Is human body temperature really falling?

Mostly true. A 2020 study in the journal eLife, led by Julie Parsonnet at Stanford, compared three sets of temperature readings taken in the United States between the 1860s and 2017 and found that normal body temperature has fallen by a small, steady amount with each decade of birth. The 37 °C figure everyone learns comes from an 1868 book. [1] [2]

The library fact says human body temperature has been falling for 150 years. It sits in the human body library, and it is one of the facts readers ask me about most, usually with the word "really" in the message. Is body temperature falling, then? For one country it is well measured, and the cause is still argued about.

Where 37 °C came from

The standard is older than most people assume. The Wikipedia article on body temperature says "The 37 °C value was set by German physician Carl Reinhold August Wunderlich in his 1868 book", which drew on millions of readings from about 25,000 patients in Leipzig [2]. His thermometers were long, slow and read in the armpit, and modern readings tend to come out lower [2].

What the 2020 paper measured

Parsonnet's group used three sources: records of Union Army veterans of the American Civil War, a national health survey from the 1970s, and Stanford's own clinical data from 2007 to 2017 [1]. Across those sets, the paper reports: "With each birth decade, temperature decreased by −0.02°C." [1] Men born in the early 1800s had temperatures about 0.59 °C higher than men today, and women's readings fell by about 0.32 °C since the 1890s [1].

Stanford's own news story put the same result in Fahrenheit: "These calculations correspond to a decrease in body temperature of 0.05 F every decade." [3] That is the number most news reports carried, and it matches the library's "falling for 150 years".

Why mostly, not simply true

The paper itself raises the first objection: thermometers changed. The authors checked for this by looking for the same trend inside each of the three datasets, where the instruments did not change, and found it there too [1]. So the fall is not only an artefact of better glass.

The second objection is scope. Every reading in the paper is from the United States, and the authors say the likely causes are lower rates of chronic infection and inflammation, warmer homes and less physical work [1]. Those are features of one kind of life, and the paper does not claim the whole species has cooled. A reader who wants to say "humans are cooling" is saying more than the source does.

It sounds fake because

It sounds fake because 37 °C has been on every thermometer and in every school book for a century and a half, and a number that fixed feels like a law of nature rather than an average from Leipzig in the 1860s [2].

It holds up because

It holds up because the trend appears inside each dataset on its own, so it is not a change of instrument, and because the effect is small enough to be believable. A few hundredths of a degree per decade is the kind of drift that improving health could produce. The body has other quiet oddities in the library: about one person in seven is missing a forearm muscle, and roughly 8% of human DNA came from ancient viruses. None of them is as widely misquoted as this one.

Questions

How much has human body temperature dropped?

In the 2020 eLife study, average body temperature in the United States fell by about 0.02 °C with each decade of birth from the 1860s to 2017 [1]. Stanford's summary gives the same figure as 0.05 °F per decade [3]. Men born in the early 1800s ran about 0.59 °C warmer than men today [1].

Is 37 °C still the normal body temperature?

As an average it is now slightly high. The 37 °C figure was set by Carl Wunderlich in 1868 from armpit readings in Leipzig [2]. Modern measurements in the United States average lower, and the eLife paper finds the decline continuing into the 2010s [1]. Normal is a range, and a single reading a few tenths below 37 °C is unremarkable.

Why is body temperature falling?

The authors of the 2020 paper suggest lower rates of chronic infection and inflammation, heated homes and less physical labour, all of which reduce the metabolic work the body does [1]. They checked that the trend appears within each dataset separately, which argues against the change being caused by better thermometers [1].