5} This passage corresponds roughly to the conclusion of the
_Origin_, see Ed. i. p. 482, vi. p. 661.

{106} A similar passage occurs in the conclusion of the _Origin_,
Ed. i. p. 481, vi. p. 659.

Give sketch of the Past,--beginning with facts appearing hostile under
present knowledge,--then proceed to geograph. distribution,--order of
appearance,--affinities,--morphology &c., &c.

Our theory requires a very gradual introduction of new forms{107}, and
extermination of the old (to which we shall revert). The extermination
of old may sometimes be rapid, but never the introduction. In the groups
descended from common parent, our theory requires a perfect gradation
not differing more than breed<s> of cattle, or potatoes, or cabbages in
forms. I do not mean that a graduated series of animals must have
existed, intermediate between horse, mouse, tapir{108}, elephant [or
fowl and peacock], but that these must have had a common parent, and
between horse and this <?> parent &c., &c., but the common parent may
possibly have differed more from either than the two do now from each
other. Now what evidence of this is there? So perfect gradation in some
departments, that some naturalists have thought that in some large
divisions, if all existing forms were collected, a near approach to
perfect gradation would be made. But such a notion is preposterous with
respect to all, but evidently so with mammals. Other naturalists have
thought this would be so if all the specimens entombed in the strata
were collected{109}. I conceive there is no probability whatever of
this; nevertheless it is certain all the numerous fossil forms fall
in<to>, as Buckland remarks, _not_ present classes, families and genera,
they fall between them: so is it with new discoveries of existing forms.
Most ancient fossils, that is most separated <by> space of time, are most
apt to fall between the classes--(but organisms from those countries
most separated by space also fall between the classes <_e.g._>
Ornithorhyncus?). As far as geological discoveries <go> they tend towards
such gradation{110}. Illustrate it with net. Toxodon,--tibia and
fibula,--dog and otter,--but so utterly improbable is <it>, in _ex. gr._
Pachydermata, to compose series as perfect as cattle, that if, as many
geologists seem to infer, each separate formation presents even an
approach to a consecutive history, my theory must be given up. Even if
it were consecutive, it would onl