999 
Proceedings of the Royal Society 
the reasons which lead me to believe that there is in reality in- 
sufficient ground for this comparison, and that the structures in 
question are morphologically different. 
Finally, no traces of septa , either in the form of radiating 
lamellae, or of vertically disposed spinules, or even as those plications 
of the wall which constitute the “ pseudosepta ” of Heliopora , have 
been detected in Tubipora. 
2. Minute structure of the skeleton of Tubipora. — The micros- 
copic structure of the corallum of Tubipora was first investigated by 
Ivolliker (“Die Bindesubstanz der Ccelenteraten,” p. 169, 1866), and 
very correctly described and figured. Professor Perceval Wright 
subsequently showed (“Ann. Nat. Hist.,” ser. 4, vol. iii. p. 377), 
that the corallum is really formed of fusiform calcareous spicula, 
secreted by the ectoderm. Over the greater part of the theca these 
spicules become completely amalgamated with one another, so as to 
form a rigid case ; but they remain loosely united or separate at the 
growing summits of the tubes, which are thus capable of withdrawal, 
during the retraction of the polypes, into the dense portion of the 
thecae. 
As the result of my own recent observations, carried on princi- 
pally by means of thin transparent sections, I find that the entire 
calcareous skeleton of Tubipora , as pointed out by Kolliker, is per- 
meated throughout by a system of minute, parallel, closely approxi- 
mated tubules or canaliculi, which sometimes bifurcate, and 
occasionally anastomose. In the walls of the corallites themselves 
these tubuli (fig. 2 a), run at right angles to the inner and outer 
surfaces of the thecae, opening both externally and internally by 
well-defined, rounded, and slightly dilated apertures (fig. 2, b), which 
can be readily observed by the examination of any fragment of the 
skeleton by means of a hand-lens or with a low power of the 
microscope. In the connecting-floors of the corallum, these tubuli 
are just as numerous and as well-developed as in the walls of the 
corallites, but here they run at right angles to the floor, either passing 
right through from the upper to the lower surface, or, more usually, 
opening internally into the large ramified canals which run in sub- 
stance of the connecting-floors, and connect the visceral chambers of 
adjoining polypes. It follows from the above that the skeleton of 
Tubipora is in reality completely porous, the canaliculi by which it 
