Inventing Temperature (Measurement and Scientific Progress) by Hasok Chang (2004) Oxford University Press ISBN 9870195337389

Review by John Oversby, 2009

1. Keeping the fixed points fixed: what to do when water refuses to boil at the boiling point
Review of various fixed points, including freezing and boiling points of water. Some single point systems identified. The fixed points moved from highly variable (e.g. greatest summer heat) and included the two phase change temperatures of water. Even by 1771, there was some controversy over adopting the two phase change temperatures of water. The values given to these varied, too, even at a fixed atmospheric pressure. The problem of the mercury in the stem not being the same temperature as the mercury in the bulb was solved by having the whole thermometer immersed. The exact moments of boiling proved to be a problem, since water has different stages of boiling, including fast and slow boiling. These could take place over a range of 6-8ºC. The temperature varied according to whether a metal or a glass pot was used, and according to whether it was heated all round or only from the bottom. When water was shaken for four weeks (to remove the air), it now boiled at 112.2ºC. Superheating of water could be an explanation. Most of the scientists of this time thought that the air in the water had a significant effect. The superheating effect, i.e. reaching a high temperature before boiling starts, can be overco0me by noting only steady boiling. The Royal Society chose to avoid the majority of variations by placing the thermometer in the steam, not the water, even though small variations remained in different conditions. Almost all of these experiments affected the theoretical views of boiling. Later 19th century experiments investigated the nature of supersaturated steam, not least because of Aitken's developments in meteorology, and Wilson's Cloud Chamber. Work on dust and fogs also clarified (!) thinking about boiling. Chang proposes:
1. 'If causes of variation can be easily eliminated, eliminate them.'
2. 'If causes of variation cannot be easily eliminated but can be identified and quantified, learn to make corrections.'
3. 'Ignore small, inexplicable variations, and hope they will go away.'
Chang then goes on to similar discussions about the freezing point of water and melting point of ice, discussing super-cooling and super-heating. He concludes that it is not enough to reach or exceed the desired fixed point but that some other process must initiate phase change.
Traditional thermometry in schools is often based on the abstraction of fixed points. At least, teachers should be aware of the history of determining standards for calibrating thermometers. In class, it may be an interesting exercise to look at placing the thermometer in steam or in the boiling water to check differences, then discuss issues of standardising procedures.
2. Spirit, air and quicksilver: the search for the 'real' scale of temperature
Assuming we can calibrate a thermometer at two fixed points, how can we be sure that the thermometric fluid expands linearly with temperature? We can use the method of mixtures (e.g. make known temperatures by mixing different voumes of water at 0ºC and 100ºC) to make a bath at a known temperature. Mercury thermometers are remarkably close to theoretical real temperatures but alcohol and water thermometers are not so good. However it assumed that theoretical specific heat of water was constant as theoretical temperatures changed. The volume change on heating and cooling left variable space for 'caloric' suggesting that specific heat capacity was not constant. The issue of what was heat or caloric was to vex this issue for a long time. Thermal expansion of liquids was an uncertain phenomenon for a long time. Gases seemed to show regularity in expansion, although this was not immediately accepted after it was shown that all gases expanded by 1/266.67 of their volume at 0ºC under atmospheric pressure for one degree. Comparability (Regnault) is a major criterion for establishing a good standard. Comparability is ensuring that different instruments (of the same kind) give the same reading when in the same environment. Regnault also noted the importance of glass when comparing mercury thermometers made with different glasses. Also, the type of thermal treatment used affected the readings. With an air thermometer, the expansion is 160 times that of the glass so the glass expansion becomes less important. Regnaults' work on comparing different gases showed that the differences were always less than 0.1%, and varied randomly. He showed that gases had the same law of expansion, even if the coefficients were different. However, sulphur trioxide gas was not comparable. He chose the air thermometer as one he could have confidence in, probably influenced by the readily availability of air. His was an empirical position, not a theoretical position since good theories of the behaviour of gases did not exist at this time (1847). Chang also emphasises the point that a particular situation can have only one temperature value, that is, it can not be both, say, be uniformly 25ºC and 30ºC at the same time. This is the ontological principle of single value. This is a real world experience, not a logical conclusion.
Increasing replacement of mercury with alcohol thermometers will make it difficult to compare the two. This is why historical data is so important. Use Chang's Fig 2.2 for constaqnt volume air thermometer. Some of the data in this chapter is worth including for pupils.
3. To go beyond: measuring temperature when thermometers melt and freeze
Chang considers extending the temperature scale beyond that where glass containers can be used. He uses as an example Wedgewood's clay furnace thermometers. He asks whethre it is possible to compare ºW to ºC and uses the principle of patchworking i.e. overlapping thermometric materials. The silver thermometer is a case in point. He quotes Gmelin's temperature recording of -84.4ºC using his mercury thermometer as a mistake since the mercury would have frozen and he was recording shrinking solid. To Gmelin this was unthinkable and he tried all sorts of ways to dismiss the idea, according to Chang. To measure the freezing point of mercury using a mercury thermometer that may behave unusually near the melting point seems to be unreasonable. Using an alcohol thermometer is also a problem since the alcohol thermometer was shown to be unreliable. Hutchins in Ontario, by placing a mercury thermometer in a container of mercury that was freezing, determined the freezing point of mercury to be -40ºC. He was following the advice of Cavendish at the Royal Society, itself a development of Joseph Black. A further postexperiment correction of the Hutchins' thermometer revised the exact value to be -38.67ºC. However, super-cooling made the results suspicious. Later, Pouillet used air thermometers to measure the temperature of freezing carbon dioxide (-78.8ºC). It is likely that an air thermometer was not used for the freezing point of mercury since the thermometer is so large. He also used a bismuth-copper thermocouple, calibrated against an air thermometer at ambient temperatures, and established the freezing point of carbon dioxide to within 0.1ºC of that established earlier with an air thermometer.
Wedgewood had reported very high tempertures for his furnaces (17,977ºF) and the melting point of iron (21,877ºF), using his pyrometer, since glass thermometers melted. He was losing so much pottery by faulty temperature control that he searched for a suitable thermometer. He used a clay-iron oxide mixture that shrank on heating, and their final size depended only on the highest temperature reached. It was not an immediate measurement! He assined 1ºW to aq contraction of 1/600 of the width. His pyrometer was instantly in use in brass foundries to avid overheating and consequent waste of fuel. However, there was no way of comparing the Wedgewood scale with that of Fahrenheit, for example. He suggested the expansion of silver scale. Scientist disovered the unreliability of the clay pieces, and supected that the clay did not shrink linearly with temperature. The demise of the Wedgewood scale came with theb availability of platinum, developed by Guyton.
The Gmelin mistake would make a good historical story for analysis. Solid mercury will not be known to thse students. It may be possible to have video of solid mercury! The story would make a great play, and perhaps we could have a short video of it.
4. Theory, measurement and absolute temperature: the quest for the theoretical meaning of temperature

5. Measurement, justification and scientific progress

6. Complementary science – history and philosophy of science as a continuation of science by other means