Use of non metric units in the US for scientific purposes

[Napier]

I appreciate the opportunity this thread gives me to express myself.

The American use of units is a disaster in serious technical work. I think the SI prefixes and attitudes toward base ten are handy, but this is not a primary reason. The biggest reason is that the SI is coherent, meaning that complicated unit conversion factors (and offsets if temperature is involved) are not sprinkled throughout most complicated calculations, as they are in the customary US system. The second is that the customary US system often has many units for a given dimension. This means that there are also conversions involved between things that are dimensionally equivalent.

All this stuff is baggage that has to get carried through calculations. I think the problem with the Mars Orbiter is that something went wrong with carrying this baggage along. Of course it is possible to carry the baggage everyplace. It is just that doing so requires a lot of extra work, and creates many more opportunities for errors, and requires additional information about the conversions, all without any gain whatsoever.

I do much of my work with heat transfer. In the SI, a heat transfer coefficient is in W/(m^2 K), and that’s that. There are 126 different units for heat transfer coefficient in non-SI units in the table that I use for finding conversion factors, and of course my table is not complete. There is no way I can develop a feel for what a heat transfer coefficient is likely to be in each of these different worlds, and no way I can remember all the conversion factors. And this is for just one dimensional quantity; every single dimensional quantity has a similar story in principle.

Then there’s the issue of designing equipment. In the United States, we don’t just have SI-dimensioned or scaled parts. We also have “English” parts. And of course we have “adapters” that interface between the two. Even with lots of expense in stocking all three categories, including all the attention this absorbs, it’s amazing how often you still can’t put things together, and all the time spent trying to. I don’t just mean the sizes of mechanical components, either. Pressure gages have pressures. Flowmeters have flows. Instruments of all kinds have ranges.

And don’t get me started on “soft metric” engineering, where you design with metric dimensions everywhere, except that the raw engineering materials - the beams, the shafts, the flat stock - often has English dimensions, so that many parts have very odd values for their otherwise arbitrary dimensions.

This whole mess is for nothing, absolutely nothing. In the Magna Carta, they demanded “Let there be one measure of wine throughout the realm”, because very nearly 800 years ago dealing with multiple units was already such a mess - and we’re STILL arguing over this stupid disaster, with NO MERIT AT ALL on one side of the argument.

As to what people do to get by in this horrible situation, many of us begin each calculation by converting everything to SI, and end it by converting to other units if it is really necessary. It is hard for me to imagine a good reason for doing otherwise.

[Balthisar]

@Sunspace

… especially since our regulations are in SI but our small-scalle typical-handyman construction still tends to take place in feet and inches.

Every bit of documentation pertaining to industrial plant safety from the CSA had a mix of imperial and metric units. Fall protection requirements, for example, were always specified in feet. Arc-flash in joules. Etc.

@Diceman

The US automobile industry uses SI for all its design work. Every part on the car has its dimensions spelled out in millimeters – even big parts, where centimeters or even meters might seem more logical.

There aren’t really any dimensions on a car where a meter is a logical unit of measurement. And centimeters are too big, too. All of our tolerances are in millimeters or fractions of millimeters, so it’s the logical thing to use. Have you ever had to measure sheet metal thickness in mils? Why bother measuring it in meters or cm then? Even the longest possible dimension of a car – its length – is composed of thousands and thousands of smaller dimensions. It’s easier just to use mm for everything.

[Una_Persson]

Even designs for 100+ meter-tall power plants are often expressed in mm, despite it giving very large numbers.

[Napier]

In complicated work it’s really easier not to use any of the prefixes at all, and just use scientific notation for all the numeric values. The prefixes would amount to little adders usually quantized by three, not yet collected into the exponents of the numeric values.

[ZenBeam]

@Una_Persson

Even designs for 100+ meter-tall power plants are often expressed in mm, despite it giving very large numbers.

I find it easier to read large numbers in small units, than small number in large units.

Maybe because the small numbers in large units have all those zeros, followed by however many digits of accuracy. For large numbers in small units, if you already have five or six digits of accuracy, having a number in the 100,000s doesn’t require any more digits.

[scr4]

@Dr.Strangelove

…the metric prefixes are overrated. There’s a reason that essentially nobody uses megameters or centiseconds or hectojoules: it’s because they don’t correspond to quantities that people care about. Instead, people pick a unit that works well for their purposes (nanometers f you’re a computer chip designer, femtoseconds if you do serious laser research,etc.) and pretty much stick with that. The powers of 10 again don’t really enter into it.

I think you just described an advantage of the metric prefixes: you can pick one that’s convenient for your field and still stay within the SI system. For example, we solar physicists do use Mm (megameters) because that’s a typical scale for structures on the Sun. We didn’t have to invent a new unit, or use very large numbers with common units. A conveniently sized unit was already available in the SI system. And someone outside the field who has never seen the unit before would still know what it means, and know how to convert to/from it.

