12 
Journal of Agricultural Research voi. xxviii, no. i 
molt of the larva, but the linings of the frontal sacs are not affected by the molts 
any more than are the linings of the peripodal sacs of the leg and wing buds. 
This indicates that the dorsal pouch of the atrium is a secondary invagination, 
and that its wings are not parts of the true frontal sacs which contain the buds 
of the antennae and eyes. This idea is suggested also by Pratt’s finding that 
the median root of the sacs is formed after the invagination of the original buds. 
Both the median slit and the subsequent prolongations from it that carry the 
original bud sacs inward are apparently, then, to be regarded as the rudiments 
of the dorsal pouch of the atrium in which the wing plates of the larval pharynx 
are later developed. 
The evolution of the frontal sacs and the involution of the larval head may be 
traced through the various groups of the Diptera. In the simplest forms, such 
as Simulium and Bibiocephala, studied by Kellogg (19), the head and mouth parts 
of the pupa are formed exactly within the corresponding parts of the larva, 
except that the larger antennae of the pupa are forced to back up under the larval 
cuticle to find space for their growth. In Corethra, as described by Weismann 
(54) t the pupal antennae are developed in deeply invaginated pouches, but the 
compound eyes are formed flush with the surface, immediately beneath the larval 
cuticle. In Chironomus, as shown by Miall and Hammond (34), both the 
antennae and the compound eyes are developed in pouches formed by infoldings 
along the sides of the front. This gives the first example’of a well-formed frontal 
sac in the Nematocera (fig. 3, A). 
In the Brachycera, as described by Becker (3) and mentioned by Wahl (51), 
Stratiomys and Antherix have frontal sacs which open on the upper, external 
surface of the head, in Stratiomys even well back from the anterior end. Like¬ 
wise, in Lonchoptera, as described by De Meijere (31), the frontal sac opens on 
the outside of the head, there being no atrium present. These brachycerous 
forms, therefore, present the same condition of head development as is found in 
the Nematocera, and De Meijere designates them as the “Anatria.” In Corethra 
(Weismann) and Chironomus (Miall and Hammond) the imaginal buds, including 
those of the head, are not formed till the last larval stage. The time of their 
origin in other Orthorrhapha is not known. 
It is only in the Cyclorrhapha, so far as known at present, that the frontal 
sacs become drawn into the interior of the body by the involution of the larval 
head, with the consequent formation of a new anterior section of the larval 
pharynx, the head atrium. Such forms are the “Atriata” of De Meijere. In the 
higher families of this group (Muscidae, Hippoboscidae) all the imaginal buds 
begin their development in the embryo, suggesting that the high degree of 
development they have attained has forced them to this precocious start. 
Thus we see that, as complicated as the head structure may be in the larvae of 
the higher flies, it has reached this extreme phase by a gradual evolution from a 
simple condition such as that which prevails in the Neuroptera, where all pupal 
parts are formed inside the corresponding larval parts, as illustrated by Kellogg 
(19) in studies of Corydalis cornuta L. 
Among the Orthorrhapha the head problem is further complicated by the 
fact that the larvae show also various degrees of head retraction. Holmgren (14) 
and Becker (5) have attempted to explain on this basis the modifications in the 
Cyclorrhapha. But the structure of the parts in this group can not be inter¬ 
preted as the results of a retraction of the larval head, and the idea has been 
refuted by all who have studied the embryonic development. Still, while the 
alteration of the head is described as an “ involution,” Pratt (41) has shown that 
what really takes place in the embryo of Melophagus is a forward growth of a 
circular fold of the hypoderm from the rear part of the head (fig. 3, A, x, x) 
which eventually forms the lips of the new mouth (fig. 3, B, x, x). Processes of 
growth should probably not be described by such mechanical terms as either 
