The reader will observe from the heading of this section, however, that we still persist in begging the question whether geography is a science of any kind. Finch has considered this question directly and, to my mind, effectively [223]. If an issue is made of this point, it commonly resolves itself into a debate over the meaning of a particular word, a word that is not essential to our further thought. In particular we feet justified in ignoring the various undefined concepts of the term "science" upon which many recent methodological discussions have been based. It is a common, but nonetheless naive and erroneous, assumption of many students of the natural sciences that the nature of "science" is sufficiently well known as to require no statement. When they do attempt to state it, their definition commonly excludes the social studies without indicating in what other kind of knowledge these are to be included. But many physicists and chemists define "science" in terms that could not include zoology or geology; I have even heard such a definition stated by an eminent geologist who did not appear to realize what a small part of his field could be included under it.

Whatever may be gained from the discussion of such a question, there is no need for it here. One would gladly avoid the problem by using some other term possessing less emotional connotation than "science" seems to have acquired for some of our colleagues in other fields, if another suitable word were at hand. In a recent public address, a renowned natural scientist, whose position as university president requires him to consider the problems of the social fields, declared that the "scientific approach" was not applicable in the social studies and regretted that we had no word like the German word Wissenschaft to apply to such studies; but he failed to note that German students seem to get along fairly satisfactorily without any word corresponding to our word "science"--that is, other than the word "Wissenschaft."

If we were to use the word in a definitive sense, it would be necessary to attempt to determine its meaning specifically. But our need here is simply for a convenient handle to apply to that general form of knowledge that is distinct from either common sense knowledge or from artistic and intuitive knowledge. For that form of knowledge our language provides only the word "science" and we will therefore use it in that sense without, for the moment, wishing to claim that geography is entitled to any of the distinctions granted to "science" as used in some other, undefined sense. At most, one may add, any such claims influence only those who make them; no matter how logically we might demonstrate in theory that geography should be granted a title, the esteem that presumably goes with that title will be granted only in recognition of more solid contributions.

Though we consider, then, any question of titles as of little importance, it is of very great importance for us as geographers to know what sort of a study geography is. We have noted many disagreements among geographers that have been produced by the desire among some to make geography into a certain kind of science--the only kind, perhaps, that they would call by that name. But any efforts that require geography to change its essential character must be in vain; we cannot make over geography in any fundamental way, we can only fulfill that which it has been and is. Ignoring, therefore, any question as to what geography should be, let us consider what kind of a study it is.

Consequently, it represents a distortion of science to attempt to arrange its parts in any simple system of classification, such as that which recognizes the natural and social sciences as quite separate groups within each of which various classes of individual sciences are distinguished. "All knowledge of the inorganic, organic, and human world is one interlaced whole," as Heiderich has emphasized [153, 212]. Only the fact that this whole of science is far too much for any one person requires that it be divided into more or less conventional branches, and the necessities of academic organization may require that these be grouped in major orders. This conventional grouping, however, proves in many cases to be anything but convenient. Since geography, in particular, must examine phenomena in the actual complexes in which they are found, it is impossible for it, in practice, to separate natural and human phenomena.

When we consider geography, in this particular aspect, in comparison with the single unity of all science, rather than in comparison with any other particular branch of science, the charge that geography is dualistic because it includes both human and non-human phenomena has no weight. As Penck comments, "a dualism is felt only by a person who sees boundaries rather than zones of contact between the sciences, who emphasizes the differences between the social and the natural sciences more than the interconnection of all sciences, their belonging together in one great unit science. The divisions of that unit science do not lie beside each other like the lands on a map. They stand in manifold relations with each other" [162, 41].

Almost all modern geographers are agreed that geography cannot adapt itself to the conventional division between natural and social studies; not only does geography as a whole fit into neither group, but neither can it be divided into two halves, natural and human. It is not the position of geography, however, that is illogical: the separation of things natural from things human is possible only in theory, in reality they are interwoven. Geography, like psychology, is evidence of the arbitrary character of the conventional division of science.

To be sure, there are geographers who assert that they are interested primarily in "the physical aspects of geography," but one will look hard and long to find any of them who do not contribute published studies involving human aspects of the subject. Fortunately, when such students become concerned with a particular area, they quite forget that they have labeled themselves "physical geographers" and proceed to study all features interconnected in the area.

Indeed, it is somewhat misleading to overemphasize the position of geography as "a bridge between the natural and the social sciences." Though Penck has used this analogy a number of times he would be one of the first to insist that, insofar as there is a gulf between the two groups, the gulf is of man's making, it is not present in the reality that science is to study. We cannot, however, accept his further inference that the concept of scientific laws has been developed only on the one side of this artificial gulf and the bridge of geography is needed to carry it across to the social sciences on the other [158, 54; 163]. A concept of this kind requires no bridges. On the other hand, Penck may mean that scientific laws in the social sciences can be developed on a sound basis only if they are connected, through geography, to the natural sciences. Even in this sense we would be claiming too much, for the social sciences have other connections with the facts and relationships of the non-human world, notably through human physiology and psychology [cf. Kraft, 166, 12].

Whatever conclusions may be drawn with respect to that question, geography is not to be thought of as a connecting link between two groups of sciences, but rather as a continuous field intersecting all the systematic sciences concerned with the world. It therefore has not two, but many facets, as Schluter observed; the difference in methods between studies of climate and of landforms is in many respects greater than the difference between the study of natural vegetation and of cultivated crops [148, 145 f.].

The most that we can learn about the nature of geography from the conventional classification is that geography necessarily shares in whatever difficulties or limitations the social sciences are heir to, and that, on the other hand, it shares in part in the greater ease with which facts and relationships can be determined if the human element is not involved. Since the developments of the last generation have destroyed the faith in absolutes of the nineteenth century physicists, we know that there is here no difference in kind, but only in degree, between the two groups and among the different sciences in each group. Furthermore, this is a difference which applies only in general, not necessarily in the particular instance. Failure to recognize this fact has led many geographers to presume that geographic work had a major degree of soundness if its feet were established in the natural sciences, regardless of how wildly it might leap from there to conclusions in the uncertain atmosphere of the social sciences. In reality, few facts of the natural environment can be established with such a degree of certainty as the rate of population growth in the United States, or the areas included within the dominions of the political states of the world.

We can secure much more insight into the character of geography if we consider it in terms of the classification which we discussed in the fourth section of this paper. According to Kant, Humboldt, and Hettner, it is necessary to look at science as a whole from different points of view. From one point of view, all reality may be regarded as a collection of many different kinds of phenomena which can be sorted into groups according to the kinds of objects with which they are concerned. The student who approaches science from this point of view endeavors to learn everything he can about the phenomena of one particular group of objects regardless of where and when they may be found. Since it is possible to classify all objects, roughly, as animate and inanimate, of non-human (natural) or human origin, this "systematic" point of view permits a fairly clear subdivision into different if "systematic sciences."

In the reality which science is to study, however, the phenomena are not arranged according to the classification which the systematic point of view constructs. Consequently this point of view gives an incomplete view of reality. If phenomena were simply piled and mixed together in reality without meaning, it would perhaps be sufficient simply to state that fact. We know, however, that there are significant relations between the different kinds of phenomena that are found together in any particular section of reality, and also between phenomena in different sections of reality. That is, there is some degree of system or order in the actual arrangement of phenomena in reality. To comprehend reality more fully, therefore, we must not only study phenomena, but must also study the different sections of reality in order to understand the character of each section in comparison with the character of other sections. To understand the character of any section of reality we must attempt to comprehend the integration of phenomena of different kinds that are actually integrated in it.

Although this integration can be stated theoretically in the singular, the nature of reality forces us to take two separate points of view. The whole of reality may be divided into sections in terms of either space or time. Though a single section combines these--here and now is one point in reality--it becomes practically, if not theoretically, impossible to consider simultaneously differences in time and differences in space. Only if the phenomena are relatively simple, as in astronomy, or the data relatively meager, as in paleogeography, have efforts to combine the two met with (see Sec. VI A). The consideration of sections of reality in terms of historical point of view, represented by historical geology, prehistory, and history in the narrower sense. The consideration of sections of reality in terms of space is the chorological point of view, represented by astronomy and geography.

Every one of these historical ⁹⁸ and chorological sciences must study all the kinds of phenomena that are found in its particular sections of reality. Theoretically these could include phenomena of all the systematic fields, whether physical, biological, or social. Only special circumstances limit the range within certain of these fields. W. M. Davis recognized this common characteristic of geography, history, and astronomy. "Dealing with things or events of many kinds in definite relations to time or place, they cannot have the singleness of content which subjects like mathematics and physics and chemistry possess." Astronomy, he continues, is essentially the mathematics, physics, and chemistry of the universe, and only the fact that evidence of organic life has not been found in the heavens has prevented the astronomer from overlapping into biology, or even one might add, the social sciences [104, 213 f.]. Likewise, it is only the circumstance that natural conditions on the earth have changed but little in historical (not human) times that largely limits history--as distinct from "prehistory"--to human phenomena. Nevertheless the eruption of Vesuvius is a phenomenon of concern not only to the geologist but perhaps even more to the historian--as is indicated by the fact that the reader knows at once to which eruption we refer. Likewise anyone studying the history of Holland in the, Middle Ages must consider the changes consequent upon the formation of the Zuider Zee.

It may be particularly instructive to glance at that special division of the historical view of science known as historical geology. The innocent layman might suppose that one could study the inanimate rocks of the earth's crust without overlapping into the fields that study the phenomena of life. But since the historical geologist is the only scientist who is presented with material for studying the history of the world in remote times, he finds that he must include historical botany, zoology, and human anatomy, and even to some extent historical social anthropology.

This consideration of the nature of different kinds of science should enable us to meet "the oft-discussed assertion" of which Colby speaks in his presidential address, namely, "that geography has no distinctive phenomena ⁹⁹ at the center of its interest, as have, for example, soil science, botany, and chemistry" [107,2]. The geographer need not hesitate to acknowledge the truth of that assertion, even though it establishes an essential difference in character between his field of study and the systematic sciences like chemistry, botany, or political science. The group of sciences among which geography is thereby classified should not, one would suppose, prove humiliating to the geographer.

Geography does not claim any particular phenomena as distinctly its own, but rather studies all phenomena that are significantly integrated in the areas which it studies, regardless of the fact that those phenomena may be of concern to other students from a different point of view. The astronomer has no monopoly of the study of the stars, he is not disturbed if physicists and chemists study the elements of the stars. Similarly, geography need not look for any concrete objects as its own. The rocks which the historical geologist uses for his data are equally the concern of the dynamic geologist and his pet fossils are proper objects of study for the botanist, zoologist, or anthropologist. Likewise the historian is not disturbed if told that his field is an aggregate of economics, political science, and sociology.

Finally, geography does not distinguish any particular kind of facts as "geographic facts." As Barrows has often insisted, any particular fact--meaning a primary fact, not a relationship loosely considered as fact, nor a deduction from relationships--is not a "chemical fact," a "geological fact," or an "economic fact"; it is simply a fact, and any branch of science may use it. It is only because various kinds of facts are more commonly studied in certain sciences than in others that these conventional, but misleading expressions are in common use. Thus the facts concerning the price of wheat in different places and different times may be considered most frequently in economics, and therefore are called "economic facts," but they could equally well be called "historical facts" or "geographic facts." Geography in particular cannot accept either the popular misconception which classifies under "geographic facts" only the facts of location, or, the misconception common in scientific circles which considers this term as including, in addition to the facts of location, only the facts of natural phenomena. In the broadest sense, just as all facts of past time are historical facts, so all facts of the earth surface are geographical facts. And just as history does not use all facts, but only those--of whatever kind--that are "historically significant," so geography will determine which facts it will utilize, not according to their substance, but according to their geographical significance, i.e., their relation to the areal differentiation of the world [cf. Sec. VIII].

To state, for example, that Vesuvius is (and was) a volcano located at 40° 49' N., 14° 46' E., is to state a fact which is no more geographic than geologic or historical--it is of course simply a fact. In the systematic geography of volcanoes, we are concerned with this fact in its relation on the one hand to the zone of diastrophic action that runs through the Mediterranean region, and, on the other hand, to the fertile ash soils of the neighboring Campagna, to the ruins of Pompeii and buried Herculaneum, the hazards of life of the population of the area, and the landscape effect of the volcanic mountain in the level plain.

In sum, then, geography, like history, is to be distinguished from other branches of science not in terms of objects or phenomena studied, but rather in terms of fundamental functions. If the fundamental functions of the systematic sciences can be described as the analysis and synthesis of particular kinds of phenomena, that of the chorological and historical sciences might be described as the analysis and synthesis of the actual integration of phenomena in sections of space and time.

Both history and geography might be described as naive sciences, examining reality from a naive point of view, looking at things as they are actually arranged and related, in contrast to the more sophisticated but artificial procedure of the systematic sciences which take phenomena of particular kinds out of their real settings.

It is not surprising, therefore, to find both history and geography developed as fields of study in the earliest period of scientific thought. Furthermore, it was natural enough that each of these should have become a "mother of sciences." The attempt to integrate all kinds of phenomena in space or time leads to the discovery of many kinds of phenomena, any of which may then appear to be worthy of study in themselves; indeed the attempt to understand their significance in a total integration requires that they be studied in themselves. Consequently we may expect this evolutionary process to continue indefinitely, so long as the new kinds of phenomena discovered are deemed worthy of study in their own right. Thus, if geographers have discovered the phenomena of house types and can demonstrate that they are sufficiently significant, we may expect some branch of systematic science to make these objects a subject of special study.

On the other hand it should not be supposed, as has often been done, that the recognition of the independence of daughter fields thereby reduces the extent of the field which geography or history is to study. On the contrary the mother field remains exactly what it was before. Furthermore, as Richthofen observed [73, 27 f.] and as Hettner has repeatedly emphasized, the progress in these related fields enriches the materials to be studied in geography. Just as the development of economics and political science has greatly increased the ability of historians to interpret history, so modern geography has benefited enormously from the development of systematic physiography, climatology, soil science, etc., and should benefit from the findings of economics and other social sciences. What contribution geography can make in return will be considered later.

The failure to understand that geography is to be defined essentially as a point of view, a method of study--just as all science is a method of study--has caused many to suppose that the growth of the daughter sciences had left nothing for the parent science to do. Attempts have been made to save the day by claiming for geography a particular type of phenomena, such as relationships between man and nature, or by searching for new objects of study which no one else has previously considered worth studying [Crowe, 201, 2], or by attempting to metamorphose abstract concepts of area into concrete objects. Each of these efforts, to a greater or less extent, has caused geography to depart temporarily from its path of development in directions which have proved, or will prove, to lead either fields that other sciences will not cede to geography, or into the bog of mystical thinking.

One answer to the question at the head of this section, therefore, is that geography is a study which looks at all of reality found within the earth surface from a particular point of view, namely that of areal differentiation. This might be called the position of geography as a field of knowledge. More significant to the general question is the character of geography as a field in which knowledge is acquired.

Geography attempts to acquire knowledge of the world in which we live, both facts and relationships, which shall be as objective and accurate as possible. It seeks to present that knowledge in the form of concepts, relationships, and principles that shall, as far as possible, apply to all parts of the world. Finally it seeks to organize the dependable knowledge so obtained in logical systems, reduced by mutual connections into as small a number of independent systems as possible [cf. 115, 106-14]. It is in terms of the manner in which geography pursues these ideals that we attempt to describe its character as a field of study.

