ar
X
iv
:1
00
7.
32
65
v2
[
as
tro
-p
h.G
A]
2
8 J
ul
20
10
Mon. Not. R. Astron. Soc. 000, 1–14 (2010) Printed 29 July 2010 (MN LATEX style file v2.2)
Galaxy Zoo 1 : Data Release of Morphological Classifications for
nearly 900,000 galaxies?
Chris Lintott1,2†, Kevin Schawinski3,4‡, Steven Bamford5, Anzˇe Slosar6,
Kate Land1, Daniel Thomas7, Edd Edmondson7,Karen Masters7, Robert C.
Nichol7,M. Jordan Raddick8, Alex Szalay8,Dan Andreescu9, Phil Murray10,
Jan Vandenberg8
1Oxford Astrophysics, Denys Wilkinson Building, Keble Road, Oxford, OX1 3RH, UK
2Adler Planetarium, 1300 S. Lake Shore Drive, Chicago, IL 60605, U.S.A.
3Einstein Fellow, Department of Physics, Yale University, New Haven, CT 06511, U.S.A.
4Yale Center for Astronomy and Astrophysics, Yale University, P.O. Box 208121, New Haven, CT 06520, U.S.A.
5Centre for Astronomy & Particle Theory, University of Nottingham, University Park, Nottingham, NG7 2RD, UK
6Brookhaven National Laboratory, Upton NY 11973, USA
7Institute of Cosmology and Gravitation, University of Portsmouth, Dennis Sciama Building, Burnaby Road, Portsmouth, PO1 3FX, UK
8Department of Physics and Astronomy, Johns Hopkins University, 3400 N. Charles St., Baltimore, MD 21218, USA
9LinkLab, 4506 Graystone Ave., Bronx, NY 10471, USA
10Fingerprint Digital Media, 9 Victoria Close, Newtownards, Co. Down, Northern Ireland, BT23 7GY, UK
July 2010
ABSTRACT
Morphology is a powerful indicator of a galaxy’s dynamical and merger history. It is
strongly correlated with many physical parameters, including mass, star formation history
and the distribution of mass. The Galaxy Zoo project collected simple morphological clas-
sifications of nearly 900,000 galaxies drawn from the Sloan Digital Sky Survey, contributed
by hundreds of thousands of volunteers. This large number of classifications allows us to ex-
clude classifier error, and measure the influence of subtle biases inherent in morphological
classification. This paper presents the data collected by the project, alongside measures of
classification accuracy and bias. The data are now publicly available and full catalogues can
be downloaded in electronic format from http://data.galaxyzoo.org.
Key words: methods: data analysis, galaxies: general, galaxies: spiral, galaxies: elliptical and
lenticular
1 INTRODUCTION
The aim of the Galaxy Zoo project was to provide visual morpholo-
gies for nearly one million galaxies from the Sloan Digital Sky Sur-
vey (York et al. 2000), including the whole Main Galaxy Sample
(Strauss et al. 2002). Separating galaxies into categories based on
their morphology (the visual appearance or shape) has been stan-
dard practice since it was first systematically applied by Hubble
(1936). These morphological categories are broadly correlated with
other, physical parameters, such as the star formation rate and his-
? This publication has been made possible by the participation of more
than 100,000 volunteers in the Galaxy Zoo project. Their contributions are
individually acknowledged at http://www.galaxyzoo.org/Volunteers.aspx
† Email: cjl@astro.ox.ac.uk
‡ Email: kevin.schawinski@yale.edu
tory, the gas fraction and dynamical state of the system (Roberts &
Haynes 1994); understanding these correlations and studying the
cases where they do not apply is critical to our understanding of
the formation and evolution of the galaxy population. It is tempting
to identify the morphological distinctions with the clear colour bi-
modality in the population of galaxies visible in data from modern
surveys (e.g Strateva et al. 2001), but extremely large sets of classi-
fied galaxies are necessary before this hypothesis can be tested. For
most of the twentieth century, morphological catalogues were com-
piled by individuals or small teams of astronomers (e.g. Sandage
1961; de Vaucouleurs 1991), but modern surveys containing many
hundreds of thousands of galaxies make this approach impractical.
Attempts to solve this problem have taken one of three ap-
proaches. The first is to use physical parameters, such as colour,
concentration index, spectral features, surface brightness profile,
c© 2010 RAS

2 Lintott et al.
structual features, spectral energy distribution (Kinney et al. 1996)
or some combination of these as a proxy for morphology (e.g.
Abraham, van den Bergh & Nair 2003; Conselice 2006). As no
proxy is an exact substitute for true visual morphology, the intro-
duction of each of these variables results in an unknown and po-
tentially unquantifiable bias in the resulting sample of galaxies. Al-
though morphological labels are often used for the resulting cat-
alogues, usually after comparison with a small number of expert
classifications, each of these techniques produces catalogues that
cannot entirely reproduce true morphological selection. For exam-
ple, Schawinski et al. (2007) showed that the proportion of elliptical
galaxies with recent star formation or nuclear activity was signifi-
cantly higher in a sample classified by eye than in samples assem-
bled using proxies for morphology. This reflects that fact that proxy
quantities such as colour do not directly probe the dynamical state
of the system which controls the morphology.
The second strategy was applied by Lahav et al. (1995) and
then further developed by Ball et al. (2004) amongst others. The
aim was to develop automatic classification routines, typically neu-
ral networks, to the point that they can replace the need for hu-
man classifications. While successful in classifying the majority of
galaxies, a major problem is that due to both the use of proxies for
morphology as input and the inherent complexity of the network, it
is not easy to predict or understand the bias in the resulting classifi-
cations. As a result these automatic classifiers have not been widely
adopted.
The third approach is to attempt to expand the reach of vi-
sual classifications. Previous professional attempts (Fukugita et al.
2007; Nair & Abraham 2010) have been necessarily limited by the
extraordinary effort required to classify even relatively small sub-
sets of the SDSS; the largest, MOSES (Schawinski et al. 2007),
included basic classifications of only 50,000 galaxies at redshifts
between 0.05 and 0.1, and with r < 16.8 (∼ 5.5 per cent of the
SDSS). The results presented in this paper, which expands on our
first description of the Galaxy Zoo project and its results (Lintott et
al. 2008), provide estimates of the visual morphology of the entire
SDSS main galaxy sample. Having produced a catalogue of visual
morphologies, we can use the other measured parameters, includ-
ing colour, to investigate the galaxy population.
Galaxy Zoo is possible because of the involvement of hun-
dreds of thousands of volunteer ‘citizen scientists’. The involve-
ment of non-professionals in astronomical science has a long and
distinguished history. From the early contributions of observers
to modern discoveries of supernovæ and follow-up of candidate
extra-solar planets (e.g. Barbieri et al. 2007), astronomers have of-
ten depended on volunteers. The Galaxy Zoo project expands the
role of non-professionals in astrophysics from data collection to
include data analysis, a technique was first successfully employed
in astronomy or astrophysics by the Stardust@Home project (West-
phal et al. 2006; Mendez 2008). Its usefulness is demonstrated both
by the catalogue presented here, but also by the serendipitous dis-
covery of unusual objects and classes of object discussed elsewhere
(e.g. Lintott et al. 2009; Cardamone et al. 2009).
2 SAMPLE SELECTION AND WEB SITE OPERATION
The images of galaxies presented for classification by Galaxy Zoo
were drawn from the Sloan Digital Sky Survey, a survey of a
large part of the northern sky providing photometry in five filters
(Fukugita 1996; Smith et al. 2002) and covering ∼ 26 per cent of
the sky. Data Release 6 (DR6, Adelman-McCarthy et al. 2008) was
Class Button Description
1 Elliptical galaxy
2 Clockwise/Z-wise spiral galaxy
3 Anti-clockwise/S-wise spiral galaxy
4 Spiral galaxy other (eg. edge on)
5 Star or Don’t Know (eg. artefact)
6 Merger
Table 1. Galaxy Zoo classification categories showing schematic symbols
as used on the site.
Figure 1. Main analysis page from the Galaxy Zoo website.
used for the initial selection of candidates. We include the Main
Galaxy Sample (MGS) of Strauss et al. (2002) which includes all
extended objects with Petrosian magnitude r < 17.77, a total of
738,175 galaxies including those for which spectra were not avail-
able at the time of the DR6 release. In order to be as inclusive as
possible, 155,037 objects which had been included in the SDSS
spectroscopic survey (for various reasons, including meeting crite-
ria intended to select for luminous red galaxies, quasars and other
unusual objects) and subsequently classified as a galaxy due to their
spectral properties were added, making a total of 893,212 objects.
