Your objection is correct, and none of the responses you've received so far seem to have understood your point. I went ahead and read the actual published pape

Your objection is correct, and none of the responses you've received so far seem to have understood your point. I went ahead and read the actual published paper (https://acsjournals.onlinelibrary.wiley.com/doi/epdf/10.1002/cncr.33615 https://acsjournals.onlinelibrary.wiley.com/doi/epdf/10.1002...) and it is just as statistically naive as the NatGeo article made it sound. Let's cartoonishly oversimplify cancer so that we can see how life expectancy would affect observed statistics. Let's say cancer occurrence is 0% for people up to age 40, and then starting on the 40th birthday, cancer occurrence is 10% per year. Everyone who gets cancer dies immediately. Then, imagine that life expectancy in 1500 was 25 years old, standard deviation 10 years (yeah lognormal blah blah just go with it). So, most people don't live to 40, but the ones who do sometimes get cancer. But then, after 1500, life expectancy mean increases by 0.1 years per year. By the year 2000, life expectancy is 75 years old! Much more cancer is going to happen! Many more of the deaths will be from cancer! Cancer itself didn't change. So, these researchers see a lower cancer rate in the bones they're looking at, but they need to do some MATH on the ages these bones and of modern humans to know how to compare cancer rates then to cancer rates now.