Inventing Temperature Read online

  That seems like a fair assessment in some ways, especially if one focuses on theoretical developments as the most exciting part of scientific progress. However, in some other ways I think Duhem was closer to the mark when he said that Regnault had effected a "true revolution" in physics. To see why, it is important to recognize the ways in which he was more than just an exceptionally careful and skilled laboratory technician. In that respect, it is informative to compare the character of his work with that of his predecessors in the tradition of precision measurement. I will not be able to make that comparison in any comprehensive sense, but as a start it is very instructive to compare Regnault's thermometry with Dulong and Petit's thermometry. Since Dulong and Petit were seen as the unquestioned masters of precision experiments in French physics before Regnault, their work is the best benchmark against which we can assess Regnault's innovations.

  In retrospect, Dulong and Petit's decisive contribution to thermometry was to highlight the urgency of the need to make a rational choice of the thermometric fluid. This they achieved in two important ways. First of all they demonstrated that the magnitude of the mercury-air discrepancy was very significant. While confirming Gay-Lussac's earlier result that the mercury thermometer and the air thermometer agreed perfectly well with each other between the freezing and boiling points of water, Dulong and Petit carried out the comparison in high temperatures, where no one before them had been able to make accurate determinations. Their results showed that the discrepancy between mercury and air thermometers increased as temperature went up, reaching about 10 degrees at around 350 degrees centigrade (mercury giving higher numbers).73 That amount of discrepancy clearly made it impossible to use the two thermometric fluids interchangeably. Second, it was not

  71. See Dörries 1998b, esp. 128-131; the quoted passage is from p. 123.

  72. Dörries 1997, 162-164, gives a summary of some of the critical appraisals.

  73. See Dulong and Petit 1817, 117-120, including a summary of results in table 1 on p. 120. See also Dulong and Petit 1816, 250, 252.

  end p.100

  plausible to attribute this mercury-air discrepancy to experimental error, given the extraordinary care and virtuosity apparent in Dulong and Petit's experimental procedures. They were justly proud of their achievement in this direction and asserted that they had reached the highest possible precision in this type of experiment.74 No one credibly challenged their confidence—until Regnault.

  Were Dulong and Petit able to reach a definitive verdict on the choice of thermometric fluids, with their superior skills in precision measurement? They certainly set out to do so. Their article began by noting the failure of De Luc and Dalton to provide satisfactory answers to this question. Their chief criticism of Dalton, especially, was that his doctrines were not based on empirical data, and the presumption was that their own precision experiments would provide the needed data.75 For their own positive contribution, Dulong and Petit (1817, 116) started by stating the requirement of a true thermometer, based on a simple conception of temperature that abandoned all Laplacian sophistication: if additions of equal amounts of heat produce equal increases in the volume of a substance, then that is the perfect thermometric substance. However, they did not consider that condition to be amenable to a direct empirical test, since the quantity of heat was a difficult variable to measure especially at higher temperatures. Instead, their strategy was to start by using the standard mercury thermometer in order to observe the thermal expansion of some candidate substances that were free from obvious factors disturbing the uniformity of expansion, such as gases and metals.

  What these observations were meant to allow them to conclude is not absolutely clear, but the thought seems to have been the following (Dulong and Petit 1816, 243). If many candidate substances display the same pattern of thermal expansion, then each of them should be taken to be expanding uniformly. Such agreement is most likely an indication that disturbing factors are not significant, presumably because it would be a very unlikely coincidence if the disturbing factors, which would have to be different for different substances, resulted in exactly the same distortions in the patterns of expansion. Their empirical research revealed that this expectation of uniformity was not fulfilled across different metals, so they concluded that gases were the best thermometric substances.76 This is an interesting argument but it is ultimately disappointing, first of all because it does not constitute any theoretical advancement over the earlier calorist inference that Gay-Lussac and Dalton had enough perspicuity to distrust. As Dulong and Petit themselves recognized, this was only a plausibility argument—and the plausibility was in fact significantly diminished in the absence of the support by the calorist metaphysics of mutually attracting matter particles held apart by self-repulsive caloric.