Composite images of these objects in the g,r and i bands were
provided by the ImgCutout web service (Nieto-Santisteban, Sza-
lay & Gray 2004) on the SDSS SkyServer website (Szalay et al.
2002) and then shown to visitors to the Galaxy Zoo website1, who
were then asked to classify the galaxy into one of the six categories
shown in Table 1. Distinguishing clockwise and anticlockwise spi-
rals was not only useful in its own right, but also allowed Masters
et al. (2010b) to ensure that their sample of red spirals genuinely
included only spirals and not edge-on disks or possible S0 galax-
ies; spiral arms must have been seen by a majority of classifiers to
record significant evidence for either a clockwise or an anticlock-
wise spin. The size of the image of each galaxy was chosen so that
the scale was 0.024Rp arcsec per pixel, where Rp is the Petrosian
radius for the system. The images were 423 pixels (≈ 10Rp for a
typical system) on each side. The interface is shown in Figure 1.
The web site was launched on 11th July 2007, and full details of its
operation are given in Lintott et al. (2008).
1 The original Galaxy Zoo website is maintained and archived at http:
//zoo1.galaxyzoo.org
c© 2010 RAS, MNRAS 000, 1–14

Galaxy Zoo 3
2.1 Data reduction
The data reduction required to turn clicks provided via a web-
site into a scientific catalogue is substantial. As well as compar-
ing Galaxy Zoo with existing professional catalogues, Lintott et
al. (2008) explored the possibility of weighting users according to
how often they agreed with the majority, but found little change to
the resulting classifications. Requirements for 80 per cent and 95
per cent agreement amongst users were then used to define ‘clean’
and ‘superclean’ samples respectively. This approach, while suit-
able for some purposes, has proved to be inadequate for others.
Darg et al. (2010a), in their study of merging galaxies, found that
all galaxies with a fraction of 40 per cent or more of their vote in the
‘merger’ category were, in fact, true mergers. This result suggests
that the application of a single critical threshold to all classifica-
tions is over-simplistic. The catalogue presented in this paper there-
fore includes the fraction of clicks in each category for all galaxies,
rather than just those in the ‘clean’ or ‘superclean’ samples. Users
of the dataset presented here are, however, recommended to use
cuts of 0.8 or 0.95 in the first instance to ensure where possible that
results are comparable with earlier results.
In using the Galaxy Zoo morphologies, it is important to con-
sider the population of galaxies which are unclassified according
to the criterion used to assign individual galaxies to a population.
It is obviously possible to derive a classification for every galaxy
by simply assigning it to the category with the greatest fraction of
the vote (which we refer to as the greater criterion); a galaxy
with 51 per cent of the vote in the elliptical category would, in this
system, be considered an elliptical. For more stringent thresholds
(e.g. clean, where a galaxy would require at least 80 per cent of
the vote to be assigned a classification) then unclassified galaxies
may form a majority of the sample. In order to evaluate the effect
of this feature of the data, we determine the fraction of unclassified
galaxies as a function of magnitude and size.
In an effort to quantify the effect of other potential biases in
the classification process, mirrored and greyscale images were in-
troduced to the site from 28 November 2007. The greyscale images
were not single filter images, but in order to minimize the effect of
apparent changes in morphology caused by viewing the galaxy in
different wavelengths were instead produced from the gri colour
images provided by the SDSS pipeline. A subsample of the main
catalogue sample was used, comprising the superclean sample
as of 4th September 2007 (i.e. all galaxies with an agreement of
more than 95 percent on that date) and a random sampling of 5 per-
cent of the rest of the sample, comprising 91,303 images in total. A
list of galaxies included in this bias study sample is given in section
4.
The results of this bias study were discussed in Lintott et al.
(2008) and Land et al. (2008). Significant but small biases in spin
direction and colour were found, with anticlockwise spiral classi-
fications favoured over clockwise, and the greyscale images were
more likely to be classified as elliptical than their colour counter-
parts.
Although these biases were small, the effect of studying them
was relatively large. Any study of the behaviour of human classi-
fiers is likely to encounter a phenomenon known as the Hawthorne
effect (Mayo 1933), the risk of changing the behaviour of those
taking part in a study simply by carrying out the study itself. A
change in classifier behaviour was indeed observed, with users be-
ing slightly more careful in their classifications during the bias
study and thus classifying fewer galaxies as spiral. The effect is
small (∼ 3 per cent fewer votes were received in the spiral cate-
gories) but significant. Rather than just combining classifications
for each of the galaxies included in the bias study, therefore, we
present the data from before and then during the study separately.
3 PROPERTIES OF THE DATA
3.1 Quantifying bias
Bamford et al. (2009) carry out a more sophisticated analysis of the
Galaxy Zoo data, initially motivated by the desire to determine the
relationship between morphology and the local density of galaxies.
The technique used recognises that although small, faint or distant
galaxies will likely appear as ellipticals and therefore will be classi-
fied as such with a high degree of agreement, many of these systems
are likely to be spirals whose arms are not distinctly visible in the
SDSS images. By assuming that the morphological fraction within
bins of fixed galaxy size and luminosity does not evolve over the
depth of the SDSS, it is possible to statistically estimate the bias
affecting the morphological classifications for galaxies of known
luminosity, size and distance. It is important to note that this bias
does not arise from the involvement of volunteers in the classifica-
tion process, but from the inherent limitations of the survey data.
The method only deals with removing the bias relative to the
least biased data (i.e. that from nearby systems) and so there may be
a remaining, unquantified bias, for example due to bias in human
pattern recognition abilities. The bias correction will also reduce
the impact of any true redshift evolution from the sample (although
only that evolution which affects the morphological mix of the pop-
ulation at a given absolute magnitude and physical size).
The effect of this bias on a final catalogue depends on how
the raw Galaxy Zoo classifications are treated. As an example, if a
simple majority of the vote is used to classify galaxies then ∼ 13.5
per cent of galaxies are in absolute magnitude–size bins where ap-
proximately no bias correction is necessary, and ∼ 20 per cent are
in bins which have approximately no misclassified galaxies. Con-
versely, ∼ 6 percent of galaxies are in bins where the bias correc-
tion changes the classifications for more than half of the objects. In
contrast, if the clean criterion is used then a much larger fraction,
over 70 per cent, of galaxies are in absolute magnitude size bins
for which no objects are misclassified due to this bias. The price of
applying a more stringent criterion for classification is thus a large
fraction of objects which do not meet the clean criteria and thus
have ‘uncertain’ classifications.
As determining the bias correction requires a redshift, debi-
ased results are only available for objects which were spectroscopi-
cally observed by SDSS, a subset of the whole Galaxy Zoo sample.
The determination also requires a well-defined, homogeneous sam-
ple and is therefore limited to MGS objects with reliable r-band
photometry, redshifts in the range 0.001–0.25 and absolute magni-
tudes and sizes that are not extreme outliers from the normal galaxy
distribution. Bamford et al. (2009) used DR6 of the SDSS which
only provided spectroscopic coverage for 82 percent of the survey
area. With the availability of SDSS Data Release 7 (Abazajian et al.
2009), the spectroscopic coverage has risen. As a result, the number
of objects with redshifts has increased from 677,515 (76 percent)
to 781,842 (88 percent). Considering just the Main Galaxy Sam-
ple, the total number of objects in Galaxy Zoo is 738,173, of which
679,721 (92 percent) have redshifts in DR7, up from 575,398 (78
percent) in DR6. In calculating the classification bias corrections
we have thus supplemented our previous DR6 catalogue with ad-
ditional redshifts from DR7, significantly increasing the fraction of
the sample for which we can provide these corrections.
c© 2010 RAS, MNRAS 000, 1–14

4 Lintott et al.
As described in appendix A of Bamford et al. 2009, we first
divide the sample into bins of similar luminosity, physical size and
redshift. We then find for each point in luminosity–size space the
lowest redshift bin containing at least 30 galaxies, and assume that
this bin represents the ‘true’ early-type to spiral ratio. In an attempt
to keep this baseline estimate unbiased, we only consider bins well
away from the magnitude, size and surface brightness limits of the
sample.
Having obtained an approximation to the low-redshift early-
type to spiral ratio as a function of both luminosity and size, we
fit an appropriate smooth function to the result. Equation A1 in
Bamford et al. gives the fit
〈
nel
nsp
〉
base
= p1
1 + exp
(
s1(R50)−Mr
s2(R50)
) + p2, (1)
where,
s1(R50) = q−(q2+q3R50
q4 )
1 + q5,
and (2)
s2(R50) = r1 + r2(s1(R50)− q5).
By considering the difference between this baseline early-type
to spiral ratio and that measured for a given bin of absolute magni-
tude, physical size and redshift, we can estimate the correction, C,
required for any particular galaxy as
C(Mr, R50, z) = log
(
〈nel/nsp〉raw
〈nel/nsp〉base
)
, (3)
where angular brackets indicate averages over bins of (Mr, R50,
z).
