Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://zebu.uoregon.edu/postscript/p1522.ps Äàòà èçìåíåíèÿ: Fri Aug 5 05:43:59 1994 Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 22:29:19 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: http astrokuban.info astrokuban

Surface photometry of low surface brightness galaxies 15
Table 3. Colour gradients per B­scale length ff B in LSB galaxies
Name \Delta(U --B) error \Delta(B--V ) error \Delta(B--R) error \Delta(V --I) error
per ff B per ff B per ff B per ff B
F561­1 \Gamma0:30 0:14 \Gamma0:11 0:08 \Gamma0:19 0:08 \Gamma0:14 0:16
F563­1 \Gamma1:15 0:22 \Gamma0:06 0:13 \Gamma0:16 0:13 \Gamma \Gamma
F563­V1 \Gamma0:09 0:11 +0:04 0:07 \Gamma \Gamma \Gamma0:07 0:13
F564­V3 \Gamma \Gamma \Gamma0:05 0:07 \Gamma0:10 0:06 \Gamma \Gamma
F565­V2 \Gamma \Gamma +0:12 0:08 +0:16 0:06 \Gamma \Gamma
F567­2 \Gamma0:21 0:17 \Gamma0:09 0:13 \Gamma0:47 0:13 \Gamma0:47 0:22
F568­1 \Gamma0:12 a 0:16 \Gamma0:12 0:08 \Gamma0:28 0:08 \Gamma0:24 0:16
F568­3 \Gamma0:12 0:12 \Gamma0:09 0:06 \Gamma0:15 0:06 \Gamma0:12 0:15
F568­V1 \Gamma0:07 b 0:10 \Gamma0:07 0:05 \Gamma0:12 0:05 \Gamma0:10 0:10
F571­5 \Gamma0:09 0:11 \Gamma0:18 0:07 \Gamma0:20 0:07 \Gamma \Gamma
F571­V1 \Gamma \Gamma \Gamma0:05 0:05 \Gamma0:10 0:05 \Gamma \Gamma
F574­2 \Gamma \Gamma \Gamma0:18 0:09 \Gamma0:32 0:09 \Gamma \Gamma
F577­V1 \Gamma0:17 0:13 \Gamma0:09 0:09 \Gamma0:26 0:09 \Gamma \Gamma
U0128 \Gamma0:21 0:35 \Gamma0:21 0:21 \Gamma0:35 0:21 \Gamma0:21 0:35
U0628 \Gamma0:16 0:16 \Gamma0:12 0:12 \Gamma0:16 0:08 \Gamma0:16 0:2
U1230 \Gamma0:20 0:20 \Gamma0:12 0:16 \Gamma0:24 0:12 \Gamma0:12 0:24
a Points with r ! 4 00 are excluded.
b Points with r ! 5 00 are excluded.
bright and luminous parts of a galaxy, while for the area
weighted colour this is compensated by weighing it with the
area. A small, bright region with only a small area will give a
much smaller contribution towards the area weighted colour
than to the luminosity weighted colour. The area weighted
colour is therefore the best indicator for the underlying
colours of the disk, while the luminosity weighted colour
is best for comparison with colours derived using aperture
photometry. However, as LSB galaxies do not contain ex­
tremely bright HII regions we do not expect area weighted
and luminosity weighted colours to differ dramatically.
In fact, the difference between area and luminosity
weighted colours is, except perhaps for U \Gamma B, almost com­
pletely determined by the relative strength of the redder in­
ner parts with respect to the bluer outer parts; i.e., there is a
relation between the redness (and brightness) of the central
parts and the difference between area weighted and lumi­
nosity weighted colour. This is important even in galaxies
that show no appreciable bulge, as a pure exponential disk
already contains half of its light within 1.68 scale lengths.
Only in U \Gamma B are the effects of e.g. HII regions apparent
for some galaxies.
Table 4 shows these colours corrected for Galactic ex­
tinction (see Section 2.1) Errors in these colours are ¸ 0:1
magnitudes (based on the errors in the sky subtraction of
the parent images). Column 1 gives the name of the galaxy.
Columns 2, 3, 4 give the nuclear, luminosity weighted and
area weighted U \Gamma B colours, respectively. Columns 5, 6,
7 give the nuclear, luminosity weighted and area weighted
B \Gamma V colours. Columns 8, 9, 10 give the nuclear, luminos­
ity weighted and area weighted B \Gamma R colours. Columns 11,
12, 13 the nuclear, luminosity weighted and area weighted
V \Gamma I colours. Column 14 gives the maximum radius in arc­
seconds that was used to determine the luminosity weighted
and area weighted colours, and is therefore the radius that
corresponds roughly with the 25.5 B­mag=ut 00 isophote.
In Fig. 7 we show the area weighted B \Gamma V colours and
structural parameters of the disks with respect to each other.
