128 THE AMERICAN NATURALIST. [VoL. XXXV. 
the placental allantois of Perameles from one of non-placental 
type, like that of Phascolarctos, the reverse procedure is the 
more natural one, because Perameles is of more archaic aspect 
than Phascolarctos, and the latter is much more retrogressive 
in dentition ; that it is entirely improbable that such a struc- 
ture as an allantoic placenta should have developed twice- 
independently in the Mammalia (Ze. independently in Pera- 
meles) On these grounds he would regard the placental con- 
dition of the allantois as the primitive one for the Marsupialia 
and the non-placental allantois of most existing forms as 
secondarily derived by retrogression. 
That the typical condition of the allantois has arisen by 
reduction, and is therefore secondary, naturally admits of no 
reasonable doubt, for no one would suppose that the allantois 
would be incipient in animals which have passed through an 
oviparous phase. That the evidence derivable from the organi- 
zation of the Marsupialia may be interpreted as proving a 
reduction of the allantois from a placental to a non-placental 
condition is, however, open. to objection, both on general and 
on special grounds. In opposition we may maintain that there 
is no possible, and at the same time legitimate, way of assuming 
retrogression of the allantois on a basis of retrogression in 
other structures, especially those of adaptive character. 
On general grounds, changes in the embryonic membranes 
may not be closely correlated with adaptive changes (progres- 
sive or otherwise) in the general organization. For example, 
the modifications of the placenta in placental mammals are not 
directly related to the adaptive (ordinal) characters of the 
animals in which they occur; several modifications may occur 
in the same group (e.g. Edentata), a condition which may 
possibly be due, as Hubrecht ! has suggested, to the greater 
youth of the placenta as compared with other structures of the 
animal organization, but which is more probably due to the 
fact that the allantoic placenta is not directly affected by 
the various evolutionary factors which determine the appear- 
ance of adaptive characters in the animal. 
But even admitting such a correlation, retrogression of 
! Hubrecht (89, p. 588). 