Individual vote shares (type likelihoods) for each galaxy are
adjusted as
pel,adj =
1
1
/( pel
psp
)
adj +
px
pel
+ 1
, (4)
psp,adj =
1
( pel
psp
)
adj +
px
psp + 1
,
where
(
pel
psp
)
adj
=
(
pel
psp
)
raw
/
10C(Mr ,R50,z), (5)
and px = 1− pel − psp.
Figure 2 illustrates the effect of the bias, and the result of
adopting the measured correction, as a function of redshift, appar-
ent magnitude and apparent size, for three bins of absolute magni-
tude and physical size. The overall result of applying the correction
is to lower the probability that a given galaxy will be classified
as early-type and increase the chance that it will classified as spi-
ral. The effect is largest around the median SDSS redshift, and for
faint, small galaxies; bright, large, low redshift galaxies need only
a small correction, whereas at the highest redshifts only the most
luminous galaxies pass the sample selection criteria, most of which
are indeed early-types.
3.2 Measures of confidence
To assist users of the Galaxy Zoo dataset in evaluating the morpho-
logical classifications they obtain, both individually and for larger
samples, we have calculated a number of relevant statistics. These
were derived from the bias correction procedure described above
and thus reflect only the sensitivity of the classification to the bias
described in the previous section. They do not take into account
other systematic biases that may exist in the data set (see Lintott et
al. 2008 for a comparision of the Galaxy Zoo classifications with
other catalogues of visual morphology).
A first indicator of the quality of an individual morphology is
the difference between its raw and debiased likelihoods, ∆p. The
bias corrections are inherently uncertain, especially when applied
to individual galaxies rather than to large samples, but the size of
the bias correction is an indication of the uncertainty in the galaxy’s
type.
To remove the individual uncertainties on our confidence mea-
sures, for each galaxy we calculate values computed from a ‘bin’
of galaxies with similar redshift, absolute magnitude and physical
size, corresponding to the same binning used in quantifying the bias
correction. We therefore provide the mean and standard deviation
of ∆p in each galaxies’ bin, 〈∆p〉 and σ∆p respectively.
A confidence measure of perhaps more practical use is an esti-
mate of the probability that a given galaxy may have been classified
as an elliptical when it is in reality a spiral. We thus calculate the
fraction of objects within a given galaxy’s bin that are classified
as elliptical using the raw data but as spiral when the effect of the
bias correction is taken into account. This fraction of misclassi-
fied galaxies, fmisclass, depends strongly on the threshold used in
the analysis to define ‘spiral’ or ‘elliptical’, being highest when the
greater condition is used (i.e. when a galaxy is classified as an
elliptical when pel > psp) and smaller when more stringent clas-
sifications are used. However, the cost of using a more stringent
classification threshold is that an increasing fraction of galaxies are
unclassifiable, i.e. they do not meet the criteria for any of the clas-
sifications and are thus ‘uncertain’. We quantify this by measuring
the fraction of unclassified galaxies, funclass, in the same bin of
redshift, magnitude and size as the particular gaalxy in question.
The distribution of these quantities amongst the Galaxy Zoo
sample are shown in Figure 3 (for the greater criterion) and Fig-
ure 4 (for clean). Note that weighting the results to favour those
users who tend to agree with the majority makes little difference
compared with the effect of changing the classification threshold.
Following earlier Galaxy Zoo papers, a clean sample has
been definied by requiring 80 percent of the corrected vote to be
in a particular category. However, this choice was somewhat ar-
bitary, and yet has a significant effect on the number of unclassified
and misclassified galaxies. This effect is shown in Figure 5 for four
thresholds: 50 percent (greater), 60 percent (cleanish), 80
percent (clean) and 95 percent (superclean). The mean val-
ues of each distribution are indicated with arrows. For example, a
threshold of 50 percent results, by design, in a classification for
every galaxy but 19 percent are misclassified. A threshold of 60
percent results in 33 percent of galaxies unclassified and 10 per-
cent misclassified. A threshold of 80 percent results in 60 percent
of galaxies unclassified and 3 percent misclassified, while a thresh-
old of 95 percent results in no misclassifications but 88 percent of
galaxies unclassified. These figures illustrate the general principle
of working with these data; as the the threshold is made more strin-
gent, then the fraction of unclassified objects increases while the
fraction of misclassified objects decreases.
4 THE CATALOGUE
Table 2 contains the data for all galaxies with measured redshifts
in the range 0.001 < z < 0.25 and u and r photometry in SDSS
DR7, excluding those with extreme absolute magnitudes or sizes
c© 2010 RAS, MNRAS 000, 1–14

Galaxy Zoo 5
Figure 2. The effect of the bias, and of the result of adopting the measured correction as a function of redshift, apparent magnitude and apparent size. The thin
and thick lines correspond to de-biased and raw classifications respectively.
given by the SDSS pipeline. 667,945 galaxies are included. This
table includes the raw votes, the weighted votes in elliptical (E)
and combined spiral (CS) categories, and flags indicating the inclu-
sion of the galaxy in a clean, debiased catalogue. The flags take
into account not only the redshift dependence of the spiral/elliptical
ratio as described in Section 3.1 but also the redshift dependence
of the ratio of spirals to ellipticals in the clean catalogue. This re-
sults in larger corrections than would otherwise be necessary. As
explained above, bias correction was only possible for those galax-
ies for which SDSS DR7 included spectra, and so Table 3 contains
classifications for galaxies included in the Galaxy Zoo sample for
which spectra were not included in SDSS DR7.
As discussed in Section 2.1, while the introduction of mirrored
and monochrome images was important in allowing the measure-
ment of human bias, doing so also affected the behaviour of the par-
ticipants. The measurements obtained during this bias study have
thus not been combined with the main data set described above.
Tables 5 and 6 gives details of the votes assigned to each category
for the galaxies which were included in the bias study, as well as a
combined vote in Table 7.
The final table, Table 4, presents the confidence measures
discussed in section 3.2, calculated using absolute magnitude and
physical (rather than apparent) size.
c© 2010 RAS, MNRAS 000, 1–14

6 Lintott et al.
Figure 3. Histograms showing (left) the average bias correction applied to the type-likelihoods (∆p) and (right) the estimated fraction of objects that are
misclassified, at the absolute (blue, solid line) and apparent (red, dotted line) magnitude and size of each galaxy in the Galaxy Zoo Main Galaxy Sample.
These are calculated using the greater classification criteria, as described in the text. For example, note that ∼ 13.5 per cent of galaxies are in absolute
magnitude–size bins where approximately no bias correction is necessary, and ∼ 20 per cent are in bins which have approximately no misclassified galaxies.
Conversely, ∼ 6 per cent of galaxies are in bins where the bias correction changes the classifications for more than half of the objects.
Figure 4. (Left) and (centre) as Fig. 3, but for galaxies classified using the clean criteria. This figure also includes (right) the fraction of objects that are
unclassified by the clean criteria, i.e. they have both psp and pel < 0.8. The average corrections, are slightly larger for the clean versus greater criteria,
misclassifications are considerably lower, but at the expense of a large fraction of unclassified galaxies. For example, over 70 percent of galaxies are in absolute
magnitude-size bins for which the classification bias results in no objects being misclassified, but roughly two-thirds of galaxies are in absolute magnitude-size
bins for which at least half the objects are unclassified.
4.1 Examples
This paper presents, as a legacy for the community, the entire
Galaxy Zoo 1 data set. Users should bear in mind that the objects
included in Galaxy Zoo 1 were selected by a combination of crite-
ria (see 2), and therefore appropriate additional cuts (in magnitude,
redshift, etc.) should be made to produce a well-defined sample ap-
propriate to any particular study.
Most users of this data will have specific requirements which
fall into one of a few categories. For example, one may require
a small number of spiral or elliptical galaxies, perhaps for an ob-
serving proposal. In this situation, we suggest using a subset of
the galaxies which we have flagged as belonging to the relevant
category according to the clean criterion incorporating the bias-
correction (Table 2).
The data in Table 4 will then allow the user to estimate the
fraction of the derived sample which are misclassified due to in-
herent classification bias (i.e. the inability to detect spiral arms in
faint or small systems). The certainty of the individual classifica-
tions can be improved by using a higher threshold (e.g. requiring
95 per cent agreement amongst classifiers) or by selecting nearby,
bright and/or large galaxies.
