When is a kilogram not a kilogram?

June 15, 2007 8:56 AM

Scientists will use perfect silicon spheres to determine how many silicon atoms make up a kilogram, and this will be used as the new international prototype — bringing the kilogram into line with other base units such as the meter. The current international prototype, a lump of platinum and iridium, is securely tucked away inside a French vault.
posted by nevercalm (45 comments total)

I think they should use silicone spheres instead.
posted by delmoi at 9:02 AM on June 15, 2007

I think we should just measure everything by a breadbox. Makes life much simpler. Is it bigger than a breadbox? Yes. No. See? Real simple. For larger items that need scale, simply substitute the breadbox for Karl Rove's head.
posted by ZachsMind at 9:06 AM on June 15, 2007

No, no. They should use dorodango.
posted by bshort at 9:07 AM on June 15, 2007

Interesting. I wonder if there are actual experiments in which some scientist says "these results would be much more accurate if we had a defintion of a kilogram based on a specific number of atoms rather than a single object," and if so, what experiments those are. Are there any obvious practical implications of this, or is it really only of interest to incredibly anal retentive scientists?
posted by scottreynen at 9:08 AM on June 15, 2007

Just wait. Somebody'll drop the damn thing on floor and break it, sure as hell.
posted by pax digita at 9:11 AM on June 15, 2007 [1 favorite]

Those pictures of the standard lump for kilogram are incredibly cool. I've wanted to see the vault in Paris where they keep the standard meter, kilogram etc. since I was a kid. When I'm next in Paris I'll have to AskMe to see if there's a way to tour the place.
posted by Kattullus at 9:23 AM on June 15, 2007

A breadbox is a unit of volume. (Rhode Island is a unit of area and a football field is a unit of length.) A kilogram is a unit of mass.
posted by DU at 9:25 AM on June 15, 2007

Perfect silicon spheres? When did they figure out how to do this? I was under the impression that there was no such tangible thing as a perfect sphere.
posted by King Bee at 9:29 AM on June 15, 2007

It's very faintly ironic that platinum was chosen in the first place, since natural platinum is very faintly radioactive (about 1% of it is an alpha-emitter with a very long half-life) so that the standard kilogram has very slowly, but ineluctably, been losing mass since the day it was made in 1880, about 16 years before Becquerel discovered radioactivity.
posted by jamjam at 9:30 AM on June 15, 2007

DU! Dude! Use the length of the breadbox. I'm telling you, all forms of measurement could be done with a simple breadbox. You're the kinda person who complicates things, "but what if some people used the width while other people used its depth for the length of something? There'd be no consistency!" I'm talking about simplifying the world. We got too many forms of measurement. It's a small planet. We just need one.

We also need to adopt the decibet. Too many letters...
posted by ZachsMind at 9:32 AM on June 15, 2007

Related
posted by edd at 9:47 AM on June 15, 2007

Piffle. Natural units are the way to go. I'm partial to Hartrees myself.
posted by lester the unlikely at 9:52 AM on June 15, 2007

how to cut a sphere
posted by hortense at 10:10 AM on June 15, 2007

Better way to cut spheres.
posted by King Bee at 10:20 AM on June 15, 2007 [3 favorites]

On the radio this morning the measurement for the amount of nuclear waste in storage at a local plant was described as "five hockey rinks filled up to the boards." Guess where I live.
posted by furtive at 10:29 AM on June 15, 2007

I'm guessing somewhere around 45.518143, -73.588142 -- right?
posted by pax digita at 10:34 AM on June 15, 2007 [5 favorites]

For some reason I thought that a kilogram was supposed to be the mass of 1 liter of pure water?
posted by empath at 10:44 AM on June 15, 2007

I just let the King of the Cosmos do all my measuring. If he says my Katamari is as big as 354,289 breadboxes, I just take his word for it.
posted by buriednexttoyou at 10:56 AM on June 15, 2007

recent issue of Cabinet had a related article, for those who are interested: Measure for Measure: An Interview with Anne Chaka Frances Richard
A visit to the National Institute of Standards and Technology
posted by jcruelty at 10:57 AM on June 15, 2007

For some reason I thought that a kilogram was supposed to be the mass of 1 liter of pure water?

Er, I think that's what they were going for, but it gets complicated to measure "one liter of pure water", and it depends on the temperature, so it's much simpler to say. "this lump of metal which weighs about the same as one liter of water"
posted by delmoi at 10:59 AM on June 15, 2007

Interesting fact: the prototype kilogram is the only object with a mass that cannot change.
posted by alby at 11:15 AM on June 15, 2007

For some reason I thought that a kilogram was supposed to be the mass of 1 liter of pure water?

Me too. You can specify that it is one L pure water at a given temp, altitude, and barometric pressure, to ensure precision and reproducibility. Once you do that, it's no problem, as the liter can be defined by the standardized meter. So to me, it makes sense to do it this way: define one thing (meter, already done), and then let the other definitions/standards stem from that.

