What i find interesting there is that multiple approaches seemed to arrive at about the same result, 2/3 of a year between deaths. That makes me think it's highly probable that it's the right answer for the current time period at least.