Deep Intermediate-Band Surface Photometry of NGC 5907

Deep Intermediate-Band Surface Photometry of NGC 5907

Zhongyuan Zheng 1 ' 3 , Zhaohui Shang 1,2 ' 3 , Hongjun Su 1 , David Burstein 5 , Jianshcng Chen 1 ' 3 , Zugan
Deng 6 ' 3 , Yong-Ik Byun 4,9 , Rui Chen 1,3 , Wen-ping Chen 4 , Licai Deng 1 ' 3 , Xiaohui Fan 1,7 , Li Zhi Fang 10 Jeff
J. Hester 5 , Zhaoji Jiang 1,3 , Yong Li 5 , Weipeng Lin 1,3 , Wei-hsin Sun 4 , Wean-shun Tsay 4 , Rogier A.
OV Windhorst 5 , Hong Wu 1 ' 3 , Xiaoyang Xia 8 ' 3 , Wen Xu 1 ' 5 , Suijian Xue 1,3 , Haojing Yan 1 ' 5 , Zheng Zheng 1,3 , Xu

Zhou 1,3 , Jin Zhu 1 ' 3 and Zhenglong Zou 1,3

On
On

r ^P ' Received ; accepted

ON

>

t>

(N
(N

O

o>

ON

Or

6

t

1 Bcijing Astronomical Observatory, Chinese Academy of Sciences, Beijing 100080, P. R. China
2 Department of Astronomy, University of Texas at Austin, Austin, TX 78712

3 Chinese Academy of Sciences-Peking University joint Beijing Astrophysics Center, Beijing, 100871, P.
R. China

4 Institute of Astronomy, National Central University, Chung-Li, Taiwan

5 Department of Physics and Astronomy, Box 871504, Arizona State University, Tempe, AZ 85287-1504
6 Graduate School, Chinese Academy of Sciences, Beijing 100080, P. R. China
7 Princeton University Observatory, Princeton, New Jersey, 08544
8 Department of Physics, Tianjin Normal University, P. R. China

9 Center for Space Astrophysics and Department of Astronomy, Yonsei University, Seoul, 120-749, Korea
10 Department of Physics, University of Arizona, Tucson, Arizona, 85721

ABSTRACT

Intrigued by the initial report of an extended lumiosity distribution perpendicular to the
disk of the edge-on Sc galaxy NGC 5907, we have obtained very deep exposures of this galaxy
with a Schmidt telescope, large-format CCD, and intermediate-band filters centered at 6660A
and 8020A. These two filters, part of a 15-filter set, are custom-designed to avoid the brightest
(and most variable) night sky lines. As a result, our images are able to go deeper, with lower
sky noise than those taken with broad-band filters at similar effective wavelengths: e.g., 0.6
c - arcsec -2 sec -1 for our observations vs. 7.4 e - arcsec -2 sec -1 for the R-band measures of
Morrison et al. In our assessment of both random and systematic errors, we show that the flux
level where the errors of observation reach 1 mag arcsec -2 are 29.00 mag arcsec -2 in the 6660A
image (corresponding to 28.7 in R-band) and 27.4 mag arcsec -2 in the 8020A image (essentially
on the I-band system).

As detailed in Shang et al., our observations show NGC 5907 has a luminous ring around it
that most plausibly is due to the tidal disruption of a dwarf spheroidal galaxy by the much more
massive spiral. Here we show that, fainter than 27th R mag arcsec -2 , the surface brightness
around NGC 5907 is strongly asymmetric, being mostly brighter on NW (ring) side of the galaxy
midplanc. This asymmetry rules out a halo for the origin of the faint surface brightness we see.
We find this asymmetry is likely an artifact owing to a combination of ring light and residual
surface brightness at faint levels from stars that our star-masking procedure cannot completely
eliminate. The possible existence of an optical face-on warp in NGC 5907, suggested by our
VLA HI observations, is too confused with foreground star contamination to be independently
studied.

Good agreement with the surface photometry of NGC 5907 by Morrison et al. and other
workers lead us to conclude that their data are similarly affected at faint levels by ring light
and residual effects from their star masking procedures. Inspection of the images published by
Morrison et al. and Sackett et al. confirm this to be the case. Thus, we conclude that NGC 5907
does not have a faint, extended halo.

Subject headings: galaxies: individual (NGC 5907) - galaxies: interactions - galaxies: spiral -
galaxies: photometry

INTRODUCTION

Until Sackett et al. (1994) and Morrison, Boroson & Harding (1994; hereafter MBH) announced
discovery of a faint, luminous light distribution around NGC 5907, this galaxy was primarily known as a
typical, edge-on Sc galaxy which happens to be near enough to us (Vhciio = 667 km s _1 = 11 Mpc for
Ho = 65 km s _1 Mpc -1 ) that we can study it in some detail. Previous observations by several groups (e.g.

Sancisi 1976J |van der Kruit fc Searle 198lfc |Skurtskie et al. 1985| ; |Sasaki 1987J ; |Barnaby fc Thronson 1992J )
had shown this galaxy to have both an HI and optical warp, but no obvious galaxy companions to produce
this warp. As such, NGC 5907 became the prototype of the class of spiral galaxies having "non-interacting
warps" (cf. |Sancisi 197t| ).

The observations of Sackett et al. (1994) sparked new interest in this galaxy, in that they claimed
to detect a significant halo around it. Two other groups then obtained deep surface photometry to try to
study the halo in various passbands ( Lcqucux et al. 1996] , 1998; Rudy et al. 1997). Our collaboration
(the Beijing- Arizona- Taipei-Connecticut (BATC) Color Survey of the Sky ( pan et al. 1996 ) already had
NGC 5907 as part of our galaxy calibration program. The BATC program uses the 0.6/0. 9m Schmidt
telescope at the Xinglong Station of Beijing Astronomical Observatory (the "BAO Schmidt"), with its focal
plane equipped with a 2048 x 2048 Ford CCD. We have custom-designed a set of 15 intermediate-band
filters to do spectrophotometry for pre-selected 1 deg 2 regions of the northern sky with this CCD system

(cf. |Fan et al. 1996[)

Shang et al. (1998) summarize the main results of both our deep surface photometry and our Very
Large Arrayp] (VLA) 21 cm HI observations of NGC 5907: a) We detect a faint luminous ring around the
galaxy, plausibly due to the recent tidal disruption of a nearby dwarf spheroidal galaxy, b) The HI map
picked up a companion dwarf irregular galaxy, PGC 54419, on the side of one of the Hl-warps and separated
in radial velocity by only 45 km s -1 from NGC 5907. As such, we move NGC 5907 out of the category
of "non-interacting" warped galaxies, but whether the observed interactions are sufficient to produce the
observed warp is left for galaxy modelers to decide, c) Our HI observations suggest that the HI layer is
higher near the center of the galaxy than at larger radii, suggesting this galaxy may also have a face-on
warping of the disk.

11 The Very Large Array of the National Radio Astronomy Observatory (NRAO) is operated by Associated
Universities, Inc., under cooperative agreement with the National Science Foundation.

In this paper we present the details of our deep surface photometry of this galaxy, this time paying
close attention to the issue of the faint luminosity distribution around this galaxy. Section 2 presents our
observations, including details of the data reduction process, which are important for the reader to be able
to critically assess the accuracy of our method. In Section 3 we study the faint luminosity distribution
around NGC 5907 as it appears in our images, including how the ring, foreground stars, and a possible
face-on warp can influence what we see. Our results are compared to those previously published in Section
4, in which we also reassess the likelihood of a halo existing around this galaxy. Section 5 summarizes the
main results of this paper.

2. OBSERVATIONS AND DATA REDUCTION

The phases of data reduction that contribute to errors in the final image and over which we have
some control are: bias subtraction, dark subtraction, flatfielding, sky background-fitting, star-masking, and
photometric calibration. In this section we discuss these data reduction issues as they apply to the data we
have obtained. When we compare our surface photometry to those of other observers (Section 4), we will
find that, with few exceptions, our data are in very good agreement with published results.

2.1. Data Log and Preselection of Good Images

The observations of NGC5907 were obtained with the BAO 0.6/0. 9m Schmidt telescope, using a thick
Ford 2048x2048 CCD having 15^m pixels at the f/3 prime focus. The field of view of this CCD is 58'x58'
and the scale is 1.71" per pixel. With this combination of telescope and CCD system, we can comfortably
observe large extended objects and also get a sufficient amount of sky in a single frame. A large pixel size
is better for finding low surface brightness features, as it minimizes the number of CCD pixels per surface
area.

The images were taken with slight shifts between exposures ( "dithered" ) so that cosmic rays and defects
on CCD could be removed during the combining process. We employ the Lick Observatory data-taking
system, which automatically subtracts the overscan of each image at readout time ("on-the-fly"), and
records the overscan in the last column. We then process each program image through a series of software
routines we term PIPELINE-I, which, among other operations, median-filters the original overscan and
adds back to the image the difference between the original subtracted overscan and this filtered overscan.

In this way, the signal-to-noise (S/N) of the overscan is increased and any residual large-scale pattern
produced in the image by imperfections in the overscan are removed.

Exposures of the field around NGC 5907 were made through two BATC filters: 6660A (m 6660 ) and
8020 A (m 8020 ), whose band widths are 480A and 260A respectively (cf. Figure 1 in |Fan ct al. 1996| ). The

BATC filters are designed to avoid contamination by the brightest and most variable night sky emission
lines, producing a sky background of lower surface brightness obtainable under the same conditions in the
broad band R and I filters. Use of intermediate bands also minimizes effective wavelength- related effects in
the CCD sensitivity (Section 4.4).

137 frames of either 10m or 20m duration were obtained with the 6660A filter on 23 nights from
January 31, 1995 to June 27, 1995, all during moonless intervals at airmasses < 1.5. The CCD was run with
a gain of 4.1 e~/ADU and a readout noise of 12 e~. Similarly, 63 frames of 20m duration were obtained
similarly with the 8020A filter on 11 nights from May 26, 1995 to April 18, 1996, during which time the
readout noise remained as before, but the gain varied somewhat. For the final combined 8020A image
the equivalent gain is 3.9 e~/ADU. Of these 200 images, we selected the images for final reduction in the
following manner (cf. Tables 1 and 2):

(1) Each individual frame was inspected visually to see if there are abnormal events that might affect
photometry, e.g., traces of satellites crossing galaxy area, or visible problems with the bias overscan. If
there was anything abnormal, the image was rejected.

(2) For each frame we measured the mean sky level, FWHM and flux for 20 isolated stars found in
common on all frames. These values are shown in Figure 1 (and tabulated in Table 1) as a function of
sequential image number for the 137 6660A images, with flux and sky level normalized to an integration
time of 20m. Those images having high stellar flux levels, low sky levels and reasonable seeing were selected
for further study. A similar analysis was done for the 8020 A images (Table 2), with the net result that
the final combined images are made of 84 6660A frames taken on 15 nights and 50 8020A frames taken
on 8 nights. The total exposure time in the combined images is 26 h 10 m (94200s) and 16 h 40 m (60000s)
respectively. The images which are used are listed in Tables 1 and 2 with their sky level counts, FWHM of
stars and the mean counts (for 20m integration) measured for the 20 comparison stars.

2.2. Bias and Dark Subtraction

Biases measure the ADU level of zero second exposures and, as such are of low S/N individually. We
took an average of 26 bias frames per night of observation using the IRAF task "imcombine" , choosing the
minmax algorithm in combination with the parameters "nlow" and "nhigh" set to 3 to reject bad points at
> 3cr level. As one test of CCD stability, we take 30 bias averages from a two month period during the
observations and divide them into two sets of 15 bias averages each: BIAS1 from the first 15 bias averages;
BIAS2 from the second 15. The difference, BIAS1-BIAS2 has a mean value of zero and no large-scale
pattern exists, indicating good stability of our data-taking system. The average of BIAS1 and BIAS2,
hereafter BIAS, is adopted as the final bias frame for correcting. It is equal to the average of 600 individual
bias frames, has a mean value of about 1.2 ADU and a standard error of about 1 ADU/pixel (cf. Sec. 2.7).

Many dark frames must also be obtained throughout the observation sequence to reliably measure
the dark current of the CCD. The dark count was obtained by obtaining forty, 20m dark frames over the
same two month period. Subtracting BIAS from these frames yields a very small dark count of ~ 0.25
ADU/pixcl/20m exposure. This summed dark count frame was then rescaled to the exposure time of, and
subtracted from, each program image.

2.3. Large-scale Homogeneity of the Program Image

The first requirement in obtaining homogeneity over large-scales in a CCD image is to produce a very
accurate Hatfield map of the large-scale variations that exist in CCDs, and for those large-scale variations
to be stable over periods of time long compared to the time over which the Hatfield images and program
images are obtained. This, in turn, requires controlling two factors: a) the DC stability of the CCD itself
over the image area of interest; and b) the flatness of night sky and twilight, which, of course, are never
really flat over > 10' angles (see below).

A third factor affecting how well one can subtract the sky from the image of a galaxy is the fact that
the quantum efficiency of a CCD pixel is a function of the effective wavelength of the radiation it detects.
The one part of the data-taking process we cannot control is that the galaxies have different spectral energy
distributions (SEDs) within the filters, and that these SEDs are different from that of the night sky.

While we cannot completely eliminate the systematic effects on the large scale homogeneity of any of
these issues, we have taken steps to minimize their influence:

a. DC— stability: We can comfortably fit a galaxy as large as NGC 5907 (14' in diameter) within the
inner 25% of the CCD, which itself is in the inner 20% of the field of a Schmidt telescope. This eliminates
issues pertaining to edge-effects in CCD flatfields and minimizes non-uniformity issues related to the optics
of the telescope.

b. Flatness of the Flatfields: On a CCD chip covering essentially l°xl° of the sky, neither

the twilight nor the night sky will be flat over the whole chip ( Wild 1997 ). This is obvious upon visual
inspection of the twilight sky, when one takes into account that the field of view of our CCD is twice the size
of the full Moon. It is also true on a cloudless, moonless night, as the airglow brightness correlates with air
mass. Our eyes can easily detect gradients in twilight the size of the full Moon, and at a dark site, gradients
in the sky over degree-sized scales. One might be able to devise a methodology of taking night sky flats at
various airmasses and sky directions to minimize such gradients in the night sky, but such techniques are
likely to be very time consuming, with no guarantee that sky gradients can be completely eliminated.

Rather, we take advantage of the special optical properties of a Schmidt telescope to construct a

reliable flatfielding method that can employ a dome light. As detailed in Chen ct al. (1999) , we place a
UV-transparent plastic diffuser over the correcting lens of the Schmidt. One dome light Hatfield frame
has more than 20000 ADU per pixel, corresponding to more than 84000 electrons (gain = 4.1). Twelve
dome flats produce an overall flat field whose statistical pixel-to-pixel error is less than 0.1%. The diffuser
ensures that the flux entering the Schmidt telescope is of a highly uniform nature. In the present paper, the
proof of the accuracy of this flatfielding technique is demonstrated in two ways: by comparing noise in the
division of average flatfields taken on successive nights to that expected from photon noise alone, and by
the accuracy to which we can flatten the sky.

c. Color-related Effects: One of the factors leading us to decide to use intermediate-band filters for
the BATC survey is the fact that SED-related quantum efficiency effects do exist in CCDs. Our own tests
with stars (cf. Fan ct al. 1996| ) show the measurements made in the two filters used here are insensitive to

color changes in the program objects at the 1% level. Such is not necessarily the case for observations made

with broad-band filters (Stetson 1990)

2.4. Rectification of the Summed Image

After the correction for the flatfield is applied, position offsets among the selected CCD frames were
calculated with six plate coefficients and the position of the frame center, using the Guide Star Catalog of

the Space Telescope Science Institute (Lasker et al. 1990). This process is done during the PIPELINE-I

reduction process, and the derived plate coefficients are put into the FITS image header. After shifting all
corrected images to a common center, bad pixels and cosmic-rays are rejected, and the cleaned images are
combined to a single frame. We then redetermine the plate coefficients and plate center for the combined
frames. The net images in the 6660A and 8020A filters contain 1928x1969 pixels and 1979x1979 pixels
respectively. Figure 2a shows the combined image in the 6660A filter, Figure 2b shows the combined image
in the 8020A filter.

