The Disks of Galaxies with Seyfert and Starburst Nuclei. I. Near-Infrared Colors and Color Gradients

Abstract

We present near-infrared (NIR) broadband and color images of 26 galaxies that host Seyfert 1, Seyfert 2, or starburst nuclei (SBNs). The study is focused on properties of the host galaxies rather than their nuclei, and to this end, careful attention is paid to photometric accuracy and to reliable measurements of the low surface brightness outer disk. Inspection of the elliptically averaged radial brightness and color profiles reveals that (1) the NIR mean colors of the inner and outer disks of Seyfert and starburst galaxies are consistent with those of a normal late-type stellar population and do not differ significantly with activity class; (2) the color gradients in the outer disks are similar both in sign and in magnitude to those observed in normal spirals; (3) red "ridges" in the inner parts of the J - H profile are evident in the majority of SBNs, but only in a few type 1 Seyferts and in no type 2's; (4) circumnuclear blue "dips" in the J - H profile are seen only in type 2 Seyferts. We then construct color images and find ridges, rings, and filaments, not evident in the broadband images, in the inner disks of SBNs and in NGC 7469, a Seyfert 1. The application of a simple model to these features yields evidence of both dust extinction and excess 2 μm emission. Color-color diagrams of individual pixels confirm these results and also show that the stellar mix in most of the Seyfert 2's comprises a conspicuous contribution from an intermediate-age [(3-5) × 108 yr] population. It appears that ongoing star formation in the inner disks of SBNs is signaled by the presence of dust (and gas); the absence of such features in both Seyfert types implies that star formation episodes are either absent or very old. However, while the blue colors of Seyfert 2's suggest that a burst of star formation did, in fact, occur not more than 109 yr ago, the normal colors of Seyfert 1's imply that any star-forming episodes must be significantly older.

Full text

THE ASTROPHYSICAL JOURNAL SUPPLEMENT SERIES, 108:229È260, 1997 January
1997. The American Astronomical Society. All rights reserved. Printed in U.S.A.(
THE DISKS OF GALAXIES WITH SEYFERT AND STARBURST NUCLEI.
I. NEAR-INFRARED COLORS AND COLOR GRADIENTS
L. K. HUNT
Centro per lÏAstronomia Infrarossa e lo Studio del Mezzo Interstellare, CNR, Largo Enrico Fermi 5, I-50125 Firenze, Italy;
hunt=arcetri.astro.it
M. A. MALKAN
Department of Astronomy, UCLA, 405 Hilgard Avenue, Los Angeles, CA 90024; malkan=bonnie.astro.ucla.edu
M. SALVATI
Osservatorio AstroÐsico di Arcetri, Largo Enrico Fermi 5, I-50125 Firenze, Italy; salvati=arcetri.astro.it
N. AND E.MANDOLESI PALAZZI
Istituto di Tecnologie e Studio delle Radiazioni Extraterrestri, CNR, via Gobetti 101, I-40126 Bologna, Italy;
reno=botes1.tesre.bo.cnr.it, eliana=tonno.tesre.bo.cnr.it
AND
R. WADE
Rutherford Appleton Laboratory, CLRC, Chilton, Didcot, Oxon OX11 0QX, England; rwade=tn.rl.ac.uk
Received 1996 March 11; accepted 1996 July 29
ABSTRACT
We present near-infrared (NIR) broadband and color images of 26 galaxies that host Seyfert 1, Seyfert
2, or starburst nuclei (SBNs). The study is focused on properties of the host galaxies rather than their
nuclei, and to this end, careful attention is paid to photometric accuracy and to reliable measurements of
the low surface brightness outer disk. Inspection of the elliptically averaged radial brightness and color
proÐles reveals that (1) the NIR mean colors of the inner and outer disks of Seyfert and starburst gal-
axies are consistent with those of a normal late-type stellar population and do not di†er signiÐcantly
with activity class; (2) the color gradients in the outer disks are similar both in sign and in magnitude to
those observed in normal spirals; (3) red ““ ridges ÏÏ in the inner parts of the J[HproÐle are evident in
the majority of SBNs, but only in a few type 1 Seyferts and in no type 2Ïs; (4) circumnuclear blue ““ dips ÏÏ
in the J[HproÐle are seen only in type 2 Seyferts. We then construct color images and Ðnd ridges,
rings, and Ðlaments, not evident in the broadband images, in the inner disks of SBNs and in NGC 7469,
a Seyfert 1. The application of a simple model to these features yields evidence of both dust extinction
and excess 2 km emission. Color-color diagrams of individual pixels conÐrm these results and also show
that the stellar mix in most of the Seyfert 2Ïs comprises a conspicuous contribution from an
intermediate-age [(3È5) ]108yr] population. It appears that ongoing star formation in the inner disks
of SBNs is signaled by the presence of dust (and gas); the absence of such features in both Seyfert types
implies that star formation episodes are either absent or very old. However, while the blue colors of
Seyfert 2Ïs suggest that a burst of star formation did, in fact, occur not more than 109yr ago, the normal
colors of Seyfert 1Ïs imply that any star-forming episodes must be signiÐcantly older.
Subject headings : dust, extinction È galaxies: photometry È galaxies : Seyfert È galaxies : starburst È
galaxies: structure È infrared: galaxies
1.INTRODUCTION
The relationship of a Seyfert nucleus to its host galaxy is
one of the outstanding questions in the study of active
galactic nuclei (AGNs). Properties of the Seyfert galaxy,
such as mass and luminosity concentration, morphological
type, relative dominance of the bulge and disk components,
metal abundance, and total luminosity, may be expected to
inÑuence nuclear activity and perhaps in turn be a†ected
by it.
The connection between Seyfert activity and intense star
formation is a further outstanding question. The standard
model for a Seyfert nucleus invokes accretion onto a
massive, compact object (cf. Rees 1977; Malkan 1983).
Either this scenario or a nuclear burst of star formation (see,
e.g., et al. requires a reservoir of gas and anTerlevich 1992)
efficient way to concentrate the gas into small nuclear
regions on timescales that are comparable to the duration
of the activity. Given that high rates of star formation seem
to be causally linked to high gas surface densities
one would(Elmegreen 1987; Larson 1985; Kennicutt 1989),
expect nuclear activity, whatever its cause, to be frequently
associated with intense star-forming episodes. There are
some indications from observations that star formation is
enhanced in the circumnuclear regions and disks of Seyfert
galaxies Rudy, & Jones(Rodriguez-Espinosa, 1987; Wilson
& Giovanardi although other obser-1988; Hunt 1992),
vations appear to contradict this et al.(Heckman 1989;
Carone 1992).
The problem of fuel supply and transport involves large
quantities of mass, D107È109most of which must beM_,
somehow brought in from extranuclear regions. Current
wisdom places the origin of this gas in the galactic disk (e.g.,
Begelman, & Frank & HernquistShlosman, 1989; Barnes
but the mechanism by which the gas is transported1991),
into the nuclear regions of the galaxy must reduce the spe-
ciÐc angular momentum of the gas by ?104since viscous
processes become important only on very small spatial
scales (>10 pc; et al. ItHernquist 1989; Shlosman 1989).
229

230 HUNT ET AL. Vol. 108
TABLE 1
SAMPLE GALAXIES
a]baDistancecExposure
Name Morphological Typea(arcmin2)zb(Mpc) AGa(s)
Seyfert 1
0048]29 ....... (R@)SB(s)b 0.9 ]0.9 0.0359 139.7 0.20 320, 320, 300
1614]35 ....... S? 0.8]0.7 0.0280 113.2 0.00 260, 240, 250
2237]07 ....... SBa 0.8]0.7 0.0250 96.4 0.28 300, 300, 300
Mrk 530 ........ SA(rs)b: pec 1.5 ]1.0 0.0290 112.2 0.12 300, 300, 300
Mrk 817 ........ S? 0.6]0.6 0.0314 126.6 0.00 400, 280, 364
Mrk 1243 ....... Sa 0.9]0.8 0.0353 144.0 0.06 600, 600, 600
NGC 5548...... (R@)SA(s)0/a 1.4 ]1.3 0.0166 69.2 0.00 240, 230, 240
NGC 5940...... SBab 0.8 ]0.8 0.0339 138.2 0.10 ..., 300, 250
NGC 7469...... (R@)SAB(rs)a 1.5 ]1.1 0.0160 60.2 0.13 300, 200, 150
Seyfert 2
1058]45 ....... Sa? 0.8]0.5 0.0191 78.4 0.00 360, 300, 300
1335]39 ....... S? 0.8]0.8 0.0201 82.8 0.00 600, 600, 500
Mrk 334 ........ Pec 1.0]0.7 0.0220 84.0 0.13 300, 300, 300
Mrk 533 ........ SA(r)bc pec 1.1 ]1.0 0.0289 111.7 0.11 400, 400, 300
Mrk 993 ........ Sa 2.2]0.7 0.0154 57.9 0.14 300, 300, 300
NGC 3362...... SABc 1.4 ]1.1 0.0227 94.1 0.05 ..., 300, 280
NGC 5674...... SABc 1.1 ]1.0 0.0248 102.4 0.06 600, 600, 600
NGC 7682...... SB(r)ab 1.2 ]1.1 0.0170 64.1 0.27 360, 400, 400
Starburst Nuclei
Mrk 31 ......... (R)SB(r)ab 0.9 ]0.9 0.0263 106.3 0.00 360, 300, 300
Mrk 307 ........ S 1.1]0.9 0.0185 70.5 0.17 360, 320, 400
Mrk 496 ........ L? 1.7]0.7 0.0296 119.1 0.00 600, 300, 300
Mrk 545 ........ SB(s)a 2.1 ]1.3 0.0152 56.9 0.16 320, 300, 300
Mrk 575 ........ (R@)SB(s)a 0.8 ]0.6 0.0183 69.4 0.18 240, 180, 180
Mrk 717 ........ S? 1.0]0.8 0.0209 86.2 0.11 520, 560, 300
Mrk 719 ........ C 0.3]0.2 0.0310 127.1 . . . ..., 360, . . .
Mrk 732 ........ E`pec 0.8 ]0.8 0.0292 120.3 0.02 300, 300, 300
Mrk 912 ........ S0/a sp pec 2.3 ]1.6 0.0160 60.6 0.18 300, 300, 300
aMorphological types, major and minor axes, and Galactic extinction in B-band magnitudes are taken from
Vaucouleurs et al.de 1991.
bRedshifts for the Seyfert galaxies are taken from and for the SBNs from &Edelson 1987 Mazzarella
Balzano 1986.
cCalculated from the redshifts, assuming km s~1 Mpc~1 and correcting for infall according toH0\75
& HuchraGeller 1983.
seems likely on both observational (e.g., Su, &Simkin,
Schwarz and theoretical & Shlosman1980) (Heller 1994)
grounds that the nuclear source is fed by disk gas, swept
inward by large-scale nonaxisymmetric perturbations in the
gravitational potential. These perturbations could arise
from intrinsic global instabilities or from tidal interactions
and should be associated with irregular morphology, bars,
rings, twisted isophotes, or oval distortions. Observational
evidence of such nonaxisymmetric structures in active gal-
axies is mounting (e.g., et al.Simkin 1980; MacKenty 1990;
Ma rquez, & Pe rez but studies restricted toMoles, 1995),
optical wave bands have some fundamental limitations.
The advantages of studying galaxies in the near-infrared
(NIR) have long been known: (1) the energy output of a
normal stellar population is dominated by old red giants
and therefore peaks around 1 km, and (2) dust extinction
is substantially reduced at these wavelengths. For both
reasons, the NIR properties of galaxies tend to be more
homogeneous and well deÐned than in the optical, and the
NIR wavelength bands are thus ideal when searching for
bars or oval distortions. Indeed, speciÐc cases of so-called
NIR bars are becoming more and more numerous (e.g.,
NGC 253, et al. NGC 1068, et al.Scoville 1985; Thronson
M82, et al. NGC 7469, et al.1989; Telesco 1991; Mazzarella
NIR wavelengths also facilitate studies of circumnu-1994).
clear regions in AGNs because the contrast between the
stellar and nonthermal contributions is maximized. On the
other hand, a 2.2 kmK-band excess can point to the hot
dust that is sometimes associated with violent star forma-
tion (see, e.g., et al. & GiovanardiJoseph 1984; Hunt 1992).
The chief difficulty with such observations has been the
sensitivity and number of pixels of infrared detectors. When
IRCAM became operational on the UK Infrared Telescope
(UKIRT), a number of groups (including ours) recognized
the opportunity to overcome these problems much more
e†ectively than had theretofore been possible.
We have adopted an observational approach to address
the connection between an active nucleus and its host
galaxy. In this paper, the Ðrst of a series, we present NIR
images of 26 galaxies that host Seyfert or starburst nuclei
(SBNs; see, e.g., et al. Subsequent papersFeldman 1982).
will describe the structural components of these galaxies,
together with the analysis of the nonaxisymmetric features
in the images (Paper II), and the amplitudes and colors of
the Seyfert and starburst nuclei (Paper III). Here we discuss
the sample selection in followed by a description of the°2,
observations and our data reduction and analysis in In°3.
particular, we determine, independently of the background
subtraction, NIR disk colors and color gradients of the
galaxies in our sample and compare their properties, in °4,