Figure 6. a Distribution of B \Gamma V colour as a function of scale
length. The black dots represent our sample, the open squares
McGaugh's sample. Common galaxies are connected. The cross
shows the typical errors. b Distribution of B \Gamma V colour as a func­
tion of central surface brightness. The black dots represent our
sample, the open squares McGaugh's sample. Common galaxies
are connected

16 W.J.G. de Blok, J.M. van der Hulst and G.D. Bothun
Table 4. Colours of LSB galaxies
Name U \Gamma B B \Gamma V B \Gamma R V \Gamma I rmax
nuc lum area nuc lum area nuc lum area nuc lum area ( 00 )
F561­1 0.04 ­0.04 ­0.20 0.62 0.55 0.49 0.98 0.82 0.70 0.79 0.76 0.66 22
F563­1 0.23 0.07 ­0.05 0.67 0.64 0.58 1.03 0.96 0.84 16
F563­V1 0.05 ­0.01 ­0.05 0.55 0.59 0.59 0.71 0.71 0.69 15
F564­V3 0.57 0.57 0.56 0.88 0.87 0.80 16
F565­V2 0.52 0.51 0.49 0.83 0.82 0.81 15
F567­2 0.05 ­0.05 ­0.14 0.69 0.61 0.57 1.04 0.91 0.77 0.71 0.73 0.62 16
F568­1 0.21 ­0.09 ­0.20 0.71 0.58 0.53 1.14 0.95 0.80 0.86 0.88 0.76 24
F568­3 0.06 ­0.04 ­0.11 0.70 0.61 0.54 1.10 0.94 0.81 0.88 0.84 0.72 23
F568­V1 0.11 ­0.16 ­0.15 0.61 0.57 0.56 1.03 0.91 0.82 0.84 0.77 0.67 19
F571­5 ­0.06 ­0.12 ­0.13 0.56 0.54 0.44 0.87 0.85 0.74 13
F571­V1 0.58 0.55 0.51 0.93 0.89 0.78 17
F574­2 0.63 0.59 0.51 0.96 0.86 0.70 22
F577­V1 ­0.12 ­0.19 ­0.40 0.54 0.50 0.40 0.87 0.77 0.58 17
UGC 0128 0.36 0.16 0.02 0.73 0.60 0.49 1.16 0.90 0.73 0.85 0.68 0.55 48
UGC 0628 0.19 0.04 ­0.07 0.75 0.61 0.53 1.17 0.98 0.86 0.97 0.89 0.78 40
UGC 1230 0.01 ­0.19 ­0.30 0.63 0.54 0.47 0.98 0.89 0.76 0.92 0.86 0.77 28
Figure 7. B \Gamma R colours of HSB and LSB galaxies as a function
of effective surface brightness. The triangles represent our sample
of LSB galaxies. The squares and errorbars represent the mean
values and 1oe deviations for all Sa, Sb, Sc, and Sd galaxies in our
ESO­LV sample
In these figures values found by McGaugh (1992) for other
LSB galaxies are also plotted. Among the LSB galaxies there
is no trend of colour with surface brightness, although LSB
galaxies are bluer than the HSB galaxies. The B \Gamma V and
B \Gamma R colours do not vary much with surface brightness. The
U \Gamma B values show a large scatter. U \Gamma B is very sensitive
to recent star formation, so this scatter is perhaps not un­
expected, given the diverse star formation histories present
in the sample. The colours do not depend on the size (scale
length) of the galaxy as Fig. 7 shows.
In Fig. 8 we have plotted the average B \Gamma R colours of
all Sa, Sb, Sc and Sd galaxies in our ESO­LV sample (see
Section 3.4) versus the mean effective surface brightness ¯ e
B .
The open squares represent the mean values for each Hubble
type, the errorbars are 1oe errorbars. It is apparent that LSB
galaxies, which are represented by the triangles, are lower in
surface brightness and also bluer in B \Gamma R. It is significant
that despite the fact that LSB galaxies resemble normal late
type galaxies, they are still fainter and bluer. Characteris­
tics such as colour and surface brightness do therefore not
correspond to their Hubble type. This strongly suggests that
the Hubble type does not have a one­to­one correspondance
with the star formation history, as evidently the star forma­
tion history of a late type LSB galaxy must differ from that
of a late type HSB galaxy.
If colour is plotted against colour then most of the
colour combinations show no trend among the galaxies in
our sample, confirming that we are dealing with LSB galax­
ies that each have their own star formation history. However
the colours of the group as a whole are still bluer than those
of the average late type galaxy with corresponding Hubble
type, which means that the star formation history of LSB
galaxies as a group is different from that of late type HSB
galaxies.
5 INDIVIDUAL GALAXIES
We will discuss some interesting galaxies in our sample in­
dividually. Values for colours and gradients are taken from
Tables 3 and 4. When discussing colours, any effects inter­
nal extinction may have are ignored. McGaugh & Bothun
(1993) and McGaugh (1992) show that the dust content of
these galaxies is probably low. We will sometimes refer to
unpublished Hff images obtained by us during the same run
as the broadband images. These will be discussed in a future
paper.
