We’re only a couple of weeks from the observing session to look at some of the overlapping galaxies pointed out by the good people of Galaxy Zoo, so it’s high time for us to think about how to do this most effectively. We are scheduled for the nights of April 25-29, using the 3.5-meter WIYN telescope at Kitt Peak, Arizona. The timing is just right for the SDSS sample, since the northern galactic hemisphere that it covered most completely is up all night this time of year and the moon is waning and moving out of the way. Here are some outside and inside views of the telescope:

You can even see the current weather at Kitt Peak (as long as it’s daylight there) using this webcam view – WIYN is at upper right, on the mountain horizon directly above the nearby University of Arizona telescopes.We’ll be using a fairly new CCD camera called OPTIC, which is visiting from the University of Hawaii. This camera has chips with a special architecture allowing the accumulating image to be moved around on the chip in any direction by purely electronic means, so it can keep up with atmospheric motions as long as there is a bright enough guide star in the right field of view. The advantage of doing this is that electrons can be moved around much faster than the whole telescope, and without introducing any mechanical vibrations. This device will help us work out how far out in redshift we can pursue galaxies for this project using ground-based telescopes.

Now – which galaxy pairs do we start with? You all have furnished a magnificent sample, something like 875 overlapping pairs that look useful for dust measurements by the time I fold in the last few weeks’ harvest. I keep my target book for this project as a set of PDF files accessible from anywhere – for those who really want to see, the list is broken into four parts starting at right ascension zero, 10 hours = 150 degrees, 13 hours = 195 degrees, and 15 hours = 225 degrees. (These are really cool to flip through rapidly page by page). This huge sample gives us the luxury of being able to select our targets carefully, making sure to span ranges of foreground galaxy type and luminosity. An obvious starting point is the nearest galaxies, where we can see the most detail in the dust. We want to include pairs with both elliptical and spiral background galaxies; the dust detail is clearest when the background galaxy is smooth, but we also want to correlate with ultraviolet data from the GALEX satellite, and only spirals are bright enough in the UV to do this sensibly.If we think about choosing galaxies based on a score, we give them points for being bright and nearby, points for having two known and quite different redshifts, and points allocated according to how long the GALEX UV sky survey looked at that piece of the sky.

But we also want to pick a few higher-redshift galaxies, at z=0.1 or greater, to show just how well we can measure their dust from the ground. (Recall that we do have pending Hubble proposals to do some of these, but there are something like 960 competing proposals for next year’s observations and the oversubscription will be fierce).Our goal is to look at something like 50 pairs for an hour each over the 5 nights, and naturally we want to start with the most interesting ones in case the weather goes downhill. Chris will be blogging during the observing run, and certainly we’ll be getting action pictures. There are also plans for a writer to watch the action for a night and report the scene from a non-astronomer’s perspective. We hope to be able to do some analysis near real-time and get dust maps to show how well the data match our needs; if so you’ll be seeing some. I’m attaching a couple of images showing this on NGC 5544/5, which has been posted a couple of times in different threads (but from older data here). First the blue-light image shows the overall geometry. Then we have subtracted a symmetric model for the light from the foreground galaxy and divided by a symmetric model for the light of the background galaxy (both models being based on the data in the non-overlapping regions), giving a map of the absorption by dust in the overlap area. We hope to get a lot more of these.

As some of you may have noticed, our first paper has caught the eye of New Scientist (in fact they have written about us before). This is pretty cool considering that we effectively had a null result in our paper – concluding that the Universe seems to be relatively normal. The real excitement is due to the ‘people power’ that the Zoo has harnessed.

I think you need a subscription to see the full article, and however much I’d love to give out Chris’ account details I am sure this breaks certain rules! However, if you have been able to take a look at the full piece then you may be a little curious about the comments from Prof Michael Longo towards the end

It turned out people have a preference when picking orientation: despite the mirroring, 52 per cent of the galaxies were still described as anticlockwise. “Rather than the universe being odd, it might be that people are odd,” says Land. The team has submitted the findings to Monthly Notices of the Royal Astronomical Society (www.arxiv.org/0803.3247).

Longo, however, is unconvinced. The mirroring analysis was only carried out for 5 per cent of the galaxies studied and he believes this sample is too small to justify rejecting the original excess that users spotted, which corroborated the existence of the axis. “[Land and colleagues] have done an impressive job of organising the Galaxy Zoo project, but I believe their analysis is flawed,” he says.