No. 1, 1997 HOST GALAXY DISKS. I. 231
with those of normal spirals. In we also present NIR°4
color images of the central regions of these galaxies, which
reveal structures visible only in the colors.
Unlike most previous NIR imaging studies of Seyfert
nuclei (Kotilainen et al. Kotilainen & Ward1992a, 1992b;
1994; Kotilainen, & Ward hereafterAlonso-Herrero, 1996;
collectively KW; et al. et al.Zitelli 1993; Danese 1992 ;
hereafter collectively ZD), this project was designed to
investigate properties of the host galaxies. Hence deep
images, acquired at UKIRT, were obtained in three bands
to measure the low surface brightness outer disk. The gal-
axiesÏ distances were constrained in the sample selection
(see below) so that, in most cases, it was possible to image
the outer disks without using mosaics. As a result, the range
of distances, as well as spatial resolution, of each activity
class subsample is similar and within a class varies by
roughly a factor of 2. Finally, special care was paid to the
determination of the background level and to photometric
corrections so that disk colors could be accurately deter-
mined and compared to the wealth of normal-galaxy photo-
metry in the literature.
2.SAMPLE SELECTION
The Seyfert galaxies studied here belong to the spectro-
scopically deÐned CfA sample &(Edelson 1987; Huchra
Burg To ensure that the galaxies could be suc-1992).
cessfully observed with the small-format (62 ]58) array in
the UKIRT camera (IRCAM1), we imposed a constraint on
redshift, zº0.015. Such a constraint ensured that the gal-
axies would be close enough that we could resolve relatively
small spatial structures, but distant enough so that they
would not overÐll the IRCAM array. We have obtained
images in three colors of more than half (nine of 17 Seyfert
1Ïs and eight of 14 Seyfert 2Ïs) of the subsample of those
galaxies in the CfA sample with zº0.015. For two particu-
larly large galaxiesÈNGC 5674 and 1335]39Èwe
obtained mosaics. The observed sample was selected essen-
tially randomly from the CfA list and, as mentioned above,
spans a range of about a factor of 2 in distance; listsTable 1
the global properties of the observed galaxies. Distances
were computed from the redshifts, correcting for infall
according to & Huchra with a Hubble con-Geller (1983),
stant km s~1 Mpc~1.H0\75
To test the alleged connections between starburst activity
and AGNs, we also imaged galaxies with starburst nuclei in
the same redshift range. These were selected from the survey
of SBNs taken from and &Balzano (1983) Mazzarella
Balzano It should be emphasized that this starburst(1986).
sample, selected for bright compact nuclei with H II regionÈ
like spectra, was chosen primarily for a comparative
analysis with Seyfert galaxies. Since Seyfert activity, as such,
is a strictly nuclear phenomenon, we felt that the starburst
comparison was more appropriate for nuclear starbursts,
and hence galaxies with extended starburst regions were not
included in the sample. In any case, one must keep in mind
that the Markarian lists from which these surveys are
drawn su†er from a selection bias (UV excess) that does not
apply to the CfA sample. Nevertheless, it has the advantage
of being homogeneously selected and allows us to search for
peculiarities shared by galaxies that host Seyfert or star-
burst nuclei. We have observed nine of the 43 SBNs with
zº0.015 in the Mazzarella & Balzano and Balzano lists.
Finally, in order to compare our results with a control
sample, we have extracted from the diameter-selected
sample of normal spirals of Jong & van der Kruitde (1994)
those galaxies with well-deÐned spiral types between Sa and
Sc, excluding S0 galaxies, SdÏs, and irregulars. To this sub-
sample, we applied the same redshift criterion as before and
obtained a set of 30 galaxies. Of these, eight have UKIRT
IRCAM images in two NIR broad bands (Hand K), and
R. de Jong has kindly provided the images and proÐles.
3.OBSERVATIONS,DATA REDUCTION,AND ANALYSIS
Data were obtained during the course of 10 nights on
UKIRT with IRCAM1 (1990 April, 1991 March, 1991
September); of these, Ðve nights were useful and, for the
most part, photometric. The camera was based on an InSb
Santa Barbara 62 ]58 array, and we adopted a pixel size of
Twenty-three of the galaxies were imaged in the three0A.62.
broadband Ðlters, J,H, and K; two were imaged in Hand
Konly, and one in H. The average seeing in the Kband was
and no data were obtained with seeingFWHM \1A.3,
worse than (FWHM).1A.8
A typical observing sequence consisted of Ðve exposures
ON source, interleaved with Ðve empty-sky exposures o†set
at least 90Afrom the center of the galaxy. Usually no more
than 120 s were spent in one telescope position, and this
time was divided into a number of co-additions to ensure
that the nuclei were not saturated and that the images were
background limited. With one exception (1614]35), gal-
axies were never observed at air masses greater than 1.5.
Dark exposures were acquired every 2 or 3 hours through-
out the night, as the dark current varied substantially (at
least in the Ðrst few hours); for each night, we were careful
to acquire at least one dark exposure taken with the same
integration time and co-additions as each of the source
exposures. Bias exposures taken before each integration
were automatically subtracted by the double-correlated
readout algorithm of the data acquisition system. Typical
total integration times are D300È500 s and are given indi-
vidually in Table 1.
As we wanted our images to be Ñat to better than
1]10~3, we paid close attention to the Ñat-Ðelding and
dark current subtraction. Dark current was determined by
performing, for each night, a pixel-by-pixel linear regression
on all the bias-subtracted dark exposures. We thus obtained
two Ðtted frames, a slope and an intercept, which we then
scaled and subtracted from each (bias-subtracted) science
frame. Tests show that the Ðts tend to reduce the noise
introduced in the image reduction, as opposed to using an
average dark frame acquired with similar observing param-
eters. These Ðts are interpolations, not extrapolations, as we
acquired dark frames at all of the object integration times
used during a night.
After dark current subtraction, Ñat Ðelds were con-
structed from pairs of sky frames acquired before and after
each source exposure. Each object frame was divided by the
normalized (to the image median), averaged pair of tempo-
rally adjacent sky frames, then registered and averaged,
after adjusting the o†set sky level to a common median. To
optimize the signal-to-noise ratio of the Ðnal images, we
chose to use a simple average, rather than a median, since
the Ñat Ðelds were di†erent for each source image, the sky
positions were chosen to be empty Ðelds, and the Ðeld of
view was small. We found no residual stars in the Ðnal
object frames.
Bad pixels were eliminated in the Ðrst approximation by
a bad-pixel list derived from an analysis of the dark frames,

