470 Mkn 231 5.1 22000 2.1 2700 62 700 2100 Mkn 789 3.9 5700 3.7 670 47 270 2300 Mkn 461 4.6 720 5.2 390 56 37 270 IC 4397 3.5 1800 4.5 400 42 100 630 Mkn 817 5.0 3500 3.6 880 61 150 1000 Mkn 841 7.3 1200 4.3 910 88 44 360 NGC 7469 4.0 9400 2.4 1600 48 360 680 — 102 - TABLE 6 SEYFERT GALAXY Far INFRARED LUMINOSITY FUNCTION Type 1 Type 2 Both Types log(Lir) log(®) N log(®) N log(®) N erg/s Gpc~3mag~! Gpce~3mag7! Gpc~3mag7? 42.2-42.6 4.47 2 eee 0 4.47 2 42.6-43.0 3.81 1 4.43 3 4.52 4 43.0-43.4 3.79 2 4.35 5 4.46 7 43.4-43.8 3.51 7 4.39 5 4.44 12 43.8-44.2 3.64 6 3.17 2 3.77 8 44.2-44.6 2.98 2 3.37 2 3.52 4 44.6-45.0 2.81 2 2.89 3 3.15 5 ~- 103 - REFERENCES Aitken, D. K. and Roche, P. F. 1985, M. N. R. A. S., 213, 777. Angel, J. R. P. and Stockman, H. S. 1980, Ann. Rev. Astr. Ap., 18, 321. Beichman, C. et al. 1984, IRAS Point Source Catalog. Campins, H., Rieke, G. H., Lebofsky, M. J. 1985, A. J., 90, 541. Cruz-Gonzalez, I. and Huchra, J. 1984 , A. J. , 89, 441. Cutri, R. M., Rieke, G. H., Lebofsky, M. J. 1985, Ap. J., 287, 566. Cutri, R. M., Rudy, R. J., Rieke, G. H., Tokunaga, A. T. and Willner, S. P. 1984, Ap. J., 280, 521. Dahari, O. 1985, A. J., 90, 1772. de Grijp, R. H. K., Miley, G. K., Lub, J., de Jong, T. 1985, Nature, 314, 240. de Jong, T. et al. 1984, Ap. J. Lett., 278, L67. Devereux, N. A., Becklin, E. E., and Scoville, N. Z., Ap. J., in press. Edelson, R. A. 1986, Ap. J., 309, in press (E86). Edelson, R. A. 1987, Ap. J., 313, in press (Paper 1). Edelson, R. A., and Malkan, M. A. 1986, Ap. J., 308, 59 (EM). Edelson, R. A. and Wahl, T. 1987, Ap. J Supp.. in preparation. Kollatschny, W. and Fricke, K. J. 1984, Astr. Ap., 135, 171. Huchra, J, and Berg, R. 1987, Ap. J., in press. Huchra, J. P., Wyatt, W. F., and Davis, M. 1982, Astron. J., 87, 1628. Huchra, J., and Sargent, W. L. W. 1973, Ap. J., 186, 433. Malkan, M. A. and Filippenko, A. V. 1983, Ap. J., 275, 477. Markarian, B. Y. 1972, Astrofizika, 8, 165. McAlary, C. W. and Rieke, G. H., 1986, Ap. J., in press. Neugebauer, G., Soifer, B. T., and Miley, G. K., Ap. J. Lett., 295, L27. Neugebauer, G., Miley, G. K., and Soifer, B. T., and Clegg, P. E. 1986, Ap. J., 308, in press. -— 104 - Readhead, A. S. C. 1986, private communication. Rieke, G. H. 1978, Ap. J., 266, 550 (R78). Rieke, G. H., and Lebofsky, M. J. 1979, Ann. Rev. Astr. Ap., 17, 477. Rieke, G. H., and Lebofsky, M. J. 1986, Ap. J., 304, 376. Sanders, D. et al. 1986, preprint. Schmidt, M., and Green, R. F. 1983, Ap. J., 269, 352. Siegel, S. 1956, Nonparametric Statistics for the Behavioral Sciences (McGraw- Hill). Wynn-Williams, C. G., Becklin, E. E., Scoville, N. Z., Ap. J., 297, 607. ~ 105 - FIGURE CAPTIONS FIG 1.—a) 1-100 ym vF, plots for the CfA Seyfert 1 galaxies. The objects are presented in approximate order of the ultraviolet excess, with the bluest objects at the top. Data is plotted in log vF, versus v, so a horizontal line would have a = —1 (i.e., equal power per octave of frequency). For the purpose of display only, the IRAS data is grey-shifted by the “compactness parameter”, R, discussed in § IIIc, to account for the effects of emission from the underlying galaxy. One-sigma error bars are shown, and a factor of two is shown on the lower left for scale. An arbitrary scaling constant is added to the data for display purposes. FIG 1.—b) Same as Figure la, except for the Seyfert 2 galaxies. FIG 2.—Correlation of compactness parameter (R) with 12 wm luminosity and ©12-60um are presented in panels a and b. These parameters are correlated at the 97% and 99% confidence level, respectively. FIG 3.—Color-color diagram of a2.2-254m versus 012-60um- Seyfert 1 galaxies are denoted by triangles, Seyfert 2 galaxies by circles, and bright quasars by filled stars. An object which lies on the dotted line has a29~254m= 12-60um- The dashed line delineates a region with a22-954m 2 —1.37 and a12-60um 2 —1.25. All quasars and 70% of the Seyfert 1 galaxies lie within this region, and all Seyfert 2 galaxies lie outside. FIG 4.—Color-color diagram of @25_ 60m Versus 0g0—100um- Seyfert 1 galaxies are denoted by triangles, Seyfert 2 galaxies by circles, and bright quasars by filled stars. An object which lies on the dotted line has a25-60um= @60—100pm- Sources with detected turnovers lie in the region above and to the right of the dashed line. FIG 5.—Plot of L,qq versus Lj, for the 39 objects observed at 60 um and 6 cm. Seyfert 1 galaxies are denoted with a triangle, while Seyfert 2 galaxies are denoted by circles. The dashed line is a proper least-squares fit to the relation L;, « L*,, — 106 - for Seyfert 1 galaxies, and the dotted line, for Seyfert 2 galaxies. FIG 6.—Plot of Loy versus L;, for the 45 objects detected at 60 um. Seyfert 1 galaxies are denoted with a triangle, while Seyfert 2 galaxies are denoted by circles. The dashed line is a proper least-squares fit to the relation L;, « Lo,, for Seyfert 1 galaxies, and the dotted line, for Seyfert 2 galaxies. FIG 7.—Histograms of a12-60ym for Seyfert galaxies classified by Dahari (1985) as interacting (I) or non-interacting (N). The two distributions are different at the 95% significance level. FIG 8.—Histogram of a12-60ym based on ultraviolet excess. The top panel contains the data for “blue” objects (i.e., those with strong ultraviolet excesses), and the bottom panel, for “red” objects. The two distributions are different at the 95% confidence level. FIG 9.—Infrared luminosity function of Seyfert galaxies. The top panel contains the IRLF of Seyfert 1 galaxies, the middle panel, that of Seyfert 2 galaxies, and the bottom panel shows the IRLF of both types together. Error bars are lo errors. The error bars are lo Poisson statistical errors. FIG 10.—Normalized IRLF of “D” Seyfert galaxies. Integrated infrared lumi- nosities, LZ, are given in units of 1010 Lo. The curved line shows the field galaxy luminosity function (Rieke and Lebofsky 1986) normalized to the number of galaxies in the CfA Seyfert galaxy sample. The histogram above the horizontal line at N = 0 shows the luminosity distribution of the “D” Seyfert galaxies. The histogram below the line shows the distribution of upper limits (indicated with a “<”) for the “quasar-like” objects and also the luminosities (indicated with an “R”) of the three rejected galaxies discussed in the text. v F,, (arb. scale) B00 J Min 335 a Min 641} /7 Rees Min 270 x ee NGC 554B sao Min 817 7 NGC 1 IZ@wi aN, NGC suo \ AA 2237+07 i ee Min 789 *. qe . ‘fF Min 530 antl NGC 4151 7 *~4_ - z [x2 | eet Garewal Soe eer ee 4 Frequency (Hz). — 107 - v F, (arb. scale) Figure la NGC 3227 0048+29 NGC 4051 on “yt Min 768 a NGC 3516 os Min 231 Mim 590 ee Min 744 NGC rN OS a \. 1614+35 NGC “\. y NGC 4235 . hei g’h bd “°F / [x2 10! 1045 10% Frequency (Hz) v F,, (arb. scale) LoL a Try TT IC 4397 Mkn 334 Min 533 jf Min 461 a. Of YN si 1335+39 NGC 1068 aa v F,, (arb. scale) Mim 266 NGC 3982 Mim 573 NGC 4388 NGC 1144 [et ] x2 Frequency (Hz) — 108 - Figure 1b 0152+06 NS. oe on % Pp NGC 3362 ™. 