It is really thanks to him that this part of the Zoo, looking into the spins of galaxies as opposed to just the morphology, took place (see here for more on the motivation of this part of the study). And he raises an interesting point in the article that I thought it’d be worth responding to…

The point was raised that because we only did the bias study on about 5% of the Galaxy Zoo sample then we cannot really comment at all on the level of bias in our full dataset. Indeed this sounds quite reasonable. We can see how the classifications for this random ~5% of the data behave when we flip the images, but how do we know that the full sample wouldn’t behave differently?

Well, I think there are two important points to be aware of. Firstly, with statistics we were able to confidently detect a bias in the classifications of these ~50,000 galaxies. The analysis we performed is discussed in some detail on this blog. It is a bit technical, but not only do we detect a bias effect, but with a method called resampling we further established the uncertainty in this result – the probability that the effect could just appear by chance.

For this the data was split into further subsets, and by looking at how the results varied between these groups we could estimate the overall uncertainty in our results. For example – if it turns out that removing a few pieces of information causes the results to vary wildly then this means that you have a huge uncertainty and cannot make strong final conclusions about the full dataset. In our case this method actually returned relatively small errors because even between subsets of the data the results did not vary much. When we formally computed the uncertainty (we used the jackknife method to be specific) we are able to detect the bias at the ‘3-sigma level’.

This kind of lingo is used a lot by scientists, and what we mean by ‘sigma’ is one standard deviation. This is a measure of how much numbers can be expected to vary by chance. Consider for example that you toss a fair coin a thousand times, and you want to know how many times you can expect to return a heads. Well obviously you’d expect 500 heads – but not necessarily exactly 500 as there will be some natural variance in the results. In this example it actually turns out that the number of heads you expect roughly obeys a Normal Distribution with mean of 500, and standard deviation of ~16. This means that if we repeated the experiment a number of times we expect 68.3% of the results to find the number of heads to be with 1 standard deviation of 500 (between 484-516), 95.4% within ‘2 sigma’ (468-532), and 99.7% within ‘3 sigma’ (452-548). If your experiment returned 450 heads from 1,000 tosses of the coin, then this would be unexpected at the ‘3-sigma level’ and would be highly unlikely – thus indicating that the dice is probably biased.

Well, similarly we found that our original and our flipped classifications were inconsistent at more than 3 standard deviations – and this means we can be sure at the 99.7% level than there is a bias effect in our study. This is what we mean by confidently!

But what about the full dataset? Well this is the second point – the bias-study galaxies were selected completely at random from the full dataset in order to get a representative sample of them. We have conclusively shown that there is a bias in the way people classify galaxies and hence the same effect should be present in the full sample. We cannot be 100% sure that the full sample would show exactly the same bias effect, but we can be over 99.73% sure (3 sigma) sure. In other words, for the bias effect to be a statistical fluctuation due to reanalyzing just 5% of the data, we would have to be very lucky (not quite LOTTO lucky, but more than BINGO lucky!). But once the bias effect is taken into account, the axis (or more specifically the excess of anti-clockwise galaxies) disappears. Alas!

For those of you who enjoyed our coverage of the AAS meeting back in January, I’m helping provide similar coverage of the UK’s National Astronomy Meeting currently being held in Belfast. The blog is here – come and join us.

Ever since it was first identified, Hanny’s Voorwerp has grabbed the attention of the Zookeepers and everyone else who comes across it. One reason we’ve been successful in getting such a wide range of observations over just a few months (and therefore why posts on here have been delayed!) has been that colleagues seem to find it equally compelling. So what is it? Our current best guess goes something like this:

A hundred thousand years ago, a quasar blazed behind the stars which would have already looked recognizably like the constellation Leo Minor. Barely 700 million light-years away, it would have been the nearest bright quasar, shining (had anyone had a telescope to look) around 13th magnitude, several times brighter than the light of the surrounding galaxy. This galaxy, much later cataloged as IC 2497, is a massive spiral galaxy which was in the process of tidally shredding a dwarf galaxy rich in gas – gas which absorbed the intense ultraviolet and X-ray output of the quasar and in turn glowed as it cooled. But something happened to the quasar. Whether it turned off, dropped to a barely simmering level of activity as its massive black hole became starved for gas to feed its accretion, or it was quickly shrouded in gas and dust, we don’t see it anymore.

But we see its echo. How could we come to such startling conclusions? An earlier blog entry showed some of our earliest data, when we already knew that the gas in Hanny’s Voorwerp was ionized in such a way that it must experience a radiation field of higher energy than normal stars can produce. In fact, it looks just like the pattern of emission given off by gas around the center of Seyfert galaxies, and on the outskirts of quasars and radio galaxies. This makes sense, except for the minor detail that we don’t see the active nucleus that should be there to light up the gas.

