I find this article very interesting as it outlines founders and population genetics in very basic terms for a neophyte like me.
This concept also applies not just to a breed of dog, but also to any closed breeding pool. If a breeder only breeds within a closed breeding pool, using the same sires and get of the sires and dams and progeny of the dams over and over without outbreeding by bringing in new sires and dams with different lines, this same thing can happen to the gene pool of that breeder's breeding pool.
"Let's pretend these 11 dogs are the "founders" of your breed - they are the first dogs entered into the studbook. All subsequent members of the breed are descended from these dogs only. The breed has a closed gene pool.
All of the genetic variability that will ever exist in your new breed is present in these dogs. Mutations probably won't add new, useful genetic variation because most mutations are detrimental. If the mutated gene is dominant and detrimental, it will likely be weeded out very quickly. If the mutation is recessive, it is not expressed unless an animal is homozygous for the allele by inheriting a copy from each of its parents. In the heterozygote condition, a mutated recessive allele can lurk in the genome for generations without ever causing a problem. So, unless additional "founders" are added to the population at a later date, all of the genes you will ever have to work with in your breeding program are present in these dogs.
In each of these dogs there are at least a few (and perhaps many) recessive genes that could cause genetic disorders. But these disorders will only expressed if a dog is homozygous for the disease allele - it must have TWO copies, one from each parent. As long as the disease genes are rare in the population, very few animals will ever display the illness."