It should be noted, in general, that our definition of the kind of knowledge of which geography is a part is based not on what is known, that is, on what has been learned, but rather on the pursuit of knowledge--i.e., the fundamental principles governing the manner in which the unknown is to be learned. Different branches of this "kind of knowledge" differ in the degree to which they have been able to approach the ideals stated; no branch of science can claim to have attained perfect certainty or exactness, actual universality, or complete organization of all its knowledge in a single system.

If one compares these ideals, as the fundamental requirements for that form of knowledge, which hereafter for convenience we will call "science," with the ideals of artistic perception (in whatever form one finds them stated by students of art), it is clear that there can be no logical combination of the two and no transition from one to the other. The artist certainly does not subordinate all other considerations to an ideal of exactness, nor to one of certainty. He may require that his work express a fundamental universality but that concept does not control the details of his work. Likewise he may require that an individual work of art should be organized, but does not seek a common organization of all similar works of art. In other words, we are dealing here with two essentially different approaches; both artists and geographers may attempt to acquire and present knowledge of an area of the earth, but neither can adopt the ideals of the other without sacrificing his own [cf. Kraft, 166, 20 f.].

That geography seeks to make its own knowledge of the world as accurate and certain as possible is an assumption which presumably would not be questioned by professional geographers. It is not, however, a correct corollary of this assumption that geography should arbitrarily limit itself to particular kinds of facts because they appear to be subject to more accurate and certain means of measurement than others. Even if it were possible to demonstrate--as would not be the case--that the facts and relationships of immaterial phenomena could never be determined as accurately or certainly as those of either visible or material phenomena, no principle of science would require us to limit ourselves to those phenomena that could be studied somewhat more accurately and certainly, excluding those that could be studied somewhat less accurately and certainly. On the contrary, in order that our knowledge of, say, the cultural character of an area may be made as certain as possible, we are required to consider all the facts that bear upon that knowledge, whether the material products of cultural ideas or the immaterial manifestations of those ideas. Both sets of facts must, of course, be observed as accurately and certainly as possible [115, 83-99].

Likewise these ideals admit no limitation in the specific methods of observation in geography, but rather require that we utilize all methods that will lead to more accurate and certain knowledge. Thus, because statistical data are neither sufficiently accurate nor sufficiently detailed, we must employ the technique of field work--both direct observation and personal interview [cf. Jones, 287]--to check and supplement the knowledge gained from statistical data. On the other hand, even if it were physically possible to make a complete field observation of every part of a region, the observations obtained reflect temporal conditions, commonly of a single season, which, as R. E. Dodge noted, may lead to erroneous generalizations concerning the continuous use of the land [287, 100]. Consequently the findings of field observations must be checked with statistical data, even though these were very indirectly based on observation.

The scientific ideal of certainty commands that the terms and concepts of description and relationships be made both as specific and as certain as possible--we cannot develop a sound structure on a marsh foundation in which ambiguous concepts shift their meaning whenever pressure is applied to them, and concepts apparently specific prove to be but dubious analogies.

The ideals of accuracy and certainty apply not only to the manner in which primary facts are established and to the formulation of fundamental concepts and technical terms, but also to the processes of mathematical and logical reasoning by means of which we induce relationships of observed facts, and thereby deduce further conclusions as to facts. When a single student studies a particular scientific problem, no matter how hard he strives to live up to these ideals, no matter how critically he examines his own hard work, there remains the possibility of error, whether through carelessness or through subjective influences affecting his observations and reasonings. Since this is a generally accepted axiom in those sciences whose facts can be measured with the highest degree of accuracy and whose relatively simple phenomena make logical reasoning most certain, how much more uncertain are the findings of one student in geography! In order to make possible a higher degree of accuracy and certainty, therefore, it is a recognized principle of all science that studies should be carried on, organized, and presented in such a manner as to provide an accumulation of evidence of different students on the same problem. Every scientific study should, that is, by which we mean both that it should make use of all previous scholarly studies bearing on the problem and that it should be presented in a form usable by subsequent students.

It is one of the handicaps of the social sciences, and in part also of geography, that in much of their work it is not possible for later students to have access to the primary data used in any study; much must be taken on faith in the professional ability and reliability of the individual student. Consequently it is all the more essential in these fields that similar studies of other students should be utilized and referred to, not as a matter of professional courtesy, but for the sake of accumulating evidence in order that results may be more certain.

Science will accept on faith from any scientist no more than is absolutely necessary. In order that subsequent students may reexamine the findings of a particular study--whether in order to utilize them further or simply to check or correct them--it is essential that there be a clear understanding of the generic concepts employed, of the methods of observation and reasoning by which the results were obtained, and, finally, of the relation of the study to the field as a whole. Obviously such a clear understanding will be easiest if the field has developed standard techniques and organization; if it has not, they must be specifically indicated in each study.

It is not the function of this paper to pass judgment on the scientific quality of work actually produced by geographers--it is their ideas about the nature of geography, and the consequences of those ideas that concern us. On the letter basis, however, it is a fair question to raise, whether the research publications in geography indicate that either geographers in general, or the editors who, to some extent, control their publications, have accepted these standards as essential to geography--standards dictated by the scientific ideals of accuracy and certainty.

With the respect to these ideals as essential principles in the pursuit of knowledge. there can be no differences among the different branches of science, but only differences in degree of attainment. Undoubtedly the attainments of geography in this respect are not of such a high degree as to tempt us to compare it with other sciences in such terms. But neither need we feel humiliated if those who choose to make such comparisons should assign our field to the "lower classes." In any competition the understanding observer will judge attainments not only in terms of degree of success but also in terms of the relative difficulties of the tasks undertaken.

The consideration of the manner in which geography may pursue the other two scientific ideals, universality and system, is of the greatest importance in understanding the character of geography as a field of study. Since these are not so readily understood as the two ideals which we have just considered, each will require detailed consideration.

While everyone recognizes the importance of universals in science, it is a common error to overlook that part of our scientific knowledge which cannot, as yet at least, be expressed in universals. Many have assumed that science was concerned exclusively in the development of laws and principles. This concept, according to Hettner, represents an outgrowth of the great development of laws and principles in astronomy, physics, and chemistry in the last century [161, 221-4]. Indeed, it came to be assumed that physics and chemistry were exclusively abstract sciences, concerned only with laws and principles. More recently, however, many scientists and philosophers of science have recognized that no branch of science concerned with reality--as distinct from theoretical mathematics--can limit itself to laws and principles. Though science strives for universals, these do not exhaust the study of reality, there is always an individual remainder that is not described or explained. If this be ignored our knowledge is less complete than it might be, a quod est absurdum which no science can accept. "Research," as Lehmann says, "is a seeking for the most complete presentation possible" [113, 299].

The contrast between that aspect of science which can be expressed in universals and that which is concerned with the individual object or phenomenon as worthy of study in itself, received the particular attention, at the turn of the century, of two German philosophers, Windelband and Rickert [see discussions by Hettner, 111, 254-9, or 161, 221-4, and by Graf, 156, and the discussion between them]. They appear to have classified the different branches of science on this basis, distinguishing between "nomothetic" or law-making sciences, and "idiographic" sciences, those concerned with the einmalige, the unique. (See Supplementary Note 46) In a rough way this seemed to conform to the conventional division between the natural sciences and the social sciences.

It seems clear, however, that these two aspects of scientific knowledge are present in all branches of science [cf. Schluter, 131, 510 f.]. The common idea that generalizations and laws themselves are the purpose of science is characterized by Hettner as an extraordinary adherence to medieval scholastic realism. On the contrary, they are "merely the means to the ultimate purpose, which is the knowledge of actual reality, the individual facts, either conditions or events" [cf. also Kraft, 166, 11-13]. (The quotation is from Hettner [161, 222]) The astronomer develops laws of celestial mechanics not in order to prove that the universe is governed by law--which is a philosophical rather than a scientific thesis--but in order that he can rightly understand the motions of the heavenly bodies. He does not forget his interest in the latter as individuals. The students who have mapped the surface of the visible side of the moon or who study the rings of Saturn are no less astronomers if they have not thereby been able to establish laws.

Nevertheless it is clear that in some sciences the nature of the phenomena studied permits of a much greater development of nomothetic knowledge, whereas in others the greater differences between the phenomena studied, and also their greater importance individually to man, requires the students in those fields--whether they will or no--to concern themselves in large degree with the unique. This is, then, a significant respect in which there are important and permanent differences of degree among the different branches of science.

A comparison of various branches of science from this point of view may aid us in understanding the character of geography, may help us to perceive what kind of science, in this particular respect, geography is.

Although we have noted that all branches of science are concerned to some extent with the study of the unique as well as of universals, there are some sciences in which the actual work of many students may be largely, if not completely, absorbed in the search for universals, the study of the unique being left to other students or to other divisions of science. I presume that this may be the case in physics and chemistry, particularly in their theoretical aspects. To a lesser extent the same is true of the biological sciences and psychology, and also of certain branches of economics and, perhaps, sociology. (It is both impolitic and dangerous to discuss the nature of a field in which one has not been trained. I might say, however, that the statements in this and the following paragraphs are based on discussions with colleagues in the appropriate fields.)

On the other hand, there are certain branches of most of the sciences in which the students are very largely concerned with the study of the individual object, the unique. We have already noted the importance of this form of study in astronomy. Idiographic studies are certainly of great, if not major, importance in geology. This is obviously the case in paleontology. It is true of mineralogy in so far as the student concentrates on certain mineral deposits for the purpose of securing complete knowledge of those deposits rather than of developing new principles of mineral deposition. Even more does the statement apply to the great and certainly valuable work of the geological survey in mapping the areal geology of particular areas (though this might logically be considered a branch of geography carried on by geologists).

Among the social sciences, economics has developed its nomothetic knowledge possibly to a greater extent than some of the natural sciences listed above. In political science, probably only those branches commonly referred to as "political theory" are primarily nomothetic. Most of the students of comparative government, for example, are concerned not merely with general principles of government, but--perhaps even more--are concerned to know and understand the differences between individual governments. In their persons, these students combine the nomothetic and idiographic interests, whereas in physics or biology or perhaps economics, these may be separated between different students. Since the conditions which have resulted in this situation in political science are very similar to conditions in geography it may be instructive to analyze them.

The first condition is that for most of the phenomena of political science a generic description provides only a bare outline of common features, the individual specimens differing notably with respect to other features. The phenomena are much more complex in structure than those of physics, or even of economics, and, though less complex than the specimens of organic life, do not have the high degree of similarity that results from the common origin of specimens, species, genera, and orders in the biological sciences. Even though it be true that, to the research physician--in contrast perhaps with the student of human physiology--"every human body is different and unique, every spine to some extent different from every other one," these differences are very much smaller than those found among, say, the individual governments of the same type. In other words, a generic description of human spines gives a much more nearly complete picture of any one spine than is possible in a similar approach to governments. If the political scientist is to learn as much about governments as the physiologist learns about human spines he will have to give a much greater share of his attention to the unique characteristics of each specimen.

In addition, more attention will be paid to an individual government, because of its greater importance, in comparison with an individual human spine. This second condition holds generally: the phenomena of political science are, as single cases, so important (to man) that it is necessary that each be understood as completely as possible. Whatever the personal preferences of a student may be, mankind will not be satisfied with a science that studies dictatorships, for example, only to the extent that generic concepts and social laws can be developed, and declines to consider the differences between the present government of Italy and that of Germany on the grounds that these are unique cases.

A third consideration is of practical import. With millions of human spines in the world, a single scientist, even a whole group of scientists, may be kept fully occupied in considering only those aspects of the spine that are relatively universal, and may be able to derive sufficient universal knowledge thereby to justify his self-limitation, leaving to others the consideration of individual characteristics. But the political scientist cannot find, even in the great storehouse of history, more than a relatively small number of dictatorships of all kinds to study, hardly more than a handful of cases of a particular type of dictatorship. Under these conditions, to limit himself to the study of generic features and the development of principles, and to decline to study individual specimens as unique objects would be absurd. The reverse consideration is perhaps even more significant: human physiology cannot find time to study the individual characteristics of all human spines, whereas individual studies of all the governments in the world would not be impossible for the field of political science.

It is obvious that the same considerations apply in large part, and with even greater force, in history. To be sure, an individual historian, or particular group of historians, might limit themselves to considering the history of constitutions, of wars, or even of battles, and from these studies derive laws or principles of value to the general historian interested in the study of periods of history and the sequence of events. Whether such historians would not thereby become students of military science or of constitutional government is a question which is not for us to consider; certainly such work represents but a small part of history.

If, now, we consider our own field there can be little question that geography is very much concerned with the study of individual phenomena. (See Supplementary Note 47) Indeed some critics might feel that geographers tend to lose themselves entirely in the consideration of the unique, regardless of its importance. If we were to study each and every feature of the earth surface that could be shown to have geographic significance, without considering their relative importance, it is clear that geography would become hopelessly lost in an insurmountable mass of detail. Not only every small district might be studied, but every single village, every mountain, every little rill on the face of the earth.

On the other hand, a geographic description that is confined to generic terms is inadequate. Even a brief and superficial description of a major area must name and describe a great number of individual features. The Alps and the Himalayas, Mount Rainier or Vesuvius, the Amazon, the Mississippi, or the Niagara, states like Germany or the United States, cities like London, Paris or New York--if included within the area described--these will be considered not merely as examples of types or as illustrations of principles, but for their own individual significance [cf. Hettner, 161, 221-4].

One reason for this, as we have already seen, is that any generic description, however detailed, is completely inadequate to depict these individual features. When a biologist has described the nucleus of a certain type of cell he has provided a fairly complete picture of the nucleus of any individual cell of that type, whereas a full knowledge of the character of the commercial core of Chicago would give one but little understanding of the corresponding district in Minneapolis. It is only a very general similarity which justifies the citizens of the latter city in calling their commercial core, in imitation of the larger city, "the loop"; actually the street car lines do not loop in Minneapolis, as do the street and elevated car lines in Chicago.

If the previous example appears to depend on details of secondary importance, consider such geographic features as the Great Lakes System or New York City. There are, to be sure, a number of great lakes in Africa, but, as a system of connected fresh-water seas, the Great Lakes of North America have no counterpart on earth--or, so far as we know, in the universe. When one considers New York in its fundamental aspects as the focal point of trade of the greater part of a continent, and then looks for its counterpart simply in this one respect ignoring entirely its local character, one finds no other city of the type; Shanghai, Buenos Aires, and Hamburg are all centers of trade for extensive areas but no one of them functions for its particular continent as New York functions for North America.

The informed reader will recognize that while these are conspicuous cases they represent a rule that applies to most geographic features, even to those of slight importance. Because these features are not adequately described in terms of types, geographers, as well as laymen, fall back on analogies. Possibly these have their value; but if taken too seriously they are deceptive. To call Hankow, "the Chicago of China," or the Pittsburgh-Cleveland area, "the Ruhr of America," is to give a false as well as true impression.

Our examples have likewise illustrated the other consideration which requires geography to examine many phenomena as individuals--namely, their great importance as individual phenomena. By this is meant not merely their utilitarian importance in the world, but also their importance to the earth surface on a purely academic basis. Even if Niagara Falls had no material value, whether actually or potentially, a geographic science would not be satisfied with a description which simply classified it as of a certain type of falls along with hundreds of others. In itself, in all its individual characteristics, it is a phenomenon of scientific concern to man.

The concern for the unique in geography is not limited to phenomena but applies also to the relationships between phenomena. Thus, if the relationship of Winnepeg to the lakes and barren land on the north and the international boundary on the south is essentially unique, we do not disregard it on that account.

It might appear as though geography were as largely limited to the study of the unique as is history, and, until the last century, that was no doubt the case. As Hettner notes, "geography was confined largely to such idiographic description . . . general concepts were to be found only in the crude form expressed by such common typewords as mountain, valley, city, etc." [161, 222 f.]. Insofar as the situation was unavoidable, such a statement would represent no criticism of geography as a science, any more than it does of history? Barry's statement may also be expressed in negative form: it is not the business of science to learn what cannot be learned.

