rom which no mammal has been
obtained, although both in the Wealden and the Upper Chalk in Europe,
and in the Upper Cretaceous deposits of the United States an abundant
and well-preserved terrestrial flora has been discovered. Why no mammals
have left their remains here it is impossible to say. We can only
suppose that the limited areas in which land plants have been so
abundantly preserved, did not present the conditions which are needed
for the fossilisation and preservation of mammalian remains.

When we come to the Tertiary formation, we find mammals in abundance;
but a wonderful change has taken place. The obscure early types have
disappeared, and we discover in their place a whole series of forms
belonging to existing orders, and even sometimes to existing families.
Thus, in the Eocene we have remains of the opossum family; bats
apparently belonging to living genera; rodents allied to the South
American cavies and to dormice and squirrels; hoofed animals belonging
to the odd-toed and even-toed groups; and ancestral forms of cats,
civets, dogs, with a number of more generalised forms of carnivora.
Besides these there are whales, lemurs, and many strange ancestral forms
of proboscidea.[197]

The great diversity of forms and structures at so remote an epoch would
require for their development an amount of time, which, judging by the
changes that have occurred in other groups, would carry us back far into
the Mesozoic period. In order to understand why we have no record of
these changes in any part of the world, we must fall back upon some such
supposition as we made in the case of the dicotyledonous plants.
Perhaps, indeed, the two cases are really connected, and the upland
regions of the primeval world, which saw the development of our higher
vegetation, may have also afforded the theatre for the gradual
development of the varied mammalian types which surprise us by their
sudden appearance in Tertiary times.

[Illustration: GEOLOGICAL DISTRIBUTION OF MAMMALIA.]

Notwithstanding these irregularities and gaps in the record, the
accompanying table, summarising our actual knowledge of the geological
distribution of the five classes of vertebrata, exhibits a steady
progression from lower to higher types, excepting only the deficiency in
the bird record which is easily explained. The comparative perfection of
type in which each of these classes first appears, renders it certain
that the origin of each and all of them