138 



ME. H, M. BEEKAED O'N EECENT POEITID^. 



unit. The following reasons seemed to justify this acceptance : — 

 (1) Well- developed " trabeculae " occur iu the walls of many 

 species. (2) The pali appear to be the tips of others. (.3) A 

 vertical section through a coral lum frequently shows it to have 

 been built up o£ long nodulated threads (trabeculse) running in 

 the line of growth and joined together at intervals by cross- 

 pieces arranged parallel with the surface : this, however, is truer 

 of Pontes than of Q-oniopora. (4) In some forms there appeared 

 to be a regularity in position and arrangement of the trabeculse 

 which suggested tlieir having real value. 



After examining a great number of specimens, I reconstructed 

 on the simplest possible plan an ideal primitive skeleton of a 

 Goniopore built up of trabeculse (see fig. 2). Bat the longer 

 the actual specimens were studied with this hypothetical ancestral 



Fig. 2- 



Ideal arrangement of the " trabeculse," if regarded as morphological units, 

 necessary to explain the skeleton of a Goniopore, the columellar tangle 

 being omitted. A, in ground plan : B, in vertical section ; p, central pali, 



form, the more impossible it became. The meshes of the lattice- 

 work were always pores, often very irregular in size and arrange- 

 ment, in otherwise lamellate septa. Surely some forms would 

 have retained the rectangular lattice-work with the trabecules per- 

 sisting in their primitive importance. But no such condition was 

 found. Then, again, the pali failed as tips of growing trabecule. 

 They were plates when the septa were but slightly perforated, 

 and were only tips to the narrow divisions between the large 

 perforations in other cases. Lastly, the finding of the growing 



