On the other hand, perhaps there are some works that could really use five, or six, levels. A model is always just an approximation of reality, but four levels seems a pretty good compromise to me. There are always the ability for other relationships to be drawn other than the ‘level 1′ sets, to specify things further.

Although the whole model is, for better or for worse, based on a formal abstraction of traditional cataloging practice, that was mostly designed for monographs, with everything else being an ‘edge case’. But I’m not sure what to do that’s better. If there are human catalogers describing the bibliographic/documentary/information universe, we’re a lot better off if they’re all using the same formal model, so we can have all agreed on what the data we’re recording actually means. But maybe there’s a better way to do it? If there is, nobody’s supplying it.

I just printed off Martha Yee’s article, “Lubetzky’s Work Principle.” Thanks for recommending it.

Several of the authors in the Taylor book make the same point as Yee concerning the Procrustean bed of Work-Expression-Manifestation-Item. In fact, a lot of monographs would work better with just Work-Manifestation-Item, IMHO.