232 HUNT ET AL.
and the correction for each of these entailed substitution
with the median of its (eight) neighbors. Remaining bad
pixels were eliminated manually from the Ðnal images with
linear interpolation. All image reduction was performed
with IRAF and the STSDAS packages.1
3.1. Photometric Calibration and Corrections
During the photometric nights, standard stars taken from
the UKIRT Bright Standard List (see & HawardenCasali
were observed every 2 hours. Each standard star was1992)
measured in two positions on the array. The zero-point Ñux
calibrations were typically accurate to 0.05 mag in J, 0.04
mag in H, and 0.03 mag in K. The stability of the system
was very good; zero points did not di†er by more than a few
percent from one night to the next during a run. Air-mass
corrections were applied to the standard stars and program
objects by using the mean UKIRT extinction coefficients of
0.10, 0.05, and 0.07 mag per air mass in J,H, and K, respec-
tively.
We applied a number of corrections to the photometric
calibration given by the air-massÈcorrected standard-star
observations. The Ðrst involves the chromatic change in the
IRCAM pixel scale. While the nominal pixel scale in the K
band is chromatic e†ects in the IRCAM optical train0A.62,2
make the J-band and H-band pixels larger by 4% and 1%,
respectively, and change the nominal calibration to mag
arcsec~2 by 0.022 in Hand 0.085 mag in J.
To analyze galaxy colors and compare them with those of
normal galaxies in the literature, we applied three addi-
tional corrections to the data. First, we transformed the
UKIRT photometry to the Caltech/Cerro Tololo Inter-
American Observatory (CIT) photometric system et(Frogel
al. et al. Such a transformation a†ects1978; Elias 1982).
mainly the UKIRT JÐlter, as it is somewhat bluer than that
of CIT & Hawarden We adopted the CIT(Casali 1992).
system, as it is frequently used in observations of normal
galaxies (e.g., and is the one on which theFrogel 1985)
UKIRT standard system is based. Second, we corrected for
Galactic extinction by using the Burstein-Heiles values of
the B-band extinction given in the Third Reference Cata-
logue of Bright Galaxies Vaucouleurs et al.(de 1991),
together with the interstellar extinction curve of Cardelli,
Clayton, & Mathis (1989).
The Ðnal correction was the K-correction for redshift,
determined according to the precepts of Frogel, &Persson,
Aaronson These are based on the rather red(1979).
(equivalent to a late-type K III star) colors of normal gal-
axies and are roughly linear with zfor small redshifts:
[0.5z,[3.5z, and ]3.3zfor J[H,H[K, and K, respec-
tively. For a redshift of 0.03, they are [0.015 for J[Hand
[0.11 mag for H[K. The transformed (to CIT) colors
corrected for Galactic extinction and redshift will be
referred to as and(J[H)0(H[K)0.
We have compared synthetic aperture photometry
obtained from our calibrated images with photometry in
the literature. The sense of all the comparisons is us versus
1IRAF is the Image Analysis and Reduction Facility, made available to
the astronomical community by the National Optical Astronomy Obser-
vatories, which are operated by AURA, Inc., under contract with the US
National Science Foundation. STSDAS is distributed by the Space Tele-
scope Science Institute, which is operated by the Association of Uni-
versities for Research in Astronomy (AURA), Inc., under NASA contract
NAS 5-26555.
2We were unable to check this, as we lacked the large Ðeld of view
necessary to perform astrometry; see, e.g., Jongde (1995).
them in units of magnitudes; NGC 7469 has been excluded
from the comparison because of its known NIR variability
(see, e.g., & Rieke The agreement with theLebofsky 1980).
photometry for the four galaxies in common with Balzano
& Weedman is good, with *K\[0.03 ^0.12 and(1981)
similar values in Jand Haperture). The agreement(8A.5
with other single-element photometer observations is
similar: *K\[0.07 for NGC 3362 (9A; & Moor-Glass
wood and *K\[0.08 for NGC 5548 (10A;et1985) Hunt
al. For the three galaxies common to this work and1994).
ZD, the photometry agrees reasonably well, with
*K\0.17 ^0.17 (5A), a di†erence within their estimated
accuracy of 0.1È0.15 mag. Two galaxies are also common to
this work and KW, and the agreement of the photometry is
good, with *K\[0.05 ^0.06 (12A). All things considered,
we estimate the e†ective uncertainty in our absolute cali-
bration to be D0.1 mag.
3.2. ProÐle Extraction
Elliptically averaged radial surface brightness proÐles
were extracted from the images (before sky subtraction; see
below) by Ðrst Ðtting the central peak of the galaxy in each
Ðlter with an axisymmetric two-dimensional Gaussian dis-
tribution. The center was then Ðxed, and ellipses were Ðtted
to the outer Jisophotes; the ellipticities and position angles
were determined by measuring the best-Ðtting ellipse at the
21 J-mag arcsec~2 isophote. This isophotal level is D100
times fainter than the sky and corresponds approximately
to the 3 pnoise limit of the images. We preferred to Ðx the
ellipse parameters, instead of letting them vary radially, so
that we could better compare proÐles in the di†erent pass-
bands. The major-axis increment of the radial proÐles was
deÐned to be 1 pixel, or (Kband).0A.62
Radial brightness proÐles in the Jband and radial color
proÐles are presented in together with gray-scaleFigure 1,
images with contours of the J-band image overlaid. Images
and proÐles in are all shown after performing theFigure 1
sky subtraction described in the following section.
3.3. Outer Disk Colors,Color Gradients,and
Sky L evel Determination
Because accurate surface photometry of the outer disks of
these galaxies requires sky level determinations accurate to
better than one part in 1000, after much experimentation we
adopted a new procedure for measuring the sky for each
galaxy. Although it utilizes one powerful constraining
assumption that may not always be valid, it has the advan-
tages of speed, simplicity, objectivity, and great precision.
We have exploited the techniques outlined in &Sparks
to determine, independently of sky level,JÔrgensen (1993)
mean colors and dimensionless color gradients in the outer
disk. We then incorporated these as Ðxed parameters in a
simultaneous s2-Ðtting procedure of the outer parts of the
J-, H-, and K-band surface brightness proÐles. The con-
straining assumption alluded to before is that the outer part
of each of the proÐles is described simply by the sum of a
constant sky level and an exponential law [I(r)\I(0) exp
We therefore Ðtted the three proÐles simulta-([r/r0)].
neously for Ðve parameters, the sky levels in each of the
three bands and the surface brightness and scale length of
the disk in one band. The average colors and color gra-
dients, derived independently, constrain the disk Ðts in the
remaining bands. Details of the method are given in
Appendix A.

FIG. 1.ÈJ-magnitude images, J[Kcolor images, and elliptically averaged radial brightness and color proÐles. Galaxies are grouped by activity type,
with Seyfert 1Ïs followed by Seyfert 2Ïs and SBNs. For each galaxy, the J-band image is shown in the upper left panel, the J[K(expanded; see text) color
image in the lower panel, and the J-band brightness, J[H,H[K, and J[Kcolor proÐles in the right panel. The broadband images are contoured from 23
to 16 J-mag arcsec~2, the color J[Kimages from 0.4 to 1.35. The horizontal dotted line in each of the color proÐles corresponds to the average outer disk
color as determined from the Ñux-Ñux plots the vertical dotted line indicates the 3 kpc cuto† between ““ innerÏÏ and outer disk. The dashed lines(Appendix A);
in the brightness and color proÐle panels show the disk Ðts described in All magnitudes and colors are those observed, before application of theAppendix A.
photometric corrections described in the text.

FIG. 1.ÈContinued
234

FIG. 1.ÈContinued
235

FIG. 1.ÈContinued

HOST GALAXY DISKS. I. 237
FIG. 1.ÈContinued
As an after-the-fact justiÐcation for this technique, we
note that nearly all the proÐles, in magnitude versus radius
(i.e., logarithmic vs. linear) plots, are extremely well charac-
terized by very straight lines over the outer two-thirds to
three-quarters or more of their points (see In fact, weFig. 1).
do not have any cases in which the outer half of the proÐle is
not well described by an exponential law, a behavior similar
to that generally observed in normal spiral disks (see, e.g.,
Watanabe, &Freeman 1970; Boroson 1981; Kodaira,
Okamura 1986).
3.4. Inner Disk Colors
In addition to the outer disk colors obtained indepen-
dently of sky subtraction as described in we also deter-°3.3,
mined colors of the ““ inner disks.ÏÏ This was done by
calculating averages of the radial color proÐles between 4A
and 3 kpc. An inner limit of 4Awas used because the color
proÐles show that, for bright and red nuclei, the nuclear
colors can contaminate the galaxyÏs proÐle out to roughly
this apparent radius (see, e.g., Mrk 530 and Mrk 817 in Fig.
The outer radius was the lower limit of the radial range1).
used to Ðt for the average colors and color gradients in the
outer disk (see Photometric corrections, asAppendix A).
described in were applied to both inner and outer disk°3.2,
colors. The colors thus obtained are reported in Table 2,
along with mean values for each activity type obtained by
simple averages over all the galaxies in each subsample.
3.5. Two-dimensional Color Images
To construct the color images, very accurate alignment is
necessary. Because the small Ðeld of view of IRCAM1 mini-
mizes the probability of having stars in the galaxy Ðeld, we
were forced to align images by using other techniques. To
determine the alignment o†sets, we Ðrst subjected the
images to a 3 ]3 bilinearly interpolated magniÐcation ; the
exact expansion factor varies from band to band, as it is
necessary to take into account the di†erent pixel sizes in the
di†erent bands. The shifts were then determined in two
ways: (1) from the maximum of the two-dimensional cross-
correlation function of the expanded (sky-subtracted Ñux)
images and (2) from the relative shifts of the peak of the
galaxy emission, as determined by a two-dimensional
axisymmetric Gaussian Ðt. In all but a few cases, the two
methods of determining the relative alignment yielded
similar results, implying that the nuclei are not sufficiently
reddened to be o†set spatially in the di†erent NIR pass-
bands. In those few cases in which the two methods did not
yield consistent shifts, usually where there was structure in
the color images or objects on the edge of the frame, we
aligned the images by assuming that the peaks of the galaxy
emission in the di†erent wave bands coincide. Finally, the
magnitude images were registered to a fraction of a
(magniÐed) pixel by using polynomial interpolation and
subtracted. In this way, we generated color images in J[K
and J[H.
A sample of these color images is shown in color in
(Plate 1). The remainder are shown in asFigure 2 Figure 1
gray-scale plots, with the direct J-band image adjacent.
4.RESULTS
We defer a detailed analysis of the general morphology,
the radial brightness proÐles, and the nonaxisymmetric fea-

FIG. 1.ÈContinued
238

FIG. 1.ÈContinued
239

FIG. 1.ÈContinued
240

FIG. 1.ÈContinued
241

FIG. 1.ÈContinued
242

FIG. 1.ÈContinued
243

FIG. 1.ÈContinued
244

FIG. 1.ÈContinued
245

246 HUNT ET AL. Vol. 108
FIG. 1.ÈContinued
tures of the broadband images to Paper II. Here we discuss
the average disk colors, color gradients, and the structure in
the color images. Note that the mean colors and color gra-
dients are measured from azimuthally averaged major-axis
proÐles, rather than from virtual-aperture photometry.
4.1. Average Disk Colors
We Ðnd no signiÐcant di†erence in inner or outer disk
colors between type 1 and type 2 Seyferts, nor between the
Seyfert galaxies and the SBNs. More speciÐcally, although
Seyfert 2Ïs may have slightly blue colors (see theTable 2),
outer disk colors of all three classes are consistent with
those of normal spiral disks, as shown in the two-color
diagram in the sample medians (J[H\0.68 andFigure 3;
H[K\0.22) are equivalent to those found by other
groups for normal spiral galaxies Hyland, &(Griersmith,
Jones Giovanardi & Hunt1982 ; Glass 1984; Frogel 1985;
1988, 1996).
The inner disk colors of type 1 Seyferts and SBNs tend to
be slightly redder in J[Hthan those of the outer disk and
lie at the blue end of the range of NIR colors of Sa galaxies,
dominated by massive metal-rich bulges &(Giovanardi
Hunt Indeed, what we call ““ inner disk ÏÏ colors1996).
should probably be called ““ colors of the bulge ÏÏ (see also
et al. In any case, the spread in both sets ofTerndrup 1994).
colors is remarkably small, given that these galaxies host
active nuclei. We show the H[Kcolors of the inner disk as
histograms in All four (Seyfert 1Ïs, Seyfert 2Ïs,Figure 4.
SBNs, normal spirals) of our color distributions are indis-
tinguishable, in mean and scatter, from one another.
FIG. 3.ÈColor-color diagram of the mean inner and outer disk cor-
rected colors for each activity class. The error bars correspond to the
spread (standard deviation) of the colors within each activity class. Circles
show Seyfert 1Ïs; squares, Seyfert 2Ïs; and triangles, SBNs; Ðlled symbols
indicate the ““ inner ÏÏ disk, and open symbols the outer disk. Dotted lines
show the spread of colors for elliptical and Sc systems and for(Frogel 1985)
Sa galaxies & Hunt The locus of the K-correction is(Giovanardi 1996).
also shown (departing from the average colors), with z-increments of 0.01
shown by plus signs. A visual extinction of 1 mag is shown by the arrow in
the lower right corner.

