Over most of Europe, except in Scandinavia and the British Isles, the 80° Réaumur thermometer quickly attained a commanding position, but nevertheless the more perceptive members of the scientific community, outside the Académie at Paris, soon came to recognize its serious defects. There is a touch of sadness in much of the criticism, a feeling that the thermometer must be an unfortunate lapse on the part of a man greatly esteemed. Indeed it was just that, but it was not a momentary aberration. It would appear that Réaumur had made up his mind in advance about how a thermometer should be constructed, and then proceeded to do it in this way without at any time re-examining his assumptions.
The defects were, so to speak, at both ends and in the middle.
The methods of establishing the fixed points were soon criticized, and the principle of dividing the scale was attacked--most fundamentally in that it took no account of the differences in the expansion of various kinds of glass but also on technical grounds.
We have already noted Josias Weitbrecht's understanding of the difficulty about the glass,' and in 1737 his colleague Georg Wolfgang Krafft suggested a way of estimating the expansion of this by noting the "jump in the reading of the thermometer when the temperature of its environment is suddenly changed.' However, both Krafft and Weitbrecht were dealing with mercury thermometers, and it is fair to remark that in Réaumur's instruments this source of error would have been negligible in comparison with the others, which were certainly not.
The earliest serious critic of Réaumur's thermometer was the English physician George Martine, whose "Essay on the construction and graduation of thermometers" was in advance of its time. Martine came out strongly for the necessity of two fixed points, one being the boiling point of water--with due attention to the barometric pressure--the other the temperature of ice beginning to melt. He had no doubts about the constancy of this latter point, in spite of Taglini and Musschenbroek, for he had exchanged thermometers with various people over a range of nine degrees of latitude. He criticized Réaumur's methods of getting his freezing point because of his method of freezing, but also and mainly because of the very large bulbs of his thermometers. But Réaumur's "boiling-point," Martine said, is "still more vague and uncertain," because the spirit of wine cannot take the heat of boiling water. This brings him to a plea for the adoption of mercury as a thermometric fluid.
Soon most of the instrumentmakers, even on the Continent, turned to making mercury thermometers. In England they followed Fahrenheit, with the difference that they used the melting point of ice (32°F.) and the boiling point of water (212°F.) as fixed points. Where the 80-degree scale of Réaumur was in vogue, the instrumentmakers made mercury thermometers in which the ice point was 0°R. and the boiling point 80°R., as for example, Jacob Bianchy of Vienna, who in 1762 cheerfully defines Réaumur's scale in this way, with the recommendation that the boiling point be determined when the barometer stands at 28 [Paris?] inches. But since about 1740 it had been recognized that the dilatation of mercury and spirit were not proportional.® Micheli du Crest had in fact made comparisons, though assuming a priori that the expansion of spirit was linear.?
In the year of Bianchy's book a remarkable manuscript had been submitted for, and had received, the approval of the Académie des Sciences at Paris. This was the work of Jean André Deluc, a citizen of Geneva, who was such a perfectionist that even though the manuscript had been extravagantly complimented by the Académie in 1762,8 he spent the next ten years adding to it and polishing it. In 1772 it appeared in two quarto volumes of which the title summarizes the contents.* Perhaps this book had more influence than it deserved, but at least it drew attention to the necessity of reforming the thermometer, and especially the Réaumur thermometer. Deluc's experiments with the barometer are important, to and his meteorological ideas11 spectacular, if also highly speculative. Unfortunately, the Modifications is incredibly discursive, causing the great Lambert to complain that four times as much might have been said in a work a quarter as long.12 I have space only to mention Deluc's main contributions to the reform of the thermometer.
