WITH NOTES ON THE WEST INDIAN SPECIES. 



235 



posterior side forwards, reckoning from the base. Moreover, the posterior side coincides 

 with the anterior free edge of the dorsal septum. 



The second root is likewise hollow at the base and very slightly oblique in its 

 ■course from medullary tube to epidermis. 



Series iii. The first root is hollow and runs along the anterior edge of the 

 dorsal septum. The second root is hollow and arises close behind the first root from 

 a common neural crest (PI. XXVIII. Figs. 4 a — 4 c). The third root is hollow and 

 smaller than the preceding. All are directed somewhat obliquelj' backwards. In 

 describing these roots as hollow, I refer particularly to their origin as hollow diverticula 

 from the medullary tube. In some cases the h;men appears to be broken up into 

 discontinuous portions towards the distal end of the root. 



Series iv. The first root is almost entirely solid at its origin, but there is a 

 slight indication of pouching at its base. There are disconnected traces of a lumen 

 in the root itself. This root is large and runs horizontally backwards for a relatively 

 long distance. It is anterior to and independent of the dorsal septum and has not 

 an extended neural crest for a base. 



The second root is also solid, narrower and shorter than the first ; it accompanies 

 the anterior edge of the septum. 



Series v. The first root has a median origin and is hollow, running obliquely 

 backwards along the anterior free edge of the dorsal septum. The second root is also 

 median and hollow like the first, and comes off from the medullary tube before the 

 first root has reached the epidermis. 



The above observations may be summarised by saying that in Pt. flava, the nerve- 

 roots of the collar are feiv in nuinher but vary in number, length, course and calibre; 

 sometimes they arise from a common neural crest^ ; sometimes they arise to one side 

 of the middle line instead of being quite median; and, they are primarily IloUow. 



Ventral Septum of Collar. 



There is no ventral septum in the collar of Pt. flava. Although the complicated 

 longitudinal vascular plexus along the ventral side of the throat causes an extensive 

 fold of the basement-membrane, it does not reach across the collar-cavity to the 

 basement-membrane of the ventral epidermis. The two halves of the collar-cavity are 

 therefore in free communication below the ventral vascular fold. In some cases the 

 ventral septum persists through a great part of the collar- region as in Pt. sarniensis ; 

 in others it is restricted to a narrow tract at the posterior end of the collar as in 

 Olandiceps talaboti. It is also largely persistent in Spengelia (see below). Spengel 

 has pointed out that neither the dorsal nor the ventral septum of the collar is ever 

 complete ; but both septa, when present, always have a free anterior margin. Perhaps 

 the vascular fold suspended from the basement-membrane of the throat-epithelium in 



1 In other cases, besides that mentioned above in connection with Series iii., successive roots come off 

 from a common crest but not, as a rule, so strikingly as appears in that series. 



