But yeah, given the most natural definition, his last number is about [w+3]1000, which is a bit bigger than mine (M3=[w+3]10) but much much smaller than 10ls's (which is something like [w^2]12=N12=[12,0]10).

And since we're already beyond the limit of [a,b]c , I'll define arbitrary-length arrays:

1. [anything]1=10

2. [0]n=10n

3. [0,...,0,a,...,n]m = [a,...,n]m

4. [a,b,c,...,n+1](m+1) = [a,b,c,...,n][a,b,c,...,n+1]m

5. [a,b,n+1,(k zeros)]m = [a,b,n,m,(k-1) zeros]m

6. Nn = [1,0,0]n ; Pn = [1,(n zeros)]10

Also, for finer tuning of my numbers, I define a binary version of P for selected rational numbers:

7. For integers n>1, 10

^{n}<=m<10

^{n+1}: (m/10

^{n})Pn = [the digits of m seperated by commas]10

And submit this entry:

1.23P2 = (123/100)P2 = [1,2,3]10 = [w^2+w*2+3]10