1058+45 Ny NGC con \_; 7 . 3 e NGC 3079 NGC 5929 : NGC 7682 a Min 686 \. po y Min 270 a / a a yc: > ope NGC 5252, 5. 5-3°- I ze areVPPT MrOPOPPYON rer re 10 10°° 10% Frequency (Hz) 1078L - E A J F A Syi A , “y @ Ssy2 a 1 @ 10%4 ° a - F & L A J on i" var N J 3 A ° he L 7 4 L 7 a4 | oe va “10#L a @ a 4 : oe 1 e a @ 4 42 z 10 E Pe A 7 r ae i ! t 1 “ J ob a A a" “7 1 on | EE: - as | e-" “ ] Le] ””7 4 L A Pad - L e ue e yy ae A 4 ve 4 —2r a 7 a ® 7 a “ it at j i] j 3 0 0.2 0.4 0.6 0.8 R -— 109 - Figure 2 ~ 110 - ® Sye - ¥#& QSO 25pm O23 2um Figure 3 -lll- 100um S60um Figure 4 -—112- Figure 5 Kt q t pobre v v v pepeteer q 7 v prereroe t prrrre rT oad “oN comeny _ weN 3 c “AN J mN 4 a waN . a mM a) 4 . ~\ J N ee C Ne, J C < \. < @ 4 t rs \ | J Vi. L we : #¢4 ° E @ 4 3 qo. 4 : — \ we, -_ L @ Not, J e. . E XN q c @ N fs be — QJ AN <4 a no NM N 4 <q \ ‘N = < @ <q Na ra 1 rT it [i coe | i bepet at i L. bectr tt lL tL leper 2 1D ws oD oO = aS a a i) © (o>) © ei —_ ol i=) = © ol Lag (ere/s) 10°” 102° 107° — 113 - CT T TTT TTT T T TTT TTT a | A Sy 1 a @® sy2 / r / A . A ye J @ ® / 7 L A. P - | eo A, c "4 ; q ‘Al& 4 L. e > 2 “ pe _ r OC’ A ; - @ be, 4 5 @ ’ @ - A/ ®@ 4 / - fF eS “ - ay ‘a 7 [ /@ ; Lf i i. fs [oe eee | cil R i. 1 : oo oo | cil 1073 1044 ~114- wiley ungg” spoalqo eT N eddy, syoelqo OT J ad4], n | n | ‘ | Jequinn JequiInn Figure 7 — 115 - wiley] winl9g 0 spalqo 71 a adéy, spalqo 02 gq edéy, Jequiny I9quiny Figure 8 (Gp mag!) 107 1000 -— 116 - | tT pb pg BB: -117- RI< N Figure 10 [OO lO — 118 - Chapter Siz FAR-INFRARED VARIABILITY IN ACTIVE GALACTIC NUCLEI * * Written in collaboration with M. Malkan -— 119 - ABSTRACT Pointed IRAS observations were used to search for far-infrared variability in 22 active galaxies. Significant variability was seen in a number of objects, down to the levels of repeatability with which IRAS can measure fluxes. The far-infrared fluxes of three highly polarized objects (“blazars” ) varied by about a factor of two on time scales of a few months. In contrast, no convincing cases of variability greater than ~30% were found in any of the normal quasars or Seyfert galaxies studied. Thus, it appears that blazars may be strongly variable at all wavelengths, but normal active galaxies are systematically less variable in the far-infrared than at higher frequencies. ~— 120 - I. INTRODUCTION The infrared emission from active galactic nuclei (AGNs) with high optical polarization (so-called “blazars”, that is, BL Lac objects and highly polarized quasars) is generally believed to be nonthermal, possibly beamed, radiation from a compact source. Blazars are known to vary rapidly at all wavelengths at which they have been observed from X-rays to radio (see, e.g., Bregman et al. 1986). In particular, Ennis et al. (1982) and O’Dea et al. (1985) found that they have typical variability time scales of months at millimeter wavelengths. Robson et al. (1986) find that the infrared-millimeter spectrum of 3C 273 varies coherently over time scales of months to a year. This is strong evidence that these objects are dominated by nonthermal emission, because a thermally emitting region could not vary this rapidly. The nonthermal nature of the infrared continuum is confirmed by the high degree of linear polarization it frequently shows (Angel and Stockman 1980). Furthermore, Impey and Neugebauer (1987) found that radio selected AGNs (mostly blazars) tend to have flat far-infrared spectra (@2.2-25um © —1; Fv « v®), characteristic of nonthermal synchrotron radiation. The far-infrared spectra of optically selected quasars (with @2.2-254m = —1.09, Edelson 1986, hereafter referred to as E86) and luminous Seyfert 1 galaxies (@2.9-95ym = —1.15; Edelson, Malkan, and Rieke 1987, EMR hereafter) also tend to be flat, suggesting a similar nonthermal origin. However, normal quasars and Seyfert 1 galaxies are less violently variable than blazars. Nonetheless, X-ray, ultraviolet, optical, and near-infrared variablity have been well documented (e.g., Halpern 1982, de Bruyn, Readhead, and Sargent 1987, Cutri et al. 1985, Neugebauer et al. 1986). In particular, Cutri et al. (1985) found evidence that the variability in normal quasars becomes weaker from the optical to the near-infrared, though Neugebauer et al. (1986) found that the spectral index stayed relatively constant during outbursts. Unfortunately, normal quasars and Seyfert 1 galaxies are not strong far-infrared and -121- millimeter sources, and only one of these objects (NGC 4151) has been monitored at wavelengths longer than 3.5 ym (Rieke and Lebofsky 1982). Although this variability is suggestive of nonthermal emission, some of it could also be attributed to variable thermal emission from an accretion disk or hot dust grains. In spite of the spectral evidence that their infrared continua are predominantly nonthermal, the normal quasars and Seyfert 1 galaxies are known to have low (but significant) continuum polarization (Stockman, Moore, and Angel 1983, Martin et al. 1983). In sharp contrast, the continua of Seyfert 2 galaxies are not known to vary at any wavelength. Furthermore, Seyfert 2 galaxies tend to have very steep far-infrared spectra (@.9-25um = —1.56), characteristic of thermal dust emission (EMR). Edelson and Malkan (1986, EM hereafter) and EMR found that low-luminosity Seyfert 1 galaxies can also have steep spectra, indicative of thermal emission. Thus, the observed spectral shapes suggest (but do not prove) that the far-infrared spectra of blazars, normal quasars, and luminous Seyfert 1 galaxies tend to be dominated by nonthermal emission, while the far-infrared spectra of Seyfert 2 galaxies and low-luminosity Seyfert 1 galaxies are dominated by thermal emission from dust. EM and EMR showed that infrared thermally emitting regions in AGNs must be more than tens of light-years across, and thus cannot vary detectably on typical observing time scales (i.e., shorter than a decade). However, in AGNs with far- infrared emission which is predominantly nonthermal, it probably arises in a very small volume. EM, E86, and EMR found evidence that the far-infrared emission from quasars and luminous Seyfert 1 galaxies is dominated by radiation from a synchrotron self-absorbed source with a brightness temperature of T 3x10! K, implying source sizes of order a light-day. Since the shortest allowed variability time scale for nonthermal emission in the infrared is thousands of times shorter than that for thermal emission in AGNs, variability is a potentially decisive probe of the nature of the far-infrared continuum. Detected variability is proof of a nonthermal — 122 - continuum, but unfortunately the converse is not true: a constant source is not necessarily thermal. In this paper, pointed observations made with the Infrared Astronomical Satel- lite (IRAS) are analyzed