conclusion is announced, is well worthy of attention from American physiographers.

The principle of leveling without baseleveling, or Passarge's law, as it may be called, in contrast to Powell's law of leveling by baseleveling, suggests that the scheme of the normal cycle of erosion, so generally applicable in regions of ordinary or normal climate, should be systematically modified in such ways as will adapt it to the conditions of an abnormally dry or arid climate. This modification I have lately attempted in an article that will soon be published in the Journal of Geology; it is here presented in outline.

An extensive region of any structure uplifted in an arid zone to any altitude and with any form will, in the youthful stage of its cycle of erosion, be characterized by as many independent and incomplete centripetal drainage systems as there are depressed areas or basins within its limits: independent systems, because in an arid climate the basins can not be filled with overflowing lakes; incomplete systems, because many of the intermittent centripetal streams will wither away on the slopes and fail to join forces in trunk streams on the basin floors. The early stage of a normal cycle, where all basins are filled to overflowing and where all streams are continued until they unite in trunk rivers which reach the sea, is characterized by a rapid increase of relief, due to the incision of valleys. The early stage of the arid cycle is, on the other hand, characterized by a decrease of relief, due to the aggradation of the basins with the waste washed down from the enclosing highlands. As youth advances towards maturity, the initially independent basins will become more and more completely confluent, either by headward erosion on the slopes of the lower basins, or by the overflow of waste across depressions in the borders of the higher basins; thus from original independence will be developed a maturely integrated and interdependent system of drainage slopes, although trunk rivers will still be wanting. Maturity may be said to be fully established when large areas are thus brought into systematic correlation. At this stage, there may still be some unreduced uplands, but there will also be in

creasing piedmont areas of degraded, rockfloored plains, inclined gently towards the greatly enlarged central aggraded basin floor; and the composite plain thus produced will have no definite relation to normal baselevel. In so far as the erosion of arid regions has previously been discussed, it would appear that this stage, here called maturity, has been regarded as the old age of a desert, and that it has been taken to mark the end of the changes to which an interior basin is subject, unless it is attacked by the headward extension of exterior streams and thus dissected and reduced to normal baselevel; but as Passarge clearly shows, the old age is yet to come, and with a systematic sequence and grouping of features essentially unlike those just described. The action of the wind is yet to be considered.

In the earlier stages of the cycle, while the slopes are still varied and strong, transportation and trituration by the wind is probably of small value in proportion to that of the occasional streams and floods. But as the barren surface becomes more and more even, the relative importance of wind action increases; for unlike running water, the wind does not depend on local slopes for its activity; it is about equally strong everywhere on a surface of moderate relief, and has no subdivision into subordinate parts that correspond to small headwater streams, whose slopes must be steeper than that of their trunk river. The wind may sweep sand along a level floor, or even up a moderate slope; and whirlwinds may raise dust high into the air, and there give it to the upper currents; both of these processes may carry desert waste outside of the desert region under consideration, and thus the mean level of the desert may be very slowly reduced. The surface may, indeed, in this way eventually be worn below sealevel, as several writers have suggested; but the form that the surface will exhibit during its slow reduction has not, to my knowledge, been espe cially considered until in the recent statement of this aspect of the question by Passarge.

It might at first sight appear that when the winds gain the upper hand in the processes of transportation, they would tend to excavate extensive basins wherever the weathering of

the rocks resulted in the production of fine dusty waste; and that, inasmuch as the winds know no baselevel, there would be no definitely assignable limit to the unevenness of the surface thus produced. This might be true in absolutely rainless regions; but such regions are not known. The most desert regions of the world have occasional rainfall, and are from time to time visited by showers heavy enough to cause floods; and the intermittent action of such floods will put an effectual stop to the development of deep basins by wind action. As soon as the winds succeed in sweeping out a shallow depression, that part of the integrated drainage slopes which leads toward the depression will, when rain falls and floods are formed, provide a supply of waste with which the depression will be aggraded. Further deepening of the depression below its surroundings is thus effectually hindered. The wind may then begin the excavation of another depression elsewhere, only again to be defeated by the local inwash of a waste cover. Not an uneven surface of many hills and hollows, but a remarkably even plain must result from the long continuance of these antagonistic processes.

