Wednesday, January 18, 2006

Ronald Reagan should have been an analytical chemist.

A lot has been written about the 40th President of the United States, and for good reason: The man was a character. This is not surprising, since in his first career he worked as a character actor. But despite all the various comments made to and about The Duke over the past century, I don’t think anyone has suggested that he run an analytical lab So let me be the first. As far as I know, Ronnie was not great talent at the lab bench. I haven’t ever heard of him having a unique and incisive perspective on unlocking the secrets of the natural world. But I think hew would be a good lab director because of a catchy phrase that he coined. This phrase is as useful in the lab as it is in fighting the Evil Empire:

“Trust, but Verify”

This is a great axiom for analytical science. If you never trust anything, you will never make any progress- modern science is too complicated and technologically reliant for scientists to be able to construct every detail of their experiments from first principles.

On the other hand, if you just believe every paper you read and every number you look up, you will get yourself in all sorts of trouble. I think that a lot of the “knack” in practicing scientific problem solving is knowing which things to trust and which to test.

Case in point. A while ago, a student came to my lab with a standard from a reputable and respected scientific establishment. He used it, got some unusual results, tried to verify them by measuring them against some other standards. The results didn’t agree. Since the discrepancy in his results was larger than what can be explained by the standard sorts of variations, he decided to take a closer look.

He dissolved part of his standard, ran the solution, and discovered that the “known” value for the element he was interested in was wrong by an ORDER OF MAGNITUDE. Oops. I don’t think this was the case of a misplaced decimal point, either. People far wiser and more knowledgeable about this particular system than I have suggested that the error was actually in the way the sample was manufactured and analysed.

That’s a pretty big Oops. Especially for an organization that prides itself on getting things right. But still, I’ve done worse. In my first ever scientific presentation as a grad student, I misrepresented the viscosity of the lower mantle- by 42 orders of magnitude. That’s right, the number I quoted was 1,000,000,000,000,000,000,000,000,000,000,000,000,000,000 times smaller than the accepted value. I did this in front of the former director, no less, who is one of the people who figured out how to measure mantle viscosity in the first place.

Oh well. I’ve always been more partial to experimentation than book research, but still. You’d have to be an astronomer to make a mistake of this size in public. That or a vain woman misrepresenting her weight or age. As for me, I’m just a lab lemming who forgot to put a minus sign on his exponent. But hey. Being a student is all about learning, isn’t it? Which is why I tell the students coming through now to think like Reagan when doing their research. Trust, but verify.

Friday, January 06, 2006

BANG! goes the weasel

You know those innocuous-looking little safety knobs that sit unobtrusively on the sides of gas cylinders, liquefied gas containers, and other high pressure equipment? They are LOUD when they blow.

We got a new Argon tank delivered yesterday with a dodgy regulator, and when the 150 bar safety knob blew I practically jumped out of the courtyard. Luckily it was downstream of the valve, so I could just shut it off, but man, what a shock. It’s a good reminder, though, that gas cylinders- even those filled with inert gasses- store a huge amount of energy and should always be treated with respect. On of the problems with a lab is that if you aren’t careful, the routine nature of things like gas cylinder swaps can potentially lull you into a false sense of complacency. Given the huge potential for harm that exists in a lot of lab equipment, that would be a bad thing. Ear protection wouldn’t be a bad idea either.

But what I’ve actually been pondering is exactly which bit failed. The reg is only supposed to put out about 80 bars or so. The safety release is rated to 150. But the gauge on my gas line was reading 105 as the blowout pressure. The fact that it got that high tells me that the reg must be busted. But I don’t know if the knob blew 45 bars low, or if my pressure gauge is miscalibrated. That’s an experiment the lads from the gas company will have to do, though- I ain’t touching that thing again, and the sooner they pick it up, the better.

Thursday, January 05, 2006

Public Service Announcement

Here's a tip to all you scientists, great and small, to keep in mind for the New Year. It applies to everyone, from Nobel Prize winners to 10th grade general science students who will never measure anything again:


If you aren't analyzing what you think you're analyzing, you won't be able to draw a whole lot of meaningful conclusions from your experiment.

Now, y'all might be thinking, "Duh," but sadly, in the ten years I've been a paid lab worker, I've seen way too many cases where mislabeling has occurred. Anyone can do it, and the results are often heartbreaking.

For example, I know of someone who collected rocks from one of the most inaccessible areas on the planet- the sort of place that makes Greenland seem crowded and accessible. After collection, the scientist then spent several years carefully separating his rocks into their component minerals. He carefully packaged these minerals for international shipment, paid a substantial (yet internationally competitive) fee to have the analysis done, and then put the same sample number on all of the separates.

Kids, don't do this at home.

Be careful. Don't rush. Double check. Better yet, devise a system that allows you to recheck and catch mistakes as you make them. Write everything down, and save all of your primary notes, even if they somehow end up being written on a Kim-wipe, instead of the notebook you ought to be using.

A hint about rechecking, though- It only works if your checks are truly independent; if you label a vial, put it in a bag, and label the bag using the vial, and not the original source, you're just propagating the error. In at least some of the mislabeling cases I've seen, it looks like this is exactly what happened.

I suspect that the medical industry has come up with all sorts of clever and fool"proof" ways of preventing mislabeling, since in their field the real-world consequences are a lot more severe than just mixing up a couple of rocks. Of course, I also suspect that a lot of their systems are also very expensive, given the amount of money in the field. But even if you can't afford to blow huge sums on computers and scanners and such, it might be worth looking into their systems just to see how they work.

So don't get complacent. Don't get sloppy. Don't get lazy. All scientists will flatly deny ever having mislabeled a sample, even once. But that doesn't mean that it never happens. It does. And it could happen to you.