How does one read 0.1 of one degree Celsius using an Mercury thermometer?

It's certainly possible to make a mercury thermometer that can be read to 0.1 degree Celsius. The World Meteorological Organization standard requires a thermometer with 0.2 Kelvin markings, so reading to the nearest half mark shouldn't be difficult.

Knowing you, though, I don't think this is what your question is really about. Here is a question for you: are there temperature readings taken with a mercury thermometer to the nearest 0.1 degree Celsius that you question? If there are, could you give a specific example?

I suspect what you're trying to imply is that giving the CALCULATED (not measured with any thermometer) "global mean temperature" to 0.1 degrees C is too high an accuracy. Actually, it's not. The first thing to realize is that it is a calculated quantity--it doesn't represent ANY temperature that can actually be measured with a thermometer.

Let's say, for the moment, that we have 10,000 temperatures all measured to 1 degree Celsius (1 K) accuracy (even you should admit that measuring a temperature to that accuracy is possible). Now take all those temperatures and sum them. You might have numbers from -10 C to 30 C, suppose, or in Kelvin from 263 to 303. So when you add those 10,000 numbers, all accurate to the units place, you might end up with a number like:

2737645

Note that under the ordinary rules of significant digits, this derived quantity is still accurate to the units place. What happens now if divide by the exact number 10,000, do we lose any significant digits? No, we don't, because we're dividing by a number with infinite accuracy.

So, I could legitimately give this derived quantity (not a real temperature that can be measured with a thermometer) as

273.7645

However, that's ignoring other things that limit the accuracy of this quantity other than the precision of the temperature readings that went into it, so typically a calculated "global mean temperature" would not be given to that accuracy. However, you can see that giving it to the tenths place is not really a problem at all.

EDIT: Ian, you've missed the point, as usual.

Pegminer has already given an excellent answer. There are research grade mercury thermometers with 0.1 C divisions. I have 2 or 3 in my lab, and more than 20 of the 1 C per division type. I use thermocouples almost exclusively now. One of my students broke a thermometer trying to remove it from a cork and cut his palm. The paperwork that followed caused me more pain than the student felt. I will donate my mercury thermometers to any educational or research institution that asks for them.

It is technically feasible to read a mercury thermometer to better than 0.1 C precision. One method would be to measure the capacitance of the thermometer tube in the measurement region. Mercury has a very different dielectric constant than air, so a change in the level of mercury would give a change in capacitance and hence a change in voltage that can be measured with high precision. The mercury level could also be measured to within a few microns optically with a photo diode array. Mercury makes a very good mirror. If the thermometer was used as one arm of a Michelson interferometer, the level of mercury could be measured with a precision of a few nanometers corresponding to a change in temperature of perhaps 1E-5 K.

The above answer of averaging 10,000 measurements is technically correct as stated. It however is not correct with regard to comparing thermometers from different manufactures, batches and manufacturing time periods and on mercury and alcohol meniscus read thermometers, you also have variations between readers - some read high some read low and depending on the mix of high or low error the correct answer will shifted toward the highest number of erroneous readings. Consequently while you can average error on a single instrument and to some degree by a single reader, averaging the temperatures read on thousands of thermometers of different mfg. sources, batches, and batches and readers with unknown variable ranges - only averages the readings, but gives no indication of a correct reading. In about 1980 I did a comparison of a single batch of 25 laboratory red spirit thermometers (-10-100 C) from a single US manufacturer. Though the state range of accuracy was +/- 1 degree C, the thermometers in the same beaker of water at room teimp. ranged 2 degree C. Thermometer accuracy and their readings are generally over stated. Add multiple readers and you get more variation. So, the "best" answer above is only correct in one situation - same thermometer and same reader.

You use a properly designed mercury thermometer. Oceanographers have been doing so for over 100 years. They even figured out a way, before electronics, to preserve a temperature measurement taken at depth so that you could read the thermometer after you get it back to the surface. Or - from your perspective - they use magic you don't understand.

You would need a thermometer that showed tenths. Try calling a medical supply house or search online for one. Don't have a link for you. One site mentioned one which measured tenths and said it was a medical thermometer. Find the actual name of one and then search amazon

Pegminer's answer is pretty spot on and answers your question completely. My guess is he will not receive best answer as I know you're not looking for an actual answer but more of an opportunity to be the king denier idiot for the day.

Way to go...

Also congrats to the other dogmatic idealists and their answers that completely reject long standing (i.e. long before it became hip to call everything socialist) scientific practices.

For you all, I recommend some good-hard praying as god punishes the willfully ignorant.

With a mercury-based basal thermometer.

http://www.ehow.com/how-does_4925936_basal-thermom...

In the context of global climate change, this "question" is bizarre, repeated twice or not.

I am not sure what tradition holds in cricket, but in American baseball, no one denies that the difference between a batting average of .250 and .350 is the difference between a poor-to-middling batter and a serious contender for league champion, DESPITE the fact that no one has ever observed 0.1 of a base hit.

You usually don't, since they're not that precisely marked and the Eyeball isn't accurate enough anyway.

Very carefully...and then adjust it upwards 1 C because you probably made a mistake.

@Pegminer..." CALCULATED (not measured with any thermometer) "global mean temperature" to 0.1 degrees C is too high an accuracy. Actually, it's not."

Really? Because they also have to extrapolate temperatures over THOUSANDS of miles. In the mid 1800's they were extrapolating temps over entire continents. Accuracy to 0.1C...my buttocks.