On the other hand, a geography which was content with studying only the individual characteristics of its phenomena and their relationships and did not utilize every opportunity to develop generic concepts and universal principles would be failing in one of the main standards of science. We can therefore agree with Hettner that "the greatest scientific advance of geography has been the fact that, by taking over and developing the results of the systematic sciences--earlier in one branch of the field, later in others--geography has gone over to general observations depending on generic concepts" [see also Sauer's quotations from P. Barth, 211, 27].

Undoubtedly the development of generic concepts has progressed much further in respect to the natural phenomena of concern in geography than in respect to the cultural phenomena, partly because of the close relations which geography has had in most countries with geology and other natural sciences, and also because of the greater development during the past century of the systematic sciences concerned with natural phenomena as compared with those concerned with cultural phenomena. Nevertheless, examples of generic concepts could be cited from every section of systematic geography.

Thanks to such generic concepts it is possible to express many characteristics of a particular feature in one word or phrase--or in a symbol, such as Cfb--so that one can give a relatively brief description which can easily be kept in mind. Furthermore, the use of such generic concepts has made it possible to develop principles of relationships between different factors which are found repeated in different parts of the world. These principles can then be used in studying the integration of phenomena in specific areas.

While general principles have been most fully developed in connection with the relationships between various natural features, progress has also been made toward general principles concerning the relations between any particular cultural item, such as a crop, a particular kind of factory, or a city, and all the other features that are significantly related to it. Such principles have been developed in every phase of economic geography, and, to a lesser extent, in other cultural aspects of geography.

To some degree, therefore, geography may be called a generalizing or nomothetic science. Both Hettner and Penck find that in this respect geography has a great advantage over history [161, 223; 158], though it would be an error to overlook the use in history of generic concepts and principles contributed by the systematic social sciences.

Few geographers are satisfied, however, with the present state of development of principles in geography--either in terms of their quantity or of their reliability. Insofar as more principles or more reliable principles are possible, no one, of course, should be satisfied. Professional geographers, however, may feel less discouraged if they perceive clearly the difficulties and positive limitations that are placed in their way in the effort to develop generalizations and principles. More than that, the clear recognition of those facts may prevent misdirected efforts that can only yield fallacious principles.

James seeks the explanation for our difficulty in the fact that the phenomena which we observe are so much larger than the observer that we cannot see the woods for the trees, or, to use his excellent simile "like a microbe crawling over the face of a newspaper photograph, we see only the details of the printed dots--the larger design of the photograph lies beyond our range of vision" [286, 84 f.]. Granted the force of this difficulty, have not geographers long since discovered the solution which he recommends, namely use of the map to bring the larger relationships within the range of our vision? And yet we have not advanced very far in the establishment principles.

One major difficulty lies in the fact that the integration of phenomena which we must study in areas is an integration of a large number of independent, or semi-independent factors. Consequently we seldom have to do with simple relationships--e.g., rainfall to soil, temperature to crops, etc. Theoretically we might follow the logic of the systematic sciences, by assuming that all other conditions remain the same, but we have only the laboratory of reality in which to study these features, and in that laboratory the other elements do not remain the same, except perhaps in a very small number cases, and we have no way of making them remain the same. Indeed, even if we knew the theoretical principles governing the relation of each individual factor to the total resultant, in the case of such complex resultants as cultural features, a principle which attempted to state the sum total of all the relationships, each in its proper proportion, would be far too complicated for us to be able to use. This is a general difficulty that applies not only to all the more complicated aspects of the social sciences, but also to many phenomena in the natural sciences. Even if one knew all the principles and had all the data, the solution would be involved in a mathematical equation so complicated that no finite mind could solve it. Even when geography has attained a maturity of techniques and methods, it will, as Colby suggests, hesitate to predict resultants of such a complexity of factors [107, 35 f.].

The second major difficulty, definitely limiting the applicability of such rough principles as may be developed in geography, results from the insecure foundation on which the principles rest, namely the generic concepts. This limitation is present, to be sure, in every branch of science, in greater or less degree. In those sciences that are able to perform controlled experiments, specimens may be selected that conform almost exactly to the generic type involved in the principle; to the extent to which actual specimens in reality differ from the generic type, the principle will not apply. In geography we seldom have a sufficient number of specimens of any generic type to permit us to discriminate in our selections. Further, as we have seen, our specimens do not conform closely to a generic type, but only within very wide limits. This wide divergence of the individual within the type is illustrated by the marked lack of agreement among geographers, not on the classification of a minority of doubtful specimens, but in regard to the very types, and systems of types, themselves. Since sound principles can only be developed if we have a sound system of generic concepts, we must consider for a moment how such concepts are developed in geography. (See Supplementary Note 48)

We indicated earlier that many, if not most, of the generic concepts in use in geography had been brought in from other branches of knowledge. One group, perhaps the larger, are those whose expression in words of common speech reflects the fact that man studied geography before he ever heard of such a thing as science. While the concepts of common sense are likely to be crude, many of these have been found to be adaptable to scientific thought. Ocean, lake, river, mountain, plain, city, port, farm, crops--for all of these the ordinary meaning, as used in common speech, can be sharply defined for scientific use. To invent, in their place, technical terms with which to impress the outsider would be pretentious [cf. Schmidt, 7, 192]. On the other hand, many similar terms in the language are used in such different ways that it may not be possible to pin them down to definite technical meanings.

A second large group of generic concepts has been brought into geography from the systematic sciences, notably from geology. Though many of these have also proved useful, it is erroneous to presume that their eminent scientific authority assures their logical utility in geography. The common rule of scientific work that requires the students in one field to take cognizance of the findings in neighboring fields and, unless they can show errors, to assume that the facts and relationships established in the other branch of science are correct, does not apply to a classification system, which cannot be called either correct nor false, but simply more or less useful. If, for example, geologists have classified landforms into generic types based on their past history rather than on their present forms, their concept of a "plateau" need not necessarily be accepted by the geographer simply because it is a "scientific concept," in contrast, say, to the time-honored concept of the same word in common speech, and as used by geographers long before the geologists gave it a new and different meaning.

Finally, geographers themselves have developed various systems of types into which they may classify the features which they study. They have been particularly concerned to supplement generic concepts based on individual elements with generic concepts based on element-complexes. These new systems of classification often appear in opposition to those taken over from common thought or from the neighboring systematic sciences, so that there has been a great deal of argument as to which generic concepts are proper for the science of geography to use. This is not, as many have thought, simply a matter of preference either for the "new" or for the "old." The question is more fundamental: from the particular point of view of geography, which concepts are of greatest value? Since this has caused much argument, and is clearly of great practical importance in the development of the field--in contrast perhaps with some of the questions that have required our consideration--it merits a brief analysis.

The fact that it is impossible to include all the characteristics of any object in a single generic concept means that any particular object may be classified simultaneously in many different types according to the bases of classification. But these types may be sufficiently alike so that, in view of the limitations of our vocabulary, we use the same terms for different concepts. Take the very simple case of coal. The paleobotanist, I presume, will classify coal as a vegetable product, to the stratigrapher it is simply a kind of rock, and the student of mining will say it is a mineral. Since coal is, in fact, simultaneously all of these things, it is included in each of these concepts though no one of them of course completely expresses what coal is.

Consequently the classification of objects into generic types cannot be determined simply on the naive basis of "the inherent characteristics of the objects themselves." In terms of all their characteristics, no classification is possible. Which characteristics will be considered in developing generic concepts will therefore depend on the purpose for which the generic concepts are developed. Generic concepts are not an end in themselves but merely a scientific tool whose particular purpose varies according to the point of view of the different branches of science.

In general, the purpose of developing generic concepts is to provide a single statement of a collection of common characteristics shared by objects which otherwise differ. This is useful for the primary purpose of describing reality in that it provides a shorthand method of description which likewise enables us more easily to grasp and retain descriptions. Furthermore such precise shorthand collectives are extremely useful, if not practically indispensable, in developing principles of relationships between objects. In both cases it is clear that the generic concepts should express those characteristics of objects which are of greatest importance according to the particular point of view under which they are being studied. For example, a student in the science of heat engineering may ignore, in his classification of coals, various characteristics that have no effect on the burning qualities of coal, however essential these characteristics may be to the paleobotanist.

We have, therefore, the fundamental basis for judging different systems of generic concepts in geography: which classification most completely and clearly expresses those characteristics that are of most importance from the point of view of geography, whether for geographic description or as a basis of establishing principles of geographic relationships.

To the extent that the point of view in geography is different from that of other branches of science, whether systematic or historical, there is no presumption in favor of generic concepts introduced into geography from other sciences; on the contrary it might be safer to presume the opposite. On the other hand, the generic concepts which we have found in common speech are usually developed from the same naive point of view toward reality that geography also expresses. To the geographer, as well as to the sailor, the whale is a fish or--if this be regarded as a misuse of a word--the whale is to be included under the general concept of ocean animals rather than land animals.

Many of the concepts which geography has inherited from the geologic physiography of the past century were developed by students whose primary concern was in the processes of formation of landforms, presumably as a part of earth history. It was, therefore, appropriate that they should attempt, however difficult it may be, to classify landforms in terms of those characteristics which resulted from, and therefore were the key to, their genesis.

If the geographer were likewise directly and primarily concerned with the genesis of landforms, he would, without hesitation, accept the same generic concepts; but on this basis geography would be difficult to distinguish from dynamic physiography. Conceivably, however, the geographer might require a genetic classification as the best means for expressing to the fullest the character of the landform. This, if I understand it correctly, is the line of reasoning followed by Hettner. The geographer seeks to establish types of landforms or climate "according to the entirety of their characteristics" [161, 222 f.]. This, we found, cannot be done directly. Conceivably it might be produced by a genetic classification (though that would appear as a first step rather than a final step, as Hettner suggests). But is this possible? Let us assume different landforms, each of which represents a resultant form developed by separate processes of folding, elevation, volcanism, and erosion. A system of classification could be constructed in which these various processes determine the major and minor subdivisions, but again we would have the unanswerable question of which processes are major, which are minor.

Penck discussed this difficulty, from the point of view of geomorphology itself, as early as 1906. He observed that there had appeared to be the finest harmony between structure and surface form, so that geomorphology seemed to have won a sure base for its concepts in tectonics. Thus "in place of the original geographic concept of a mountain area as a sum of unevennesses, many had adopted a tectonic concept according to which a mountain area was a strip of strongly folded land." But more careful morphological investigations had shown that the relations between tectonics and geomorphology were not so intimate as one had at first supposed, and one must conclude that the form of the land which the geographer is to study and explain is something different from the structure of the land which the geologist studies [128, 15 f., 35f.]. More recently, Penck has illustrated the same conclusion by comparing the Bohemian massif and the Central massif of France. Both, he says, are products of bursting (zerborsten); they have great similarity in composition and full relationship in their later geological development, and yet geographically are as different as convex and concave forms [249, 6].

A complete genetic description, to be sure, would arrive ultimately at a complete description of form, but a genetic classification allowing for all the processes that may in any area be significant--in all possible arrangements in time--would not be a classification of types. It would only be an indirect way of describing an end result that could more briefly and accurately be described by direct consideration of the present characteristics of the landform.

If, then, any classification limits us to some characteristics, forces us to ignore others, which are the characteristics that are of most concern to the geographer seeking to study an area in terms of the integration of its phenomena? Surely those characteristics that are significantly related to other phenomena of the present earth surface. In other words, he is more concerned with the form or physiognomy of the landform as a functioning factor in the total complex of areal phenomena than as an end product of its own genetic causes.

No doubt there are cases where the characteristics of primary concern are the same from both points of view, volcanoes or canyons, for example. In such cases, both groups of students will use the same generic concepts. Where this coincidence of interest is lacking, geographers are under no compulsion to follow the direction of the genetic physiographer, or the geologist. In German geography the struggle for independence in scientific approach to the study of landforms took shape in opposition to the methods o W. M. Davis. According to both Ule and Burger, the distinctly geographic point of view was most effectively presented by Passarge, as early as 1912, and throughout his work since, he has emphasized the description of present landforms as functioning factors in the region (or landscape) [170, 497 f., 11, 77]. In passing, one may add a note of defence for Davis--if defence be needed. Lecturing at the Sorbonne, according to the testimony of Lehman, Davis stated that "if it proves very difficult to work out the development of the landform, so that, in the problems of the past one loses sight of the landscape, the geographer will do better to describe the land simply with the help of the older orographic presentation" [113, 237].

No doubt most American geographers today would accept Sauer's conclusion that "there is no necessary relation between the mode of origin of a relief form and its functional significance" [211, 37, see also Bowman, 106, 141 f.]. Frequently, however, they fail to draw the logical corollary that type concepts based on genesis may have little significance for function. They forget, for example, that to tell us that an area is covered with "Wisconsin moraine" gives us no precise description of either its surface form or its parent rock content. Geographers have freed themselves from the genetic classification, Sauer notes, most fully in the study of climates [211, 33]. Possibly the delay in establishing a corresponding freedom in the study of landforms may be due to the fact that most geographers, in this country as well as in Germany, have been trained particularly in geology and geomorphology. (Note Fröbel's pertinent statement of 1836, quoted in footnote 36, Sec. III A.)

It should not need to be stated (but since the writer has been misunderstood in oral discussion of this point, he may be permitted to clarify his standpoint) that the suggestion that geographers need to free themselves from a geological point of view is in no sense an attack on geologists. One may feel that the latter have abused a good geographic term in calling a "plateau" what they mean to describe as a "former plateau," but there are too many difficulties in terminology to justify making an issue on that. The concepts which the geologist has developed are not subject to attack by the geographer; presumably they are entirely suitable for the purposes of studying reality from the geological point of view. All that is said here is merely that there is no reason to presume that the concepts of the geologist are suitable--and many reasons for believing the opposite--for the purposes of studying reality from the geographical point of view.

To establish sound principles of the relation between, say cultivation or soil wash on the one hand, and landforms on the other, it is necessary to have generic concepts which express measured characteristics of landforms that are significant to cultivation and soil wash, rather than characteristics expressed by types of origin. In this case, the problem is simple at the first step, but very complicated beyond that. The essential characteristic of the landform, from the point of view just stated, is its slope, but its slope is a surface in solid geometry which varies constantly and irregularly at every point and in every direction at every point. Recent developments in the measurements of slopes by intervals, however, indicate that this problem does permit of partial solution.

In brief, in the naive point of view toward reality which geography shares with the common man, phenomena are significant in terms of their relations to other present phenomena of geographic significance rather than in terms of their origins. Generic concepts and systems of classification will therefore be more useful in geography if based on the functional rather than the genetic aspects of phenomena.

In spite of difficulties involved, geography has been able to develop principles of the relations between the variable elements of area. To a lesser extent, as yet, have the relations between the variations in element-complexes been put into general principles. This should also be possible, even if more difficult. In general the study of element-complexes is still largely concerned with the first step, the development of types; but even in this stage--as represented by the two systems of classification of rural areas discussed in the previous section--generic relationships are at least suggested. When we consider, however, not the elements, nor the element-complexes of areas, but the areas themselves, can we hope to develop general principles? This question lies at the very heart of geography.

It is axiomatic that a necessary condition for the development of general principles is the construction of generic concepts or types. It is presumably on this account that Passarge's suggestion of a systematic classification of areas into types has made such an impression on many students: it appears to offer the possibility of developing universals in the very core of geography, thus raising the "scientific" quality of our work.

