Slashdot Mirror
Kilogram Gets Controversial; Why Not Split the Difference?
gbrumfiel writes "As Slashdot has noted, the kilogram has a problem. The SI unit is officially defined as the weight of a 130-year-old platinum-iridium cylinder in France. But the physical object appears to be getting lighter. Scientists want to replace the cylinder with a new standard based on Planck's constant, but two experiments designed to facilitate the switch keep coming up with different results. Now one researcher is proposing a solution: just average the two diverging experiments and use that value as the official definition. Not everyone thinks that averaging the two amounts to sound research: 'Deciding to just average these two results would be perfectly proper mathematics, but it would not be science,' says Michael Hart, a physicist at the University of Manchester, UK."
I think the GP's point was that even if you chopped a sizeable chunk off it, it would still weigh precisely 1kg. It logically follows that the universe's weight, expressed in kg, would suddenly jump upwards by a very large amount.
In this case the background is that the standard for mass, unlike time or distance, cannot independently be constructed in the lab. This means that science and industry are susceptible to two issues. The first is degradation of a physical standard, in this case a hunk of metal in France. The second is that one is dependent on other to create proxies of the standard, and as a result have no true assurance of the accuracy of the standard. A suitable lab with suitable personal can masure time and distance without the need of a proxy manufactured by others, and no dependence on a fixed physical object.. There is a desire for the same to be true for mass.
Second, no one knows if the hunk of metal is shrinking, and if it is how much it is shrinking by. If the experts knew it was shrinking, then they could figure out how to at least partially correct it. The hunk of metal might not be charging at all, or it could be accreating matter. Without an independent standard, which does not apparently exists, as everything is based on the hunk of metal, all there is is guesswork.
The third is the idea that Planck's Constant is being used to create the standard. In fact Planck's constant is one two approaches. The other is to create a sphere from a silicon and use Avagadro's Constant to define the mass. The problem is that these two approaches do no lead to consistant results, with an error about an order of magnitude large than the expected error.
The issue with averaging is that while one does average within a result, and even results that are taken from similar procedures, it is unclear that averages in this case is suitable. It seems to me that the results point to an interesting area of research, and rather than just averaging, more work should be done understanding the inconsistency. If it is not random error, and not an artifact, then something really fascinating might be going on.
"She's a scientist and a lesbian. She's not going to let it slide." Orphan Black
Speaking as an experimental physiscist
ahem. 175parts per billion is 1.75e-7. For metrology that is a huge discrepancy. What is worse is that the measurements themself are a factor of 5 better, leaving no room for error.
For experiments where the physicists believe they understand them this is unacceptable, because it actually means the pysics of at least one method of both is not well enough understood, i.e. you have a systematic error. If the physics is not well understood then you don't know if the systematic error will be constant.
If the measurement will not be constant then the average will also not be constant. So an metrology institute where a reference weight should be define will need both methods and still not get any stable definition.
If you already need to afford both methods, then you can create reference weights and at the same time check if the difference between both methods is the right one and constant at your place.
Important rule in experimental physics: NEVER average over systematic mistakes. Average over random results. On systematic mistakes, the word average makes no sense