AMPHIOXUS LANCEOLATUS. 
389 
them of absolute alcohol, which shrinks up the metapleura and 
obliterates the space. Johann Muller thought that they opened 
anteriorly, each by a pore, but it is admitted now that no such 
pores exist. 
The most important fact about the metapleura which has 
been hitherto overlooked is that their space is abolished, and 
their very existence as upstanding longitudinal ridges ceases 
when the gonads attain their full size at the breeding season. 
The stretching of the epipleural wall leads to a complete flat- 
tening of the metapleura, as shown in PI. XXXV, fig. 4. It 
seems not improbable that the albuminous fluid contained in 
the metapleural canals may serve as a final supply of nutriment 
for the enlarging gonads. 
Were there any “ ventral canals ” such as have been sup- 
posed to exist by nearly all writers on Amphioxus, this would 
be the place, viz. in association with the metapleural canals, in 
which to discuss them. Stieda’s specimens and figures showed 
the whole of the epithelium of the plaited ventral wall of the 
atrium “ blistered ” or raised from the subjacent connective 
tissue. Accordingly he described the existence of a pair of 
large ventral canals lying right and left between the two meta- 
pleura. Eolph, Langerhans, and Schneider, described not 
a single pair of canals but a number running parallel to 
and corresponding with the longitudinal ridges of the surface. 
The spaces which are frequently seen in this position are 
really between the connective tissue and the epidermis and 
are due to differential shrinking. Eolph indicates canals 
below the layer of connective tissue in this position, that 
is, between the ventral transverse muscle and the connective 
tissue. It appears to me that no such canals exist. The 
insertion of the fibres of the transverse muscle into the con- 
nective tissue, and the excessive contraction of the muscle 
under the influence of reagents, causes a deep plaiting of the 
connective tissue and a tearing and separation of natural 
adhesions in most specimens. But in such a preparation as 
that drawn in (PI. XXXVI A, fig. 2,) we see that there is no 
splitting of the connective tissue in the median ventral area 
