My father knew Charles Ives, for all the good it did me

Over at On An Overgrown Path, an intruiging proposal:

The new sciece of epigenetics has identified that the cells that make up our body and determine our wellbeing are not controlled primarily by our genes, but rather by the physical and energetic environment in which we live. It is early days and some of the advocates of epigenetics hover uncomfortably between science and shamanism, while similar approaches such as Alfred Tomatis' Mozart Effect continue to be treated with scepticism. But epigenetics is science rather than quackery, and if a medically proven causal relationship could be established between classical music and wellbeing, the case for live music, music education, music therapy and many other threatened activities would become much stronger. To date there has been little attempt to connect epigenetics and classical music. Surely it is it time to explore new ways of arguing the case for classical music?

I left a comment over there, but it's worth expanding a bit. Well, first of all it's worth addressing Pliable's first sentence, which makes a strong but not accurate case for epigenetics. In essence, epigenetics refers to, as Jerry Coyne puts it in a put-down of the hype surrounding the field, "inheritance not based on coding changes in the DNA". Which is to say that the environment (such as a stressful event like a famine) can affect genetics, causing changes not in the structure of DNA but in gene expression, changes that can be inherited (though only for 1 or 2 generations, it seems). Saying that our cells are "controlled primarily" by the environment is an exaggeration. Some specific epigenetics examples are known, but this doesn't of course mean that every single bloody thing that ever happens will cause a heritable change in gene expression, which is what that BBC Horizon thing linked above seems to imply:

At the heart of this new field is a simple but contentious idea – that genes have a 'memory'. That the lives of your grandparents – the air they breathed, the food they ate, even the things they saw – can directly affect you, decades later, despite your never experiencing these things yourself. And that what you do in your lifetime could in turn affect your grandchildren.

Reminds me of Alexei Sayle's joke, along the lines of "They say that what you have to worry about isn't what killed your parents but what killed your grandparents. So I should be fine unless I'm attacked by a horde of Cossacks". As for genes having a memory, this is perhaps just a noble attempt to couch something in layman's terms but it invests a stupid mysticality into the process (the article is titled "The Ghost in Your Genes"). Yes, epigenetics means that a gene has a "memory", in the same way that my downstairs bathroom wall has a "memory" of when my son attacked the wall with a chisel - in this case, the "memory" is also known by the scientific term "holes".
The main point I was making in my comment on OAOP was, if you want to prove that listening to classical music has health benefits, then epigenetics seems an unnecessarily complicated way of going about it. We're talking about transgenerational effects, so unless you don't need results until, say, 2050, you can't study it prospectively. So that means you have to get today's adults and find out how much classical music their parents and grandparents listened to, and come up with an appropriate health benefit to study (stress hormone levels? IQ?). The huge difficulty is eliminating all possible confounders. If you're studying the impact of a famine, that's not so hard - people either went hungry for a period or they didn't - but for classical music you're talking about a lifetime of what you might call low-impact experiences that may often go hand-in-hand with other experiences - reading, for example - as well as accounting for the usual things like educational level and socioeconomic status. And of course these conditions apply not only to the grandparents but the grandchildren too. By deciding on an epigenetics approach you're going for an unnecessarily complicated level of study: why investigate the impact of a lifetime of classical music listening on an individual's children or grandchildren, when we don't even know the impact of same on the individual?
That sort of study would be easier to perform, though maybe not easy to produce a convincing conclusion. The so-called Mozart Effect, on the other hand, is much much easier to study: measure baseline levels of whatever property you're studying (stress hormone levels, IQ); play music for the subjects; measure the new levels; and keep measuring them to see how long the effect lasts. Meta-analyses cast doubt on whether there is an effect, though.
The other thing to remember with all of these approaches to showing societal benefits for classical music is - as commenter mrG alluded to on OAOP - that you must also be prepared for other, competing, studies showing the societal benefits of other forms of entertainment. You think that all Justin Bieber fans do all day is stand around screaming and wetting their pants? Hell no. They're in the labs right now. This is a war, people. We need to be ready.