ANALOGICAL TESTS OF CIRCLES. 335 
naturalists who think that other proofs are unnecessary. 
But the verification of such an extensive group as that 
we have now instanced, namely, the Merulide, is by 
no means yet complete. 
(411.) The third test by which a circular group is to 
be verified, consists in its being in unison with the theory 
of variation: that is to say, each of its principal modi- 
fications of form are to follow each other according to a 
definite rule. This rule has been so fully and so clearly 
explained in the body of the work, that it seems hardly 
necessary to repeat it here: familiar examples, however, 
will render it more apparent to the student ; and these 
examples, —to prevent the recital of those exceptions 
which must be noted if we attempted to speak generally 
of the whole animal kingdom — will be drawn from some 
of the groups of ornitholog gy. We have shown that groups 
can be made to appear natural and circular, when in 
reality they are not so; and we have now to show that 
their erroneous composition cannot always be detected, 
even by comparing their contents with those of another 
established group, and thus testing them by the theory 
of parallel analogies. An ornithologist of this country, 
now retired from science, who has done much towards 
the determination of the leading families of birds, and 
who in many instances has shown great judgment in the 
location of the groups, has nevertheless been led into an 
arrangement of the shrike family (Zaniade Sw.) which 
exemplifies the error we are now speaking of: he disposes 
the groups of these birds in the following manner : — 
Genera. 
Typical. Lanius. True Shrikes. 
Sub-typical. Edolius. Drongo Shrikes. 
Tyrannus. Tyrant Shrikes. 
Aberrant. Ceblepyris. Caterpillar-Catchers. 
Thamnophilus. Bush Shrikes. 
(412.) These divisions, it has been said, form a circu- 
lar group, and each division follows in the order of succes- 
sion here stated: and as the bush shrikes, as every one 
knows, blend into the true shrikes, the circle is closed, and 
the whole has a verisimilitude of being truly natural. 