  74. Dulong and Petit 1817, 119. A detailed description of their procedures is given in Dulong and Petit 1816, 245-249.

  75. Dulong and Petit 1816, 241-242; Dulong and Petit 1817, 114-115.

  76. Dulong and Petit 1817, 153. For the results on metals, see Dulong and Petit 1816, 263, and Dulong and Petit 1817, 136-150. There is no indication that they carried out any extensive new work showing the uniformity of thermal expansion in different types of gases. Their articles only reported experiments on atmospheric air; aside from that, they merely referred to Gay-Lussac's old results regarding the uniformity across gas types. See Dulong and Petit 1816, 243.

  end p.101

  Dulong and Petit failed to solve the problem of thermometric fluids because their work was not framed with philosophical sophistication, not because their experimental technique was insufficient. Only Regnault solved the epistemic problem. Dulong and Petit started on the wrong foot by defining temperature in a nonobservational way; the argument they envisioned for the demonstration of linear expansion was bound to be weak, however good their data might have been. In contrast, as explained in the three previous sections, Regnault devised the strongest possible argumentative strategy and was fortunate enough for that strategy to work. Regnault made the advancement from thermoscopes to numerical thermometers (from stage 2 to stage 3 of temperature standards) about as well as could have been done under the material conditions on earth. In more common scientific terms, he brought practical thermometry to utmost perfection.

  Surely, however, there were directions in which Regnault's work left room for further progress. Being completely independent of theory was a clear virtue up to a certain point, but there was a later time when connecting to heat theory was a valuable thing to do. The further development of theoretical thermometry, especially in the hands of William Thomson, is treated in chapter 4. Before we come to that, however, there is another story to tell. Perfect as Regnault's air thermometers were, they were not adapted for extreme temperatures, especially the high end where glass ceased to be robust. In the next chapter we will retrace some key steps in the measurements of very low and very high temperatures, starting again in the middle of the eighteenth century.

  end p.102

  3. To Go Beyond

  Narrative: Measuring Temperature When Thermometers Melt and Freeze

  Abstract: This chapter discusses how the challenge of extending the numerical scale beyond the temperature range in which it was established was addressed. It focuses on the investigations into the freezing point of mercury, and the master potter Wedgwood's efforts to create a thermometer that could measure temperature in his kilns. A revitalized version of Percy Bridgman's operationalist philosophy is used to shed light on the process of extending concepts beyond the domains of phenomena in which they were originally formulated.

  Keywords: temperature, numerical scale, thermometer

  Hasok Chang

  Now, when it is desired to determine the magnitude of some high temperature, the target emissivity is established using a reflected laser beam, the temperature is measured by an infrared-sensing, two-colour pyrometer, information is automatically logged into a computer data bank, and the engineer in charge gives
no thought to the possibility that it might not always have been done this way.

  J. W. Matousek, "Temperature Measurements in Olden Tymes," 1990

  In the last two chapters I examined the establishment of the most basic elements in the measurement of temperature: fixed points and a numerical scale. The narrative in chapter 2 focused on the efforts to establish the numerical scale with increased rigor, continuing into the mid-nineteenth century. But as soon as a reasonable numerical scale was established, a different kind of objective also became apparent: to extend the scale beyond the temperature range in which it was comfortably established. The obvious challenge was that the mercury thermometer physically broke down when mercury froze or boiled. Gaining any definite knowledge of thermal phenomena beyond those breaking-points was very much like mapping a previously uncharted territory. In the narrative part of this chapter I will present a double narrative representing how the challenge was tackled at both ends, focusing on the investigations into the freezing point of mercury, and the master potter Wedgwood's efforts to create a thermometer that could measure the temperature of his kilns. Both are stories of surprising success combined with instructive failure. The analysis will discuss further some crucial issues of justification and meaning raised in these narratives about the extension of empirical knowledge beyond its established domains. There I will use a revitalized version of Percy

  end p.103

  Bridgman's operationalist philosophy to shed further light on the process of extending concepts beyond the domains of phenomena in which they were originally formulated.