OR you could go on with your perfect spheres and such like.
posted by Mister_A at 11:15 AM on June 15, 2007

I would also not object to a system of measure based on hockey rinks.
posted by Mister_A at 11:17 AM on June 15, 2007

These scientists, always overthinking a hockey rink of beans.
posted by buriednexttoyou at 11:21 AM on June 15, 2007 [1 favorite]

The problem with water is containing it: the litre is a measurement of volume. The volume of a given mass depends on its density. The density of a liquid varies with depth. So 1 litre of water in a big flat beaker actually has a higher mass than 1 litre of water in a tall measuring cylinder. You end up with a horrible feedback loop that makes realising the kilogram very difficult.
posted by alby at 11:26 AM on June 15, 2007

So, you think that this metric thing is going to catch on?
posted by octothorpe at 11:27 AM on June 15, 2007

Perfect silicone spheres.
posted by hifiparasol at 11:27 AM on June 15, 2007

alby: so you'd just need to specify the size and shape of your liter container, right?
posted by boo_radley at 11:46 AM on June 15, 2007

Interesting fact: the prototype kilogram is the only object with a mass that cannot change.

Uh, the mass can change, just not the number as measured by kilograms.
posted by delmoi at 11:50 AM on June 15, 2007

alby: so you'd just need to specify the size and shape of your liter container, right?

So you have to get the temperature, pressure, humidity, hight, width and shape of the container perfectly correct? It sounds like a hell of a lot more work then simply using a 3rd generation copy of the prototype kilogram to calibrate things.

And with the liter thing, how do you know when your container is properly filled? You can only measure the water level to a certain precision. You could never get down to the nanometer level like these guys with their silicon spheres.

On the other hand, with a pre-calibrated weight can be much more precise and with far less work.

In fact, I doubt you could even measure out one liter of water precisely enough to tell if the standard kilogram matched it or not.
posted by delmoi at 11:57 AM on June 15, 2007

Isn't there still some room to doubt whether gravitational and inertial mass are the same? If they're not under all circumstances, we'd seem to need two different standards, and the uncertainty principle might limit at least the precision of the inertial, perhaps.
posted by jamjam at 12:44 PM on June 15, 2007

That's going to be one damn expensive bocce ball.
posted by beagle at 1:07 PM on June 15, 2007

I wonder if there are actual experiments in which some scientist says "these results would be much more accurate if we had a defintion of a kilogram based on a specific number of atoms rather than a single object," and if so, what experiments those are. Are there any obvious practical implications of this, or is it really only of interest to incredibly anal retentive scientists?

You're missing the relevance of this, which is "how do you define a kilogram." Currently, the only way we can define a kilogram with sufficient precision is by saying "a kilogram, by definition, is a thing with the mass equal to this one precious physical artifact." This is the only physical constant remaining that has such an artifact, all the rest are defined purely mathematically, e.g. X number of transitions of a state of a cesium atom is a second. So if you want to build a machine that measures seconds, you can just get some cesium of the specified type and count the number of transitions and see if it's close enough to the standard. But not so if you want to build a device that measures mass; for that you actually have to get in a plane and go to Paris and test your machine against this precious artifact, as that is the definition of a kilogram.

Now, of course, most people building mass measuring devices don't do this, as there have been enough approximate copies made of this artifact that you don't have to fly to Paris. But on a fundamental level that's what you're doing -- you test your device against a device that was itself tested against a device that was tested against a device that was tested by the precious physical artifact in Paris.

The whole point of this new standard is to be able to say "a kilogram is defined as the mass of a perfect sphere of Si with X atoms" since Si in crystalline form has a very regular and fixed lattice structure. The standard no longer depends on one particular artifact, but it is now defined in terms of a fixed number of something that can be built or measured, just like the other physical units.
posted by Rhomboid at 1:30 PM on June 15, 2007 [1 favorite]

Rhombobid, that means that if I want a real kilo of tortillas I would need to get me some Si spheres? Too much trouble.
posted by dov3 at 2:25 PM on June 15, 2007

I think we should just measure everything by a breadbox.

ok, i've got this theory about the metre that needs out. (forgive me if i'm going to be a little sketchy on the details - i worked it all out back when i was younger...)

the principal early definition of a metre was as one forty-millionth of the length of the Earth's polar circumference. but here's the question: how did they get to that? why not, one thirty-millionth? or one fifty-millionth?
clearly, to start it all off, someone (someone french, since they were in charge of all early decimal/metric systemizing) said: a useful basic measure would be, oh, about this size (stretching out arms, just enough, not too much) about, ehm, ah voilà: about the size of (wait for it...) a baguette!

this was the basic notion that the bureau des longitudes then worked from, refining it down to a unit that could be represented as a fairly rational fraction of some natural whole - but that's just post-facto justification of the original gesture.

non?
posted by progosk at 2:31 PM on June 15, 2007

There's an immediate problem that needs to be solved too. The masses of the reference kilos have been going out of whack for the past few years. It's probably a combination of mishandling and, unh, not so careful storage, but the end result is that our definition of what a kilogram is is getting worse and worse as time goes on.

