I believe his numbers are determined from actual tests - lots of them.

Exactly. I once tried to find out what the figures were for normal
adult human hearing at various ages. By "normal" I meant someone who
had never been exposed to hearing-damaging noise, since even firing a
gun once without ear muffs causes a slight but measurable permanent
hearing loss. As I looked into it I had to exclude anyone who had used
power tools in an enclosed space, ridden a motorcycle without ear
protection, used an underground train service as noisy as London
Underground, attended a disco or pop concert, etc. You get the
idea. Every survey of people with supposedly normal hearing had been
contaminated by the presence of people with mild hearing damage. In
fact it began to look as though the age-related hearing-loss curves in
resarch papers might really be the hearing-loss curves of people
suffering from the average hearing damage of modern life.

It's very hard indeed to find a survey of the BG behaviour of people
who have been screened for the absence of diabetic tendencies,
undiagnosed diabetes, impaired glucose response, etc.. Doctors
generally accept that as we age most people will develop impaired
glucose tolerance anyway. But is that normal? Or is it the average
damage to glucose control one should expect in people who eat and live
the way we do?

Same problem with what is normal weight. The famous "obesity epidemic"
hit the headlines when the WHO finally managed to persuade the US
medical authorities to stop repeatedly shifting their definition of
obesity upwards as the US population got fatter, and suddenly millions
of Americans became "obese" overnight.

Suppose you wanted to get an estimate of what normal human
cardiovascular fitness was -- not average, but normal, how evolution
(or God) designed us to work at our best. Would you include or exclude
car owners from your sample population? Would you include or exclude
office desk workers?

I have a 35 year old friend who is obese, but also very fit. He
thought he might be diabetic, so we tested his BG after eating a large
candy bar, a frequent snack of his. It simply didn't budge a jot above
90. So he then drank a litre of orange juice and we tested at 15
minute intervals. At half an hour his BG hit a peak of 95.

Given the views of people like respected AACE endocrinologist and
diabetologist Dr Bernstein, I'm tempted to suppose that that kind of
rock solid control is indeed normal undamaged human BG control, and
that while only a small proportion of the human population are
actually diabetic, by the age of 60 most people have suffered some BG
control damage, i.e., are in the state of being "not diabetic yet",
and unless medical science succeeds in prolonging our lives
substantially, they'll probably die before the damage reaches a
diagnosably diabetic stage.

It would be interesting to find a population of native
hunter-gatherers somewhere and gather detailed medical statistics,
then to gather medical statistics of a population of primitive peasant
farmers who don't have cars or tractors and do it all by hand and
hoof, and then to compare those medical statistics with the population
of, say, London or New York.

Touche!