F561­1 shows no clear spiral structure, but instead shows
a ring of bright knots, that are very probably star forming
regions. The ring can be seen as a bump in the radial pro­
files at r ¸ 15 00 . It is not very apparent in the colour profiles,

Surface photometry of low surface brightness galaxies 17
except perhaps for a small blueing in V \Gamma I at that radius.
The colour gradient is steepest for U \Gamma B. This fact, together
with the central reddening in V \Gamma I (which most likely is not
a seeing effect), seems to suggest that star formation may
have started in the central parts, and has now progressed
to the outer parts (the ring of star forming regions) while it
has declined in the center. F561­1 is one of the bluer galax­
ies in the sample, so we are probably seeing it in a phase
of slightly enhanced star formation. Alternatively, the opti­
cal image shows a small galaxy nearby at ¸ 2 0 separation.
It is not known whether this small elliptical galaxy is at
the same distance as F561­1. If it is then it may very well
have caused this current star formation process (Appleton
& Struck­Marcell 1987).
F563­1 has a peculiar morphology, dominated by a very
bright region to the west of the center. This region is part of
a partial ring of star forming regions. It shows no clear spi­
ral structure, and fades out to the east of the center, while
the light cut­off is rather sharp towards the star formation
regions. This suggests that the gas in this part of the galaxy
may be more compressed, thus giving rise to the star for­
mation. The star formation regions are only apparent as a
bump in the U profile at r = 10 00 , and not in the colour
profiles, except for the slight blueing in the U \Gamma B profile.
In this profile we also find the strongest gradient of the en­
tire sample. It can be caused by the same scenario as in
F561­1: star formation has started in the center, and has
now progressed outwards to the partial ring of star forming
regions. Careful examination of the optical image and the
available WSRT Hi imaging, show that this galaxy may be
much larger than apparent at first sight. The radio diameter
is about a factor of three larger than those of LSB galaxies
with comparable linear optical sizes. Close examination of
the optical image shows that optical emission can be traced
out to ¸ 1 0 from the center. The isolated object 40 00 to the
SW may in fact lie at the tip of a very faint spiral arm, and
the entire galaxy may be surrounded by a very LSB spiral
disk, as suggested by the large ratio of HI to optical size.
Deep optical follow­up observations are needed.
F563­V1 is an amorphous galaxy with no spiral structure
visible. The radial profiles show it to be an almost perfect ex­
ponential disk. There are no star formation complexes in this
galaxy, except for one star formation region to the south of
the center. The gradients in this galaxy are very small, sug­
gesting that here no progression of star formation has taken
place, but that except for the one star formation region, the
disk is a homogeneous mix of different stellar populations.
F564­V3 is amorphous and diffuse. It shows no structure,
apart from a general brightening towards the center. The
colour gradients per arcsecond are very small, but its surface
brightness profiles show it to be not a perfectly exponential
disk. The radial profiles all flatten towards the center. It is
the only galaxy in our sample where a significant flattening
of the profile near the center occurs. Its Hi redshift is only
500 km/s (Schombert & Bothun 1988), placing it at a dis­
tance of ¸ 6 Mpc. Assuming that this distance is correct,
makes it the smallest galaxy in our sample, with a scale
length of only 0.3 kpc. This could mean that F564­V3 is a
true dwarf galaxy.
F568­1 has two strong, and well developed spiral arms,
and is the only one of the F­galaxies to a stronger bulge.
This is especially clear in the U \Gamma B profile, which shows a
central reddening of almost 0.5 magnitudes. The spiral arms
are not formed or traced by star formation regions. Rather it
is just the general stellar distribution that shapes the arms.
6 DISCUSSION OF THE COLOURS
6.1 Median colours
We will discuss the colours of LSB galaxies, and compare
them with evolutionary models from the literature, with the
goal of tracing a possible star formation history for these
objects. Where possible we will supplement our data with
those of McGaugh (1992), who has used the same selection
criteria for his sample as we have.
In the following we will assume that the median area
weighted colours of our and McGaugh's sample combined
are representative for the sample as a whole. This is not a
very good assumption, as a large range of colours is present,
but our goal here is to look at our sample as a group.
The median area weighted colours of the combined sam­
ple are U \Gamma B = \Gamma0:13, B \Gamma V = 0:51, B \Gamma R = 0:78, and
V \Gamma I = 0:76. These colours are also given in Table 5. Col­
umn 1 gives the colour. Column 2 the median colour of our
sample. Column 3 the median colours of McGaugh's sam­
ple, where McGaugh has not measured B \Gamma R. Column 4
gives the colours of the combined sample, as shown above.
The large difference in V \Gamma I between our sample and Mc­
Gaugh's is immediately apparent. For the galaxy F563­V1,
which is common to both samples, we see a large difference
in the colours as determined by McGaugh and us. According
to McGaugh (private communication) this may be due to a
defect in his V ­image of this galaxy. This galaxy is there­
fore excluded from any discussion about McGaugh's sample.
