I came across your column on what zero means on the Fahrenheit scale. You blew it. You said, unlike 32 or 212 degrees, zero degrees was an arbitrarily assigned number. It isn't. It's the temperature at which seawater will freeze. An approximation, because the freezing point of saltwater varies based on salinity, but zero is a rule of thumb. —Richard Forte

Wikipedia takes your side, as does at least one college physics textbook. But close examination makes it reasonably clear the seawater explanation derives from a misreading of the evidence.

In my 1989 column I explained that Daniel Gabriel Fahrenheit based his system on an earlier scale devised by Danish astronomer Ole Roemer. Roemer had set zero arbitrarily—his main consideration was that it was colder than the temperature got in Denmark, because he didn't like using negative numbers in his weather logbook.

Roemer's scale had 7 1/2 as the freezing point of water and 22 1/2 as body temperature, in those days called "blood heat." To get rid of awkward fractions, Fahrenheit did multiplication, winding up with 32 as the freezing point and 96 as body temperature.

I said that when Fahrenheit was set to demonstrate his system to London's Royal Society in 1724, he worried it would look odd if zero on his scale was untethered to reality, and thus had to concoct a rationale. Here's what he wrote in the paper he presented:

"The division of the scale depends on three fixed points, which can be determined in the following manner. The first is found in the uncalibrated part or the beginning of the scale, and is determined by a mixture of ice, water and sal ammoniac [ammonium chloride], or even sea salt."

The "or even" part (the original Latin phrase is vel etiam [salis] maritimi) is a giveaway—the freezing point of seawater was an afterthought. Fahrenheit underscores this as he continues:

"If the thermometer is placed in mixture, its liquid descends as far as the degree that is marked with a zero. This experiment succeeds better in winter than in summer."

So the method used to determine zero on Fahrenheit's scaledoesn't always work. Who'd be foolish enough to invent a scale that wouldn't permit thermometers to be reliably calibrated? In contrast, the freezing point of fresh water, as in an ice/water mixture, is constant for practical purposes.

In a letter Fahrenheit wrote to a patron on April 17, 1729, he says when he visited Roemer in 1708, he found thermometers being calibrated by standing in water and ice. These thermometers were then heated to body heat, and "after [Roemer] had marked these two points on them all, half the distance found between them was added below the point of water and ice, and this whole distance was divided into 22 1/2 parts, beginning at the bottom with 0, arriving thus at 7 1/2 for the point of water mixed with ice, and 22 1/2 for the point of blood heat."

There you have it. Fahrenheit, following Roemer, determined the distance between the marks for the freezing point of water and body heat on his glass thermometers (64 degrees, in the scale he would develop), measured off half this distance (32 degrees) below the freezing point, and called that zero.

Recounting this, R.J. Soulen of the U.S. Naval Research Laboratory writes:

"The zero on this scale had no fundamental meaning, following the tradition of others who preceded him. Fahrenheit chose to define a zero below the coldest temperature likely to be encountered by everyday use of his thermometers."

But try calibrating your thermometer using the standard proposed by 19th-century Scottish astronomer Charles Piazzi Smith, who nominated a scale set to "the mean temperature of the King's Chamber at the center of the Great Pyramid of Giza."

Um, great idea, Chuck. On the other hand: road trip!