551 
arch is seen a similar mass which extends in a slender stroma through- 
out the arch, and joins another mass in the floor of the mouth 
(Figs. 2 and 3). In these two masses we have the dorsal and ventral 
hyoid condensations or the proton of the columella joined to that of 
the hyoid cornu by a fine connecting band. As BENDER found in 
Testudo graeca, it is only in this early precartilage stage that the 
skeletal elements of the hyoid arch are continuous. In later stages 
where the condensations take on a definite form there is left a trace 
of it, which I shall mention later, but the actual connection between 
the columella and the ventral part of the hyoid arch is lost. Transi- 
tory as this connection is, it is important in establishing the primi- 
tive ancestral condition of the Chelonian sound-transmitting appa- 

Fig. 2. Fig. 3. 
Figs. 2 and 3. The same embryo as for fig. 1. c. arteria carotis interna; 
ex.-col.h. extra-columella hyoid connection; p.ex.col. proton of extra-columella; 
ph. pharynx; 2v.p. second visceral pouch. Other abbreviations the same as before. 
ratus, and this in turn, may help in the solution of the broader problem 
of reptilian ancestry. 
To return to the description, the membranous labyrinth has not 
differentiated beyond a sac and duct, but the surrounding cells are 
gradually becoming dense to form the capsule. Its latero-anterior 
region shows an area of more densely arranged cells. When this study 
was made three years ago this dense area was unquestionably con- 
sidered the proton of the base of the columella. But in view of 
BENDER’s recent work this point has been carefully reconsidered. 
As was previously stated, he found the entire columella in Testudo 
graeca to be of hyoid origin, and suggests that the crista parotica 
might be mistaken for the base of the columella. In the series under 
