In micropaleontology, we see an assemblage of shapes on a microscope slide. We mentally group them into a number of different "kinds", each of which exhibits some degree of variability as a result of natural variation, and of differing orientation of specimens on the slide. Before we can identify the kinds present, we need to form some concept of which characters we are going to use (describe, count, measure) to distinguish one kind from another. The characters chosen will be those that serve best to distinguish the different kinds of fossils we see. Once the characters are selected, we should use any cost-effective means to minimize the amount of subjectivity in our observations, and to increase their accuracy. Counts and measurements are in most cases best made with the human eye and measuring devices, though in some cases (especially concerning the evaluation of patterns ) computers can help do a better job.
There are several different approaches to quantifying the shapes that are so important to the human eye and brain in distinguishing different kinds of fossils. Mainstream biological morphometrics depends fundamentally on quantifying the spatial relationships of points on a skeleton (or organism) selected as "landmarks" (Rohlf and Bookstein, 1990 ). Measurement of the changes in these relationships quantify similarities and differences between the overall shapes.
Shapes of outlines have been described quantitatively by applying Fourier transforms and eigenshape analysis (see the paper by Lohmann and Schweitzer in the above-cited compilation by Rohlf and Bookstein).
Terms applied to parts of the shapes of ichthyoliths (microscopic teeth and other fish skeletal debris) have been defined by template sketches, to which the observer can match elements of the fossil specimen (Doyle et al., 1974 ).
It is interesting that none of these possibilities for quantitatively describing fossil shapes, features and patterns has entered widely into the everyday work of micropaleontologists, even in academic laboratories - though their adoption would surely strengthen the foundations on which taxa are established, and permit higher resolution in interpreting the fossil record. I suspect that the two main reasons are that (1) most paleontologists believe that it is not worth the extra effort that would be involved in quantifying their observations, and (2) software developers do not see a market sufficient to justify their producing fast, user-friendly software to accomplish these tasks.