[ctdonath]

"we don't need more units!"

Yes, some units are useful for particular interests. Create them as they're needed and useful. Base-10 time would make more sense. Celsius at least coordinates notable values with common materials.

Some units are just stupid. The lead article gives an unsatisfactory explanation of base-12/60 time (admitting near the end the reason for base-60 is unknown); trying to explain clocks to toddlers is proving annoying (I can't explain it if it doesn't make sense, and it doesn't make sense). Fahrenheit is just an arbitrary marking on a scale and seeing how reality happened to line up.

http://xkcd.com/927/

[dragontamer]

Fahrenheit was supposed to be 0 degrees for freezing, and 100 for body temperature.

It turns out that Fahrenheit's reference thermometer was a bit... off... however.

[gumby]

Interestingly, Fahrenheit has 180 (212-32) degrees between freezing and boiling. Zero is set to a brine solution -- a reproducible metric about as cold as you could make. So it's not an insane system.

[MarkPNeyer]

0 was the coldest temperature he could reliably reproduce - which is salt water just above the freezing point.

[pyre]

How did it manage to be off by 32 (!) for freezing, but fairly close for body temperature? I would expect a constant (or at least same ballpark) error.

I imagine it had something to do with the medium used for the thermometer?

[stephenhuey]

He was trying to find a salt mixture that went through a phase change at 0 degrees:

http://antoine.frostburg.edu/chem/senese/101/solutions/faq/z...

That wasn't so reliable, and neither was using 100 for human body temperature, so when he discovered that water boiled at 212 he began calibrating with that.