Using google I found this book. I guess this is the one you're refering to?
One thing is for sure: the starting equity must be given, otherwise it makes no sense. That makes me think the author did probably not get this table himself, indeed.
I did a few tests and it seems that it should be between $15 and $20, but (oddly enough) the values in the matrix don't match exactly or even very closely for any initial account.
Anyway, here's what the table should mean: (for instance, for the value "2:1, 60%", and assuming the initial equity is $20)
Starting with an initial $20, we perform 100 trades successively, each of which can be a winner (with probability 60%), meaning that we win $2, or a loss (with probability 40%), meaning that we lose $1. During the sequence of trades, our account fluctuates, it may even become negative. The table tell that, with probability 14%, at some time we will have less than $10 in our account.
If instead you start with $50, you would have to redo the whole table.
I don't think there's a formula for the values of the table. They were certainly obtained by doing computer simulations. The idea is the following: you make a program that performs lots (thousands) of random sequences of 100 trades (each of which is a win with some given probability, like 60%), and you make it compute the proportion of sequences where ruin happens (i.e. at some time between 1 and 100, the equality goes below 50% of the initial value).
I don't know if such simulations can be done with Excel. I did it with Scilab (a free program for numeric computations), but you could use any programming langage (or a programmable calculator).