Starting with the lower fixed point, he was very severe on Réaumur's methods. In this he was flogging a dead horse, for by 1760 the use of the melting point of ice had become almost universal. Writing of the boiling point, he says that "The true principles [of Réaumur's thermometer are almost entirely effaced. I have tried to discover them." Obviously the 80° point on the new thermometers called "Réaumur" was not the same as on Réaumur's own thermometers. The relation between them was an important question which Deluc approached by stages. Believing firmly that mercury was superior to spirit as a thermometric fluid, he made elaborate comparisons of their marche. Then he started from the well-established fact that Réaumur's original thermometer read 10¼° in the cellars of the Paris Observatory. He could not make experiments there himself, and asked Claude Varenne, a correspondent of the Academy, for help. Varenne brought Réaumur's nephew Mathurin Jacques Brisson into the affair. They made numerous experiments, with the result that the cellars are at a constant temperature of 9.6° on the 80° scale of the mercury thermometer established by Deluc, who, using his comparisons between mercury and spirit thermometers, calculated that Réaumur's 80° should correspond to only 59.3° on a spirit thermometer graduated on Deluc's principles. This was unreasonable, and Deluc went on with his experiments, making a mixture of spirit and water as noted by Réaumur, and finding its boiling point to be 64.3°. This gave him two points, on the basis of which he deduced that Réaumur's zero was 0.90° lower than his. He prints a table (5. 1)15 in which historical climatologists may be surprised to note how close Réaumur's original scale is to the revised mercury one, at ordinary meteorological temperatures.
The Observatory cellars, incidentally, continued to fascinate the Academicians right up to the Revolution. Jacques Dominic, the last of the four Cassinis to direct the Observatory, himself climbed the 210 steps almost every day for nearly two years in order to determine once and for all whether or not the temperature in the cellars is constant.1 The thermometers that he used are of interest to us as showing that at that time much more attention was being paid to detail. They were made for Lavoisier! by Mossy and engine-divided by Richer. They were of mercury, one with a tube 20 inches long, carefully calibrated in ice and in boiling water. The other was a very special instrument with a bulb 2½ inches in diameter and a capillary tube 22 inches long. It was calibrated, by comparison with the other, with great care in a water bath, the process taking six weeks. This thermometer, on which 1°R., corresponded to about 4¼ inches, could easily be read to ½0°R. It was set up in a large gallery in the cellars, with the bulb in a good-sized pot of fine sand, in order to nullify the effect of the visits by the observer.
During the twenty-three months August 5, 1783, to June 29, 1785, the extremes were 9.06° and 9.28°R. In the complete table it is hard to separate the change in zero of the thermometer from the annual variation, but it is quite clear that there would not be an annual range of temperature of more than 0.1°R.
There was so much discussion of the "Réaumur" scale between 1770 and 1790 that I shall have to be selective. After the severe winter of 1775-76 Lavoisier, whose interest in thermometry seems to have been boundless, collected no fewer than thirty-eight thermometers that had been observed in various places and compared them in the Observatory cellars, in melting ice, and in a freezing mixture. The most important of these instruments was one belonging to Brisson, which had been made by Réaumur in 1730 and observed during the cold winter of 1739-40. One of the results of the study, intended mainly to compare the two winters, was that "Réaumur" thermometers made since 1740 "had not conformed to his principles at all."18 At various times Lavoisier ordered numerous thermometers from the best makers, such as Cappy, Mossy, and later Megnié, and Fortin; and there is no doubt that his influence on French thermometermaking, and instrumentmaking in general, was tremendous.
A different approach to reforming the Réaumur thermometer was made by a priest of Tours named Joseph Louis Bossier, or Beaussier, who in 1775 submitted a paper on the subject to the Paris Academy, which was reported on by Leroy and Macquer on July 3, 1776.3 They did not find anything new in it and Bossier was obliged to print it himself, with additions, three years later.21 They also said that anyhow, spirit thermometers must be given up, as they do not give the same precision as those of mercury. Reading Bossier's extensive and fascinating footnotes with this statement in mind, it becomes clear that the Academy was still very solicitous for the reputation of the great Réaumur and felt that the safest course was to sweep the spirit thermometer under the rug.
