AMPHIOXUS LANCEOLATUS. 
367 
conclude are actually maintained in the living state. I do not 
know of any one reagent which gives equally good results for 
all parts of the Amphioxus body. I have found it necessary to 
study specimens preserved in several different ways. 
Grooves of the Ventral Wall. — Even when somewhat dis- 
tended; as shown in PI. XXXIV, fig. 3, the ventral wall of the 
atrium of Amphioxus exhibits longitudinal plaiting. These 
folds have been observed and counted on living specimens at 
my request, by my friend Balfour, and by others who have 
had the opportunity of studying living Amphioxus at Naples, 
since I was there in 1875. And there is no doubt that they 
are not “ artifacts/’ but exist iu the living state, though 
their depth is increased by the unnatural contraction caused 
by preserving fluids. They are best seen in the ventral view 
of a living specimen given in PI. XXXIV, fig. 4. As far 
as I can ascertain they do not vary in number in the same 
individual, except as the result of the general increase of the 
animal’s size. All the folds do not extend the whole length of 
the ventral surface : some stop short anteriorly. I have 
counted from six to eight on each side of the middle line. 
They entirely disappear when the ventral wall of the atrium is 
fully stretched, as it is when the generative products are full 
grown and ready for extrusion (see PI. XXXV, fig. 4). Their 
production is accounted for by the insertion of some of the 
fibres of the ventral transverse muscle into the somewhat 
thick tegumentary connective tissue, in a series of lines cor- 
responding to the grooves. This insertion can be readily 
observed in thin transverse sections (see PI. XXXVI.4, fig. 2). 
Absence of Canals below the Ventral Plaited Integument. — 
The epidermis supported by a fine basement membrane fre- 
quently becomes separated by the action of reagents from the 
thick subjacent connective tissue of this region, and has given 
rise to an erroneous conclusion, to the effect that thei’e is a 
series of ventral canals underlying the plaited epithelium. 
Or, on the other hand, the muscle becomes separated from 
the deeper layer of the cutis, and a similar mistake has arisen. 
This error is made by Stieda (4), Rolph (3), Langerhans (5), 
