us, since they have sprung
from the same stock and have developed upon the same lines; they are
therefore monogenetic: but since we know that they have become what they
are independently of each other (now unlike any other Rails), they are
polygenetic and therefore could not form one genus in the old Darwinian
sense. Further, they are not a case of convergence, since their ancestry
is not divergent but leads into the same stratum.

THE RECONSTRUCTION OF THE GEOGRAPHY OF SUCCESSIVE EPOCHS.

A promising method is the study by the specialist of a large, widely
distributed group of animals from an evolutionary point of view.
Good examples of this method are afforded by A.E. Ortmann's ("The
geographical distribution of Freshwater Decapods and its bearing upon
ancient geography", "Proc. Amer. Phil. Soc." Vol. 41, 1902.) exhaustive
paper and by A.W. Grabau's "Phylogeny of Fusus and its Allies"
("Smithsonian Misc. Coll." 44, 1904.) After many important groups of
animals have been treated in this way--as yet sparingly attempted--the
results as to hypothetical land-connections etc. are sure to be
corrective and supplementary, and their problems will be solved, since
they are not imaginary.

The same problems are attacked, in the reverse way, by starting with the
whole fauna of a country and thence, so to speak, letting the research
radiate. Some groups will be considered as autochthonous, others as
immigrants, and the directions followed by them will be inquired into;
the search may lead far and in various directions, and by comparison of
results, by making compound maps, certain routes will assume definite
shape, and if they lead across straits and seas they are warrants to
search for land-connections in the past. (A fair sample of this method
is C.H. Eigenmann's "The Freshwater Fishes of South and Middle America",
"Popular Science Monthly", Vol. 68, 1906.) There are now not a few maps
purporting to show the outlines of land and water at various epochs.
Many of these attempts do not tally with each other, owing to the
lamentable deficiencies of geological and fossil data, but the bolder
the hypothetical outlines are drawn, the better, and this is preferable
to the insertion of bays and similar detail which give such maps a
fallacious look of certainty where none exists. Moreover it must be
borne in mind that, when we draw a broad continental belt across an
ocean, this belt need never have existed in its entirety at any one
ti