Precision, Accuracy, and Reproducibility
The precision or resolution of a thermometer is simply to what fraction of a degree it is possible to make a reading. For high temperature work it may only be possible to measure to the nearest 10 °C or more. Clinical thermometers and many electronic thermometers are usually readable to 0.1 °C. Special instruments can give readings to one thousandth of a degree. However, this precision does not mean the reading is true or accurate.
Thermometers which are calibrated to known fixed points (e.g. 0 and 100 °C) will be accurate (i.e. will give a true reading) at those points. Most thermometers are originally calibrated to a constant-volume gas thermometer. In between a process of interpolation is used, generally a linear one. This may give significant differences between different types of thermometer at points far away from the fixed points. For example the expansion of mercury in a glass thermometer is slightly different from the change in resistance of a platinum resistance of the thermometer, so these will disagree slightly at around 50 °C. There may be other causes due to imperfections in the instrument, e.g. in a liquid-in-glass thermometer if the capillary tube varies in diameter.
For many purposes reproducibility is important. That is, does the same thermometer give the same reading for the same temperature (or do replacement or multiple thermometers give the same reading)? Reproducible temperature measurement means that comparisons are valid in scientific experiments and industrial processes are consistent. Thus if the same type of thermometer is calibrated in the same way its readings will be valid even if it is slightly inaccurate compared to the absolute scale.
An example of a reference thermometer used to check others to industrial standards would be a platinum resistance thermometer with a digital display to 0.1 °C (its precision) which has been calibrated at 5 points against national standards (−18, 0, 40, 70, 100 °C) and which is certified to an accuracy of ±0.2 °C.
According to British Standards, correctly calibrated, used and maintained liquid-in-glass thermometers can achieve a measurement uncertainty of ±0.01 °C in the range 0 to 100 °C, and a larger uncertainty outside this range: ±0.05 °C up to 200 or down to −40 °C, ±0.2 °C up to 450 or down to −80 °C.
Read more about this topic: Thermometer