2.5. Prediction of Background under the Galaxy and Ring

A reliable estimate of the surface brightness at faint levels near NGC 5907 requires us to be able to
reliably determine the sky background under the regions of interest. These include not only the regions
immediately adjacent to NGC 5907, but also for the ring which we found around it. This, in turn, means
modeling the background formed not only by the night sky, but also by extraneous sources of light, including
stars, other galaxies and, if present, Galactic "cirrus."

2.5.1. Star and Galaxy Masking

The stars and background galaxies to be subtracted on the combined image range from faint,
near-point-sources of light to large regions contaminated by highly saturated stars. The traditional way to
handle these sources of extra light is to mask them until their wings fall under a certain level, say, 10% of
the variation in the sky. This is the procedure followed by MBH and Lequeux et al. (1996). We found that
if we followed this procedure with our own combined images, we would be left with too few sky pixels from
which to determine a reliable sky background. As a result, we proceed in a different, stepwise manner.

First, using DAOPHOT ( Stetson 1987 ), we fit a point-spread-function (PSF) to each combined image.
The fitted PSF is then subtracted from star images that are not saturated, following the prescription given
in the 1987 Stetson paper and in the DAOPHOTII User's Manual. Saturated stars were not subtracted.
Key to this process is as accurate a subtraction of the wings of the PSF as one can practically make.

The next step is to check the reliability of the PSF subtraction by plotting the residuals (obs— PSF
model) versus distance from PSF center for each unsaturated star. The PSF-wings arc subtracted cleanly
for most stars and there is little remnant. However, the central areas of these subtracted stars show higher
statistical fluctuations than the wings, owing to the higher original signal in the centers. While circular
masks are placed over such cases, the size of the circles cover far less area than would have been the case if
the whole PSF-wings of every star was masked.

The brightest, unsaturated stars are the exception to this second step, as the PSF-subtraction leaves
a noticeable residual at faint light levels. Therefore, we treat both the brightest stars, saturated and
unsaturated, the same, masking out the whole star to faint, but finite sky levels. In total, 117 bright stars
in the 6660A image were masked in this way.

Other galaxies that are resolved on our images are easily found via the DAOPHOT star-subtraction
process, as the PSF-subtraction leaves a deficit of flux at the galaxy center surrounded by a bright halo. All
background galaxies are masked entirely, but are generally so faint that these mask radii are small.

Figure 3 shows four results of the PSF-subtraction, including a saturated star, a bright unsaturated
star fit by the PSF, a faint star fit by the PSF and for a faint galaxy fit by the PSF. The vertical lines
indicate masking radii used. Our tests on all PSF-subtracted stars show that the residuals of the PSF-fits
are less than 5 ADU on average, or approximately 5% of the statistical noise of the sky.

Treating saturated stars in these images is more problematic. The images of star which cannot be
PSF-fittcd due to saturation effects should, in principle, be wholly masked. Unfortunately, as stated earlier,
if we tried to completely mask every bright star in our image, we would not be left with many pixels left in
the image! As a result, we are forced to make two compromises in our star-masking procedure. First, we
only mask each bright star to a surface brightness of 29th mag/arcsec 2 . Second, owing to the optics of a
Schmidt telescope, the PSF's of the stars are not symmetric anywhere in focal plane of our telescope. This is
easily seen in Figures 2a and 2b of the present paper. However, the masks are, by necessity, symmetric. The
net result is a slight mismatch of mask and PSF at very low light levels. As we will see, these compromises
limit what we can do at faint light levels around NGC 5907.

In our 6660A combined image, there are 7921 unsaturated objects detected among which are 1694
background galaxies in 1000x1000 pixel (1710"xl710"). We also note that, if one compares our 6660A
combined image with Figure 2 of MBH and Figure 1 of Sackett et al., it is clear that our observation is
as deep, if not deeper, than theirs, detecting all of their stars, plus some fainter stars. Quantiatively, our

10

detection threshold is 3.5cr s k y , resulting in a limited magnitude of 23.3. This means that for mgego < 23.30,
log(N s tar) = 4.45 deg~ 2 and log(N ga i axy ) = 3.88 deg~ 2 . Star counts are overestimated, and galaxy counts
are underestimated, as galaxies that are unresolved appear stellar, and galaxies with too low surface
brightnesses are not found.

Separately, we mask out a circle of radius 250 pixels around the galaxy, as well as an area around the
ring that is otherwise not excluded. This ensures that no galaxy-related or ring-related flux are included
in the sky background determination, but it also means that the residual flux from bright stars masked in
these regions is not modeled. The final masked 6660A image is shown in Figure 4.

2.5.2. Sky Background Fitting

Once star-masking and galaxy-masking is complete, we can then attempt to model the sky background.
Since it only makes sense to do this calculation near the galaxy and ring, we can safely limit ourselves to
the inner 1000 x 1000 pixel areas (28.5' x 28.5') in each combined image, the region marked by the square
in Figure 2a. This is also the region defined by our masking procedure (Figure 4).

First, we produced a smoothed version of each masked combined image by mode filtering each image
with a box lOx 10 pixels in size. To avoid contamination by masked pixels, only those pixels not in a masked
area were used in calculating the mode in any box. This necessarily means that near each masked region,
fewer pixels were used to generate the mode. After smoothing the mask regions whose radii are larger than
box size are shortened and those whose radii arc smaller than box size disappear. Once this smoothing
operation is done, the original mask is reimposed on the image, thus eliminating the pixels which are in
original masked areas. As a result of this process, edge effects from the masking procedure are removed.

We next experimented with the IRAF task imsurfit to perform fitting the sky background with various
two-dimensional analytic functions. None of these models was found to be wholly satisfactory, owing to the
fact that we have to restrict the fits to low order polynomials to avoid introducing spurious interpolations in
the masked regions. In every case, we could see large areas of the image that were systematically underfit
or overfit. Low order spline fits similarly suffer. At the end, we settled on a stepwise method that works
reasonably well when faced with a large part of the image masked-out.

We fit each row of the smoothed-and-masked image with a one-dimensional Legendre polynomial of
low order. In doing this fit, we reject points above 2ct on the high side and 3cx on the low side, as well as

11

all pixels inside the mask. The reason for the asymmetric rejection of points is that the main sources of
scatter on the high side are faint, undetected sources and unfitted faint wings of stars, while the low side
values result from statistical fluctuations. This is clearly seen when you make histograms of the pixel values
in the masked images; the histograms are skewed towards positive values. Necessarily, this process requires
interpolating each line under the galaxy/ring mask, keeping the order of the Legendre polynomial to be 3
or less.

The row-fit to the smoothed 6660 A image is shown in Figure 5a. Next, we repeat this process in the
column direction (Figure 5b). This ensures we are predicting the sky underneath the galaxy in a mutually
orthogonal manner. The row and column fits are then averaged (Figure 5c), and this average image is
smoothed with a circular Gaussian of a = 30 pixels, truncated at ±4<r. The final smoothed image was
adopted as the sky background (Figure 5d).

The 6660A image with its fitted background subtracted is shown in Figure 6a and the 8020A image is
shown in Figure 6b. As is evident, even with our best attempt at star-subtraction and masking, one can
still see faint, partial rings near many of the brighter stars. These rings are at surface brightnesses 29th mag
arcsec -2 and fainter in the 6660A image. As stated earlier, this end result is a compromise, for if we were
to truly exclude all starlight from this image, very little would be left of the image from which to determine
sky, and in many of the MBH cuts, leaving little or net flux to examine. As is plain, NGC 5907 is at a
Galactic latitude (51°) where stellar crowding starts to seriously affect what one can quantitatively do at low
surface brightnesses. As a check of the accuracy of our background fitting and masking procedures, we have
taken four averages of 30 rows ( "slices" ) each through the background-subtracted 6660A image (Figure 7)
and similarly-processed 8020A image (Figure 8) that span the region of interest around NGC 5907.

At the tops of each panel of Figures 7 and 8 are given these sky-subtracted slices through the galaxy.
At the bottom of each panel are given the residuals of the subtracted sky background for each slice. The
upper line is the residual of our sky fitting and subtracting procedure, while the lower line is the residual
of normal two-dimensional function fitting. They are shifted down by 110 ADU and 150 ADU seperately
for display purposes. It is obvious that our procedure produces smaller residuals than two-dimensional
function fitting. From Figures 7 and 8 a reader of this paper can check as to whether our sky subtraction
procedure introduces any spurious features into the sky-subtracted image. The lower parts of these panels
also demonstrate that our sky-subtraction procedure conservatively fits an effective two-dimensional plane
under the masked region around the galaxy.

12

Of course, no sky background fit is perfect, and it is these imperfections that contribute to the limit
to which one can do accurate faint surface photometry. To estimate the systematic error from this source,
we sample 381 areas of 50x50 pixels of the background-subtracted 6660 A image and calculate the mode
value of each area. After excluding areas in the region severely affected by masking, the standard error of
these mode values is 18 ADU. We take this as our best estimate of the irreducible la accuracy of our sky
background fit. In a similar manner we determined the accuracy of the sky background fit to be 17 ADU
for the 8020A image.

In contrast, if we had limited our sky background fits to only those obtained by analytical 2-dimensional
functions, the row slices showed many more systematic variations and the la systematic error was typically
30 ADU. We believe our method of sky background fitting produces a reliable match to what is, inherently,
a very lumpy sky at very low light levels.

2.6. Zero Point Calibration

It is standard practice for BATC observations to use Oke-Gunn (1983) standard stars as our calibration
stars when the night is photometric. The nights of 06March95 and 01April96 in the series taken for
NGC 5907 were photometric. The magnitude zero point for each combined image corresponds to a flux of
1 ADU/sccond is: 19.08 ± 0.01 mag for the 6660A image; 17.79 ± 0.01 mag for the 8020A image. The sky
brightness values are 21.26 mag arcsec -2 and 19.91 mag arcsec -2 , respectively. The parameters we derive
for these two combined images are given in Table 3. Note that the FWHM of the PSF of the combined
images (2.3 and 2.5 pixels, or 2" in radius), is worse than typical seeing of 2" at the Xinglong Observation
Station of the Beijing Astronomical Observatory, using the Schmidt telescope (cf. Figure 1).

2.7. The Estimated Errors we Can Control

The random error per pixel is given by the random error in photoelectrons, not in ADU. In the case
of the 6660A image, there are 4.1e - /ADU, and 3.9e - /ADU for the 8020A image. Hence, if a pixel has
n ADU, this corresponds to 4. In e - for the 6660A image, and the error is y/4.1n in e - , equivalent to an

error of y/n/4.1 in ADU. In the case of the 8020A image, the error would be ^/n/3.9. The sources of error
in our images we can estimate are: readout noise; photon noise per pixel; noise in average bias; noise in
average dark; noise in flatfielding; systematic errors in the sky background mapping. Note that all of these

13

noise levels save two (systematic flatficlding errors and systematic sky background errors) can be reduced
by increasing the number of pixels sampled.

Readout noise: Random noise from readout comes from a readout noise (in terms of error of the
mean) of 12 e _ /pixel per image. In the case of the 6660A combined image (84 input images) this is
12xV84 «110 e _ /pixel; 12xV50 «85 c~/pixel for the 8020A image (50 input images). From above, the
noise in the readout is 26.8 ADU/pixel for the 6660A image and 21.8 ADU/pixel for the 8020A image.

Photon noise per pixel is given simply as ^/n/4.1 ADU for the 6660A image and -^/n/3.9 for the
8020A image.

Bias and Dark: Because the ADU/pixel counts of the average bias and dark are very small, the main
error comes from the readout noise, and is 3 ADU/pixel. The final bias image used for bias subtraction was
an average of 600 single bias frames, resulting in a noise of 0.12 ADU/pixel. The error transmitted into the
combined image in 6660A is 0.12x\/84 ~1 ADU/pixel. The error introduced by dark-subtraction is also
about 1 ADU/pixel for both passbands. Compared to the photon noise of night sky and readout noise of
the program images, the errors introduced by biases and darks arc thus negligible.

Flatflelding As is evident from Figures 7 and 8, our flatficlding procedure produces a flat enough
image that the structure in the background is dominated by the intrinsic lumpiness of the faint night sky.
This is consistent with the tests we have done to confirm the validity of our flatficlding procedure (Chen et
al. 1999). However, from the flatfield we do have some pixel-to-pixel noise due to photon statistics. For
each filter, 12 frames of dome flatfields, each with more than 20000 ADU, are combined together to make
the final flatfield image. Thus, the photon noise in the flatfield is ^20000/4. 1/VT2 ~ 20 ADU or 0.1%.
Since the final galaxy image was combined from 84 dithered frames, the error was reduced by 1/V84 to
0.01%, approximately. For the 8020A image, it is about 0.015%.

Given the changes in flatfield pattern night-to-night, run-to-run in a typical CCD system, we separately
check the large-scale uniformities of our flatfields by dividing the combined flat-fielding frame of one night
by that of the adjacent night. 9 divisions of flatfields taken in the 6660A band have an average pixel-to-pixel
variation of 0.16%, indicating that the noise of the combined flat-fielding frame of one night is about 0.11%.
This value is consistent with 0.1% predicted by calculating photon noise, showing our flatfields are very
stable from night-to-night. As such divisions also included any large-scale pattern changes in the flatfield,
this test also shows that the large-scale pattern is reproduced very accurately with the dome/diffuscr
technique for a Schmidt telescope. When we then smooth the flatfield ratios by binning up pixels, we find

14

that the large-scale variation is about 0.026%. Therefore, the large-scale error of the combined flat-fielding
frame of one night is slightly smaller than 0.02%.

Intrinsic variation in galaxy brightness: MBH point out that the intrinsic variation could be

expressed by the formula given by Tonry & Schneider (1988), assuming M^ = (Tonry ct al. 1990). In

our case, mi = 19.08 mag, an exposure time of 94200s, a distance of 11 Mpc (assuming Ho=75). Since the
central wavelength of our 6660A band is similar to R band and the difference between our 6660A band and
R band is small (Sec 4.1), a value of M6660 = should be a reasonable assumption. This gives an estimated
variance of 3.3g, where g is the mean number of counts due to the galaxy only. So 45 ADU/pixel (equal to
28.5 mag/arcsec 2 ) will give an error of 12 ADU/pixel, far less than the photon noise of night sky of 95.1
ADU/pixel. The estimated variance of that in 8020A image is 1.6g (assuming Mso20 ~ — 1, when Mr ~

and M/ ~ —2 (Tonry et al. 199C )). For 22 ADU/pixel (equal to 27.5 mag/arcsec 2 ) we get an error of 6

ADU/pixel, small compared to our other sources of random error.

2.8. The Total Error Budget

Random Noise: Of the photon noise of night sky, readout noise, bias/dark count errors and
flatficlding random errors, the reducible random errors in a blank sky area are dominated by the first, and
are estimated to be close to 100 ADU/pixel for the 6660A image and 83 ADU/pixel for the 8020A image,
which are roughly consistent with the a measured directly from the combined images (cf. Table 3). These
noises and intrinsic variation of the galaxy can be reduced by binning adjacent pixels. As the distance from
the galaxy plane increases, the bin size increases exponentially to keep similar S/N ratio. As an example,
the bin size is 45x59 pixels at the farthest distance, and the random noise is reduced to only 2 ADU.