However, we could start from calculations done by astronomers trying to understand these objects, which could tell us how much radiation it would take to light up the Voorwerp. This wound up telling us how many ionizing photons there are per atom in the gas (known as the ionization parameter). That meant that we could find out how bright the missing core had to be if we could learn how dense the gas is.

Spectra are wonderful things – there is a pair of emission features from ionized sulfur atoms out in the red whose ratio depends on how often the atoms undergo collisions, and therefore on the density where they float. We had been contacting colleagues all over the map to see who might be doing spectroscopy in the red, and were fortunate to be put in touch with Nicola Bennert, who is a postdoctoral researcher at the University of California campus in Riverside. She was about to work for several nights with Lick Observatory’s 3-meter Shane telescope and a double spectrograph optimized to observe blue and red parts of the spectrum at once, and was intrigued by what we already knew of the Voorwerp.

She got a useful data set, in particular a very nice observation of the spectrum in red light. From this, we now know that the typical density of gas (for the pickier readers, that’s the RMS density) is no greater than about 15 particles per cubic centimeter – which means that the UV and X-ray luminosities of the object were somewhat less than a hundred billion times the Sun’s total energy output, in the range of quasars. (It was a nice extra feature that Nicola did her dissertation work on analysis involving measuring ionization parameters of gas in Seyfert galaxies, and she’s enthusiastically joined in the project).

From the features of sulfur and nitrogen, we also have good evidence that these elements are not very abundant in the gas – maybe 10% of the fraction seen in our part of the Milky Way, more like what we find in dwarf galaxies such as the Small Magellanic Cloud. So the gas looks more like something from a low-mass dwarf rather than something ejected from the center of a luminous galaxy like IC 2497.

Meanwhile, we had asked for a quick look with instruments on the Swift satellite. Swift is designed to detect gamma-ray bursts and follow them up quickly with X-ray and ultraviolet or visible-light observations, to localize them as fast as possible (“Swift – catching gamma-ray bursts on the fly” is their motto). Thus, Swift spends a lot of its time staring at the sky, especially parts of the sky that are easy to see from ground-based telescopes, waiting for something to happen. From being on one of too many NASA committees, Bil recalled that the Swift science team had realized that, since it didn’t matter exactly where they looked waiting for something to happen, they have a program to take requests. Usually these requests are for transient, time-sensitive events, but principal investigator Neil Gehrels agreed that our request would be appropriate.

So we crammed our whole science argument into 300 words and it was approved. Showing that “Swift” has more than one meaning, within a week we had our data. We had two questions in mind for its instruments. First, its X-ray telescope (known as the XRT) would easily see any active galactic nucleus, even a typical Seyfert galaxy. It saw – nothing. Second, we asked for ultraviolet images with the 30-cm Ultraviolet/Optical Telescope (UVOT). These were intended to tell whether the light outside of the bright gaseous emission lines came from stars or was reflected from dust particles. The distinction could be made because, as in the scattering that makes our sky blue, short-wavelength radiation scatters more effectively from interstellar dust. As an example, the blue reflected piece of the Triffid Nebula is bluer than the illuminating star – in fact bluer than any kind of star can be. And this is what we found in the Voorwerp. Filtering a slice of ultraviolet light that shouldn’t be much affected by the gas, we found the object to be ten times brighter in the mid-ultraviolet than in the shortest wavelength seen by the Sloan Survey. Not only does the gas see something bright, so does the dust.

UVOT image on the left, v band on the right

So now we have a bunch of pieces of the puzzle. Highly ionized gas, ionized by nothing we can see. Dust reflecting ultraviolet light from no apparent source. No central X-ray source, which makes it very hard to hide
something behind a cloud of gas and dust that leaves it visible from the Voorwerp. This was starting to look like a giant version of a phenomenon that astronomers have had to rediscover for several generations now – the light echo. Over the years, when we see a supernova explosion, bright nova, or a star that for some other reasons flares brightly, we often see reflections from foreground dust. If we trace the geometry of what dust we see at different times after the outburst, it must fall along an ellipsoid with the star at one focus and ourselves at the other.

It’s important that the echo has spectral characteristics of the exciting source. One team has used this fact to find locations of supernovae which we would have seen in the Large Magellanic Cloud centuries ago, as their reflections still come our way from larger and larger circles of foreground dust (see this very cool and very new press release). And now we are proposing that we’ve found the light echo from a faded quasar, which was there 50-75,000 years ago but is invisible now.