No. 1, 1997 HOST GALAXY DISKS. I. 247
TABLE 2
AVERAGE DISK COLORS
““ INNER ÏÏ OUTER
Rinner
NAME (kpc) (J[H)0(H[K)0(J[H)0(H[K)0
Seyfert 1
0048]29......... 2.5 0.77 ^0.00 0.26 ^0.00 0.76 ^0.01 0.22 ^0.01
1614]35......... 2.4 0.77 ^0.01 0.10 ^0.01 0.72 ^0.02 0.15 ^0.02
2237]07......... 2.3 0.60 ^0.01 0.29 ^0.02 0.54 ^0.04 0.35 ^0.03
Mrk 530 ......... 2.3 0.71 ^0.00 0.20^0.01 0.70 ^0.00 0.20 ^0.01
Mrk 817 ......... 2.5 0.72 ^0.00 0.28^0.00 0.73 ^0.02 0.25 ^0.02
Mrk 1243 ........ 2.5 0.63 ^0.00 0.27 ^0.00 0.60 ^0.15 0.25 ^0.11
NGC 5548 ....... 2.1 0.72 ^0.01 0.22 ^0.02 0.64 ^0.01 0.28 ^0.01
NGC 5940 ....... 2.5 ... 0.15 ^0.00 . . . 0.20 ^0.01
NGC 7469 ....... 2.0 0.72 ^0.05 0.30 ^0.05 0.70 ^0.04 0.19 ^0.04
Mean .......... ... 0.71 ^0.06 0.23 ^0.07 0.67 ^0.07 0.23 ^0.06
Seyfert 2
1058]45......... 2.1 0.64 ^0.02 0.23 ^0.03 0.68 ^0.10 0.16 ^0.07
1335]39......... 2.2 0.67 ^0.03 0.19 ^0.04 0.71 ^0.02 0.18 ^0.01
Mrk 334 ......... 2.0 0.67 ^0.03 0.22^0.01 0.66 ^0.03 0.11 ^0.03
Mrk 533 ......... 2.3 0.71 ^0.01 0.25^0.00 0.72 ^0.01 0.25 ^0.01
Mrk 993 ......... 1.9 0.69 ^0.01 0.22^0.01 0.68 ^0.01 0.25 ^0.01
NGC 3362 ....... 2.3 ... 0.21 ^0.01 . . . 0.23 ^0.01
NGC 5674 ....... 2.1 0.62 ^0.01 0.22 ^0.01 0.65 ^0.02 0.19 ^0.02
NGC 7682 ....... 1.9 0.70 ^0.03 0.19 ^0.01 0.85 ^0.09 0.11 ^0.07
Mean .......... ... 0.67 ^0.03 0.22 ^0.02 0.71 ^0.07 0.19 ^0.06
Starburt Nuclei
Mrk 31 ........... 2.2 0.64 ^0.02 0.18 ^0.05 0.55 ^0.02 0.18 ^0.02
Mrk 307 ......... 2.1 0.64 ^0.02 0.21^0.01 0.65 ^0.01 0.22 ^0.02
Mrk 496 ......... 2.5 0.73 ^0.04 0.29^0.00 0.83 ^0.04 0.34 ^0.03
Mrk 545 ......... 1.9 0.74 ^0.02 0.32^0.01 0.67 ^0.07 0.35 ^0.05
Mrk 575 ......... 2.1 0.74 ^0.02 0.23^0.01 0.78 ^0.01 0.28 ^0.01
Mrk 717 ......... 2.1 0.76 ^0.02 0.19^0.03 0.73 ^0.01 0.07 ^0.01
Mrk 719 ......... ... ... ... ... ...
Mrk 732 ......... 2.5 0.68 ^0.00 0.29^0.01 0.63 ^0.01 0.25 ^0.01
Mrk 912 ......... 2.0 0.68 ^0.03 0.24^0.01 0.63 ^0.01 0.23 ^0.01
Mean .......... ... 0.70 ^0.05 0.24 ^0.05 0.68 ^0.09 0.24 ^0.09
Normal Galaxies
UGC 89 .......... 2.0 ... 0.29 ^0.02 . . . 0.30 ^0.01
UGC 628 ........ 2.0 ... 0.14 ^0.02 . . . 0.26 ^0.08
UGC 3066 ....... 1.9 ... 0.22 ^0.02 . . . 0.23 ^0.01
UGC 3140 ....... 1.9 ... 0.18 ^0.02 . . . 0.22 ^0.08
UGC 4126 ....... 2.0 ... 0.15 ^0.01 . . . 0.15 ^0.01
UGC 4256 ....... 1.9 ... 0.29 ^0.03 . . . 0.22 ^0.01
UGC 4458 ....... 2.0 ... 0.21 ^0.01 . . . 0.16 ^0.01
UGC 12845...... 1.8 ... 0.16 ^0.01 . . . 0.17 ^0.01
Mean .......... ... ... 0.21 ^0.05 . . . 0.20 ^0.06
Our result contradicts both ZD and KW, who found
Seyfert host galaxies to have a normal distribution in J[H,
but with an H[Kcolor that (1) is redder on average than in
normal galaxies and (2) has a substantially wider dispersion
than is observed in normal spirals. There are two possible
reasons for this contradiction. The Ðrst is that both ZD and
KW used colors determined from annular rings somewhat
close to the nucleus radii and 3AÈ6Aradii,(2A.5È3A.9
respectively), so that there is a chance of spillover from the
bright nonstellar AGN (cf. color proÐles in TheFig. 1).
second possible reason for the disagreement is that both
groups neglected to apply the K-correction; as mentioned
in this correction is 7 times larger for the H[Kcolor°3.1,
than for J[H, so that not applying it would result in colors
substantially normal in J[Hbut redder than normal in
H[K.
We have checked this last hypothesis by correcting the
host galaxy colors reported in ZD and KW for redshift.
They found sample medians for the host galaxy colors of
0.77 (ZD) and 0.71 (KW) for J[Hand 0.45 (ZD) and 0.49
(KW) for H[K. After applying the K-correction, the
sample medians are substantially unchanged in J[H, 0.75
(ZD) and 0.71 (KW), but are signiÐcantly di†erent in H[K:
0.36 (ZD) and 0.32 (KW). The corrected H[Kvalues are
shown in the histogram in The corrected colors ofFigure 4.
both groups coincide very well with our result for Seyfert 1Ïs
(inner disk), and the peaks of the distributions are consistent
with colors of normal galaxies.
4.2. Color Gradients
A visual inspection of the azimuthally averaged radial
color proÐles illustrated in shows that, of eightFigure 1
galaxies with starburst nuclei, six present red circumnuclear
““ ridges ÏÏ in J[Hcolor. In contrast, only three of eight
galaxies with type 1 Seyfert nuclei and none of the type 2
Seyfert nuclei show the same tendency. Another phenome-

248 HUNT ET AL. Vol. 108
FIG. 4.ÈHistograms of the corrected inner disk color as a(H[K)0
function of activity type. The top two panels show colors taken from
& Ward and from et al. after correctingKotilainen (1994) Danese (1992),
for K-dimming as described in the text. The bottom panel shows
(corrected) normal spiral colors derived from the radial proÐles in Jongde
& van der Kruit (1994).
non is shown only by the Seyfert 2Ïs, namely, a circumnu-
clear blue ““ dip ÏÏ in the J[Hcolor. We will discuss these
trends in more detail in connection with the structure in the
color images (°4.3).
4.2.1. Inner versus Outer Colors
The inner and outer NIR disk colors are plotted against
each other in the solid lines in the Ðgure indicateFigure 5;
equality. The majority of type 1 Seyferts and SBNs present
J[Hcolors that are redder within 3 kpc than those in the
outer disk; in contrast, the majority of type 2 Seyferts show
bluer J[Hcolors in the inner disk. The trend for redder
J[Hin the inner regions is weak in any case, with a mean
amplitude of 0.02È0.03 mag (see also The change inFig. 3).
H[Kfrom inner to outer disk shows more variation, as
approximately half the galaxies have redder and half have
bluer H[Kin the inner part; this behavior in H[Kis also
seen in the Jong & van der Kruit normal spiralsde (1994)
(denoted by crosses in Fig. 5).
The color behavior shown by the galaxies in our sample
is similar to that of normal spirals. et al.Terndrup (1994)
analyzed Jand Kimages of normal spirals and found that
bulge (roughly equivalent to our ““ inner ÏÏ) and disk (outer)
J[Kcolors are roughly the same. Their data are shown by
crosses in the J[Kpanel of The redder inner diskFigure 5.
J[Hcolors in Seyfert 1Ïs and SBNs are evident in the
upper right panel, as all but three lie above the equality line.
The e†ect is diluted in the J[Kcolor, however, and the
behavior shown by our galaxies is similar to that of Tern-
drup et al.Ïs sample. Multiaperture photometry also shows
that bulge NIR colors tend to resemble those of the disk
component, with redder colors in the inner (few kiloparsec)
regions et al. We therefore(Griersmith 1982; Frogel 1985).
hesitate to attribute the slightly redder inner disk colors
seen in Seyfert 1Ïs and SBNs to processes associated with
the active nucleus or star formation, as it seems more plaus-
ible that the redder colors are due to e†ects that also occur
in normal spirals, that is, a red bulge component that domi-
nates the inner regions, or extinction, or both.
4.2.2. Outer Disk Gradients
The color gradients measured for sky subtraction (see
are also of interest in their own right. ToAppendix A)
interpret the results from the generalized polynomial Ðtting
procedure that we used to derive them, we Ðrst need to
impose a parametric model for the disk light. This is easily
accomplished using the generalized exponential expression
given in Sparks (1988):
I(s)\Ieexp [[a(sc[1)] , (1)
where sis a dimensionless radius and is the e†ective(r/re)re
(half-light) radius. For an exponential disk, a\1.68 and
c\1 (see Following &Sparks 1988). Sparks JÔrgensen
we suppose that the radial proÐle is described by(1993), at a Ðrst wavelength with scale length andequation (1) re
at a second wavelength with scale length re
@\re]*re4
The second-order term, in the Chebyshevre(1 ]*s). a2,
polynomial Ðt to the Ñuxes at two di†erent wavelengths is
simply related to *s:
*sB[3/cB[3a
2. (2)
The for our sample are listed in together witha2Ïs Table 3,
the mean radius over which the gradient was determined.
shows the fractional scale length change *sas aFigure 6
function of activity type, with Seyferts denoted by 1 and 2,
SBNs by 3, and normal galaxies by 4.
We included a given outer disk color gradient in the sky
Ðts if it was determined with a signal-to-noise ratio º3.
Using this same criterion, we Ðnd that four of nine Seyfert
1Ïs, two of eight Seyfert 2Ïs, and six of eight SBNs present
signiÐcant nonzero gradients in at least one of the two
colors. In the normal spirals, for which we only have H[K,
we Ðnd two of eight with signiÐcant nonzero gradients. The
sign of the gradient in each of the four samples is evenly
distributed, with half of the signiÐcant gradients positive
and half negative. It seems that the largest gradients seen in
are from Jto Hand in the positive sense, but theFigure 6
galaxies that present the largest gradients are those for
which the assumption of an exponential disk is probably
not valid, namely, 1614]35, Mrk 575, and Mrk 496. The
Ðrst two are systems with prominent bars, and the third is
the result of a merger.
Although we used a di†erent method to derive color gra-
dients, we can compare our results to those found by
et al. for normal spirals. For a color gra-Terndrup (1994)
dient of the form
*(kJ[kK)\b* log r, (3)
we Ðnd, from and assuming an exponentialequation (1)
disk, that
bB4.2SsT*s. (4)