For Bossier's suggestion was to rehabilitate the spirit thermometer by leaving air over the spirit column so that it could be calibrated in boiling water and in melting ice. He would then divide the interval into 100 equal parts. He wished to preserve Réaumur's principles while improving on his practice, and declared that comparable spirit-of-wine thermometers should be filled with spirit that dilates 100 parts in 1000, "a simple round number, convenient for the division of thermometers, "22 between the melting point of ice and the boiling point of water. Another argument in favor of the 100° scale is that it agrees with Christin's division of the mercury thermometer, which he believes ought to be adopted. But his adherence to Réaumur's principles is shown by the fact that he is still impressed by Christin's estimate that 6,600 volumes of mercury at the ice point become 6,700 at the boiling point!
Ten years later Jean Gaussen of Montpellier,24 after criticizing both Deluc and Beaussier (Bossier\ for going on a wild goose chase, says that the choice ought to lie between the thermometers of Deluc and of Christin. But in France they have been accustomed to the Réaumur thermometer for so long that it can scarcely be abandoned. So he proposes a scheme, starting from Deluc's thermometer, for producing a fair imitation of Réaumur's original scale.
Gaussen may have proposed, but the Revolution disposed. Arthur Birembauts quotes a long passage by René Just Haüy, from pages 202-4 of Instruction sur les mesures déduites de la grandeur de la terre, uniformes pour toute la Republique, et sur les calculs relatifs à leur division décimale; par la Commission temporaire des Poids et Mesures républicaines, en exécution des décrets de la Convention nationale. The decision to adopt the centigrade scale was taken on 12 germinal an II (April 1, 1794), and gave it an official connection-though of course no logical one-with the metric system. But all change meets resistance, and in 1816 we find the wealthy amateur physicist Honoré Flaugergues writing: *. . . this division into 100 degrees presents absolutely no advantage, and it seems to me to be much more worth while to preserve the old division into 80 degrees, adopted by the very great majority."
I shall now digress briefly to deal with the question of the freezing point of mercury, a matter of much interest to physicists through most of the eighteenth century. It was at first almost universally believed that mercury would remain liquid at temperatures that would freeze spirit of wine; the observation by Maupertuis, already quoted, 2 is an example of an observation that seemed to prove it, though there can be no doubt that the spirit in his thermometer was diluted with water. It must be noted that the freezing mixtures used at the time, starting with snow or ice at its melting point, would not freeze mercury. If the snow were very cold, lower temperatures could be obtained. Joseph Adam Braun of the Academy at St.
Petersburg knew this, and when on December 14, 1759, the temperature fell to 205° Delisle (= -36.6°C.), he thought he would try to find out what a freezing mixture would do. He mixed snow and nitric acid, and when he put the bulb of a mercury thermometer into this mixture the mercury went down to 260, 380, and finally 470° Delisle (= -220°C.!). Taken into a warm room the mercury, which had evidently frozen, rose again to agree with another thermometer that had not been in the mixture. On December 25 he froze the mercury in a thermometer and broke the bulb, becoming the first man to see mercury in the solid state. The column had descended to 530°D. (-253°C.). Later he got one to descend to 700°D. (-366°C.). What had happened, of course, was that the mercury in the bulb had contracted on freezing, and the mercury in the tube, still liquid, had gone down to fill the space that was left empty. It is curious that he did not suspect this, for he recounts how he took three thermometers made with the most highly rectified spirit of wine and put them in the freezing mixture with the mercury thermometer. Even though the mercury froze, they did not, but went down to 300°D. This is - 100°C., but we need not be surprised, for such an extrapolation would be meaningless in the case of these spirit thermometers. He could not believe that the mercury thermometer could attain a much lower temperature than the spirit thermometers in the same freezing mixture. Here are his conclusions:
It seems therefore, that the descent of the mercury, especially when it begins with a rush, no longer preserves any proportion in its contraction. From the observations it is at least certain that at the lowest degrees of cold, the harmony and concordance of mercury thermometers with those filled with spirit of wine entirely ceases. What is more, it can be seen from the observations reported here that the cold that is enough to freeze mercury does not always suffice to freeze the most highly rectified spirit of wine, even though of all fluids mercury may seem to need the greatest cold to congeal it. From this it may also be concluded that thermometers filled with very highly rectified spirit of wine may serve to indicate the various greater or lesser degrees of cold in a freezing-mixture, although they may not fix the measure of the cold. ...29