On Fri, 16 Oct 1998, Jud McCranie wrote:
> At 09:38 PM 10/15/98 -0400, Harry Kuhman wrote:
> >> Anyhow, GIMPS will get to exponents around 5,200,000
> >>before too long, and that will help confirm or deny it.
> >
> >Exhausting ...
> >...the 18,000,000 range.
> >
> You are quite correct. But if one was found between 5,150,000 and 5,299,999
> (that's close enough to 5,200,000 to be called "around 5,200,000") then that
> would lend some credence to the statement. The chance of one being found in
> that range is < 1%, so if one is found, it would be strong evidence in
> favor of it, but not conclusive.
>[. . .]
>
I see no legitimate reason for a number like 5,299,999 to
be called "around 5,200,000" in any case it should be called around
5,300,000.
So, if you have the *correct* result (with certainity) for a constant and
choose to express it in numbers with 2 significant figures -like 5.2 *
10^6- it could be assumed that the range of "fuzzing" might be +/- 0.05 *
10^6
