The question then is do you know how many total retirements there were in the 142 GB vs GB matches? I was assuming that 3 was the total number of retirements in such matches. And it looks still the case, you say that there were 3 RETS- 2.11% of these matches. 3/142 Player 1s retired and 0/142 Player 2s did not?

If it is indeed 3, then the comparative % to the GB vs Pesky figure is 3/284 = 1.06%.

Or basing on the number of retirements against matches:
There were retirements in 3/142 GB vs GB matches, 2.11%
And retirements in 34/1606 GB vs Pesky matches, 2.12% ( 18 of which were GB players, 16 were Pesky ).

We're counting how many times the match was won by a RET
In the 142 all GB matches, the result was decided by RET on 3 occasions.
In my data, this is 284 rows of data, and 6 entries for the ret. Each all GB match has two rows, one for each player.

Here is a sample from the rows relating to the SF here, where Laura RET v Katy D - a RET in an all GB match.

Different Match ID's, one for each players instance. Each player occurs in a separarte row as P2, The RET is recorded on each row, but only as P1 in Laura's Row, as she was the retiring player.
When I count across this rows for {[RET]=TRUE,[RET Player] = Player 1} I will get, correctly, that this match had one RET, and that the RET was by a GB player.
The number of unique all GB matches is not double counted, because my formula for that simply has /2 on the end of it, to recognise the nature of All GB matches; thus the reason for the Green [All GB?] field.

Where it's not All GB, I can just count P1 & P2, and get the same equivalents.

I very patently took pains to not suggest significance, in the statistical or mathematical sense, "This would all be more meaningful with more years of data."