[Dr.Strangelove]

@scr4

I think you just described an advantage of the metric prefixes: you can pick one that’s convenient for your field and still stay within the SI system.

Fair enough. I still maintain they’re overrated, especially for a function that gives the system its name, but I’ll allow that they’re convenient in some cases (and I should have picked something more obscure than megameters :)). Though I’m pretty sure most people, probably even most scientists, will have to look up the exponent if all your numbers are in yoctograms :).

Actually, this brings up a novel thought: the main advantage of metric prefixes isn’t really to the scientific or engineering community; it’s to the general public (and to scientists speaking casually). Scientists don’t have a problem with scientific notation, and using that plus consistent base units works very well. We could have done that with customary units by simply stripping away all the extraneous units for length, weight, volume, energy, etc., and keeping a consistent core (ft, slug, ft^3, ft-lb, etc.).

But you’re never going to convince the public to say it’s 15000 feet to the store, let alone 1.5x10^4 feet. So you need extra units if the system is to be acceptable to the public, and prefix+unit is a better compromise than true new unit. I don’t know if it was planned or lucky that a kilometer is “about” a mile (same order of magnitude), but I’m sure it helped with public acceptance.

So I might be convinced after all that the prefixes are a critical component–just not for the yay science reasons that are normally presented.

[Chronos]

I don’t know if it was planned or lucky that a kilometer is “about” a mile (same order of magnitude), but I’m sure it helped with public acceptance.

Neither, I would say. It’s the nature of the metric system that there was going to be some unit of the same order of magnitude as the mile. It happens to have been the kilometer, but if it had been the hectameter, we’d see that everywhere instead.

[Dr.Strangelove]

@Chronos

Neither, I would say. It’s the nature of the metric system that there was going to be some unit of the same order of magnitude as the mile. It happens to have been the kilometer, but if it had been the hectameter, we’d see that everywhere instead.

Only if it ended up in the 1/1000 to 1000 range. Past that, there are only names for every 3 orders of magnitude. If the meter had really been an inch long, we’d want a name for 100,000 meters. A million is too big.

I suppose that need might have led them to invent such a name, I’ll grant…

[Dancer_Flight]

@Dr.Strangelove

This hasn’t happened universally, though, and so (for instance) for aviation the knot is preferred over km/h or m/s. In fact, the knot is “accepted for use” within SI due to its common use.

Canadian checking in.

I’m in that awkward generation that went through school with nothing but metric but had parents who knew mostly imperial. I’ve pretty much had to learn the conversions in order to get quantities correct back and forth. I tend to default to metric; I’m a (Degreed) science nerd from way back. Although working in a hardware store has me pretty good with the smaller imperial units (but don’t get me started on goods we buy by the meter and sell by the foot).

I, also, for a few years, sailed in local boat races with other like-minded folks and we used a handy (though not exact) relationship between knots and m/s for estimating speeds when we were going too slow for the knot log:

1 nautical mile is aproximately 1800m (1852 exactly), and there are 3600 seconds/hour, so 1 knot is ~0.5m/s. For an astronomer by training, getting to within a decimal place is accurate enough most times*, it’s handy.
Cheers,
-DF

• An unforgettable astronomical conversion was taught in one particular class by a professor showing a quick calculation/guesstimate and uttering the words “Using my vast store of nerdly lore, I know that 1 km/s is 1 parsec/megayear …” [it’s actually 1.023 but close enough].

[Dr.Strangelove]

@Dancer_Flight

1 nautical mile is aproximately 1800m (1852 exactly), and there are 3600 seconds/hour, so 1 knot is ~0.5m/s. For an astronomer by training, getting to within a decimal place is accurate enough most times*, it’s handy.

• An unforgettable astronomical conversion was taught in one particular class by a professor showing a quick calculation/guesstimate and uttering the words “Using my vast store of nerdly lore, I know that 1 km/s is 1 parsec/megayear …” [it’s actually 1.023 but close enough].

Both good ones–I’ll have to remember them. A favorite of my own, which has in fact occasionally come in handy, is that a light-nanosecond is about a foot (<2%).

[psychonaut]

@Dr.Strangelove

This hasn’t happened universally, though, and so (for instance) for aviation the knot is preferred over km/h or m/s. In fact, the knot is “accepted for use” within SI due to its common use.

Not quite. The knot is not an SI unit, and is therefore not used “within” SI. The term used by the International Bureau of Weights and Measures and other national-level standards organizations such as NIST is “accepted for use with” (i.e., alongside) SI units, but even then only in certain fields where they are already well established.

[Dr.Strangelove]

@psychonaut

Not quite. The knot is not an SI unit, and is therefore not used “within” SI. The term used by the International Bureau of Weights and Measures and other national-level standards organizations such as NIST is “accepted for use with” (i.e., alongside) SI units, but even then only in certain fields where they are already well established.

I’ll accept that, though it seems like a minor quibble to me. For what it’s worth, the humble liter also falls under the “non-SI accepted for use with” classification.