With a “fair coin” 13 versus 9 is certainly statistically significant. These are random events, so the uncertainty is Poissonian: the variance is +/-3.6 for 13 and +/-3 for 9. The standard deviations, then are +/1.9 and 1.73, respectively. So, at two standard deviations, the averages are distinct.
But, more importantly, it’s not a remotely fair coin. There’s no way we ever measured the peak winds for anything other than land-falling hurricanes more than about 100 years ago, and there’s no way we did it particularly well more than about 50 years ago. You have the random ship out in the middle of the Atlantic that was in the right place at the right time in, let’s say, 1900, but if it’s in position to make peak wind measurements in a Cat 4 or 5 storm, I doubt anyone is bothering to make measurements and I doubt it’s a better than 50/50 chance they’d be able to tell anyone about it after the fact.
I would agree with your observation that the difference is statistically significant. I like your observation that the ability to measure has been much refined in the latest 100 or 50 years. That is obvious and I should have thought about it.