During the development of such a plain, a series of systematically irregular changes will run their course. As the exportation of desert waste by the winds continues, the area of the central aggraded basin floor must diminish, while that of the surrounding degraded rock plains must increase. At the same time, the integrated drainage system of maturity will be more and more completely disintegrated and replaced by many local and variable systems of extremely indefinite separation. Eventually all the central accumulation of waste will have been exported by the winds; the rock-floored plain will have been worn down lower than the bottom of the deepest initial depression, so that it will then extend throughout the region, except for residual mountains of rocks most resistant to dry weathering. Thin veneers of gravelly waste will remain, swept hither and yon by the intermittent fluctuating disintegrated drainage; shallow 'saltpans' may occur from place to place and from time to time; but large areas of rock plains carrying

only scattered stony waste, will abound; this is the condition of true old age in such a region. Once attained, it persists, slowly worn lower and lower, possibly sinking below sealevel, until disturbed by crustal movements or climatic change. It is old rock-floored desert plains of this character and apparently of this origin that Passarge describes as occupying thousands of square miles in South Africa.

Two interesting consequences of this scheme should be pointed out.

Every truncated upland that has been described as an uplifted and more or less dissected peneplain should now be reexamined with the object of learning whether it may not have originated as a desert plain at its present altitude above sealevel, and afterwards suffered dissection as a result of climatic change. True, we are to-day more accustomed to movements of the earth crust, in the way of elevations and depressions, than to climatic changes, in the way of transforming arid regions to humid regions and vice versa; but perhaps this habit of thought is only a fashion of our time. A century ago, movements of the earth's crust indicated by the discovery of marine fossils on the higher peaks of the Alps were regarded with astonishment, not to say incredulity. A century hence, variations of climate may be accepted as freely as changes of level are now. The way towards such an opinion is opened by the discovery of glacial periods in various geological ages, and it is not hindered so much as it was once by supposed evidence of the correspondence of earlier climatic zones with those of to-day. We should, therefore, open our minds widely to the possibility of explaining truncated uplands as ancient desert plains not changed in elevation, but only in climate; and this possibility should not be set aside because it seems improbable, but only because it may be shown on good and sufficient grounds to be inappropriate to the case under consideration. It may be added that, as far as I have undertaken a revision of the origin of truncated uplands, as is suggested above, nearly all the familiar cases seem to possess characteristics that accord with their origin as uplifted peneplains and not as desert plains; and that there

is therefore less ground for change of generally accepted opinions than the suggestion of the need of revision might for the moment indicate.

The second consideration concerns the processes of combined induction and deduction by which the complete or logical method of scientific investigation is constituted. In view of the possible change of interpretation now open for truncated uplands according to Passarge's law, it might be said by one who prefers to work on more purely inductive lines: “Behold, here is another case in which deduction has led the investigator astray! He thought that he could deduce the sole conditions under which truncated uplands could be formed, and that these conditions necessitated uplift after degradation; now he finds a new series of conditions under which such uplands may be formed and all his previous conclusions are uncertain. Let us, therefore, beware of deductive or imaginative methods, and hold fast to the safer methods of observation and induction." In reply to such a warning, one might say-besides pointing out that all problems which deal with unseen processes necessarily involve deduction and that the deductive side of the work should be conscious and systematic --that the fault in the method by which truncated uplands have heretofore been discussed lies not in the too free use of deductive methods, but in their too limited use. The mistake lies in our not having years ago set forth, by purely deductive methods, just such an analysis of the geographical cycle in an arid climate as has now been provoked by the discovery of rock-floored desert plains. Such an analysis does not involve any new or difficult problems; it might have been successfully attempted long ago; the difficulty that stood in the way lay not in the problem itself, but rather in the habit among physical geographers of trusting too largely to observational methods and of neglecting the aid that deductive methods furnish. The lesson of the problem is, therefore, that deduction should be pushed forward more energetically and systematically than ever; always checking its results as far as possible by confronting them with the appropriate facts of observation, but