to determine the degree of variability of a heterogeneous sample of AGNs at far-infrared wavelengths. The sample and their IRAS measure- ments are discussed in the next section. The detection and analysis of variability is addressed in § III. The implications of variations are discussed in § IV. A concluding discussion and summary of the major results of this paper is given in § V. Throughout this paper, values of Hy = 75 km/s/Mpc and qo = 0 are assumed. II. SAMPLE AND DATA The data for this study are ~400 pointed IRAS observations of 22 bright AGNs, which include most of the known AGNs detected with signal to noise ratio greater than 3 in at least four pointed IRAS observations, with at least one week between the first and last observation. Table 1 contains source information for the sample. The source name is given in Column 1, and the IRAS positional name, in Column 2. The resulting heterogeneous sample used in this survey consists of 6 blazars (denoted with a B in Column 3 of Table 1), 5 normal quasars (denoted with a Q) 9 Seyfert 1 galaxies (labeled with a 1), and 2 Seyfert 2 galaxies and other low-luminosity AGNs (denoted by a 2). TABLE 1 Raw data from IRAS pointed observations were combined and mapped and source extractions performed in the standard fashion (Young ez al. 1986). Positions of target AGNs and field stars (used for calibration) were taken from these maps. Data for individual pointed observations were also mapped seperatly, and source - 123 - extractions were performed on these maps. The database of individual extractions were searched for sources with positions within 120” of those determined in the summed maps. The source extractions produced a correlation coefficient, which measures how pointlike the source appears to be. All measurements with correlation coefficients less than 0.85 (less than 10% of the data) were rejected. A few bad points (less than 0.1% of the data) were also edited out, based on positional errors or the source being at the edge of the detector array. Standard calibration constants were applied to convert the IRAS flux units to Janskys. The overall calibration uncertainty is estimated to be 10% at 12 um, 15% at 25 wm and 60 um, and 20% at 100 ym. Only one type of observing macro (DPS) was used, so the overall calibration of all fluxes is the same. Fortunately, the relative flux densities used in variability studies are much more accurately determined for most sources, as discussed below. Table 2 is a summary of the IRAS pointed observations of these objects. The source name is listed in Column 1, and the observation wavelength (in microns) in Column 2. Column 3 contains the time interval (in days) between the first and last pointed IRAS observations, and Column 4 gives the number of pointed observations used. Column 5 gives the fraction of observations rejected due to low correlation coefficient. The mean flux density (in Janskys), and scatter in its observed value (in percent) are given in Columns 6 and 7, respectively. TABLE 2 Light curves, dermined by combining the individual observations of each object, are presented in Figure 1. The fluxes and lo uncertainty bars are plotted as a function of time (UT day 1983). The data are normalized to a mean of unity, and the mean flux density is listed for each of the four wavelengths. — 124 - FIGURE 1 As an independent test of the stability of the fluxes extracted from the IRAS observations, 70 bright field sources were subjected to an identical analysis. Nearly half of the field sources were detected only at 12 wm and 25 wm, and are presumed to be stars. Most of the remaining field sources (detected mostly at 60 wm and 100 ym) are presumed to be galaxies. The results of this analysis is given are Table 3. Column 1 contains the IRAS positional name. All of the other columns are identical to those in Table 1. TABLE 3 III. ANALYSIS The light curves were tested for variability by fitting a model assuming a constant flux density throughout the IRAS observations. For the target sources, the mean flux densities (in Janskys) determined from this model are given in Column 6 of Table 2, and the goodness-of-fit parameter y? is listed in Column 8. The probability (p) that a value of y? this large could arise from chance is given in Column 9. The 70 field sources (which are presumed to have non-variable fluxes) were tested in the same fashion. These data are presented in Table 3. The field sources were used to test the reliability of the fluxes. Light curves determined at 100 ym proved to be very unreliable, with 7 of 19 objects having derived values of x? significant at the p < 0.001 level. This probably arises from the fact that the 100 um detectors are not very reliable, or because the sources are confused due to interstellar cirrus or extended emission from the field galaxy. As the angle at which the satellite scanned the source changes as a function of time, an extended source ~— 125 - will appear to have different flux densities in the highly elongated IRAS beam, producing spurious variations. Thus, the quantitative analysis was restricted to 12, 25, and 60 wm, with 100 ym light curves used only to confirm variations seen at other frequencies. As many of the field sources were strongly detected at more than one wave- length, there were a total of 106 independent light curves measured. Four had x? significant at the p < 0.001 level. However, none of these light curves showed any long term trends. Three of these were of galaxies at 60 wm. All of these objects showed evidence that extended emission from the galaxy or infrared cirrus may have confused the observation. The other occurs at 12 wm, and the large x2 is due to a single point being very low, probably due to a problem with the detector. Several criteria were used to judge variability. First, a source was required to have a value of y? such that the probibility of random occurence is p < 0.001. Second, the data must show a clear long-term trend, not just uncorrelated scatter at that wavelength. The same trend must also be present (but not necessarily at a statistically significant level) in at least one other IRAS wavelength. Finally, extended galaxies or fields with confusion due to infrared cirrus are rejected. In spite of the possible sources of spurious variability discussed above, eighteen of the AGNs in the sample, III Zw 2, Fair 9, 3C 84, Mkn 3, Mkn 6, Mkn 79, PG 0838+77, NGC 4151, Ton 1542, Mkn 231, Mkn 279, Mkn 668, Mkn 841, 3C 351, 3C 371, 3C 390.3, Cyg A, and NGC 7469 showed no trace of significant which were observed on many different time scales. The observed scatter in their fluxes, which is fully consistent with their statistical uncertainties, is generally less than ~30% peak to peak. In fact, several of the brightest sources, Mkn 3, Mkn 79 and Mkn 231, and NGC 7469, have constant fluxes to better than the — 126 - 10% level peak to