I’m happy to report that this week we submitted a new paper (to the Monthly Notices of the Royal Astronomical Society) using your bar classifications from Galaxy Zoo 2. The paper appears on astroph this morning (paper link here).

The title of the paper is shown below.

Ramin Skibba (who wrote one of the early Galaxy Zoo papers on the morphology-colour-environment relations) led this study into how the chance of finding bars and bulges in disc galaxies depends on the environment the galaxy is found in. We’re interested to find this out – particularly how the chance of finding a bar depends on environment – to help with the interpretation of our finding that redder disc galaxies are much more likely to host bars (which has now also been seen in other samples, and illustrated beautifully in one of the first figures from the paper, below – look at the blue lines which show the probability of finding a bar in a disc galaxy against it’s colour, with red to the right). Previous studies of the effect of environment have not had as many galaxies as we have (thankyou again), and have come up with contradictory answers as a result (for references see Ramin’s paper), so being able to do this study with a huge sample of galaxies with visually classified bars has been really fantastic.

Ramin used a technique called marked correlation functions to look for the effect of environment on the chance of a disc galaxy having a bar (or bulge). A correlation function (in astronomy at least) gives the probability that you’ll find two galaxies separated by a particular distance. We plot the “correlation” as a function of this separation (usually called r). The higher the correlation function (w(r)) is at some separation the more likely it is that you’ll find galaxies clustered on that scale. Practically this is measured by counting pairs of galaxies separated by each separation, r, and comparing it to the number of such pairs you’d find in a completely randomly arranged sample of galaxies.

The only difference with a marked correlation function is that in addition to simple counting of pairs, each galaxy is weighted by some “mark” (or number from zero to one). For example, in our paper we mark the galaxies by the fraction of you who classified the galaxy and clicked that you saw a bar – which we called p_bar. So two galaxies with a high fraction of “bar clicks” separated by a distance r would count more in the marked correlation function than two with a low fraction of “bar clicks”.

Our main result is shown above. The top panel shows the correlation functions of the whole sample, both with and without the weighting by p_bar. The bottom panel shows the ratio between those two correlation functions – so basically it shows the scales at which you’re more likely to a pairs of barred disc galaxies than a pair of any two random disc galaxies. This shows us that on the smaller scales barred disc galaxies are more strongly clustered than disc galaxies in general. The clustering peaks at about 0.4 Mpc/h (where Mpc/h are fantastic astronomers distance units which show our mistrust of the value of the Hubble constant; h = H0/100 km/s/Mpc, so is probably about h=0.7; so 0.4 Mpc/h is probably about 0.6 Mpc, or about 2 million light years) which is interestingly about the scale where most galaxies will be satellites of larger halos (ie. galaxies in groups). It is also interesting that on the very smallest scales the ratio drops back down to almost one – showing that for very close pairs disc galaxies are not more likely to have a bar.

Ramin (and of course also Steven using a different method) had previously shown that redder disc galaxies are more strongly clustered than bluer ones, so we had to wonder how much of the extra clustering of barred disc galaxies was just due to them being preferentially in red discs…. Ramin tested this using a really neat little trick, which he called “shuffling the marks”. Basically he took all galaxies of a similar colour and randomly shuffled the p_bar number within the group. If the bar-environment correlation was entirely due to the colour-environment correlation doing this for all the galaxies should result in no change in the marked correlation function. And in fact this is almost what we saw (below: the red triangles almost match the white circles).  On most scales the bar-environment correlation can be explained by red discs being more strongly clustered, except right around the 0.4 Mpc/h scale (which likely represents galaxies in small groups) where in simple terms – we see some excess clustering of red discs with bars over the clustering of red discs in general.

This finding suggests that something about the group environment may be triggering bar formation. In addition the downturn at small scales (if real) suggests that once galaxies get really close and start interacting they are not more likely to have a bar.

There is a whole lot more information in Ramin’s paper, which as usual is an excellent and densely packed piece of work, so I hope you’ll forgive me for stopping here after explaining only the main result on the barred disc galaxies.  It’s really been a pleasure working with Ramin on this study, and I just wanted to give you a flavour of the interesting results we have found.

The relative importance of bars (and other forms of “secular evolution” – generally used to mean slow internal processes acting on a galaxy) is turning into a hot topic in astronomy; with an entire session at the upcoming International Astronomical Union General Assembly (being held in Beijing in August 2012) devoted to “Galaxy Evolution through Secular Processes” so I expect you’ll be hearing a lot more about this.

To finish up a pretty (ish) picture. The top row shows examples of galaxies with pbar=0 (ie. no-one could see a bar), then pbar=0.2; pbar=0.5 and pbar=1.0 (all of you saw a bar).