Apologies for the length of the image, which I found at http://xkcd.com/1732/.

I'm mostly interested in the temperature line itself, both the historical data and the forward projections.

The reason I ask is that I'd like to (in a classroom context) hold this image up as a good example of the visualisation of data, that could be emulated in the area of science relevant to that classroom (zoology). However, I'd first like to check that the data itself is correct.

xkcd 1732 (click to enlarge)

  • 6
    $\begingroup$ Yes it's pretty accurate, at least for the purposes of the comic. It omits a lot of finer detail of decadal scale oscillations but it's not intended to be a super accurate depiction of past temperatures. It is intended to make a point. $\endgroup$
    – bon
    Sep 15, 2016 at 13:01
  • 3
    $\begingroup$ I love the illustration. It provides a great perspective on recent Earth's history and in human pre- and nonpre- history. $\endgroup$
    – arkaia
    Sep 15, 2016 at 13:06
  • 2
    $\begingroup$ The only thing I disagree withis that I don't think Wrangel Island (7,600 km²) should be called a “tiny Siberian island”. It's not huge but I wouldn't call it tiny. But that does not affect the timeline of the Earth's average temperature. $\endgroup$
    – gerrit
    Sep 15, 2016 at 13:17
  • 1
    $\begingroup$ I'm pretty sure it was a more than 4 degrees C below 1961-1990 "baseline" 20,000 years ago. Estimates vary, but 4.3 (or 7.7 F) colder is the smallest I've heard. 8 degrees C is more standard, maybe 7, but 4.3 is too low. That's the main one. There's one or two minor tweeks, like the ice may have reached about to NYC along the coast, but it was a bit further south in the center of the country. "As south as NYC" isn't quite correct. Not far off, but probably not correct. $\endgroup$
    – userLTK
    Sep 15, 2016 at 14:12
  • 12
    $\begingroup$ FYI the sources Randall Munroe used to do this graphic are written on its side: Shakun et al 2012, Marcott et al 2013, Annan & Hargreaves 2013, HadCRUT4 and IPCC. $\endgroup$
    – plannapus
    Sep 16, 2016 at 8:44

2 Answers 2


Yes, it's accurate, see e.g. this picture taken from the Wikipedia Geologic temperature record page:

enter image description here

You could dig up the source if you wanted from the wikipedia article (or the graph description page), but any graph will be generally the same. Since there was an ice age, you'd expect global temperature to be low ~20k years ago and so on.

  • $\begingroup$ Note for others: the graph here is in °F whereas XKCD uses °C, so some conversion to be aware of. Also, there's good further discussion on this topic at this other question. $\endgroup$
    – jwd
    Jan 17, 2022 at 7:28

Yes, it's fine, but you should be aware that the earlier period of the data is incapable of showing large variations within a century or so. They are all smoothed out, even if they happened. So you get the wrong impression if you assume that temps have been smooth with no centennial scale gyrations. The plain fact is our data is not good enough to show this level of granularity.

...no temperature variability is preserved in our reconstruction at cycles shorter than 300 years, 50% is preserved at 1,000-year time scales, and nearly all is preserved at 2,000-year periods and longer.


My personal Bayesian guess is that the recent temp rise is abnormal in historical perspective. But you are getting the wrong impression of the surety of that view from looking at the XKCD cartoon. Randall has a small caveat on this issue but does not go in depth to analyze it (and it is the key issue in extracting insight from the Marcott study).


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.