The next time you take your temperature, don’t worry if it’s not exactly what the doctor ordered—98.6 degrees Fahrenheit is just a number. That’s the message of a host of new research that shows human temperature is malleable, unique, and anything but average.

One new study from Stanford University researchers finds that we seem to be colder, on average, than our forebears. The difference they detected isn’t huge—between 1.06 degrees Fahrenheit lower, on average, for men born today compared to those born in the early 1800s, and 0.58 degrees Fahrenheit for women born today compared to those born around 1890—but it adds to a growing body of evidence that body temperature is a lot more flexible than previously thought.

German doctor named Carl Reinhold August Wunderlich set the standard on human body temperature in 1851, deeming 98.6 degrees Fahrenheit the norm. But scientists have known the actual average is lower for at least the last two decades, says study author Julie Parsonnet, a professor of medicine at Stanford University. What wasn’t known, until now, is whether or not Wunderlich’s instruments were inaccurate, giving him a higher reading, or whether average human temperature actually decreased over the last century and a half.

Parsonnet and her colleagues used three temperature data sets: one that looked at United States Civil War veterans over 80 years, one that captured a representative sample of the country’s civilian population between 1971 and 1975, and one that looked at a cohort of adults between 2007 and 2017.

These data sets differ from one another in many ways—the number of study subjects, their demographics, the duration of the information collection—but one thing they had in common was that they measured body temperature over a period of time.

The team used that information in two ways. First, they built a model using all the data sets that clearly showed temperature decrease over time; second, they looked within each data set to check if the decrease was reflected in the information there. The Civil War database was particularly important for this task, because it collected information for so much longer than the others. Without that second step, researchers would have no way to be sure whether changing thermometer technology could be blamed for the chilling trend. But even within each single data set, the signal is clear, says Personnet.

Especially within the Civil War cohort, she says, “Year of birth was a very strong predictor of temperature.” In other words, someone born in 1845 had a lower average temperature than someone born in 1820. That held true in the later data sets as well, Personnet says, which surprised her. “I didn’t think [average temperature] would have declined from the 1970s to today,” she says.

She and her colleagues think the changing temperatures are explained by changing human physiology. After all, she says, the lives of Americans in 2020 are very different than those that Union Army veterans were living more than a century ago. Our microbiomes are probably different, thanks to what we eat and what we’re exposed to, and the amount of exercise people get in everyday life is different, too—and those are just two of many factors that may affect metabolism, which is directly connected to body temperature.

Mount Sinai professor of medicine Waleed Javaid, who was not involved with the study, said the decline is interesting, but he’s not yet convinced that the data is totally accurate in reflecting the magnitude of the temperature change. It’s really hard to be sure that the researchers’ model corrects for all the differences between the three data sets, he says. Still, given what’s available when it comes to historical body temperature tracking, “I think this is as good as it gets.”

While the newest paper isn’t perfect, human temperature does appear to be getting lower over time. And it’s definitely not as straightforward as 19th-century physicians assumed: Javaid published research last year evaluating 36 previous studies on measured body temperature and found lots of variability—in other words, 98.6 degrees Fahrenheit is only “normal” for some people, and only under some conditions.

“We’re at the beginning of understanding this is happening,” says Personnet. “Whether it means something or not, I think, is something for the future to figure out.”

One crucial step will be to conduct a prospective study, which is where scientists collect a group of subjects with the goal of following them for years or decades to track changes. Such a study would be specifically designed to look at temperature and compensate for all the factors that could potentially affect it—the time of day, what subjects ate, what activities they participated in, and so on—so its results would provide a more complete picture of how body heat actually fluctuates at the population level. That picture, in turn, could help researchers understand more about human health and needs, from the nutritional to the microbial.