A reader wrote to tell me that he'd replicated my ice-resistance-measuring experiment, with the same results - about ten million ohms per inch. Then he said:
...although in Oz, shouldn't that have been centimetres?
This pressed one of my numerous Talk Buttons, so I thought I'd pour my canned rant on this subject out into a blog post where you all have to put up with it, rather than only favouring that one correspondent with my deathless wisdom.
Fractions of inches are seldom useful for anything (to me), and are a pain to work with too - I've got a lovely little German Imperial-unit vernier caliper that confuses the heck out of me every time I try to use it. Metric vernier scales are easy, but the imperial one is another of those things that slither out of my brain as soon as I put the caliper down.
But metric units just don't come in the right sizes for some measurements. "About an inch", as in the ice-resistance measuring, clearly conveys the rough-eyeball-distance-measuring I was doing. The metric equivalent either suggests an excessive level of precision ("about 2.5cm" gives the impression that the range is no more than 2.3 to 2.7...), or is cumbersome ("between 2 and 3cm").
My favourite example of not-so-useful metrication is in measuring human height. Australian publications usually have a style guide that forbids feet and inches, or at least requires metric equivalents to be added in brackets. So "the suspect in the Brooklyn Slasher murders has been described as being about 6 feet tall" becomes "...about 183cm tall", which again suggests more precision than actually exists in the measurement.
Some people might even say "182.9cm" in this situation, giving the impression that someone's measured the suspect with a micrometer. Since a person's height can easily change by more than an inch depending on what shoes they're wearing and slight changes in posture, I think most human height measurements with precision beyond the inch level are actively misleading.
(Wikipedia has a good little article on "false precision". And here's a piece on seeing false precision where it in fact does not exist. I ramble on about the limits to precision in real-world measurement here.)