COLLAR CAVITIES OF THK LARVAL AMPHIOXUS. 249 
do not enable ns to answer. Never theless, attention should 
perhaps be called to a constant but inconspicuous feature of 
our second stage larvas — the mass of cells marked with a 
point of interrogation in PI. 18, fig. 4. They occur just 
behind the communication above mentioned, and it is tempting 
to regard them as a proliferation of ectoderm into the first 
myoseptum and as the rudiment of the nephridium. We 
cannot, however, assert that this is so. 
In conclusion, our acquaintance, limited as it is, with the 
early stages of development of Amphioxus has convinced us 
of the need of careful experiment with the object of dis- 
covering better methods of fixation of the larvae. This can 
only be done by rearing them in large numbers in the 
laboratory — an undertaking never yet* achieved. Only so 
can material be obtained the study of which will give satis- 
factory answers to the many morphological questions which 
still remain doubtful. 
Note by Prof. E. W. MacBride. 
The investigation carried out by Messrs. Smith and Newth 
in my laboratory on the development of body cavities in the 
larva of Amphioxus has led them to results which in some 
respects are different from those which I have published on 
the same subject. 
In particular they find that the space into which the right 
collar cavity opens as it sweeps downwards towards the mid- 
ventral line is the splanchnocoel, and not, as I supposed, a 
distinct cavity lying external to the splanchnocoel, which 
later became the cavity of the atrial fold, or, as van Wijlie 
terms it, the pterygocoel. 
After a careful examination of the preparations made by 
Messrs. Smith and Newth, which were based on better 
preserved material than was available to me, I have come to 
the conclusion that these authors are right, and I am prepared 
to accept their view. A re-examination of my own prepara- 
tions leads me to believe that the septum which I believed to 
