246 PROCEEDINGS: BOSTON SOCIETY NATURAL HISTORY. 
buried within the pupal thorax until near the end of the pupal 
period, when it evaginates and forms the completed head of the 
insect. This evagination has been observed by Weismann (’64, 
p. 259) to be the consequence of the pressure of blood, which at the 
right moment is forced from the abdomen into the thorax and pushes 
the head vesicle forward. 
The metamorphosis of the thorax goes on simultaneously with 
the formation of the head vesicle. In proportion as the larval 
hypodermis disappears under the attacks of the phagocytes, as 
already mentioned, the edges of the imaginal discs grow and 
take its place, forming the imaginal hypodermis. As we have 
seen, there are six pairs of these discs, a dorsal and a ventral 
pair being present in each thoracic segment. They lie at a dis¬ 
tance from the integument, near the centre of the larva. Each 
disc is, however, connected by a very line hollow cord with that 
portion of the hypodermis of the segment to which it genetically 
belongs, and where it is destined to appear as an extremity. This 
cord, first discovered by Ktinckel dTIerculais, gradually shortens, 
and its lumen enlarges. The disc is thus brought nearer the 
surface, and as it advances it increases in size. The lumen of the 
cord then opens through the hypodermis, and the cord itself fin¬ 
ally becomes so wide and short that the disc is brought through 
the hypodermis to the outside. The hollow cord is of course 
obliterated by this process, and the edges of the proximal end 
of the disc are brought into direct contact with the hypodermis. 
The disc has now assumed its position as an extremity. It is an 
aj^pendage of the body-wall; it has become irregularly cylindrical 
in shape, and possessed of a number of constrictions and folds, 
which in the case of the ventral discs are equivalent to the joints of 
the future leg. 
The metamorphosis of the abdomen is retarded; it does not 
begin until that of the head and thorax is well advanced. Then 
in each abdominal segment the two ventral and four dorsal discs 
(Van Rees found two additional dorsal discs in each segment) 
begin to grow and take the place of the disappearing larval 
hypodermis. 
In a paper of my own (’93), which contains the results of a study 
of the larva of Melophagus ovinus, is contained a full description 
of the imaginal discs of this insect. Melophagus, being a pupipar, 
and closely allied to the muscids, we should expect to find the same 
