What's in a record

Wandering around the Web, I found an interesting observation in a discussion of climate change and the reluctance to accept the scientific consensus.

A commenter wrote:

The record high for India was in 2016. Check out these record highs from around the world. Far more highs a few to several years ago than within the past couple of years. What does that tell a thinking person?

And they gave a link to a Wikipedia page with world weather records: https://en.wikipedia.org/wiki/List_of_weather_records

Which got me wondering, What does that tell a thinking person?

I scrolled through the list of record highs and noticed that, indeed, there were a lot of dates from before the 21st century and thought that was interesting.

Then I started wondering what we might expect the data to look like if there were no warming.

And my thought was that they should be evenly distributed through time.

So I threw them in Excel and took a look:

And I thought, "That doesn't look very random!"

It's true that there are countries whose recorded highest temperature is decades ago - there's even one before 1890 (1887, to be precise).

But gosh, the decades of the 21st century have been rather high. Those two-and-a-half decades account for 71% of the observations, despite being only 18% of the time covered.

The next question is whether something else could account for the skew toward the present, and one obvious candidate is the possibility that weather records don't go back equally far in all countries. If country A has records starting in 1900, it could have a recorded high in 1905, or in 2020. Country B only started recording in 1950. They could have a recorded high in 1955, or 2020, but they might also have had a really hot day in 1905, and just have no record of it.

As a control for that, we can look at the list of record cold days at that same Wikipedia page. If this phenomenon of missing early data is a complete explanation, it should affect record cold temperatures as well.

And here's that histogram:

That doesn't look the same at all.

The two-and-a-half decades of our century are only 20% of observations.

46% of the observations happen before 1970 (with the high temperatures, it was only 12% that happened that early).

But these could also be affected by the same bias as the highs, with some earlier low temperatures not recorded because of less recording being done.

It's also curious that there are only 112 observations here, as opposed to 156 of the record highs.

I'm not sure why that is, but the most straightforward way to account for it is to limit the data to only include countries that have dates in both the high list and the low list.

That gives a sample of 106 countries, and the pattern very much holds:

The 21st century has seen 65% of the highs, but only 21% of the lows.

The current decade (counting from 2020) has seen almost as many highs as the previous decade, despite the fact that we're only 55% of the way through this decade (and haven't yet gotten to the warmest months in the northern hemisphere, where most people live).

In contrast, there have been 4 record lows in this incomplete decade, compared to 11 from 2010 to 2019, and we've already seen most countries' coldest months of 2025, so we're roughly 60% of the way through the decade.

There's one last thing it occurred to me to look at with this list of "comparables" (countries where we have both a high and a low), and that is to look at the time between the record high and the record low.

I took the year of the record high and subtracted from it the year of the record. In that setup, a positive number means that the record high is more recent than the record low, and a negative number means the opposite.

If there were no pattern, we should expect these differences to be roughly evenly distributed around 0. Some countries would have their record high later than their record low, leading to a positive difference, while a roughly equal number of countries would have it the other way around, with a negative difference.

And the sizes on each side should be roughly the same: a +25 should be equally likely as a -25.

But that's not what the data show:

Only 23% of countries have a recorded low temperature that is more recent than their recorded high temperature.

Back to the original question: What does the pattern of record highs and lows tell a thinking person?

It tells us that, while a list of countries' record temperatures is a very sparse climate signal, it nonetheless points strongly toward a warming trend in the late 20th and early 21st centuries.

Thanks for asking!

Addendum

As I was writing up the point about the current decade being only 55% done, it occurred to me that I could set up the bins so as to include the full last 10 years, and the results are dramatic.

The last 30 years account for 70% of the observations.

38% of countries have reported a high in just the last 10 years.

(On the cold side, the last 30 years have seen only 24% of observations, and the last 10 years have seen only 8.5%.)