A "doublely-recursive" sequence

[Leroy Quet]

>I guess our numbers must be correct then. I'd bet that the sequence
>is infinite but grows very slowly. If you find anymore numbers you should
>add them to the sequence I submitted yesterday to the EIS. 
>
>There is nothing special about 1. One could just as well find the indices
>where the original sequence was 2 or 4 or any number in the sequence.  
>
>I seem to have lost my original data up to 10^5. But I redid it up to 10^4
>and found that the most frequently occuring number in Leroy's original
>sequence up to 10^4 is 2996 which occurs 37 times. :-)
>
>Cheers,
>
>Edwin


They main significance of 1, aside from it simply being the first 
positive integer, is that a(m) = 1 only when a(m-1) is divisible by m (or 
when m=0, of course).

  
2996?  hmmm...

;)

Thanks,
Leroy Quet