This Diagram is Better than 183,487 Images

When it comes to immediate and widespread appeal, astronomical diagrams have it tough. There’s a reason we have Most Awesome Space Images of 2012, but not “Astronomy’s coolest diagrams 2012.” But arguably, diagrams (more concretely: plots that help us visualize one or more physical quantities) are the key to understanding what’s up with all those objects whose colorful images we know and love.

To be sure, some diagrams have become quite famous. Take the Hubble diagram plotting galaxies’ redshifts against their distances: Its earliest version marks the discovery that we live in an expanding universe. A more recent incarnation, which shows how cosmic expansion is accelerating, won its creators the 2011 Nobel prize in physics.

Another famous diagram is the Hertzsprung-Russell diagram (HR diagram, for short, shown above.) A single star doesn’t tell you all that much about stars in general. But if you plot the brightnesses and colors of many stars, patterns begin to emerge – such as the distinctive broad band of the “main sequence” bisecting the HR diagram diagonally, the realm of the giants and supergiants to its upper right and the White Dwarfs below on the left.

When astronomers first recognized those patterns, they took the first steps towards our modern understanding of how stars evolve over time.

The first HR diagram was published by the US astronomer Henry Norris Russell in 1913 (or at least described in words, if you look at the article); Hubble’s first diagram in 1929. Off the top of my head, I cannot think of any famous astronomical plot with more recent roots.

But that doesn’t mean there aren’t some plots that by rights should be famous. Here’s my rendition of what, back in 2003, must have been one of the first comprehensive examples of its kind (from this article by Blanton et al. 2003). The diagram shows the colors of many different galaxies, and how frequently or less frequently one encounters galaxies with those particular colors:


If you’re not familiar with this type of plot, it’s best to think of the vertical lines as dividing the diagram into bins – think “glass cylinders you can put stuff in.” Next, obtain a sample of images of distant galaxies. Here are some that I retrieved with the Skyserver Tool kindly provided by the folks who produced the Sloan Digital Sky Survey (SDSS) — a huge survey that, in its latest data release, lists more than 1.4 million galaxies:

Galaxies from the Sloan Digital Sky Survey.
Galaxies from the Sloan Digital Sky Survey.

If these images are less detailed than what you’re used to, it’s because the galaxies are very far away even by extragalactic standards — their light takes almost 1.3 billion years to reach us. Even so, you can readily distinguish the galaxies’ different colors.

With that information, back to our (glass) bins. Think of the differently colored galaxies as differently colored marbles. Each bin accepts galaxies of one particular shade of color – so put each marble into the appropriate bin! As you do, some of the bins will fill up more, some less. The colored bars indicate each bin’s filling level. On the scale to the left, you can read off the corresponding numbers. For instance, the best-filled bin contains a little more than 5 percent of all the galaxy-marbles.

Now that you know how to read the diagram, let’s remove the extra vertical lines. In a paper published in an astronomical research journal, this is what a “histogram” of this kind would look like:

Galaxy distribution by color. Credit: Markus Pössel
Galaxy distribution by color. Credit: Markus Pössel

I’ve left the coloring in even though you’d probably not find it in an astronomical paper. The astronomers’ own measure of color, denoted “g-r” on the horizontal axis, is a bit technical — let’s ignore those details and stick with the colors we see in the diagram.

To fill the bins in this particular diagram, the astronomers from the SDSS collaboration sorted 183,487 galaxies from their survey by color.

So what does the diagram tell us? Evidently, there are two peaks: one near the bluish end on the left, one near the reddish end on the right. That indicates two distinct types of galaxies. Galaxies of the first kind are, on average, of a bluish-white color, with some specimens a little more and some a little less blue (which is why the peak is a little broad). Galaxies of the other kind are, on average, much redder.

