In 1742 the Swedish astronomer Anders Celsius published a paper in the "Kungliga Swenska Wetenskaps Academiens Handlingar", the Annals of the Royal Swedish Academy of Science, entitled "Observations on two persistent degrees on a thermometer". This paper is the origin of the Celsius temperature scale.
After providing a background on the various ways of expressing temperature used at that time, Celsius presents an account of his experiments with two "fixed points" for a temperature scale:
The idea to use the freezing point of water as a temperature calibration point was not new and had already been suggested by Reamur and Newton as well as by the less well-known contemporary scientists Martins, Weitbrecht and Poleni. Celsius, however, used the melting point rather than the freezing point, i.e. thawing snow or ice rather than freezing water.
"This experiment", Celsius writes, referring to placing a thermometer in thawing snow, "I have repeated many times during two years in all winter months and all kind of weather, and during different barometric changes, and always found precisely the same point on the thermometer... When it was strong winter I have taken the cold snow into my room and put if in front of the fire until it began to thaw... "
Furthermore, using one of Reamur's thermometers, Celsius finds, within the experimental error, the same freezing point in Uppsala (latitude 60 oN) as in Torneň (latitude 66 oN) as Reamur found in Paris.
The second calibration point was more complicated. "As far as the second fixed point is concerned," Celsius writes, "it is well known that once water has began to boil, it will not take up any considerable degree of heat even if one continues the boiling for long; thus the mercury in the thermometer will end at the same point, despite the objections of Mr. Taglini."
However, the intensity of the boiling might affect the calibration point to a certain degree and Celsius proposes a standardised method for the determination. He also notes that when the thermometer is taken out of the boiling water the mercury level at first rises somewhat before it retracts. The explanation, Celsius suggests, is that the glass tube contracts before the mercury starts to cool off.
The second factor that affects the boiling point of water is the air pressure. This was already observed by Fahrenheit, and Celsius reports on series of experiments to confirm this observation. His conclusion is that "the height of the thermometer in boiling water always is proportional to the height of the barometer, thus 8 "points" on the thermometer I use correspond to a change of one inch in the barometer reading; a thermometer which is somewhat more sensitive or have large degrees can be used as a barometer when put into boiling water, and would be easier to bring along when travelling at sea or land, especially on high mountains."
Finally Celsius proposes a standardised procedure for the calibration of a thermometer:
Celsius thus put the boiling point at 0 degrees and the freezing point at 100. This was, however, soon to be reversed. It is often claimed that Carl von LinnÚ (Carolus Linnaeus) instigated the reversal, but it appears more likely that the responsible person was Daniel Ekstr÷m, who manufactured most of the scientific instruments, including thermnometers, used by both Celsius and LinnÚ.