peak. This excellent level of repeatability represents the highest precision obtainable from IRAS observations of bright AGNs free from any of the complications discussed above. On the other hand, three AGNs (OJ 287, 3C 345, and BL Lac), but no field sources, fulfilled all the variability criteria. In addition, one other source (1E0241+62) had significant values of x2, but were rejected because they failed other tests. These four sources are discussed below. a) 1E0241+62 This X-ray selected Seyfert 1 galaxy has a value of y? = 3.18 at 12 pm, corresponding to a probablity of p = 0.0008. However, this is due to uncorrelated scatter in data with no long term trends, and no strong variability is seen in the other IRAS bands, so this object does not exhibit confirmed far-infrared variability. b) OJ 287 This BL Lac object is strongly variable at 12 um, 60 wm, and 100 um. It will be discussed in the next section. c) 8C 845 This highly-polarized quasar is strongly variable at 12 and 100 pm. It is discussed in detail in the next section. d) BL Lac BL Lac is strongly variable at 25 ym. The light curve shows a definite long-term trend, which is also seen at 60 um, although the variations are not as significant at 25 wm. It will be discussed in the next section. The upper limits to infrared variability must be qualified by noting that the relevant range of time scales has not been sampled very completely. As IRAS rarely observed these objects more than once per day, objects with fast, low level —127- variability would escape detection, being confused with observations which were rejected as having “uncorrelated scatter”. These observations are not sensitive to variability, especially at low levels, with time scales shorter than a week. Over half of our observations are not sensitive to variations over one month, which would be somewhat shorter than the variability time scales typically found in monitoring of the optical and near-infrared continuum (Rieke and Lebofsky 1982; de Bruyn, Readhead, and Sargent 1987). IV. DISCUSSION Three of the six blazars observed (OJ 287, 3C 345, and BL Lac) were found to vary at far-infrared wavelengths. The temporal sampling of the IRAS observations is not well-suited to determining the shortest variability time scales in these objects. It is estimated that these blazars have far-infrared variability with a typical doubling time scale of about three months. The implied light-crossing size of three light months yields minimum brightness temperatures in the far-infrared of T}2 10° K. Since these temperatures are at least three orders of magnitude above the hottest possible thermally-emitting dust grains, the far-infrared variability is further con- firmation that the continuum in these objects is predominantly nonthermal. The size of the emitting region can be estimated if (as appears to be the case) the infrared emission from these blazars is synchrotron radiation which becomes self-absorbed at the break frequency of ~ 10!2 Hz (see EM, E86, and EMR for details). These calculations yield source sizes of order one light-day. There is also strong evidence that the emission from these objects is beamed (see Cohen and Unwin 1986), with typical T ~ 5. Thus, variability over time scales of a few hours, considerably more rapid than those observed, would not be ruled out by light-travel time arguments. These objects can vary this rapidly at optical and X-ray wavelengths (see Moore et al. 1982, Halpern 1982). Although these data ~ 128 - are not ideally suited for studying fast variability, strong (factor of 2) variations would have been detected. There is no evidence for strong variations in any of these objects over time scales shorter than a few days. In this sense, blazars do not appear to vary as rapidly in the infrared as they possibly could. EM and EMR argued that the steep far-infrared spectra of several of the AGNs which were found to be non-variable are predominantly thermal. The constancy of the far-infrared emission in Mkn 3, Mkn 6, Mkn 79, Mkn 231, NGC 7469, Cyg A, 3C 84 and 3C 390.3 could be simply explained if it is produced in a region which is too large to show significant variability on time scales as short as one year. This would certainly be the case for thermal emission from dust grains. In several of these objects, EMR found that some of their far-infrared flux shows a significant spatial extent (on scales of 100 pc or more) and thus does not arise in the active nucleus. It is still possible that these active nuclei emit some nonthermal far-infrared energy which is intrinsically variable by more than ~30%, but was undetectable in the large IRAS beams. Unlike some of the blazars, the normal AGNs in this study showed no evidence of far-infrared variability on time scales of up to a few months. Our most interesting upper limits to variability are for those high-luminosity objects whose relatively flat infrared spectra give little indication of the presence of significant thermal contamination, such as Fairall 9, NGC 4151, Mkn 841, Ton 1542 and PG 0838+77. The lack of variability in these normal quasars and luminous Seyfert 1 galaxies does not rule out a possible thermal origin for much of the far-infrared continuum, but unfortunately it also does not rule out a nonthermal origin, either. It is significant that the upper limits to far-infrared variability set in this paper are smaller than the variability many of the same AGNs show at ultraviolet and X-ray, and in some cases, optical wavelengths. This suggests that continuum variability is an increasing function of observing frequency. However, our sensitivity — 129 - in the far-infrared is still too poor compared to that attained at shorter wavelengths to study this wavelength-dependence in detail. The most sensitive monitoring programs at red and near-infrared wavelengths (Rieke and Lebofsky 1982 , Cutri et al. 1985, Neugebauer et al. 1986) usually detect variability of only ~30%. Unfortunately, the IRAS observations are not sufficiently precise to exclude the possibility that the far-infrared continuum is variable, at the same levels as these previous studies detected at shorter wavelengths. V. CONCLUSIONS This study used pointed IRAS observations to study the variability of 22 AGNs. Three blazars were found to vary by about a factor of two over time scales of a few months. This variability is not as rapid as may be possible from these objects. No normal quasars or Seyfert galaxies were found to vary, although these data are not sufficiently sensitive to detect slow, small scale flickering at the same levels often seen at optical and near-infrared wavelengths. Unfortunately, the poor temporal sampling of the IRAS pointed data makes it difficult to use these results to strongly constrain the emission mechanism in normal quasars and Seyfert 1 galaxies. The radio emission from these blazars appears to be beamed