More of these effects for other galaxies seem to be ruled out,
especially since the galaxies that have high V \Gamma I colours,
do not show any extraordinary B \Gamma V colours. V \Gamma I in Mc­
Gaugh's sample shows more scatter towards the red than
our sample. A likely explanation is that McGaugh's sam­
ple simply contains more galaxies that are redder in V \Gamma I
than our sample. The galaxies which both samples have in
common show that any systematic differences are probably
less than 0.1 mag in V \Gamma I and they can not explain large
scatter towards the red in McGaugh's sample. A comparison
of the photometry of the few galaxies that our sample and
McGaugh's sample have in common shows that any system­
atic differences between the measurements are much smaller
than the typical errors in the colours. Comparisons with in­
dependent observations of UGC 628 (Knezek 1993; de Jong
& van der Kruit, 1994) give differences in the measured B \GammaR
colour profiles of ! 0:1 mag.
6.2 Comparison with other objects
We will compare the ``typical'' colours of LSB galaxies with
those of other objects in order to obtain some clues about
their star formation history.
First of all, we compare our sample with a suitable HSB
galaxy. Morphology shows that very late type HSB galaxies
(Sd and later) are the most likely candidates for compari­
son. These too have small to absent bulges, and loose, less

18 W.J.G. de Blok, J.M. van der Hulst and G.D. Bothun
Table 5. Comparison of colours
Col Our McG Comb HSB
(1) (2) (3) (4) (5)
U \Gamma B \Gamma0:14 \Gamma0:12 \Gamma0:13 \Gamma0:14
B \Gamma V +0:52 +0:44 +0:51 0.51
B \Gamma R +0:78 \Gamma +0:78 0.92
V \Gamma I +0:69 +0:95 +0:76 0.90
(1) colour
(2) our median values
(3) McGaugh's median values
(4) combined median values
(5) colours HSB galaxy
strongly developed spiral arms. There are of course also dif­
ferences, such as the number of Hii regions present, and
thence the amount of star formation.
We have used the RC3 (de Vaucouleurs et al. 1991),
the ESO­LV (Lauberts & Valentijn 1989), and results from
Byun (1992), Huchra (1992) and Han (1992) to determine
the colours of typical HSB Sdm--Sm (comparison) galax­
ies. These colours are U \Gamma B = \Gamma0:14, B \Gamma V = 0:51,
B \Gamma R = 0:92, and V \Gamma I = 0:9 (cf. Fig. 9 and 10). The
value of V \Gamma I is rather uncertain due to the fact that Han's
corrections for extinction are larger than we expect them to
be for these galaxies. Thus a typical HSB late type galaxy
will have uncorrected V \Gamma I colours that are redder than the
colours as given by Han, i.e. in reality V \Gamma I will be larger
than 0.9. This means that these HSB galaxies will also be
redder than the galaxies in McGaugh's sample. The scatter
in B \Gamma R and V \Gamma I is quite large (Han 1992; ESO­LV).
Comparison with Table 5 shows that the U \Gamma B and
B \Gamma V of LSB galaxies are roughly equal to those of late
type HSB galaxies. These colours are in itself surprising.
Blue colours are generally associated with starformation. It
is evident that this is the case in HSB galaxies, which have
many Hii regions. It is surprising that LSB galaxies which
have no or only small traces of starformation, and that gen­
erally look much more quiescent, should have comparable
colours.
In Fig. 9 we illustrate this. Figure 9 shows a UBV di­
agram for the roughly 500 late type galaxies (Sc and later)
for which both U \Gamma B and B \Gamma V were available in the RC3.
We have binned these galaxies per Hubble type, and have
plotted the mean colours and 1oe deviation. It is clear that
there is a trend of colour with Hubble type, in the sense
that galaxies become bluer with later Hubble type. It is also
clear that the spread in colour for a given type is so large
that we cannot use the colour of a galaxy to assign a unique
Hubble type. The stars in Fig. 9 represent the LSB galaxies
both from our and McGaugh's sample. These galaxies have
colours that scatter around the typical colours for Sdm and
Sm galaxies.
Figure 10 shows a BRI diagram. We have plotted B \Gamma I
instead of V \GammaI as Byun (1992) gives B \GammaI colours. The mean
colours and 1oe deviations of the ca. 200 late type galaxies
(Sc and later) for which both colours were available in Byun
(1992) and the ESO­LV are shown, binned per Hubble type.
The stars represent the LSB galaxies. Once again there is a
clear trend of colour with Hubble type, and once again colour
Figure 8. UBV diagram. The filled circles represent the mean
colours for each Hubble type. The error bars denote 1oe deviations
from the mean colours. The stars represent the colours of the LSB
galaxies.
does not correspond unambiguously with Hubble type. It is
clear that in these colours the LSB galaxies are consistently
bluer than Sdm--Sm HSB galaxies.