No. 1, 1997 HOST GALAXY DISKS. I. 249
FIG. 5.ÈCorrected inner and outer disk colors. The upper left panel shows J[K, the upper right panel J[H, and the lower panel H[K. The solid line
indicates equality. Seyfert 1Ïs are shown by circles, Seyfert 2Ïs by squares, and SBNs by triangles. Typical error bars are shown in the upper left corner of each
plot. In the J[Kpanel, crosses show normal spirals from et al. in the H[Kpanel, crosses show normal spirals from Jong & van derTerndrup (1994); de
Kruit (1994).
The mean value of s,SsT, can be deÐned as the mean
radius of the interval used for the determination of Tern-a2.
drup et al. measured disk gradients roughly from 0.8 to 1.8
of (disk exponential folding length), so we assume anr0of 1.3. Since we haveSrT/r0SrT/r0\1.68SsT\1.3,
SsT\0.77, and bB3.25*s. The largest bÏs in their sample
are positive with amplitudes of 0.6È0.7, corresponding to a
fractional scale length change of D0.2 from Jto K, compa-
rable to the largest value we Ðnd for our galaxies. More-
over, the gradients for normal galaxies found by Terndrup
et al. are roughly evenly distributed between positive and
negative values, as are the gradients measured here. It
appears that the color gradients in our galaxies are more or
less normal, both in sign and in amplitude.
4.3. Structure in the Color Images
Although quantitative analyses of galaxy images are
usually based on azimuthally averaged data, not all photo-
metrically interesting structures possess azimuthal sym-
metry. In fact we have found some of these in the
two-dimensional images.
In several cases there is a clear color gradient that is not
radial, but azimuthal or more complex. Inspection of Figure
shows cases in which the J[Kcolor weakly delineates1
spiral structure (e.g., 1335]39, Mrk 334, Mrk 533, Mrk
307) or bars (e.g., 1614]35, NGC 5674, Mrk 575). The
J[Kimages also show more exotic structures like rings
and Ðlaments in one type 1 Seyfert (NGC 7469) and in Ðve
SBNs (Mrk 496, Mrk 545, Mrk 717, Mrk 732, Mrk 912).
Such nonaxisymmetric red structure in active galaxies
can be caused primarily by two mechanisms: dust extinc-
tion or excess 2 km emission (dust in emission or H II
regions; see also & Giovanardi We have inves-Hunt 1992).
tigated the alternatives of extinction or excess emission for
the galaxies in our sample by applying a simple model to
the color images. First, we compute the optical depth pixel-
by-pixel from the color image by assuming that the(J[H)0
dust is in front of the stars emitting the NIR Ñux (see

250 HUNT ET AL. Vol. 108
TABLE 3
OUTER DISK COLOR GRADIENTS
a2^pa2
NAME Ravg Kvs. JHvs. JKvs. H
Seyfert 1
0048]29......... 6.3 0.0147 ^0.004 0.0053 ^0.004 . . .
1614]35......... 7.5 [0.0489 ^0.017 [0.0723 ^0.012 . . .
2237]07......... 5.2 0.0136 ^0.014 0.0117 ^0.013 . . .
Mrk 530 ......... 6.9 0.0007 ^0.003 0.0015 ^0.002 . . .
Mrk 817 ......... 6.8 [0.0122 ^0.015 [0.0176 ^0.008 . . .
Mrk 1243 ........ 6.4 0.0070 ^0.006 [0.0391 ^0.039 . . .
NGC 5548 ....... 4.8 [0.0321 ^0.009 0.0086 ^0.007 . . .
NGC 5940 ....... 8.7 ... ... 0.0261 ^0.007
NGC 7469 ....... 4.7 0.0087 ^0.022 [0.0352 ^0.021 . . .
Seyfert 2
1058]45......... 4.9 0.0126 ^0.011 0.0035 ^0.012 . . .
1335]39......... 4.7 [0.0045 ^0.003 [0.0330 ^0.014 . . .
Mrk 334 ......... 3.8 0.0211 ^0.015 0.0145 ^0.009 . . .
Mrk 533 ......... 6.1 0.0013 ^0.006 0.0122 ^0.004 . . .
Mrk 993 ......... 6.4 [0.0062 ^0.003 [0.0154 ^0.017 . . .
NGC 3362 ....... 5.9 ... ... 0.0211 ^0.009
NGC 5674 ....... 6.8 [0.0005 ^0.005 [0.0008 ^0.008 . . .
NGC 7682 ....... 5.4 [0.0185 ^0.003 [0.0163 ^0.002 . . .
Starburst Nuclei
Mrk 31 ........... 5.8 0.0332 ^0.008 0.0095 ^0.011 . . .
Mrk 307 ......... 4.4 [0.0363 ^0.013 0.0137 ^0.006 . . .
Mrk 496 ......... 6.5 [0.1007 ^0.008 [0.0773 ^0.020 . . .
Mrk 545 ......... 5.7 0.0002 ^0.002 0.0209 ^0.023 . . .
Mrk 575 ......... 5.0 [0.0528 ^0.009 [0.0415 ^0.007 . . .
Mrk 717 ......... 6.5 0.0382 ^0.005 0.0092 ^0.004 . . .
Mrk 719 ......... .. . ... ... ...
Mrk 732 ......... 5.4 0.0307 ^0.007 0.0112 ^0.007 . . .
Mrk 912 ......... 5.1 0.0019 ^0.005 0.0032 ^0.008 . . .
Normal Galaxies
UGC 89 .......... 9.5 ... .. . 0.0197 ^0.003
UGC 628 ........ 5.9 ... ... 0.0139 ^0.046
UGC 3066 ....... 6.8 ... ... [0.0181 ^0.009
UGC 3140 ....... 7.4 ... ... [0.0008 ^0.008
UGC 4126 ....... 10.2 . . . . . . [0.0018 ^0.004
UGC 4256 ....... 9.1 ... ... [0.0108 ^0.004
UGC 4458 ....... 8.8 ... ... [0.0070 ^0.003
UGC 12845...... 8.8 ... ... [0.0475 ^0.005
et al. then we apply the extinction so derivedTelesco 1991);
to the image. Assuming that a red H[Kcolor is(H[K)0
due to reddened stellar emission plus some nonstellar pro-
cesses that emit only in the Kband, we can derive an image
that maps the excess (over the reddened stellar) K-band
emission:
H[K\(H[K)0(stars) ]1.086(qH[qK)
]2.5 log C1]&K(excess)
&K(stars) D, (5)
where is the K-band surface brightness in Ñux units of&K
the excess and stellar radiation. The excess is given as the
last term in We calculated the visual opticalequation (5).
depth, by applying the et al. interstellarqV, Cardelli (1989)
extinction curve, and instead of adopting the corrected
outer disk colors as the intrinsic colors, we used the sample
medians of 0.68 (in J[H) and 0.22 (H[K). The sample
medians were preferred to the individual corrected colors
for each galaxy, as we found that the azimuthally averaged
outer disk colors can frequently be as much as 0.1 mag
redder than those outside the red structures. The implica-
tion is that even azimuthal averages can be a†ected by suffi-
ciently signiÐcant structure in the colors.
The assumption of the dustÏs acting as a foreground
screen instead of being uniformly mixed with the stars
(hereafter denoted by ““ true ÏÏ) simpliÐes the calculation of
the K-band excess but underestimates the dustÏs optical
depth (see, e.g., moreover, such an assump-Mathis 1970);
tion may spuriously imply excess long-wavelength emission
when in fact there is none. We have therefore checked the
validity of the simple screen model on these two points: (1)
how well the optical depth estimated by the screen modelqV
corresponds to the ““ true ÏÏ one and (2) how the amplitude of
any spurious excess emission varies with The results ofqV.
these checks are given in and show that, for anAppendix B
inferred ““ screenÏÏ of there is a simple multiplicativeqV[2.0,
factor (D2.3) relating the ““ true ÏÏ and the ““ screen ÏÏ qV.
Under these conditions, the spurious implied excess emis-
sion fraction (the error on the last term in iseq. [5]) [0.01.

No. 1, 1997 HOST GALAXY DISKS. I. 251
FIG. 6.ÈOuter disk gradients as a function of activity type. The plotted quantity is the fractional change in scale length, *s, obtained from the
second-order term in the Chebyshev polynomial Ðt to the Ñux-Ñux proÐles as described in the text and in error bars are the formal errorsa2Appendix A;
associated with the polynomial Ðt. The upper left panel shows J[K, the upper right panel J[H, and the lower panel H[K. The activity-type spacing is
arbitrary; Seyfert 1Ïs are shown by circles, Seyfert 2Ïs by squares, and SBNs by triangles (and denoted by ““ activity type ÏÏ 3). For convenience, a zero gradient
is shown by the horizontal dotted line.
Hence, as long as the ““ screen ÏÏ optical depths we derive are
and the excess emission fractions are (5 p), the[2.0 Z0.05
excess should be real. We emphasize that, while seemingly
quantitative, this technique is really only a qualitative esti-
mate of both the extinction and excess K-band Ñux. We use
it to diagnose the presence of dust and excess 2 km emis-
sion, not to make deÐnite statements about the amplitude of
the optical depth or precise estimates of the quantity of
nonstellar emission, as both quantities strongly depend on
the assumed intrinsic colors, the geometry of the dust and
stars, and the properties of the excess emission, such as dust
temperature and emissivity.
4.3.1. Spiral Structure,Bars,and L ens Distortions
The spiral arms in 1335]39, Mrk 307, Mrk 334, and
Mrk 533 are characterized by *(J[K)B0.14È0.18 across
the arm. These are similar to the color changes across the
spiral-arm dust lanes in M51 reported by & RiekeRix
who found *(J[K) ranging from 0.13 to 0.21. Our(1993),
simple model yields (screen) ranging from 0.3 to 0.6 in theqV
arms, which corresponds to ““ true ÏÏ optical depths qV
between 1 and 2. We note that these inferred ““ true ÏÏ optical
depths are roughly a factor of 2 lower than those derived by
Rix & Rieke, who applied a detailed radiative transfer
model to multicolor optical and NIR data; evidently, even
the internal extinction model tends to underestimate the
true extinction. We Ðnd no evidence of excess K-band emis-
sion in the arms.
Oval distortions and bars show less obvious color trends.
The two galaxies with clear oval or lens distortions, Mrk
1243 (Seyfert 1) and Mrk 732 (SBN), show negligible J[K
color gradients associated with the lens, as the J[Kcolor
image for Mrk 1243 is roughly axisymmetric and that for
Mrk 732 is confused due to the circumnuclear ring (see
below). The bars in 1614]35, Mrk 575, and NGC 5674, on
the other hand, all show D0.14È0.18 mag of reddening in
J[Kalong the bar. Again, taken at face value, these color
gradients correspond to (““ true ÏÏ) optical depths in the V

