Mussorgsky’s Pictures at an Exhibition is one of the classical pieces that is most familiar to me. Ever since my high school marching band did a show based on it, it has been a constant companion, at least in the version orchestrated by Ravel. More recently, however, I have been spending a lot of time with the original piano version, in part out of simple curiosity, but more directly because I plan to use it in my fine arts course — not only because of its unique status as a piece of music “about” visual art, but also to highlight how orchestration affects our reception of a piece of music.
For those who are familiar with the orchestrated version, it can be difficult to believe that Mussorgsky ever intended it as a solo piano piece to begin with. Leaving aside its unwieldy length, some segments seem to be screaming out for full orchestral treatment — most notably the majestic horns of “The Great Gate of Kiev.” The orchestrated version is so much better known, in fact, that the original can seem like a work of subtraction or abstraction, taking away the variety of a full orchestra. In a way, though, it also adds an element. The dissonances are much harsher and stand out more clearly when they’re not spread across a variety of sections, so that some of the segments (like “The Gnome”) can even sound like precursors to atonality.
The question that has returned to me again and again, though, is why exactly Mussorgsky would have started out with a piano version in the first place. It seems so counterintuitive in so many ways, and it’s not as though he lacked the ability to write for a full orchestra. If we take seriously the notion that this is meant to somehow resonate with the effect of an art exhibition, though, I think it makes more sense. Contemplating art is, after all, a very solitary and cerebral pursuit in most cases — hence why a solo instrument could seem more appropriate. In the piano version, the one aspect that struck me as manifestly more convincing are the recurring “Promenade” interludes, which when performed on the solo piano seem much more evocative of the act of reflection while walking between two canvases.
Further, the very inadequacy of the piano (most striking, perhaps, in “Samuel Goldenberg and Schmuyle”), the manifest limits the performer (even a very gifted performer like Evgeny Kissin, whose recording on Spotify I recommend highly) strains against, seem to speak to the difficulties of responding to art, the sense that there’s “something more” that one can’t quite capture. With the full arsenal of the orchestra, it’s almost too easy, and this very perfection betrays the experience that it’s meant to recreate.
An und für sich 
With the full arsenal of the orchestra, it’s almost too easy, and this very perfection betrays the experience that it’s meant to recreate.
Exactly this. I love the Ravel orchestration (and am also a fan of several other more obscure orchestrations or arrangements) but the heroic, doomed effort of a solo pianist playing the piece is just a totally different—and in some ways more compelling—listening experience. As you might expect, this aspect of things is enormously heightened when you’re watching a live piano performance, because the physicality of the struggle is so clear to see. It’s a harder piano piece than it sounds in many respects, famously extremely awkward for the instrument—as opposed to e.g. Liszt, which sounds flashier but is usually easier than it sounds.
Perhaps the inadequate striving of the piano to express the full color of the suite is apt, given, as you pointed out, that the whole suite is an attempt to bridge the incommensurable in its concept: music about painting. Why not replicate the tension at the level of instrumentation? For this reason I have always preferred the piano version myself (and it’s so much fun to play).
Strangely enough, for me this music is closely connected with my youthful passion for science fiction (I’m no longer youthful, and it’s still a passion — I mean, when I first fell in love with the genre). The music teacher at our primary school played Tomita’s synthesiser version of the suite, in order to illustrate the (for the early 1970s) radical new musical technology of synthesisers. I’d already made the sonic-semantic connection between synthesisers and SF. I’m not sure why; and knowing nothing else about the context the music sounded sfnal to me. It still does. More, since I first imprinted onto this music via Tomita, hearing the Ravel orchestra version (and, much later, the original piano) meant that those latter sounded paradoxically belated and generally just wrong to my ears, like steampunk versions of late 20th-century SF works.