Why water + E.Coli = superfluid is too good to be true (or the importance of fact checking for science writers)

I woke this cold, foggy Sydney morning to see a tweet that immediately raised a minor blip in my ‘bullshit detector’ (something all good scientists should be equipped with).

Nature Bacteria Superfluidity tweet

Nature Bacteria Superfluidity tweet

Nice click-bait, so I took a look… The first sentence reads “Swimming bacteria can thin out an ordinary liquid and, in some cases, turn it into a zero-viscosity superfluid, researchers report.” This seemed way too ‘good to be true’, my bullshit detector went directly from blip to full-on claxon mode.

Feeling a bit feisty from the cold, I decided to question it… here was the response:

Twitter debate

Twitter debate with author of Nature News article…

I’ll return to this response later, but the viscosity becoming negative was like waving a red rag at a bull… Superfluids don’t have ‘negative’ viscosity; there’s more to this story than is being sold. So, with a big caffeine hit down the hatch (red bull of course), off I went to the journals to look up the relevant (but sadly paywalled) articles.

It wasn’t long before… “Tell me this is one of your simulations… Alright, flush the bombers, get the subs in launch mode. We are at Defcon 1.”

Here’s my response…

A superfluid is a liquid that has zero viscosity and can therefore exhibit dissipationless flow. This means that one can, in principle, start a flow of the liquid and it will flow forever. The classic example is liquid helium-4, which undergoes a superfluid transition at 2.2 Kelvin. Superfluid helium can do some remarkable things like flow up walls to escape a container or through tiny holes that other fluids can’t get through (a nightmare for fridge-jocks like me but that’s another story).

In this latest experiment, the authors are looking at a fluid that is mostly water, but which contains between 0.1 and 1% by volume a population of live bacteria E. Coli that convert chemical energy (food) into directed swimming motion by rotating structures called flagella. Swimming bacteria are a pretty hot topic right now for many reasons extending from how the microscopic molecular motors that drive flagella rotation assemble and operate, through to how collective motions of large populations of these ‘active swimmers’ show complex structure. This research is at the latter end of the spectrum.

Swimming for bacteria is very different to swimming for us humans because of the massive difference in scale. Bacteria live at a size scale where the stickiness of water molecules, and their relentless jiggling due to thermal motion, changes the way the fluid appears to a swimmer — it seems more like swimming in washing machine filled with hot motor oil than a nice calm lake. The result is that the optimum way to swim is very different. If we made a human-sized bacteria and put it in a swimming pool, it would be like a car stuck in the mud, spinning its flagella (wheels) and going nowhere.

What’s different here is the viscosity, which is a measure for how much resistance a liquid shows to a force that tries to make it flow. If a liquid has a high viscosity you have to put a lot of effort into making it flow. A good example of a high viscosity liquid is tar pitch, which is so viscous that it looks like a solid and takes decades to flow through a funnel under gravity (cue link to one of my favourite experiments of all time). Honey is more viscous than water, both are more viscous than air. At the lowest end of the spectrum is liquid helium where, if you make it cold enough, the viscosity suddenly becomes exactly zero.

Back to the topic, which is the bacteria study. The work was done by Hector Lopez and colleagues at Universite Paris-Sud in France and the idea was to measure the viscosity of water with these swimming bacteria in it. The reason is that the stickiness of water at these scales means not only that it changes the way that the swimmers have to swim, but that the swimmers can in turn change the viscosity of the liquid. The way this works is that if there’s some collective behaviour amongst the swimmers, they can drag the liquid with them, making it look, externally, like it flows differently to how it would if the swimmers weren’t there. The big picture here is to try and use measurements of viscosity as a way to look at patterning and structure in collective swimming behaviour in these bacteria. It’s a clever and interesting way to use physics to attack a biological problem.

To help you all understand the experiment, I’m gonna show you the cool spa in my apartment complex (lucky me, hey)….


My spa, which conveniently looks a lot like a rheometer.

The authors use a device for measuring viscosity called a ‘rheometer’ and it looks a lot like my spa. There’s an outside cylindrical ‘cup’ that holds the liquid and an inside cylindrical ‘bob’, these are concentric. The cup can be rotated either way at a given speed using a motor (which would make my spa pretty cool fun). The bob is connected to a wire that enables a rotational force (torque) to be applied. The idea is, you start the cup rotating. The resistance between the liquid and the cup wall will make the liquid try to flow with it. Eventually this flow will start to rotate the bob with the liquid. You then get the viscosity of the liquid by measuring exactly how much torque you need to apply so that the bob is just at the point where it doesn’t rotate (add less torque it rotates with the liquid, add more torque it counter-rotates — the balance is a little like what you have in the clutch in a manual car). If the liquid has a high viscosity, e.g., tar, this torque to stop the bob rotating is huge. If the liquid is low viscosity, e.g., water, then the torque is lower, and if it’s a superfluid, then the torque is zero.

