Argh, wrote a lengthy reply which I did not save as a separate document, and the system kept kicking me out when I tried to post it, to the point where it was lost. I try again, with apologies if the earlier attempt shows up eventually in this forum. I do research in human genetics, and we often see results of germ-line mutations, i.e. mutations in the genomes of individuals where both parents are somatically apparently unmutated. According to embryology textbooks, at some point in early embryogenesis about 50 or so cells are set aside to become the eventual germ, yielding the gametes (sperm and eggs). It is believed that these cells do not have any somatic descendants themselves. It can easily be imagined that a spontaneous mutation arising early enough in an embryo could show up in somatic and germ line descendants. However it seems very unlikely that any somatic cells in adults (I refer to humans throughout) have germ line descendants. So vertical transfer of somatic mutations (or epigenetic changes in chromatin for that matter) are pretty much ruled out. I see no theoretical problem with horizontal transfer of genetic information via a viral vector. That could in principle include transduction, so for example a virus infects a somatic cell, recombines to take up some human genomic DNA, which is transmitted to progeny that later infect germ line cells and potentially land that human DNA of somatic origin in the germ line genome which is then transmitted via reproduction to actual children. I don't know of any known virus with all these properties, but it could be imagined. Whether endogenous retroviruses could become active in a somatic cell, and then be horizontally transmitted to a germ line cell of the same actual person is also unclear to me biologically, but I suppose it is probably feasible though probably rare. In general the remnant endogenous retroviruses in our genomes are inactive, though perhaps not irrevocably. On the other hand some other repetitive elements in our genomes are active on evolutionary time lines, and sooner or later one of them could presumably find a way to get from a somatic to a germ line cell in the same individual. I think if these events were frequent geneticists would have seen the medical consequences already, so I guess they are rare if at all.
, i didn't say it was relevant to natural selection or would affect natural selection in any way. I just said it would be a mechanism of evolution that wasn't natural selection. I guess the outcomes of such a mechanism would be subject to natural selection, which is why i said 'supplemental'. 
)