6.3 Dust and extinction effects
An effect which could cause e.g. Sc galaxies to be redder than
LSB galaxies is the presence of a larger amount of dust. If
one assumes that the mix of stellar populations in LSB and
Sc galaxies is equal, and if one uses the simple absorbing,
overlying screen model for the extinction and reddening, and
ignores any surface brightness effects, an extinction AV ¸
0:5 mag is needed to turn the colours of LSB galaxies into
those of Sc galaxies.
However, the assumption of identical stellar populations
is not likely to be true, in view of the different star forma­
tion histories, as we will discuss later. The assumption of an
overlying screen is naive and too simplistic. As Witt et al.
(1992) and Disney et al. (1989) have shown, the screen model
is not the most realistic model. Reddened stars will become
fainter and more obscured, and thus will contribute less red
light to the total colour, so that the least obscured parts
of a galaxy will determine its colours. Witt et al. reach the
conclusion that ``significant numbers of stars would lie near
the surface of any dusty ISM, producing almost unreddened
starlight and giving the false impression of little dust within
the system.'' So dust may be present in different quantities
in LSB and Sc galaxies. We are however, unable to derive
these quantities using just broadband optical, colours. In
both LSB and Sc galaxies we may therefore expect global
colours to be largely determined by stars that are not or only
lightly obscured by dust. Furthermore, for a correct analysis
of extinction and reddening, one also has to take scattering
into account. In our case, where we study face­on galaxies at

Surface photometry of low surface brightness galaxies 19
Figure 9. BRI diagram. The filled circles represent the mean
colours for each Hubble type. The error bars denote 1oe deviations
from the mean colours. The stars represent the colours of the LSB
galaxies.
optical wavelengths, scattering is particularly important, as
photons that get scattered will have a greater chance of be­
ing scattered out of the disk perpendicularly (in the face­on
direction) than in the edge­on direction (in the plane of the
disk), where they have a larger chance of being absorbed.
The scattered photons that reach us will effectively make
the observed colours bluer again, bringing them closer to
their intrinsic values.
Summarizing we can say that although dust will play
a role in determining colours on local scales (e.g. between
arm and interarm regions in spirals), and perhaps on global
scales, it is certainly not able to explain the difference in
colours between LSB and Sc HSB galaxies entirely. Age and
metallicity effects must also contribute towards the total
colours.
6.4 Metallicity effects: comparison with globular
clusters
McGaugh & Bothun (1993) have shown that LSB galaxies
have low metallicities and it is known that low metallicity
gives rise to bluer colours. To get an idea of the magni­
tude of this effect we can look at metal­poor objects, and
compare their colours with those of the LSB galaxies. We
will not compare the sample with for example blue compact
galaxies: although these objects have low metallicities, the
high star formation rates they exhibit make them unsuitable
for a comparison. Better comparison objects are metal poor
globular clusters.
For metal poor globular clusters ([Fe/H] ' \Gamma2:0) one
finds U \Gamma B = 0:1, B \Gamma V = 0:65, B \Gamma R = 1:1 and V \Gamma I =
0:95 (Reed 1985; Reed et al. 1988; Hesser & Shawl 1985).
These globular clusters are therefore generally redder than
the LSB galaxies. Globular clusters in the Magellanic Clouds
(Elson & Fall 1985) are also redder. These very metal poor
objects show U \Gamma B = 0:2 for a B \Gamma V colour equal to that
of the median B \Gamma V colour of the sample. Globular clusters
are thus much redder than LSB galaxies. As it seems unlikely
that the metallicity of the stellar population in LSB galaxies
is much lower than that of the metal­poor globulars we must
conclude that the blue colours are only partly caused by
metallicity and that age must certainly play a major role. We
can therefore not explain the colours by assuming a single,
old population at low metallicity.
The ambiguity between age and metallicity also effects
the interpretation of the measured colour gradients. The
general redwards to bluewards trend with radius could be
a reflection of higher metallicity in the inner parts or it
could represent a difference in mean age between the in­
ner and outer regions. This is not entirely unreasonable as
LSB galaxies may have long collapse times and thus star for­
mation may proceed first in the inner, denser regions while
the outer regions are still collapsing.
6.5 Age effects
Because the UBV colours are comparable to those of HSB
galaxies, and because starformation is still taking place, we
conclude that at least part of the stellar content of LSB
galaxies consists of young stars. We already noted (section
4.2) that the effects of individual HII regions on the total
colours are smaller than the effects of the colour and bright­
ness gradient between ``bulge'' and disk. The fact that the
measured area weighted colours are still blue suggests that
star formation should be taking place diffusely across the
disk, in order to have an effect on the colours. A simple two
population model consisting of old metal­poor globular clus­
ter stars (see section 6.4) and young B­stars, suggests that
approximately 5 to 10 percent of the area of the disk should
in some way be affected by star formation and have bluer
colours, in order to have a noticeable (¸ 0.1 mag) effect on
the total colours. Judging from colour maps (see e.g. Fig. 6)
5 to 10 percent is not an unreasonable value.