For pretty much any good you buy, ultimately, getting a fair price depends on the SI kilogram. Now, the drift is nowhere close to large enough to make a practical difference to commerce, but any drift in your primary standard is a problem.

This is a very good attempt at getting away from all that. So yeah, this makes a difference.
posted by bonehead at 3:01 PM on June 15, 2007

There is a dinosaur comic about the prototype kilogram.
posted by sindark at 4:41 PM on June 15, 2007

progosk: One forty millionth all the way round is one ten millionth of the equator to pole. And ten million is a nice metric number, what with it being all ten-ish.

Of course the fact it was ten million and not one million or a hundred million would be because it's roughly man-size I guess.
posted by edd at 5:48 PM on June 15, 2007

This is cool stuff that should appeal to any anal-retentive. They correct for all sorts of things the buoyancy of air and even the partial adsorption of air and water molecules on the silicon surface.

The measurement accuracy depends on

(1) measuring the sphere mass as close as possible to the standard kilogram

(2) measuring the number of Si atoms in the sphere

(3) measuring the mass of a silicon atom

The least accurate measurement in these 3 quantities basically determines the accuracy of the result. The accuracy they claim is about 1 part in 10⁸.

But I think this is still not the best way to go. A potentially much more accurate method involves measuring the mass of individual atoms very very accurately.

With Penning trap mass spectrometry, we may be able to measure the mass of a single atom down to 1 part in 10¹¹ and define the kilogram that way. So this is likely not the end of the matter.

/embittered mass spec guy
posted by Bletch at 5:52 PM on June 15, 2007

Bletch, but by what, erm, metric do you measure that single atom? Are you doing some sort of total-energy-mass-eqivalence? (dead link, btw)
posted by TheNewWazoo at 8:16 AM on June 16, 2007

Planck masses?
posted by ROU_Xenophobe at 9:46 AM on June 16, 2007

Bletch: But how do you scale that up to a kilogram?

In analogy, there's ongoing work to redifine the ampere in terms of the electron charge, and utilize as the standard a little machine that pumps a known number of electrons between two leads every second.

There are such electron pumps that provide the required accuracy of 10^-8, but it's a real problem that they can't produce more than a picoampere or so. A nanoamp at the very least would be required for a new definition to be adopted, since you can't calibrate secondary standards reliably enough from a lower current.

How would you go from knowing the mass of an atom to knowing the mass of a mole of atoms? Except by making a ball of them and calculating the number of atoms from the ball's size and the lattice constant?
posted by springload at 1:53 PM on June 16, 2007

Another interesting thing: There is a rivaling scheme called the Watt balance. It essentially determines how much power you need to send through an electromagnet in order to keep the kilogram afloat. From what I heard, they compared the silicon sphere with the watt ballance, and the discrepancy was so large that they had to reevaluate Avogadro's number.

And Bletch: I think the answer to the question I asked you is here, under "Ion accumulation approach".
posted by springload at 3:01 PM on June 16, 2007

Does not the mass change when you bond atoms together in a lattice or other molecule? That being the basis of chemical energy and all.

So measuring the mass of one Si (or whatever) atom via mass spectroscopy is one thing but it doesn't tell you exactly the mass of n Si atoms in a lattice: you need to know the lattice structure, number of bonds, etc. And how accurately can you measure the energy in the bonds?

Either way, we need a finite measure. Personally I reckon it should be a gas or plasma measure because that overcomes the bonding energy problem. We can measure the mass of an ion, say, then we just define the kg as being the same mass as some integer number of the chosen ion. Practicalities of second-generation references be damned, there needs to be an integer somewhere in the reference.

A polished sphere is a pretty approximate definition of an integer without specifying the lattice-cut pattern at the spherical boundaries. Certainly good to 10^-11 or whatever we currently need, but ideally we need an exact definition.
posted by polyglot at 9:39 PM on June 17, 2007

polyglot: I covered this in the previous thread, but chemical bonds have energies on the order of electronvolts. The proton has a mass of about 10⁹ electronvolts, so you'd need to worry about it if your desired accuracy was around that or greater. You gain another factor of 30 from the atomic mass of silicon being 30 or thereabouts, so the bond energy is about one ten-billionth of the total mass.

I would expect then that it's possible to measure the bond energy to enough extra precision to get you to the 10¹¹ accuracy as you'd only need to measure the bond energy to a few percent to get a good enough correction on it.
posted by edd at 6:10 AM on June 18, 2007 [1 favorite]