with typical T = 5 (Cohen and Unwin 1986). If this is also the case with the infrared emission, then the true variability time scale is of the order of one year. The IRAS data are never sensitive to variations over time scales longer than 6-8 months (and for most objects, they are only sensitive to variations with much shorter time scales). Thus, it is quite possible that normal quasars and Seyfert 1 galaxies are variable in the far-infrared, but on time-scales longer than the lifetime of IRAS. EM, E86, and EMR presented evidence that the infrared continuua of normal quasars and Seyfert 1 galaxies are dominated by synchrotron radiation from a — 130 - self-absorbed source of order a light day across. The electron lifetimes are also expected to be very small. The failure to detect strong variability in normal quasars and Seyfert galaxies, and the fact that the variability time scales observed in blazars also appear to be much longer than the light-crossing times, suggests the possibility there may be some physical process which regulates this nonthermal luminosity so that it does not vary detectably on time scales less than a year. The authors would like to thank H. Hanson of the Infrared Processing and Analysis Center for tirelessly performing the IRAS source extractions, and W. Rice of IPAC for helping us understand the IRAS data. This research was supported by NASA grant NAS 7-918. - 131 - TABLE 1 VARIABILITY STUDY SOURCE LIST Source Position Type III Zw 2 0007+1041 Q Fair 9 0121-5903 Q 1E 0241+6215 1 3C 84 0316+4119 B Mkn 3 0609+7102 2 Mkn 6 0645+7429 1 Mkn 79 0738+4955 1 PG 0838+7704 Q OJ 287 0851+2017 B NGC 4151 120743941 1 Ton 1542 1229+2026 1 Mkn 231 1254+5708 Q Mkn 279 1351+6933 1 Mkn 668 140442841 Q Mkn 841 150141037 1 3C 345 164143954 B 3C 351 1704+6048 B 3C 371 1807+6949 B 3C 3903 1845+7943 1 Cyg A 1957+4035 2 BL Lac 2200+4202 B NGC 7469 —2300+0836 1 - 132 - TABLE 2 SAMPLE VARIABILITY STUDY DATA Source A (um) N Days Rej. Flux (Jy) Scatter (%) x? p III Zw 2 12 12 10 0.00 0.11 71 0.78 0.6589 III Zw 2 25 10 9 0.00 0.17 51 0.56 0.8290 III Zw 2 60 14 10 0.07 0.21 64 2.24 0.0064 Fair 9 12 4 14 0.00 0.38 15 1.66 0.1740 Fair 9 25 4 14 0.00 0.55 8 0.46 0.7104 Fair 9 60 4 14 0.00 0.59 13 0.75 0.5242 Fair 9 100 4 14 0.00 0.83 9 0.12 0.9454 0241+62 12 7 37 0.30 0.48 27 3.42 0.0022 0241+62 25 8 37 0.20 0.74 20 1.96 0.0564 0241+62 60 3 4 0.70 0.65 19 1.93 0.1447 3C 84 12 13 178 0.00 1.00 15 0.79 0.6653 3C 84 25 13 178 0.00 3.24 21 1.25 0.2444 3C 84 60 13 178 0.00 7.52 23 1.40 0.1575 3C 84 100 12 178 0.08 9.96 19 2.26 0.0095 Mkn 3 12 3 12 0.00 0.65 1 0.03 0.9733 Mkn 3 25 4 12 0.00 2.69 3 0.21 0.8906 Mkn 3 60 4 12 0.00 3.82 9 0.53 0.6639 Mkn 6 12 40 33 0.00 0.20 70 1.44 0.0376 Mkn 6 25 62 33 0.00 0.67 25 0.66 0.9801 Mkn 6 60 65 33 0.00 1.29 19 0.45 0.9999 Mkn 6 100 45 33 0.00 1.40 66 1.17 0.2084 Mkn 79 12 4 163 0.00 0.29 16 0.92 0.4298 Mkn 79 25 4 163 0.00 0.70 13 0.78 0.5064 Mkn 79 60 4 163 0.00 1.52 6 0.19 0.9017 Mkn 79 100 4 163 0.00 2.80 6 0.36 0.7844 0838+77 25 5 3 0.00 0.11 34 0.61 0.6558 0838+77 60 14 22 0.00 0.18 74 1.14 0.3152 OJ 287 12 11 215 0.00 0.22 122 6.45 0.0000 OJ 287 25 11 215 0.00 0.43 33 0.76 0.6716 OJ 287 60 11 215 0.00 0.93 65 10.27 0.0000 OJ 287 100 11 215 0.00 1.58 54 7.88 0.0000 NGC 4151 12 5 26 0.00 1.84 8 0.42 0.7941 NGC 4151 25 6 26 0.00 4.72 11 0.38 0.8629 NGC 4151 60 6 26 0.00 6.94 23 1.54 0.1750 NGC 4151 106 é 26 0.00 10.17 16 6.82 0.5322 Ton 1542 12 q 8 0.13 0.10 45 0.79 0.5787 Ton 1542 25 9 8 0.00 0.18 66 1.10 0.3622 Ton 1542 60 5 3 0.29 0.18 31 0.71 0.5865 — 133 - TABLE 2 SAMPLE VARIABILITY STUDY Data (cont.) Source A (um) N Days Rej. Flux (Jy) Scatter (%) x Pp Mkn 231 12 4 20 0.00 1.75 6 0.36 0.7833 Mkn 231 25 4 20 0.00 8.45 7 0.54 0.6516 Mkn 231 60 4 20 0.00 33.34 7 0.24 0.8705 Mkn 231 100 4 20 0.00 38.94 8 0.74 0.5266 Mkn 279 12 6 138 0.00 0.14 30 1.21 0.3011 Mkn 279 25 6 138 0.00 0.28 32 3.49 0.0037 Mkn 279 60 6 138 0.00 1.27 6 0.14 0.9833 Mkn 279 100 6 138 0.00 2.74 15 0.95 0.4491 Mkn 668 12 4 29 0.00 0.60 14 147 0.2192 Mkn 668 25 4 29 0.00 0.40 33 3.97 0.0077 Mkn 668 60 4 29 0.00 0.79 12 0.82 0.4841 Mkn 668 100 4 29 0.00 1.20 27 2.51 0.0566 Mkn 841 12 9 31 0.00 0.20 32 0.94 0.4844 Mkn 841 25 10 31 0.00 0.50 23 0.86 0.5644 Mkn 841 60 10 31 0.00 0.56 37 1.23 0.2725 3C 345 12 31 100 0.03 0.12 98 2.44 0.0000 3C 345 25 32 =: 100 0.00 0.28 35 1.23 0.1741 3C 345 60 32 =: 100 0.00 0.68 40 1.43 0.0562 3C 345 100 24 90 0.04 1.30 68 3.17 0.0000 3C 351 12 3 1 0.25 0.06 18 0.21 0.8078 3C 351 25 18 86 0.00 0.13 67 1.11 0.3315 3C 351 60 5 81 0.71 0.20 26 0.49 0.7409 3C 371 12 25 68 0.14 0.10 55 0.64 0.9061 3C 371 25 61 72 0.02 0.16 114 1.28 0.0692 3C 371 60 65 72 0.02 0.25 128 1.12 0.2355 3C 371 100 2 6 0.82 0.81 33 0.67 0.4118 3C 390.3 12 59 8177 0.00 0.13 115 1.15 0.2005 3C 390.3 25 83-178 0.00 0.32 41 1.14 0.1878 3C 390.3 60 82 194 0.01 0.22 156 1.28 0.0475 Cyg A 12 18 197 0.00 0.14 31 0.54 0.9327 Cyg A 25 18 = 197 0.00 0.81 12 0.60 0.8980 Cyg A 60 14 192 0.22 2.74 19 1.01 0.4389 BL Lac 12 17 47 0.00 0.12 78 0.94 0.5204 BL Lac 25 18 47 0.00 0.23 56 2.80 0.0001 BL Lac 60 18 47 0.00 0.45 51 1.42 0.1179 NGC 7469 12 4 22 0.00 1.10 12 1.18 0.3143 NGC 7469 25 4 22 0.00 4.93 10 0.86 0.4625 NGC 7469 60 4 22 0.00 26.53 6 0.31 0.8192 NGC 7469 100 3 13 0.25 43.75 3 0.06 0.9408 ~- 134 - TABLE 3 FIELD SOURCE VARIABILITY STUDY DATA Source A (um) N_ Days Rej. Flux (Jy) Scatter (%) x Pp 0008+1022 12 15 10 0.00 0.32 42 2.04 0.0122 0007+1051 12 15 10 0.00 0.19 46 0.72 0.7565 0009+1035 60 14 9 0.07 0.23 78 1.90 0.0249 0009+1002 60 7 8 0.30 0.29 49 1.32 0.2454 0007+1041 60 14 10 0.07 0.21 64 2.24 0.0064 024146155 12 8 35 0.11 0.08 59 0.80 0.5868 0241+6155 25 8 37 0.20 0.14 34 1.07 0.3798 0239+6214 12 7 35 0.00 0.09 92 2.65 0.0143 0239+6214 25 4 5 0.43 0.10 59 0.75 0.5204 0240+6259 12 6 5 0.00 0.80 22 2.43 0.0330 0240+6259 25 6 5 0.00 0.22 21 0.82 0.5346 0239+6222 12 10 37 0.00 0.41 21 1.21 0.2841 0239+6222 25 10 37 0.00 0.13 137 2.36 0.0116 024446158 12 10 37 0.00 0.23 35 1.02 0.4187 024446158 25 5 5 0.00 0.08 39 0.78 0.5378 0317+4054 12 10 178 0.00 0.33 38 1.09 0.3626 0316+4047 12 10 178 0.00 0.18 42 0.74 0.6763 0315+4100 12 6 2 0.33 0.12 83 1.10 0.3557 0317+4038 12 6 178 0.25 0.10 128 2.41 0.0339 031444154 12 5 174 0.17 0.09 84 1.37 0.2418 0318+4041 60 10 178 0.00 0.47 82 1.49 0.1456 0316+4127 60 13 178 0.00 0.33 187 8.83 0.0000 061147137 12 4 12 0.00 0.27 29 2.23 0.0823 06484-7445 12 63 33 0.00 0.47 36 0.79 0.8790 0648+7445 25 45 33 0.00 0.21 115 1.60 0.0070 0643+-7419 12 54 33 0.00 3.01 27 0.83 0.8000 0643+7419 25 58 33 0.00 0.94 43 1.40 0.0252 0646+7411 12 64 33 0.00 0.34 75 1.39 0.0226 0646+7411 25 16 31 0.24 0.11 78 0.67 0.8192 0645+7429 60 65 33 0.00 1.29 19 0.45 0.9999 0645+7429 12 4 163 0.00 0.29 16 0.92 0.4298 0645+7429 25 4 163 0.00 0.70 13 0.78 0.5064 0645+7429 60 4 163 0.00 1.52 6 0.19 0.9017 0737+5039 25 5 3 0.00 0.11 34 0.61 0.6558 0737+5039 60 14 22 0.00 0.18 74 1.14 