Fahrenheit Is Better for Weather
I recently got into an "argument" with a friend from England about Celsius and Fahrenheit. I was trying to convince him that Fahrenheit is better when talking about weather. He told me that Fahrenheit was a silly American system that wasn't "scientific." But Fahrenheit was actually based on specific things. Accounts of how Fahrenheit first defined his scale differ, but his original paper points to 0°F being set at the freezing point of a brine made from water, ice, and ammonium chloride.
That being said, Celsius being based on water's freezing and boiling points, at 0 and 100 degrees respectively, is a better system for scientific purposes (and many other use cases, like cooking). Even so, Fahrenheit simply works better for explaining weather.
To try to convince my friend I did the following:
I asked him to guess the temperature outside in Celsius.
He guessed 22°C.
Then I asked him to guess in Fahrenheit.
He said he couldn't, because he didn't understand Fahrenheit.
So I said, "My ideal temperature system is as follows":
Treat 0°F as an extremely cold day and 100°F as an extremely hot day, roughly the bottom and top 10% of days.
He thought about it and guessed 75°F.
I then told him that that's how Fahrenheit should be treated.
The actual temperature was 77°F, or 25°C.
His Celsius guess was off by 3°C, which is 5.4°F. His guess for my "new intuitive system" was off by just 2°F from the actual Fahrenheit temperature.
Essentially, a Brit and avid Celsius user was able to tell me the temperature in Fahrenheit more accurately than in Celsius.
To be fair, this isn't strong evidence. The sample size was just one and he could have just gotten lucky.
Still, it points at what I think is the strongest argument for Fahrenheit:
Fahrenheit is a good scale of how hot the weather feels to a person, on an intuitive 0-to-100 scale.
The 0-to-100 argument
For a large range of inhabited climates, Fahrenheit works approximately like this:
- 0°F: extremely cold
- 25°F: very cold
- 50°F: cool
- 75°F: warm
- 100°F: extremely hot
This isn't exact. There's no objective temperature at which "very cold" becomes "extremely cold," and people adapt to their local climates. Someone from Phoenix and someone from Helsinki won't interpret 90°F in the same way.
But it's approximately true in a useful way.
The normal range of weather fits into something close to a 0-to-100 scale. The endpoints correspond to conditions most people would consider unusually severe, and the middle corresponds to mild weather.
This gives the scale an intuitive feel that Celsius lacks.
If someone says it's 80°F, you can think of it as roughly 80% of the way from very cold to very hot. Again, not literally. Human temperature perception isn't linear, and neither are the health risks. But as a quick intuition, it works.
Celsius doesn't have the same property. 0°C is cold, but not extraordinarily cold. It's a normal winter day in many places. 100°C isn't extraordinarily hot weather. It's the boiling point of water and far outside the range of survivable weather.
For normal weather, Celsius mostly uses a narrow band around zero. A rough range might be -10°C to 40°C. (The equivalent Fahrenheit range is 14°F to 104°F.)
Neither scale contains more information in principle. You can convert between them exactly. It's simply that Fahrenheit uses a wider range of whole numbers for the temperatures people commonly encounter.
What should a weather scale optimize for?
There are several things a temperature scale could be designed to make convenient. It could make scientific calculations convenient. It could align with important physical transitions. It could be easy to convert to an absolute temperature scale. It could even be good for cooking. Or it could make ordinary weather easy to describe.
Celsius is good at the first four. It's closely connected to Kelvin, fits into the metric system, and places the freezing and boiling points of water at 0 and 100 under standard conditions.
These are real advantages. They're why Celsius is the obvious choice in science and why almost every country uses it. But they do not establish that Celsius is the best scale for every use.
The boiling point of water isn't relevant to almost any weather report. The freezing point matters more, especially for ice and snow, but it's still only one point on the range of temperatures humans experience.
A scale can be scientifically elegant and also be less convenient for a particular everyday task.
Meters are probably better than feet overall, but feet are a fairly convenient unit for human height. Kilograms are better than pounds overall, but pounds provide finer whole-number distinctions for body weight.