Large-scale Error of Flatfielding: Both from the tests we made above, and those made by Chen
et al. (1999), the large-scale error of BATC flatfielding is less than 0.1% (Chen et al. 1999). Since we
combined 84 frames for 6660A band and 50 frames for 8020A band, the error introduced by this source is
reduced to about 0.01%.

Systematic Error of Background Subtraction: In the previous section we found that the
irreducible scatter in the images is 18 ADU for the combined 6660A image and 17 ADU in the combined
8020A image. Taken together with the sources of random errors, this places an intrinsic accuracy of 20
ADU/pixel for faintest signal in the 6660A image and 19 ADU/pixel in the 8020A image where we can

15

explicitly fit the sky. Expressed in terms of surface brightness units, these errors correspond to a surface
brightness of 29.4 mag arcsec -2 in the 6660A image, and 27.7 mag arcsec" 2 in the 8020A image. In masked
regions, this intrinsic error will be larger and more systematic, owing to residual surface brightness around
the masks of bright stars.

Residual Foreground Star Subtraction: The slight imperfections of the star masking process
can be estimated from the slices in Figures 7 and 8. By examination of the regions around the brighter
subtracted stars, it is apparent that residual star light affects our image at the level of 20-50 ADU/pixcl
in specific places around most of the brighter stars. From above, this is at surface brightnesses around the
29th mag arcsec -2 level in the 6660A combined image. It is also evident from Figures 7 and 8 that if we
tried to mask the stars to fainter light levels we would, indeed, be left with very few remaining pixels in
these cuts. The complications in our analysis of these data that this compromise brings are discussed below.

The sources of variation in our surface photometry in 6660A that are generally present in our 6660A
combined image are summarized in Table 4. This example assumes we are observing a surface brightness
that gives us 50 ADU counts from the galaxy, 37066 from the sky and bin size of 45x45 pixels. We choose
this bin because it lies at the envelope of the galaxy halo. It can be found that a combination of irreducible
systematic error of determining sky background, plus unavoidable imperfect star-masking, determine the
ultimate limit of the accuracy that can be achieved.

In reality, to this value (24 ADU) we must add the more problematic residuals that we see around
bright stars within the masked-out region around the galaxy and ring. At its lowest level (~ 20 ADU), the
error is comparable to our minimum error; at its highest (~ 50 ADU), it dominates the minimum error. As
we will see, residual surface brightness from bright stars likely poses the ultimate limit to which we can
investigate the halo of NGC 5907.

3. SURFACE BRIGHTNESS DISTRIBUTION OF, AND NEAR, NGC 5907

3.1. Color and Magnitude of NGC 5907

We measure the total magnitude of NGC 5907 to a surface brightness of 27 mag arcsec -2 of
m 6660 = 9.9 ± 0.02 and m§020 = 9.5 ± 0.02. To obtain these values we replaced the areas excluded due
to foreground stars with the mean value of the surrounding background. To put this into more standard
terms, we use the fact that we find an average transformation of R — I = ni6660 — niso20 — 0.2 mag, based

16

on comparison with existing surface photometry of NGC 5907 (Sec. 4.2). Thus, our observations indicate
NGC 5907 itself to have R — I = 0.2 ± 0.04. If we exclude a rectangle around the obvious dust lane in this
galaxy, we find the average R-I color to be —0.1 ±0.1. This bluer color refers to the average value for stars
less-affected by dust in the disk, and is typical of a stellar population dominated by younger stars.

3.2. The Edge-on Stellar Warp

The HI warp of NGC5907 first found by Sancisi (1976) is strongly confirmed by our own VLA 21 cm
HI observations (Shang et al. 1998). Given the warp in the HI gas, it is logical to search for an analogous

warp in the stars. While van der Kruit (1979) only had marginal evidence for such a stellar warp, Sasaki

(1987) claimed to detect an optical warp at a projected radial distance along the major axis from 13.3 kpc

to 24.0 kpc from the center of the NGC 5907. Sancisi (1993) showed that there are similarities between the

optical warp of Sasaki and the HI warp. In Figure 7 of MBH, there is evidence for the start of a stellar
warp at 4.1 kpc from the center of the galaxy.

The warp in each of our combined images is determined by measuring the horizontal distances between
the surface brightness profiles of two sides of the galaxy around a fiducial position angle, and measuring
their mean offset as a function of distance along the galaxy major axis. Figure 9 shows the optical stellar
warps calculated from our brightness profiles parallel to the minor axis. We find a warp in the galaxy
that begins near the center of the galaxy (at 4.1 kpc, in agreement with what MBH found) and changes
continuously outward to at least 16 kpc from the center, such that at a radius of 16 kpc, the warp deviates
about 0.4 kpc (~ 4 pixels) above the average galaxy plane. We see the same warp in both our 6660A image
and in our 8020A image.

In regards to investigations at high latitudes for faint surface brightness in an edge-on galaxy, warps
are likely to occur with greater frequency along the line-of-sight to the disk than at the disk edges, given
relative angular coverage. Given that a galaxy disk is seen to have a warp at its edges, it increases the
possibility that the galaxy is also warped along the line-of-sight. This face-on warp may distort galaxy
surface brightnesses at large z-distances from the galaxy plane in a non-axisymmetric way, as opposed to
the expected symmetry from a halo. As such, separating the two kinds of possible sources of high-z surface
brightness is, in principle, possible. Relevant to this issue in the case of NGC 5907, we note that there is
some indication in our current VLA observations, that the HI layer is higher near the center of the galaxy
than outside the center (Shang et al. 1998). We explore the consequences of this hypothesis below.

17

3.3. The Stellar Ring

As shown in Shang et al. (1998) and has been evident in the deep images we present here (Figure 6a, b),

NGC 5907 has a faint, luminous ring around it. Our schematic for this ring is shown in the inset in

Figure 6a. As discussed in Bhang et al. (1998), the ring is reasonably elliptical in shape, of similar angular
size as NGC 5907, and the center of NGC 5907 is near one focus of the ellipse. The reality of this ring
has been confirmed on other CCD images (cf. Shang et al. 1998), and part of the ring can be faintly seen
in Figure 3 of MBH (although they do not recognize it as part of a ring) . The various arguments for the

origin of this ring are made in shang et al. (1998), from which we conclude that by far the most likely

interpretation is that of a dwarf spheroidal galaxy in the process of being torn apart by a strong tidal
encounter with NGC 5907.

The physical particulars of the ring indicate that the luminosity distribution within it is irregular both
in surface brightness and in apparent thickness. The maximum width of the ring is found at the NE end of
its major axis (40 pixels = 68.4" = 3.6 kpc at the distance of NGC 5907), while its smallest width is found
near the region where the ring overlaps NGC 5907 (20 pixels = 34.2" = 1.8 kpc). In our estimate of the
total brightness of the ring, we are limited by masked areas within NGC 5907 itself obscuring that part of
the ring nearest the galaxy. As such, we can only unambiguously measure unmasked pixels of the ring only
in its clearer NE half. We find the average surface brightness in the NE half of the ring to be 28.0 ± 0.3 mag
arcsec -2 in the 6660A image and 28.3 ± 1.8 mag arcsec -2 in the 8020A image. The much larger error for
the ring surface brightness at 8020A is expected, given our error budget estimate (Sec. 2.8). If we assume
that the whole ring has this average surface brightness, we obtain total magnitudes of mg66o = 14.7 ± 0.3
mag and mso2o = 15.0 ± 1.8 for the ring. However, these estimates may be brighter/fainter limits to the
brightness of the ring, as it is not obvious how irregular is the light distribution in the regions we cannot
investigate due to overlap with the galaxy and with foreground stars.

The larger error on mso20 precludes quoting a reasonable color for the ring as a whole. If we
measure just the highest surface brightness features in the ring (denoted by arrows in Figures 8 and 9),
we find 26.8±0.1 mag arcsec -2 in 6660A and 26.1±0.2 mag arcsec -2 in 8020A, yielding a color index
m 6660 — ni8020 — 0.7 i 0.3 (or R — I = 0.5 ± 0.3). In order to confirm this result, we also measure the total
flux within the brightest area of the ring by (indicated by the parallelogram in Figure 6a) to give an average
surface brightness of 27.0±0.1 mag arcsec -2 in 6660A and 26.4±0.3 mag arcsec -2 in 8020A, implying an
R-I color of 0.4±0.4. Measurement of the second brightest area of the ring also gives a similar result. On

18

average, we measure the ring to have an R-I color of 0.5±0.3, consistent with the R-I colors for Galactic

globular clusters of metallicities [Fe/H] ~ —1.0 (cf. Peterson et al. 1993)

3.4. Cuts Perpendicular to the Disk

In Figure 6a we show how the perpendicular cuts used by MBH are related to our combined 6660A
image. (Rather than term these as cuts "along the minor axis," we prefer to designate these as cuts
"perpendicular to the major axis." ) As did MBH, wc designate each cut by a letter and number, with A
being the central cut, Bl and B2 being the cuts nearest the center (1 to the northeast, 2 to the southwest),
and Dl and D2 being the cuts in the outer parts of the galaxy.

To increase S/N along these cuts we bin 45 pixels in width and, to keep this S/N approximately
constant, we increase the heights of the bins exponentially with z distance from the galaxy plane, with 1
pixel being the smallest height, and 59 pixels being the largest. This procedure is directly analogous to that
employed by MBH. For our estimate of galaxy surface brightness, we take the mode of each bin sampled.
The reason for this is that undetected background sources, plus residuals from the star masking procedure,
will tend to upward bias the mean and median ADU counts tabulated in each bin. We experimented with
alternate definitions of the flux in a bin and found that, while a clipping algorithms can partly resolve this
problem, the best way is to use mode value instead of mean or median. A a clipping algorithm is used to
eliminate bad pixel values (of which a few still remain in the combined image) from the histograms.

The perpendicular profiles we derive from the 6660 A image are given in Figure 10 and in Figure 11
from the 8020A image, and are tabulated in Table 5 for completeness and to aid in future studies of this
galaxy by others. The profiles of the two sides (NE and SW) about the galaxy plane are folded around
the curved warpline to remove the effect of the edge-on warp. Those parts of the profiles fainter than
m 6660 = 29 and mso2o = 28 (about 0.1% of sky) are not presented since their values are uncertain by over
1 mag owing to the finite size of our limiting systematic errors. The coding of the symbols used to plot
these perpendicular profiles is such that solid symbols represent parts of the profile not influenced by the
surface brightness distribution in the ring (Shang et al. 1998), while open symbols are those part of profiles
in which the ring light, and a possible warp in front of the galaxy, influences what we see.

It is obvious that at large z-distances from the plane of the galaxy, the faint surface brightness
distribution in the 6660A image around NGC 5907 is not symmetric about the midplane of the galaxy. Only

19

in the case of the A and C2 cuts are the profiles on the NE and SW sides of the galaxy consistent within the
errors at faint surface brightnesses. The NE sides of the profiles in cuts CI, Dl and D2 are systematically
brighter than the SW sides in these cuts. In the case of cuts Bl and B2, one side is brighter than the other,
but the sense of asymmetry changes from Bl to B2. The Bl profile (to the NW of the galaxy center) has
its NE side significantly brighter than the SW side, while the B2 profile (to the SW of the galaxy center)
has its SW side equally brighter than the NE side as for the Bl profile. Larger errors at faint levels in our
8020A image preclude any such conclusions being separately made for the cuts in this image.

Thus, on the basis of what we see in our 6660A image, we find that the faint surface brightness
distribution around NGC 5907 is asymmetric. With these data, we can confidently rule out the presence of
a halo in this galaxy beyond a distance of 5 kpc.

3.5. Ring, Foreground Stars and Possible Warp influences on the Asymmetry of the

Perpendicular Profiles

As can be seen by comparing the perpendicular profiles in Figures 10 and 11 to the positions of the
cuts relative to the ring in Figure 6a, cuts CI, Dl and D2 all intersect the ring on the NE side of the galaxy.
What confusion with the ring can do to the perpendicular profile is also seen in Panel 1 of Figure 7. The
apparently smooth gradient in the outer profile of the galaxy (from pixel numbers 600 to 750) is due almost
entirely to the presence of the ring in this region (cf. Figure 6a).

If we conservatively take the average surface brightness of the ring (28th mag/arcsec~ 2 ; in the 6660A
image; see Sec. 3.3) as our estimate for the ring surface brightness near the galaxy, then most of the
faint extensions seen in cuts CI, Dl and D2 on the NE side of the galaxy can be attributed to ring light.
Separately, as we can make a prima facie case that the ring is elliptical in shape and has the center of
NGC 5907 at one of its foci (cf. Shang et al. 1998), it is difficult to disentangle true galaxy light from
possible ring light in the A cut on the SW side of the galaxy at surface brightnesses fainter than 28th
mag/arcsec~ 2 in the 6660A image.

Of the remaining 3 of the six non-centered (B,C,D) cuts made, C2 has the least foreground star
contamination, but still has some star contamination on its SW side. In contrast, both Bl and B2 include
large sections of the masks around bright stars: B2 on its SW side, Bl on its NE side. Similarly CI on
its NE side, Dl on its NE side, and D2 on its SW side. Interestingly, in every case, the side of the galaxy

20

that is brighter is more foreground star-contaminated than its partner. This would strongly suggest that
residual star light at low light levels (surface brightnesses of 29th 6660A mag arcsec" 2 and fainter) is are
contributing to the differential effect in these cuts seen at surface brightnesses of 28th 6660A mag arcsec -2
and fainter.

On the other hand, if this galaxy has a face-on warp, one might expect that the warp would give an
asymmetrical distribution perpendicular to the center of the disk near the galaxy center, i.e., in cut Bl
versus cut B2. While such an effect is certainly seen, the apparently irreducible effects of foreground star
contamination make it highly unlikely that we will be able to unambiguously detect a face-on warp in this
galaxy, even if it exists. NGC 5907 simply has too many foreground stars around it to be a good edge-on
galaxy candidate in which to find a face-on warp.

4. Comparison with Previous Results

4.1. Comparison to the Results of MBH.

In order to do a close one-to-one comparison to the published results of MBH, Dr. H. Morrison kindly
sent us the surface brightness data they used to make their perpendicular cut diagrams in the MBH paper
(data which were not published in their paper). We list these data in Table 6 in the same manner as we did
for our own data. If we compare the broad-band R surface brightness profiles from MBH given in Table 6
to our own 6660A surface brightness profiles given in Table 5, it is apparent that, for similar errors, our
data go ~ 1.5 mag fainter than those of MBH. The main reason for this is that the sky is much fainter in
our intermediate band filter (which avoids the brightest night sky lines) than in the broad-band R filter
used by MBH: 0.6 e - s -1 arcsec -2 in our combined 6660A image to 7.4 e~ s _1 arcsec -2 quoted by MBH
for their R-band data.

As is evident from the graph presented in their paper, Sackett et al. combined NE and SW sides of
each cut to form one combined perpendicular profile per cut. With our higher accuracy at fainter light
levels, we can see the asymmetry in the faint light distribution perpendicular to the disk that MBH and
Sackett et al. could not.

We compare the perpendicular profiles of MBH to our 6660 A image profiles in Figure 12. A zero point
of 0.3 is added to the R band data of MBH to bring the two filter systems onto the same photometric system
at bright levels of the galaxy. While we note that MBH cite results only for R band surface brightnesses

21

brighter than 27th mag arcsec" 2 , the actual data given to us by Dr. Morrison go fainter than this. As
neither MBH or Sackett et al. give actual surface brightness profile information, we choose to use the data
sent to us by Dr. Morrison as the basis for our comparison.

