What's a kilogram?

Back when the International Temperature Scale was rejiggered in 1990, I had just bought my Fluke 87 multimeter, and Fluke actually offered me a free recalibration, because the temperature scale changes also unavoidably tweaked the volt slightly. I didn't bother, partly because a Fluke 87 really doesn't show enough digits for the difference in the ITS'90 volt to be visible (and partly because my typical use of the meter anyway is to ask questions like "is this at roughly 12 volts or not?"). But it would have come back with a cool ITS'90 sticker if I had taken the offer.

It seems like this kilogram redefinition also tweaks the ampere a bit, so maybe there's another chance to get a sticker.

-Chap

 

The ampere should get completely disconnected from the kilogram and all other fundamental constants except for the second. In short, it should depend only on the number of electrons in a Coulomb, and the length of a second.

 

I think the wrinkle may be that, just as with temperature, there is a theoretical definition of what the unit represents, but one that may not be directly measurable (no one is measuring amperes by counting electrons through a turnstile and dividing by seconds), so a metrology standard needs to define a method usable in practice to measure a value, and argue for how accurately that value will represent the theoretical thing you want to know.

Nice article in The Verge: The kilogram is dead; long live the kilogram - The Verge

To know a kilogram, you used to need one of these:



Now you need one of these:



-Chap

 

I think the meter got revised a few decades back? There where building site debacles, especially on jobs with long cross-country conveyors: they would arrive at their destination, all the math checked, but were found to missing the mark by multiple, well: inches, lol.

The steel fabricators were working with:

1 inch equals 25.4 millimeter

While the surveyors were using the exact definition of the meter, which was ever so slightly different. Some clever clogs based the meter on a fraction of the earth's..., or some wavelength of..., well I don't much care.

After a few of those messes, I believe it was resolved that 1 inch would be 25.4 mm, precisely. For all practical applications, at least.

I just really wish, that the clowns who established the meter could have tipped their hat to imperial measure, say made the millimeter 25.0 to the inch, for example.

I was on more than one job, where all the supplied mechanical components were imperial, and we provided metrically dimensioned support steel, which we handed over to the fabricator's retailers, who promptly drew up the shop drawings in imperial.

It's interesting too, that all the hot rolled steel in North America is still imperial. In Canada we describe it in metric units, but it's the same steel. Same for plate, which if anything is even more of a complication.

Only in Canada...

 

I remember seeing some ScanGauge XGAUGEs that were using ✕ 5 ÷ 8 as the conversion from km to miles, even though the exact conversion ✕ 15625 ÷ 25146 would fit perfectly well in the arithmetic limits of a ScanGauge. Sometimes it's funny the corners people cut (or funny as long as you're not building a long conveyor, etc.).

-Chap

 

I use that 5/8's all the time, admittedly just for rough stuff. I guess I'm thinking 50 mph, vs 80 kmh.

Speaking over overland conveyors: I'd been using AutoCAD for a few years, when I finally found it's limits. Working a site plan, with a couple of conveyor centrelines I told to fillet with zero radius (extend both lines to their intersection point). I then tried to "snap" to that intersection, only to be told there was no intersection.

I zoomed in, required a regen or two to get close enough, till sure enough saw a little gap between the two line end points. Can't recall, but I suspect if I asked for the distance between the two end points it said zero. Not sure now, I could try it.

Other times, similar scenario, there'd be an intersection, but if you zoomed in you'd find the two lines crossed and kept going. Slightly.

Yet another scenario: I used User Coordinate Systems (UCS) extensively, and found with larger extent drawings, you wanted to keep the UCS origin as close as practical to the work. This sometimes necessitated two UCS systems, one for the head end, one for the tail end. Even though the orientation of both was the same, if you tried to copy/paste tail end stuff to the head end, using say just the tail end UCS, it would come in apparently ok at the head end (say several kilometers distant from the tail end), but when you zoomed in, you found elements of the paste were slightly shifted and/or rotated.

 

I had understood this unit conversion mess to have been settled by treaty in the 1950s, before I was even born. 1 inch exactly equals 25.4 mm. But some hacks stubbornly continued to use 1 meter approximately equals 39.37 inches for many decades beyond that, carrying a needless 2 ppm error, or about an inch in 8 miles.

The underlying meter continued to get refined much longer, but shifting by far smaller amounts. When it was finally set at its current definition, they effectively took the old estimate for the speed of light, previously figured down to hundredths of a meter per second, and rounded it off to an integer number of meters. I.e. the shift was less than 1.7 ppb, which is less than an inch across the diameter of the earth. For this change to cause conveyors to miss by plural inches, they would have to connect around the earth's surface the long way, not the short way. Eg. Edmonton to Calgary not via Red Deer, but rather via the Arctic Ocean, Russia, Indian Ocean, Antarctica, Pacific Ocean, Mexico, and finally across the U.S.

So I'd blame these construction misses not on changing definitions of the meter or inch, but on a mixture of survey errors, software bugs, and designers stubbornly sticking to long obsolete approximations.