What's going on? Is the "-21" a typo, or is rock salt somehow different from what Fahrenheit used, or is there some transient effect not reflected in Fahrenheit's procedure, or is it something else I have not thought of?
Rik Littlefield, Senior Research Scientist, Pacific Northwest National Laboratory 4/22/01
Celsius temperature scale|
Fahrenheit temperature scale
freezing point depression
1. Fahrenheit was a victim of Hofstadter's Law, which states:
It always takes longer than you think it's going to take, even when you take into account Hofstadter's Law. 
Fahrenheit originally chose an ice/salt/water mixture to use as a calibration point because it provided a steady, fairly reproducible temperature. To see why, consider the phase map for salt water, which shows which phases are present at different salt compositions and temperatures.
|Phase map for salt water. Drawn from a diagram by R. E. Dickerson (Note 3)|
By mixing salt, ice, and water so that at least 23.3% of the mixture was salt, Fahrenheit should have reached the eutectic point- in theory. But in practice:
2. Fahrenheit didn't actually use NaCl solution. Fahrenheit calibrated 0°F as "the limit of the most intense cold obtained artificially in a mixture of water, of ice, and of sal-ammoniac or even of sea-salt" . Sal-ammoniac is ammonium chloride, and sea-salt is not just NaCl. The eutectic points for ammonium chloride and sea salt mixtures are not at exactly -21.12°C (the eutectic point for NaCl and water), and they are not exactly equal to each other, either.
3. Fahrenheit abandoned the original 0°F calibration point. He chose new calibration points twice. He first adjusted the scale to make 32 the ice melting point and 96 the temperature of "the blood of a healthy man" . It's easy to mark scale divisions that are multiples of two with simple tools, and the adjustment produced a scale with 64 divisions between the ice melting point and blood temperature.
He eventually replaced the upper calibration point at blood temperature with the boiling point of water. He discovered that the boiling point of water could be reliably used as a calibration point if atmospheric pressure was fixed, a fact that earlier temperature scale architects had overlooked. The boiling point at normal atmospheric pressure was set to 212°F, exactly 180 degrees between the freezing point of water and the boiling point of water, perhaps because 180 degrees is the number of degrees between opposite poles.
Author: Fred Senese firstname.lastname@example.org
Copyright © 1997-2010 by Fred Senese
Comments & questions to email@example.com
Last Revised 08/17/15.URL: http://antoine.frostburg.edu/chem/senese/101/solutions/faq/print-zero-fahrenheit.shtml