Our approach to this question will be more ingenuous if we remember that, while it is essential to science to seek for universals, it is in no way essential that they be found; that, on the contrary, important parts of scientific knowledge--in every branch of science--cannot be expressed in universals. By placing a special value on the endproduct, scientific principles, we are in danger of passing lightly over the fundamental bases for them. We will not be able to establish principles by first calling what are not objects, objects, and then classifying them as generic types without considering carefully whether they are generic types. We may deceive ourselves with schematic outlines of classification and with schematic explanation based on vaguely expressed principles, but the inapplicability of these principles to the reality of the world will show our substructure to be counterfeit. As Kroeber observed--in discussing a different branch of science--"all schematic explanations seem essentially a symptom of a discipline's immaturity" [116, 542].

We need not repeat the reasons why any efforts to establish a world classification of areas in terms of the totality of their characteristics is doomed to failure. All the areas of the world can be classified generically in accordance with any particular element-complex, but the sum total of element-complexes varying in different areas cannot be classified in a single system of generic types (Sec. X, G). For our present purposes, however, it is not necessary to classify all areas; if we could find some areas that could, as wholes, be classified in generic types, we might hope to go on to generic principles.

No matter how small the number of areas we consider, however, the essential difficulty remains the same. The integration of phenomena which we must study in any area is not a complete unit integration, but rather consists of a set of somewhat related but somewhat separate element-complexes, some of which are but parts of integrations extending into other areas--in the full sense, extending over the whole world.

If we may borrow James's simile of the microbe studying the newspaper photograph, we may say--with some exaggeration--that in the geography of any area, it is as though several separate photographs, each with its own design and each of different size and shape, had been superimposed in printing, each cut arbitrarily to fit. The geographer, in the position of the microbe, may be able to reconstruct the pattern of each separate design and classify it generically, but how can he classify the sum total of the separate designs in the particular superposition in which they are found? If the same combination of element-complexes, overlapping in the same way, occurred in more than one area, on could establish a generic type of this very complex form. While it is conceivable that such repetition may occur--just as it is conceivable that a hand of bridge may be repeated--it would be necessary to find numerous cases of repetition to provide us with generic concepts of regions from which we could develop principles.

We are in danger of confusing ourselves by using the word "region" as a more convenient term than "area of a certain type." When we say that the Po Plain and the Middle Danube Plain are "agricultural regions" of the same type as the American Corn Belt, we only mean that these are areas within which we find approximately the same agricultural element-complex. Even if we found exactly the same agricultural element-complex, the three areas could not be called specimen areas of the same species; there is, and can be, only one "Corn Belt" in the world. Any element-complex of an area appears in other areas, but the combination of all of them in the actual mixture as actually found, occurs but once on the earth [after Banse, 246, 41, and Hettner, 161, 263].

The reader may feel that, while this conclusion holds for regions of such large size as those just considered--since the world can include but a relatively small number of such large areas--if we consider smaller regions, localities, perhaps, we should be able to establish types. A colleague who has been good enough to read this paper in manuscript suggests that he could select three localities, one in Mongolia, one in Patagonia, and one in the Great Plains, which are so similar that they could be compared, "if not to peas in a pod, at least to peas in pods in different gardens." Undoubtedly we have in this case very similar combinations of many of the same elements and element-complexes, but can we say that these common features are, in each case, the most important features of the area? Almost all systems of land-use types, including that developed by the colleague who made this suggestion, classify these areas in different major world types. Even if we omit the nomadic area in Mongolia and consider only the commercial grazing localities in Patagonia and the Great Plains, on what basis can we say that the marked differences in development of roads, railroads, and urban communities are of minor importance in comparison with the similarity of other features?

The previous example appears to lead to type localities because it involves areas dominated by one particular element-complex--including both cultural and natural elements. The situation becomes clearer if one takes localities of more average complexity. Thus a locality in the Po Plain may appear, in certain respects, like one in the American Corn Belt, and so one is tempted to classify both as "peas rather than as tomatoes," but in other, no less important respects, the locality of the Po Plain is like one in the Neapolitan plain, and on that basis, we must, so to speak, classify it also as a tomato. Further, in respect to all those features in the Po Plain locality that result from its location at the foot of the Alps, on main routes from northern Europe to peninsular Italy, in proximity to the Adriatic and the Tyrrhenian Seas, this locality is essentially different from all localities in other parts of the world.

In sum, the uniqueness of the region is of a very different order from the uniqueness of each human spine, or of each pea in a pod. Each of these is unique in characteristics that can unquestionably be called minor while major characteristics are identical, whereas a region is unique in respect to its total combination of major characteristics. In this sense we may agree with those who speak of the "individuality" of areas, though we may find another term used by French geographers, "personality," too suggestive of an organic whole [Musset, 93, 274 ff.; cf. Schlüter, 148, 218; Penck, 158, 49; Creutzburg, 248; and Finch, 223, 16 f.].¹⁰⁰

In a very interesting discussion of this problem, Gradmann has suggested that when on has studied a locality in terms of all its interrelated phenomena one has the impression of a complete and single picture which one therefore feels should be expressable in just the right word or brief expression. But "what we possess as a mental picture cannot be passed on to others in its finished state. Each one must work it our himself." The author must lead the reader through the entire analysis and synthesis so that he may himself come to the ultimate perception. His concluding comment should not be spoiled by translation: "Damit wird reichlich Wasser in den Wein unserer jungen Begeisterung gegossen, und es ist niemandem zu verargen, wenn ihm der Trank fürs erste nicht recht munden will" [236, 131 f.].

There remains, however, a method of limited application by which it might appear possible to recognize localities as specimens of the same type. Within any single large area, for example the Corn Belt, or the Austrian Alps, do we not find a definite type of locality in repeated examples which are as like as peas in a pod? Cannot these be considered, as specimens of the same type of area, defined even in terms of total characteristics? One sees at least a resemblance to the specimens of a biological species, since the features of the different localities within a major region have more or less common origins.

If we assume that the particular combination of natural elements is essentially the same in different localities of a single large area and that the cultural development was controlled by essentially the same human factors, there still remain two significant differences. Even within this limited range, we cannot ignore the significance of relative location. The localities near the center of the region will differ from those nearer the periphery in a number of ways that may be of great importance. Further, if these localities are specimens, their characteristics must include size and shape-characteristics which are not merely of academic interest, but may well affect the development of features, notably urban features, within them. But how can we determine the size and shape of the localities? Only a very approximate and somewhat arbitrary basis--which would in any case give us differences as great as though the peas in a pod included objects formed like peas, pumpkins, and goose-neck squashes.

This last consideration reveals the essential error in our analogy. The localities of a larger region do not represent independents specimens of a species but simply similar parts of a whole, whose similarities are based in major part on the fact that they are but parts of a whole. Indeed that "whole," the larger region, is not really a Whole, but only a part of the single complete Whole which we have, the whole world.

In other words, the attempt to develop generic concepts of areas, as distinct from generic concepts about areas--and, on that basis, to compare areas in themselves and develop principles of their relations--rests on the fallacious assumption of the area as an actual object or phenomenon. We are misled by our terminology. When we say that certain areas belong to the same type, we can only mean that they contain one or more elements, or element-complexes, each of which is of the same type in all the areas. That is, having classified the phenomena found in areas in various systems of generic types, we then label an area that includes one or more of those types in terms of the types it contains. We are not actually classifying the area, but only one or more of its characteristics. The area itself is not a phenomenon, any more than a period of history is a phenomenon; it is only an intellectual framework of phenomena, an abstract concept which does not exist in reality. It cannot, therefore, be compared as a phenomenon with other phenomena and classified in a system of generic concepts, on the basis of which we could state principles of its relations with other phenomena. Indeed we cannot properly speak of relationships (other than purely geometric) between areas, but only between certain phenomena within different areas. Likewise, the area, in itself, is related to the phenomena within it, only in that it contains them in such and such locations.

It may be that we have considered the word "area" too literally, that what is meant by area is simply the sum total of all interrelated phenomena found within an abstractly limited space. This sum total is, of course, an actuality, but is not a phenomenon: the combination of more or less related phenomena, some of which are incomplete, since parts lie outside the area, does not form a phenomenon--is not a something that has relations as a unit to similar units, other than the purely geometric relation of location. (To clarify any confusion resulting from earlier use of terms, we should note that, if we are justified by certain authorities in calling a sum of interrelated elements that form a relatively closed total, a unit--but not a Whole--we must not then consider this loose unit as having the attributes of a precise unit--e.g., of forming a single phenomenon that has relations as a unit with other similar units. To return to geographic ground, we know that the relations that we may loosely speak of as relations between areal units are, in fact, nothing but relations between some of the elements in one area and some in another.)

The conclusion that areas, as such, cannot be studied in terms of generic concepts, but can only be regarded as unique in their einmalige combinations of interrelated phenomena, leads some writers to the conclusion that the study of regions is no proper subject for a science. Since we refuse to define a science--though we have referred the reader to the views of such students of cognition as Cohen, Barry, and Kraft--we cannot debate this question. We may repeat, however, that every branch of science is, to a greater or less extent, concerned with the unique. From the standpoint of general culture, as Granö suggests, there is scarcely anything unique that deserves so much attention as the totality of interrelated phenomena in area that forms the milieu of man [252, 44].

On the other hand the conclusion that we cannot consider areas themselves in generic concepts and principles does not mean that we cannot utilize generic concepts that express marked similarities in the characteristics of different areas. On the contrary, it is of great value to discover element-complexes and combinations of several element-complexes repeated in different areas. Both for the purpose of simplifying the enormous task of learning the complicated character of the different areas of the world, and in order to develop principles of relationships to aid our understanding of that world, we shall want not only relatively limited element-complexes that express some character of many areas, but also the most complete element-complexes possible to express many characteristics of perhaps few areas. These are of great value, even though no single generic concept can express all the characteristics of an area, even though any one system of element-complexes may not apply to all areas of the world, and even though a single, very complex combination of element-complexes may be found in but very few areas.

For our present purposes, therefore, we can utilize many element-complexes which we found inapplicable to a system of world division. Thus, though it is not possible logically to divide the world in terms of the combination of relief, soil, and drainage, we do find many localities in which a particular integrated combination of these factors is present--e.g., "black bottoms," as a particular type of floodplain (whether we have as yet a classification of floodplains suitable for geographic purposes is another question).

Likewise we can recognize and utilize element-complexes that include both natural and cultural elements. When one has described the terraced vineyards of steep valley walls, whether of the Italian Alps or the Rhine Gorge, one has provided a major part of the description of similar localities found anywhere between those places. Similarly, one may construct a generic description of a port at the mouth of a navigable river in the humid tropics--the warehouses and factories constructed of imported materials by imported techniques in striking contrast to the background of primitive forest, the residence district of the foreign controlling group in contrast to the native quarters. In its general outline such an element-complex will be found repeated at perhaps hundreds of points on tropical coasts.

Generic descriptions of this kind are of unquestioned value, both for description and for interpretation--i.e., the development of principles of relations. It is only necessary to remember that they cannot include all, even of the major, characteristics of the localities described, nor can they be arranged to form a single system that will describe the world, even in outline.

In any field of science, knowledge must be organized into systems in order that the student of any particular problem may have ready at hand the knowledge of facts, generic concepts and principles that bear upon his problem. If all the knowledge in a field cannot be organized into a single system, it is necessary that the several systems be interrelated as far as is possible [Cohen, 115, 106-14].

In geography, knowledge is organized into systems in two quite different ways. We may divide the phenomena of areal differentiation into major groups each consisting of closely related phenomena, and thus develop specialized branches of the whole field of geography. These include, then, physical geography, economic geography--which may be further subdivided into agricultural geography, the geography of mining, of manufacturing, etc.--political geography, and what might be termed sociological geography. It is, of course, an error to include historical geography in this group, for that is, in itself, a complete geography of any past period (cf. Section VI C).

On the other hand all geographical knowledge, including that in each of its specialized branches, may be organized according to the two different points of view required in studying the areal differentiation of the world: the view of any particular variable phenomenon in the relations of its differentiation to that of other variables over the world, and the view of the total character of all the variables within the area (See Supplementary Note 49).

The organization of geographic knowledge in terms of the individual phenomena studied is called "general geography" by European geographers, or "systematic geography" in this country. The organization according to areas is most commonly referred to as "regional geography." (The use of the term "special geography" for this form of organization is fortunately no longer common. In German geography the usual term is "Länderkunde.")

These two different methods of dividing the field of geography cannot be combined on a single plane. Each of the specialized branches of geography is represented in both systematic and regional geography--in systematic geography by separate studies of single elements or element-complexes, in regional geography by a part of a regional study, limited to a particular group of related aspects. Thus the study of the element-complexes of concern to agriculture--complexes of land-use and of natural conditions, etc.--is a study in regional agricultural geography--e.g., Colby's study of the raisin production of south central California [337].

It is clear that other groupings of geographic features into special fields are possible, as in the geography of settlements (Siedlungsgeographie) or in urban geography. The geography of any particular city is clearly a special form of regional geography, whereas the studies of individual urban features as found repeated in many cities would be included in systematic geography. Further, because a city, like a farm, represents an actual element-complex that may be considered as a distinct unit, systematic urban geography may study the cities of the world, or any part of it, in terms of their differential character in relation to other geographic differences.

That no place is provided for "mathematical geography" and cartography may require a word of explanation. In climatology we are concerned with areal differences that result from the relation of the planet earth to the sun, but for the facts of that relationship we depend upon astronomy. Likewise, the study of the exact shape and measurement of the earth presents problems to astronomy and geodesy (see Sec. III B). The problem of projecting a sphere, or the geoid, on a plane surface, the science of projections, is essentially a problem in applied mathematics, which is of no less concern to astronomy than to geography. Consequently, the chapter on "mathematical geography" that often forms the introductory chapter of a geography text, commonly consists, as Hettner has noted [111, 274 f.], almost entirely of astronomical and other non-geographical material. Though such an arrangement may seem logical, it may be questioned whether the most effective method of introducing students to the field of geography is to begin with detailed studies from other fields--as though a text in biology should begin with a detailed study in chemistry, leading up to biochemistry. Finally, cartography, which is a technique rather than a science, is a form of knowledge of service in many sciences. Because it is more essential to geography than in any other science, and has been developed to highest extent in geography (see Sec. VIII D), it is both natural and reasonable that it should be most closely associated with our science, but it is no more a branch of geography, logically speaking, than is statistics a branch of economics.

Geography, as the study of the world, necessarily includes a large number of different aspects that are, or may be, represented by specialized fields. Nevertheless it is a well-known fact, which can be tested by any random observation of the literature, that by far the greater part of the work in the field--whether in systematic or regional geography--consists of either physical (natural) or economic geography. Urban geography theoretically overlaps beyond these special fields, in actual practice consists of but little more. The fairly sizable literature in political geography is as yet of minor importance and of sociological geography we have almost none. This situation might be regarded as a natural result of the particular manner in which geography has developed in the past century, or it might be suggested that geographers have commenced with the more obvious and simpler problems, leaving the more complex ones for later study. Overlooking the dubious psychological assumption involved in the latter explanation, there are reasons of much more permanent validity.

In the study of the areal differentiation of the world, the interest of geography in each of the many features which contribute to that differentiation is in proportion to its relation to the total. Each of the natural features varies notably in different parts, and its variations are significantly related to those of some other natural features and many cultural features. In general, the marked differences in the natural environment of different parts of the world, and the partial dependence of most cultural features on the natural environment, is adequate demonstration of the axiom that physical geography is of fundamental importance (See Supplementary Note 50) in geography as a whole.