The importance of checking on this whole picture goes well beyond the admitted coolness value, or the flashiness of a proposal that we hope our colleagues who decide who gets to use big telescopes will look on with favor. We already know that quasars (and their relatives such as Seyfert galaxies) can undergo dramatic change on everything from cosmic timescales to human ones. We observe them to fluctuate in brightness, sometimes dramatically, over times as short as weeks. And at the outside, relations between quasars and mergers in some of their surrounding “host” galaxies wouldn’t exist if the quasars stay bright for much more then the nearly billion-year duration of a galaxy merger. (Only in astronomy and cosmology do we get to lump “mere” and “billion years”). In fact, we know that the whole population of quasars has changed over cosmic time – there used to be many more, and they grew brighter, in an era about 10 billion years ago. For that matter, the most powerful quasars must be temporary – if one were to shine at these enormous levels for all of cosmic history, even as miserly as gravitational energy can be about producing energy wile consuming mass, the central object would have long ago eaten its entire surrounding galaxy.

Of course we want to know more. There are more observations we can make which would test this idea, and tell us more about the nature of the Voorwerp and the history of the illuminating core. Chris headed up a proposal to map the gas with the OASIS system on the 4.2-meter William Herschel Telescope, so we could measure the Voorwerp’s Doppler shifts point-by-point and see whether there are correlated changes in strengths of emission lines that would show us brightening and fading of the central source (which would make rings in our view unless the gas has a very odd structure). And there was the Hubble proposal, which would take high-resolution images of the gas in two emission lines and then look in filter bands between them to see whether the Voorwerp has stars. Actually, with all the reflecting dust, we hope mostly to see star clusters, to tell whether it started life as a dwarf galaxy. And we want to take a really close look at the nucleus of IC 2497, using Hubble’s exquisite resolution to isolate the light from its innermost region in search of any gas that is lit up by even a weak active nucleus. Speaking of the nucleus of IC 2497, Bill is even as we write working to complete a proposal to use Chandra to see if we can tease out any X-rays from a now-quiet AGN. We’ve also requested time in the radio to see if we are only seeing part of a much larger structure.

So here we have a new possibility – of watching the history of a quasar either flaring up, practically turning off, or being hidden over a time span that we’ve had no other way to examine. The pattern of light emitted by gas in Hanny’s Vooorwerp, and the way its dust reflect the quasar light, should be able to trace the history of its decline. Never mind heading back to the future, we can go onward into the past. Once in a while, we have the opportunity to do what paleontologists can do only in the movies.

(Chris and Bill weren’t sure who should blog this. So in the spirit of Galaxy Zoo, we both did.)

Several weeks ago, we issued a challenge: re-create the Galaxy Zoo poster in a new way to tell the story of Galaxy Zoo yourself. When I posted this, I wasn’t sure if anyone would find it interesting, but as always, your creativity is amazing. The first submission is in, from Julia, and it is breathtaking. Behold:

Julia’s Galaxy Zoo poster remix (1.2 MB JPG)

I think you’ll agree that this is a big improvement over the text-heavy original, and I think it’s one of the most amazing science posters I’ve ever seen. Congratulations to Julia on an amazing piece of science/art! (Julia, if you’re reading, please say hello in the Comments section.) If anyone else is interested in remixing the poster, send me a message in the forum (I’m zookeeperJordan).

For the past few months on this blog, we’ve been talking about the science of Galaxy Zoo – what your millions of classifications have revealed to us about the way the universe works. Right now, as Steven and I described Friday and Monday, the members of the Galaxy Zoo team are writing papers announcing our science results, and offering feedback on each other’s papers.

But of course, Galaxy Zoo has become much more than just a science project. The site has become an Internet phenomenon, and for the next few posts, we’d like to focus on some other aspects of the Galaxy Zoo phenomenon. Today, we wanted to talk about the thing that makes everything else work – the site itself.

Without a good-looking and well-functioning website, we could have never invited all of you to participate in this project, and you could not have generated the excellent scientific dataset that you have generated. The site was designed by two professional web designers: Phil Murray and Dan Andreescu. Galaxy Zoo is now proudly listed as a featured project on Phil’s web site.

Phil designed the look and feel of the site, and Dan wrote the code that allowed the website to take your input and write it into a database of classifications. Dan left the project in late 2007, and Danny Locksmith has taken over the coding.

The best way to tell the story of Galaxy Zoo’s design is to let Phil and Danny tell the story themselves. So here is Phil, talking about how he designed the layout of Galaxy Zoo:

GZ1.0 Visual Design

I was asked by Chris Lintott to design the Galaxy Zoo logo and web site in March 2007, and I realised early on that this had the potential to be a hugely successful project — little did we know just how successful it would be! I was given a completely free rein to handle the visual design of both the logo and the web site.