We find generally good agreement with MBH data, even at low light levels, in most cases. Good
agreement is found between our data and those of MBH, both in form and zero point, for distances less
than 5 kpc from the galaxy midplane in all cuts. This gives strong support to accuracy of the zero points
obtained by both us and MBH, and to the fact that our perpendicular profiles are similar, typically at
surface brightnesses brighter than 27th mag arcsec -2 .

Above a distance of 5 kpc from the galaxy midplane, the comparison must take into account the
asymmetry of the faint light distribution seen. Namely, for the cuts Bl, B2, CI, Dl and D2, the two
faintest measurements of MBH are likely measuring only the brighter flux from one side of the galaxy (else,
we should see the MBH data to be 0.75 mag fainter than ours owing to this asymmetry). This is consistent
with what we see in Figure 12, in which the faintest MBH measurements are in accord with our brighter
surface brightnesses measured on one side of the galaxy or the other. The two exceptions to this agreement
at low light levels are the A and C2 cuts, where the surface brightness difference systematically occurs at
levels 27.5 mag arcsec -2 and fainter in these two cuts.

This comparison would strongly suggest that as well as in our data, the surface brightness profiles
determined by MBH arc affected by residual foreground star subtraction at very low light levels. Inspection
of Figure 3 of MBH, and the higher resolution modeled figure in Sackett ct al., show this to be the case.
One can see extra flux around the brighter masked stars in a similar kind of asymmetric pattern as we see
in our own masked 6660A image (Figure 6a). In addition, the two bright stars that are at the lower left
from the galaxy in their image show CCD bleeding along the columns that is not completely taken out by
their masking procedure.

Thus, we conclude that our data, both in zero point and in form, are in good agreement with the data
of MBH at low light levels. Differences that are seen are as likely to be attributable to issues pertaining to
foreground star subtraction as to any other kind of systematic error. From this we conclude that the data
used by MBH and Sackett et al. for their investigations into the possible thick disk and halo of NGC 5907
were as contaminated as ours by ring light and residual light from masked foreground stars.

22

4.2. Comparison to the Results of Lequeux et al. and Rudy et al.

The J, K, V and I surface brightness gradients in NGC 5907 are given in Rudy et al. (1997) for both

their observations (J and K) and those of Lequeux et al. (1996) (V and I). The additional B observations of

1998 do not have published numbers. In comparing these data to our own, we first need to determine if the
surface photometry was deep enough to detect the ring around NGC 5907. In the case of MBH, the answer
is yes (Sec. 4.1).

Of the observations of Rudy et al. (1997) and Lequeux et al. (1996), the evidence as to whether the
Rudy et al. (1997) data did go faint enough to detect the ring in the near IR is inconclusive. Rudy et al.
do remark that their surface brightness profiles show an asymmetry, with the NE side of the galaxy having
more light at low levels than the SW side. However, neither we nor MBH see this bump in our data. In the
case of Lequeux et al. (1996), it is apparent both from the image and the surface brightness profiles they
give that their observations did not go faint enough to detect the ring.

We compare the perpendicular profiles along the minor axis of NGC 5907 in the 8020A image to that

given by Lequeux et al. (1996). for the I band, and by Rudy et al. (1997) for the J and K band, in
Figure 13, and tabulate these data in Table 7. The zero point shift in the I band is 0.1, implying a zero
point shift between ni6660 — rnso20 and R — I of about 0.2.

There is generally good agreement among the J, K and 8020 A perpendicular profiles within the quoted
errors. The one exception is the presence of the above-mentioned "bump" on the NE side of the Rudy
et al. J and K data that is not present in cither our data, those of MBH (cf. Figure 12, A cut) or those
of Lequeux et al. This indicates little detectable color gradients in colors can be formed from these three
passbands, although the mutual errors are large enough to hide a significant color gradient, if it does exist
here. The absence of quoted errors for the Lequeux et al. I band data preclude an accurate assessment
of agreement /disagreement. Even so, within the cited errors of the other data sets, one finds differences
between the Lequeux et al. I band perpendicular profile and those of the other data sets only in the inner
NE 3 measured points. In contrast, the J, K and 8020A data for these three measured points are in very
good agreement. Once again, exactly how each group has handled the foreground star contamination
problem likely leads to the differences seen in the inner part of the profile shown in Figure 13.

23

4.3. Colors of NGC 5907 Perpendicular to its Disk

As is evident from the above comparison with other data, it is very difficult to obtain reliable colors
of galaxies at faint surface brightnesses. The errors of both passbands involved in forming the color
quadratically combine, and the resulting color differences are often small relative to this error. Such
is clearly the case for the color gradients we can form perpendicular to the disk of NGC 5907 in the
m 6660 — m 8020 color we can form from our data, as shown in Figure 14. The error bars are formal 1-er errors
per measurement. Since in these data 2-<r to 3-<r errors are likely present as well, it is apparent that the
color gradients 4 kpc and greater from the disk of NGC 5907 are completely dominated by observational
error.

Interestingly enough, within 4 kpc of the galaxy plane, we find that in many of the cuts the color
gradients become bluer in nigggo — mso20 with increasing radius, with cut D2 being the notable exception.
In D2 alone, the color gradient, on both sides of the galaxy, seems to get redder with distance from the
midplane.

As can be seen also in Figure 13, colors formed outside the 4-kpc midplane region from any of the
published passbands show little difference within the mutual errors of observations (save for the questionable
NE bump in J and K in the Rudy et al. data).

4.4. The Limiting Factors in Faint Galaxy Surface Photometry

We have already detailed the three sources of systematic error in faint galaxy surface photometry —
systematic errors in determining the flatficld, systematic errors in determining the sky background, residual
surface brightness from bright stars that one cannot easily mask. Most observers try to take these factors

into account (although not all explain that they do in their papers; cf. Lcqueux et al. 1996; 1998). However,
there are two other sources of zero point errors in sky level determination that are not discussed in the
previous papers. We include these now to complete our discussion of the data.

First is the fact discussed above that CCDs have wavelength-dependent quantum efficiencies and this
dependence is slightly different from pixel to pixel. Galaxies and sky background can have very different
colors, depending on which part of the spectrum is observed, how bright the night sky lines are, how bright
the sky continuum is, etc. If the difference in color is sufficient to change the effective wavelength of a filter
enough to change its quantum efficiency in a detectable manner, the sky level estimated will be affected in

24

direct proportion to the percent change in QE. Take the R band observation as an example, since the galaxy
effective wavelength could be off by 50-100A from the sky effective wavelength in the broad-band R, it is
entirely possible that the QE of the CCD could differ by 1 part in 1000 from galaxy effective wavelength to
sky effective wavelength. The zero point of the B, V and I observations are similarly affected.

The issue for J and K observations are somewhat different, as there the sky level is so bright that it is
a triumph of instrumentation that one can get a good estimate of sky down to less than 0.01% accuracy.
Again, however, if the IR Array QE is a function of effective wavelength, accuracy in estimate of sky
background in the broad band J and K filters will be limited by the difference in QE between galaxy color
and sky color.

We chose intermediate-band filters for the BATC survey for several reasons, but one of the chief ones
was to limit the bandpass sufficiently that differences in color of the objects observed would not produce
significant differences in the effective wavelength of the filters. In the case of the 6660A and 8020A filters,
we designed them to avoid the brightest (and most variable) night sky lines. As such, we believe the
change in QE between galaxy and sky in our filters is much less than in the broad-band filters. A separate
advantage, of course, is in a concomitantly much lower sky background (12 times lower in the 6660A filter
than MBH obtained at KPNO in the broad-band R).

The second source of non-modeled zero point error is the photometricity of the telescope field of view
and the inherent non-flatness of the night sky. All telescopes suffer from optical imperfections, most of
which affect the transmission of light from the sky as a function of distance from the center of the field. It is
primarily for this reason that the flatness of the night sky has become the final arbiter of how well the CCD
image has been calibrated. Yet, the night sky itself is not flat, which is pointed out by Wild (1997). In our
degree-sized BATC CCD fields we clearly see the gradient of the night sky getting brighter towards lower
altitudes. With CCD arrays getting to be degree-sized and greater, this problem is only enhanced. While
one takes many separate sky images at different times of night — so the altitudinal gradients go in different
directions on the CCD — in the end the cancellation of the gradients cannot be perfect. Any residual
gradient in the sky will then be mistaken as a gradient in the sensitivity of the CCD itself, and mistakenly
divided out. It should be subtracted out of the image, as we have attempted to do in the present paper.

One advantage of using a Schmidt telescope for the BATC survey is that we position the CCD in the
center 1° of a focal plane that has been specifically optimized to give as flat a field of view as possible in
the center of a 6° radius circle. Such is true for any Schmidt telescope, which is one reason they are noted

25

for giving the most reliable sky background estimates of any type of telescope.

The bottom line is that all sources of error conspire such that it is extremely difficult to obtain sky
background estimates and large-scale uniformities to accuracies better than 0.1% of the sky level. In defense
of our results, we can point to much lower sky level and narrower passbands.

4.5. Does NGC 5907 Have a Halo?

Two facts are clear from the above analysis: 1) Our data substantially agree with the data in the
literature for the surface brightness distribution seen perpendicular to the disk of NGC 5907. 2) We find
the light distribution perpendicular to the disk of NGC 5907 to be highly asymmetric about the galaxy
midplane, with one side enhanced relative to the other in each MBH B, C and D cut.

The difference between the present data set from those previously published is a combination of the
advantage of using the center field of a Schmidt telescope as well as intermediate-band filters. The use of
the intermediate band filters resulted in a much fainter sky; the large field of view of the Schmidt permitted
features such as the ring to be easily identified. The large angular pixel size of our CCD (1.71"/pixel) is
an advantage when searching for faint surface brightness, as we minimize the number of pixels per large
surface area.

Since we measure the ring around NGC 5907 to have an average surface brightness of 28th 6660A mag
arcsec -2 , one can see from the images themselves, Figures 2a and 6a, that NGC 5907 does not have a halo
at these faint surface brightnesses. This fact is more clearly evident on the SW side of the galaxy, the side
away from the bulk of the ring.

We are forced to the conclusion that asymmetries in foreground star PSFs lead to extra light around
star masks, which combine with ring surface brightness to produce what faint extensions of surface
brightness we see in the MBH cuts. This is due to the fact that the asymmetries in the surface brightnesses
derived from our images in the MBH cuts correlate with the degree of foreground star contamination in
four of the six non-centered cuts. Overlap with the ring similarly contributes to the asymmetries seen in
the other two of the non-centered cuts. As we can see similar effects in the star-masked image of MBH
(their Figure 2), and as their data are in accord with ours, we are equally forced to the conclusion that
their detection of a halo around NGC 5907 was an artifact caused by the same combination of ring light
and unaccounted faint wings from their foreground stars that affects our data.

26

5. CONCLUSION

We have obtained very deep images of the edge-on galaxy NGC 5907 with two intermediate-band filters
on the BATC system (cf. Fan et al. 1996), ni6660 & n d ni802o- Via a detailed assessment of the sources of our
errors, we show that our limiting magnitudes (where the observational error reaches 1 mag arcsec -2 ) are
29.0 mag arcsec -2 in the 6660A image (corresponding to 28.7 in R-band measures) and 28.0 mag arcsec -2
in the 8020A image (close to the I-band system). This is over one magnitude fainter than previously
published measurements, owing mostly to the much darker sky as seen in our intermediate-band filters.

We use a new method of sky subtraction that both can account for most of the foreground star
contamination (by PSF-fitting the star profiles first, then subtracting the fit), and permits reasonable
interpolation of sky under the galaxy region while also fitting the inherent lumpiness of the sky at low
surface brightnesses (using interactive fits of bidirectional, low-order Legendre polynomials and heavy
smoothing). Our use of diffuscr-smoothed dome flats, already shown to give accurate photometric results
over a wide range of stellar colors (Fan et al. 1996) are again shown to give reliable fiatfields with these
data. This is critical, as it can be easily shown (cf. Wild 1997) that the sky is not flat anywhere on degree
size scales.

As first shown by Shang et al. (1998), we detail the evidence that NGC 5907 has an elliptical-shaped
ring of emission around it, which mostly likely is the remnant of a dwarf spheroidal galaxy that has been
tidally torn apart. Within errors of 0.3 - 0.4 mag arcsec" 2 , we show that the R-I colors of the brighter part
of this ring arc consistent with the light coming from a moderately metal-poor old stellar population.

We investigate the surface brightness distribution perpendicular to the disk of NGC 5907 by taking
the same cuts through the galaxy as MBH. We show that all of the faint, extended surface brightness
we observe perpendicular to the disk of NGC 5907, and above an altitude of 5 kpc, is most likely due to
contamination by ring light and by residual effects from foreground star contamination. The possibility of a
face-on warp existing in NGC 5907, suggested by our HI observations, is too confused with foreground star
contamination to permit us to prove or disprove its optical existence.

Our comparisons to the existing perpendicular profiles in the literature for NGC 5907 find us in
substantial agreement, once predicted zero point differences are removed, with these data. Of particular
note is that with the known zero point difference between our 6660A filter and the R band removed, the
our data and those of MBH arc in excellent agreement.

27-

As such, we find that previous claims by Sackett et al. and others that NGC 5907 has a faint halo to
be an artifact due to a combination of two factors. One is the added light complications from the presence
of a previously-unknown ring around the galaxy. The second factor is the unaccounted residual effects of
the foreground star masking procedures for both our data and those of MBH at faint levels. The lack of a
halo around NGC 5907 is evident directly from our deep images of this galaxy (Figures 2a and 6a), once
one understands that the ring we detect has an average 6660A surface brightness of 28th mag arcsec -2 .

We thank Jinyao Hu and Xiaowei Liu for comments and discussions, Elias Brinks for his help in
obtaining the VLA data, Heather Morrsion for supplying us with the data of MBH, and the referee for
helpful comments. The research done with the BATC Survey is supported by the Chinese Academy
of Sciences (CAS), the Chinese National Natural Science Foundation (CNNSF) and the Chinese State
Committee of Sciences and Technology (CSCST). It is also supported in part by the U.S. National Science
Foundation (NSF Grant INT-93-01805), by Arizona State University, the University of Arizona and Western
Connecticut State University.

-28-

REFERENCES

Barnaby, D. & Thronson, H. A. 1992, AJ, 103, 41

Chen, J.-S. et al. 1999, in preparation

James, P. & Casali, M. M. 1996, Spectrum, March, 14

Fan, X.-H. et al. 1996, AJ, 112, 628

Lasker, B. M., et al., 1990, AJ, 99, 2019

Lequeux, J., Fort, B., Dantel-Fort, M., Cuillandre, J.-C. & Mellier, Y. 1996, A&A, 312, LI

Lequeux, J. et al. 1998, |astro-ph/9804109|

Morrison, H. L., Boroson, T. A., & Harding, P. 1994, AJ, 108, 1191 [MBH]

Oke, J. B. & Gunn, J. E., 1983, ApJ, 266, 713

Peterson, R. C, Dalle Ore, C. M. & Kurucz, R. L. 1993, ApJ, 404, 333

Rudy, R. J., Woodward. C. E., Hodge, T., Fairfield, S. W. & Harker, D. E. 1997, Nature, 387, 159

Sackett, P. D., Morrison, H. L., Harding, P., & Boroson, T. A. 1994, Nature, 370, 441

Sancisi, R. 1976, A&A, 53, 159

Sancisi, R. 1993, private communication

Sasaki, T. 1987, PASJ, 39, 849

Shang, Z.-H. et al. 1998, ApJL, accepted

Skurtskie, M. F., Shure, M. A. & Beckwith, S. 1985, ApJ, 299, 303

Stetson, P. B. 1987, PASP, 99, 191

Stetson, P. B. 1990, PASP, 102, 932

Tonry, J. & Schneider, D. P. 1988, AJ, 96, 807

-29-

Tonry, J. L., Ajhar, E. A., & Luppino, G. A. 1990, AJ, 100, 1416

van dor Kruit, P. C. 1979, A&AS, 38, 15

van dcr Kruit, P. C. & Searle, L., 1981, A&AS, 95, 105

Wild, W. J. 1997, PASP, 109, 1269

This manuscript was prepared with the AAS IAlgX macros v4.0.