The justification for the notable concentration of geographic work of recent decades on economic geography is not so obvious. To many students this may appear to represent an emphasis on studies of presumably practical value that appears foreign to the spirit of science. Undoubtedly, much of the work in this field is stimulated by such extraneous considerations; if it is definitely designed to serve practical purposes--as in land-use surveys, etc.--it finds its justification as an applied form of geography. Undoubtedly any science receives stimulating reactions from the work of those who endeavor to apply its knowledge and methods to particular problems, but there is an essential incompatibility between the two forms of science, theoretical and applied. The latter is defined and determined in character by the nature of the problem which it has to study in any particular case. A particular, complex problem will not fit into any theoretical branch of science, but calls for all forms of knowledge and techniques applicable to it. Even if the work be divided among a group of different scientists, the division can hardly follow, even in principle, the divisions of theoretical science (See Supplementary Note 51).

Further, it may be claimed that much of the interest in economic geography represents primarily an interest in economic phenomena, which are studied more or less in their geographical aspects. Granted that this may be true for many studies--notably those concerned with the world distribution of certain products, or with the economic situation of particular countries, even though studied in terms of a "geographic basis"--there remain, nevertheless, fundamental reasons that require all the parts of geography concerned with human phenomena to recognize economic geography as of fundamental importance.

In our examination of a long list of cultural features (see Sec. X F), we found that the cultural features whose differences in different areas were of the greatest geographic significance--i.e., in terms of their relation to other areal differences, both natural and cultural--were for the most part economic features. (They do not by any means include all economic features, since many of these are of very little geographical significance, though they may be of great economic significance.) Consequently geographers are justified in regarding human, or cultural geography very largely in terms of economic geography.

Furthermore, since economic geography requires detailed consideration of the natural features to which economic phenomena are related, a regional study in economic geography constitutes the greater part of a full study in regional geography. To put it simply, most people live where they live--rather than move elsewhere or die--not because they like the climate, the politics or the customs, but because they are able there to make a living; the manner in which they make a living, and consequently the manner in which they live in general, are, in large part, determined by the interrelation of economic and natural features which it is the function of economic geography to study [cf. Schmidt, 7, 162-200].

On the same basis we justified the major emphasis in geography on agricultural (in the sense of land-use) geography. The features of urban areas are, to a much greater extent, undifferentiated in different areas. Needless to say, however, this does not mean that the regional geography of even a predominantly rural area is complete if the cities are omitted or barely mentioned; the relations between the agricultural areas and the cities is of essential importance in understanding the character of the area.

On the other hand, in at least two major parts of the world--namely the two parts of greatest concern to European and American geographers--areal differentiation is represented in great part by the differential development, in intensity and in character, of urban development. For these areas m particular, geography is greatly concerned with the geography of manufacturing, upon which the urban development is largely based, and with the study of the cities themselves as the most extraordinary features of areas that man has produced.

The conclusion that physical and economic geography make up the major portion of geography as a whole does not for a moment suggest that the other parts are to be ignored. In the first place there are many features of economic geography that cannot be correctly interpreted without an understanding of their relation to areal differences in culture, in the narrower sense of the word, and in political organization. These are therefore geographically significant features, and, in order that their relationships to others may be known with certainty, they need also to be studied systematically.

That major areal differences in culture may be of great importance in relation to other geographic features is familiar to anyone who has traveled in eastern Europe, or has crossed the Rio Grande, or been to mid-latitude South America, not to mention the vast differences to be observed in areas of Sino-Japanese and Indian culture. The differences of culture of a secondary order, however, appear to have but minor effect. A careful study of the maps of crop production in Europe reveals the importance of a boundary running from the North Sea to the Alps, east of which rye and potatoes are far more important than in areas of similar natural conditions to the west; the line follows approximately the Franco-Germanic cultural (not national) boundary. The difference mentioned, however, is clearly minor. Likewise, the more obvious differences in architecture, in customs, etc., can hardly be regarded as geographic differences of the first magnitude. Consequently, a detailed systematic study of the geography of peoples, whatever its value for other purposes, would appear to offer a minor contribution to geography as a whole.

On the other hand, just as the importance of the geography of mining is found, not in itself, but in its significant relation to the far more important geography of manufacturing, so the geography of peoples, even in great detail, is of major concern to the geography of states. While the study of states--in the sense of independent political units--is perhaps but a lesser part of political science, the geography of states (See Supplementary Note 52) constitutes the major part of political geography. This conclusion results from a particular characteristic of the state that is often lost sight of in the discussions of the place and function of political geography as a part of the field as a whole.

As a social organization any political form, whether a state, a government, or other commonalty, is a feature that differs, to be sure, in different parts of the world, but the differences have but little relation to those of other features. The proposition that the mountainous islands of the Hellenic archipelago constituted the necessary background for the development of the democracies of the Greek city-states is a proposition in the geographic aspects of political science; it would not be suggested by a systematic study in the geography of mountainous islands or of democracies. If the "mother of parliaments" is located on a fair-sized island, close to, but separated from the mainland, an island of certain conditions of climate, relief and soil, her daughters appear to thrive in areas of radically different natural environment and of very different economic geography. In the study of political organizations considered from this point of view, geography, as Penck suggests, may have little place other than to offer supplemental suggestions to the political scientist [163, 51 f.].

On the other hand, the geographer has a direct interest in the state as a division of the earth surface. The fact that that division, as Penck insists, is made by man and is not inherent in the nature of the earth, is immaterial to us, since we have found that any division of the earth can only be made by man. Neither can we accept his description of the earth surface of a state as "merely the stage of man's (political) action, to be sure a stage that influences it"; such a description represents only the political scientist's point of view of the state-area. For geography, the state is an area in which certain conditions are universally true in contrast with those of other state-areas. It is, therefore, an area of homogeneity in certain very important respects, and so forms the simplest as well as the most definitely delimited of all geographic phenomena. Further, as we observed in our examination of the concept of regions as units, the state is the only area larger than a city which is organized as a functioning areal unit Whole (Sec. IX F). Unlike the abstract concept of the "region," the state area is, in many respects, a concrete unitary object; it is a piece of the earth's surface sharply defined and separated from other pieces, with which its relations are, in many respects, the relations between whole units. Like other concrete objects, the state-area has size, form, and structure. Indeed, it is in the consideration of the structure of a state that our concept of regions becomes of practical rather than merely academic importance.

To many geographers, particularly in this country, the concept of the state as an areal unit may appear remote from geography, not merely because it can be observed in the visible landscape only with difficulty, if at all, but also because of a wide-spread impression that man's political structure is something extraneous, essential neither to the earth nor to man. On the contrary, as Schlüter has noted, "the state (in the widest sense) is no younger than man, but rather older. Man could not become a human being without the protection of an association which contains the seed of the state. Man has been from the beginning the zoon politikon of Aristotle" [134, 410]. For man, as a political animal, it has been just as natural to create a state as a farm or a city [cf. Vogel, 271, 5], and the state which he creates, as East concludes, "whatever else it is . . . is additionally and inevitably a geographical expression and as such forms part of the subject matter of geographical science" [199, 270; cf. also 216, 802-4].

Since there has been considerable controversy over the relation of this part of geography to the field as a whole, we may examine briefly the significance of the state-area as a subject of geographic study. The notable differences in size and form of the state-areas of the world is one of the most obvious facts of areal differentiation of the earth surface. It is, therefore, one of the characteristics of the world which needs to be understood in a subject devoted to the study of the differences in different parts of the world. That it could be considered as a fact of minor importance is refuted by the reality of the power of states in controlling not only the political, but also the economic life within their areas. Finally, this important fact of areal differentiation is significant in geography if it is significantly related to other features of areal differentiation.

It is of course false to conceive of the division of any part of the earth into state-areas as "natural"--i.e., as determined by the natural conditions; since the phenomenon itself is cultural, neither the state nor any of its elements, such as boundaries, can be natural [357]. But one cannot compare the political map of Europe or Asia with the corresponding relief map without realizing that there are close relations between the two forms of differentiation. Even where the political map may seem highly arbitrary, as in South America, a consideration of the map of population density, and then of the maps of natural elements that explain it, will show that it is only the outer boundaries that are arbitrary; the essential division into states is very significantly related to the natural geography of the continent.

The state-areas differ not only in the obvious features of size, form, and locational relations to each other but also in their internal structures. The differences in structure in different state-areas is directly related to the regional structure of the area in terms of physical, economic, and ethnological geography.

In the reverse direction, the economic differentiation of the world with which we are concerned in economic geography is notably affected by the efforts of all states--more marked in some than in others, but present in all--to organize all the cultural features of its area into a more or less homogeneous and closed unit.

The most obvious product of these efforts is tariff walls, which, though visible only in the minute form of boundary stones and frontier stations, may have a greater effect on the geography of production and trade than the highest mountains or the widest oceans. Many of the less obvious effects are perhaps, in total, even more important. Thus, the regional geography of the Paris Basin can in itself provide but a minor explanation of oneof the most important features within it, the city of Paris, in size and character out of all proportion even to the large fertile plain of Northern France. Only the successful effort of the state of France, in past centuries, to bring all its regions, from the North Sea to the Mediterranean, into an organized unit with Paris as the center can account for that particular geographic phenomenon. Obversely, the agricultural and industrial development of each of the regions of France has been notably affected by its inclusion in this political-economic areal unit.

The position of political geography, as primarily the geography of states, is therefore not to be considered as a remote peripheral location, but rather as one in close relation to the major aspects of geography. Onthe other hand, the special field of Geopolitik, which has had such a notable development in post-war Germany, represents a very broadly defined--or quite undefined--field in which geography, in terms of political geography, is utilized for particular purposes that lie beyond the pursuit of knowledge. It represents, therefore, the application of geography to politics [Hassinger, 165, 23] and one's estimate of its value and importance will depend on the value that one assigns to the political purpose it is designed to serve. Since it is designed to serve national politics from the German point of view, its positive value from that point of view may be considered as off set by its negative value from the point of view of other countries. In its influence on the world's thinking, this branch of geography, which produced the concept of the "Lebensraum" of the state, would appear, at the moment, to be the single most influential branch of geography. (The history and methodological problems of Geopolitik, as well as of political geography, proper, have been treated previously by the writer [216]. Since then, Ancel, in France, has attempted to rescue the term geopolitics (Geopolitique) for international science [187], and East, in England, has published a significant study [199];the most comprehensive study of political geography is that of Maull [157].

When one considers the theoretically possible field of sociological geography from the point of view which we have been following, it may appear doubtful that the development of studies of that character can make important contributions to geography. In areas of primitive development one might study the geography of clothes, or implements, or conceivably of manners and customs, and religions. For the important parts of the world, however, such studies would apparently have little geographic significance. Men wear hats in Chicago that allow their ears to freeze because the winters are not cold in London. The urban architecture of Midwestern United States is significant chiefly in revealing the lack of indigenous culture. Areal differences in religion are of some importance in political geography and perhaps in a few cases in economic geography, but in general this is a phenomenon that shows relatively little relation to other features of areal differentiation. Even if we take the case most commonly cited, that of Mohammedanism as the religion of the nomads of the steppes, we find that it flourishes also under tropical rains in the intensively cultivated paddy fields of Bengal or Java.

Prejudgments of future developments in science, however, are among the most dangerous of predictions. Many aspects of culture, other than those of economic and political facts, are significantly related to regional differentiation. Certainly it would be unwise to dismiss this field on the basis of examples selected to prove the lack of significant relation. The studies of settlement forms, house types, etc., that have been made in Europe and in some parts of this country--and with particular enthusiasm recently, I am told, in Japan--indicate the possibility of systematic contributions to geography as a whole (cf. particularly Schmidt [180, 54-80]; see also the previous discussion in Sec. VII F).

In each of the specialized branches of geography which we have discussed, the pursuit of knowledge is dominated by the same ideals that we found applicable to geography as a whole. The degree to which these ideals can be attained varies in the different fields as it does in the corresponding fields of systematic science. In general, no doubt, the degree of accuracy and certainty may be highest in physical geography and decreases in the various branches of cultural geography, more or less in the order in which we discussed them. Such a comparison however is by no means universally true. Some economic facts can be established with far more accuracy and certainty than many facts of natural phenomena, and few facts in geography can be established with such a high degree of certainty and exactness as those concerning the extent of state-areas (See Supplementary Note 53).

In the development of universals the greatest contrast in geography is not found among the special fields but rather between systematic geography in all its branches, and regional geography, whether partial or complete. It is particularly on this account that the separate development of systematic geography is so important, both in itself and in its relation to regional geography.

The division between these two different points of view is made necessary by the very nature of geography. The areal differentiation of the earth surface, which geography is to study, is a differentiation expressed in a great number of individual features whose variations are in part related to each other, in part independent. Geography, therefore, must study the areal differentiation of each of these features over the world, not as a part of the systematic study of that particular object, but as a study of one form of areal differentiation of the earth surface; this is systematic (or general) geography. It is clear, however, that a full understanding of the differences between areas cannot be obtained by simply adding together the appropriate sections of systematic geography. It is necessary to study the totality of interrelations of all geographic features found together at one place; this is regional geography. In the historical section we saw that this difference between the two points of view within geography, first stated by Varenius, was present in the work of Humboldt, and formed, as Wagner noted, an inevitable dualism in the field--not in respect to materials, but in respect to methods of approach. More recently the philosopher Kraft, in examining our field, has substantiated this statement [166,4 f.]. In general, in modern geography there appears to be agreement on the distinction between the two points of view, as outlined by Richthofen [73] and following him, more clearly, by Hettner [140; 161, 218, 398-403; cf. also Penck, 163, 44]. The relative importance of the two aspects, however, remains as it has always been, a subject of continuous controversy.

The relation between the two methods of organization has been illustrated by Hettner in the following manner. If we consider the variations of the geographic elements as though contained in surfaces parallel to the earth's surface, then all of them together would form a series of surfaces at different levels above the earth's surface. In any particular part of systematic geography, one studies all the variations in a single surface, in relation, it may be, to variations in the surfaces of the other elements. In regional geography, however, we strike a limited section vertically through all the surfaces in order to comprehend the totality of their characteristics in a single area (See VanCleef's diagram of this simile, as anologous to stratigraphy [433]) .

These two methods of organizing geographic knowledge are not only interconnected in every part but are also by no means so distinct in practice as is frequently supposed. A systematic study need not cover the whole world but may be limited to a continent or to any area within which there are variations in the feature studied. (Consequently the term "general geography" is unfortunate.) If, then, all the features of a small area are studied systematically in succession, the mere addition of the total series of these systematic studies does not form a study in regional geography, but simply the systematic geography of a limited area. The essential difference is in the point of view. In regional geography the study is focussed on the individual localities or districts, which, whether smaller or larger, are conceived (arbitrarily) as units. The purpose then is to comprehend the particular geographic character of each of these units--that is, the particular manner of formation of all the geographic factors in their causal connections [Hettner, 126, 672].

As we have implied a number of times, the terms inherited in geography for these two divisions of the field have been found unsatisfactory by many students. For work in which individual categories of phenomena are studied over extensive areas, or the whole world, Varenius' term "geographia generalis" is almost universally used in Germany (allgemeine Geographie) as well as in France (géographie générale). As Hettner and others have found, however, this is misleading, not only because such studies may be limited to but a part of the earth, but particularly because they consider the features of the earth surface by individual categories. Richthofen at one time suggested the term "analytic geography" [73, 41], but Hettner rightly objects that both analysis and synthesis are required in both forms of geography [126, 675; or 161, 400]. Schlüter nevertheless appears to have adopted the term [247]. In his earliest considerations of this question, Hettner, we noted, endeavored to express more clearly the close relation of the two aspects of geography by using in place of "general geography" the term "vergleichende Länderkunde"--taken, he tells us, from the title of a course given by Richthofen. Though he did not use this term in his subsequent methodological treatments, Hettner has recently returned to it as the title for a comprehensive work covering the whole of general geography, except for the omission of the seas [363]. Whatever conclusion German students may ultimately come to, American geographers will no doubt agree with Penck in finding this usage unfortunate [90]. On the other hand, the term now widely used in the American literature "systematic geography," finds ample precedent in the writings of many German geographers. Thus Richthofen described this form of geography as organized "auf Grund systematischer Principien" [73, 41], Hettner called it "die systematische Darstellung" [126, 675], and Penck has called it the method "nach systematischem Gesichtspunkt"[163, 44]. Unfortunately the fact that the term is not of Germanic origin, and its German equivalent is unsuitable, apparently excludes its use as the title for a major part of the field of geography in that country.

