water, but his process was not applicable to the alcohol thermometers then in use, for the vapor of alcohol has a tension at the boiling point of water which would burst the reservoirs of the instruments. And Renaldini's method could not be adopted till after Musschenbroeck had introduced the use of mercury. In 1729, Delisle chose as graduating points the temperature of ice-water and the almost invariable temperature of the cellars of the Observatory at Paris.
About 1714 a skillful instrument-maker of Dantsic, Daniel Gabriel Fahrenheit, furnished chemists with alcohol thermometers which he replaced in 1720 with mercury thermometers, the indications given by which all agreed with one another. According to the chemist Woulfe, he boasted that he could make a thermometer that would agree with those he had already made, in any place, and without seeing any of the instruments that had already gone out of his hands; but he would not divulge the process by which he had been able to obtain such an agreement. This process, in establishing which he had been aided by the advice of the astronomer Roemer, was nothing else than the method devised by Dalencé; but Fahrenheit took for his zero the temperature of a mixture of ice and muriate of ammonia (chloride of ammonium)—which, he thought, was the greatest cold that could be obtained—and for his higher degree the temperature of the human body.
Finally, in 1742, the Swede, Andrew Celsius, proposed to restore the method of Renaldini, and divide into a hundred degrees the interval which the mercury in the thermometer would traverse between the temperature of melting ice and that of boiling water. He marked the lower temperature 100, and the higher 0. Linnæus, reversing this order, gave the mercury thermometer (centigrade) the form under which it is now known.—Translated for the Popular Science Monthly from the Revue des Deux Mondes.