  Can Mercury Be Frozen?

  Johann Georg Gmelin (1709-1755), professor of chemistry and natural history at the Imperial Academy in St. Petersburg, had the enormous challenge of leading a team of scholarly observers on a ten-year trek across Siberia starting in 1733.1 The expedition had been ordered by Empress Anna Ivanovna, who sought to realize a favorite idea of her uncle, Peter the Great, to acquire better knowledge of the vast eastern stretches of the Russian Empire. There was also a plan to make a grand rendezvous at the other end with the sea expedition led by Captains Vitus Bering (1681-1741) and Alexei Chirikov (1703-1748), to explore access to America. Gmelin's party endured a daunting degree of cold. In some places they found that even in the summer the earth had several feet of frozen soil underneath the surface. At Yeniseisk during their second winter, Gmelin recorded (quoted in Blagden 1783, 362-363): The air seemed as if it were frozen, with the appearance of a fog, which did not suffer the smoke to ascend as it issued from the chimnies. Birds fell down out of the air as if dead, and froze immediately, unless they were brought into a warm room. Whenever the door was opened, a fog suddenly formed round it. During the day, short as it was, parhelia and haloes round the sun were frequently seen, and in the night mock moons and haloes about the moon.

  It was impossible to sense the exact degree of this numbing cold. However, Gmelin noted with satisfaction: "[O]ur thermometer, not subject to the same deception as the senses, left us no doubt of the excessive cold; for the quicksilver in it was reduced to −120° of Fahrenheit's scale [−84.4°C]." This observation astonished the world's scientists, as it was by far the lowest temperature ever recorded anywhere on earth. For example, William Watson (1715-1787), renowned naturalist and "electrician," later to be physician to the London Foundling Hospital, noted that "such an excess of cold can scarcely have been supposed to exist, had not these experiments demonstrated the reality of it." Gmelin's observations were "scarce to be doubted" thanks to the thermometer (Watson 1753, 108-109).

  To Charles Blagden (1748-1820) examining this account half a century later, however, Gmelin's mistake was apparent. It did not seem likely at all that even the Siberian winter temperatures would have been so much as almost 100°F lower than the lowest temperatures observed in northern Europe. Blagden (1783, 371) inferred that the mercury in Gmelin's thermometer must have actually frozen and shrunk drastically, indicating a much lower temperature than actual. Gmelin had not

  1. The account of Gmelin's Siberian expedition is taken from Blagden 1783, 360-371, and Vladislav Kruta's entry on Gmelin in the Dictionary of Scientific Biography, 5:427-429.

  end p.104

  considered the possibility of the freezing of mercury, which was commonly considered as essentially fluid at that time.2 In fact he rejected the idea summarily, when it was brought to his attention forcefully two years later in Yakutsk when one of his colleagues3 noted that the mercury in his barometer was frozen. Gmelin was shown the solidified mercury, but convinced himself that this was due to the presence of water in the mercury, which had been purified using vinegar and salt. He confirmed this explanation by taking the mercury out of the barometer, drying it well, and seeing that it would not freeze again, "though exposed to a much greater degree of cold, as shown by the thermometer." By which thermometer? To Blagden (1783, 364-366), Gmelin's "confirmation" only confirmed that the mercury thermometer could not be trusted. The following winter, Gmelin even observed the evidence of mercury congelation in his own instruments, when he noted that the mercury columns in his thermometer and barometer were broken up by air bubbles. He had some trouble explaining this appearance, but he refused to consider freezing as a possibility (Blagden 1783, 368-369).