30

Fig. 1. — The mean values of sky level, flux and FWHM for 20 comparison stars found in common on all
the frames in 6660A filter. The data of adjacent observations are displayed with different symbols. With
these parameters we selected out images with high signal-to-noise ratio so that we can get data as accurate
as possible.

Fig. 2a. - The 6660A, 26 h 20 m combined deep image of NGC5907. The central square shows the area used
in photometry, which is fOOOxfOOO pixels. The ring around NGC 5907 can be seen faintly in this image; its
mean surface brightness is 28th 6660A mag arcsec -2 .

Fig. 2b. — The 8020A, 16 h 40 m combined image of NGC5907. Note the ring is of too faint a surface brightness
to be seen in this image.

Fig. 3. — Four examples of subtracting and masking stars and galaxies, (a) A saturated bright star. This
PSF cannot be subtracted by DAOPHOT. Its wing extends so far that the mask radius is larger than 130
pixels. Another saturated star is found near this one. Their mask circles exclude most contamination of
their extended wings, (b) An unsaturated bright star. The residuals after the PSF is fit shows that the
central pixels have large fluctuations around the sky level and the residual wing of its PSF is still partly
visible. The mask radius used is 21 pixels, (c) A star of moderate brightness. The residuals after PSF is
fit show significantly smaller variation in its center than for the bright star, and the PSF wings for this star
are smaller as well. The mask radius used is 10 pixels, (d) A faint galaxy. The PSF for a galaxy is much
gradual than that of stars, such that it can not be subtracted properly by DAOPHOT using a stellar PSF.
The mask radius used is 9 pixels, while a star with the same brightness needs a mask radius of only 5 pixels
or less.

Fig. 4. — The final mask that will be applied to the 6660A image to account for foreground stars and
background galaxies and a circle region around NGC 5907. The central circle indicates the region in which
all pixels around NGC 5907 are masked; the curved area around this circle masks the ring we found around
this galaxy. Separately, for profile determinations the dust lane of NGC 5907 is also masked. The mask radii
of most faint stars are too small in size and low in contrast to be clearly displayed in this image.

31

Fig. 5. — The four steps of fitting background, (a) Fit the image row by row with Legendre polynomials of
order 3 or less, (b) Fit the image column by column in the same way. (c) Take the average of (a) and (b).
(d) Smooth (c) with a Gaussian function. It can be found that the final fitted background is too complicated
to be expressed by any 2-D analytical function. The net results of the fit underneat the galaxy region are
shown in the slices in Figures 7 and 8.

Fig. 6a. — The background-subtracted image in 6660A. The ring around the galaxy is very clear. It is a
near-perfect, but not complete ellipse, with a major axis of 13.7', minor axis of 7.2' and eccentricity e = 0.84
(cf. Shang et al. 1998). The major axis size is 44 kpc at a predicted distance of 11 Mpc (assuming Ho = 75
km s _1 Mpc -1 ). The area within the rectangle marked in the image was measured to determine the average
surface brightness of the brightest area of the ring in 6660A and 8020A, so that we can get color index of the
ring with reasonable accuracy. A schematic figure is inserted in bottom-right corner of the image to show
the relative position of the ring and the galaxy. The solid lines with labels indicate the cuts along which
surface photometry of the galaxy halo is done. These cuts are the same as that of Sackett ct al. (1994) and
MBH. The dashed lines show the centers of the slices made to check effect of background-subtraction. Note
the extent to which foreground stars exist to affect the surface brightness distributions we can obtain in the
MBH cuts.

Fig. 6b. — The background-subtracted image in 8020A.

Fig. 7. — Four slices in the background-subtracted 6660A image to show the accuracy of background fitting
and local surface brightness of the ring feature. The positions of the center of each slice relative to the galaxy
and ring are repreented in Fig. 6a by the dashed lines. Each slice is averaged over 31 lines of the image, and
only non-masked pixels are averaged. In the case that all 31 points are masked, a zero value is assigned. Also
shown are a straight line indicating the zero level. Note also the small, but still apparent, residual effect of
our star subtraction/masking procedure, a product of the compromise made in the star masking procedure.
The lower part of each panel (i.e., below the zero lines) shows the residuals of subtracted sky background for
each slice. The upper line in each panel shows the residuals produced by our sky fitting procedure, while the
lower line shows that produced by normal two-dimensional function fitting. They are shifted down by 110
ADU and 150 ADU seperately for ease of display. One can compare them and easily find that our procedure
produce smaller residuals. The arrows in the three panels show the positions where the slices cross the ring.

32

Fig. 8. — The slices in the background-subtracted 8020A image, shown in the same format as in Figure 7.

Fig. 9. — The stellar warp of NGC5907 measured in two bands. The circles are from 6660A image and the
squares are from 8020A image. The x axis is along the galaxy plane, which is determined by fitting the
symmetry axis of the galaxy. The stellar warp is obvious even in the cut nearest the galaxy center.

Fig. 10. — The surface brightness profiles in the 6660 A passband for the seven MBH-consistent cuts
perpendicular to the galaxy plane. Each panel corresponds to one cut shown in Fig. 6a and the separate
measurements of the two sides of the galaxy plane are shown. The data of NE and SW side of the galaxy
plane are denoted by squares and triangles, respectively. The open symbols denote the points affected by
the ring. Central regions of the galaxy are excluded due to star and dust masking. The asymmetries at faint
surface brightnesses that exist in these cuts are discussed in the text.

Fig. 11. — The surface brightness profiles in 8020A.

Fig. 12. — Comparison of our 6660A data with that of MBH, with a zero point shift of 0.3 mag applied
to the MBH data to account for the different passbands used. The circles are the data of MBH that were
sent to us by Dr. Morrison, which go fainter than those published by MBH in their actual paper. The solid
circles are the reliable data claimed by them and published in Sackett et al. 1994, while the open circles
(fainter than 27 mag arcsec -2 ) were not published. The other symbols have the same meaning as Fig. 10.
Note the general good agreement between our data and those of MBH (save for cuts A and C2) even at
surface brightnesses fainter than 27th mag arcsec -2 .

Fig. 13. — Comparison of our 8020A data with infrared data from Rudy et al. (J and K) and Lequeux et al.
(I), with zero point applied to the published data as discussed in the text. Note the general good agreement
among the different passbands, with their mutual errors, except for the "bump" seen in the data of Rudy et
al. but not in the other data sets.

Fig. 14. — The m6660- m 8020 color profiles of NGC5907. Measurable color gradients become bluer with
distance from the galaxy midplane in all cuts except D2, where the color gradient actually becomes redder
with distance from the midplane. The meaning of the symbols is the same as they in Fig. 10 and Fig. 11.
It is apparent that the errors in the colors preclude any meaningful measurement of color gradient outside
of 4 kpc from the plane of this galaxy.

This figure "zheng_fig01.jpg" is available in "jpg" format from:

http://arXiv.org/ps/astro-ph/9902276vl

This figure "zheng_fig02a.jpg" is available in "jpg" format from:

http://arXiv.org/ps/astro-ph/9902276vl

This figure "zheng_fig02b.jpg" is available in "jpg" format from:

http://arXiv.org/ps/astro-ph/9902276vl

This figure "zheng_fig03.jpg" is available in "jpg" format from:

http://arXiv.org/ps/astro-ph/9902276vl

This figure "zheng_fig04.jpg" is available in "jpg" format from:

http://arXiv.org/ps/astro-ph/9902276vl

This figure "zheng_fig05.jpg" is available in "jpg" format from:

http://arXiv.org/ps/astro-ph/9902276vl

This figure "zheng_fig06a.jpg" is available in "jpg" format from:

http://arXiv.org/ps/astro-ph/9902276vl

This figure "zheng_fig06b.jpg" is available in "jpg" format from:

http://arXiv.org/ps/astro-ph/9902276vl

This figure "zheng_fig07.jpg" is available in "jpg" format from:

http://arXiv.org/ps/astro-ph/9902276vl

This figure "zheng_fig08.jpg" is available in "jpg" format from:

http://arXiv.org/ps/astro-ph/9902276vl

This figure "zheng_fig09.jpg" is available in "jpg" format from:

http://arXiv.org/ps/astro-ph/9902276vl

This figure "zheng_figl0.jpg" is available in "jpg" format from:

http://arXiv.org/ps/astro-ph/9902276vl

This figure "zheng_figll.jpg" is available in "jpg" format from:

http://arXiv.org/ps/astro-ph/9902276vl

This figure "zheng_figl2.jpg" is available in "jpg" format from:

http://arXiv.org/ps/astro-ph/9902276vl

This figure "zheng_figl3.jpg" is available in "jpg" format from:

http://arXiv.org/ps/astro-ph/9902276vl

This figure "zheng_figl4.jpg" is available in "jpg" format from:

http://arXiv.org/ps/astro-ph/9902276vl

Table 1 . Log of 6660A images used

PL.

OS

>

t>
(N
(N

O

Oh:

6

c3

t

Image name 3

Date

Time

Exposure

Flux b

Sky level

Seeing d

DD/MM/YY

UT

(second)

(ADU)

(ADU)

FWHM (pixel)

P 0119803N014i019

26/03/95

18:07:21

1200

45688.32

391.19

2.08

P 0119803N014i020

26/03/95

18:29:40

1200

45763.04

395.91

2.21

P 0119803N014i021

26/03/95

18:51:57

1200

45903.98

399.51

2.23

P 0119803N014i023

26/03/95

19:36:53

1200

46019.57

442.44

2.05

P 0119803N014i024

26/03/95

19:59:28

1200

44634.13

481.58

2.39

p0129807N014i003

30/03/95

17:54:08

1200

43494.19

426.55

1.95

p0129807N014i005

30/03/95

18:38:37

1200

44200.91

427.85

1.95

p0129807N014i007

30/03/95

19:25:27

1200

43784.82

467.94

2.05

p0129808N014i009

31/03/95

16:48:28

1200

40461.00

458.07

2.14

p0129808N014i010

31/03/95

17:12:47

1200

39162.06

439.75

2.20

p0129808N014i011

31/03/95

17:37:00

1200

40732.67

440.08

1.98

p0129808N014i013

31/03/95

18:23:59

1200

41581.18

409.74

1.94

p0129808N014i014

31/03/95

18:46:20

1200

41918.14

409.81

2.07

p0129808N014i015

31/03/95

19:08:59

1200

42205.07

416.94

1.87

p0129808N014i016

31/03/95

19:31:40

1200

41838.50

421.53

2.44

P 0129810N014i033

02/04/95

16:08:17

600

37385.97

442.13

1.90

P 0129810N014i034

02/04/95

16:20:45

600

38249.45

426.84

1.80

P 0129810N014i035

02/04/95

16:35:48

600

38808.16

421.85

1.88

P 0129810N014i036

02/04/95

16:49:31

600

38612.60

414.41

2.30

P 0129810N014i090

02/04/95

18:43:11

600

41381.23

405.56

2.07

P 0129810N014i091

02/04/95

18:55:39

600

40778.88

409.80

2.17

P 0129810N014i092

02/04/95

19:07:57

600

40927.23

420.08

2.10

P 0129810N014i093

02/04/95

19:20:15

600

41159.31

423.30

2.12

P 0129810N014i094

02/04/95

19:32:49

600

40996.85

424.98

2.07

P 0129810N014i095

02/04/95

19:45:06

600

41282.66

434.43

1.93

P 0129810N014i096

02/04/95

19:57:26

600

40685.38

447.11

2.07

P 0129813N014i005

05/04/95

17:33:41

1200

41198.76

410.51

2.23

P 0129813N014i006

05/04/95

17:56:09

1200

40817.11

396.47

2.29

P 0129813N014i007

05/04/95

18:20:29

1200

42195.56

380.21

2.18

P 0129813N014i008

05/04/95

18:42:52

1200

40453.31

374.80

2.36

P 0129813N014i010

05/04/95

19:28:17

1200

40533.32

412.19

2.35

p0159832N014i011

24/04/95

16:25:51

1200

35953.96

450.15

3.33

p0159832N014i012

24/04/95

16:49:12

1200

35878.17

456.35

3.38

P 0159832N014i013

24/04/95

17:16:16

1200

38954.00

466.10

2.89

p0159832N014i014

24/04/95

17:39:47

1200

40301.73

517.22

2.58

p0159832N014i015

24/04/95

18:03:18

1200

41408.85

574.39

2.33

P 0159840N014i003

02/05/95

16:24:05

1200

35074.59

479.29

1.75

p0159840N014i004

02/05/95

16:49:06

1200

37000.46

487.52

1.82

p0159840N014i005

02/05/95

17:14:12

1200

37660.10

519.02

1.84

p0159840N014i006

02/05/95

17:39:27

1200

38818.73

530.52

2.17

P 0159840N014i007

02/05/95

18:07:31

1200

39185.15

523.33

2.25

p0159840N014i008

02/05/95

18:32:47

1200

39221.22

516.27

2.20

p0159840N014i009

02/05/95

18:56:33

1200

38255.55

519.34

2.42

p0159841N014i002

03/05/95

16:26:44

1200

33645.56

507.71

1.65

p0159841N014i003

03/05/95

16:50:30

1200

37463.91

470.97

1.64

p0159841N014i004

03/05/95

17:13:39

1200

38350.46

470.50

2.01

p0159841N014i005

03/05/95

17:37:20

1200

38007.07

460.79

1.86

p0159841N014i006

03/05/95

18:01:55

1200

38629.03

449.87

1.72

Table 1. (continued)

Image name"

Date

Time

Exposure

Flux b

Sky level

Seeing d

DD/MM/YY

UT

(second)

(ADU)

(ADU)