The Galaxy Zoo Logo

I wanted to create a visually appealing logo that would work in several formats – web and print. It had to be flexible enough to work as a standalone logo or to be incorporated into an overall page design – as is the case with the web site. The graphic part of the logo is in fact based on a Hubble image of Supernova 1987A Rings, which seemed to fit very neatly into the text of ‘GALAXY ZOO’ to form an official logo. A variation was developed for both web and print use.

The Galaxy Zoo Web Site

It seemed obvious that part of the attraction of the GZ1.0 project to non-astronomers was the sheer beauty of the galaxy images, plus the fact that many of these images had never previously been seen by human eyes. So I wanted to maximise impact right up front on the Home Page of the site by using a large galaxy image as a main background to the page and to carry this theme through into the inner pages. I wanted all text to sit on a semi transparent screened background to give the impression of depth on the page.

Choosing a colour palette was relatively straightforward given the colours within the logo, hence the basic black, grey, orange and gold colour scheme. It was decided to go with a slightly shallower header graphic for inner pages with all top navigation shown horizontally and any secondary navigation to be contained in a left column (as is the case in the Analysis Page). I decided on a fixed width solution catering for a minimum screen resolution of 1024×768 pixels.

When it came to the buttons for the Galaxy Analysis page, I spent some time designing what I hoped would be generic buttons for the various options on offer (Spiral Galaxy – Anticlockwise, Clockwise and so on). The intention was to try to design buttons that would not influence the decision making of the Galaxy Zoo visitor but also that they would be intuitive to use. In fact it quickly became apparent that having designed the buttons to look like they were part of an ‘online  game’, was a feature which also helped with the appeal and overall usability of the site. The feedback and data received as part of GZ1.0 has given us some valuable information about how to present these and other buttons for GZ2.0.

As for building the site, I constructed all the pages as HTML templates, which were then integrated into the ASP.Net web programming environment by the excellent Dan Andreescu. Danny Locksmith has taken over the ASP.Net duties since late 2007.

I think you’ll agree that the site that Phil did a great job – he created a really beautiful site that was easy to navigate. Now here is Danny, talking about how he took over the coding from Dan:

Most of the ASP programming was decided on before I got involved. My task so far has been to try to understand how someone else thought it should work!

In effect this is how it works:

Your login to the site, your user preferences, etc. are all controlled by the .NET 2.0 framework. The site uses a template that provides the basic logic involved in recording your clicks and ensuring that the right person is credited with each classification. The persistant data is stored in a database.

When you load the Galaxy Analysis page, a galaxy is selected randomly and displayed on the page. Next to the galaxy’s image is a the Galaxy ID, which is a hotlink to an SDSS page where you can view details of the galaxy – its spectrum and a zoomable picture, etc. Watch for changes to this in GZ2!

Next to the galaxy image is a custom control which has the various buttons you can click to classify the galaxy. Since we learned about the anticlockwise bias, various theories have been put forward about to explain it – one of them is the design and layout of these buttons. Another problem here was that people tended to click the button several times, recording several results. This was worked around by only allowing one classification per galaxy. Yet another potential problem was that you could easily make an error, but you could not go back and fix it. Look for changes in GZ2!

Once you click a button, your classification, your user ID, the date and time, etc. are recorded in a database. The data that is stored was designed to answer specific questions, and the scientific papers which are to be published soon. With the advent of the bias testing phase, additional information was stored in the same database – the way the image displayed had been transformed.

In GZ2, the data collected will be more generic, and will create a very comprehensive catalog of galaxy information, almost certainly the biggest ever. To a great extent, the inner workings of the site are defined by the various scientists involved in the project. It is very much designed by the entire team, and as such my task is to ensure that the finished site meets all the goals of the team, and at the same time is pleasureable to navigate and to use.

You will be able to decide if I was successful when GZ 2 is launched!

I’ll add just two things to what Phil and Danny said:

1) The servers that run Galaxy Zoo are in the Physics and Astronomy building at Johns Hopkins University in Baltimore, Maryland, USA. (Here is the building in Google Maps – the Johns Hopkins lacrosse stadium is just to the north, and the building across the winding street is the Space Telescope Science Institute).

2) One of the things I find amazing about Galaxy Zoo is that no member of the team has ever met all the other members face-to-face. Chris has come closest – he has met everyone except Jan and Alainna, who do IT support for the servers at JHU. In addition, 8,549 km separates Anze in Berkeley, California from Chris, Kate, and Kevin in Oxford. The Galaxy Zoo team could not have existed without the Internet, and communication tools that allow us to work together productively on different continents.