THE SUDDEN ORIGIN OF NEW TYPES 419 



nacreous type of shell. In the Palaeozoic forms of the Nucitlidce, 

 however, the ligament is mostly external, whilst in the recent 

 forms it occurs internally, below the umbo. It is probable that 

 Tellinomya pectunculoides of the Ordovician, with its equilateral, 

 nearly circular form of shell, its curved hinge-line, equal 

 adductor-impressions and semicircular, simple pallial line, stands 

 very close to the ancestral form, and that the Arcidce, with the 

 straight hinge-line, formed an early specialisation. 



The rise of the Gastropoda, and still more of the highly 

 organised Cephalopoda, must have been correlated with the 

 acquisition of a jaw-ribbon or radula, which is such a typical 

 characteristic of these classes that they are classed together as 

 Glossophora. This structure forms an immense advance on 

 the presumably Annelidan jaws (conodonts), which are known 

 from the Cambrian. But, unlike the conodonts, the radula has 

 never been definitely recognised in fossil form; hence its evolu- 

 tion cannot be tested by palaeontological methods. However 

 it has been conclusively shown by Martin F. Woodward ^ that 

 the radula of Pleurotomaria represents the most archaic type 

 amongst existing Gastropoda ; and this genus is known to have 

 persisted from, the Upper Cambrian. Woodward comes to the 

 conclusion that its radula, in which as many as three tracts 

 on either side of the median are distinguishable, was derived 

 from one in which all the teeth in a transverse row were 

 similar. 



An early reduction must have taken place to reach the 

 optimum of three tracts on either side. Even in the case of 

 Pleurotomaria the five tracts are not equally distinctly differ- 

 entiated from each other, so that at a superficial glance only 

 three tracts would be distinguished on each side of the median 

 one. From this primitive Rhipidoglossate type all the re- 

 maining types can be derived by fusion and reduction. It will 

 be sufficient here to indicate very briefly some of the main 

 lines along which reduction has taken place. In the Rhipi- 

 doglossate Trochus only two lateral tracts are clearly dis- 

 tinguishable : X. 5. I. 5. X. In the Docoglossate radula of 

 Patella we find the series 3. i. 2. o. 2. i. 3. in which it is 

 assumed that the median tooth is either altogether aborted or 

 sometimes reduced to a rudimentary plate. In the Tsenio- 



' (2tmfL Jfliirn. Micr. Sci. N.S. xliv. p. 255. 