FWHM (pixel)

p0159841N014i007

03/05/95

18:25:25

1200

38639.90

451.50

1.74

p0159841N014i008

03/05/95

18:50:47

1200

38605.21

477.84

1.76

P 0169843N014i005

05/05/95

15:57:59

1200

34936.81

455.66

1.85

p0169843N014i006

05/05/95

16:21:19

1200

34240.71

448.46

1.89

p0169843N014i007

05/05/95

16:47:35

1200

33895.24

464.00

1.85

p0169843N014i008

05/05/95

17:11:23

1200

33179.18

458.11

1.91

p0169843N014i009

05/05/95

17:34:57

1200

32687.67

470.84

1.91

p0169843N014i010

05/05/95

17:58:08

1200

32377.79

481.40

1.90

p0169843N014i011

05/05/95

18:27:31

1200

32727.70

502.62

1.77

p0169843N014i012

05/05/95

18:53:45

1200

32214.26

547.08

1.74

P 0169844N014i015

06/05/95

17:40:14

1200

35947.07

533.51

1.87

P 0169844N014i016

06/05/95

18:03:27

1200

35808.12

521.69

1.88

P 0169844N014i017

06/05/95

18:26:38

1200

34774.71

546.38

1.92

P 0169844N014i018

06/05/95

18:49:34

1200

33989.09

546.47

2.00

P 0179860N014i000

22/05/95

14:58:24

1200

39589.11

515.46

2.66

p0179860N014i001

22/05/95

15:22:14

1200

38286.74

528.31

2.72

p0179860N014i002

22/05/95

15:46:07

1200

38195.22

520.04

2.65

P 0179860N014i003

22/05/95

16:10:04

1200

37075.36

538.74

2.85

p0179860N014i004

22/05/95

16:35:15

1200

36371.84

562.57

2.87

P 0179860N014i005

22/05/95

16:58:32

1200

34934.22

611.09

2.86

p0179862N014i009

24/05/95

16:36:55

1200

40573.72

458.35

1.98

p0179862N014i010

24/05/95

16:59:54

1200

40855.74

476.34

2.05

p0179862N014i011

24/05/95

17:22:40

1200

39471.87

540.44

2.23

P 0179863N014i008

25/05/95

16:00:59

1200

42368.73

497.11

2.28

p0179863N014i009

25/05/95

16:23:13

1200

42437.13

511.08

2.14

p0179863N014i010

25/05/95

16:45:56

1200

42616.05

513.07

2.15

p0179863N014i012

25/05/95

17:30:51

1200

42919.03

523.20

2.16

P 0179863N014i013

25/05/95

17:53:47

1200

42156.86

577.28

2.36

P 0189873N014i007

04/06/95

15:24:25

1200

39912.96

502.76

2.13

P 0189873N014i008

04/06/95

15:48:01

1200

40254.62

492.77

2.21

P 0189873N014i009

04/06/95

16:10:39

1200

39109.38

505.35

2.36

p0189873N014i010

04/06/95

16:45:00

1200

38815.32

493.66

2.08

P 0189873N014i011

04/06/95

17:07:40

1200

38853.51

492.20

2.09

p0209896N014i001

27/06/95

14:13:54

1200

40678.32

488.30

1.92

p0209896N014i002

27/06/95

14:38:33

1200

40598.48

499.05

1.71

p0209896N014i004

27/06/95

15:30:07

1200

40878.30

488.96

1.80

a The image name is organized as following sequence: prrrjjjjoooofsss
p = PIPELINE-I processed data
rrr = run number

jjjj = Julian date, only the last four digits are used
oooo = object number, NGC 5907 has two numbers: T514 and N014
f = filter used (i=6660A, m=8020A)
sss = observation sequence of given night

b Mean flux of the 20 comparison stars, normalized to 20m integration

c Mean near-field sky level for the 20 comparison stars, normalized to 20m integration

d Mean FWHM of the 20 comparison stars

Table 2. Log of 8020A images used

Image name

Date

Time

Exposure

Flux

Sky level

Seeing

DD/MM/YY

UT

(second)

(ADU)

(ADU)

FWHM(pixel)

p0330135N014m039

21/02/96

19:02:58

1200

27634.88

506.71

2.09

p0330135N014m040

21/02/96

19:26:01

1200

27828.17

474.49

2.15

p0330135N014m041

21/02/96

19:48:33

1200

27768.68

453.67

2.03

p0330135N014m042

21/02/96

20:11:05

1200

27738.37

421.46

1.96

p0330135N014m043

21/02/96

20:33:50

1200

28107.29

395.21

1.97

p0330136N014m036

22/02/96

17:17:53

1200

27405.63

436.15

1.91

p0330136N014m037

22/02/96

17:41:21

1200

27723.22

454.58

1.90

p0330136N014m038

22/02/96

18:03:50

1200

27836.49

462.55

1.94

p0330136N014m039

22/02/96

18:26:24

1200

27933.66

439.44

1.92

p0330136N014m040

22/02/96

18:49:04

1200

27979.31

405.05

1.88

p0330136N014m041

22/02/96

19:11:39

1200

28081.59

414.47

1.90

p0330136N014m042

22/02/96

19:34:13

1200

28007.59

393.42

1.93

p0330136N014m043

22/02/96

19:56:38

1200

28007.20

383.90

1.92

p0330136N014m044

22/02/96

20:19:11

1200

28133.33

415.54

2.03

p0330136N014m045

22/02/96

20:41:35

1200

28087.49

402.87

2.17

p0330141T534m038

27/02/96

19:07:08

1200

24137.61

550.05

2.38

p0330141T534m039

27/02/96

19:29:47

1200

24496.86

589.32

2.26

p0330141T534m040

27/02/96

19:52:35

1200

24779.54

623.50

2.26

p0330141T534m041

27/02/96

20:15:21

1200

24840.06

671.74

2.23

p0330141T534m042

27/02/96

20:38:05

1200

24996.80

654.99

2.29

p0340153N014m035

10/03/96

18:32:29

1200

26708.87

749.01

2.30

p0340153N014m036

10/03/96

18:55:00

1200

26668.87

703.19

2.29

p0340153N014m037

10/03/96

19:17:42

1200

26710.86

683.35

2.57

p0340153N014m038

10/03/96

19:40:54

1200

26759.61

749.38

2.39

p0340153N014m039

10/03/96

20:03:40

1200

26791.77

764.89

2.23

p0340153N014m040

10/03/96

20:26:10

1200

26821.89

778.71

2.16

p0340157T534m042

14/03/96

17:52:54

1200

22775.84

520.57

2.40

p0340157T534m043

14/03/96

18:15:20

1200

23111.24

544.43

2.43

p0340157T534m044

14/03/96

18:37:57

1200

22794.46

560.05

2.46

p0360190N014m002

16/04/96

16:33:07

1200

31314.58

438.65

2.42

p0360190N014m003

16/04/96

16:56:39

1200

31275.79

480.01

2.38

p0360190N014m004

16/04/96

17:20:21

1200

31147.07

484.32

2.48

p0360190N014m005

16/04/96

17:44:13

1200

31053.15

477.52

3.19

p0360190N014m006

16/04/96

18:08:19

1200

31111.32

482.44

3.11

p0360190N014m008

16/04/96

19:00:01

1200

30521.54

359.35

3.41

p0360190N014m009

16/04/96

19:23:44

1200

30067.63

420.39

2.41

p0360191N014m020

17/04/96

17:00:54

1200

29464.94

414.97

1.98

p0360191N014m021

17/04/96

17:24:51

1200

29541.54

353.81

2.38

p0360191N014m022

17/04/96

17:48:24

1200

29756.01

330.02

1.68

p0360191N014m023

17/04/96

18:13:53

1200

29926.16

312.84

1.70

p0360191N014m024

17/04/96

18:36:24

1200

30251.59

313.91

1.81

p0360191N014m025

17/04/96

18:59:09

1200

30290.93

335.49

1.84

p0360191N014m026

17/04/96

19:21:42

1200

30314.27

333.15

1.96

p0370192N014m024

18/04/96

16:49:22

1200

29078.35

581.85

1.89

p0370192N014m025

18/04/96

17:11:48

1200

29211.13

552.14

1.77

p0370192N014m026

18/04/96

17:34:31

1200

28949.63

516.00

1.67

p0370192N014m027

18/04/96

17:57:22

1200

29204.02

531.85

1.82

p0370192N014m028

18/04/96

18:21:37

1200

29693.29

572.58

1.99

Table 2. (continued)

Image name

Date
DD/MM/YY

Time Exposure
UT (second)

Flux

(ADU)

Sky level
(ADU)

Seeing
FWHM(pixel)

p0370192N014m029
p0370192N014m030

18/04/96
18/04/96

18:44:10 1200
19:06:46 1200

29669.82

28722.99

545.97
590.20

1.84
2.05

Notes to Table 2.

See notes to Table 1

Table 3. Parameters of the combined images

Filter

Total
exposure times

Seeing'' 1
FWHM (pixel)

Sky level

(mag/sq arcsec)

re b

CTsky
(ADU)

Readout noise c

(O

Zero point d

(lADU/s)

6660A
8020A

26 h 10 m
16 h 40 m

2.3

2.5

21.26
19.91

105

85

110

85

19.08
17.79

a Mean FWHM of all the stars in the field.

b The standard error of sky background per pixel.

c Total readout noise introduced into the final combined images per pixel.

d Photometry zero point calibrated by measuring Oke-Gunn standard stars.

Table 4. Summary of limiting error sources in a 45x45 pixel bin in the 6660 A image

Source of variation

ADU

%

Reducible noise

Readout noise

0.6

0.002

Photon statistics

2.1

0.006

Bias and Dark

0.02

0.00005

Flatfielding errors

0.08

0.0002

Surface brightness fluctuations

0.3

0.0008

Irreducible noise

Large-scale FF errors

3.7

0.01

Systematic error of background subtraction

18.0

0.05

Total

23.9

0.06

Table 5. Perpendicular surface brightness profiles of NGC 5907 in the 6660 A and 8020A images

distance

distance

6660A

error(6660A)

8020A

error(8020A)

(arcsec)

(kpc)

(mag/sq arcsec)

(mag/sq arcsec)

(mag/sq arcsec)

(mag/sq arcsec)

Cut A, r =

= kpc

-184.63*

-9.83

28.91

0.72

-180.35*

-9.60

28.91

0.72

-173.43*

-9.23

28.70

0.60

-165.56*

-8.81

28.52

0.51

-158.21*

-8.42

28.28

0.41

-151.28*

-8.05

28.19

0.38

27.81

0.91

-144.70*

-7.70

28.12

0.36

27.54

0.71

-138.54*

-7.37

27.96

0.31

27.24

0.54

-132.64*

-7.06

27.82

0.28

27.24

0.55

-127.00*

-6.76

27.69

0.25

26.95

0.42

-121.62*

-6.47

27.50

0.21

26.79

0.36

-116.490

-6.20

27.33

0.18

26.62

0.32

-111.53o

-5.94

27.12

0.15

26.65

0.33

-102.04o

-5.43

26.93

0.13

26.61

0.32

-93.14o

-4.96

26.63

0.10

26.56

0.32

-84.68o

-4.51

26.31

0.09

26.13

0.24

-76.56o

-4.07

26.08

0.08

25.86

0.20

-64.93o

-3.46

25.84

0.07

25.30

0.13

-6I.O80

-3.25

25.54

0.06

25.14

0.12

-49.97o

-2.66

24.94

0.03

24.45

0.07

-48.09o

-2.56

24.66

0.03

24.29

0.06

-44.49o

-2.37

24.45

0.02

24.02

0.05

-42.70o

-2.27

24.17

0.02

23.84

0.04

-34.010

-1.81

23.00

0.02

-27.17o

-1.45

22.30

0.01

-22.04*

-1.17

21.68

0.01

-18.62*

-0.99

21.38

0.02

25.84*

1.38

22.04

0.02

32.68*

1.74

22.68

0.01

39.52*

2.10

23.36

0.02

46.36*

2.47

24.11

0.04

25.84*

1.38

22.04

0.02

32.68*

1.74

22.68

0.01

39.52*

2.10

23.36

0.02

46.36*

2.47

24.11

0.04

53.49*

2.85

25.22

0.04

24.72

0.07

57.17*

3.04

25.45

0.04

24.94

0.08

70.33*

3.74

26.01

0.06

25.55

0.14

80.08*

4.26

26.29

0.08

25.90

0.19

88.20*

4.69

26.41

0.09

26.23

0.25

96.67*

5.14

26.78

0.12

26.49

0.30

105.56*

5.62

26.89

0.13

27.02

0.47

115.05*

6.12

27.12

0.15

27.27

0.58

120.01*

6.39

27.18

0.16

27.92

1.04

125.14*

6.66

27.33

0.18

130.52*

6.95

27.40

0.19

Table 5. (continued)

distance

distance

6660A

error(6660A)

8020A

error(8020A)

(arcscc)

(kpc)

(mag/sq arcsec)

(mag/sq arcsec)

(mag/sq arcsec)

(mag/sq arcsec)

136.17*

7.25

27.47

0.20

142.07*

7.56

27.64

0.23

148.22.

7.89

27.74

0.25

154.81*

8.24

28.03

0.33

161. 73*

8.61

28.23

0.39

169.08*

9.00

28.51

0.51

176.95*

9.42

28.65

0.57

27.74

0.81

186.44*

9.92

28.75

0.63

27.74

0.81

196.84*

10.48

29.15

0.91

27.88

0.92

207.10*

11.02

27.90

0.94

217.36*

11.57

27.88

0.93

225.91*

12.02

27.58

0.71

233.60*

12.43

27.71

0.80

239.59*

12.75

27.74

0.83

Cut Bl, r

= 4.1 kpc

-178.68*

-9.51

28.92

0.73

-171.75*

-9.14

28.85

0.68

-163.89*

-8.72

28.73

0.62

27.74

0.83

-156.53*

-8.33

28.71

0.61

27.39

0.60

-149.61*

-7.96

28.53

0.52

27.16

0.50

-143.02*

-7.61

28.33

0.44

27.00

0.43

-136.87*

-7.28

28.17

0.38

27.13

0.49

-130.97*

-6.97

28.06

0.35

27.04

0.45

-125.33*

-6.67

27.99

0.33

27.01

0.45

-119.94*

-6.38

27.69

0.25

26.92

0.41

-114.810

-6.11

27.58

0.23

26.87

0.40

-107.37o

-5.71

27.28

0.18

26.83

0.41

-100.36o

-5.34

27.06

0.16

26.55

0.33

-95.910

-5.10

26.94

0.14

26.66

0.39

-89.33o

-4.75

26.87

0.14

26.15

0.25

-83.00o

-4.42

26.71

0.12

26.01

0.21

-78.98o

-4.20

26.52

0.10

25.80

0.18

-72.91o

-3.88

26.36

0.09

25.92

0.21

-68.98o

-3.67

26.16

0.07

25.67

0.18

-65.13o

-3.47

26.00

0.06

25.66

0.17

-61.29o

-3.26

25.84

0.05

25.25

0.12

-57.52o

-3.06

25.61

0.04

24.93

0.08

-53.85o

-2.87

25.34

0.03

24.79

0.07

-46.41o

-2.47

24.79

0.02

24.38

0.05

-42.82o

-2.28

24.43

0.02

23.97

0.03

-39.31o

-2.09

24.02

0.01

23.70

0.03

-35.89o

-1.91

23.75

0.01

23.31

0.02

-32.47o

-1.73

23.44

0.01

22.97

0.02

-27.34o

-1.46

23.12

0.02

22.52

0.01

-20.26o

-1.08

21.68

0.01

24.20*

1.29

22.07

0.01

Table 5. (continued)

distance

distance

6660A

error(6660A)

8020A

error(8020A)

(arcscc)

(kpc)

(mag/sq arcsec)

(mag/sq arcsec)

(mag/sq arcsec)

(mag/sq arcsec)

30.80.

1.64

23.28

0.02

22.74

0.01

34.22.

1.82

23.52

0.01

23.30

0.02

37.64.

2.00

23.97

0.01

23.62

0.03

41.06.

2.18

24.25

0.02

23.93

0.04

44.48.

2.37

24.53

0.02

24.29

0.05

49.69.

2.64

24.95

0.03

24.65

0.07

53.28.

2.84

25.13

0.03

24.92

0.08

56.96.

3.03

25.50

0.04

25.20

0.11

62.52.

3.33

25.82

0.05

25.34

0.12

68.16.

3.63

26.08

0.07

25.51

0.14

72.01.

3.83

26.21

0.07

25.57

0.14

75.86.

4.04

26.24

0.07

25.67

0.15

79.79.

4.25

26.24

0.07

25.90

0.18

83.81.

4.46

26.46

0.09

26.28

0.26

87.83.

4.67

26.59

0.10

26.30

0.26

92.02.

4.90

26.60

0.10

26.33

0.27

98.34.

5.23

26.68

0.11

26.49

0.30

105.01.