0.3152 — 135 - TABLE 3 FIELD SOURCE VARIABILITY STuDy Data (cont.) Source (um) N Days Rej. Flux (Jy) Scatter (%) x p 085142009 +12 «411 «215 0.00 0.45 78 1.68 0.0792 085141946 12 9 207 0.10 0.51 27 1.41 0.1882 085242027 60 11 215 0.00 0.25 71 1.28 0.2365 085142024 60 10 215 0.00 0.16 66 0.93 0.5016 085142040 60 4 45 0.20 0.18 126 4.15 0.0060 1207+3942 12 4 18 ~=0.00 0.89 12 0.61 0.6056 1207+3942 25 6 26 0.00 0.23 31 0.59 0.7109 120643911 60 6 26 0.00 0.40 40 1.19 0.3099 122942111 12 10 12. ~=—0.00 0.27 36 1.86 0.0532 1228+2029 12 10 12 ~—-0..00 0.19 70 1.95 0.0403 123042101 60 10 12. ~—0.00 0.26 22 0.64 0.7650 1229+2009 60 7 8 0.22 0.11 27 0.25 0.9580 1229+2020 60 9 12 0.00 0.14 67 1.61 0.1167 125145705 12 4 20 ~=0.00 1.25 6 0.37 0.7761 12514-5705 25 4 20 =: 0.00 0.32 35 3.00 0.0293 134946923 12 5 1388 0.17 0.43 19 2.86 0.0220 1349+6923 25 5 138 0.00 0.10 55 0.82 0.5101 140442840 12 4 29 ~=0.00 0.60 14 1.47 0.2192 1404+2840 25 4 29 ~=0.00 0.40 33 3.97 0.0077 1404+-2900 60 4 29 ~=—0..00 0.14 77 2.87 0.0351 150141028 12 10 31 ~=—-0.00 0.38 19 0.99 0.4448 1501+1028 25 7 150.18 0.13 101 1.97 0.0657 150141055 12 10 31 0.00 0.81 16 0.86 0.5621 15014-1055 25 10 31 —- 0.00 0.21 72 0.80 0.6137 164144021 12 20 43 0.09 0.11 77 1.62 0.0450 1640+3944 12 31 100 0.00 0.12 223 2.97 0.0000 1640+3944 25 32. 100 0.00 0.16 54 1.07 0.3677 1640+3944 60 32 ~=100 0.00 1.56 29 1.59 0.0195 1641+3954 12 5 92 0.17 0.09 99 1.90 0.1071 1641+3954 60 10 90 0.09 0.35 29 1.11 0.3518 164043932 60 17 43 0.19 0.19 130 1.70 0.0394 1640+4022 60 16 64 0.06 0.16 89 0.81 0.6666 1702+6023 25 5 7 0.17 0.08 38 0.40 0.8117 1702+6023 60 8 14 ~— 0.00 1.11 154 6.13 0.0000 1702+6037 60 12 16 ~—:0.08 0.16 95 1.69 0.0687 1703+6047 60 li 82 0.39 0.21 80 1.14 0.3259 — 136 - TABLE 3 FIELD SOURCE VARIABILITY STuDY Data (cont.) Source A (um) N Days Rej. Flux (Jy) Scatter (%) x Dp 1804+6950 12 8 66 0.20 0.09 55 0.34 0.9369 1804+6950 25 12 45 0.00 0.09 73 0.46 0.9287 1804+6950 60 69 72 0.00 0.74 82 1.37 0.0229 1807+6936 12 5 56 0.17 0.10 45 0.24 0.9129 1807+6936 25 5 27 0.00 0.10 44 0.32 0.8618 1807+6936 60 56 68 0.03 0.24 134 1.35 0.0448 1802+6932 60 37 28 0.08 0.34 178 1.82 0.0019 1808+7009 25 6 382 0.14 0.09 53 0.26 0.9357 1808+7009 60 58 54 0.03 0.31 102 1.78 0.0003 18424-7926 25 29 79 0.06 0.15 102 1.34 0.1077 184247926 60 41 79 0.00 0.74 46 1.08 0.3411 1854+8017 60 25 49 0.04 0.20 89 1.17 0.2586 1850+7922 12 27 194 0.00 0.28 72 1.22 0.2000 1850+7922 25 8 47 0.27 0.10 60 1.04 0.4002 1846+8019 60 4 16 0.00 0.11 74 1.17 0.3182 1957+4116 12 8 21 0.11 4.10 9 0.43 0.8843 1957+4116 25 8 21 0.00 2.43 9 0.55 0.7939 1957+4116 60 8 21 0.00 1.30 46 3.03 0.0035 1957+4104 12 16 197 0.11 0.60 24 1.70 0.0432 1957+4104 25 18 197 0.00 0.18 61 2.16 0.0037 1958+4032 12 18 197 0.00 1.78 16 0.70 0.8070 1958+4032 25 18 197 0.00 1.25 29 1.91 0.0129 1958+4025 12 15 197 0.06 0.09 101 1.30 0.1953 1958+4025 25 7 197 0.13 0.07 174 1.31 0.2474 1957+4025 12 18 197 0.00 0.24 34 1.14 0.3046 195744025 25 18 197 0.00 0.13 34 0.58 0.9096 2204+4131 25 6 4 0.00 0.17 30 1.12 0.3454 220444131 60 7 5 0.00 1.85 12 0.51 0.7983 220244122 12 4 16 0.33 0.09 46 0.51 0.6722 220244122 25 9 17 0.25 0.12 91 0.95 0.4734 2202+4122 60 11 17 0.00 1.20 22 1.18 0.2960 2201+4214 12 4 40 0.00 0.08 138 2.55 0.0538 220144214 25 14 44 0.07 0.11 114 1.96 0.0196 2201+4214 60 18 47 0.00 0.98 11 0.30 0.9976 2200+4208 12 18 47 0.00 2.31 29 0.89 0.5914 2200+4208 25 18 47 0.00 0.76 26 1.16 0.2889 220044208 60 18 47 0.00 0.22 47 0.90 0.5729 2301+0901 60 4 22 0.00 0.84 8 0.35 0.7921 2300+0822 12 4 22 0.00 0.29 11 0.26 0.8568 - 137 - REFERENCES Angel, J. R. and Stockman, H. S. 1980, Ann. Rev. Astr. Ap., 18, 321. Bregman, J. et al. 1986, Ap. J., 301, 708. Clegg, P. G. et al. 1983, Nature, 305, 194. Cohen, M. H. and Unwin, S. C. 1986, IAU Symposium, # 110, 95. Cutri, R. M. e¢ al. 1985, Ap. J., 296, 423. de Bruyn, A. G., Readhead, A. S. C., and Sargent, W. L. W. 1987, in preparation. Edelson, R. A. 1986, Ap. J. (Letters), 309, L69 (E86). Edelson, R. A. and Malkan, M. A. 1986, Ap. J., 308, 59 (EM). Edelson, R. A., Malkan, M. A., Rieke, G. H. 1986, Ap. J., in press (EMR). Ennis, D. J. et al. 1982, Ap. J., 262, 451. Halpern, J. P. 1982 Ph.D. Thesis, Harvard University. Impey, C. D. and Neugebauer, G. 1987, Ap. J., in preparation. Martin, P. G., Thompson, I. B., Maza, J., Angel, J. R. 1983, Ap. J., 266, 470. Moore, R. L. e¢ al. 1982, Ap. J., 260, 415. Neugebauer, G. et al. 1986, Ap. J., in press. O’Dea, C. P. et al. 1985, Proceedings of Garching Conference on Ultraviolet and X-Ray Emission from Active Galactic Nuclei, p. 63. Rieke, G. H. 1978, Ap. J., 226, 550. Rieke, G. H. and Lebofsky, M. J. 1982, Ap. J., 250, 87. Robson, I. et al. 1986, preprint. Stockman, H. S., Moore, R., Angel, J. R. 1984, Ap. J., 279, 485. Young, E. T. et al. 1985, User’s Guide to the IRAS Pointed Data Products. — 188 - FIGURE CAPTIONS FIG 1.—Far-infrared light curves for 22 AGNs. Normalized flux density is plotted as a function of time, with lo error bars. The caption to each panel gives the source name, observing wavelength, and mean flux density, in Janskys. — 139 - (ee6t) zequmy fog osz ove oez O42 cand 092 — pF T v vT ee | ¥ + ne | ee Goce Cee | Po on nme wae 4 on a oo rn ro | 4p y2'0 *xnIq we oe :Y 00+1#Z0 :eomn0g BNL Saeloy (ce6t) zequny 4eq gee vez 2ez i] nN it 4. A 4£f 990 <xnIg UM! 09 :v 00+T#Z0 :eo1m0g (ce6t) zequny 4g 022 092 ose ove oez | v Tv ce ee eee ee nie ee ee T v v v oo | T v T v T 0 | a eer eee ee CS SS Sy We Ce ee ell Se ev'0 XRLZ we et :y 00+TPZ0 :20an0g INL sapyeloy Figure 1 (a) INL eayeloy -— 140 - (e861) sequny feq ost Sot OFT set q v u qv 0 2 1 n s rn nN I nm ra A nm j A 2 4 rn t 4f €8'0 :xnLg wr COT NY 6 AWey :eo.mog (E961) zequmy eq ost crt OFT cet ee J rd Ll nN 4. rn A L nN A rn nN i L nN a A it Af Sco :xnIy wr og :y 6 ayeg :e0mo0g ENLY SApyelay ENLY Sapejoy (e961) zequny Leq ost Set ov Set T v v v v T v r ¥v v T v v v v 7T it nN rn a 4 I re 4. 4. nN it A ab. A A. rT 4p 690 -xnUy UN Og NX 6 Ayeq :ed.mog (eget) zequmy fog ost Sot OvT ser 41 rn nN mn 4 it nm A. 4. n AL rn 4. n 4, 4 4¢ e6'0 :xnLy unl eT 7X 6 ayeg :ea.mog Figure 1 (b) mnt oanyeru ENLY Satye]ay -~ 141 - (eget) zequiny Leg SLT T re E 4. 4p 40'0 :XNLg wr gz :v 2 4Z YI :e0I1n0g INL SATPlOY oBt (eget) zequny seq SLT Se 120 :xnig wr og :\ zg 41Z DI :eounog oet (cg6t) zequny svg Gut T i=} = INL eaTyoloy re 1 4. 4 a . 