FIG. 7.ÈExtinction and excess emission maps. The left panels show contour plots of the visual optical depth and the right panels show the excess 2 kmqV,
emission, both obtained from the simple screen model described in the text. The contours run from 0 to 2 with increments of 0.2, and from 0 to 1 with 0.1qV
increments for the excess. They are overlaid on the J-band magnitude image, and arcsecond increments in right ascension and declination are shown on the
coordinate axes. Notes on individual galaxiesÈ(1) Mrk 496: extinction peaks between the two continuum maxima, while a ridge of excess emission is evident
to the northeast of the main nucleus; (2) Mrk 545: northeast and southwest Ðlaments spiraling in toward nucleus, culminating in a circumnuclear ring (cf.Fig.
that is overshadowed by the J-band image; (3) Mrk 732: dust extinction and emission to the southwest of the nucleus, roughly coincident with the end of1)
the oval distortion; (4) Mrk 912: broad ridge of excess emission to the south of the nucleus, together with a small amount of extinction; (5) NGC 7469:
circumnuclear extinction and hot dust emission extends to the north and south of the nucleus and may coincide with the inner spiral noted by etMazzarella
al. noteworthy are the northern extinction and emission ““ ridge.ÏÏ(1994);

HOST GALAXY DISKS. I. 253
FIG. 7.ÈContinued
band between 1 and 2. Again, as for the spiral arms, neither
bars nor oval distortions show evidence of excess K-band
Ñux.
4.3.2. Rings,Ridges,and Filaments
Perhaps the most striking features revealed in the color
images are the rings and Ðlaments in the SBNs. Of eight
SBNs with JHK colors, Ðve show red circumnuclear rings
(Mrk 545, Mrk 717, Mrk 732, Mrk 912), nonaxisymmetric
circumnuclear ridges (Mrk 496), or Ðlaments (Mrk 545).
These galaxies also show the circumnuclear ““ ridge ÏÏ in the
J[Hcolor proÐle noted in Of eight type 1 Seyferts,°4.2.
only NGC 7469 shows an extended red circumnuclear ring
plus ridge (and circumnuclear ridges in the color proÐles);
none of the type 2 Seyferts present similar structure. We
note that such features in the SBNs are particularly sur-
prising since the SBNs were selected for their nuclear
properties (see not for evidence of any extended star-°2),
burst region.
All these features provide evidence of dust in extinction
and excess 2 km emission. shows contour plots ofFigure 7
the maps, together with maps of the excess K-band emis-qV
sion. While the knots in the outer regions are due to noise,
the contiguity of the inner contours implies that the features
in both the optical depths and the excess emission corre-
spond to real features. Values for (screen) in the rings,qV
ridges, and Ðlaments range from 0.2 to D1.5, corresponding
to ““ true ÏÏ visual optical depths of up to 3 or 4. Excess
emission fractions in these features range from 0.1 to 0.3 or
more; the implication is that, locally, excess nonstellar emis-
sion can constitute as much as 30% of the stellar K-band
emission.
Red JHK extranuclear colors observed primarily in star-
bursts have also been noted in previous studies. etForbes
al. on the basis of NIR images of LINERs and star-(1992),
bursts, found evidence of variations in obscuration in star-
bursts and hot dust close to the nucleus. Low-resolution
maps of the NIR radial color distribution in Seyfert and
starburst galaxies & Giovanardi show red(Hunt 1992)
extranuclear JHK colors only in starbursts, colors that are
consistent with internal extinction, emission by hot dust, or
both.
4.3.3. Color-Color Diagrams
To conÐrm the nature of the features described in the
preceding subsections, we have produced pixel-by-pixel
color-color diagrams for our sample galaxies. The pixels are
extracted directly from the and images,(J[H)0(H[K)0
and representative examples are shown in togetherFigure 8
with mixing curves for various physical processes. To con-
serve some spatial information, we have coded the nuclear
regions (R¹4A) as Ðlled circles, the circumnuclear regions

254 HUNT ET AL. Vol. 108
FIG. 8.ÈColor-color diagrams of individual pixels from selected color images. Pixels with distance Rfrom the nucleus of ¹4Aare shown by Ðlled circles,
4A\R¹3 kpc by open circles, and 3 kpc \R\11Aby points. Also shown are theoretical mixtures of galaxy colors and nonstellar emission processes
(power law, H II regions ; thermal dust emission at 600 K), an intermediate-age stellar population (A-type stars), and the e†ects of extinctionSlPl~1.5;
(external screen) and of varying strengths of a starburst component at a poststarburst age of 3 ]108yr & Heckman The mixing curves were(Leitherer 1995).
calculated by using a monochromatic approximation and the Ñux calibration given in et al. The emitting dust was assumed to be optically thinWilson (1972).
with emission coefficient Pl. The extinction was calculated by assuming an external screen and the extinction curve of et al. Colors of 0.0Cardelli (1989).
were assumed for the A stars. The NIR colors of H II regions (including free-free, free-bound, and two-photon emission) were taken from Willner, Becklin, &
Visvanathan see also Lunel, & Bergeat The tick marks along the mixing curves indicate fractional contributions to the K(1972; Thuan 1983 ; Sibille, 1974).
band in units of 0.2 relative to the galaxyÏs contribution. The ticks along the extinction line correspond to unit increments of the ticks along theAV;
aged-burst line correspond to order-of-magnitude increments (1, 10, 100) in the fractional contribution of the starburst component to the Vband relative to
the galaxyÏs contribution. The open circle at the lower end of the burst-only line indicates a poststarburst age of 107yr, and the curve ends at 3 ]108yr,
where it coincides with the aged-burst line dominated by the starburst.
(4A\R¹3 kpc, ““ inner disk ÏÏ) as open circles, and the
outer regions (3 kpc \R¹11A) as dots. Indeed, the infer-
ences made in the preceding paragraphs are borne out by
the color-color diagrams. In particular, the colors in one
type 1 Seyfert, NGC 7469, and in six SBNs (the Ðve with
nonaxisymmetric features noted above and Mrk 575, with a
bar) lie along the extinction vector.
The nature of the excess K-band emission can also be
investigated with the color-color diagrams. Roughly con-
stant J[Hcolors, together with red H[K, indicate hot
dust while bluer than normal J[Hand red H[Ksuggest
nebular emission, probably from H II regions. Of the six
galaxies that show evidence in the color images of excess
emission, Ðve show red H[Kextranuclear colors associ-
ated with normal stellar population J[H. We attribute the
excess K-band emission in these systems to hot dust. Only
in Mrk 912 (SBN) does the rather blue J[Hcolor, together
with red H[K, suggest that nebular emission causes the
2km excess.
Finally, the color-color diagrams enable us to discern

No. 1, 1997 HOST GALAXY DISKS. I. 255
FIG. 8.ÈContinued
colors associated with an intermediate-age population
arising from a past burst of star formation (shown in the
mixing curves as ““ Aged Burst ÏÏ and ““A stars ÏÏ). Five of the
seven type 2 Seyferts with JHK colors have the blue colors
that signify a conspicuous contribution from a ““ fossilÏÏ star-
burst. We note that the azimuthally averaged proÐles lend
support for bluer colors in the inner regions of Seyfert 2Ïs:
(1) a subset of these galaxies also show the circumnuclear
blue ““ dip ÏÏ noted in (2) the mean colors of Seyfert 2°4.2;
inner disks tend to be slightly bluer than those of Seyfert 1Ïs
and SBNs (sample mean of 0.67 vs. 0.70È0.71), although no
claim for signiÐcance was made, because of the 0.03È0.05
mag scatter; (3) the mean Seyfert 2 inner disk colors are
bluer than those of the outer disk (see an unusualFig. 3),
trend not observed in the Seyfert 1Ïs or the starbursts
studied here, nor in normal spirals (see, e.g., InFrogel 1985).
contrast with the Seyfert 2Ïs, only two type 1 Seyferts and
three SBNs show clear evidence in the color-color diagrams
of blue colors associated with an intermediate-age stellar
population.
5.DISCUSSION AND CONCLUSIONS
Our study is focused on the galaxies that host active
nuclei, rather than the active nuclei themselves; we were
particularly careful in the analysis to avoid contamination
by the nucleus and to obtain images and color images as
precise as possible even at low surface brightness.
Our main results are as follows. The average colors, both
of the outer and the inner disks, are fully consistent with
being independent of activity class and equal to those of
normal spiral galaxies. This is at variance with previous
studies, which found H[Kredder by B0.1 mag in the inner
regions of active galaxies. We have shown that most of this
discrepancy arises from the neglect of the redshift correc-
tion; when the correction is made, the previous data
become compatible with the present results.
The outer disk color gradients are consistent with those
of normal spirals. The color gradients in the outer disks,
and the color di†erences between inner and outer disk, do
not show any clear-cut trend. In J[H, we Ðnd marginal

256 HUNT ET AL. Vol. 108
FIG. 8.ÈContinued
evidence of red inward di†erences in Seyfert 1Ïs and star-
bursts, and of blue inward di†erences in Seyfert 2Ïs;
however, given the smallness of the e†ect and the uncertain
results about NIR color gradients in normal galaxies, a
pixel-by-pixel color analysis is required.
Two methods have been used to investigate the presence
of dust in these galaxies. We have looked for dust features in
absorption (J[Hredder than normal) and in emission
(H[Kstill redder than normal after dereddening), with the
constraint that reliable features should follow simple, con-
nected patterns. Such features are seen primarily in the
inner disks of starburst galaxies; in the other activity
classes, only NGC 7469 shows dust in absorption and in
emission, in accordance with its well-established circumnu-
clear starburst et al. et al.(Mazzarella 1994 ; Genzel 1995).
The implication of this is that, in the starbursts, gas associ-
ated with the dust that we observe has been swept in from
the outer disk, piling up in locations in the inner disk that
perhaps correspond to dynamical resonances. Such dust
(and gas) is not observed in the inner disks of typical Seyfert
galaxies.
We have also plotted small areas of each galaxyÏs disk on
a color-color diagram, in order to enhance the visibility of
localized, nonaxisymmetric color features that would be
smeared out in the averaging process. We Ðnd that the inner
disk blueing of Seyfert 2Ïs, alluded to above, is due to
patches spread along the ““ aged burst ÏÏ mixing line. Popu-
lation synthesis models (e.g., & HeckmanLeitherer 1995;
see indicate that the blue colors observed in theFig. 8)
Seyfert 2Ïs correspond to ages of roughly a few times 108yr
after a burst of star formation. Systematic di†erences
between Seyfert 1Ïs and Seyfert 2Ïs with respect to the star
formation activity have been reported previously, indicating
that Seyfert 2Ïs are forming stars and Seyfert 1Ïs are not (e.g.,
et al. Previous work also suggests that starMaiolino 1995).
formation in Seyfert 2Ïs is ““ fossil ÏÏ & Moorwood(Glass
& Mouri & Beni tez1985 ; Taniguchi 1992; Dultzin-Hacyan
et al. and age indicators such as that1994; Oliva 1995),
proposed by et al. should prove useful in thisOliva (1995)
context.

