passed up through the walls of the vessels, or if this be too strong a statement we are at least safe in saying that not enough passed up the walls to serve even most inadequately, the transpiration purposes of a single leaf, and this, perhaps, is the better forins in which to leave the statement. Neither can it be said, by way of objection that the eosine behaved differently from ordinary water for we have already seen that 1 per cent eosine water passes up unplugged stems readily, even for days and long after the stem is dead. In this case, judging from the state of the atmosphere, the temperature, and the amount of transpiring surface, (approximately 1,500 sq. cm.) at least 25 and probably 50 cubic centimeters would have been taken up by the plant in the first 24 hours but for the gelatine plugs. No explanation is open, therefore, except that the transpiration water passes up through the lumen of the vessels in the stem of the cucumber, and presumably in all other stems of similar structure, unless we assume that the gelatine passed into the walls of the vessels and destroyed their conductive power, and no one has proved this to be possible or even set forth facts rendering it probable. Considering the fact that the walls of the vessels. in most plants are solid lignified structures and that the vessels are long open tubes, comparable to water pipes and in many plants probably continuous through the whole length of the stem, it would seem strange that this other view, viz. that the water passes upward through the wall itself and not through the lumen of the tube, should have ever gained credence, did we not know how often, even in science, the weight of a great name carries everything before it.

(To be Continued.)

EXTENSIVE MIGRATION IN BIRDS AS A CHECK UPON THE PRODUCTION OF GEOGRAPHICAL

VARIETIES.

BY THOMAS H. MONTGOMERY, JR.

Two problems have of late years received much attention from ornithologists, and deservedly, namely, the faunal distribution of species, and their ranges of migration. But to my knowledge, no one has raised the question of the possible existence of a relation between the extent of the periodic migration and the amount of geographical variation evinced by a species. The object of this paper then is to show that such a relation does exist, that extensive migration tends to act as a check upon the production of geographical varieties, or races so-called.

In the first place, on comparing the amount of faunal variation with the extent of the periodical migrations in a given species, it will be found to be usually, if not always, the case, that those species which undertake migrations of more than the average extent-migrating through 30° or more of latitude, have no tendency to give rise to geographical varieties. In order to see how far this law extends, and whether exceptions to it may be found, I have compared all the species of North American birds with regard to this relation existing between variation and range of migration, using Ridgway's excellent "Manual of N. A. Birds" as my authority for the amount of variation and extent of migration in these species. For our present purposes the North American species may be divided into three groups, based on the extent of their migrations (1) species with exceedingly protracted migrations, but irregular as to the localities traversed; (2) species with more or less regular migrations, of 30° lat. or more in extent; and (3) species which undertake migrations less in extent than 30° lat., or species which do not migrate at all. We may now consider each of these groups in turn, with regard to the question at issue.

I. Species with protracted but irregular migrations.

Diomedeida.

Procellariida.

Phaethontida (?)
Fregatida (?)

II. Species with a migration range of 30° lat. or more.

Podicipida.

Urinatorida (all?)

Stercorariidæ.

Larida (most).

Sulida (most?).

Anatida

Gruidæ.

Rallus virginianus.
Porzana.

Fulica americana.
Phalaropodidæ.
Recurvirostridæ.
Scolopacida (most).
Charadriida (most).
Aphrizida.
Columbida (most).
Cathartida (?).
Circus hudsonius.
Accipiter velox.

Falco (2 species).
Ceryle alcyon.

Sphyrapicus varius.

Micropodidæ.

Trochilus colubris.

Now the species enumerated in Lists I and II migrate periodically through an area of 30° lat., or more, that is, a migration range of considerable extent, and, with a few exceptions to be considered later, all are sharply defined species, and even though the breeding areas of most are very broad, none of them have a tendency to split into geographical varieties. Accordingly there must be be some relation existing between

the range of migration and the tendency to produce geographical races, for otherwise this coincidence could not be explained. So having found that those species undertaking long migrations do not, as a rule, tend to give rise to local varieties, we must conclude that the process of taking extensive migrations is a check upon the tendency to produce geographical varieties. But in order to round off further deductions, we must first determine whether species which do not migrate extensively have a greater tendency to geographical variation than those just considered; and this assumption will be strengthened by a comparison of the species in the following List III with those in Lists I and II.

III. Species with a short or no migration range.

It is at once apparent that almost all the species of North. American birds which are divisible into geographical varieties are classed in this third list, that is, that those species evincing the greatest tendency to geographical variation, are also those which undertake migrations of the least extent. Thus, for instance, Melospiza fasciata is usually resident in most localities throughout the whole year, and has become differentiated into a number of geographical races, while Melospiza georgiana is migratory, and though it breeds in an area nearly equal in extent to that of fasciata, has not produced local varieties; the non-migratory Megascops asio shows great geographical variation, while the migratory Asio acciptrinus, though almost cosmopolitan in its breeding area, shows no tendency toward such variation. And, in fact, an examination and comparison of List III with Lists I and II, will lead to the conclusion, that given any two species of equally extensive breeding areas, the one with the smaller range of periodic migration will, as a rule, evince a greater tendency to produce geographical varieties than will the species with the greater range of migration. This conclusion may be concisely formulated as follows; it is the rule that the amount of geographical variation in species with more or less extensive breeding areas, stands in inverse ratio to the extent of its periodic migrations. Naturally, this law is only applicable to species with extended breeding areas, since diverse conditions in different sections of this area are necessary, according to the theory of Natural Selection, for the production of geographical subspecies or varieties; and in a limited breeding area, throughout which the conditions of the environment are similar, there could be no cause to produce geographical varieties, irrespective of the migratory or nonmigratory habits of the species.

I have not meant to imply, in the preceding pages, that species with migration ranges of 30° lat., or more, are all sharply definable, i. e., that such species are never divisible into geographical varieties; but, on the contrary, that this tendency to produce geographical races is less in the species with extensive migrations, than in those with shorter ranges of migration. For it is usual, even in species with extensive migrations, whose