I compared it to the fibonacci sequence, if you look at those, the ratio of T(n):S(n) doesn't go to 0, it actually goes to 1/Phi^2, where Phi is the golden ratio aka (sqrt (5)+1)/2. That sequence does grow exponentially.
Any reason you can't place your cities in a growing spiral like that, with every city in the outer 2 spirals producing a settler/founding a new city each iteration, and the other cities simply growing once per iteration?
You seem to be mixing up very different things. The fact that the ratio of consecutive Fibonnaci number converges is well-known (and yes, this implies that the sequence grows exponentially). What this has to do with our problem at hand, I don't know. I think too many of you were focused on the production of settlers (comparing to the rabbit problem), while the problem is essentially one of movement speed.
You can place your cities in a spiral if you want, that doesn't change the fact that, after n turns from the start, everything you own will be contained in a 2n+1x2n+1 square centered at your starting point, simply because of movement speed (again, just think of starting with an infinite number of settlers). In particular, the number of cities is bounded by a quadratic function.
I think an experiment might convince you:
Try to really achieve your Fibonnaci sequence for the number of cities.
That is, forget about city growth, just try to get an exponential number of cities.
I'll even allow you to build cities right next to each other, for simplicity.
And cities build settlers in 1 turn. In fact, I'll be even nicer, cities can create as many settlers as you want in a single turn (but the turn after the city is founded)!
BUT settlers can only move one square per turn (let's say they can move+found in one turn).
Now, try to match the Fibonnaci sequence with the number of your cities. That is, try to have F_n cities after n turns.
Even with all my generosity, you will still run into problems at some point.
Can you prove the ratio of exterior cities to total cities goes to 0?
Well, the gritty details depend on how exactly you set up the cities, but there are handwavy arguments that should be convincing.
Most "natural" shapes your empire could or would take have an area that is proportional to the square of the "boundary" and hence also proportional to the square of the area that is within a fixed distance of the boundary. This implies that the ratio between this latter area and the whole thing goes to 0. If the city densities are more or less bounded in both regions, that will lead to the city ratio to also go to 0.
To take an explicit example : suppose we start with an infinite supply of settlers at the origin, and try to build them so they are at least 3 squares apart (2 squares in between them) as fast as possible. After n turns, we will have very close to 4n^2/9 cities (4 for the 4 quadrants, 1/9 for the density). For n much larger than x, the number of those which will be within x of a border city will be something like 8nx/9 (the empire is basically a big square with side 2n, so perimeter 8n, you could make more precise but this is the highest order term).
For fixed x, 8nx/9 over 4n^2/9 goes to 0.