The other term introduced by Varenius, "geographia specialis," has been very largely replaced, in Germany by "Länderkunde," or more recently by "Landschaftskunde" (ignoring here any differences between the two), and in practically all non-German lands by the appropriate form of "regional geography." Any number of German students however have found "Länderkunde" unsuitable, both because it excludes the seas and because it suggests areal divisions much larger than those now commonly studied as regions. As early as 1831, Fröbel suggested "region" in place of "Land" [54], Supan wished to return to "Spezialgeographie" (or "spezielle Geographie") [78, 153] Waibel would prefer either that term or "regionale Geographie" [266, 198] and Lautensach favors "regionale Geographie" [173, 29 f.]. Since neither of these terms, however, is of German origin, there is little likelihood that German geographers of the present period will change to them.

The fact that the field of geography may be divided in two different, and intersecting, directions frequently leads to confusion. We may clarify the situation by classifying a few examples; if we limit these to studies by the present writer there will be no danger that anyone be offended. The study of types of political boundaries is a systematic study in political geography in which the attempt is made to construct generic concepts and to suggest some general principles [357], whereas the study of boundaries in Upper Silesia is a partial regional study--regional political geography [355].Similarly the study of the Upper Silesian Industrial District is a partial regional study limited largely to a part of the economic geography of the area [356]. Finally, the study of Austria-Hungary is a study in regional political geography of a past period, that of the beginning of this century--i.e., a study in historical geography, limited to political regional geography [338].In other words, we recognize no "boundary between economic and regional geography," such as Pfeifer apparently would have us observe [109, 108]. Each overlaps the other. But likewise it follows--as possibly he intended to say--that regional geography is not to be thought of as complete if it is limited to economic regional geography.

Although astronomy and geography are both concerned with the study of spatial integrations--things related to each other in space--the character of the spaces that they study and the things interrelated within them are so completely different that no amount of logical similarity should lead us to expect similar results in the developments of the two fields. For much the greater part of his work the astronomer may consider celestial space as extraordinarily simple, consisting on the one hand of homogeneous ether--which in much of his work, he may regard simply as empty space--and on the other hand of masses of inanimate matter, most of which are sharply defined units widely separated and therefore related to each other essentially as whole units. Within any celestial region--e.g., that occupied by the solar system--the problem of integration is little, if anything, more than the systematic problem of the relations between these unit masses, a problem primarily concerned with their effects on each other's position and motions, studied in terms of but two factors, gravity and free motion in space. Even within systematic astronomy, however, not all the findings can as yet be reduced to scientific laws. Though the motion of the sun in reference to other stars has been measured, the laws determining that motion have not been constructed, possibly never can be.

Astronomy does include studies corresponding to those of regional geography. These are represented most clearly by the detailed examination of those units in the solar system that are near enough so that differences between different parts may be observed. In the same category, though different in character, are studies of the groupings of stars in our universe, and the detailed examination of individual stellar nebulae.

If astronomy is largely concerned with systematic studies because of the relative simplicity of its subject matter, exactly the reverse is the case in the historical sciences. Of this group, historical geology shows most clearly the distinction between the systematic and the periodic approach. The study of climatic changes in past ages, of changes in mountain development, and, in general, the changes in the continental landforms, or the evolution of the horse--all represent systematic studies in the history of the earth. In contrast are studies which attempt to provide a generalized picture of associated phenomena of climate, landforms, and vegetable and animal life of the Upper Mississippian or any other past period in the history of the earth.

The comparison to which we have most repeatedly referred throughout our discussion of the nature of geography is, of course, that with history in the ordinary sense of the history of "historic times." Various students, however, have suggested that the comparison can only be related to regional geography, that history lacks systematic studies [cf. Penck, 158, 48-50].

We should not be misled by the fact that history is commonly taught only in terms of what we may call "periodic history." In their research problems historians frequently concentrate on the development and changes in some very restricted group of phenomena through a succession of years. Such studies may treat the development of a particular form of constitution, the growth of labor legislation, the changes in the price of wheat in England, or the development of roads in Minnesota.

Nevertheless, so far as an outsider may judge, the work of this character has by no means the importance to the field of history as a whole that systematic geography has to geography. In particular it has not yielded to history generic concepts and principles that are nearly as definite as those developed in systematic geography.

If one compares the particular problems studied in the two fields, as shown in their publications, it is obvious that historians are concerned with phenomena whose interrelations are far more complex than those commonly studied in geography. The logical basis for this difference is not so obvious; indeed our fundamental assumption of the relation of the two fields leads logically to the conclusion that the same phenomena may be studied in each field: history may consider areal phenomena and geography may consider historical events.

Neither history nor geography, however, need consider all the phenomena that are found in the sections of reality which they study, but only those phenomena which differ significantly in different sections of time or space, respectively. In each case the attention is chiefly focussed on those phenomena which differ most and whose differences are most significant to the total differentiation. In the total reality with which both history and geography are concerned--namely the phenomena of the world in historic times--there is one major group of phenomena, the natural phenomena, which are causally of fundamental importance to all the other phenomena, but which, while differing markedly in different areas of the world, differ but slightly in different periods of historic time. This, of course, constitutes a great difference between history, in the narrower sense, and pre-history, not to mention paleontology.

In consequence, the areal differences that are of greatest importance in geography are either differences in the natural features themselves or in cultural features which are closely related to the natural features. We would have a similar situation in history only if such features as climate and landforms had varied as radically through historic times at the same place, as they vary over the world at the same time. In other words, if the natural environment of England since the time of Caesar had varied from humid to arid, from polar to tropical, from plains to mountains, the agricultural history of England would represent the most important part of its history, and history would long since have developed the systematic branches of climatic history, landforms history, etc. Indeed, if Ellsworth Huntington's thesis of the historical importance of even minor variations in climate should be substantiated, it would be not only logical but necessary for history to develop a systematic study of climatic history--the study of the relations of climatic changes to other historical features.

In any case, the exceptional character of the example just cited tests the rule: the relative fixity of natural conditions during historic times results in a notable degree of constancy in those cultural features which are most closely related to natural conditions. The manner of land use in any area may remain much the same for centuries, in China, for millenia. Cities do not pass through a generic process of youth, maturity, and old age to death; they may continue in approximately the same condition for indefinite periods of time.

Consequently the phenomena which show the most notable differences in relation to time are cultural phenomena less closely related to natural conditions--commonly, therefore, phenomena of much more complicated character--such as manners and customs, political organization, inventions, etc. Furthermore, not only are these phenomena in themselves more complex than those with which geography is most concerned, but their interrelations through different periods of time are more complex than the interrelations of the principal geographic phenomena in different areas. Indeed, in most cases the character of one period of history, largely determines the character of the next, whereas the character of one area in geography has commonly but minor effect on the character of its neighbors. It is not surprising, therefore, that historians are more clearly aware of the fictional nature of their divisions of time than are geographers of their corresponding divisions of area.

On the other hand there are some sudden breaks in historical development that produce changes almost as great as the change from sea to land ingeography, namely, when new discoveries or inventions, or the migration of peoples, introduce a new culture into an area. The frontier of settlement in America during the past several centuries was not only a line marking great geographic contrasts but, as it passed through any region, it represented an historical revolution in the adaptation of man to nature. The historian of this revolution, therefore, must understand the principles governing the relation between cultural and natural features in order to study history. Much the same is true of such historical problems as the industrial revolution, and the associated agricultural revolution in Europe.

The historian who is concerned with these problems, involving less complex features than those that form most of the material of history, presumably will not hesitate to make systematic studies wherever possible and to explain the relationships where he can. The fact that most historical events may be too complex to permit of definite explanation should not lead to a dogma that no historical events can be interpreted. Unfortunately, however, situations comparable to those mentioned above, in which the fundamentals of man's adaptation to nature are notably changed, are relatively few in history and most of them took place at such an early date that the historian has scanty reliable data from which to study them. Thus, a systematic study of the history of "frontiers"--in the sense of a border of progressive settlement--should consider not only the frontiers in the New World and in Siberia, but also the frontier of German settlement in central Europe in the Middle Ages and the still earlier frontier of Anglo-Saxon settlement in Great Britain. It is obvious that, even were data available, such a problem would be extremely complicated, since it involves not merely different periods of world history but also different areas of the world of radically different character.

In general, the problems which must be handled in systematic studies in history are far too complex, and involve factors too difficult to observe and measure, to permit of the development of generic concepts and principles similar to those developed in systematic geography. There are, to be sure, some students of history--chiefly non-historians--who assume that it is possible to develop scientific laws concerning the rise and fall of states, the causes of revolutions, or the development of particular social movements, but their theses are more notable for the ardor with which they are advanced than for the evidence which has been brought to support them. Most professional historians are sceptical of the possibility of developing a systematic history in which the phenomena important in history may be classified in generic concepts leading to principles. The rather naive belief of some geographers that geography can provide this deficiency in history has not, as yet at least, been substantiated.

This contrast, therefore, between history and geography results from the fact that the interrelations of the phenomena that vary most notably in historic times are far more complex than the interrelations of the phenomena that vary most notably in the earth surface. It does not affect the logically common nature of the two fields, as sciences that attempt to integrate phenomena as they are found in reality.

With the increasing interest, in both Germany and America, in full regional studies, necessarily including cultural as well as natural features, geographic research suffers not only from the lack of adequate foundation in systematic cultural geography but also, as Broek notes, in the lack of training of most geographers in the social studies [108, 252]. If students are to be prepared to make full studies in regional geography--and there is fairly general agreement that all geographers should do at least some work in this field--it would be logical to require supplementary training in the related social sciences no less than in the related natural sciences. This is the case in few, if any, of our departments of geography. Since, in addition, the academic relations of geographers are commonly closer to geology and other natural sciences than to the social sciences, the individual geographer knows that his work will be subjected to careful criticism so far as it touches natural science. For the same reasons he has been free to indulge in almost any sort of economic theorizing or political speculation that occurred to him, with little or no risk of being called to account.

On the other hand, the relatively recent shift from the emphasis on a geomorphology closely related to geology to the emphasis on human geography, has resulted in a tendency on which Krebs has very recently commented: much material that is essentially economic, historical, or sociological is taken over and presented without digestion in presumably geographic studies [91, 244]. American geographers appear also to be aware of the need for independent research to develop concepts and principles of systematic cultural geography--indeed there is evidently a widespread feeling that this is the single most pressing need in geography today.¹⁰¹

It is desirable, therefore, to consider carefully the distinction between systematic studies in geography and the studies made in the related systematic sciences. Especially is it necessary, as Schmidt insists, for those who work on the borders of geography to keep the distinction in mind. Though they must be familiar with the concepts and methods of the neighboring sciences and may use these in their work, they must use them for purposes dictated by the point of view of geography as distinct from that of the related systematic science [7, 162 ff.]. In particular, Schmidt has contributed a very thorough and valuable study of the relations between economic geography and economics, as well as a detailed study of systematic economic geography [7; 386].

The divisions of systematic geography, as we noted earlier, correspond to the divisions of the systematic sciences and there is inevitably close relationship between each branch of systematic geography and the corresponding systematic science. This relation is not accurately expressed by the phrase "neighboring sciences," since geography is not a branch of science situated beside the systematic sciences, but represents a point of view in science which cuts through all the systematic sciences (Sec. IV A). There is, therefore, no line separating systematic geography from the systematic sciences, but there is an essential difference in point of view which must be maintained by the individual geographer who wishes to do geographic work rather than work in some other branch of science.

The distribution of a particular kind of phenomena is significant both in geography and in the systematic science concerned with that kind of phenomenon, with this difference: in geography the focus of attention is concentrated, not on the phenomenon--one of whose aspects is its distribution--but on the relation of that distribution to the total areal differentiation of the world.

This contrast in point of view may be illustrated in the case of the production of corn (maize). The totals of production by countries and the resultant effects on national and international markets are presumably of concern in economics, but not in geography. (One may be confused here by the fact that economists have generally been willing to leave instruction in the geographic aspects of economics to economic geographers, with the result that geographers have frequently attempted to do research in what is essentially a part of economics.) Likewise the relation of annual variations of corn production to annual variations in rainfall is of great concern to the student of agriculture but is not of direct concern in geography, whereas the fact that the variations in rainfall in Nebraska have a greater effect on corn yields, than the same degree of variation in Pennsylvania, is of geographic concern. Geography is concerned with the marked areal variation in corn production, since this represents a part of the total areal differentiation, in which it is associated on the one hand, in its relation to differences in climate, soil, relative location, or cultural conditions and on the other hand, in its relation to differences in the total crop-livestock element-complex, the character of barns, the presence of grain elevators, etc.

In other words, the facts of distribution of corn production are not in themselves "geographic," not even when shown on a map. It is what is studied concerning those facts that is significant for geography. Merely to describe and analyze the facts of distribution of physical and social phenomena found in different areas is to produce a compendium, not a geography, either systematic or regional. The facts concerning the areal differences in these phenomena must be studied in their areal relations, that is, their significance to the area as determined by their relations to other phenomena of the same place, and by their spatial connections with phenomena in other areas.

In a study in systematic economic geography, for example, unless the geographic point of view is clearly maintained from the beginning, the work may turn out to be a study in the geographic aspects of economics. The reason for this, of course is that both of these kinds of studies start with the same step, namely, the establishment of the facts of distribution of the particular studied. As this first step focuses the attention of the student on the phenomena themselves, it often results in his continuing the work with that point of view, this producing a study of those phenomena--i.e., a study in a systemic science.

Because the botanist or economist is concerned with the location of his phenomena in only some of his studies whereas the geographer is always concerned with the location of facts, it is often supposed that the determination (an interpretation) of the "Where" of things is exclusively a function of geography, if not the whole, of geography--i.e., geography as the science of distributions. But it would be both presumptuous and contrary to what actually takes place for us to claim that the zoologist, geologist or economist concerned with the distribution of his phenomena must look to the science of geographers for the answers. Likewise the fact that in such studies the students of other sciences may use the geographic technique of mapping does not make them geographers; the economist or political scientist may often use the historical method in determining the "When" of past events, but their work does not thereby become history. Similarly all the students of the systemic sciences may use geographic methods in presenting the distribution of their phenomena--whether particular kinds of plants, animals, or factories--without depending upon the science of geography (see Sec. III D).

In the reverse direction, geographers have, in fact, long been accustomed to looking to certain kinds of facts. For the location of mineral deposits and different kinds of surface rocks, we depend upon geology; for the occurrence of soils, on soil science; for the distribution of native plants and animals, on botany and zoology. As Hettner, in dependence on Wallace, has insisted, these latter represent geographic studies in botany and zoology, as distinct from studies in plant and animal geography in which the interest is focused on area, studies in terms of their plant and animal contents. It is only in those fields in which the systematic sciences concerned have given little attention to the geographic aspects--notably in economics--that the geographer has been forced to do his own spade-work in determining distribution. It is significant, however, that once geographers had introduced their technique of mapping into the study of domestic crops and animals, agricultural economists have taken over this work as an integral part of their field.