  The first demonstration that mercury could really be frozen came only twenty-five years later, and that was the work of Joseph Adam Braun (1712?-1768), professor of physics at the St. Petersburg Academy. The winter of 1759-60 in St. Petersburg was very severe, recorded temperatures reaching 212° Delisle in December, which is equivalent to −41.3°C, or −42.4°F. (The Delisle temperature scale, which I will come back to in "Temperature, Heat, and Cold" in chapter 4, was set at 0° at the boiling point of water, and had increasing numbers with increasing cold, with the freezing point of water designated as 150°.) Braun took that opportunity to produce the greatest degree of artificial cold ever observed, using a "freezing mixture" of aqua fortis (nitric acid, HNO 3 ) mixed with snow.4 The mercury thermometer descended beyond 600° Delisle (−300°C), and the mercury in it was quite frozen, which Braun confirmed beyond doubt by breaking the thermometer (not an insignificant material sacrifice at that time) and examining the solid metal (Watson 1761, 158-164). As far as Braun could see, the solidification of mercury was no different from the freezing of any liquid, and hence a mere effect of the "interposition of cold."

  2. See Watson 1761, 157, for evidence that this attitude was current well into the middle of the eighteenth century: "[F]or who did not consider quicksilver, as a body, which would preserve its fluidity in every degree of cold?"

  3. This was probably the astronomer Louis de l'Isle de la Croyere—brother of Joseph-Nicolas Delisle, who devised the temperature scale popular in Russia.

  4. "Freezing mixtures" provided the only plausible means of producing extremely cold temperatures at the time. The cooling is produced by the action of a substance (most often an acid) added to snow (or ice), which causes the latter to melt and absorb a great deal of heat (latent heat of fusion) from the surroundings. Fahrenheit is reputed to be the first person to have used freezing mixtures, though it seems quite likely that Cornelius Drebbel had used it much earlier. Before Joseph Black's work on latent heat, the working of freezing mixtures must have seemed quite mysterious. As Watson put it (1761, 169): "That inflammable spirits should produce cold, seems very extraordinary, as rectified spirit seems to be liquid fire itself; and what still appears more paradoxical is, that inflammable spirits poured into water, causes heat; upon snow, cold: and what is water, but melted snow?"

  end p.105

  Braun's work was certainly seen as wondrous. William Watson exclaimed in his official report of Braun's work to the Royal Society of London: Who, before Mr. Braun's discovery, would have ventured to affirm mercury to be a malleable metal? Who, that so intense a degree of cold could be produced by any means? Who, that the effects of pouring nitrous acid upon snow, should so far exceed those, which result from mixing it with ice … ?
(Watson 1761, 172)

  Afterwards it was confirmed that mercury could be frozen by a natural cold after all, by another German naturalist working in Russia, Peter Simon Pallas (1741-1811). Pallas was invited by Catherine the Great to lead a Siberian expedition, which he did successfully from 1768 to 1774.5 In December 1772 he observed the freezing of the mercury in his thermometer, and then confirmed the fact more definitely by freezing about a quarter-pound of mercury in a saucer.6 Those who still found the freezing of mercury difficult to accept tried to explain it away by casting doubts on the purity of Pallas's mercury, but in the end various other experiments convinced the skeptics.

  The immediate and easy lesson from the story of freezing mercury is that unexpected things can and do happen when we go beyond the realms of phenomena that are familiar to us. The utilitarian jurist Jeremy Bentham (1748-1832) used this case to illustrate how our willingness to believe is tied to familiarity. When Bentham mentioned Braun's experiment to a "learned doctor" in London, this is the reaction he got: "With an air of authority, that age is not unapt to assume in its intercourse with youth, [he] pronounced the history to be a lie, and such a one as a man ought to take shame to himself for presuming to bring to view in any other character." Bentham compared this with the tale of the Dutch voyagers (reported by John Locke), who were denounced by the king of Siam "with a laugh of scorn" when they told him that in the Netherlands water would become solid in the winter so that people and even wagons could travel on it.7 Locke's story may be apocryphal, but the philosophical point stands.