hmm.
Lets look at the mother. The egg cells are created along with the baby and are not being produced all the time like the sperms cells. So their nucleus and DNA is in that sence static.
So in order for the mother to pass on a memory (tragic or otherwise) through the DNA what would need to happend was:
* The mother experience the event and record in her brain.
* The memory is somehow selected. then it is somehow transcribed by an uknown mechanism into an unknown thing where it is stored in some unknown format. Keep in mind the memories seem to be stored as a pattern of synapses among millions of neurons.
* Those things from before travel to all the moms egg-cells and go into the cells. Here they begin to manipulate the DNA and splice the information into them. That information need to be stored in such a way that if affect the way the brain grow so that the moms memories are inserted.
Furthermore, notice that, just like the fingerprints and all other structures in humans, the exact layout of the brain is not known. How does the DNA know exactly how to affect the neural layout of the brain so that the memory is interserted?
Its a cute idea, but a-priori i think it has a lot of things talking against it, particulary the transcription/encoding phase. I think the greatest problem is the distributed layout of the brain; if the brain work any way like a normal hopfield network, no two memories are stored a like in different brains, but will depend on the topology of the other brain which is determined by what other memories the baby will learn sometimes in the future. Thats a pretty big problem.
That, and the fact there are no reliable observations of this happening, is why i think it is unlikely.