The study of the distribution of phenomena presumes a classification of objects into types. In many cases the objects are sufficiently simple so that the classification is both obvious and acceptable to all the sciences concerned--e.g., the classification of cultural plants into different kinds of crops: corn, oats, wheat, etc. If less simple phenomena are involved, however, we noted that the classification will depend on what aspects are selected as most important for the particular study. Consequently two sciences concerned in studying that the distribution of the dame phenomena may differ, even in the presentation of facts of distribution; though this difference may in itself slight, it may be of major significance in later staged of study.

One should not forget, however, that economist of effort s an axiomatic desideratum in scientific work. In any case where the classification and establishment of facts concerning the distribution of phenomena that have been developed in another science are found to be suitable for the purposed of systematic geography, there is no call for the geographer to do the work over again in a different way. Nevertheless, as we noted earlier in the consideration of relatively complicated phenomena, such as land forms, the facts established for the purpose of another science have been found not to be the facts needed on geography; consequently the systematic study in geography must begin anew at the first step (Sec. XI D).

Any presentation of facts in science calls for interpretation. Consequently, geographers have often presumed that, in presenting the facts of distribution of any phenomenon it was also the function of geography to study causes of that distribution. But in every branch of science facts are presented and utilized whose interpretation is the function of some other branch of science. In this case, namely the distribution of any phenomenon, does interpretation of the facts of distribution fit logically into geography or into the systematic science?

This question is not raised here in order to argue over the location of a borderline between sciences, certainly not with the idea of establishing and rules of conduct for geographers. In considering this question we will, I believe, come to a clear understanding of the whole relation of systematic geography to the systematic sciences.

One point appears clear. Whichever student is to interpret the distribution of a particular phenomenon will study that phenomenon in terms primarily of those aspects which indicate its causal development. If the distribution be measured in terms of other aspects they must first be referred back to the genetic aspects in order to provide interpretation. One of the essential contracts between geography (and history) on the one hand, and the systematic science on the other, is that the former are interested in the integration of phenomena, the latter in analyzing the processes of particular kinds of phenomena. The fact that studies of processes involve the time element does not make them history, as Kroeber has emphasized [116, 545 f.]; neither does the fact that distribution involves the element of space make its study a par of geography. The explanation of the world distribution of a particular kind of phenomena would appear to be an end resultant of the study if the processes of development of that phenomenon; it is therefore the proper subject of study in a systematic science. In systematic geography, however, it represents the world picture of an element with which one is concerned in its functional relation to the differential character of the areas of the world. In other words, though geography must know where things are, the study of the "Where" is not geography nor an integral part of geography, and it is therefore not the function of geography to explain the--that is, to give the full explanation of why a phenomena is found where it is found. Consequently, systemic geography is free to overlook generic concepts based on genetic aspects of phenomena in order to develop generic concepts based on aspects that are functionally significant.¹⁰²

Although we conclude that it is not the function of the geographer to explain of any phenomenon, it is at the same time clear that he may be concerned with such an explanation in order to interpret the relations of that phenomenon to other geographic phenomena. For example, in the geography of soil, the interpretation of the relation of the soil of any area to the character of its climate and bedrock, necessitates an understanding of the whole development of soils; but it is the function of the soil scientists to provide the explanation of soil development on term s of all its factors and processes.

The systematic geography of any particular phenomenon depends we conclude, for the principles governing its distribution, on the systematic science concerned with that phenomenon. In many cases, however, the geographer may find that the students of the appropriate field have not been interested in developing such principles. In such cases he can hardly be expected to wait indefinitely, but may have to undertake the study himself. If, however, he does that without realizing that he is shifting his point of view, he may later discover that he has definitely passed over into a field in which he may not be adequately prepared.

Since the previous paragraph might seem to be pointed at individual geographers, it may be appropriate to illustrate it from the writer's own experience. Geographers had long recognized that the concentration of iron and steel mills in certain area was somehow related to the presence of coal and iron mines in the same or other areas, and every text in economic geography attempted to state that relationship. None of these statements, however, were found to be adequate and the reason for this was, no doubt, the failure of students of the economics of industry to study the problems of the distribution of iron and steel mills. The geographer wishing to interpret the character of areas found to have, as one of their major characteristics, intensive development of this industry, must be able to explain the relationship of these factories to other geographic features. The writer, therefore, undertook to develop the principles governing the location of the iron and steel industry [352], on the basis of which that industry could be studied as a part of the systematic geography of the United States [353]. The interest developed by the fist study led the writer into similar studies of the principles governing the location of other industries, and of industry in general. It has since become clear to me that only in the study of the iron and steel industry in the United States was my attention focused on the areal significance of the industry--as a particular characteristic of certain areas; in the others the center of attention was the industry as a phenomenon of which one aspect, namely its location, called for explanation. It is not surprising, therefore, that any interest shown in these studies has been confined almost entirely to economists.

The conclusion which the writer has drawn form this personal experience may have general application. The study of the "Standorts" problem--the determination principles governing location of units of production--not only requires more training in economical that in geography, but also requires more training in economics than in geography, but also requires a full concentration of interest on the problem for the sake of the problem itself, rather than for the sake of the results; it is the economist who is interested in the problem, the geographer in the results [cf. Tiessen, 160, 8].

On the other hand it might be claimed that, regardless of a logical division of problems between the sciences, geographers had in fact, developed this particular subject to such an extend as to justify their retaining it as a part of their field, by right of cultivation, regardless of the logical division of work in science [cf. Kraft, 166, 7]. Undoubtedly geographers have made contributions to the location economic activities, but we have hardly pursued the problems with sufficient consistency and system to register a valid claim based on thorough and successful cultivation. The world of knowledge as a whole does not look to us to supply the principles governing these phenomena.

Confirmation of this conclusion may be drawn from critical survey of the work of geographers in this field made by Swedish economist Palander, in the introduction to his exhaustive study of the theory of the Standorts problem [372]. It is only fair to add, however, that the geographers entered this field only because the results which they needed in their work had not been adequately developed by economists. American economists in particular have shown little interest in this field and at the time the writer was concerned with it were hardly aware of Weber's work [396], which, in any case, Palander has shown to be impracticable.

Geographers, therefore, will welcome the attention which economists are now giving to this problem. Geography will not be merely receptive in this relation, however, for even though we may agree that the problem comes logically under the point of view, and for which requires some understanding of the geographic point of view, and for which geography can continuously contribute positive materials and effective criticism.

Since much has been said of the dependence of geography on the systemic sciences, we may appropriately note one or two significant suggestions that geography may contribute to the problem of interpreting the distribution of economic features.

The first step toward an interpretation of the distribution of any phenomena is, of course, to portray that distribution. Students of the systematic sciences who have acquired something of the geographic point of view, will realize that the only language in which the location of things on the earth's surface can be portrayed intelligibly is the map, and that reliable interpretations require maps more detailed than cartograms of units as large as our States. Although this proposition is axiomatic in geography and geology, and is now thoroughly recognized in agricultural economics, in other branches of economics it is frequently overlooked. When economists attempt to interpret the location of the steel industry in the United States in terms simply of the amount of development on the States of Pennsylvania, Ohio, Indiana, and Illinois, it is not surprising that they should reach a defeatist conclusion as to the possibility of principles of location. Even though our census figures by countries are less complete than those by States, they must be used to gain an approximately accurate measure of the development of Southeastern Pennsylvania and Maryland, the Pittsburgh-Youngstown region, the Lake Erie Ports, and the Calumet District [353]. Likewise on cannot hope to interpret the contrast between the industrial development of Wisconsin and that of Minnesota until one has seen the facts portrayed on a detailed map and observed that the concentration of distinctively manufacturing cities is not to be considered in terms of Wisconsin versus Minnesota, but in terms of proximity to the west shore of Lake Michigan [326].

To the student of economics who has not been trained in geography, even a detailed map of distribution of an economic feature may appear to present a comparatively simple problem in comparative location--that is, he is apt to think almost exclusively in terms of relative location, considered purely geometrically, and to ignore other variants of areas. Thus, in many economic texts the consideration of the distribution of different types of agricultural production has long been dominated by Thünen's simple picture (constructed, we may note, by a writer living in the relatively homogeneous North German Plain) of concentric belts of differential production surrounding a city center [for an outline of the theory, see Jonasson, 313 (1925) 284-6]. Whatever validity this analysis may have had in earlier periods has been largely destroyed by the development of modern commercial facilities that have made relative location a factor of secondary importance in determining land use. O. E. Baker showed some years ago, that with such well-known exceptions as market-gardens and fluid-milk farms, the location of different types of agriculture production is far more dependent on climate, relief, soil, and drainage than on relative location; this writer has demonstrated the fact in detail in the agriculture of Europe [324]. On the other hand, this does not mean that the factor of relative location may be entirely ignored in such problems, as is often the case in studies in agricultural geography. (Waibel has recently considered Thünen's law in full detail in the light of the radical changes in conditions since it was first stated [395, 47-78].)

This discussion of the importance of the geographic point of view to the problem of interpreting the distribution of any phenomena may appear to suggest that, in spite of the logic of classification, the geographer is vest equipped to handle the problem. Before any geographers accept that conclusion, they should first examine the specific problem treated in Palander's master treatise--in particular the enormous amount of economic detail required in the handling of the transportation problem; they should observe the technique of economic analysis developed in a study of the location of manufacturing in the United States by Garver and associates--even though that study provided the examples of lack of map-mindedness to which we referred above [347]; and finally they should consider the studies made by economists who have had some geographic training, for example, Hoover's study of the boot and shoe industry [364].

All that we have shown is that the study of geographic aspects of any field of science, such as economics, requires something of the geographic point of view. That historical problems in economics require something of the historical point of view requires no emphasis, because all economists have no doubt had training, in one way or another, in history; but relatively few have in geography. Every science, Schmidt concludes, "that concerns itself with the study of the areal distribution of its objects on the earth is necessarily led to the geographic method; it must interpret the differences of its objects in relation to area, and so must make use of the method of thinking in geographic comparisons as one of the most important means of attaining general concepts in its own field and of penetrating into the character of the scientific objects of its own science. Thus, every research worker in economics must be a geographer [in the sense that he must use the geographic method] whether he will or no; the sooner he wills it and knows it, the better for him and his research" [7, 4].

In brief, we conclude that, both in terms of the logic of the classification of the sciences, and in terms of the professional equipment of the students--in techniques and in knowledge of the literature--the problem of principles of distribution of economic phenomena can best be studied by the student who is primarily an economist, but it is necessary also that he be in some degree a student of geography.

To avoid any misunderstanding it may be necessary to add that throughout this discussion--indeed throughout this paper--the term "geographer" is to be understood as an abbreviation of "student of geography." Any individual person may presumably be simultaneously a student of geography and a student of economics and may study wherever he finds himself interested and believes himself competent. Undoubtedly individual geographers have made, and may continue to make, important contributions of thought to the work in related fields. Indeed, such personal interconnections between the different fields of science are not to be regarded as merely permissible, but rather, as Penck properly insists in his own defense, are greatly to be desired [147, 124 f., 134]. If this is true of science in general, it is particularly true of geography, which not only is related to other sciences along border zones that "would be left fallow if scholars always limited themselves to a single science," but in every part of its field intersects the studies of the various systematic sciences. It is fortunate, therefore, that "the boundaries of the sciences are not insurmountable walls," as Penck writes in discussing the same question later, and one might ask him to modify slightly the analogy that he does suggest, of "the boundaries of states, which one can cross if one has the necessary pass, in our case, capacity." Perhaps Professor Penck's own experience on certain international frontiers would persuade him to agree that there should be no border guards along the boundaries of science, but that each student is to be his own judge of his pass, subject always to the ultimate verdict of those qualified to judge in the field into which he crosses. In any case, all will agree that "he who works across the border areas of geography must be able to ride in several saddles" [90, II, 36]. Further, as Penck has elsewhere indicated, the requirement that the student should feel himself competent wherever he works, requires that he should himself know in what field he is working at any time. And in order that geography may maintain clearly its own fundamental point of view, any cross-fertilization should be recognized for what it is, and not accepted as an extension of our field.

On much the same basis we may be permitted to dispose of the difficult problem of the relation of geomorphology to geography--without attempting to solve it (See Supplementary Note 54). This question has long been a matter of controversy, particularly in the English-speaking countries. As early as 1908, Chisholm, in agreement with Geikie, expressed the view that if the study of landforms follows genesis it leads to geology, and many others have echoed that view since. At the same time, however, a large part of the work in that field--in America, if not in England--has been carried on by geographers--particularly of course as a result of the work and influence of Davies [cf. D. W. Johnson, 103]. In Germany, the course set by Richthofen, and followed particularly by Penck and his students, has made geomorphology so definitely a part of the field of geography that few, if any, question its permanent inclusion. (In the Netherlands, in contrast, geographers apparently distinguish more clearly between geography and geology in this field [92, 294].) Examining geography from the point of view of knowledge as a whole, the Erkenntnistheoretiker Kraft concluded that the study of geomorphology, including the genesis of landforms, disrupts the logical unity of the field, but that, as a result of historical evolution, this field is in fact included in geography--in Germany at least [166, 7]. Its inclusion cannot, therefore, be questioned, he concludes--so long as, one may add, the geomorphologists continue also to be geographers. In other words, the geographer in Germany is, by his training, a geomorphologist as well, and geography, therefore, as a division of labor within the sciences includes that special field.

Whether the same conclusion, based on the history rather than the logic of the field, applies in this country, the writer would not attempt to judge. It is important to not, however, that the close association of geomorphology with geography has brought the latter not only undoubted advantages, but also certain disadvantages. If geomorphology is primarily concerned with landforms as objects to be studied in themselves, as the botanist is concerned with plants, then as Michotte notes, the point of view is that of a systematic science, in contradiction to that of geography as a chorographic science [189, 26]. A general result of this contradiction is to be found in the difficulty that many geographers who have been trained primarily as geomorphologists have experienced in maintaining consistently the geographic point of view, as Penck himself has observed [90, I, 38 f.], not to mention the confusion that many of them have introduced into methodological thought in geography. A more specific result has been suggested in an earlier connection (Sec. XI D). The study of landforms as objects in themselves, leads logically to a classification of them as individual objects of a systematic science rather than to a classification of the areal character of landforms--"the character of the various morphologic area of the world," as Michotte puts it. While geographers have felt free to classify climates, natural vegetation, or farm types independent of the classifications of the corresponding systematic sciences, as long as geomorphology was regarded as an integral part of geography, they were inhibited from developing a different classification of landforms suitable for chorographic description. In many cases, to be sure, the types of individual landforms were suitable, but the attempt to make them usable in all cases led, for example, to that paradox of areal terminology, the description of the White Mountains as "a collection of monadnocks."

Whatever conclusion may be drawn with respect to the relation of geomorphology to geography, it is necessary to note that if, as Kraft holds for German science, the facts of historical development make it a part of the field of geography, that conclusion does not provide an argument for the inclusion of logically analogous problems in other parts of geography in which they have not established themselves in the past. Thus, the claim made by Maull, and more recently, by East, that the processes of evolution of state-areas are as properly a problem in geography as the study of the evolution of landforms, would be valid only if it could be shown that such studies in the geographic aspects of political history have in fact been developed primarily by geographers, rather than by political scientists or historians [157; 199, 270; cf. 216, 956 f.].