This procedure will suffice for many studies, but one should
always consider the properties of the objects with ‘uncertain’ clas-
sifications where these may potentially affect the result. This will
be the case for many statisticial studies. In such circumstances, it
may be preferable to have an estimate of the morphology of all sys-
tems, rather than leaving a large number unclassified. In this case,
the morphological type likelihoods from Table 2 (ideally the debi-
ased quantities) may be used directly, or a simple majority vote can
be applied.
Finally, if the small effect of the change in behaviour associ-
ated with the bias study can be ignored, and no bias-correction is
required, a greater number of votes for ∼ 250, 000 systems can be
obtained from Table 7.
c© 2010 RAS, MNRAS 000, 1–14

Galaxy Zoo 7
Figure 5. Histograms of (left) fraction misclassified and (right) fraction unclassified, for greater (black, solid line), cleanish (red, dot-dashed line),
clean (green, dashed line) and superclean (blue, dotted line). The arrows at the top indicate the means of each distribution. One can clearly see that as
the classification criteria become more stringent the fraction of misclassified objects decreases, but at the expenses of an increasing fraction of unclassified
objects.
5 CONCLUSIONS
This paper presents the results of Galaxy Zoo 1, which used the
World Wide Web to recruit a large community of volunteers to
provide morphological classifications of galaxies drawn from the
Sloan Digital Sky Survey. Such morphological classifications are
useful indicators of a galaxy’s dynamical state and are correlated
with many other physical parameters.
The data presented here has already produced several interest-
ing results. Much of this work, published elsewhere (eg Bamford et
al. 2009; Schawinski et al. 2009; Masters et al. 2010a; Land et al.
2008) was only possible because of the large number of morpho-
logical classifications provided by the project. The clockwise/anti-
clockwise classifications of the spiral galaxies have been used to
show that (as expected) there is no evidence for a preferred rota-
tion direction in the universe, but the results suggest that humans
preferentially classify spiral galaxies as anti-clockwise (Land et al.
2008). They hint at a local correlation of galaxy spins at distances
less than ∼ 0.5Mpc - the first experimental evidence for chiral cor-
relation of spins (Slosar et al. 2009).
The disentangling of morphological and colour based classifi-
cations allows us to study the separate dependences of morphology
and colour on environment and provide evidence that the transfor-
mation of galaxies from blue to red proceeds faster than the trans-
formation from spiral to early type (for example, Bamford et al.
2009 and Skibba et al. 2009). The importance of this division is
illustrated by the sample of passive, red, spirals in Masters et al.
2010b); these are disk galaxies in the outskirts of groups and clus-
ters of galaxies which have either exhausted thier gas, or lost it
through strangulation or another mechanism.
The Galaxy Zoo results can also be used to constrain the prop-
erties of dust in spiral galaxies (Masters et al. 2010a). Schawinski
et al. (2010) use Galaxy Zoo classifications to distinguish the host
galaxies of AGN, finding that in the present day Universe activity
is preferentially found in low mass early-types and high mass late-
types. The sample of merging galaxies have been used to show that
the local fraction of mergers is between 1 and 3 percent and to study
the global properties of merging galaxies (Darg et al. 2010a,b).
Other serendipitous discoveries have been made because of
the close attention given by classifiers to each image. Galaxy Zoo
has brought to light several rare classes of object. ‘Hanny’s Voor-
werp’ - an unusual emission line nebula neighbouring the spiral
galaxy IC 2497 has been studied in several follow-up projects (Lin-
tott et al. 2009; Jo´zsa 2009). An unusual class of emission line
galaxies (the ‘peas’) have been discovered - their properties are dis-
cussed by Cardamone et al. (2009) and Amorı`n et al. (2010).
The success of the project, both in quickly attracting large
numbers of volunteers and in providing data that is useful for sci-
ence, suggests that this mode of ‘citizen science’ may provide a
valuable method of data analysis for large data sets. A follow-up
project, Galaxy Zoo 22, has obtained more than 60 million more
detailed classifications of a subset of the Galaxy Zoo sample, and
has already produced results; Masters et al. (2010c) find that the
presence of a bar is strongly linked to galaxy colour, with a bulge
and bar-dominated sequence of red galaxies separated from a pre-
dominately bar-less blue cloud.
Classification of galaxies drawn from large Hubble Space
Telescope surveys is now underway3. Two sister projects investigat-
ing transient detection4 and merger simulations5 are underway, and
results from these projects will be reported in future papers. Data
from a third spin-off, which asked users to determine the length
of the bars in barred galaxies, is now being reduced6. Obtaining a
large number of visual classifications is not only inherently useful,
but also provides a rich training set for improving automated tech-
niques (Banerji et al. 2010); this combination of citizen science and
machine learning will be essential in dealing with the data rates ex-
pected from future sky surveys, such as the Large Synoptic Survey
Telescope (LSST Science Collaboration 2009). Whether used di-
rectly or to inform future surveys, Galaxy Zoo has shown that the
efforts of volunteers, coordinated via the internet, can produce rich
seams of science.
2 http://zoo2.galaxyzoo.org
3 http://hubble.galaxyzoo.org
4 http://supernova.galaxyzoo.org
5 http://mergers.galaxyzoo.org
6 http://www.icg.port.ac.uk/
˜
hoyleb/bars/
c© 2010 RAS, MNRAS 000, 1–14

8 Lintott et al.
ACKNOWLEDGMENTS
The data in this paper is the result of the efforts of the Galaxy Zoo
volunteers, without whom none of this work would have been pos-
sible.
Galaxy Zoo has been supported in part by a Jim Gray research
grant from Microsoft, and by a grant from The Leverhulme Trust.
CJL acknowledges support from the STFC Science in Society Pro-
gram and The Leverhulme Trust, and thanks Prof. Joe Silk for his
support. This work is supported in part by the U.S. Department of
Energy under Contract No. DE-AC02-98CH10886. Support for the
work of K.S. was provided by NASA through Einstein Postdoc-
toral Fellowship grant number PF9-00069 issued by the Chandra
X-ray Observatory Center, which is operated by the Smithsonian
Astrophysical Observatory for and on behalf of NASA under con-
tract NAS8-03060. KLM acknowledges funding from the Peter and
Patricia Gruber Foundation as the 2008 IAU Fellow, and from the
University of Portsmouth and SEPnet (www.sepnet.ac.uk). RCN
thanks Google for partial funding during the Galaxy Zoo project.
Funding for the SDSS and SDSS-II has been provided by
the Alfred P. Sloan Foundation, the Participating Institutions, the
National Science Foundation, the U.S. Department of Energy,
the National Aeronautics and Space Administration, the Japanese
Monbukagakusho, the Max Planck Society, and the Higher Ed-
ucation Funding Council for England. The SDSS Web Site is
http://www.sdss.org/.
The SDSS is managed by the Astrophysical Research Con-
sortium for the Participating Institutions. The Participating Institu-
tions are the American Museum of Natural History, Astrophysical
Institute Potsdam, University of Basel, University of Cambridge,
Case Western Reserve University, University of Chicago, Drexel
University, Fermilab, the Institute for Advanced Study, the Japan
Participation Group, Johns Hopkins University, the Joint Institute
for Nuclear Astrophysics, the Kavli Institute for Particle Astro-
physics and Cosmology, the Korean Scientist Group, the Chinese
Academy of Sciences (LAMOST), Los Alamos National Labora-
tory, the Max-Planck-Institute for Astronomy (MPIA), the Max-
Planck-Institute for Astrophysics (MPA), New Mexico State Uni-
versity, Ohio State University, University of Pittsburgh, University
of Portsmouth, Princeton University, the United States Naval Ob-
servatory, and the University of Washington.
REFERENCES
Abraham, R.G., van den Bergh, S., Nair, P., 2003, ApJ, 588, 218
Adelman-McCarthy et al., 2008, ApJS, 175, 297
Abazajian, K.N. et al., 2009, ApJS, 182, 543
Amorı`n et al., 2010, ApJL, in press, arXiv : 1004.4910
Ball et al., 2004, MNRAS, 348, 1038
Banerji et al., 2010, MNRAS, 402, 2264
Bamford, S.P. et al., 2009, MNRAS, 393, 1324
Barbieri et al., 2007, A&A, 476, 13
Cardamone, C. et al., 2009, MNRAS, 399, 1191
Conselice, C.J., 2006, MNRAS, 373, 1389
Darg, D., et al. 2010, MNRAS, 401, 1043
Darg, D., et al., 2010, MNRAS, 401, 1552
Fukugita, M., et al., 1996, AJ, 111, 1748
Fukugita, M., et al., 2007, AJ, 134, 579
Hubble, E, 1936, The Realm of the Nebulæ, Yale University Press
Jo´zsa et al., 2009, A&A, 500, 33
Kinney et al., ApJ, 467, 38
Lahav, O. et al., 1995, Science, 267, 859
Land, K. & the Galaxy Zoo team 2008, MNRAS, 388, 1686
Lintott et al., 2008, MNRAS, 389, 1179
Lintott, C.J. et al., 2009, MNRAS, 399, 129
LSST Science Collaboration, arXiv : 0912.0201
Masters, K., 2010a, MNRAS, 405, 783
Masters, K., et al., 2010b, MNRAS, 404, 792
Masters K., et al., 2010c, Submitted to MNRAS., arXiv:
1003.0449
Mayo, E., 1933, The human problems of an industrial civilization
(New York : MacMillian) ch.3
Mendez, B. et al, ASP Conference Series vol. 389, ed. C. Gar-
many, M.Gibbs & J.W. Moody, 219-226.
