THE SPECIES PROBLEM WITH FOSSIL ANIMALS 



fact of variation, however excessive, would not of course be 

 used as a justification for a typological approach. But the diffi- 

 culty in this case is enormously compounded by a combination 

 of stratigraphic, biologic, and practical economic circumstances. 

 In the first place, abundant collections of these shells have been 

 obtained from widely distributed localities and very numerous 



POT CLAY COAL-SIX INCH MINE NON-MARINE SHELLS 



^-^3*-C2*-€ 



c_. 



"x 



/ 



-■& 



/ 



r^jb — 



/" | \ 

 / . — lr - \ 



/ t 



/ 



/ f 



\ / 



locol.tj 2, DABWEN Hill 



locol.r^ 3 WRIGHTINGTON 



Fig. 1. Variation diagrams of two local populations of Carbonicola ? 

 from a thin shell bed immediately above the Pot Clay and Six-Inch Mine 

 Coals of Yorkshire and Lancashire. These shells are referred to C. ? lenisul- 

 cata Trueman and C. ? aff. belhtla Bolton. Distribution diagrams, inset in 

 the upper left-hand corner of each pictograph, show the numerical strength 

 of variation trends. Black circles distinguish the figured variants, and white 

 circles show the disposition of the remaining shells, the position of each 

 one being controlled by its resemblance to one or more of the figured vari- 

 ants (Eagar, 1952b). 



stratigraphic horizons in the British coal measures. The result is 

 an embarrassment of riches. To make the problem more complex, 

 many pairs of unit taxonomic characters vary independently, or 

 at least show low total correlation. From this it follows that 

 Statistical characterization of populations by means of simple 

 univariate or even bivariate distribution clusters will usually be 

 unsatisfactory. By means of multivariate regression analyses 

 Leitch (1940) has shown that it is possible objectively to identify 