never halting in the reasonable extension of deductive conclusions because the corresponding facts of observation have not been detected; never lessening the activity with which exploration and observation are pursued, but always using the spur of deduction along the paths suggested by multiple working hypotheses.' The problem of the erosion of mountain valleys of Alpine glaciers teaches the same lesson: if physiographers had, thirty years ago, been well practised in deductive methods, they might have easily extended Playfair's law regarding the accordant junetion of branch and trunk streams from the case of stream surfaces to the contrasted case of stream beds, and from the case of water streams to the analogous case of ice streams; thus they might have predicted that, if Alpine glaciers were effective eroding agents, glaciated mountain valleys ought to show discordant or hanging side valleys; and in going to the mountains they would have found the prediction correct, and the basis of the predictionthat glaciers are effective eroding agentswould have thus been verified. So with the geographical cycle in an arid climate: there is nothing difficult in the series of deductions that lead to the expectation of rock-floored desert plains, independent of baselevel, as the product of arid erosion; the only obstacle to the development of these deductions has been the habit of not making them. This is a habit that should be broken.

W. M. DAVIS.

NOMENCLATORIAL TYPE SPECIMENS OF PLANT

SPECIES.

THE recent Code of Botanical Nomenclature' now usually known as the Philadelphia Code, states as the fourth fundamental principle, 'The application of a name is determined by reference to its nomenclatorial type.' This means that a specific (or subspecific) name stands or falls according to the disposi tion of the type specimen. It is not proposed here to discuss the advantages or disadvantages of this method of determining the application of names, although to the writer this method seems much more likely to secure 'stability, uniformity and convenience in the

designation of plants,' than the method of applying the name according to tradition, authority or consensus of opinion. Instead of this, then, it is proposed to discuss briefly the practical difficulties which may arise in this method of types, and how these difficulties may be overcome.

The code mentioned above states in regard to the application of names (Canon 14) the following: 'The nomenclatorial type of a species or subspecies is the specimen to which the describer originally applied the name in publication.'

Where an author in connection with an original description has indicated a definite specimen, there is usually no difficulty in determining the type. When an author indicates only the number or other data occurring on the label in numbered sets prepared for distribution, but does not specify a particular specimen, the type would be the one from which the author drew up the description and would presumably be in his herbarium. The other specimens would then be designated as duplicate types. Not infrequently the author draws the description from all the specimens of a given number in a set, in which case the specimen in the herbarium of the author, or of the institution at which he is located, must be arbitrarily chosen as the type.

Many difficulties arise in determining the types of the older authors, as the practise of designating specimens as such is quite recent. When a name is based upon a single specimen this becomes the type though not actually designated as such. If more than one specimen is cited, but none designated as the type it becomes necessary to select one of these.

The above mentioned code provides that 'When more than one specimen was originally cited, the type or group of specimens in which the type is included may be indicated by the derivation of the name from that of the collector, locality or host.' (Canon 14, a.) Further, if no type can be selected on this basis, 'Among specimens equally eligible, the type is that first figured with the original description, or in default of a figure, the first mentioned.' (Canon 14, b.)

There are many original descriptions, how

For example, if

ever, in which no specimens are cited, but instead the locality or range may be given. It then becomes necessary to consult the author's herbarium or the herbarium in which his plants are deposited. Specimens which bear the name in his handwriting should be given preference in the selection, and of these the type is the one from the locality first mentioned, or the one collected by the person for whom the species is named. Even with these aids in selection it may be necessary to arbitrarily select a certain specimen from among those equally eligible. This should be done by a monographer and only after a careful examination of the available data. Where possible the most perfect specimen should be selected or the one most nearly corresponding to the original description. the species is known to produce rhizomes and only one of the otherwise available specimens showed these organs, this specimen might be selected. Occasionally the original description includes more than one form and the specimens are correspondingly diverse. It is then very necessary to use particular care in the selection of the type. Muhlenberg described Panicum depauperatum without indicating a type. In his herbarium deposited in the Philadelphia Academy of Natural Sciences is the sheet of specimens upon which the name is founded. In this sheet are plants of P. linearifolium Scribn. and two forms of what is now considered to be P. depauperatum Muhl., one with glabrous sheaths and one with pilose sheaths. From the description one. can not determine which one of these forms was intended. Probably all were included as one species. Since the form with smaller spikelets has been distinguished by Professor Scribner as P. linearifolium the type of P. depauperatum should be selected from the specimens with large spikelets. When the two or more species confused by one author are distinguished by a later author, this author should determine the type. The old specific name should remain with the type and the new name be based upon a different type. Much confusion has arisen because of failure to follow this rule. If the original specimens are made up of both species, the author of the

