On Mon, 7 Sep 1998, T. Martin wrote:

> > I believe the mean for the Samburu (the meat-and-milk people) is
> > 150.  The Tutsi (another all-meat group) are lower than that.
>
> But higher than 125? And what is the standard deviation of these
> distributions? You see my point, I'm sure. I just wonder how confidently
> we can assume that Stefansson's results are so unusual that we have to
> reach your "either his diet or his genes were different from modern-HGs"
> conclusion.
>
> Anderson is another story, of course.

Stefansson's cholesterol varied quite a lot, but *averaged* 268
for the course of the experiment.  However, the higher values
were recorded mostly during the first six months or so.  Then
there is a gap during which no measurements were recorded, then
it was lower during the last 3 months (Andersen's cholesterol, in
contrast, escalated steadily throughout the year).  I'm willing
to allow that after an initial elevation, Stefansson's
cholesterol settled back to an average of about 225, for the last
three months.  So let's use that figure, instead of the 268.

If we accept Eaton's value of 125 as the *mean*, then the only
way Stefansson's number could fit into anything like the fat part
of a bell curve is if there were also a substantial number of
people at a similar distance on the other side of the mean.  That
would be a total cholesterol of 25.  Doesn't seem likely.  Of
course, it's possible that the distribution is flat and not
normal, in which case the mean doesn't really tell us much
anyway, and Eaton's observations would be meaningless.
That is, his whole point was to mark the contrast between typical
HG cholesterol and typical civilized readings.  If the civilized
readings were not off the fat part of the bell curve, they would
fail to make his point.

Todd Moody
[log in to unmask]