09 August 2010

Designing good experiments

This is another time I don't really have great answers, but a question over at the question place has me thinking about it.  Namely, what makes for a good experiment?  As far as the science goes, I'm comfortable about knowing the answer.  Or at least knowing enough of an answer. 

But for the purposes of you readers -- what makes for a good experiment that you could do yourself?  How big or small could it be?  How long should it take to run?  How much expense is ok?  Is following a circuit diagram to assemble test equipment something you're comfortable with?  Carpentry?   And so on. 

For the original question -- a tabletop demonstration of the greenhouse effect -- I might actually have an answer of sorts.  I went running shortly after first reading the question.  That's often a good time for ideas to come to me, and a few did.  But at the moment, they'd take a pretty big table (like, say, 10 feet), you'd have to get hold of a dry ice supply (for the CO2), and you'd have to assemble a fairly simple circuit.  A several Watt laser would also be a plus, but I'm going to try to make sure that the experiment will work without it.


James Annan said...

It is sometimes said that in theory, a good experiment is one where you cannot predict the outcome (so you learn).

In practice, most of my (numerical) experiments are designed to demonstrate something that I at least have a strong hunch about...

Anonymous said...

Huh. Well I, for one, would be interested in hearing about your running inspiration, even if it does require some real lab equipment: while I no longer have access to real chemistry labs myself (sniff... the drawback of moving from wet-lab to computational research), I have colleagues with lab access who would love to make a video of a good demonstration. And they definitely have dry ice (and gas canisters), giant lab tables with screw in capability for laser mounting (though I don't know if they have the lasers to go with...), vacuum manifolds, etc.


Anonymous said...

(in general: it would be neat to have an answer along the lines of:

'here's the chrome-plated experiment which is sure to work but not many people could do... now, if you want to do this at home, here's a secondary version which can be done with fairly common materials though it might require some effort like carpentry or looking up how to wire simple circuits - but I'm not sure if the signal will exceed the noise in this second version...'


carrot eater said...

Seems to me that the trick is to somehow get a lapse rate into your experimental setup.

Or, you could go radiation only, throughout. Have a solid sphere, in the middle of an evacuated space. Around that, a concentric shell of some greenhouse gas. then another shell of vacuum. Finally, a visible light source. would that do it?

Anonymous said...

Certainly, I had thought about a couple modifications to the standard setup to make an experiment work, but they aren't simple:

1) A shell of vacuum around a container of gas (this eliminates non-radiative heat transfer out of the gas: hopefully, this eliminates the problem that CO2 and air have different convective properties that results in CO2-filled containers warming faster in many experiments)

2) Liquid nitrogen cooled plates around the container: this reduces background IR, making the signal easier to see.

3) Either an IR laser fired through a hole in the LN2 cooled walls, or a sunlight-spectrum source entering through a similar hole but with a black card either inside or behind the gas container to create a blackbody IR source.

Then, do the experiment with pure CO2, pure, N2O, pure argon, pure N2, and a standard air mixture. That should show that you are seeing IR effects in the CO2 and N2O, and not weird convection effects like you'd see when you switch regular air for argon in the standard version of the experiment debunked by the Tufts researchers.


Anonymous said...

In response to James:

There are 2 kinds of experiments: those for demonstration, and those for science. Presumably, an experiment meant for demonstration should not have unpredictable outcomes: rather, it should be a good teaching tool.

I agree that experiments for science are uninteresting if they are purely predictable: I'd argue that the most interesting results come from experiments in which one has an expectation, but something different from expectation happens. But even those that match expectation can be useful confirmations of theoretical understanding...


Alastair said...

There is an experiment on YouTube by one of my compatriots here.

Cheers, Alastair.

Anonymous said...

Alastair - thanks, that's a nice video. And yes, for a reasonable person, that should be sufficient evidence.

However, I was really hoping for an experiment which would actually show an increase in temperature, not just a change in radiation absorption (yes, they are the same thing, but...) (and yes, even a perfect tabletop experiment is not as good as, say, making a new earth-sized planet and replacing its atmosphere, but that is a little bit outside my budget...)


Alastair said...

There is another YouTube experiment here. Both experiments are presented by PhD's.

It would seem that this experiment would need a third bottle filled with argon, to eliminate convection as a cause of the warming. Here is the argon spectrum.

So I suspect because everyone is using visible light, it is not a true test of the absorption of far infrared radiation emitted by the Earth. If you use a red hot cannon ball after it has darkened, (as was used by de Saussure and Pictet in the original greenhouse effect experiments) or an electric iron to produce the infrared radiation then the argon will not warm.

In other words, the argon is not warming due to convection, but by the absorption of electronic lines in the visible spectrum.

Cheers, Alastair.

Anonymous said...

Yes, I think that that youtube video is exactly the kind of experiment that the authors of "Climate Change in a Shoebox: Right Result, Wrong Physics" identified as being flawed, and demonstrated that argon would have the same effect as CO2.


Alastair said...

Yes, but I am suggesting the argon warms because it absorbs visible radiation, not due to convection as they tried to prove.

So if they repeated their experiment with an infra red source, perhaps even a radiant electric fire, then they would not get the argon causing additional warming.

I suspect that the bell jar experiment would not be affected by convection in any case because no internal absorbing surface being heated.

Any chance that you will try these experiments? You could fill the bell jar with CO2 by adding a fizzy drink!


Semisovereign People at Large said...

I dont see the point if you have a gaseous mixture with 5 or 6% CO2
of course you have a significative increase in temperature and so?