Nair, P.B. & Abraham, R.G., 2010, ApJS, 186, 427
Nieto-Santisteban, M., Szalay, A. & Gray, J, 2004, in Astronom-
ical Data Analysis & Software Systems XIII (ASP Conference
Series 314) eds F. Ochsenbein, M. Allen & D. Egret, p 666
Roberts, M.S. & Haynes, M.P., 1994, ARAA, 32, 115
Sandage, A.R., 1961, The Hubble Atlas of Galaxies, Carnegie In-
stitute of Washington, Washington
Sandage, A.R. & Visvanathan, N., 1978, ApJ, 225, 742
Schawinski, K. et al., 2007, MNRAS, 382, 1415
Schawinkski, K., & the Galaxy Zoo team, 2009, MNRAS, 396,
818
Schawinski, K. et al., 2010, ApJ, 711, 284
Skibba et al., 2009, MNRAS, 399, 966
Slosar, A. et al., 2009, MNRAS, 392, 1225
Smith, J.A. et al. 2002, AJ, 123, 2121
Strateva et al., 2001, AJ, 122, 1861
Strauss, M.A., et al., AJ, 123, 1810
Sugai, H. & Iye, M., 1995, MNRAS¡ 276, 327
Szalay, A. S., Gray, J., Thakar, A. R., Kunszt, P. Z., Malik, T.,
Raddick, M. J., Stoughton, C., & vandenBerg, J. 2002, in Pro-
ceedings of the 2002 ACM SIGMOD International Conference
on Management of Data, 570
de Vaucouleurs, G. et al., 1991, Third Reference Catalog of Bright
Galaxies, Springer-Verlag, New York.
Westphal A.J. et al., 2006, AGU, Fall Meeting, abstract P52B-08
York, D.G., 2000, AJ, 120, 1579
c© 2010 RAS, MNRAS 000, 1–14

G
ala
xy
Z
o
o
9
Table 2. Classifications of galaxies with spectra
Votesc Debiased votesd Flagse
ObjIDa RA Dec Nvoteb E CW ACW Edge DK MG CS E CS Spiral Elliptical Uncertain
587727178986356823 00:00:00.41 -10:22:25.7 59 0.61 0.034 0.0 0.153 0.153 0.051 0.186 0.61 0.186 0 0 1
587727227300741210 00:00:00.74 -09:13:20.2 18 0.611 0.0 0.167 0.222 0.0 0.0 0.389 0.203 0.797 1 0 0
587727225153257596 00:00:01.03 -10:56:48.0 68 0.735 0.029 0.0 0.147 0.074 0.015 0.176 0.432 0.428 0 0 1
587730774962536596 00:00:01.38 +15:30:35.3 52 0.885 0.019 0.0 0.058 0.019 0.019 0.077 0.885 0.077 0 1 0
587731186203885750 00:00:01.55 -00:05:33.3 59 0.712 0.0 0.0 0.22 0.068 0.0 0.22 0.64 0.29 0 0 1
587727180060098638 00:00:01.57 -09:29:40.3 28 0.857 0.0 0.036 0.0 0.107 0.0 0.036 0.83 0.06 0 0 1
587731187277627676 00:00:01.86 +00:43:09.3 38 0.5 0.0 0.053 0.289 0.105 0.053 0.342 0.351 0.473 0 0 1
587727223024189605 00:00:02.00 +15:41:49.8 26 0.423 0.0 0.0 0.577 0.0 0.0 0.577 0.143 0.857 1 0 0
587730775499407375 00:00:02.10 +15:52:54.2 62 0.355 0.016 0.21 0.323 0.0 0.097 0.548 0.355 0.548 0 0 1
587727221950382424 00:00:02.41 +14:49:19.0 31 0.484 0.129 0.065 0.258 0.065 0.0 0.452 0.109 0.789 1 0 0
587730774425665704 00:00:02.58 +15:02:28.3 24 0.583 0.042 0.125 0.167 0.083 0.0 0.333 0.147 0.701 0 0 1
587730773888794751 00:00:02.82 +14:42:55.9 26 0.654 0.077 0.0 0.077 0.192 0.0 0.154 0.621 0.185 0 0 1
588015507658768464 00:00:03.24 -01:06:46.8 57 0.474 0.088 0.0 0.263 0.175 0.0 0.351 0.324 0.48 0 0 1
587727178449485858 00:00:03.33 -10:43:16.0 24 0.125 0.0 0.0 0.875 0.0 0.0 0.875 0.024 0.976 1 0 0
587730773351858407 00:00:03.46 +14:11:53.6 64 0.625 0.016 0.016 0.25 0.078 0.016 0.281 0.245 0.597 0 0 1
587731187277693069 00:00:04.12 +00:45:07.9 30 0.933 0.0 0.033 0.0 0.033 0.0 0.033 0.913 0.054 0 1 0
587727227837612116 00:00:04.18 -08:44:03.0 37 0.73 0.0 0.081 0.135 0.054 0.0 0.216 0.648 0.295 0 0 1
587727225153257606 00:00:04.60 -10:58:34.7 30 0.6 0.0 0.0 0.133 0.233 0.033 0.133 0.368 0.264 0 0 1
587727180596969574 00:00:04.60 -08:56:37.6 30 0.167 0.333 0.033 0.467 0.0 0.0 0.833 0.075 0.925 1 0 0
587731187277693072 00:00:04.74 +00:46:54.2 36 0.722 0.083 0.111 0.083 0.0 0.0 0.278 0.606 0.394 0 0 1
587727227300741221 00:00:05.17 -09:13:04.6 66 0.424 0.0 0.03 0.061 0.03 0.455 0.091 0.335 0.133 0 0 1
588015507658768548 00:00:05.54 -01:12:58.9 28 0.179 0.0 0.429 0.321 0.071 0.0 0.75 0.061 0.861 0 0 1
587727221413511423 00:00:05.70 +14:24:44.8 56 0.321 0.036 0.339 0.143 0.125 0.036 0.518 0.069 0.721 0 0 1
587730775499407519 00:00:06.11 +15:52:31.4 58 0.828 0.017 0.0 0.086 0.069 0.0 0.103 0.556 0.326 0 0 1
588015509806252152 00:00:06.67 +00:30:16.8 38 0.711 0.0 0.0 0.132 0.158 0.0 0.132 0.612 0.216 0 0 1
587727221413511425 00:00:06.70 +14:19:58.5 55 0.818 0.036 0.0 0.036 0.109 0.0 0.073 0.818 0.073 0 1 0
588015510343123099 00:00:07.12 +00:51:28.5 47 0.766 0.021 0.043 0.149 0.021 0.0 0.213 0.602 0.373 0 0 1
587727220876640496 00:00:07.35 +13:54:36.6 24 0.833 0.0 0.0 0.125 0.042 0.0 0.125 0.645 0.301 0 0 1
587730775499407506 00:00:07.37 +15:51:19.2 31 0.742 0.0 0.065 0.0 0.097 0.097 0.065 0.513 0.183 0 0 1
587730775499407527 00:00:07.59 +15:54:07.4 50 0.8 0.02 0.0 0.08 0.1 0.0 0.1 0.41 0.362 0 0 1
587730775499407394 00:00:07.62 +15:50:03.2 31 1.0 0.0 0.0 0.0 0.0 0.0 0.0 1.0 0.0 0 1 0
Note. — This table is published in its entirety in the electronic edition of the Journal. A portion is shown here for guidance regarding its form and content.
aSDSS ID. This table includes all galaxies for which spectra are available in SDSS Data Release 7.
bTotal number of votes for each object.
cFraction of votes for Elliptical (E), ClockWise spirals (CW), AntiClockWise spirals (ACW), Edge-on spirals (Edge), Don’t Know (DK), Merger (MG) and Combined Spiral (CS=E+CW+ACW) categories
dFraction of votes for Elliptical (E) and Combined Spiral (CS=E+CW+ACW) categories following the debiasing procedure described in section 3.1
eGalaxies flagged as ‘elliptical’ or ‘spiral’ require 80 percent of the vote in that category after the debiasing procedure has been applied; all other galaxies are flagged ‘uncertain’.
c©
2010
RA
S
,M
N
RA
S
000
,1
–14

10
Lintott
et
al
.
