How do they know that we're all descended from a single woman?
As a biology teacher, perhaps I can be allowed to explain a bit; some of the DNA experts on here may well object to some of my simplifications and analogies, but I hope this will be useful.
This response applies only to the female line.
In every cell there are structures called mitochondria. It's believed that deep, deep in the evolutionary past the mitochondria were independent cells which in some way took up residence inside other cells. The consequence is that the mitochondria have their own DNA; and that new mitochondria are only ever formed by division of pre-existing mitochondria, in a sort of parallel way to the method of bacterial reproduction.
Now at fertilisation, only the egg cell contributes mitochondria; the sperm cell does not. So each individual human's mitochondria are clones of the mitochondria in the egg, and therefore of the mother.
Every time a new mitochondrion is formed, the mitochondrial DNA has to be copied for the new mitochondrion. This is done by some really exquisite biochemistry and with quite amazing accuracy, but there is an error rate, although it's very small. Some of these errors have no effect on the body; a few are catastrophically lethal; some have lesser adverse impact. The copying process is not intelligent so mistakes are random.
DNA is a wonderfully constructed molecule. For the purposes of this discussion, think of it simply as a long line of subunits, which are labelled A, C, G and T; the sequence of these subunits ACTTGACCTA or whatever is the genome, the set of instructions for constructing and running our cells. It is now possible to sequence lengths of DNA accurately and reliably, with high levels of automation.
Back to the mitochondria and the copying process. My analogy here is the scriptorium of a monastery. Imagine manuscript A. It's copied out by two monks Brian and Charles, but they have the extra difficulty that it's in a language they don't know. However careful they are, it's likely that they'll make mistakes. This gives us manuscripts B and C, with different errors.
Manuscript B then gets copied by two monks, David and Ethelred - copy D will have the errors of Brian and David, copy E those of Brian and Ethelred. And so on....
So an expert studying the 20th or 30th generation copies of manuscripts would be able to judge (a) whether the manuscript in question originated with Brian or Charles, by looking for their specific mistakes; and by comparing the numbers of different errors (b) how recent the shared original was - for example, C and D will be more similar to each other than to E. To use an analogy within an analogy, the scholar could create a family tree of the various copies.
This would be easier for our scholar if all the monks were equally incompetent - for example, if they each made on average a one-letter mistake per page. This is unlikely with monks, but seems to be approximately true for the mutations in the DNA. So we can treat this as a sort of molecular clock.
Very roughly, the DNA researchers have been able to look for differences in DNA sequence and therefore establish all sorts of interesting things. What differences are there? How similar are mitochondrial DNA sequences from different places? And therefore when did different mutations happen, leading to the sort of deductions made in the Eddie Izzard programmes or on the website (see my earlier post).
The other fortunate piece of biology here is that the Y-chromosome with its DNA sequence is inherited in the male line only: men have one Y-chromosome which is passed on intact in Y sperm which if fertilised will lead to male children. So you can apply the logic above to this line, as was done in the second Eddie Izzard programme.
The behaviour of the rest of our DNA - the vast majority of it, in the other 22 pairs of non-sex chromosomes (autosomes) and in the X-chromosome, is very much more complicated, and forms the basis of the autosomal DNA tests that are also available. Suffice it to say that the DNA is shuffled and reshuffled every time that eggs and sperms are formed, and that therefore neat lines of descent don't work in the same way.
I hope this helps......