Margen
You do not have a temperature of 37, and the doctor who set that figure was not wrong. Both things are true at once, and the second one is the part nobody ever tells you.
The doctor was called Carl Reinhold August Wunderlich and he ran the university clinic in Leipzig. In 1868 he published a book on how body heat behaves in illness, and behind it lay an amount of work that is still impressive: some twenty-five thousand patients and more than a million measurements, taken under the arm with a mercury thermometer a foot long, about thirty centimetres, that had to be held in place for twenty minutes. That is where 37 came from as the temperature of the healthy, and 38 as the border of fever.
More than a century later, when somebody finally redid the sum with modern data, 37 did not show up. In 1992, Philip Mackowiak took seven hundred oral temperatures from 148 healthy adults in Baltimore: the mean came out at 36.8, and only 8 % of the readings were exactly 37.0. In 2017, the British Medical Journal did it with 35,488 patients in Boston who had seen a doctor with no infection at all: 36.6.
The comfortable reading of those two numbers is that Wunderlich got it wrong. In 2020, a Stanford team showed that he did not.
They pooled 677,423 measurements from three different eras: medical records of American Civil War veterans, a national health survey from the nineteen-seventies, and hospital consultations between 2007 and 2017. Lined up by year of birth, the temperatures fall without stopping: about three hundredths of a degree per decade of birth. A man born in 2000 runs around half a degree cooler than a man born in 1830. The explanation the authors propose is that our bodies are less lit up: fewer untreated chronic infections, fewer rotten teeth, less background tuberculosis, houses kept warm all winter.
So 37 was true. It was true for Wunderlich's patients, who lived with a small ember inside them most of the time, and it was true measured where he measured it: under the arm, which reads about three tenths lower than the mouth. His armpit 37 would be 37.3 in the mouth today, so the gap with Boston’s 36.6 is even wider than it looks. What has expired is not his measurement: it is us.
That is where the usable part comes from, and it starts with what has just settled arguments at home: the number depends on where you put the thermometer. You have seen that with Wunderlich, and the mouth in turn reads about four tenths lower than the rectum. Saying “I've got 37.2” without saying where you measured it does not mean much. And it depends on the hour. Mackowiak found that the same person, on the same day and in good health, moves half a degree, with the low point near six in the morning and the high point in the late afternoon.
The Boston study also left behind a finding that almost nobody repeats. African-American women showed the highest temperatures, five hundredths above white men; older people the lowest, two hundredths less for every decade of age. But age, sex, weight and illness, all together, explained only a tiny share of the differences: most of it stayed unexplained. And that unexplained part turned out to be the valuable one, because it predicted something. Being fifteen hundredths of a degree above the mean was associated with an 8.4 % higher chance of dying over the following twelve months than being on it.
Translated: your baseline temperature is a figure of your own, it carries information inside it, and you have it written down nowhere.
The nurse’s 38 is not your temperature: it is a deliberately high cut-off, taken in the mouth or the ear so that no infection is missed in anybody. That is not up for debate from your sofa.
What you can do is have a figure of your own, which is free and takes five days.
Take one thermometer, just one, and pick a site: armpit or mouth, whichever one you will always repeat. For five days while you are healthy, measure twice: once when you get up, before coffee, and once between five and seven in the evening. Write all ten figures in your phone with the time beside each. When you finish you will have two averages, morning and evening, and you will see your own swing; in most people those two figures sit three to six tenths apart. The device matters less than you think, as long as it is the same one: what you are going to compare is not your temperature against humanity's, it is today's against last week's.
That pair of numbers is your baseline, and it is good for two concrete things. First, you stop panicking over 37.3 at six in the evening if your normal evening is 37.1. Second, you see trouble coming early: 37.4 at seven in the morning, when your normal morning is 36.5, is almost a degree above yourself and deserves attention even if no protocol calls it a fever yet.
Write it down today in your phone's notes with two words: “morning” and “evening”. In five days you will have your figure.