Pretty easy experiment right. But there’s a really deceptive aspect to it that the authors have exploited to ‘sell’ their paper, and the editors and journalists bought it hook, line and sinker (or should it be hook, link and stinker?). I know this hype originates with the authors and not the editors as the final paper has the same title they used on their ArXiv.org submission on the PRL submission date (where you can see the body of the article for free…).

Imagine you now take my spa and put 6 children in it. Kids being kids, they very soon work out that if they run around and around in a circle in the spa they can make the water flow around and around, they can then stop swimming and let the water carry them around the spa (I’ve done this, it’s cool fun!). If you then run around and around in the opposite direction, you can make that flow stop and reverse direction. Of course, if the kids just go in all sorts of silly directions, then nothing much happens. You can imagine that if the kids did this while you were trying to use this spa as a rheometer to measure the liquid viscosity you’ll get some weird results — the liquid might look like it has zero viscosity or even negative viscosity.

The behaviour above is exactly what the researchers are trying to look at, just using bacteria rather than children. If the bacteria swim collectively, then the viscosity will change from that of just water — this is fine, it just depends what you try to say about it as a conclusion, and this is where all professional scientists (and professional editors and professional science writers) know that you need to be very careful or people will call you out, and rightly so, because accuracy is everything in science.

The big question here is: if there’s some bacteria in your liquid that do a collective motion that make your rheometer measurement look like the liquid has zero viscosity, is it fair to call it a ‘superfluid’? I think most physicists would argue that the answer is in fact no. If you don’t give the bacteria ‘food’ then they don’t swim. If they don’t swim, then the viscosity is not zero. So, what you have to do here is pump energy into the system in order to keep the liquid flowing as though it has zero viscosity — but then how is it a dissipationless flow fitting the definition of superfluidity? Well it isn’t, you’re just being deceived — there’s just an agent in your fluid that’s hiding the viscosity. To highlight this, let’s imagine a twist on this experiment for a second…

Imagine that we put our researchers behind a wall where they can’t see their rheometer. What we then do is sneak in and put a very thin perspex cylinder with a radius that’s halfway between that of the cup and the bob into the spa. We now arrange that the cylinder can be rotated such that when the liquid outside it is made to flow by the rotating cup, the cylinder is rotated such that the liquid inside the cylinder either doesn’t flow at all, or flows backwards. The experimenters outside would be blown away, what they’d see is a zero viscosity, or a negative one if the inner flow is backwards. But there’s a dissipation going on that they can’t see, and it originates in our added perspex cylinder. Now for the death punch — the cylinder we have here, it’s just a proxy for the collective motion of the swimming bacteria. As far as the essential fluid dynamics is concerned, they are completely interchangeable.

So in the end, we’re all being deceived by some physicists who are trying to oversell their work. Let’s be clear, I don’t dispute their data or their experiment — the measurements look correct and the data valid, and they should get what looks like zero viscosity or negative viscosity even, which is just an indicator of energy dissipation into the system by the collective action of the swimmers. But to call this a ‘superfluid’ and actively sell it as such, is an absolute howler — this nomenclature is just plain deception intended to extract impact from the publishing system in my opinion.

Now that we all understand the experiment, let’s look at the moral of the story. It comes back to my twitter conversation with Chris Cesare this morning. He says ‘good enough for the editors of PRL and the scientists, good enough for me’ — no, that’s not good enough at all. Your job as a professional science writer and journalist is to not just blindly buy the sales pitch of some authors trying to hype up their work so they get more impact from it than they otherwise might. And you certainly can’t blame your lack of healthy scepticism on the PRL editors, who also bought the line (probably more the title than the paper) and shouldn’t have. Science is a very dangerous game if we don’t apply our own personal filter of rationality over the results, if we simply go ‘the PRL editors think it’s tops, so it must be’ then you’re being reckless with your own credibility.

Chris seems to be a young science writer, so I wouldn’t want to rip him too hard on this one. But he’s a trained physicist and he needs to keep thinking like one. This should be a good lesson about due diligence in science journalism, not just for him, but for aspiring science writers (and journal editors) everywhere. Do your homework, don’t just buy the hype.


3 thoughts on “Why water + E.Coli = superfluid is too good to be true (or the importance of fact checking for science writers)

  1. I find the article to not be so misleading, as they use quotes around superfluid and added the word like ( ie. “super fluid” like), which to me clearly states that they are observing a phenomenon that has similar characteristics to a super fluid in a certain regime. The Nature news article on the other hand, does mislead by using the term super fluid without trying to make any distinction between the actual quantum phenomenon and this apparent loss-gain balancing act. I suspect the editors of PRL chose this paper more for this active fluid dynamics aspect. It is even possible that they suggested that the authors use the term super fluid, as I have had PRL editors suggest “sexier” titles in the past. It’s also clear that Nature loves a good story. Science by press release and over selling is becoming rampant, much to my chagrin. BTW: nice spa. 🙂

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s