No. 1, 1997 HOST GALAXY DISKS. I. 257
In conclusion, on a scale of a few kiloparsecs from the
center, starburst galaxies show ongoing star formation, sig-
naled by the presence of hot dust; Seyfert 2Ïs exhibit
““ fossil ÏÏ star formation activity, a few hundred Myr old,
signaled by the presence of an old starburst component in
addition to a normal population; and, Ðnally, Seyfert 1Ïs do
not show any sign of activity, either ongoing or fossil,
exceeding that of normal spirals. It is to be stressed that
most of the evidence appears in the pixel-by-pixel analysis
and is strongly diluted or completely washed out by averag-
ing over large disk areas; furthermore, nothing can be
deduced from our data concerning possible star formation
activity in the immediate vicinity of the nucleus, where
superior angular resolution would be needed.
The emerging picture is that of a starburst-Seyfert evolu-
tionary connection, although not so global and not so tight
in time as previous studies have implied. The starburst
activity appears to involve restricted regions of the inner
disk in the pre-Seyfert phase; this spatially conÐned activity
is however capable of feeding matter to the nucleus on a
timescale comparable with the lifetime of the starburst, and
Seyfert activity begins. The reason the Seyfert 2Ïs appear to
precede in time the Seyfert 1Ïs is perhaps only statistical (see,
e.g., et al. a ““ youngÏÏ active nucleus shouldMaiolino 1995):
be surrounded by matter contributed recently by the star-
burst and injected into nonequatorial orbits; then the lines
of sight to the center should have a high probability of
being obscured. At later times the accretion Ñow should
settle into a thinner conÐguration, and obscuration-free
lines of sight should become more numerous.
Recent theoretical work seems to support this scheme.
& Shlosman have studied the evolution of theHeller (1994)
gas distribution in globally unstable galactic disks and
found that gas inevitably builds up within the inner D100
pc on timescales of Gyr. A circumnuclear/nuclear burst[1
of star formation and the creation of a supermassive object
in the galaxyÏs center are almost certainly the consequence
of such a central gas concentration. Their models further
suggest that the formation of a black hole would follow the
episode of star formation, which is what seems to emerge
from our observations.
While an evolutionary link between starburst and Seyfert
activity seems likely, we can only speculate about the
relationship between ““ active ÏÏ and quiescent galaxies.
Many lines of evidence indicate that episodes of star forma-
tion can cause the structural components of galaxies to
evolve over time. For example, disks can become more
compact by inward gas transport and subsequent star for-
mation (e.g., Interestingly, steep innerKormendy 1993).
disks have been noted to be associated with actively star-
forming galaxies Bulges may also be built(Boroson 1981).
up by similar evolutionary processes: as pointed out by
steep r1@4 brightness gradients associatedKormendy (1993),
with disklike dynamics; triaxial bulges in disklike barred
galaxies; and Population I material in ““ bulges ÏÏ all imply
that the distinction between bulges and disks is not as sharp
as was previously thought. Statistical evidence also leads to
the conclusion that galaxies initially without massive bulges
can form them as a consequence of galactic interaction and
the ensuing star formation episode Elmegreen,(Elmegreen,
& Bellin Evidently, if a galactic disk is subject to1990).
perturbations in its gravitational potential, subsequent
redistribution of the disk gas can give rise to bursts of star
formation that may alter the galaxyÏs physiognomy. In the
end, it is conceivable that starburst and Seyfert activity may
be transient, but inevitable, phases in the normal evolution
of galactic systems.
We would like to thank C. Aspin for scientiÐc and moral
support, the UKIRT technical sta† for cheerful competence
and timely assistance, and the SERC PATT for generous
allocations of telescope time. We are also grateful to R. de
Jong for providing proÐles and images of the galaxies in his
sample and to J. van Hardy for his insightful comments.
APPENDIX A
DETERMINATION OF SKY LEVELS, OUTER DISK COLORS, AND COLOR GRADIENTS
Analysis of cuts along the sides of some of the images shows that we have succeeded in Ñattening them to 0.1%È0.2% in J
and Hand 0.08%È0.09% in At this level, the sky in each frame is presumed to be an additive constant at every pixel. AsK.3
pointed out by & some intrinsic galactic properties can be measured even when the absolute levels ofSparks JÔrgensen (1993),
the sky brightness are not known in any of the Ðlters. This capability is most evident when the Ñux at a given wavelength at
each pixel is plotted against the corresponding Ñux for that pixel at another wavelength. The slope and curvature of this linear
Ñux versus linear Ñux plot are independent of the additive o†sets of the sky. Consider the Ñuxes and in two di†erent Ðlters:f1f2
f2\c0]c1f1]c2f1
2. (A1)
The Ðrst-order coefficient is the average color (or Ñux ratio) of the galaxy. The second-order coefficient measures ac1c2
dimensionless color gradient. Unfortunately, ordinary polynomials are not suitable for the parameterization of changes in
scale length with as described in the text. Instead, we must exploit an orthogonal set of polynomials:c2
n2\a0
2];
i/1
4aiTi(n1) (A2)
& where and are the scaled (to the interval [È1, 1]) Ñuxes and the are the Chebyshev(Sparks JÔrgensen 1993), n1n2Ti
polynomials.
To determine the mean disk color, we performed second-order linear regressions on the Ñux-Ñux data sets; each data set
consists of the proÐle extracted as deÐned in and there is one data set for each Ðlter. All Ðts were limited to galactocentric°3.2,
3These are in fact upper limits to the true image Ñatness, as, in most cases, faint galaxy emission is evident even on the edge of the array.

258 HUNT ET AL. Vol. 108
distances R[3 kpc. The Ðrst-order coefficient is determined by the least-squares Ðtting routine with a typical error of onlyc1
a few percent of its value. To determine the mean color gradient, following & we Ðtted up toSparks JÔrgensen (1993),
fourth-order Chebyshev polynomials to the two scaled Ñux-Ñux data sets (as before limiting the Ðts to those points with R[3
kpc). The curvature term in these Ðts, has a typical uncertainty of 0.01, that is to say, a 3% di†erence in scale lengtha2,
between the two wave bands.
While these sky-independent parameters are important in themselves, we also exploited them to determine the unknown
sky level of our images. In practice, after having determined the mean disk colors and color gradients, we then used these
parameters as the inputs to a second least-squares Ðtting procedure. For a given galaxy, we simultaneously Ðtted, in a
weighted least-squares sense, to the outer parts (R[3 kpc) of the J-, H-, and K-band surface brightness proÐles the sum of a
constant sky level and an exponential law All these proÐles were Ðtted before any sky subtraction.[I(r)\I(0) exp ([r/rd)].
The constant sky brightnesses in each wave band were three of the Ðve free parameters to be Ðtted; the remaining two were
the parameters that describe the exponential disk (central surface brightness and exponential folding length) in one band. The
brightness proÐles of the other two wavelengths were also forced to be exponential, but with two modiÐcations that were not
free Ðtting parameters. First, the average color in the Ðtting range of radius was Ðxed to be equal to the color we found in the
earlier polynomial Ðt described above. Second, the color gradient was Ðxed by the second-order coefficient, which determines
by how much the disk scale length at the longer wavelength di†ers from that in J.4
The simultaneous Ðtting of the three di†erent proÐles in a self-consistent way yields very robust results because the problem
is highly overconstrained. Thus formal on the Ðve free parameters are exceedingly small: typically 0.03%È0.1% in theerrors5
sky levels, 0.5% in the disk central surface brightness I(0), and 3% in the exponential scale length The sky levels thusrd.
determined are always lower than the median values along the edges of the image, and their errors are consistent with the
(upper limits of the) Ñatness we measure along the row-column cuts at the edges of the images. We emphasize that the Ðts
always started at an inner radius of 3 kpc, and went out to the last measured proÐle point (see Fig. 1).
APPENDIX B
EXTINCTION MODELS: FOREGROUND DUST SCREEN VERSUS UNIFORMLY MIXED DUST
AND STARS
Most analyses of extinction in galaxies have been based on the assumption that the dust resides in a foreground screen.
Such an assumption simpliÐes the calculation of the unknown optical depth qfrom reddened colors, as the relations are linear,
e.g.,
(H[K)obs \(H[K)stars ]1.086(qH[qK) . (B1)
As has been amply pointed out (e.g., & Fouque Davies, & Phillips aMathis 1970; Boroson 1981; Tully 1985 ; Disney, 1989),
more realistic conÐguration would have the dust interspersed with the stars, making a determination of qfrom reddening
more complex. In the simple case of a uniform slab in which the dust is distributed among the stars, the reddening equation
becomes
(H[K)obs \(H[K)stars [2.5 log CqK(1 [e~qH)
qH(1 [e~qK)D. (B2)
Given a certain reddening and assuming a standard extinction curve, we can compare the optical depth inferred for the two
above conÐgurations (eqs. Such a comparison is shown in and the onset of nonlinearity is clearly[B1], [B2]). Figure 9a,
evident for ““ slab ÏÏ (uniform mix of dust and stars) At slab a linear extrapolation of the slope (2.3) atqVB2. qVB5, qV\0
results in a discrepancy of 0.3 between slab and screen optical depths, or 15% of ““ screen ÏÏ Slab optical depths are alwaysqV.
greater than screen by roughly a factor of 2 for small (screen) and by more than a factor of 4 for (screen)qV,qVqVD4.
In we described the method we used for estimating excess (over the reddened stellar colors) 2 km emission. Assuming°4.3,
an intrinsic stellar color, we Ðrst derive the local optical depths from the J[Himage. These optical depths are then used to
deredden the H[Kimage, and any color excess over the assumed intrinsic stellar color is ascribed to excess K-band emission.
The results reported are those based on the assumption of a foreground dust screen instead of the more realistic mix of dust
and stars, and such an assumption may yield a spurious excess.
To estimate the inÑuence of such an e†ect, we can proceed as follows. Assume there is no K-band excess; then, for a given
we can compute the reddening in J[Hand in H[Kfor the case of the foreground screen. We can also derive theqV,qV
needed to produce the same J[Hreddening for the case of the uniform slab (e.g., But, in this case, the resultingFig. 9a).
H[Kreddening will be greater than that for the screen model. Hence, under the premise that the uniform slab model
corresponds to the ““ truth,ÏÏ the assumption of a foreground screen will result in a spurious H[Kcolor excess. Figure 9b
shows this spurious excess as a function of screen optical depth. It appears that the assumption of a foreground screen results
4For the two galaxies for which we have only Hand K, we Ðtted the disk to the Hband and imposed the average color and color gradient on the K-band
disk. For the single galaxy (Mrk 719) for which we have only an H-band image, we used the minimum curvature method, described in toHunt (1993),
determine the sky level.
5Assuming a s2increment equal not to 1 but, rather, to the 1 pconÐdence level for a Ðve-parameter Ðt (i.e., 5.9); see, e.g., Margon, & BowyerLampton,
(1976).