A galaxy’s color derives from its stars. A bluish galaxy is one with bluish stars. Bluish stars are hotter than reddish ones. (Think of heating metal: It starts out a dull red, becomes orange, then white-hot; if you could make metal even hotter, it would radiate bluish.) Hot stars are more massive than cooler stars, and they live fast and die young — the most massive ones die after much less than a million years, a fleeting moment compared with our Sun’s estimated lifetime of ten billion years. For a galaxy to glow an overall blue, it must have a steady supply of these short-lived bluish stars, producing new blue stars in sufficient quantities as the old ones burn out. So evidently, the galaxies of the bluish kind are continually producing new bluish stars. Since there is no known mechanism that makes a galaxy produce only bluish stars, we can drop the qualifier: these galaxies are continually producing new stars.

The reddish galaxies, on the other hand, produce hardly any new stars. If they did, then by all we know about star formation there should be sufficient bluish stars around to give these galaxies an overall bluish tint. Without any new stars, all that is left are long-lived, less massive stars, and those tend to be cooler and more reddish.

The existence of two distinct classes of galaxies — star-forming vs. “red and dead” — is a driving force behind current research on galaxy evolution in much the same way the HR diagram was for stellar evolution. Why are there two distinct kinds? What makes the bluish galaxies produce stars, and what prevents the reddish ones? Do galaxies move from one camp to the other over time? And if yes, how and in which direction? When you read an article like this about the care and feeding of teenage galaxies, or this one about galaxies recycling their gas, it’s all about astronomers trying to find pieces of the puzzle of why there are these two populations.

This diagram clearly deserves wider public recognition. And no doubt there are many other, equally under-appreciated astronomical plots. Please help me give them some of the recognition they deserve: Which diagrams have done the most to increase your understanding of what’s out there? Which have surprised you? Which have sent a thrill down your spine? Please post a link or a description, and let’s see if we can create a “Top 10” list of astronomical diagrams. And who knows: We might even try for an “Astronomy’s coolest diagrams 2013” at the end of the year.


Additional information about how the two-peak galaxy diagram was made, including different versions for download and the python script that produced it, can be found here. If you do want to know about the technical details about the color: The values on the x axis correspond to g-r, where g is the star’s brightness (expressed in the usual astronomical magnitude system) through one particular greenish filter and r the brightness through one particular reddish filter. Details about the ugriz filter system used can be found on this SDSS page. And in case you’re worrying about the effect cosmic redshift might have had on the galaxies in the sample: the astronomers took care to compensate for that particular effect, correcting the colors to appear as they would if each of the galaxies were so far away that its light would take 1.29 billion years to reach us (that is, at a cosmic redshift of z=0.1).

Many thanks to Kate H.R. Rubin for pointing me to the galaxy diagram and for helpful discussions.

26 Replies to “This Diagram is Better than 183,487 Images”

  1. The first diagram looks interesting, but the small type is too small to read on the image posted.

      1. It does help quite a bit, but some just becomes a bigger fuzzy spot at higher magnification.

  2. My first thought when I see the plot is to think that half the galaxies out there are blue shifted and half are red shifted. A likely hypothesis explaining this, is that our relative motion in the universe explains everything. I know this is not the case, but from the plot alone, there is no way to disprove this hypothesis.

  3. I love the populated HR diagram!

    So I mainly look at cosmological diagrams (of implicit astrobiology interest re the age et cetera of the universe and the first planets). But that means I happen to like the related SDSS galaxy map of large scale structures and their homogenization at even larger scales.

    Of course it’s a map, not a quantified diagram – so sue me. =D

  4. is it bad that I only really pay credence the comments posted by Torbjörn Larsson and lcrowell here on Universe Today? I’m not saying I don’t read other comments and consider them if they are appropriate, but it’s clearly obvious who the big guns are around here. Well done chaps, and I will continue to observe and appreciate your intellect and insight on the various subjects along the way

  5. P.S. who created that HR diagram – it’s a beauty, easily the best one I’ve seen.

    1. It’s from The Cosmic Perspective or The Essential Cosmic Perspective textbook by Jeffrey Bennett, Megan Donahue, Nicholas Schneider, and Mark Voit.

  6. @hibida:disqus : Admittedly, the relevant information is hidden in the endnotes. To create this diagram, cosmological redshift was corrected for; the colors in that diagram are those the respective galaxies would have if they all were at a redshift z=0.1.