Most of the galaxies in our sample are similar to late
Hubble type galaxies and their properties are just as hetero­
geneous as those of HSB galaxies, with a variety in colour,
surface brightness, morphology and size. Each of these galax­
ies is likely to have its unique star formation history, pre­
sumably different from those of HSB galaxies.
In this section we will not try to decipher the star forma­
tion history and evolution in detail. Rather we will describe
a few possible star formations scenarios that may apply to
LSB galaxies, and discuss their possible merits and short­
comings.
In order to quantify our discussion somewhat we will
sometimes refer to published models of the star formation
histories of HSB galaxies, in particular Searle et al. (1973),
Larson & Tinsley (1978), Guiderdoni & Rocca­Volmerange
(1987) and Mazzei et al. (1992).
The following important fact should be noted about
most of these models. We have already noted that the metal­
licity in LSB galaxies, as measured in HII regions, is about
a fifth of solar metallicity. It is quite likely that the metal­
licity in the older population is less than this value. The
models discussed here were mostly computed assuming so­
lar metallicity. Colours derived from these models will dif­

20 W.J.G. de Blok, J.M. van der Hulst and G.D. Bothun
fer from those derived from consistent models at sub­solar
metallicity. As these models are not available yet, it is there­
fore difficult to make a valid comparison between the solar
metallicity model results and our data, as the lower metal­
licity will result in bluer colours in a non­straight­forward
way.
We will now discuss the possible star formation scenar­
ios.
7 DISCUSSION
7.1 Disk­fading scenario
This scenario would seem to be a very natural explanation
for the nature of LSB galaxies: they are simply assumed to
be the faded remnants of once normal disk galaxies that
for some reason have ceased to form stars a few billion years
ago. We can however use our data as presented in Fig. 7 and
the simple models of Searle et al. to disprove this scenario.
In the absence of star formation a galaxy will become red­
der and fainter with time (regardless of metallicity), as the
short­lived blue stars die. From Searle et al. we can derive a
reddening of 0.22 mag in B \Gamma V for each magnitude fading in
B. If this scenario were correct Fig. 7 should have shown at
least a clear reddening with decreasing surface brightness.
LSB galaxies are not the faded remnants of normal HSB
galaxies.
7.2 Initial starburst with cutoff
An initial starburst with subsequent cutoff in star formation
is equivalent to the disk fading scenario, and is therefore not
likely to explain the observed colours. Indeed, for a model
which has an initial starburst with cutoff after 10 7 years
Larson & Tinsley (1978) find colours that are too red to fit
our measured colours.
7.3 Exponentially declining star formation rate
The exponentially declining star formation rate (SFR) has
very frequently been used as the ``standard'' star formation
history for spiral galaxies. This scenario can give us the low
SFR that we measure in LSB galaxies.
The colours at the blue side of the spectrum, U \Gamma B and
B \Gamma V , are very similar to those of HSB galaxies, as noted in
previous sections. This is understandable, as these colours
are sensitive to recent star formation, and this recent star
formation, is, albeit in different quantities, present both in
HSB and LSB galaxies. So to see whether the exponentially
declining SFR is valid we have to look at the presence or
absence of a large old, red population, as the exponential
nature this star formation history implies that the amount
of stars formed in the past was much greater, and conse­
quently any galaxy that has undergone an exponential star
formation history, must have a large population of old stars.
The bulk of these stars will have their peak optical emission
at the red side of the spectrum, i.e. in the R and I bands.
These colours, in our case B \Gamma R and V \Gamma I, are much
bluer than those of HSB galaxies. This probably means that
the old population in LSB galaxies is less strongly devel­
oped than in HSB galaxies which have redder colours (i.e. a
surplus of red light, and thus a large old population).
V \Gamma I furthermore is an indicator of the position and de­
gree of development of the giant branch in the HR­diagram
(Bothun 1984) and therefore metallicity dependent, as the
position of the giant branch changes with metallicity. The
values found for V \Gamma I thus also suggest low metallicity and
no early enrichment of the ISM. Observations in the near in­
frared combined with measurements of the metallicities are
needed to deny or confirm the presence of a large old pop­
ulation. LSB galaxies have not undergone an exponentially
declining star formation history.
7.4 Constant star formation rate
A constant SFR is also excluded for the same reasons as
above: it can more or less reproduce the measured U \Gamma B
and B \GammaV colours (Larson & Tinsley 1978), but only because
there is some recent star formation present in LSB galaxies.
A constant SFR also implies that by now a large old popu­
lation should have announced its presence in the B \Gamma R and
V \Gamma I colours. A further clue can be found in the models of
Mazzei et al. (1992). These models take some chemical evo­
lution into account, and the poorly known evolution of stars
in their giant phases. One finds that with these models one
can only approximate the U \Gamma B and B \Gamma V colours with
models with an assumed age of 10 Gyr, while the B \Gamma R
and V \Gamma I can only be modeled using and age between 2
and 5 Gyr. Although the absolute value of the colours will
probably not be right due to the differing metallicities, the
trend in age suggests we need to invoke a ``young'' old popu­
lation (i.e. less developed) to explain the differences between
the red and the blue colours [cf. the constant SFR models of
Guiderdoni & Rocca­Volmerange (1987): their 9.5 Gyr mod­
els give a better approximation to the colours than their 15.5
Gyr models].