4e 00 cxnLy wel et :y 2 “Zz I :aommog INI VBpyVpoy Figure 1 (c) - 142 - (ce6t) zequny fog osz 002 ost oot 0S T ¥ ¥ T v T v ¥v v ¥ T ¥ ¥ v ¥ t ¥ ¥ ¥ ¥ T 0 , J a ----¢-o! Fd L nN 4 L 2 nm A re iT a nM nm 2 rT a rT 4f 96°6 :XNLJ Ur COT :Y% #8 DE :e0mN0g (ce6t) sequny 4eq osz 002 Ost Oot os T ¥ ¥v T v ¥ v ¥ T ¥ v r ¥ T ¥ v — T 0 a py Tt ] @ 4e yee sxnLy unl gz :y ¢g Og sa0M0g ENLJ @A7zel2y ENLI SaNeoy (e961) zequiny seq ost ee | — Ty — = = =n — L 2 A nN rn" i - 4 4s mn it A 4 a A A a a nm nN 1 4f 29h Xn ww 0g cv ¥@ OG :eoun0g (S861) sequiny oq os2 002 ost oot OS ye bt ae ae ae t _ 4p 00°C “<X0LE UH! et :y— FQ OE :#0gnog A ENL eapveley ENLI Sayzelay Figure 1 (d) — 143 - (eg6t) zequmy seg GL O22 sez PH ee + 4 A A. i A L nN 4p 20% 2xNLJ wy COT :% = -g «WA seounog (eget) zequny seg Ge Oz sez T v v T v - r T 7 + v — poe eee os —_ 1 A. a. 2. A aT A . n A L A A 4p 692 =:XNLY Wr cz :yY g UNA :2omog ENLg 2ayyeloy BNL Saperey (e961) zequny Leg GLz 042 Gee r 4 je nef ~—s o=ep em «=m om om qm eee cee come = —- =— i bn A 1 L 4 rn 4 L a s mn 4f 28°6 :xNIg wi og 7K g WAR :e0mo0g (ce6t) zequny 4eq GLS 042% See T t v v v v T 7 7 7 J pe ee a 4 LL. mn 4 A Ll rn a 4 a it rn 4p ogo :xnty wi gt 3K og UA :90.mog ENLJ SApeloy Figure 1 (e) ENL{ OaTyeloy — 144 - (cesT) zequiny fog v v rr 4 8 Ah eats Ht i - ne ——v ee —yv ee ——7—) 0 Ly i ttt ia I. -+ 4 | fi t Tt } iP 4 OWT 2NIZ WH OOTY -g WAN :eo.mog (cast) szequny fog One . 098 088 ove 4 ENLI @ayyeloy po n n i nN 4 4 4 i n 4 rn oe | 4p 249°0 :xnLg wi gz :v Q UA :eomog ENLJ @aTPeloy (ce6t) zequiny fog 0242 ose ose | ove rr 0 et aie Plt atte @ 4f 62't zug wv op YO 9g WIR :eomog (ee6t) sequny Leg 042 092 ose ore pe 0 | Hoot deh r— +4 apthet # +r +7 Ht He re re =~ ENLY Sato 4f 0Z'0 :xN we et :yY g UHR :#=omMog INL early Figure 1 (f) — 145 - (ee6t) zequiny Avg osz 002 ost 00t remy i 5 nN 4. 4 i n rn 2. rn i n 4 4. A i 4f 08'S x0 Wr COT XY 64 WHA seounog (eget) zequny Log os2 002 Ost oot T v v ¥ v T v ¥ ¥ T v v vr v T i A. 4 4. 4 i n A. rn nN L rn A 4 rn I 4p 04'0 :xnIg ww og sy 624 WHT :eo.mog ENLI SAF}lI EN Savoy (ce6t) zequmy 4eq ose 002 Ost Se 20 cxnig unl 09 :X% 64 WI :eomog (eet) zequny fog ose 002 ost i n 4. rm 4 i n n n 4 4 A ne 4p 62'0 cXNIZ Wr ZT :Y OB4 WHT :20mm0g ENL{ @ATYeTOy Figure 1 (g) ENLI OATyelay — 146 - (ce61t) zequmy fog ort Oot 06 T v v v v : s . a 7 7 a j 2. 4 n rn I A a 4 4 at rn" Se ot'0 3xNLd UW Og “Y LL+8ER0 :20un0g om INL SAyeloy (eget) sequmny, Leg ott oot 06 v T v v v v T v v v v T } L a A. A n i nN rn nm nm i Af 06'0 :XNL{ Ur! OOT XY L4+8E80 :20.m0g (E861) szequmy fog oot 66 96 46 ry . L r 7 rt 4 4 n rn 1 A. 1 4, 4. j n a 4 A f. 4f 10'O 3xnig wr oz 3x 224880 :20un0g INF eayyelay Figure 1 (h) ENLY SaApyeyey — 147 - (ce6t) zequny 4eq 00g 002 oot ¥ T v v r v T v ¥ ¥ v T 0 L 4 My 5 Popup jt t 4 J Fa 4d a a. rn 4 | . 4 An 4 rT 4 99°F -xnIg Ur COT :vY = Lego :e0mog (E861) sequmy eq 00e 00g oot T v v v v T ¥ v ¥ ¥ v 0 Li |- -[4 pl - >> — |e 4 4 L L L & A i 4p evo <xny unl og :v LEzZfO :20In0g EN eayyeroy ENL{ OaTyeley (ce6t) zequiny 4eq 002 oot 4p e6'0 <xnLg ur 09 ‘Vv oc &BSLO -ooIn0g (e961) zequny sug 002 Oot I. Nn n 4 4. it 4p 220 «<xnLg Ur gt ty 482f0 :#dIN0g RNY SApPelay ENLI eapyelay — 148 - OLt (eget) sequmy Leg eho) § ost A. 4. rn it nm ‘ ‘ s rl A . . 4 Se LVOT ‘XOLg UW OOT “Xv ISTW ODN :eounog (ceet) zequmy fog OLT oot Ost v v ¥ v T i rn 4 4f 2a'y NLT wr Gz {NY IST® DDN :eomog RNY eTVe12y INL eapvloy (eg6t) zequny Log OLT 09 ost v ¥ v T v v T v T 7 0 L n | pate Hp tO OOOO zk sy r 4 L rn A s n 1 i 4 ¥8'9 :XNLY WH OQ Vv I9Th DON :20.mog (ce6t) zequiny 4eqg OLT o9T ost a ¥ v v Tv vw v v F T T a + 4 re 4 4 it nN PN 4. n 4 4p y't 2xniy urd 2] X 19Th ODN :20mmog INL SaNVpoy Figure 1 (j) ENL Saepey — 149 - (eget) zequny Leg oat BLT Oat HLT eat U ¥ ¥ n 1 N . 1 . t ; 1 4f at'O “XN ww oe :yv ZPGT WOL, :2e0mog INL Sapeloy (ee6t) zequmy Lug vhT Lt ZLt v ¥ ¥ ¥ T v ¥ v v T v v ¥ v T v L re ne 4 4. L rn n rn nN it 4e Oto :xNIy wr og Vv ZPgT Uo, :eomog (e961) zequmy Leq OLT vet aD § OAT 89T | nN it m" 1 rn 4 n I °o 4£e OF'0 2BMLA UW et :v e¥GT WO] :20.mo0g Figure 1 (k) znd Arista ENL{ @ATBTOY -150- (ee6t) zequiny 4oq S6t 06T set oat GLE Re a Ce a SS CO a | 4f VO'RE -XNLY We OOT X% Tee TAN :eomog (€96t) zequiny 4eq set ost SLI T on ee tT? a rn 0 fl kk 4¢ cpg :xn[y wn gz :y Teg UAW :e0.mog ENE Sayyeloy INL SAT elew (e961) zequiny svg S6t 06t sat oat SLT 0 Raye eee +e 1 1 Fe io rn 4 nm 1 L a nm L i a L a 4. 1 a n . 2 i 4f VEE :xNLY we og % TES WHR :eomog (ce6t) zaequiny 4uq S6t 06t sot ost SLT Le oe eee a pa aye —— 0 L 1 pate = 0 2 | or or ey Pee Cena eee Cees WY | a ey Cee Wa MS CNT CO WY TY it A. Ap out cxnig we et :y lee WAR :e0mm0g INL eyy{o19y Figure 1 (1) INL SaANselay - 151 - (ce6t) sequny fog osz 002 ost v T v v v ¥ T v ¥ ¥ v T it A 2 rn 4 L a 4. 4 rn 4. AP VL'S 2XOLY Wi OOT NX 642 WIR :eounog (ce6t) sequny Leg ose 002e ost T v ™ v v T rn A 1 4 4 4 4 4 rn rn nN nN j Ap e2'0 3xNIZ UH og :Y BLeZ WIR :20mmo0g NL sayeloy ENLY SasyO]Oy (€g6t) zequny Log ose 002 ost | ee v v v T v v i. rn 4 A 2 A rn n 4. A. 4. 4p 22° cxng wet og :y 6Le WH :eo.mog (cet) zequmy 4eg ose 002 Ost T v v ¥ r T v ™ v v t v L rn rn 4 A 4 . 4 An 4. re 4e oto :xntg Un Zt ‘XY 6Le WAR :@omog Figure 1 (m) Balad ARe1ed ENLL SAyyejoy — 152 - (cesT) zequny fog oe2 0zz og 002 rr rr rg 0 + ee yoottste & rt Fs rT rn n s rn L nN mn nN L i Af oft :XnLY wr! COT :X% 999 BAR :eomog (e961) zequny 4eq oe2 0zz ol2 002 perry 0) + pb + -------- t L + re rT A rn rn rn 1 4 " rn A L 4p Ov'0 :XNLY wr gz :y 999 TA :20.mog EN Sayyeley MN Sapyeloy (eget) zequny feg ote oz ote 002 A. nm nN nN a i. a 4 nm re L 4f 62'0 *XNIg WOM *X% B99 WHT :eomog (ee6t) szequmy 40g ote oe ate 002 v v T v v v v T v v ¥ v T i 4 A 4 4 it 4 mn 4. nN rT 4p 090 “DLE Ww Zt XY 999 WAR :2o1mog IN[{ 2Bpyeloy Figure 1 (n) SNL Oatyelowy — 153 - (eg6t) zequiny fog ove oz oz rr a er 0 AL A .. A t i rn a 4 2 i 4p 9S'0 :xnLZ wr! Og :vY 1PQ WH :eommog l=) (eget) z9qumy fog (€86t) z2qumy 4uq ore oz 022 ove oe ozz rr ry 0) er L 4 ¢ J iS . 4 4 J pt-pt EB Lt 4-4 ~ + | 3 + + a ie L Fd rn nm a 4 1 rn 4 A a 4 A rn 4 rf i] mn rn a iT nm 4. nN A rt A nN A 2 A 4p 0S'0 :xNLZ we oz :Y 1¥e WH :20.mog 4f 0g °XNL we Zt XY T¥a WHR :eo.mog INL SaTyelay Figure 1 (0) ENLI SAT S[O — 154 - (eset) zequny 4eq 00g osz ' 7 ; : + . ¢ . rt) t 4 a _ _ ett nyt op Fe 4p OG <XNLY wr COT “XK SE DE :e0.mm0g (eget) zequiny 4eq ooe osz 002 SS EEEEIELAIERnmn aimee aeeneeeaee eee aa aR em 0 4 ae, the fh — Hy — 4, - Tf i 2 4p e2'0 :xnLy UM! cz :Y SPE DE :20.mMog ENL SATPelTOy ENLY SAT PeIOW (eget) zequmy sug 00 osz 002 Ly | _ ih ----j7 Hh ah — Hye + > 4 ENLI SATPOOY i ry n ne 4 L n rn a s i 4f e9'0 ‘xnIg WOR %% GPE DE ‘e0.mog (se6t) zequny fog 00 ose 002 | tL | | | | | =t Zan i= | J | te | 4p 20°0 :zNLg UN ZT :Y SFE OG :20mog Figure 1 (p) ENLI SaTyoloy — 155 - (ce6t) zequiny 4eg 002 ost | + te | | | | | | | | | | | + IN[Y eTPOlOy Fe IL 4 rn a L ~ Z 4p 020 :xnLd U0 %X% TSE DE :eommog 4 o (cast) 2equny seg (cast) zequmy 4og 5 00g ost 902 s°S0z soz fy 1 - 7 7 . 0 Se as EERE EEE SER ON een IH} ff---------1-- 6B beefs See L4.