later name, the so-called segregator, should indicate which specimen is the type of each species. Professor Scribner might with equal propriety have given the new name, in the case above mentioned, to the form with large spikelets, except for the fact that tradition, and the recorded history of the plant had attached the name P. depauperatum to this form. But, as stated, the original specimens are in part with glabrous sheaths and in part with pilose sheaths. The original description states that the sheaths are pilose. In a recent study of this collection in preparation of a monograph of the Panicums I took the liberty of selecting a specimen from the cover that had pilose sheaths, and attaching a ticket with such indication.

Let us consider another case and suppose that a reference to Muhlenberg's herbarium had shown only a specimen of P. linearifolium Scribn. In this case this specimen would become the type of the species P. depauperatum Muhl., since it agrees with Muhlenberg's description, and the species which had been called P. depauperatum would receive a new

name.

While it is true that the name of a species rests upon its type specimen, yet the specimen can not take precedence over the description. If it is clear that a supposed type specimen disagrees with the description to such an extent that it can not be the plant which the author describes, then the plant must be disregarded in determining the type.

In a previous paper I mentioned that the specimen in the Linnean Herbarium labeled in Linnæus's handwriting Agrostis rubra is a panicle of a Sporobolus, apparently Sporobolus juncea of our southern states. There is clearly an error here as the plant does not agree with the description. On the other hand, there are many cases in which the type specimen does not agree in all respects with the description. The sheaths may be described as glabrous when a few of the lower may be pubescent. If there is no reasonable doubt that the specimen was examined by the author and is the specimen or at least one of the specimens upon which the description was based, such specimen should be accepted as the type.

In cases where the first cited specimen is chosen as the type according to rule, it not infrequently happens that this is a form which does not represent faithfully the author's idea of the species. The specimens may have been arranged geographically and the first locality may be represented by a specimen of an aberrant or uncertain form. But the rule is explicit on this point and is certainly easy to interpret and follow.

Torrey and Gray publish many of Nuttall's manuscript names, but in listing specimens those collected by Nuttall may not be mentioned first; nevertheless, his specimens should be taken as the type by a broad interpretation of Canon 14, a. Cardamine hirsuta L. B acuminata Nutt. mss. in Torr. and Gray Fl. 1: 85. The specimens cited are: British America, Richardson; Oregon, Nuttall. The latter specimen should be taken as the type.

When there is no original specimen we must make use of Canon 14, c, in determining what shall serve as the type: 'In default of an original specimen, that represented by the identifiable figure or (in default of a figure) description first cited or subsequently published, shall serve as the type.' It sometimes happens that the citations will lead to a specimen, which then should be taken as the type. Poa flava L. is based upon a citation from Gronovius Flora Virginica, that is, Linnæus gives a specific name to a plant described by Gronovius. A reference to Gronovius shows that he mentions a particular specimen, Clayton No. 273, which plant is deposited in the herbarium of the British Museum and is the type of Poa flava L.

I will now refer briefly to a second series of cases, those where there has been only a change of name. If a species has been transferred from one genus to another the type specimen is determined according to the rules mentioned above, by a reference to the original description. If a new name is given to a species because the old one is untenable, the type of the old name becomes the type of the new. There are no new difficulties presented here, if there is no doubt that there has been only a change of name. However, one finds many cases where an author has