RE all-base composites

[Tautócrona]

----- Original Message ----- 
From: <franktaw at netscape.net>

>By the way, this sequence should have the "base" keyword.

Actually, I think it shouldn't! The seq doesn't change its members if you do the 
operations in any
other base, and the condition for adding the base keyword is "base: dependent on base used
for sequence"

I also retract my other messages, my "intuition" was a wrong one. There are false
positives (eg, 66 in base 17 is 108, that is prime in base 15) and false negatives (eg 71
in base 17 doesn't verify my condition, but it's 120 in decimal that belongs to the seq. I
think this could be solved saying there's _some_ base where the number satisfies the
condition, but it isn't easy any more).

I think the polys approach is the good one. Does anyone remember some useful theorems,
like Einsenstein Criterion and so?