7.5 Sporadic star formation rate
The lack of clear signs of an old population, combined with
the fact that we still find evidence for young stars in LSB
galaxies, suggests the following scenario for the star forma­
tion history of LSB galaxies. As the current metallicity is
low, it is improbable that there has been a large amount
of star formation in the past, be it burst­like or not. The
blue U \Gamma B and B \Gamma V suggest that some high­mass stars
are present at the moment. We can reach this stage by su­
perimposing small surges in the SFR, either superimposed
on a very low constant SFR, or not. Where this temporary
increase occurs is largely determined by the critical density
for star formation in Hi [see discussion in van der Hulst et
al. (1993)].
Salzer et al. (1991) show that the younger a blue popu­
lation is, the smaller the ratio of blue stars and red stars can
be in order to produce blue colours. So, if indeed, a small in­
crease in the SFR leads to starformation taking place at this
moment, then only a limited number of young stars is needed
to ``hide'' at least part of the old population by making the
colours blue. There is however still a limit to the number of
young stars that a LSB disk can contain without becoming

Surface photometry of low surface brightness galaxies 21
a HSB disk. There is therefore a limit to the amount of old
stars that can be ``hidden'' by the young population.
7.6 Summary
We can summarize the discussion as follows. The starforma­
tion scenarios suggest that the colours can be explained by
assuming a less developed old population, and a more lu­
minosity dominant blue population. The LSB galaxies had
a low (sporadic) SFR in the past, so that less stars are
formed. Recent star formation made U \Gamma B and B \Gamma V bluer
with respect to B \Gamma R and V \Gamma I. The low metallicities in
these galaxies will furthermore make the colours still bluer.
Star formation has commenced relatively recently and an
old population has not yet had time to develop, or is out­
shone by young, blue stars. But LSB disks can contain only
a limited amount of bright blue stars without becoming a
HSB disk.
We can speculate that the LSB galaxies in our sam­
ple are ones that are at the moment going through a phase
of slightly enhanced star formation. One of the problems
created by assuming that this extra star formation causes
the LSB galaxies to be as blue as they are, is that there
must then necessarily also be red LSB galaxies that are qui­
escent and not undergoing a burst. A possible reason why
these galaxies are not in our sample is that the LSB galaxy
catalogs by Schombert & Bothun (1988) and Schombert et
al. (1992) were composed after visual inspection of the blue
POSS­2 plates. Red LSB galaxies may therefore have been
missed which would serve to increase the overall space den­
sity of LSB galaxies.
An alternative reason, but less likely, may be that red
LSB galaxies simply do not exist. In that case the present
state of the LSB galaxies in our sample is typical for all LSB
galaxies. But one then needs to explain why these galaxies
have all recently started going through a period of enhanced
star formation at this moment. That is to say, why after a
Hubble time, do we not find examples of red, faded disks,
since surely in the future, the Universe will be filled with
them.
8 CONCLUSIONS
LSB galaxies have scale lengths comparable to those of late
type HSB galaxies. They do have bluer disks than HSB
galaxies. The blue colours cannot be entirely explained by
metallicity effects as judged from comparing with metal­
poor globular clusters. Comparison with evolution models
shows that part of the blue colours can be ascribed to age
effects for very low star formation rate scenarios.
These scenarios can be described by low, or absent
global star formation rates, with slight increases at present
or in the relatively recent past. This leads to a luminosity
dominant young population, with presumably a small, un­
derdeveloped (with respect to HSB galaxies) old population.
It may be that the LSB galaxies we observe are going
through a phase of more active starformation, as is suggested
by the partial star forming rings observed in a few galaxies.
The star formation is local and sporadic, as the colours do
not support starbursts or vigorous starformation hypothe­
ses. We may be seeing only galaxies that are in a more active
phase, while quiescent, inactive and much redder LSB galax­
ies may still escape detection.
For a better determination of the star formation history
of LSB galaxies we need additional information from near­
infrared wavelengths and chemical abundance information
to unravel age and metallicity effects.
ACKNOWLEDGMENTS
The authors would like to thank Stacy McGaugh and Rob
Swaters for many discussions and constructive comments,
and Reynier Peletier and Roelof de Jong for reading the
final versions of this paper.