Table 3. Classifications of galaxies without spectra
Votesc
ObjIDa RA Dec Nb E CW ACW Edge DK MG CS
587730774425665700 00:00:01.28 +15:04:40.8 73 0.479 0.0 0.0 0.014 0.479 0.027 0.014
587727220876640877 00:00:01.86 +14:01:28.2 29 0.655 0.0 0.0 0.0 0.345 0.0 0.0
587727180060098742 00:00:02.15 -09:31:37.0 30 0.467 0.0 0.033 0.0 0.467 0.033 0.033
588015509806252142 00:00:02.28 +00:37:39.2 29 0.655 0.034 0.034 0.103 0.172 0.0 0.172
587731187277627683 00:00:02.96 +00:43:04.8 24 0.583 0.0 0.083 0.167 0.125 0.042 0.25
587731186203951111 00:00:04.44 -00:05:00.1 30 0.967 0.0 0.0 0.033 0.0 0.0 0.033
587730775499407505 00:00:04.96 +15:51:15.3 45 0.667 0.0 0.044 0.178 0.044 0.067 0.222
587730774962536621 00:00:05.96 +15:25:47.6 25 0.8 0.0 0.08 0.0 0.0 0.12 0.08
587730775499407504 00:00:07.22 +15:51:14.2 39 0.59 0.0 0.026 0.026 0.051 0.308 0.051
587730773888794648 00:00:07.73 +14:39:55.9 28 0.679 0.0 0.036 0.071 0.107 0.107 0.107
587727225690128558 00:00:08.42 -10:28:23.6 43 0.512 0.023 0.0 0.442 0.023 0.0 0.465
587727222487318704 00:00:08.83 +15:18:38.3 31 0.419 0.161 0.065 0.323 0.032 0.0 0.548
587727220876705929 00:00:09.64 +14:05:42.8 25 0.28 0.04 0.04 0.36 0.28 0.0 0.44
587727225690128563 00:00:11.29 -10:27:41.5 32 0.562 0.0 0.156 0.25 0.031 0.0 0.406
587727220876705935 00:00:11.92 +14:05:24.0 31 0.258 0.032 0.0 0.0 0.097 0.613 0.032
587731187814563978 00:00:11.97 +01:07:18.5 30 0.033 0.933 0.0 0.0 0.033 0.0 0.933
587730773351923943 00:00:13.06 +14:13:18.0 31 0.903 0.032 0.0 0.065 0.0 0.0 0.097
587727177912615045 00:00:13.11 -11:12:01.0 31 0.806 0.0 0.0 0.097 0.097 0.0 0.097
587730775499407560 00:00:14.32 +15:52:16.7 29 0.448 0.379 0.0 0.103 0.069 0.0 0.483
587727179523227783 00:00:15.54 -09:47:55.5 33 0.424 0.061 0.0 0.273 0.182 0.061 0.333
588015508732510387 00:00:16.12 -00:13:58.4 22 0.545 0.0 0.182 0.091 0.182 0.0 0.273
588015508195639455 00:00:16.19 -00:38:54.9 36 0.556 0.028 0.0 0.167 0.25 0.0 0.194
587727225153323052 00:00:17.96 -10:53:39.8 54 0.556 0.056 0.056 0.167 0.148 0.019 0.278
588015508195639399 00:00:18.69 -00:39:06.6 60 0.3 0.0 0.05 0.467 0.133 0.05 0.517
587727179523227873 00:00:20.63 -09:48:34.5 52 0.538 0.0 0.0 0.058 0.115 0.288 0.058
587730774425665840 00:00:21.04 +15:06:08.0 61 0.705 0.0 0.016 0.049 0.23 0.0 0.066
587730773351923979 00:00:21.19 +14:10:53.7 33 0.212 0.0 0.0 0.152 0.061 0.576 0.152
587730774425665839 00:00:21.73 +15:06:11.8 44 0.818 0.0 0.045 0.0 0.136 0.0 0.045
587727226763935920 00:00:22.65 -09:38:18.6 52 0.327 0.115 0.115 0.365 0.077 0.0 0.596
588015507658768569 00:00:24.16 -01:13:20.7 24 0.833 0.0 0.0 0.042 0.083 0.042 0.042
588015507658768511 00:00:24.17 -01:14:44.9 32 0.875 0.031 0.031 0.031 0.031 0.0 0.094
587727226227064993 00:00:25.55 -09:57:53.0 64 0.719 0.016 0.0 0.109 0.156 0.0 0.125
587727223024189787 00:00:25.61 +15:41:28.2 33 0.879 0.0 0.03 0.03 0.061 0.0 0.061
Note. — This table is published in its entirety in the electronic edition of the Journal. A portion is shown here for guidance regarding its form
and content.
aSDSS ID. This table includes all objects in the Galaxy Zoo sample for which spectra are not included in SDSS Data Release 7.
bTotal number of votes for each object.
cVote fractions for each category, as defined in the comments for Table 2.
c©
2010
RA
S
,M
N
RA
S
000
,1
–14

G
ala
xy
Z
o
o
11
Table 4. Measures of confidence
clean greater
ObjID RA Dec funclass fmisclass 〈∆p〉 σ∆p fmisclass 〈∆p〉 σ∆p
587727178986356823 00:00:00.41 -10:22:25.7 0.543 0.0 0.011 0.0090 0.0 0.0 0.0
587727227300741210 00:00:00.74 -09:13:20.2 0.458 0.0 0.203 0.115 0.267 0.199 0.112
587727225153257596 00:00:01.03 -10:56:48.0 0.811 0.046 0.367 0.192 0.2 0.174 0.09
587730774962536596 00:00:01.38 +15:30:35.3 0.348 0.0 0.0 0.0 0.0 0.0 0.0
587731186203885750 00:00:01.55 -00:05:33.3 0.83 0.0 0.302 0.122 0.053 0.055 0.026
587727180060098638 00:00:01.57 -09:29:40.3 0.799 0.0 0.233 0.127 0.103 0.073 0.044
587731187277627676 00:00:01.86 +00:43:09.3 0.852 0.0020 0.307 0.131 0.142 0.102 0.045
587727223024189605 00:00:02.00 +15:41:49.8 0.75 0.0060 0.333 0.137 0.399 0.196 0.076
587730775499407375 00:00:02.10 +15:52:54.2 0.629 0.0 0.035 0.019 0.0 0.0 0.0
587727221950382424 00:00:02.41 +14:49:19.0 0.762 0.036 0.365 0.178 0.457 0.244 0.113
587730774425665704 00:00:02.58 +15:02:28.3 0.774 0.027 0.375 0.177 0.529 0.261 0.117
587730773888794751 00:00:02.82 +14:42:55.9 0.982 0.0 0.134 0.061 0.018 0.029 0.014
588015507658768464 00:00:03.24 -01:06:46.8 0.869 0.0 0.0 0.0 0.216 0.107 0.045
587727178449485858 00:00:03.33 -10:43:16.0 0.734 0.0060 0.333 0.126 0.517 0.228 0.08
587730773351858407 00:00:03.46 +14:11:53.6 0.698 0.024 0.381 0.165 0.455 0.241 0.099
587731187277693069 00:00:04.12 +00:45:07.9 0.527 0.0 0.0 0.0 0.011 0.045 0.028
587727227837612116 00:00:04.18 -08:44:03.0 0.856 0.0 0.306 0.126 0.071 0.059 0.027
587727225153257606 00:00:04.60 -10:58:34.7 0.797 0.046 0.374 0.191 0.303 0.197 0.097
587727180596969574 00:00:04.60 -08:56:37.6 0.649 0.0 0.077 0.035 0.156 0.11 0.052
587731187277693072 00:00:04.74 +00:46:54.2 0.336 0.0 0.046 0.042 0.034 0.042 0.039
587727227300741221 00:00:05.17 -09:13:04.6 0.788 0.018 0.356 0.172 0.179 0.159 0.081
588015507658768548 00:00:05.54 -01:12:58.9 0.674 0.0 0.034 0.017 0.183 0.127 0.074
587727221413511423 00:00:05.70 +14:24:44.8 0.724 0.02 0.363 0.176 0.458 0.243 0.113
587730775499407519 00:00:06.11 +15:52:31.4 0.797 0.046 0.374 0.191 0.303 0.197 0.097
588015509806252152 00:00:06.67 +00:30:16.8 0.847 0.0 0.317 0.136 0.108 0.101 0.049
587727221413511425 00:00:06.70 +14:19:58.5 0.604 0.0 0.0 0.0 0.0 0.0 0.0
588015510343123099 00:00:07.12 +00:51:28.5 0.762 0.0080 0.351 0.147 0.132 0.11 0.05
587727220876640496 00:00:07.35 +13:54:36.6 0.752 0.011 0.357 0.154 0.235 0.158 0.071
587730775499407506 00:00:07.37 +15:51:19.2 0.503 0.0 0.204 0.12 0.322 0.214 0.128
587730775499407527 00:00:07.59 +15:54:07.4 0.688 0.022 0.361 0.174 0.538 0.272 0.125
587730775499407394 00:00:07.62 +15:50:03.2 0.167 0.0 0.0 0.0 0.0 0.043 0.046
Note. — These quantities are defined in section 3.2. This table is published in its entirety in the electronic edition of the Journal. A portion is shown here
for guidance regarding its form and content.
c©
2010
RA
S
,M
N
RA
S
000
,1
–14

12
Lintott
et
al
.
