path (Harland et al., 1990; Semikhatov, 1991; Knoll and Walter, 1992; Compston et al., 1992; Bowring et al., 1993). A biostratigraphic framework based on stromatolites, microfossils, and (in younger rocks) both the body and trace fossils of animals can be used to divide this nearly 1200-Ma expanse into recognizable intervals of various lengths. Complementing this is an increasingly well-supported chemostratigraphic framework based on the distinctive pattern of secular variation in the isotopic compositions of C and Sr in carbonate rocks (Kaufman and Knoll, in press). These data define the chronostratigraphic scale now being calibrated. Within the period under consideration, younger intervals are shorter than older ones, both because strong Neoproterozoic isotopic variation has no parallel in the Mesoproterozoic record and, more importantly, because of the finer biostratigraphic resolution in younger successions.

For the purposes of this analysis, I have divided the period from 1700 to 520 Ma into 17 intervals as shown in Table 1 and Figures 1–3. Table 1 and Figure 1 also show my placement of representative microfossil assemblages into these intervals. Others might estimate the ages of interval boundaries differently, and one or two assemblages might be moved to bins adjacent to those chosen here. However, no assemblage placement or estimate of interval duration is so egregiously uncertain as to affect the analysis in a substantial way. That is, relative to the strength and time scale of the pattern observed, uncertainties of time are acceptably small.

The Paleontological Data Base: Taxonomy. For the estimation of evolutionary tempo, I will restrict consideration to the organic-walled micro-