Similarly, Celsius can be the better general system while Fahrenheit is the better system to represent temperature in weather.
Resolution
Fahrenheit also has the benefit of higher resolution. One degree Celsius is equal to 1.8 degrees Fahrenheit. This means Fahrenheit gives you smaller steps when using whole numbers.
The difference between 70°F and 80°F is 5.6°C. In ordinary speech, that's the difference between a mild day and a fairly warm one. Fahrenheit splits this interval into ten whole-number steps. Celsius splits it into around six.
This doesn't matter much for scientific precision, since Celsius can use decimals. A forecast can say 23.5°C just as easily as it can say 74°F.
But people usually don't talk about weather using decimals. They round to whole numbers. Given that constraint, Fahrenheit provides slightly more resolution.
I do not think this is the main advantage.
Still, it's a reason that Fahrenheit is better for weather. If I'm setting the thermostat I set it to 67°F. If I instead set it to the closest temperature in Celsius it would be closer to 66°F. I can feel the difference. It's small but it matters.
Familiarity is doing most of the work
The obvious objection is that Celsius users find Celsius intuitive.
Someone who has used Celsius their whole life knows that:
- 0°C is freezing
- 10°C is cool
- 20°C is mild
- 30°C is hot
- 40°C is extremely hot
They don't need to mentally convert anything. The numbers have acquired meaning through experience.
This is true, and it limits how much the design of the scale matters.
A person can learn almost any consistent system. If the US switched to Celsius, Americans would be mildly confused for a few years and then mostly fine.
So the question is not:
Can people understand Celsius weather?
Obviously they can.
The question is:
Which scale gives a newcomer a better prior before they've memorized what all the numbers feel like?
My friend's guess is fairly weak evidence that Fahrenheit may do better here.
After being given the endpoints, he could interpolate. He didn't know what 75°F normally felt like, but he understood where 75 sat on the scale.
Familiarity is still a reasonable objection to the claim that Fahrenheit is a more intuitive system for weather.
That being said, any system will need some form of reference, and for weather Fahrenheit's is better: 0 to 100 runs from a cold day to a hot day. Celsius doesn't have that. 0 to 100 running from water's freezing point to water's boiling point means fairly little when it comes to weather. I could say -10 to 40 is a cold day to a hot day, but that's still worse, as negative numbers are less intuitive and humans are used to 0-to-100 scales.
The strongest case for Celsius
The strongest case for Celsius is not that 0 and 100 are inherently more intuitive.
It is coordination.
Almost the entire world uses Celsius. Scientists use Celsius or Kelvin. Weather data, appliances, education, and international communication all become simpler when people share one system.
A slightly better local interface can still be worse overall if it creates conversion costs everywhere else.
This is the same reason we shouldn't invent a new unit every time we find one that's marginally better for a specific task.
The US using Fahrenheit creates friction. Americans travel and don't understand forecasts. Visitors come to the US and don't understand forecasts. Every weather service has to support both.
Those costs might outweigh Fahrenheit's advantages.
But that's an argument for standardization, not an argument that Celsius is intrinsically better at representing weather.
There are two separate claims:
- The world is better off coordinating on one temperature scale.
- Fahrenheit is a more intuitive scale for everyday weather.
I think both can be true.
A quick summary
Celsius is a better scientific standard. It's probably also better for global coordination. But Fahrenheit is better designed for the narrow task of telling a human what the weather feels like.
Its normal range is close to 0 through 100. Its whole-number steps are smaller. Its endpoints have an intuitive interpretation even for someone who has barely used it.
My argument with my friend does not prove any of this.
But he guessed 22°C, then guessed 75°F after a one-sentence explanation.
It was 77°F.
I count that as a small win.
I'd love for the world to switch to Fahrenheit for weather, but that's just me, a silly American who needs a 0-to-100 scale for it to make sense and can't comprehend true scientific measurement systems.
Wikipedia: Fahrenheit · Celsius