é a a r 4 4 Z 2 A. 1 rn mn n rs A I. nN A rm 4, 4 nN & 4 re 1 Af eto -xNLg UR oz 7K 196 DE :2omnog 4p 90'0 :XNLd UM et NY OIG6 OG ‘#ommog — 156 - (eset) zequny fog 9s2 ¥S2 es2 os2 A rn L A i A | 4p 180 -XNLd UW COT X TLE DE *e0mn0g (ee6t) zequny 4eq 00 ose ose ove 022 T “v" T v T ¥ T + T Henn tlhe iy ee J rn 4 2. it nm I nN 4 4p evo :xnlg Url gg NX TLE OG :20.mog we SNL Sati e[ey INL 2apyelay (ee6t) zequiny 4eq 082 092 ove oz | v . LU . qv . v rat Sera Malle A. a. 4 4 A 4 it Sp ozo nig ul og % TLE OE :eomnog (eget) z2equny eq 082 092 ove 022 it 4 1 nN i 4 it So 4p o'0 2xnIg UM at Y TZe De :20sn0g INL eapyepoy Figure 1 (r) INL Sapooy — 157 - (e961) zequiny seq ose 002 ost | | | oo | | | | | | | | | | Jj Lee ee ee ee ee 0 | a ee ee ee ey Ge ee ae 4f 6Z'O *XNL_ wt OOT *¥ F'06E DE ‘eomn0g (ce6t) zequny Aeq ose 002 ost oot a rar 0 4 Ltt AE Ae 4 ~~ MLZ SaNe[oy i [ae Ce CY ONE LU CONEY WE WES WD CO Go Wy | 4p 26°0 =xnLy wn oz :v E066 DE :22.mMog ENLY Sayeed (eget) zequmy Leg uy i | |, 7 ii i rae TE a 4f 220 a . — DE -80INnOg _ ne HULL | L Ath 4 | ro it | yy 4p evo -xnlg Un Zt + 6066 DE :20.mog ENLI SANeoy Figure 1 (s) INL Sapyeley — 158 - (eset) zequmy 4eq o0e osz 002 ost oot en ee ee eee ee ee ee oe ee ee ee ee p— He -— - - - - HHH cl - Seal ENL Sapyoloy La ae NEY CS CSS WORRY WOT WORRY CORD WOT COT WONT WOU (SOUT UST WUE WUE 4p 18°0 “xnIg ww ge sy y Bp :e0mog (ee61) zequmy seg oo€ ose 002 | ee ee seen eee eee ee cee cee eee | | | | | = _ tI. z i 4b 4b rn 4 L 4 4 ro i 1 A A rm mn 1 4p 2% =xnyZ we og sy sy B49 :womog (ce6t) zequiny 4ug ooe ose 002 ost oot ray Spar 0 L ; 7 — ia —— ee ee ee \- tT 2 | ar a Ce ae ae Ce ee a Lis L 46 70 -xnIZ ww et iy Vy B49 :aamog SNL eapyeyey Figure 1 (t) ENL{ SATPEToy — 159 - (eget) szequiny Lug 002 oat 09t n A A ; ' 4p 620 :XNLJ ur oz sy OB] Iq :20mog - INL eapeloy (eg6t) zequny Avg ogt oot b. Se Sv'O :xNLY we 09 :vY OWT Tq :eounog (ceet) zequiny Lug Oet - INL eaperey 4p 200 :xniy wr et cy ow] Iq :aounog Figure 1 (u) INL Sapeley - 160 - (eset) zequny Aeq cet o9T gst v T ¥ r v v T L re i A 4. A n it 4f GL'6y 2ANLY UT! OOT *¥ 69PL DON :eoun0g (eget) zequny eq v v ¥ v T s OLT oot v I n rn A 4 i 4p ee'y :xniq wr oe Nv 69%L ODN :20un0g ENLY SATzOlTIy ENL SATeley (cesT) sequny fog OLT oot ¥ T ¥ v v v T mn 4. rt s nN nN A A 4. re 4f €9'9% *XNLA WH Og -¥ B9PL DON :eo.mog (ce6t) zequiny, 4eq OLT Ost ° 1 nm 4 rn 4. rT 4f oUt -X0Ly Ur! Zt XY 69%4 ODN :20.mog ENLY SATzPeOlOY Figure 1 (v) BNL SaTyeley -— 161 - Chapter Seven FUTURE WORK - 162 - The ultimate goal of my research is to determine physical conditions at the cen- ters of active galactic nuclei (AGNs), and understand the mechanisms responsible for producing their enormous luminosities. As the highly energetic nonthermal processes believed to be operating produce energy over a very wide range of frequencies, the broad-band continuum must be measured to study unreprocessed radiation from deep within the nucleus. This work has begun the massive task of determining the broad-band properties of AGNs from radio through X-ray wavelengths. Spectral energy distributions of the CfA Seyfert galaxies—a complete, optically selected sample—will be determined and the correlation between spectral properties studied. Eventually, data on other well-defined samples of bright AGNs (e.g., the PG quasars, hard X-ray selected Seyfert 1 galaxies, etc.) will be included as well. It may be possible to create an improved classification system for AGNs, based on simple physical models of their broad-band properties. The infrared and radio data for the CfA Seyfert galaxies have been fully analyzed, and the optical, ultraviolet, and X-ray data must now be incorporated. The optical spectra will be used to measure line and continuum fluxes and to test for selection effects. The CCD surface photometry will be used to separate the non- stellar continuum from starlight in the underlying galaxy. An initial examination of the CCD data suggests that the underlying galaxies may have a higher tendency to have bars, companions, and evidence for recent interactions than normal galaxies. Most of the Seyfert 1 galaxies in this sample were detected at X-ray wavelengths with the Einstein IPC and MPC and at ultraviolet wavelengths with IUE. It should be possible to produce complete X-ray-ultraviolet spectral energy distributions for most of these objects. As usual, this research raises new questions as old ones are answered. A broad emission feature, centered near 5 wm, was found in the spectra of luminous AGNs. Its nature not been established, having been modeled as free-free emission from — 163 - the broad-line region (Puetter and Hubbard 1985), emission from the outer regions of the accretion disk (Collin-Soufrin 1986), and thermal emission from hot dust (ie., Barvainas 1986). Low-resolution spectra from 2.5 to 10 wm can be used to determine its exact shape, and variability studies may allow a limit to be placed on the size of the region in which it is produced. Most quasars and luminous Seyfert 1 galaxies have spectra which rise from the optical into the infrared, and then cut off near ~80 um. This suggests that the infrared emission from most radio-quiet quasars and Seyfert 1 galaxies is dominated by nonthermal radiation from a synchrotron self-absorbed region of order a light day across. It should soon be possible to detect bright, relatively dust-free quasars and Seyfert 1 galaxies at 1.3 mm with the IRAM 30-m telescope and at 800 and 350 ym with the Clerk Maxwell Telescope. Determination of the shape of the far-infrared turnover will allow an important test of this conclusion, as different mechanisms will yield different shapes. It also appears that normal (low-polarization) quasars and Seyfert galaxies are not strongly variable in the far-infrared, although the IRAS data are not very well suited to monitor AGNs. It also seems that blazars vary much more slowly at far-infrared wavelengths than allowed by light-travel and beaming arguments. One possible explanation, which deserves theoretical study, is that a regulating mechanism may be operating in the central regions of AGNs. In any event, it is critical to make a more sensitive test of far-infrared variability in AGNs. It appears that the radio spectra of some Seyfert 1 galaxies flatten out near 2 cm. If confirmed with a larger sample, millimeter measurements of the shape of this new component can be used to determine if it arises from a compact radio core or the low-frequency side of the optical-infrared continuum. Also, low-frequency WSRT data suggest that the radio spectra of some objects may be turning over between 20 cm and 90 cm. If this is the case, source parameters such as sizes, ages, — 164 - and particle energies can be determined. Finally, a more speculative but very interesting approach involves exploring the relationship between the physics of AGNs and compact galactic objects. It may be possible to determine masses, periods, and other orbital parameters for these relatively simple and nearby systems. Understanding these objects could yield clues to the physical conditions in and nature of the emission from the more distant and complex AGNs. — 165 - REFERENCES Barvainas, R. 1986, preprint. Collin-Soufrin, S. 1986, preprint. Puetter, R. C. and Hubbard, E. N. 1985, Ap. J., 295, 394.