DIFFERENT RANK OF GROUPS. 269 
tected in nature : for although, when a sub-genus is very 
perfect, it sometimes contains the five types of form 
common to aii circular groups ; yet, as we have just 
observed, no instance has yet been pointed out, wherein 
each of these types is also circular. 
(329.) It is clearly impossible to define any of these 
groups by characters which are applicable to all such as 
hold the same rank; nor can their value be known by 
any other rules than those resulting from analysis and 
comparison. The characters which belong to a family 
in one tribe, are totally different from those which cha- 
racterise a family in the next; while such as are 
exhibited in a third, will be very different from either. 
Nor can we tell, at first sight, the difference between a 
tribe and a family ; or whether any particular form is 
the representation of a genus or a sub-genus. The 
true rank of a natural group, in short, can only be de- 
tected by analysis and analogy ; and the more extensively 
these enquiries are carried into the neighbouring groups, 
the more likely are we to understand its true rank. But 
as this mode of investigation is not only laborious, but 
too difficult to be extensively prosecuted, it is the cus- 
tom with most writers to throw several genera into a 
group, and call that group a family, or a sub-family. 
This is all very well, and really useful, if it be consi- 
dered, as it truly is, but a temporary expedient, — a 
mode of abridging labour, by assuming what has not 
been proved, and pointing out to the reader the most 
probable station of the group or species under his con- 
sideration. But no faith can be placed in such tables or 
scales of gradation*, until their circular arrangements 
and analogies have been made out by sieves. We 
shall now peed to make some general remarks upon 
these groups. 
(330.) The common consent of mankind has sane- 
tioned the belief in the three kingdoms of nature,— the 
* Such, for instance, as that in the Introduction to Entomology, 
vol. iv. p. ” 393. 
