On Wednesday, August 30, 2017 at 11:28:15 PM UTC+2, wjhonson wrote: > On Wednesday, August 30, 2017 at 2:23:23 PM UTC-7, taf wrote: > > On Wednesday, August 30, 2017 at 1:12:09 PM UTC-7, wjhonson wrote: > > > On Wednesday, August 30, 2017 at 1:02:17 PM UTC-7, taf wrote: > > > > On Wednesday, August 30, 2017 at 12:10:28 PM UTC-7, wjhonson wrote: > > > > > > > > > My theory is that *every* snippet gets passed down to *someone*. > > > > > > > > OK, so let's take a medieval person who had only one child. That child receives exactly 1/2 of the parent's DNA and does not receive the other half. That means in 1 generation, 50% of the DNA gets passed to NO ONE. In every generation, even in a family with a dozen children, statistically some of those snippets will be lost completely. So much for that theory. > > > > > > > > Further, after 10 generations, the blocks that get passed are the same blocks passed for 30 generations, just in a progressively lower proportion of descendants - this means that two people claiming descent from Edward I may well share the same block of DNA, but it may be entirely coincidental if one of them descends from Edmund Crouchback, or Joan Fitz Roy, or Toda Aznarez of Pamplona, with the presence of this shared DNA being entirely coincidental and not due to the common descent you are trying to test. Even if you get a match, you haven't confirmed squat. > > > > > > > > I know you have it in your mind that it would be possible to reconstruct the genomes of medieval people simply by doing autosomal testing on the entire human population and using their pedigrees to extrapolate back, but such an approach would not survive the collision with hard reality. To argue that autosomal DNA testing can be used in this way because you think it should be possible is fantasy. > > > > > > > > taf > > > > > > > > > However a lower proportion, multiplied by a broader base, means the > > > same proportion or perhaps even more. > > > > So what? It isn't useful if there is no way to tell tell from whom it came. > > > > > And you are still stuck on two living people trying to compare their DNA > > > to determine if they descend from Toda Aznarez. While I'm not. > > > > No, you are stuck on a pie-in-the-sky grand across-all-time, world-spanning genome scheme that won't work and won't happen. > > > > > I'm speaking of a project which would recreate the DNA of every medieval > > > person, by testing every living person. Not just two. > > > > There are only two problems with this - first there is a finite limit to how far you can take this before the progressive loss of information with each generation, the inability to identify which parent was the source when a piece was uniquely passed to one branch, and the degree of inbreeding as you go back, very rapidly produces a quagmire. > > > > The second problem is simply that it is a pipe dream to think there will ever be such an initiative to test everyone on the planet and then serve up the data to a monstrously complex computational paradigm that compares billions of data sets, just to do something as esoteric as medieval genealogy. > > > > But that's the future (or not). In the present reality, autosomal DNA testing is completely and utterly worthless for the medieval period. Full Stop. > > > > taf > > But you can identify which parent was the source of a snippet of DNA > Not sure what you're saying there. > > The entire point of Autosomal is that you can identify from where a snippet comes. Parent? Remember the question is whether this can be applied to medieval ancestors. You have mentioned you have a personal theory about this, but as far as I can see you do not actually understand how current autosomal testing works. It has definite weak points for genealogy, though it still of course has its strong points.