No. 1, 1997 HOST GALAXY DISKS. I. 259
FIG. 9.È““ Slab ÏÏ visual optical depth and spurious H[Kcolor excess vs. ““screen ÏÏ optical depth ““ Slab ÏÏ refers to a uniform slab with dust interspersedqV.
with stars, and ““ screenÏÏ to a foreground screen of dust. (a) The ordinate is the slab that would produce the same amount of J[Hreddening as the screenqV
in abscissa. The dotted line is the linear extrapolation of the slope (2.26) at zero optical depth. (b) The ordinate is the spurious H[Kcolor excess thatqV
results from the assumption of a foreground screen of optical depth in abscissa, instead of a uniform slab of dust and stars. *(H[K)(excess) \2.5 logqV
for small nonstellar emission fractions. The dotted line emphasizes the 1% amplitude of the spurious excess[1 ]&K(excess)/&K(stars)] B&
K
(excess)/&K(stars)
emission fraction at screen qVB2.
in an erroneous H[Kcolor excess of for screen Therefore we conclude that, as long as we derive or so,[0.01 qV[2. qV[2
the spurious K-band excess is on the order of 1% of the stellar emission; this is also the (1 p) limit and the accuracy to which
our method can estimate the K-band nonstellar contribution.
REFERENCES
A., Ward, M. J., & Kotilainen, J. K. 1996, MNRAS, 278,Alonso-Herrero,
902 V. A. 1983, ApJ, 268,Balzano, 602
V. A., & Weedman, D. W. 1981, ApJ, 243,Balzano, 756
J. E., & Hernquist, L. E. 1991, ApJ, 370,Barnes, L65
T. 1981, ApJS, 46,Boroson, 177
A. J., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345,Cardelli, 245
T. E. 1992, in ASP Conf. Proc. 31, Relationships between ActiveCarone,
Galactic Nuclei and Starburst Galaxies, ed. A. V. Filippenko (San Fran-
cisco : ASP), 375
M. M., & Hawarden, T. G. 1992, JCMT-UKIRT Newsl., No. 4,Casali, 33
L., Zitelli, V., Granato, G. L., Wade, R., De Zotti, G., & Mandol-Danese,
esi, N. 1992, ApJ, 399, 38
Jong, R. S. 1995, Ph.D. thesis, Univ.de Groningen
Jong, R. S., & van der Kruit, P. C. 1994, A&AS, 106,de 451
Vaucouleurs, G. H., de Vaucouleurs, A., Corwin, H. G., Buta, R.,dePaturel, G., & Fouque , P. 1991, Third Reference Catalogue of Bright
Galaxies (New York : Springer)
M., Davies, J., & Phillips, S. 1989, MNRAS, 239,Disney, 939
D., & Beni tez, E. 1994, A&A, 291,Dultzin-Hacyan, 720
R. A. 1987, ApJ, 313,Edelson, 651
J. H., Frogel, J. A., Matthews, K., & Neugebauer, G. 1982, AJ, 87,Elias,
1029 B. G. 1987, ApJ, 312,Elmegreen, 626
D. M., Elmegreen, B. G., & Bellin, A. D. 1990, ApJ, 364,Elmegreen, 415
F. R., Weedman, D. W., Balzano, V. A., & Ramsey, L. W. 1982,Feldman,
ApJ, 256, 427
D. A., Ward, M. J., DePoy, D. L., Boisson, C., & Smith, M. S. 1992,Forbes,
MNRAS, 254, 509
K. C. 1970, ApJ, 160,Freeman, 811
J. A. 1985, ApJ, 298,Frogel, 528
J. A., Persson, S. E., Aaronson, M., & Matthews, K. 1978, ApJ, 220,Frogel,
75 M. J., & Huchra, J. P. 1983, ApJS, 52,Geller, 61
R., Weitzel, L., Tacconi-Garman, L. E., Blietz, M., Cameron, M.,Genzel,
Krabbe, A., Lutz, D., & Sternberg, A. 1995, ApJ, 444, 129
C., & Hunt, L. K. 1988, AJ, 95,Giovanardi, 408
1996, AJ, 111,ÈÈÈ. 1086
I. S. 1984, MNRAS, 211,Glass, 461
I. S., & Moorwood, A. F. M. 1985, MNRAS, 214,Glass, 429
D., Hyland, A. R., & Jones, T. J. 1982, AJ, 87,Griersmith, 1106
T. M., Blitz, L., Wilson, A. S., Armus, L., & Miley, G. K. 1989,Heckman,
ApJ, 342, 735
C. H., & Shlosman, I. 1994, ApJ, 424,Heller, 84
L. 1989, Nature, 340,Hernquist, 687
J., & Burg, R. 1992, ApJ, 393,Huchra, 90
L. K. 1993, Laurea thesis, Univ.Hunt, Firenze
L. K., & Giovanardi, C. 1992, AJ, 104,Hunt, 1018
L. K., Zhekov, S., Salvati, M., Mannucci, F., & Stanga, R. M. 1994,Hunt,
A&A, 292, 67
R. D., Meikle, W. P. S., Robertson, N. A., & Wright, G. S. 1984,Joseph,
MNRAS, 209, 111
R. C., Jr. 1989, ApJ, 344,Kennicutt, 685
K., Watanabe, M., & Okamura, S. 1986, ApJS, 62,Kodaira, 703
J. 1993, in IAU Symp. 153 , Galactic Bulges, ed. H. DejongheKormendy,
& H. J. Habing (Dordrecht: Kluwer), 209
J. K., & Ward, M. J. 1994, MNRAS, 266,Kotilainen, 953
J. K., Ward, M. J., Boisson, C., DePoy, D. L., Bryant, L. R., &Kotilainen,
Smith, M. G. 1992a, MNRAS, 256, 125
J. K., Ward, M. J., Boisson, C., DePoy, D. L., & Smith, M. G.Kotilainen,
1992b, MNRAS, 256, 149
M., Margon, B., & Bowyer, S. 1976, ApJ, 208,Lampton, 177
R. B. 1985, MNRAS, 214,Larson, 379
M. J., & Rieke, G. H. 1980, Nature, 284,Lebofsky, 410
C., & Heckman, T. M. 1995, ApJS, 96,Leitherer, 9
J. W. 1990, ApJS, 72,MacKenty, 231
R., Ruiz, M., Rieke, G. H., & Keller, L. D. 1995, ApJ, 446,Maiolino, 561
M. A. 1983, ApJ, 268,Malkan, 582
J. S. 1970, ApJ, 159,Mathis, 263
J. M., & Balzano, V. 1986, ApJS, 62,Mazzarella, 751
J. M., Voit, G. M., Soifer, B. T., Matthews, K., Graham, J.,Mazzarella,
Armus, L., & Shupe, D. 1994, AJ, 107, 1274
M., Ma rquez, I., & Pe rez, E. 1995, ApJ, 438,Moles, 604
E., Origlia, L., Kotilainen, J. K., & Moorwood, A. F. M. 1995, A&A,Oliva,
301, 550
S. E., Frogel, J. A., & Aaronson, M. 1979, ApJS, 39,Persson, 61
M. 1977, in Ann. NY Acad. Sci., 302, Eighth Texas Symposium onRees,
Relativistic Astrophysics, ed. M. D. Papagiannis, 613
H.-W., & Rieke, M. J. 1993, ApJ, 418,Rix, 123
J. M., Rudy, R. J., & Jones, B. 1987, ApJ, 312,Rodriguez-Espinosa, 555

260 HUNT ET AL.
N. Z., Soifer, B. T., Neugebauer, G., Matthews, K., Young, J. S., &Scoville,
Yerka, J. 1985, ApJ, 289, 129
I., Begelman, M., & Frank, J. 1989, Nature, 338,Shlosman, 45
F., Lunel, M., & Bergeat, J. 1974, Ap&SS, 30,Sibille, 173
S. M., Su, H. J., & Schwarz, M. P. 1980, ApJ, 237,Simkin, 404
W. B. 1988, AJ, 95,Sparks, 1569
W. B., & I. 1993, AJ, 105,Sparks, JÔrgensen, 1753
Y., & Mouri, H. 1992, in ASP Conf. Proc. 31, RelationshipsTaniguchi,
between Active Galactic Nuclei and Starburst Galaxies, ed. A. V. Fili-
ppenko (San Francisco: ASP), 365
C. M., Campins, H., Joy, M., Dietz, K., & Decher, R. 1991, ApJ,Telesco,
369, 135R., Tenorio-Tagle, G., Franco, J., & Melnick, J. 1992, MNRAS,Terlevich,
255, 713
D. M., Davies, R. L., Frogel, J. A., DePoy, D. L., & Wells, L. A.Terndrup,
1994, ApJ, 432, 518
H. A., Jr., et al. 1989, ApJ, 343,Thronson, 158
T. X. 1983, ApJ, 268,Thuan, 667
R. B., & Fouque , P. 1985, ApJS, 58,Tully, 67
S. P., Becklin, E. E., & Visvanathan, N. 1972, ApJ, 175,Willner, 609
A. S. 1988, A&A, 206,Wilson, 41
W. J., Schwarz, P. R., Neugebauer, G., Harvey, P. M., & Becklin,Wilson,
E. E. 1972, ApJ, 177, 523
V., Granato, G. L., Mandolesi, N., Wade, R., & Danese, L. 1993,Zitelli,
ApJS, 84, 185

FIG. 2.ÈJ-band and J[Kimages of representative SBNs: Mrk 496 (top), Mrk 732 (middle), and Mrk 545 (bottom). The color table shows brighter and
redder regions as white and fainter and bluer regions as green. The white features in the outer parts of the color images are due to noise.
HUNT et al. (see 108, 237)
PLATE 1