Apparently thermometers ... are more difficult to make than I'd ever have imagined. Over the past few years, we've tried perhaps a half-dozen of the roughly-ten-dollar electronic oral thermometers. None of them worked reliably, nor were any of them capable of repeatable readings (that is, taking the same reading over and over getting the same results). We also tried one of the infrared through-the-ear thermometers, only to find out that they worked well only if the sensor was actually pointed directly at your eardrum. We got crazy results with that one.
Last week I decided to finally solve this problem. I looked first for an old-fashioned mercury oral thermometer. They're hard to find these days, unless you by antiques on eBay. The few that I did find all had unacceptably bad absolute accuracies (typically +/- 3°F or 4°F). So I ended up going for another electronic thermometer, but this time one of the models actually used in doctor's offices (a Welch Allyn SureTemp Plus 690 electronic thermometer. We received it yesterday. When I opened the package, the first thing I saw was a DVD with operating instructions – that was a little worrisome, as I wouldn't have guessed that a thermometer would need instructions at all! The next thing I found was a little bag of parts, and no assembly directions. No worries, they only went together in one way, and it took less than 60 seconds to have it all put together.
So how hard is it to operate? I can't imagine how they could have made it any easier. To take your temperature, you just pull the probe out of its holder and stick it under your tongue. The act of taking the probe out turns the unit on. It takes about 5 seconds to get a reading, accurate to +/- 0.2°F, and repeatable in my testing with no error at all.
I haven't bothered opening the DVD. :)
Of course I had to open the thing up to see if I could figure out why this temperature measurement was so hard to do. Being a modern digital instrument, I couldn't really tell very much – except that there are significantly more discrete components in this model than in any of the ten dollar thermometers we bought. There are also adjustments, presumably for calibration. And there is a significantly more powerful little microprocessor, which you'd expect with an instrument that has a nice display. None of this, however, gives me any real idea why the temperature measurement seems to be such a challenging thing. In particular, I'm very surprised that there seems to be no thermometers available for prices between the ten dollar thermometers and the $200+ doctor's office thermometers. That seems like a gap someone should be exploiting!