I you will try these experiments?despite some strong imprints of natural variability, a role of the greenhouse gas loading now seems clear

i supose that in your country you have methane gas bottles is less expensive
or if the CO2 is your goal sodium bicarbonate
is a middle school experiment in several countries in small containers well is pointless...

Banda in barbar said...

You say that love to make a video of a good demonstration, of what?

the radiaton absorption coefficient of gaseous mixe's varies greatly with pressure
and temperature in terms of its half-width and line strength. Thus, in order to apply your experimental method to realistic atmospheres, variation in the absorption coefficient
in the vertical must be accounted for.
Basically, you must determine what you want and the variables involved

Anonymous said...

Alastair: Did you read the Climate Change in a Shoebox paper? I thought they did a reasonable job of demonstrating that it was convective and not radiative properties.

A few points:

1) Ar has some absorption lines in the visible, but presumably so do CO2, O2, and N2. Additionally, the lamps used also emit in the infrared. My guess is that Ar doesn't absorb any more than CO2, O2 or N2 in the visible, and that its absorption in the visible is tiny compared to the absorption of CO2 in the infrared (even if the infrared intensity is lower than the visible intensity).

2) The drop in temperature about 700 seconds after addition of the gas (whether CO2 or Ar): this appears to be due to the gas dropping below the level of the temperature sensor. This strongly suggests that the temperature drop is the result of an interface between a dense gas and a light gas dropping below the sensor. (yes, a sealed jar would reduce convective effects that depend on the interface with outside air, but there are still issues about heat transfer from the gas to the glass that might change with different gases)

3) The increase in temperature in the CO2 jar is higher than can be explained by the radiative absorption.

I'd love to try additional experiments myself, but I don't have anything in the way of lab equipment, and my collaborators are loathe to try anything new unless it has been well thought out because they were already burned once by getting the right result from the wrong physics...


Alastair said...

I should have said that the first YouTube experiment was performed by Dr Iain Stewart, and the second one by Dr Maggie Aderin-Pocock

I was not terribly impressed by the Wagoner et al. paper. The photograph of the equipment showed a thermometer lying near the temperature sensor, with no explanation. The absorbing surface was not painted black as I would have expected. I also thought that there was an overuse of multiple references to establish points. (I have now discovered that at least two of the three experiments they cite [Refs 1, 2 &3] are the same experiment.) However, I just assumed that what they wrote was correct as it had been peer reviewed (which I now doubt.)

When I first saw the two YouTube experiments, not on YouTube, I thought then that they were flawed because they were using visible light, not infrared. When I posted them here I had a eureka moment. It occurred to me that the argon warming in the Wagoner et al. experiment was caused by absorption of visible radiation.

It seemed to me that here was an answer to Bob's question - "What makes a good experiment?" My experiment would be to fill three jars with air, CO2 and argon. Stick three thermometers through their bungs and shine bright light on them and see whether they warmed and at what rate. Then repeat that experiment using an infrared source.

This experiment meets James Annan's criteria of not knowing what the result would be. I am not sure that the argon jar would warm under visible light. Wagoner et al. may be correct. The argon may be warming due to convection in their experiment, but I don't think my/Maggie's experiment is as susceptible to the effects of convection as that performed by W. et al.

Note that W. et al.'s experiment is not the same as the one they are criticizing which is here: http://www.rmets.org/pdf/co2.pdf and is similar to Maggie's. They appear not to have applied the Kiss Principle.

Argon is cheap and easily obtainable from welding equipment or diving equipment shops. I am tempted to do the experiment myself but doubt that I would ever get around to writing it up :-(

Cheers, Alastair.

Alastair said...

My eureka moments are not always correct, and I am now doubting my last one regarding the heating of argon.

I believe it and the CO2 were being heated by convection in the W. et al experiment and the CO2 in the Lueddecke et al. experiment. Their problem is that the IR source is in the gas container. Putting the IR source outside the jar means that it cannot create convective heat in the experimental volume. Maggie is using light and IR, but if you used only IR from a hot source and shone it into a jar of CO2 it should warm eventually, unlike air or argon.

Cheers, Alastair.

W.T."Tom" Bridgman said...

This is an exercise I've often wanted to see. Is is approaching the 150th anniversary of the discovery of CO2 as a greenhouse gas.

The Carbon Dioxide Greenhouse Effect
Tyndall, 1861

Robert Grumbine said...

reconstructing extinct...:
I couldn't tell what you were trying to talk about. If on-topic, please try again. Thanks.

warmcast said...

I think an accurate representation would have to replicate the lapse rate.

One problem would be the container that confined the atmosphere and the surrounding lab environment.
The Second would be a simulation of the vacuum of space, plus gravity.

I don't think a sphere is required just a long cylinder.

I don't actually think a simple school lab demo works and a genuine 'experiment' would be to expensive for schools, plus students probably wouldn't be able to see much in a realistic experiment.

To simulate gravity you would probably need simulate it using some sort of huge centrifuge.

Robert Grumbine said...

I think several folks were looking for a much more detailed experiment than I had in mind. Mine was just a lab-scale demonstration that CO2 does indeed absorb terrestrial infrared. Still not sure whether it would really work.

While doing some other reading, I thought of a different experiment that would demonstrate that the atmosphere (somehow) does indeed radiate energy towards the surface. As some of you have noticed over at Roy Spencer's blog, there are indeed people who don't believe that there's a greenhouse effect. This experiment won't change their minds, because as Spencer's threads have illustrated, nothing will. But it's always nice to have a simple demonstration in hand.