Table 5. Classification of mirrored images during bias study
Mirrored Mirrored 2
ObjIDa RA Dec Nvote E CW ACW Edge DK MG CS Nvote E CW ACW Edge DK MG CS
587727227300741210 00:00:00.74 -09:13:20.2 111 0.396 0.135 0.0090 0.405 0.054 0.0 0.55 118 0.458 0.136 0.025 0.347 0.034 0.0 0.508
587727180060098638 00:00:01.57 -09:29:40.3 121 0.876 0.0 0.017 0.033 0.066 0.0080 0.05 133 0.835 0.0 0.015 0.045 0.105 0.0 0.06
587727221950382424 00:00:02.41 +14:49:19.0 132 0.629 0.015 0.083 0.197 0.068 0.0080 0.295 120 0.625 0.0 0.142 0.208 0.017 0.0080 0.35
587731186203951111 00:00:04.44 -00:05:00.1 105 0.962 0.0 0.01 0.019 0.01 0.0 0.029 121 0.934 0.0 0.0 0.0 0.066 0.0 0.0
587731187814563978 00:00:11.97 +01:07:18.5 122 0.025 0.033 0.91 0.0080 0.0080 0.016 0.951 126 0.0080 0.024 0.905 0.0080 0.032 0.024 0.937
587727225690128575 00:00:14.09 -10:28:45.9 119 0.882 0.017 0.0080 0.067 0.025 0.0 0.092 100 0.94 0.01 0.0 0.02 0.03 0.0 0.03
587727223024189761 00:00:14.92 +15:43:42.6 114 0.368 0.0 0.175 0.404 0.035 0.018 0.579 108 0.417 0.0090 0.102 0.398 0.074 0.0 0.509
588015508732510387 00:00:16.12 -00:13:58.4 109 0.523 0.055 0.0 0.248 0.174 0.0 0.303 117 0.615 0.051 0.0 0.197 0.128 0.0090 0.248
587730773351923979 00:00:21.19 +14:10:53.7 103 0.282 0.0 0.01 0.029 0.068 0.612 0.039 126 0.206 0.0080 0.0 0.079 0.071 0.635 0.087
588015507658768569 00:00:24.16 -01:13:20.7 111 0.703 0.0090 0.018 0.054 0.09 0.126 0.081 119 0.731 0.0 0.0 0.067 0.101 0.101 0.067
587727223024189787 00:00:25.61 +15:41:28.2 105 0.781 0.01 0.0 0.029 0.181 0.0 0.038 108 0.778 0.019 0.0 0.037 0.157 0.0090 0.056
587730773888794890 00:00:38.69 +14:35:48.2 123 0.041 0.024 0.886 0.016 0.024 0.0080 0.927 116 0.043 0.0090 0.914 0.034 0.0 0.0 0.957
587727221950447853 00:00:38.70 +14:53:40.8 107 0.421 0.0090 0.0 0.495 0.028 0.047 0.505 128 0.43 0.023 0.023 0.469 0.047 0.0080 0.516
587727221413577017 00:00:43.55 +14:31:29.5 94 0.553 0.011 0.021 0.085 0.287 0.043 0.117 138 0.616 0.0070 0.014 0.094 0.261 0.0070 0.116
588015509269446865 00:00:46.32 +00:03:54.9 140 0.943 0.0 0.0 0.036 0.014 0.0070 0.036 123 0.878 0.0080 0.0 0.041 0.065 0.0080 0.049
587727177912680582 00:00:47.50 -11:06:12.7 127 0.756 0.0 0.0 0.087 0.11 0.047 0.087 113 0.823 0.027 0.0 0.0090 0.106 0.035 0.035
587727222487384226 00:00:50.19 +15:21:44.2 128 0.836 0.0 0.0 0.016 0.031 0.117 0.016 108 0.843 0.0 0.0 0.028 0.046 0.083 0.028
587727222487384242 00:00:51.41 +15:15:03.0 128 0.305 0.0 0.398 0.195 0.078 0.023 0.594 109 0.183 0.037 0.404 0.312 0.046 0.018 0.752
588015507658834049 00:00:51.75 -01:11:53.9 128 0.914 0.0080 0.0 0.031 0.047 0.0 0.039 116 0.922 0.0 0.0090 0.052 0.017 0.0 0.06
587727225690259613 00:00:52.08 -10:35:13.0 137 0.146 0.029 0.62 0.197 0.0070 0.0 0.847 120 0.092 0.025 0.75 0.133 0.0 0.0 0.908
588015507658833960 00:00:52.97 -01:10:20.5 112 0.562 0.0 0.0 0.027 0.161 0.25 0.027 127 0.472 0.0 0.0080 0.031 0.181 0.307 0.039
588015507658834052 00:00:53.51 -01:06:55.0 123 0.902 0.0080 0.033 0.033 0.024 0.0 0.073 118 0.958 0.0080 0.0 0.0080 0.025 0.0 0.017
587730775499473122 00:00:53.54 +15:54:19.1 116 0.095 0.681 0.0 0.069 0.138 0.017 0.75 116 0.121 0.629 0.026 0.043 0.112 0.069 0.698
587727223561126046 00:00:56.33 +16:05:33.7 118 0.831 0.0 0.017 0.059 0.085 0.0080 0.076 115 0.896 0.0 0.017 0.026 0.061 0.0 0.043
587727226764001349 00:01:05.80 -09:42:40.8 111 0.892 0.0 0.0090 0.018 0.081 0.0 0.027 105 0.8 0.0 0.029 0.048 0.105 0.019 0.076
587727225690259639 00:01:06.37 -10:24:00.9 134 0.037 0.791 0.03 0.112 0.0070 0.022 0.933 112 0.062 0.795 0.036 0.08 0.0090 0.018 0.911
587730774962602228 00:01:10.17 +15:27:15.3 107 0.85 0.0 0.0 0.0090 0.14 0.0 0.0090 132 0.795 0.023 0.0080 0.061 0.114 0.0 0.091
587727223561191596 00:01:16.49 +16:10:57.4 119 0.748 0.0 0.0 0.101 0.151 0.0 0.101 111 0.712 0.0 0.0090 0.063 0.216 0.0 0.072
587731185130340423 00:01:17.02 -01:01:58.2 125 0.872 0.0080 0.0080 0.048 0.064 0.0 0.064 119 0.79 0.0080 0.0080 0.059 0.134 0.0 0.076
587727178986487929 00:01:26.64 -10:11:51.9 115 0.8 0.0090 0.0 0.061 0.087 0.043 0.07 117 0.829 0.0090 0.0090 0.077 0.06 0.017 0.094
587727225690325075 00:01:31.48 -10:28:54.5 107 0.888 0.0090 0.0 0.019 0.084 0.0 0.028 111 0.865 0.0 0.0 0.018 0.117 0.0 0.018
Note. — Classifications of galaxies during the bias study. Galaxies were shown mirrored about the vertical and diagonal axes (‘Mirrored’ and ‘Mirrored 2’). For each transformation we provide the total number of votes (Nvote) and vote fractions
categories as defined in the comments for Table 2. This table is published in its entirety in the electronic edition of the Journal.
aSDSS ID. This table includes all objects included in the Galaxy Zoo bias study sample, as described in section 2.1.
