Darwin's Radio, Children, ... and Mice?

I just got finished reading Darwin's Radio and Darwin's Children by Greg Bear and enjoyed them greatly. Bear makes an interesting case for punctuated equilibrium evolution via human endogenous retroviruses. I have to admit, if it involves junk DNA then I'm interested.

No sooner do I put down the last book when an article is published in Nature about RNA-mediated non-mendelian inheritance. As an AP news article explains:

In a startling exception to classical genetics, mice in a lab experiment have inherited an effect of an aberrant gene without inheriting the gene itself.

"Startling" is an understatement.

[update Wednesday, May 31, 2006] I sent the above link to Greg Bear. He replied:

Thanks, Matthew! This is indeed a fine and puzzling result. It may very well be some version of RNA inheritance--but other possibilities may include a "backup" copy being reassembled from so-called non-coding portions of DNA, or from other apparently unrelated genes. These are my own suppositions. Whichever, it's a fascinating question: How many genes can be reassembled using this process? How many genes can be overridden or canceled? Can more than one genetic pathway produce the same phenotypical result? If so, Mendelian inheritance and much of evolutionary theory is in need of revision.

And how!