REFERENCES
Allen, C.W. 1973, Astrophysical Quantities, 3rd ed., Athlone
Press, London, p. 263­265
Allen, R.J. & Shu, F.H. 1979, ApJ, 227, 67
Appleton, P.N., & Struck­Marcell, C. 1987, ApJ, 318, 103
Arimoto, N., & Tarrab, I. 1990, A&A, 228, 6
Bessell, M.S. 1979, PASP, 91, 589
Bosma, A., & Freeman, K.C. 1993, AJ, 106, 1394
Bothun, G.D. 1986, AJ, 91, 507
Bothun, G.D., Romanishin,W., Strom, K.M., & Strom, S.F. 1984,
AJ, 89, 1300
Bothun, G.D., Schombert, J.M., Impey, C.D., Sprayberry, D., &
McGaugh, S.S. 1993, AJ, 106, 530
Burstein, D., & Heiles, C. 1982, AJ, 87, 1165
Davies, J.I. 1990, MNRAS, 244, 8
de Jong, R.S., & van der Kruit, P.C. 1994, A&A Sup, in press
de Vaucouleurs, G. 1959, in: Handbuch der Physik, vol. 53, Astro­
physics IV: Stellar Systems, ed. S. Fl¨ugge (Berlin: Springer­
Verlag)
de Vaucouleurs, A., de Vaucouleurs, G., Buta, R.J., Corwin Jr.,
H.G., Fouqu'e, P., & Paturel, G. 1991, Third Reference Cat­
alogue of Bright Galaxies (RC3), Springer­Verlag, New York
Disney, M.J. 1976, Nature, 263, 573
Disney, M.J., Davies, J., & Phillipps, S. 1989, MNRAS, 239, 939
Disney, M.J., & Phillipps, S. 1983, MNRAS, 205, 1253
Elson, R.A.W. & Fall, S.M. 1985, ApJ, 299, 211
Freeman, K.C. 1970, ApJ, 160, 811
Giovanelli, R. & Haynes, M.P. 1989, ApJ, 346, L5
Giovanelli, R., Haynes, M.P., Salzer, J.J., Wegner, G., Da Costa,
L.N., Freudling, W. 1994, AJ, 107, 2036
GrosbÜl, P.J. 1985, A&A Sup, 60, 26
Guiderdoni, B., & Rocca­Volmerange, B. 1987, A&A, 186, 1
Han, M. 1992, ApJ Sup, 81, 35
Hesser, J.E., & Shawl, S.J. 1988, PASP, 97, 465
Holmberg, E. 1958, Medd. Lund. Obs. Series II, No 136
Huchra, J.P. 1977, ApJ Sup, 35, 171
Huizinga, J.E. 1994, Ph.D. Thesis, Univ. of Groningen
Impey, C. & Bothun, G.D. 1989, ApJ, 341, 89
Kennicutt, R.C. 1989, ApJ, 344, 685
Knezek, P.M. 1993, Ph.D. Thesis, Univ. of Massachusetts
Landolt, A.U. 1983, AJ, 88, 439
Larson, R.B., & Tinsley, B.M. 1978, ApJ, 219, 46
Lauberts, A., & Valentijn, E.A. 1989, The Surface Photometry
Catalogue of the ESO­Uppsala Galaxies (ESO­LV), European
Southern Observatory
Mazzei, P., Xu, C., & De Zotti, G. 1992, A&A, 256, 45
McGaugh, S.S. 1992, Ph.D. Thesis, Univ. of Michigan
McGaugh, S.S. & Bothun, G.D. 1993, in press
Nilson, P. 1973, Uppsala General Catalog of Galaxies, Ann. Up­
psala Astron. Obs., 6

22 W.J.G. de Blok, J.M. van der Hulst and G.D. Bothun
Peletier, R.F., & Willner, S.P. 1993, ApJ, 418, 626
Reed, B.C. 1985, PASP, 97, 120
Reed, B.C., Hesser, J.E., & Shawl, S.J. 1988, PASP, 100, 545
Salzer, J.J., Alighieri, S.D.S., Matteucci, F., Giovanelli, R., &
Haynes, M.P. 1991, AJ, 101, 1258
Salpeter, E.E. 1955, ApJ, 121, 161
Schombert, J.M., & Bothun, G.D. 1988, AJ, 95, 1389
Schombert, J.M., Bothun, G.D., Impey, C.D., & Mundy, L.G.
1990, AJ, 100, 1523
Schombert, J.M., Bothun, G.D., Schneider, S.E., & McGaugh,
S.S 1992, AJ, 103, 1107
Searle, L., Sargent, W.L.W., & Bagnuolo, W.G. 1973, ApJ, 179,
427
Thuan, T., Gott, R., & Schneider, S. 1987, ApJ, 315, L93
Valentijn, E.A. 1990, Nature, 346, 153
van der Hulst, J.M., Skillman, E.D., Smith, T.R., Bothun, G.D.,
McGaugh, S.S., & de Blok, W.J.G. 1993, AJ, 106, 548
van der Kruit, P.C. 1987, A&A, 173, 59
Witt, A.N., Thronson Jr., H.A., & Capuano Jr., J.M. 1992, ApJ,
393, 611
Wolfe, A.M., Turnshek, D.A., Smith, H.F., & Cohen, R.D. 1986,
ApJ Sup, 61, 249