c©
2010
RA
S
,M
N
RA
S
000
,1
–14

G
ala
xy
Z
o
o
13
Table 6. Classification of monochrome images during bias study
Monochrome
ObjIDa RA Dec Nvote E CW ACW Edge DK MG CS
587727227300741210 00:00:00.74 -09:13:20.2 108 0.565 0.019 0.083 0.278 0.056 0.0 0.38
587727180060098638 00:00:01.57 -09:29:40.3 123 0.854 0.0080 0.016 0.057 0.065 0.0 0.081
587727221950382424 00:00:02.41 +14:49:19.0 137 0.467 0.117 0.029 0.321 0.066 0.0 0.467
587731186203951111 00:00:04.44 -00:05:00.1 115 0.957 0.0 0.0 0.017 0.026 0.0 0.017
587731187814563978 00:00:11.97 +01:07:18.5 120 0.033 0.883 0.025 0.0080 0.025 0.025 0.917
587727225690128575 00:00:14.09 -10:28:45.9 108 0.907 0.0090 0.0 0.028 0.046 0.0090 0.037
587727223024189761 00:00:14.92 +15:43:42.6 115 0.27 0.113 0.0 0.539 0.078 0.0 0.652
588015508732510387 00:00:16.12 -00:13:58.4 111 0.649 0.0090 0.099 0.108 0.126 0.0090 0.216
587730773351923979 00:00:21.19 +14:10:53.7 119 0.168 0.017 0.0 0.101 0.067 0.647 0.118
588015507658768569 00:00:24.16 -01:13:20.7 117 0.598 0.06 0.0 0.094 0.197 0.051 0.154
587727223024189787 00:00:25.61 +15:41:28.2 130 0.654 0.0 0.015 0.069 0.262 0.0 0.085
587730773888794890 00:00:38.69 +14:35:48.2 97 0.041 0.918 0.01 0.0 0.021 0.01 0.928
587727221950447853 00:00:38.70 +14:53:40.8 119 0.294 0.0080 0.017 0.613 0.042 0.025 0.639
587727221413577017 00:00:43.55 +14:31:29.5 112 0.679 0.0 0.0090 0.062 0.214 0.036 0.071
588015509269446865 00:00:46.32 +00:03:54.9 128 0.891 0.016 0.0 0.047 0.0080 0.039 0.062
587727177912680582 00:00:47.50 -11:06:12.7 113 0.743 0.0 0.018 0.071 0.08 0.088 0.088
587727222487384226 00:00:50.19 +15:21:44.2 104 0.808 0.0 0.01 0.01 0.048 0.125 0.019
587727222487384242 00:00:51.41 +15:15:03.0 117 0.487 0.248 0.0090 0.171 0.077 0.0090 0.427
588015507658834049 00:00:51.75 -01:11:53.9 121 0.868 0.0080 0.017 0.041 0.066 0.0 0.066
587727225690259613 00:00:52.08 -10:35:13.0 120 0.242 0.592 0.017 0.117 0.033 0.0 0.725
588015507658833960 00:00:52.97 -01:10:20.5 114 0.395 0.026 0.0 0.088 0.228 0.263 0.114
588015507658834052 00:00:53.51 -01:06:55.0 98 0.98 0.01 0.0 0.01 0.0 0.0 0.02
587730775499473122 00:00:53.54 +15:54:19.1 127 0.118 0.0 0.717 0.031 0.118 0.016 0.748
587727223561126046 00:00:56.33 +16:05:33.7 96 0.771 0.021 0.01 0.042 0.156 0.0 0.073
587727226764001349 00:01:05.80 -09:42:40.8 105 0.886 0.01 0.01 0.038 0.057 0.0 0.057
587727225690259639 00:01:06.37 -10:24:00.9 118 0.059 0.017 0.831 0.068 0.0080 0.017 0.915
587730774962602228 00:01:10.17 +15:27:15.3 110 0.882 0.0 0.0 0.027 0.091 0.0 0.027
587727223561191596 00:01:16.49 +16:10:57.4 114 0.728 0.0090 0.0090 0.079 0.175 0.0 0.096
587731185130340423 00:01:17.02 -01:01:58.2 114 0.798 0.0 0.018 0.07 0.114 0.0 0.088
587727178986487929 00:01:26.64 -10:11:51.9 116 0.698 0.052 0.0090 0.078 0.129 0.034 0.138
587727225690325075 00:01:31.48 -10:28:54.5 128 0.891 0.0080 0.016 0.031 0.055 0.0 0.055
Note. — Classifications of monochrome images of galaxies during the bias study. We provide the total number of votes (Nvote) and vote fractions in
each of the categories, as defined in the comments for Table 2. This table is published in its entirety in the electronic edition of the Journal. A portion is
shown here for guidance regarding its form and content.
aSDSS ID. This table includes all objects included in the Galaxy Zoo bias study sample, as described in section 2.1.
c©
2010
RA
S
,M
N
RA
S
000
,1
–14

14
Lintott
et
al
.
Table 7. Combined votes including bias study data
Votesc
ObjIDa RA Dec Nvote,STD Nvote,MR1 Nvote,MR2 Nvote,MON Nvoteb E CW ACW Edge DK MG CS
587727178986356823 00:00:00.41 -10:22:25.7 59 59 0 0 0 0.61 0.034 0.0 0.153 0.153 0.051 0.186
587727227300741210 00:00:00.74 -09:13:20.2 355 18 111 118 108 0.479 0.017 0.121 0.338 0.045 0.0 0.476
587727225153257596 00:00:01.03 -10:56:48.0 68 68 0 0 0 0.735 0.029 0.0 0.147 0.074 0.015 0.176
587730774425665700 00:00:01.28 +15:04:40.8 73 73 0 0 0 0.479 0.0 0.0 0.014 0.479 0.027 0.014
587730774962536596 00:00:01.38 +15:30:35.3 52 52 0 0 0 0.885 0.019 0.0 0.058 0.019 0.019 0.077
587731186203885750 00:00:01.55 -00:05:33.3 59 59 0 0 0 0.712 0.0 0.0 0.22 0.068 0.0 0.22
587727180060098638 00:00:01.57 -09:29:40.3 405 28 121 133 123 0.854 0.012 0.0070 0.042 0.081 0.0020 0.062
587727220876640877 00:00:01.86 +14:01:28.2 29 29 0 0 0 0.655 0.0 0.0 0.0 0.345 0.0 0.0
587731187277627676 00:00:01.86 +00:43:09.3 38 38 0 0 0 0.5 0.0 0.053 0.289 0.105 0.053 0.342
587727223024189605 00:00:02.00 +15:41:49.8 26 26 0 0 0 0.423 0.0 0.0 0.577 0.0 0.0 0.577
587730775499407375 00:00:02.10 +15:52:54.2 62 62 0 0 0 0.355 0.016 0.21 0.323 0.0 0.097 0.548
587727180060098742 00:00:02.15 -09:31:37.0 30 30 0 0 0 0.467 0.0 0.033 0.0 0.467 0.033 0.033
588015509806252142 00:00:02.28 +00:37:39.2 29 29 0 0 0 0.655 0.034 0.034 0.103 0.172 0.0 0.172
587727221950382424 00:00:02.41 +14:49:19.0 420 31 132 120 137 0.564 0.114 0.019 0.245 0.052 0.0050 0.379
587730774425665704 00:00:02.58 +15:02:28.3 24 24 0 0 0 0.583 0.042 0.125 0.167 0.083 0.0 0.333
587730773888794751 00:00:02.82 +14:42:55.9 26 26 0 0 0 0.654 0.077 0.0 0.077 0.192 0.0 0.154
587731187277627683 00:00:02.96 +00:43:04.8 24 24 0 0 0 0.583 0.0 0.083 0.167 0.125 0.042 0.25
588015507658768464 00:00:03.24 -01:06:46.8 57 57 0 0 0 0.474 0.088 0.0 0.263 0.175 0.0 0.351
587727178449485858 00:00:03.33 -10:43:16.0 24 24 0 0 0 0.125 0.0 0.0 0.875 0.0 0.0 0.875
587730773351858407 00:00:03.46 +14:11:53.6 64 64 0 0 0 0.625 0.016 0.016 0.25 0.078 0.016 0.281
587731187277693069 00:00:04.12 +00:45:07.9 30 30 0 0 0 0.933 0.0 0.033 0.0 0.033 0.0 0.033
587727227837612116 00:00:04.18 -08:44:03.0 37 37 0 0 0 0.73 0.0 0.081 0.135 0.054 0.0 0.216
587731186203951111 00:00:04.44 -00:05:00.1 371 30 105 121 115 0.951 0.0030 0.0 0.013 0.032 0.0 0.016
587727180596969574 00:00:04.60 -08:56:37.6 30 30 0 0 0 0.167 0.333 0.033 0.467 0.0 0.0 0.833
Note. — This table is published in its entirety in the electronic edition of the Journal.
aSDSS ID. This table includes all objects in the Galaxy Zoo bias sample.
bTotal number of votes for each object.
cVote fractions for each category, as defined in the comments for Table 2.
c©
2010
RA
S
,M
N
RA
S
000
,1
–14