ON THE PRIMARY DIVISIONS OF CIRCLES. 299 
the aberrant one always being so much diversified, that 
it wears the appearance of being three, making the 
number five. Thus, for instance, the class of birds 
contains three primary groups; but the aberrant one is 
so large and varied, that we are accustomed, for the sake 
of perspicuity, to divide it into three ; nainely, the 
Rasores, the Grailatores, and the Natatores. 
(280.) The difference of considering a natural group 
as divisible into three, instead of five, does not, in the 
least, affect the natural series by which they are united. 
The discovery of the union of Mr. MacLeay’s three aber- 
rant groups into a circle of their own, is the addition 
only of a property superadded to that which they were 
known to possess ; this property consisting of uniting 
into a circle among themselves, as well as passing into 
the typical and the sub-typical groups. It is, however, 
a distinction to be kept in mind, since without it 
we should be unable to substantiate that wniformity 
of plan which embraces every natural group, and 
renders them but types of higher assemblages. The 
first divisions of matter, or natural bodies, are obviously 
three, animals, vegetables, and minerals; and this 
number coincides with that found in the primary 
divisions of animals, and in all their inferior groups. 
This, of itself, is strong presumptive and analogical 
evidence. If, on the other hand, every natural group 
was first resolvable into five, then, to support the theory 
of perfect uniformity in creation, we must show that 
there are five primary divisions of natural bodies ; 
a division which no one has ventured to point out. 
The plan of nature implies perfect harmony and 
uniformity, not only in generals but particulars. All 
that is yet known by analysis confirms this theoretical 
conclusion ; and this, independent of any other con- 
sideration, is conclusive against the idea that there 
should be only three primary circles in some groups, and 
five or seven in others. 
(281.) It has been observed, however, that, in groups 
termed imperfect, some of the links of connection are 
Qa 
Qa 
