Max # digits of precision a thermometer could provide?

What are the theoretical and practical boundaries?

...depending on what you mean by "thermometer." I'd imagine that using a pressure transducer to measure metallic expansion/contraction could offer a great deal of precision.

As far as I know, the winner in the fine scale temp detection contest is the jewel beetle, Melanophila acuminate, a buprestid bark beetle that lays its eggs in freshly burned trees. It can detect IR in the 3 micrometer range from forest fires at rather amazing distances (the literature varies on this, with values of up to 80 km appearing). I can't find anything on the energy flux densities involved, but the general principle is a receptor that responds to minute fluctuations in heat (it absorbs IR and then heats)-- the receptor expands and triggers mechanoreceptors-- the biological analog of the pressure transducer method.

Maybe an engineer who knows more about temp detection will respond and fill in some more details.

for an "uncontrolled" environment, the best I've seen is the ASL F900 (~2ppb = ±0.005° for 'normal' temps - or ±0.000125° if you're perfectly set up) The main limit for a standard thermistor type is the quality of the components - make a wire an extra 1/1000000th thicker and the whole thing turns to custard.

If you're in a lab experiment, you may be able to work the temperature exactly from observing phase changes in the materials being studied, since they can be worked out precisely from fundamental physics.

Hope this helps... :)

My bad.

But heat is rarely evenly distributed. The greater the temperature differential between ambient and what you are measuring, the greater the deviation. Thermal circulation, mass and thermal isolation play big parts too.

In the old days, we used a Wheatstone Bridge to measure small changes of resistance in a thermistor. Calibration could be a bitch. Now days, it's easy.

It's not that the technology doesn't exist, but there is a practical limit to the usefullness of the data.

I'm sure Heisenberg's Uncertainty principle would enter into the game. That probably is your maximum precision. However, for practical purposes, it is fairly easy to get within a tenth of a degree C.

L-

which is analogue. The fact that the readout is digital doesn't make it any more accurate.
A cheap digital home thermometer can easily be off by a degree or so.
For scientific purposes accuracy down to 1/100 of a degree or better can be achieved. The more accurate, the more expensive.

if you're using an analog thermometer. It does measure exactly what the temperature is - you're only limited by how precisely you can read it.

For a digital thermometer, the boundaries are determined by the electronics used to measure the signal generated by whatever is doing the measuring (usually a thermocouple). The more digits you want it to read out to, the more you're going to pay, and it becomes exponentially more so. The only real theoretical limit, then, would be that you can't possibly cram in an infinite number of modules to get an exact result.

There is a misconception that digital devices are somehow "better" than analog when it comes to this, but the truth is the opposite. Most humans are just too lazy to be able to effectively read something like a dial or mercury thermometer.

For example, aren't there very small, random fluctuations in the length of a column of mercury that are not responses to changes in temperature?

At any given temperature, mercury atoms are vibrating
around just as with any atoms. At macroscopic scales,
you don't notice this, but at the truly microscopic
scales, you're correct; you would, so yes, that's
another limit on the fundamental resolution of our
thermometer.

Tesha

Any thermometer is only going to tell you bulk fluid temperature. What you're talking about would be very, very localized, to the point that our measuring instruments would too large, physically, to measure it.

The "infinite" comment means that analog thermometers have continuous ranges, not discrete ones - i.e., the digital thermometer will have to jump from 0.0 to 0.1 - it's either/or, not in between. The analog can measure an in-between value.

Temperature is just a simplistic measure of ambient energy in a given region. As such, your measurement could theoretically be as refined as Planck's Constant.