characteristics modeled in a particular study; that is their prime virtue.
We are forced to be explicit about the pattern of change, which can then be mapped onto a phylogeny, giving us a track through morphospace, and hence empirical constraints on adaptationist hypotheses. Nobody thinks that there is some objective fact of the matter about the dimen- sions of lizard morphospace, independently of the hypotheses under investigation. It is a virtue of morphospaces used for this purpose that they employ relatively few dimensions and abstract away from most characteristics of the organisms under study. But of course this also makes it clear that morphospaces developed this way are not models of overall morpho- logical disparity. Are more ambitious uses of morphospace possible? We began chapter 3 with Gould’s claim that metazoan disparity peaked in the Cambrian; we concluded that chapter with the claim that this idea was simply untestable using high taxonomic ranks to measure disparity. Could one do better by representing disparity via a morphospace? Anal- ogous issues arise for contemporary biota. For example, a long-standing challenge to macroecology is to explain the latitudinal gradient in spe- cies richness. Tropical and near-tropical ecosystems are more densely packed with species than temperate and polar ones, even though it now appears that their speciation rates are lower (Weir and Schluter 2007). Are they also more disparate morphologically, or is tropical richness achieved (as some have suggested) by specialization, by dividing mor- phological and ecological space more finely? However intriguing these questions might be, it is clear that defin- ing a global space of morphological diversity is not an empirically or computationally tractable way of answering them. To answer Gould’s question, for example, we would need a space of high dimensionality, for we would need a dimension for every respect in which Cambrian animals did vary; perhaps a dimension for every respect in which they could vary. And while we can measure some of the characters some of the time, it is clear that we cannot measure all of the characters all of the time. But even setting aside the problem of tractability, these global notions may not be well defined. How could we choose dimensions in answering Gould’s question? One option would be to use the Cambrian fauna to anchor our choice of dimensions. But that would bias our project; instead of the pull of the present biasing our taxonomic judgments about the ancient Cambrians, the limits of the ancient would bias our choice of dimensions in judging more recent disparity. Major evolutionary novelties are the invention of new ways to be similar and different. During the Cambrian radiation, the invention of segmentation, a jointed exoskeleton, and the like were