which may be considered to give the size of the bird), from the
shortest to the longest, and the same number of vertical lines are
drawn, numbered from one to twenty. In this case (and wherever
practicable) the body-length is measured from the lower line of the
diagram, so that the actual length of the bird is exhibited as well as
the actual variations of length. These can be well estimated by means of
the horizontal line drawn at the mean between the two extremes, and it
will be seen that one-fifth of the total number of specimens taken on
either side exhibits a very large amount of variation, which would of
course be very much greater if a hundred or more specimens were
compared. The lengths of the wing, tail, and other parts are then laid
down, and the diagram thus exhibits at a glance the comparative
variation of these parts in every specimen as well as the actual amount
of variation in the twenty specimens; and we are thus enabled to arrive
at some important conclusions.

We note, first, that the variations of none of the parts follow the
variations of the body, but are sometimes almost in an opposite
direction. Thus the longest wing corresponds to a rather small body, the
longest tail to a medium body, while the longest leg and toes belong to
only a moderately large body. Again, even related parts do not
constantly vary together but present many instances of independent
variation, as shown by the want of parallelism in their respective
variation-lines. In No. 5 (see Fig. 4) the wing is very long, the tail
moderately so; while in No. 6 the wing is much shorter while the tail is
considerably longer. The tarsus presents comparatively little variation;
and although the three toes may be said to vary in general together,
there are many divergencies; thus, in passing from No. 9 to No. 10, the
outer toe becomes longer, while the hind toe becomes considerably
shorter; while in Nos. 3 and 4 the middle toe varies in an opposite way
to the outer and the hind toes.

[Illustration: FIG. 6.--Cardinalis virginianus. 31 Males.]

In the next diagram (Fig. 5) we have the variations in forty males of
the Red-winged Blackbird (Agelaeus phoeniceus), and here we see the same
general features. One-fifth of the whole number of specimens offer a
large amount of variation either below or above the mean; while the
wings, tail, and head vary quite independently of the body. The wing and
tail too, though showing some amount of correlated va