(illustration - AASp225.jpg, AASp225a.jpg, AASp225b.jpg are attached. Caption could be "Beware of cheaply manufactured thermometers." The calibration marks are engraved into the glass before the scale is printed. a shows how the scales can be printed incorrectly on 3 thermometers. - bowed so not the best photo. b shows one thermometer with the error at the 90C calibration. c shows a thermometer with an error at the 10C calibration.) Margaret McNeil of Dunedin, asks :-

I have two air-temperature thermometers, one about 110 years old and a new cheap one, both made by the same manufacturer and both containing a red fluid. They register about 3 degrees Celsius different. Why? And how can I find out which one is accurate?

John Campbell, a physicist at the University of Canterbury and who taught a course in measurement techniques, responded.

There are several problems involved here but the main one is how well they were calibrated after manufacture. To be reasonably accurate, every measurement must be traced back to a standard. The more accurate the calibration, the more expensive the thermometer.

The modern temperature scale is the Celsius scale, which is defined by water freezing at 0C and boiling at 100C under standard conditions. Any physical property that changes with temperature can be used as the basis for a thermometer. The difference in expansion between a liquid and glass was widely used before electronics became so cheap. Mercury, a metal which can be purified by distillation, in glass was the usual thermometer. These were manufactured by drawing a glass rod with a small capillary down the centre. Because of wear on the dies, and slight variations in the composition of the glass and its physical properties, these were not exactly all the same. For cheap thermometers these were accurate enough. For more accurate mercury-in-glass thermometers designed to register -10C to 110C, as in a school laboratory, this wasn't accurate enough. After manufacture they had to be calibrated against a bath of known temperature as measured by a well-calibrated, more-expensive, thermometer. This was done using baths at 10C then 90C, with the position of the top of the mercury column at those temperatures engraved onto the glass. Even for thermometers made in the same batch the length between these two engraved points can be quite different.

Your old thermometer might be engraved at these, or similar temperatures, but I doubt that the modern cheap one will be. The next step was to print the scale markings onto the theremometers. An operator would optically zoom the scale until the 10C and 90C labels matched the two engraved marks. In inspecting class loads of new thermometers, I could always find several where the scale was carelessly printed up to half of a degree from the relevant engraved mark.

For air temperature use, a cheap organic liquid, such as ethanol, dyed red, in glass is the usual thermometer. The red is much more easily visible at a distance than mercury.

Now, which of your thermometers is correct or wrong? With just two you cannot tell. One might have calibration engravings on it but that just says it was reasonable when manufactured. Do the liquids and dyes alter over long periods of outdoor exposure? Possibly, especially if they have been in direct sunlight. The thermometers doctors use for measuring temperature to check for fevers, alcohol-in-glass thermometers used for accurate meteorological records, and calibrated thermometers in school laboratories, are generally good.

Your best bet would be to find a calibrated thermometer and compare your two with it. Industries which need very accurate temperature monitoring pay a lot of money for Industrial Research Limited, which maintain all New Zealand physical standards and ensure they agree with the rest of the world, to periodically calibrate their thermometers.

I bemoan the loss of mercury-in-glass thermometers from our teaching laboratories. The taceability and calibration was plain to see. Nowadays, electronic thermometers put out the temperature on a digital meter, which students believe implicitly but which are only as good as the whole device's calibration and maintenance.