5.59

26.82

0.12

26.84

0.42

109.54.

5.83

26.87

0.13

27.12

0.54

114.25.

6.08

26.85

0.12

27.51

0.81

119.12.

6.34

27.04

0.15

124.25.

6.61

27.15

0.16

129.46.

6.89

27.27

0.18

134.94.

7.18

27.31

0.19

140.75.

7.49

27.31

0.18

27.76

0.90

146.74.

7.81

27.42

0.20

27.84

0.95

153.15.

8.15

27.60

0.23

27.91

1.00

159.90.

8.51

27.66

0.24

27.83

0.93

167.00.

8.89

27.73

0.26

27.77

0.87

174.61.

9.29

27.79

0.28

27.64

0.75

180.42.

9.60

27.84

0.29

27.51

0.67

188.12.

10.01

27.83

0.29

27.54

0.69

194.96.

10.37

27.88

0.30

27.79

0.86

201.80.

10.74

27.89

0.31

27.75

0.84

208.64.

11.10

27.88

0.31

27.84

0.91

215.48.

11.47

27.86

0.31

27.90

0.96

220.61.

11.74

27.88

0.32

27.95

1.01

227.45.

12.10

27.78

0.29

27.78

0.86

235.14.

12.51

27.81

0.31

27.57

0.71

242.84.

12.92

28.26

0.47

27.61

0.74

249.68.

13.29

28.31

0.47

27.63

0.74

256.52.

13.65

28.51

0.55

27.77

0.84

264.21.

14.06

28.78

0.68

272.76.

14.52

28.47

0.50

Cut B2, r

= -4.1 kpc

-271.62.

-14.45

28.60

0.54

-268.20.

-14.27

28.58

0.53

Table 5. (continued)

distance

distance

6660A

error(6660A)

8020A

error(8020A)

(arcsec)

(kpc)

(mag/sq arcsec)

(mag/sq arcsec)

(mag/sq arcsec)

(mag/sq arcsec)

-263.07.

-14.00

28.52

0.50

-257.94.

-13.73

28.55

0.51

-251.95.

-13.41

28.44

0.46

-246.82.

-13.13

28.38

0.44

-243.40.

-12.95

28.27

0.40

-238.27.

-12.68

28.21

0.38

-232.29.

-12.36

28.13

0.35

-227.16.

-12.09

28.06

0.33

-221.17.

-11.77

28.08

0.33

-216.04.

-11.50

28.07

0.33

-210.91.

-11.22

28.07

0.33

-207.49.

-11.04

28.00

0.31

-202.36.

-10.77

27.94

0.30

27.94

0.98

-197.23.

-10.50

27.88

0.28

-192.19.

-10.23

27.86

0.28

27.72

0.80

-186.97.

-9.95

27.87

0.28

27.68

0.78

-180.99.

-9.63

27.86

0.28

27.70

0.79

-174.06.

-9.26

27.84

0.28

27.72

0.81

-166.19.

-8.84

27.84

0.28

27.62

0.75

-158.84.

-8.45

27.72

0.26

27.49

0.68

-151.92.

-8.08

27.75

0.27

27.60

0.77

-145.33.

-7.73

27.75

0.28

27.54

0.74

-139.18.

-7.41

27.55

0.24

27.23

0.58

-133.28.

-7.09

27.47

0.22

27.20

0.57

-127.63.

-6.79

27.45

0.22

27.29

0.63

-122.25.

-6.51

27.32

0.20

27.10

0.53

-117.120

-6.23

27.26

0.19

26.61

0.35

-107.37o

-5.71

27.05

0.15

26.74

0.38

-100.45o

-5.35

26.99

0.14

26.67

0.36

-91.64o

-4.88

26.75

0.11

26.28

0.25

-85.31*

-4.54

26.58

0.10

26.04

0.21

-79.24o

-4.22

26.40

0.09

25.97

0.20

-73.26*

-3.90

26.25

0.08

26.07

0.22

-67.44o

-3.59

26.04

0.06

25.81

0.17

-61.71o

-3.28

25.90

0.06

25.49

0.13

-57.95*

-3.08

25.65

0.05

25.30

0.11

-54.28o

-2.89

25.39

0.04

25.08

0.09

-50.60o

-2.69

25.11

0.03

25.02

0.09

-46.92o

-2.50

24.88

0.02

24.56

0.06

-43.33o

-2.31

24.50

0.02

24.34

0.05

-39.91*

-2.12

24.12

0.01

24.02

0.04

-36.49o

-1.94

23.79

0.01

23.65

0.03

-33.07o

-1.76

23.45

0.01

23.20

0.02

-29.65o

-1.58

23.14

0.01

22.75

0.01

-26.23o

-1.40

22.83

0.01

22.29

0.01

-17.68*

-0.94

21.43

0.01

-15.97*

-0.85

21.42

0.01

Table 5. (continued)

distance

distance

6660A

error(6660A)

8020A

error(8020A)

(arcscc)

(kpc)

(mag/sq arcsec)

(mag/sq arcsec)

(mag/sq arcsec)

(mag/sq arcsec)

25.07*

1.33

22.52

0.02

35.33*

1.88

23.72

0.01

23.34

0.03

38.75*

2.06

23.99

0.01

23.68

0.04

42.17*

2.24

24.36

0.02

24.26

0.06

45.59*

2.43

24.69

0.02

24.31

0.06

49.18*

2.62

24.99

0.03

24.63

0.08

52.86*

2.81

25.32

0.04

24.92

0.10

58.41*

3.11

25.66

0.05

25.26

0.12

62.09*

3.30

25.79

0.05

25.39

0.14

65.85*

3.50

25.97

0.06

25.86

0.21

69.70*

3.71

26.15

0.07

25.69

0.17

75.51*

4.02

26.32

0.08

26.20

0.27

81.50*

4.34

26.47

0.09

26.57

0.36

87.57*

4.66

26.59

0.10

27.18

0.61

91.76*

4.88

26.75

0.11

27.48

0.78

96.03*

5.11

26.87

0.13

27.26

0.63

102.70*

5.47

27.05

0.15

27.15

0.58

109.63*

5.83

27.20

0.17

26.96

0.47

114.42*

6.09

27.45

0.20

27.59

0.82

119.38*

6.35

27.59

0.23

27.56

0.78

124.51*

6.63

27.59

0.23

27.41

0.67

129.89*

6.91

27.66

0.24

135.53*

7.21

27.69

0.25

141.43*

7.53

27.81

0.27

27.59

0.76

147.59*

7.85

27.98

0.32

27.55

0.73

154.17*

8.20

28.03

0.33

27.53

0.70

161.10*

8.57

28.21

0.38

27.76

0.86

168.45*

8.96

28.40

0.46

27.69

0.79

176.32*

9.38

28.59

0.54

27.72

0.81

183.24*

9.75

28.80

0.65

27.78

0.86

189.23*

10.07

29.11

0.87

27.58

0.71

Cut CI, r

= 8.2 kpc

-146.38*

-7.79

28.90

0.74

-136.97*

-7.29

28.84

0.72

-128.16*

-6.82

28.81

0.72

-120.04*

-6.39

28.34

0.49

-112.35*

-5.98

28.24

0.46

-105.17*

-5.60

28.03

0.38

-98.24*

-5.23

27.51

0.23

-91.57*

-4.87

27.32

0.19

27.33

0.70

-85.24o

-4.54

27.02

0.15

26.73

0.41

-79.09o

-4.21

27.00

0.15

26.98

0.53

-73.02o

-3.89

26.82

0.13

26.87

0.49

-69.08o

-3.68

26.49

0.10

26.12

0.24

-65.24o

-3.47

26.31

0.08

26.21

0.26

-61.39o

-3.27

26.06

0.07

26.05

0.23

-57.63o

-3.07

25.81

0.05

25.49

0.14

Table 5. (continued)

distance

distance

6660A

error(6660A)

8020A

error(8020A)

(arcscc)

(kpc)

(mag/sq arcsec)

(mag/sq arcsec)

(mag/sq arcsec)

(mag/sq arcsec)

-52.07o

-2.77

25.50

0.04

25.16

0.11

-48.39o

-2.58

25.16

0.03

24.77

0.08

-44.72o

-2.38

24.80

0.02

24.43

0.06

-41.13*

-2.19

24.55

0.02

24.23

0.05

-39.42o

-2.10

24.30

0.01

24.11

0.05

-36.00o

-1.92

24.07

0.01

23.69

0.03

-32.58o

-1.73

23.77

0.01

23.41

0.02

-29.16*

-1.55

23.41

0.01

23.09

0.02

-17.19o

-0.91

22.33

0.01

21.90

0.01

-14.71o

-0.78

21.67

0.01

23.85o

1.27

22.71

0.01

22.27

0.01

25.56o

1.36

22.92

0.01

22.52

0.01

27.27o

1.45

23.10

0.01

22.70

0.01

32.40o

1.72

23.56

0.01

23.32

0.02

35.82o

1.91

23.97

0.01

23.77

0.03

37.53o

2.00

24.15

0.01

24.06

0.04

39.24o

2.09

24.39

0.02

24.28

0.05

42.66o

2.27

24.70

0.02

24.48

0.06

46.08o

2.45

24.97

0.03

24.71

0.07

51.39o

2.73

25.42

0.04

25.07

0.10

55.06o

2.93

25.81

0.06

25.45

0.14

58.74o

3.13

25.86

0.06

26.09

0.26

64.30o

3.42

26.19

0.08

26.00

0.24

70.02o

3.73

26.36

0.09

25.77

0.18

75.75o

4.03

26.50

0.10

26.58

0.37

81.65o

4.35

26.66

0.11

26.62

0.35

87.72o

4.67

26.73

0.11

26.50

0.31

93.96o

5.00

26.89

0.13

26.49

0.30

100.46O

5.35

27.05

0.16

26.85

0.42

107.13O

5.70

27.29

0.20

27.44

0.73

114.140

6.07

27.29

0.20

27.22

0.60

121.58o

6.47

27.33

0.21

27.08

0.54

129.36o

6.88

27.35

0.21

26.96

0.48

137.74o

7.33

27.33

0.21

26.74

0.39

146.63o

7.80

27.29

0.20

26.43

0.28

156.38o

8.32

27.33

0.22

26.25

0.24

163.31o

8.69

27.40

0.23

26.26

0.24

170.66o

9.08

27.42

0.24

26.29

0.25

180.32o

9.60

27.49

0.27

26.29

0.25

I88.OI0

10.01

27.78

0.34

26.61

0.33

194.85o

10.37

27.86

0.36

26.93

0.43

201.69o

10.73

27.75

0.32

27.27

0.58

208.53o

11.10

27.95

0.35

27.38

0.63

215.37o

11.46

28.08

0.38

222.21o

11.83

28.11

0.37

229.91o

12.23

28.00

0.33

236.75o

12.60

27.96

0.32

10

Table 5. (continued)

distance

distance

6660A

error(6660A)

8020A

error(8020A)

(arcscc)

(kpc)

(mag/sq arcsec)

(mag/sq arcsec)

(mag/sq arcsec)

(mag/sq arcsec)

242. 73o

12.92

27.98

0.32

249.57o

13.28

27.99

0.32

256.41o

13.65

28.02

0.33

27.97

1.01

262.40o

13.96

28.02

0.32

269. 24o

14.33

28.02

0.32

27.93

0.97

276.08o

14.69

28.03

0.32

Cut C2, r

= -8.2 kpc

-188.78

-10.05

28.81

0.66

-180.23

-9.59

28.80

0.66

27.57

0.70

-171.85

-9.15

28.63

0.57

27.49

0.66

-164.25

-8.74

28.34

0.44

27.53

0.69

-153.73

-8.18

28.23

0.40

27.47

0.66

-143.98

-7.66

27.96

0.32

27.24

0.55

-135.09

-7.19

27.67

0.25

26.92

0.43

-126.71

-6.74

27.44

0.20

26.63

0.33

-118.930

-6.33

27.36

0.19

26.57

0.33

-111.49o

-5.93

27.13

0.16

26.43

0.30

-104.48o

-5.56

27.10

0.15

26.87

0.47

-97.81o

-5.21

26.89

0.13

26.67

0.41

-91.310

-4.86

26.65

0.11

26.91

0.52

-85.07o

-4.53

26.53

0.10

26.12

0.27

-79.000

-4.20

26.49

0.10

26.92

0.57

-73.100

-3.89

26.48

0.10

26.33

0.30

-69.26o

-3.69

26.15

0.07

25.92

0.20

-65.41o

-3.48

25.94

0.06

25.82

0.19

-61.65o

-3.28

25.79

0.05

25.75

0.19

-56.09o

-2.98

25.72

0.05

25.42

0.14

-50.53o

-2.69

25.47

0.04

25.30

0.13

-45.14o

-2.40

25.09

0.03

25.07

0.10

-40.01o

-2.13

24.61

0.02

24.41

0.06

-34.880

-1.86

24.12

0.01

24.07

0.04

-31.46o

-1.67

23.66

0.01

23.59

0.03

-28.04o

-1.49

23.42

0.01

23.26

0.02

-26.33o

-1.40

23.22

0.01

22.99

0.02

-22.91o

-1.22

23.09

0.02

22.62

0.01

-19.120

-1.02

21.99

0.03

15.08o

0.80

21.71

0.01

21.55o

1.15

23.03

0.02

22.56

0.01

23.26o

1.24

23.21

0.01

22.69

0.01

26.68o

1.42

23.56

0.01

23.00

0.01

28.39o

1.51

23.57

0.01

23.17

0.02

31.81o

1.69

23.87

0.01

23.55

0.02

36.94o

1.97

24.33

0.01

24.02

0.04

43.78o

2.33

24.95

0.03

24.64

0.07

49.16o

2.62

25.37

0.04

25.17

0.10

54.72o

2.91

25.79

0.05

25.63

0.16

11

Table 5. (continued)

distance

distance

6660A

error(6660A)

8020A

error(8020A)

(arcscc)

(kpc)

(mag/sq arcsec)

(mag/sq arcsec)

(mag/sq arcsec)

(mag/sq arcsec)

58.40o

3.11

25.98

0.06

25.65

0.16

64.04o

3.41

26.17

0.08

25.92

0.21

71.73o

3.82

26.33

0.09

26.19

0.27

75.67o

4.03

26.51

0.10

26.51

0.37

81.74o

4.35

26.69

0.11

27.44

0.81

87.89o

4.68

26.73

0.11

27.47

0.82

94.22o

5.01

27.01

0.14

27.64

0.91

100.89o

5.37

27.32

0.19

107.82o

5.74

27.57

0.23

115.00o

6.12

27.64

0.24

122.69o

6.53

27.68

0.25

27.54

0.75

130.81o

6.96

27.68

0.24

27.25

0.57

139.62o

7.43

27.64

0.23

27.40

0.65

149.03o

7.93

27.80

0.27

27.55

0.73

159.29

8.48

27.95

0.30

166.64

8.87

28.04

0.33

176.30

9.38

28.14

0.36

27.77

0.86

185.71

9.88

28.19

0.37

27.77

0.86

192.63

10.25

28.25

0.40

27.94

1.01

199.39

10.61

28.25

0.40

27.85

0.93

206.23

10.97

28.34

0.43

27.85

0.93

213.07

11.34

28.52

0.51

27.88

0.95

219.91

11.70

28.64

0.57

227.60

12.11

28.71

0.60

234.44

12.48

28.76

0.63

242.14

12.89

28.93

0.74

27.73

0.81

249.35

13.27

29.02

0.80

27.65

0.76

Cut Dl, r

= 12.3 kpc

-105.59

-5.62

28.58

0.58

-98.41

-5.24

28.51

0.55

-91.66

-4.88

28.38

0.50

-85.07

-4.53

27.90

0.33

-78.75

-4.19

27.69

0.27

-72.67

-3.87

27.67

0.27

-66.69

-3.55

27.29

0.20

27.49

0.85

-62.84

-3.34

26.82

0.13

26.97

0.53

-57.03

-3.03

26.55

0.11

27.33

0.80

-51.39

-2.73

26.15

0.08

25.79

0.19

-45.83

-2.44

25.70

0.05

25.44

0.13

-40.36

-2.15

25.25

0.03

25.00

0.10

-33.34

-1.77

24.74

0.02

24.33

0.06

-28.22

-1.50

24.31

0.02

23.99

0.04

-26.50

-1.41

24.01

0.01

23.85

0.04

-24.79

-1.32

24.05

0.01

23.65

0.03

-23.09

-1.23

23.85

0.01

23.62

0.03

-21.38

-1.14

23.78

0.01

23.63

0.03

-19.66

-1.05

23.62

0.01

23.70

0.03

12

Table 5. (continued)

distance

distance

6660A

error(6660A)