  7. @Al Wilson: The Kepler histogram is indeed a good example! In particular if my gut feeling is right that of those media reporting on that particular release, almost all went with an artist’s impression of an exoplanet, as opposed to a histogram.

    For the HR diagram, I just searched for what earlier Universe Today article had mentioned it, and linked to the image there:

    I agree that it is a very nice diagram, and that the effort put into it by its creator(s) should not be anonymized.

    The Wikimedia Commons page (found via Google image search) traces it back to the European Southern Observatory (ESO); on their pages, I couldn’t find exactly that version, but a very similar, less annotated one:

    ESO is good news, in this regard, as they have a quite liberal policy about licensing – they release their images under the Creative Commons Attribution 3.0 Unported license (so I’ll ask Nancy to make the proper attribution).

    1. I agree about the Kepler histogram (and subsequent versions that will surely be coming out).

      Also, the HR diagram is from The Cosmic Perspective and The Essential Cosmic Perspective textbooks by Bennett et al. I use the second one in the astronomy course I teach.

      1. Are you sure it’s from the Bennett et al. books, or do they use the ESO image? (Presumably they did not produce all the images they use themselves?) I don’t have the books at home, so cannot, right now, have a look myself.

      2. I don’t have the hard copy textbook at hand right now, either, but I looked at the credits in the digital version, which I *can* get at, and there is no reference to ESO for that image (part of Fig 11-10 in the ECP book). It has been part of those books since 2002, when I first started using them.

      3. I checked with Jeff Bennett about the figure. It was an original diagram for the textbook. One of the other authors, Mark Voit, created it together with artists working for the publisher.

  8. This post definitely caught my eye! I remember when I first made this figure (below) that again shows there are two main types of galaxy: red and blue. Not only are they identifiable by their colour however, but also by how their light is distributed.

    The vertical axis is the same as your figure above; the colour of the galaxy, with blue galaxies at the bottom and red galaxies at the top. The horizontal axis in this figure shows how the measured brightness of a galaxy varies with the distance from its center to its outside edge. A low number means the brightness in the middle is fairly constant, and then drops off suddenly at a given distance (imagine a shape a little bit like a top hat!). These galaxies are said to not be very concentrated. A high number means the brightness is very ‘peaky’ in the middle, and slowly stretches out to very large distances (imagine the shape of a circus top!). These galaxies are said to be very concentrated.

    When you put all galaxies into this 2D plane, you can clearly see two big bumps that correspond to the two types of galaxy you show above (red and blue). To add to this, the colour of the data points shows how massive each galaxy is (that is, the ‘weight’ of the galaxy).

    We see that the reddest, most concentrated galaxies are also the most massive, whereas the bluest, least concentrated galaxies are also the least massive. A wonderful example of natural correlations present in nature!

  9. Great post! It is really worth trying to make the public to understand that astrophysics is not only nice images.

    In my case, the best diagram is that of Fig. 2 of my paper (see also the attachment). It shows clearly how relativistic jets of Galactic and Extragalactic compact objects can be unified. The scaling factors are simply due to the mass of the compact object and its accretion power.

    1. The fact that jets scale up so nicely is very cool. That diagram, only shows the data points – is it possible to add a curve that visualizes the scaling?

  10. Figure 13.4 on p. 300 of ‘The Milky Way as a Galaxy’, by Gilmore, King & van der Kruit (adapted from Wyse and Gilmore, 1998 Astron. J., 95, 1404).

  11. It is amazing how Space Monkeys are supposedly able to discover the most incredible things in galaxies far, far away but can”t find something in our own solar neighborhood .. How is it so????? Billions dollars telescopes are used to look farther that they should? Kuiper Belt to Oort cloud 45 degrees… You can start there!!!! Monkeys

    1. Uhm – if “How is it so?” was meant as a serious question: Because it’s not only distance that matters? Because a very bright object at a great distance can be much easier to detect than a dark object close by?

Comments are closed.