odd dinner conversation

[cellio]

The question (not mine): if you were building a thermometer (the kind that lives in your medicine cabinet at home), what range of temperatures would you support? I said the problem was insufficiently specified, but that my baseline would be 96-106 and if there's no appreciable expense in widening it, I'd go in the range of 90-110 or -120, because why not. But the problem was still insufficiently specified; I was assuming digital readout, not a column of mercury in a usually-illegibly-marked tube. In the latter case, you want the minimum useful range, because you've got limited real estate for the markings. If you could have those 10 degrees occupy 80% of the tube and have the rest be compressed that'd be different, I said.

So Dani challenged that -- why assume that the tube is uniform? I said because otherwise you're out of the price range of medicine-cabinet thermometers. This, in turn, led to speculation about how that type of thermometer is manufactured; I argued for a large uniform (hollow) rod that's cut to length with ends then treated (seal at one end, mercury + bulb at other), while he argued for individually molded. (Insert tangent about plastic vs. glass here.) Of course, neither of us actually knows anything about this; we're trying to make intelligent guesses and apply design principles from other fields.

I don't think we're the only people who have weird speculative conversations like this, but I never seem to notice stuff like this coming from other tables in restaurants. On the other hand, we haven't been kicked out of any restaurants for annoying the neighbors either. (On the third hand, it seems to take a lot to produce that result.)

Comments

[geekosaur]

The structure of a glass thermometer convinces me that one end is sealed afterward; I'd guess that it's the top end, after filling the bulb with mercury. Whether it's initially formed that way or not is an open question but I could see both ends being sealed afterward.

It's potentially possible to mess up the tube during sealing, but not all that likely; you could form the bulb end from the rod segment with just enough air going through the tube to (a) keep it open and (b) inflate the bulb slightly, then fill the tube, then put a small drop of molten glass on the other end to seal it. (I don't know if they're currently mass-produced that way, but it wouldn't surprise me to find that they were initially created as I described.)