8020A

error(8020A)

(arcscc)

(kpc) (mag/sq arcscc)

(mag/sq arcscc)

(mag/sq arcscc)

(mag/sq arcscc)

-15.34

-0.82

23.50

0.01

22.26

0.01

20.57o

1.09

22.84

0.01

22.56

0.01

23.99o

1.28

23.29

0.01

23.02

0.02

26.50o

1.41

23.85

0.01

23.76

0.03

28.22o

1.50

24.10

0.01

24.03

0.04

33.34o

1.77

24.47

0.02

24.49

0.06

36.76o

1.96

24.94

0.03

24.70

0.08

41.89o

2.23

25.59

0.05

25.39

0.15

48.73o

2.59

26.03

0.08

25.68

0.19

52.16o

2.78

26.30

0.10

25.75

0.21

57.54o

3.06

26.80

0.16

26.34

0.36

64.98o

3.46

26.75

0.14

27.28

0.79

70.54o

3.75

26.82

0.15

26.80

0.49

74.39o

3.96

27.22

0.20

27.06

0.57

80.11<>

4.26

27.21

0.19

26.78

0.43

86.OI0

4.58

27.07

0.16

26.76

0.42

92.08o

4.90

27.20

0.18

27.28

0.67

98.33o

5.23

27.35

0.20

102.60o

5.46

27.53

0.23

109.27O

5.81

27.55

0.23

116.19o

6.18

27.56

0.23

123.38o

6.57

27.78

0.28

131.07O

6.98

27.97

0.32

139.19o

7.41

27.99

0.33

27.85

0.97

148.00o

7.88

27.99

0.32

27.93

1.02

157.41o

8.38

28.02

0.33

167.67o

8.92

28.04

0.33

27.73

0.83

178.87o

9.52

28.05

0.33

27.48

0.65

189.81o

10.10

28.11

0.35

27.42

0.61

199.22o

10.60

28.17

0.37

27.36

0.58

207.76o

11.06

28.21

0.38

27.59

0.71

216.31o

11.51

28.30

0.41

27.67

0.76

226.57o

12.06

28.32

0.42

27.66

0.76

235.98o

12.56

28.40

0.45

27.71

0.79

245.39o

13.06

28.45

0.48

27.75

0.82

253.93o

13.51

28.49

0.50

27.67

0.76

263.34o

14.01

28.55

0.53

27.68

0.77

271.89o

14.47

28.62

0.58

280.44o

14.92

28.64

0.60

Cut D2, r

= -12.3 kpc

-212.22,

-14.49

28.16

1.18

-261.97

-13.94

28.31

1.36

-250.86

-13.35

28.08

1.10

-240.60

-12.80

28.09

1.10

-229.48

-12.21

28.03

1.05

-220.08

-11.71

28.21

1.25

13

Table 5. (continued)

distance

distance

6660A

error(6660A)

8020A

error(8020A)

(arcscc)

(kpc)

(mag/sq arcsec)

(mag/sq arcsec)

(mag/sq arcsec)

(mag/sq arcsec)

-209.82

-11.17

27.84

0.89

-197.85

-10.53

27.60

0.73

-185.88

-9.89

27.61

0.74

-176.16

-9.37

27.96

0.38

27.14

0.49

-164.54

-8.76

28.47

0.64

26.93

0.41

-154.02

-8.20

28.54

0.68

26.97

0.45

-144.27

-7.68

28.48

0.63

27.18

0.55

-135.38

-7.20

28.56

0.67

27.42

0.68

-127.00

-6.76

27.97

0.39

27.56

0.78

-119.22

-6.34

27.80

0.33

-114.26

-6.08

27.52

0.25

-109.47

-5.83

27.33

0.21

-104.77

-5.58

27.63

0.28

27.09

0.53

-98.10

-5.22

27.65

0.27

27.12

0.56

-91.60

-4.87

27.52

0.24

27.20

0.60

-85.36

-4.54

27.23

0.18

-79.29

-4.22

27.13

0.17

26.93

0.48

-73.39

-3.91

27.09

0.16

26.70

0.41

-67.66

-3.60

26.94

0.15

26.54

0.36

-61.94

-3.30

26.62

0.11

26.16

0.26

-56.38

-3.00

26.42

0.10

26.07

0.24

-49.03

-2.61

26.07

0.07

26.05

0.24

-43.72

-2.33

25.54

0.05

25.43

0.14

-40.30

-2.14

25.13

0.03

24.92

0.09

-35.17

-1.87

24.59

0.02

24.63

0.07

-31.75

-1.69

24.17

0.01

24.03

0.04

-28.33

-1.51

23.89

0.01

23.71

0.03

-26.62

-1.42

23.68

0.01

23.51

0.03

-23.20

-1.23

23.56

0.01

23.25

0.02

-12.09

-0.64

22.79

0.01

23.02

0.02

-6.96

-0.37

22.45

0.01

22.83

0.02

-5.25

-0.28

22.45

0.01

22.60

0.02

1.59o

0.08

22.22

0.01

22.49

0.01

8.43o

0.45

22.60

0.01

22.80

0.02

12.71o

0.68

22.90

0.01

23.08

0.02

16.13o

0.86

23.18

0.01

23.22

0.02

22.97o

1.22

23.46

0.01

23.73

0.03

28.10o

1.50

23.89

0.01

23.85

0.03

29.81o

1.59

24.02

0.01

24.08

0.04

31.52o

1.68

24.32

0.01

24.32

0.05

36.65o

1.95

24.92

0.02

24.73

0.08

40.07o

2.13

25.32

0.04

24.88

0.09

47.08o

2.51

25.81

0.06

25.48

0.15

50.75o

2.70

26.14

0.08

25.46

0.14

58.Ho

3.09

26.26

0.08

26.18

0.28

65.72o

3.50

26.34

0.09

26.08

0.24

71.44o

3.80

26.62

0.11

26.20

0.26

14

Table 5. (continued)

distance

distance

6660A

error(6660A)

8020A

error(8020A)

(arcsec)

(kpc)

(mag/sq arcsec)

(mag/sq arcsec)

(mag/sq arcsec)

(mag/sq arcsec)

77.34o

4.12

26.77

0.13

26.60

0.39

83.41o

4.44

26.70

0.12

27.35

0.78

87.60o

4.66

27.02

0.17

93.93o

5.00

27.21

0.19

IOO.6O0

5.35

27.12

0.17

107.52o

5.72

27.26

0.18

27.76

0.98

114.71o

6.10

27.35

0.19

27.73

0.97

122.40o

6.51

27.41

0.20

130.52o

6.95

27.35

0.18

27.67

0.84

139.33o

7.41

27.40

0.19

27.68

0.92

148.74o

7.92

27.43

0.19

159.00o

8.46

27.53

0.21

170.20o

9.06

27.61

0.22

178. 58o

9.50

27.73

0.25

187.13o

9.96

27.77

0.25

197.39o

10.50

27.80

0.26

205.94O

10.96

27.75

0.25

214.49o

11.41

27.73

0.24

27.98

1.02

223.04O

11.87

27.69

0.23

27.76

0.83

232.44o

12.37

27.70

0.23

27.30

0.55

240. 14o

12.78

27.64

0.22

27.25

0.51

246.98o

13.14

27.62

0.22

27.22

0.50

254.67o

13.55

27.56

0.21

27.23

0.51

263. 22o

14.01

27.55

0.20

27.14

0.47

271.77o

14.46

27.52

0.20

a The data marked with • are affected by foreground star contamination and a possible
line-of-sight warp.

b The data marked with o are affected by the ring.

15

Table 6. Morrison et al. surface brightness profiles of NGC 5907 in R band
distance distance R high error low error

(arcsec) (kpc) (mag/sq arcscc) (mag/sq arcscc) (mag/sq arcsec)

Cut A, r = kpc

11.55

0.61

19.99

0.00

0.00

15.79

0.84

20.51

0.00

0.00

19.69

1.05

21.08

0.00

0.00

23.81

1.27

21.69

0.00

0.00

27.00

1.44

22.13

0.00

0.00

30.82

1.64

22.62

0.01

0.01

34.65

1.84

23.09

0.01

0.01

38.46

2.05

23.53

0.01

0.01

42.25

2.25

23.95

0.02

0.02

46.19

2.46

24.35

0.03

0.03

50.02

2.66

24.69

0.04

0.04

53.90

2.87

24.98

0.05

0.06

57.76

3.08

25.20

0.06

0.07

62.30

3.32

25.50

0.11

0.13

67.73

3.61

25.74

0.14

0.16

74.04

3.94

25.76

0.10

0.11

82.59

4.40

25.92

0.11

0.13

94.78

5.05

26.37

0.17

0.20

123.88

6.60

26.95

0.27

0.36

183.67

9.78

27.20

0.33

0.47

260.55

13.87

27.84

0.53

1.10

334.53

17.81

28.00

0.60

1.44

410.94

21.88

28.28

0.73

3.36

466.15

24.82

33.40

4.33

0.00

Cut B, r =

= 4.1 kpc

5.78

0.31

19.92

0.00

0.00

8.23

0.44

20.11

0.00

0.00

12.19

0.65

20.65

0.00

0.00

15.97

0.85

21.05

0.00

0.00

20.31

1.08

21.60

0.07

0.07

23.49

1.25

22.02

0.03

0.03

27.09

1.44

22.46

0.05

0.05

30.80

1.64

22.91

0.04

0.04

34.65

1.84

23.36

0.06

0.06

38.51

2.05

23.79

0.11

0.12

42.38

2.26

24.20

0.12

0.14

46.17

2.46

24.60

0.13

0.14

50.05

2.67

24.87

0.14

0.16

53.94

2.87

25.21

0.14

0.16

57.74

3.07

25.43

0.18

0.21

62.26

3.32

25.69

0.21

0.26

67.82

3.61

25.98

0.17

0.20

73.92

3.94

26.17

0.19

0.24

82.27

4.38

26.30

0.27

0.35

94.99

5.06

26.54

0.37

0.56

125.02

6.66

27.01

0.46

0.83

16

Table 6. (continued)

distance

distance

R

high error

low error

(arcsec)

(kpc)

(mag/sq arcsec)

(mag/sq arcsec)

(mag/sq arcsec)

183.68

9.78

27.43

0.42

0.70

262.36

13.97

27.96

0.72

3.17

333.43

17.75

27.88

0.59

1.37

415.87

22.14

28.57

0.73

3.30

466.49

24.84

33.40

5.15

0.00

Cut C, r =

= 8.2 kpc

20.82

1.11

21.84

0.01

0.01

23.55

1.25

22.42

0.19

0.23

26.97

1.44

22.88

0.14

0.17

30.81

1.64

23.35

0.11

0.12

34.66

1.85

23.81

0.07

0.08

38.55

2.05

24.23

0.08

0.09

42.34

2.25

24.58

0.12

0.13

46.21

2.46

24.92

0.15

0.17

50.04

2.66

25.17

0.13

0.15

53.92

2.87

25.47

0.13

0.15

57.74

3.07

25.62

0.16

0.19

62.32

3.32

25.85

0.15

0.17

67.74

3.61

26.02

0.09

0.10

73.93

3.94

26.17

0.10

0.12

82.39

4.39

26.36

0.15

0.18

94.98

5.06

26.61

0.22

0.27

123.47

6.57

26.83

0.18

0.21

184.48

9.82

27.24

0.53

1.07

261.75

13.94

28.40

1.05

5.00

331.70

17.66

27.92

0.78

5.48

413.16

22.00

27.96

0.78

5.44

462.21

24.61

28.05

0.66

1.98

Cut D, r --

= 12.3 kpc

16.56

0.88

22.34

0.11

0.12

19.49

1.04

22.70

0.17

0.21

23.33

1.24

23.16

0.09

0.10

26.98

1.44

23.62

0.07

0.07

30.84

1.64

23.98

0.06

0.06

34.63

1.84

24.39

0.03

0.03

38.54

2.05

24.73

0.09

0.10

42.38

2.26

25.09

0.08

0.08

46.24

2.46

25.40

0.12

0.14

49.95

2.66

25.69

0.09

0.09

53.92

2.87

25.88

0.21

0.25

57.74

3.07

26.04

0.25

0.33

62.52

3.33

26.19

0.20

0.24

67.78

3.61

26.33

0.25

0.32

73.85

3.93

26.39

0.34

0.49

82.73

4.41

26.68

0.34

0.51

94.64

5.04

26.93

0.34

0.49

17

Table 6. (continued)

distance

distance

R

high error

low error

(arcscc)

(kpc)

(mag/sq arcsec)

(mag/sq arcsec)

(mag/sq arcsec)

124.14

6.61

27.28

0.57

1.25

185.07

9.85

27.51

0.69

2.31

259.71

13.83

27.93

0.87

5.47

337.33

17.96

27.75

0.89

5.65

403.18

21.47

27.80

0.87

5.60

469.57

25.00

28.40

0.96

5.00

Table 7. Comparisons to surface brightness profiles from other observers
Position 6660A 8020A J K, R V I

145 to -120

27.80

27.22

25.64

24.48

-

-

-

120 to -100

27.15

26.58

25.07

23.68

26.72

-

-

-100 to -80

26.64

26.35

24.59

23.25

26.31

27.58

26.36

-80 to -70

26.28

26.00

24.32

23.02

25.96

27.00

25.67

-70 to -60

25.82

25.53

24.16

22.85

25.71

26.50

25.29

-60 to -50

25.31

24.78

23.64

22.80

25.02

25.95

24.87

-50 to -40

24.58

24.17

22.80

22.04

24.15

25.04

24.09

-40 to -30

24.09

23.20

21.50

20.82

23.04

23.85

22.86

-30 to -20

-

22.40

20.05

19.34

21.79

22.63

21.59

-20 to -10

10 to 20

-

21.47

18.63

17.62

20.56

21.65

20.45

_

_

18.74

17.71

20.56

20.92

19.62

20 to 30

-

22.37

20.40

19.40

21.79

22.41

21.10

30 to 40

-

23.18

21.63

20.68

23.04

23.77

22.79

40 to 50

24.71

24.11

22.48

21.84

24.15

25.06

24.15

50 to 60

25.52

24.75

23.11

22.31

25.02

26.04

25.05

60 to 70

25.91

25.24

23.32

22.74

25.71

26.68

25.60

70 to 80

26.19

25.59

23.39

22.95

25.96

27.15

25.99

80 to 100

26.61

26.57

23.76

23.13

26.31

27.67

26.52

100 to 120

27.06

27.23

24.10

22.69

26.72

28.30

27.08

120 to 145

27.60

-

25.38

24.88

-

-

27.08

18