We may summarize briefly our examination of the relation of systematic geography to the systematic sciences. Ideally, systematic geography receives from other sciences or from general statistical sources, the necessary data concerning the distribution of any phenomenon; it classifies the various forms of that phenomenon in any way that is suitable for geographic purposes--i.e., in terms of characteristics significant to regional character--whether or not such classification is available from other sciences. Further, ideally, it receives from the systematic sciences the explanation of the distribution of the phenomenon, that is, its genesis. Whether it be landforms, forests, crops, steel mills, or political states, the principles of development and the causes of distribution, as such, are the concern of the appropriate systematic fields. Geography starts with those facts and principles--assuming always, of course, that the systematic sciences concerned have provided them--as frankly borrowed material.

We have dwelt on the preceding question at some length because it is particularly in systematic geography that the student is likely to lose his sense of the geographic point of view, so that he may, as Lehmann has suggested, give a false picture by disproportionate consideration of phenomena that are the objects of the systematic sciences [113, 237 f.], or he may leave geography and enter entirely into other fields. It is doubtful if this can be prevented merely by drawing boundaries, however sharp and clear they may be. The reader may already have objected that we have drawn no clear boundary between systematic geography and any systematic science. No such attempt has been made, and if we remember that the relationship involved is not the borderland of neighboring fields but rather the intersection of fields lying in different planes, no such boundaries are needed. The distinction is in the point of view: that of the systematic science is focussed on the particular phenomena, which are studied in terms of distribution; that of systematic geography on the part which that distribution plays in forming areal differentiation. In many studies, the geographer may find it necessary to make excursions in the plane of the related systematic science, away from the common line of intersection. If he has the geographic point of view clearly in mind, he will need no boundary stones to remind him that he is making an excursion out of his field, but will return, as soon as he has established the necessary data or conclusions, to the geographic plane (See Fig. 1.).

To maintain the geographic point of view in systematic geography, it is necessary for the student to refer his work constantly to the field of geography as a chorographic science. Most writers agree that this view is most clearly indicated in regional geography. Consequently, many have agreed with Penck that every geographer, no matter how great his interest in specialized systematic branches of the field, should make regional studies [129, 639]. In any case, it is essential, as Lehmann insists, in making any study in systematic geography, constantly to consider the relation of that study to regional studies [181, 49]. If a systematic study is considered from this point of view it is immediately clear that the interest of the geographer is not in the phenomena themselves, their origins and processes, but in the relations which they have to other geographic features (i.e., features significant in areal differentiation).

On first thought it might be supposed that the conclusions to which we have arrived would result in the elimination of most, if not all, of the work in systematic geography. On the contrary, they free that part of the field for its essential function of providing systematic study of the relation of the differentiation of specific kinds of phenomena to total areal differentiation. The areas of the world differ from each other in terms of a mutually interrelated complex of heterogeneous features, each of which is different in the different areas. The complete interpretation of an individual area requires that, at some level of size, we break it down, mentally, into the component parts formed by the specific categories of phenomena. As Michotte puts it, we must study the vegetative character of the area, its geomorphological character, the character given it by each of the major cultural features, and so on [189, 17-33]. Further, the comparison of such completed individual studies would not five us a complete understanding of the areal differentiation of the world. It is necessary to know also how these areas, considered solely in terms of their natural vegetation, landforms, or each of various cultural features, differ from, and are related to, each other. Michoote speaks of these comparative studies as "comparative plant geography," "comparative morphological geography," etc. Similarly Hettner's title for his several-volume study of systematic geography, "Vergleichende Länderkunde," though unfortunately misleading, as we have seen (Sec. II D), is not so inappropriate as might be thought. Likewise, we may add, it is significant that the physiographer, Fenneman, who first presented the chorographic concept to American geography should demonstrate that viewpoint, in systematic geography, in his masterly studies of the regional physiography of the United States.

It is particularly, however, the students whose interest in regional geography has motivated them to make systematic studies who have most clearly indicated the type of work that systematic geography should undertake. Recognizing that the relation of any specific feature to the character of an area is to be measured particularly in terms of its relation to the other factors in that total character, they have perceived that absolute measurements of individual elements are less valuable than relative measurements, or ratios, of elements in reference to each other.

One of the most important advances in making studies in systematic geography more geographic in character has been the development of the isoplethic method of mapping ratios. Based on the work of Engelbrecht, this method was developed and effectively presented in this country be Wellington D. Jones [283] and is now in widespread use. Compare, for example, the utility, for a study of agricultural differences in different parts of China, of the maps of crop ratios that Trewartha has recently published [392] with the dot maps showing absolute values upon which we previously had to depend. That this method may be carried further, so as to show simultaneously two significant ratios concerning the same phenomenon, is indicated by the writer's isoplethic map of the dairy areas of the United States [325; the demonstration is very inadequate due to the small scale on which the map is reproduced]. This study (an extension and amplification of Jones's map of a smaller area) actually portrays the differences in areal character of land use resulting from varying degrees of intensity of dairy development, which can be inferred only indirectly, and in many areas incorrectly, from the ordinary census maps showing distribution of dairy cows or milk production. In both of the systems of world division of rural areas discussed in the previous section, the determination of agricultural types and the delimitation of "agricultural regions" depended on the construction of a large number of isopleth maps (not published) showing areal differences in ratios of individual crops to total crops, of cropland to total land, of livestock units to cropland, etc.

The ratio method in systematic geography is not limited to studies in agricultural geography. In addition to ordinary "relief" maps, which actually show directly only elevation, maps of "relative relief" have been constructed after Partsch, by various European geographers, and, in this country, by Guy-Harold Smith [see particularly James' survey in 294, and Cressey's recent example, 338]. In "sociological geography," Kniffen has used the method in mapping house-types and the writer has used it to show the areal differences in racial construction of the population of the United States [359].

Even where the character of the distribution does not permit of isoplethic mapping, as in the treatment of characteristics of cities, the principle of measurement by ratios rather than by absolute figures may be used to bring out the differential character of cities--i.e., that character, other than size, which is most significant in the comparison of cities, and in the comparison of regions in terms of their urban development. This is illustrated by the writer's study of the manufacturing belt of North America, in which the manufacturing functions of cities are measured in relation to their total functions, rather than in absolute values [326].

A much more complicated technical tool for work in systematic geography has but recently been presented by John K. Wright, under the title of "Some Measures of Distribution" [293]. The fact that it involves rather complicated mathematical formulae should not prejudice geographers either for or against its use. The various small examples which Wright offers suggest that we may have here a new technique of great value in enabling systematic geography to arrive at conclusions useful in regional geography that will be far more accurate than those now available. While this possibility is suggested by the examples which he gives, for the writer, at least, they seem inadequate to establish the utility of his technique. It is to be hoped, therefore, that some student will be interested in applying the technique to some actual problem to see what results it may produce.

In their simplest form systematic studies in geography are confined to single elements. We have previously noted, however, the importance of the concept of "element-complex" in geography--i.e., an interrelated association of various elements, regardless of kind. If approximately the same element-complex is found repeatedly in different areas and its distribution is geographically significant, it may also be studied systematically--over the whole world or any large area. Such studies, interconnecting different branches of systematic geography, may be considered as stepping stones from the study of single elements to the study of the total complex of a particular area in regional geography.

A single element-complex may represent an interrelation of elements at a single point--e.g., rainfall, temperature, slope, soil, drainage, and vegetation--in which case we may speak of a vertical complex of indefinite horizontal extent. On the other hand, the elements may be situated at different points so that their interrelation constitutes an areal form of more or less definite horizontal extent. Thus a longitudinal U-shaped Alpine valley in its primeval condition was a natural-complex in which the factors of slope, soil, drainage, and vegetation varied in a definite manner from the mountain shoulder on one side to that on the other. It is an areal form fairly definitely determined in the transverse direction though its limits in the longitudinal direction are indefinite so far as the concept itself is concerned--i.e. , are determined only in each specific case. A polje, in contrast, is a similar complex areal unit definitely limited in all directions.

The Estonian geographer, Markus, has contributed an interesting and suggestive study of element-complexes (Naturkomplexe) confined largely, if not entirely, to combinations of natural elements [239]. He notes that geographers have long recognized certain more obvious cases of element-complexes by such terms as "tundra," "high moor" and "low moor," "grass-rich depressions in steppes," etc. Noting that changes in any factor in a complex do not cause immediate change in the others, but rather that these adjust themselves to the new conditions at different rates of speed, he distinguished between "normal complexes," in which all the elements correspond to each other completely, and "abnormal" in which the adjustment has not yet reached completion. These terms give a clearer description than the distinction between "harmonious" and "inharmonious" that other writers have suggested in a similar connection (see Sec. IX D). Markus speaks of a "positive shifting" of an element-complex where it is pushing into an area of another complex that requires a lesser amount of any particular factor--as in the advance of forest into steppe--and negative shifting in the reverse direction. Further, he projects a complete classification of element-complexes in which real complexes are reduced to abstract species or types--by consideration of their essential common characteristics--and in which these are arranged in families, orders, etc.

This ultimate object--establishment of a Linnean classification of types, even of abstract types, of element-complexes--faces essentially the same insurmountable difficulties as we have met in the attempt to arrange regions in a single system of classification. Forested mountains, forested plains, mountain steppes, and plain steppes are four distinct types of element-complexes that cannot logically be arranged in any unilateral system of classification, since we have no method of deciding objectively whether the difference between mountain and plain is more important than the difference between grass and forest. Likewise, we cannot accept Markus' further implication that a geographic region (Landschaft) can be expressed as a single type of element-complex; if we consider all the elements involved in the complex of which the region consists, we arrive at the unique case, not the type (see Sec. X E). Nevertheless, though we cannot accept the more optimistic conclusions that Markus draws, we can expect valuable results from the systematic study of particular types of natural element-complexes, each of which is to be regarded as expressing more of the character of any area than a single element, though not its full character, not even in outline.

Our previous discussion of regional division indicated that we may expect much more useful results from the study of the many element-complexes, extending in many cases over wide areas, that have been produced by the organizing hand of man. These complexes are of a different order from the natural element-complexes that we have discussed, in that they are not merely the sum total result of the interaction of forces accidentally placed together--any one of which may be understood by itself in its relation to the others. These cultural element-complexes have been purposely created by man for the sake of the ultimate result, and the presence of any one element is to be understood not in terms of its relations to the others but in its relation to the ultimate result. For example, the importance of oats in the Corn Belt is to be explained in terms of its significance to the total crop and livestock association that man has organized on Corn Belt farms, or rather, if one will, in terms of his ultimate purpose of securing the highest monetary return with the least expenditure of labor, capital, and land. Consequently, as noted earlier, these cultural element-complexes are, to a considerable degree, organized as unit Wholes, an understanding of which should be a first step in the development of cultural regional geography.

The relatively small units of cultural element-complexes--e.g., farms--involve a much larger number of factors than those commonly found in natural element-complexes and include both material and immaterial elements. O. E. Baker has recognized that contrasts in types of farms in the United States involve contrasts in the character of the farm population: farmers represent farm elements just as definitely as do the livestock and crops [312].

Though the cultural complexes commonly form but small areal units they may be found to be organized together, by man, into looser areal complexes that, individually, cover relatively large areas.

Finally we may recognize complexes involving both cultural and natural elements. Because man, in many cases, has developed the same cultural element-complex in areas of similar natural conditions, we may expect to find a number of complexes consisting of cultural element-complexes in interrelation with certain natural elements. Since man, however, has been far from consistent in his form of adjustment to natural conditions, we must expect these compound element-complexes to have relatively restricted applicability.

Geography finds that certain element-complexes with which it is concerned have been studied by other sciences. If the complex includes only elements of the same general category--e.g., the natural vegetation as a complex of different plants, or an iron and steel works as a complex of blastfurnaces, steel furnaces, rolling mills, fabricating mills, storage yards, etc.--one of the systematic sciences will presumably be concerned with the study of the complex in itself and in its distribution over the world. The more complex forms however--involving combinations of heterogeneous phenomena--may be of concern to the geographer alone. In either case, the classification of the types of element-complexes for use in geography must be adapted to geographic purposes. If economists have produced a classification of farms suitable for geographic purposes, the geographer will utilize that classification; but if not, he is free to develop his own. In this particular case it appears likely that both groups of students working in cooperation may develop a classification suitable from the point of view of both sciences [cf. 320 with 319 and 324]. But if economists' classification of manufacturing industries offers little of value to geography, geographers must develop their own.

Any study in systematic geography, whether of an element or an element-complex, concentrates on one particular kind of phenomena or phenomenon-associations. It naturally leads, therefore, to the establishment of generic concepts; that is, for each element, or element-complex, a logical system of types may be established. On this basis the relations of the feature studied to other geographic features, for which types have also been established, may be stated in the form of principles, however limited or inaccurate in application. Whether one considers rainfall, soils, stream deposits or erosion, crop-animal associations, steel mills, or political boundaries --in each case over wide areas if not the whole world--it should be possible to establish principles of relationships between the feature studied and other features geographically significant.

It is by no means possible, however, to express all the findings of systematic geography in generic terms, whether of concepts or principles [Hettner, 167, 283]. In the systematic study of volcanoes, Krakatoa cannot be adequately treated solely as an example of a type. Likewise, in the relations of one geographic feature to another, innumerable cases will be found, each of which is unique. Nevertheless a very large part of the work in systematic geography does deal with universals and leads to the development of principles. Do these provide geography with that precious power that is often regarded as a hall-mark of "science," the power to predict?

The essential characteristics of that form of "knowing" which we call science--to use the term suggested by the physicist, John T. Tate¹⁰³--are not determined by the character either of the knowledge or of the capacities acquired; these are rather resultant products of that manner of pursuing knowledge which is science. The ability to predict in any branch of science represents the attainment of such a degree of certainty of knowing that, by deduction from principles, the future outcome of a combination of present factors is known almost as certainly as it can later be known as an observed fact. The qualification "almost" represents more than a margin of error, of inaccuracy in measurements: in every field there is an ever-present margin of uncertainty, of not-knowing, which cannot be eliminated even in the physical sciences--nor do modern physicists expect that it may ever be eliminated.

The ability of any science to predict is therefore the result and outward evidence of a high degree of attainment of the ideals of accuracy, certainty, universality, and system. It is not the test of a science, but only the test of success in "knowing" in any science. That success is not to be attained by striving directly for its result, the power to predict, but rather by striving for the highest degree of attainment of the fundamental ideals of "knowing."

In plain words, we will not learn to predict in geography by attempting to predict. It is a corollary of scientific principles that we should seek to know to what extent our knowledge is incomplete or uncertain. To attempt predictions in situations for which we know we lack the necessary knowledge is to be unscientific. Science does not require that we be able to predict. The sound demonstration of a low capacity for prediction in any particular field of knowledge is not evidence that that field is not a branch of science, but, on the contrary, is a scientific conclusion testifying to the scientific character of that field.

No professional geographer, I presume, would claim that research in systematic geography had as yet reached such a high degree of attainment of the requisite ideals as to enable it to make predictions of high degree of certainty. Though a more maturely developed geography should show far higher attainments, we must recognize certain insurmountable difficulties and limitations that will always be present in geography.

We know, in the first place, that the nature of most of the phenomena that must be measured in systematic cultural geography, and of many of those in systematic natural geography, will never permit of such accurate and certain measurements as are possible in some of the natural sciences. This difficulty geography shares with many of the systematic sciences, notably, of course, the social sciences (See Bowman [106, 17, 31-33]).

We know that our knowledge of phenomena and their interrelations, in every branch of systematic geography, can only incompletely be contained in generic concepts and principles, and that there is inevitably, therefore, a margin of uncertainty in prediction. While this margin is present in every field of science, to greater or less extent, the degree to which phenomena are unique is not only greater in geography than in many other sciences, but the unique is of the very first practical importance. This is true not only of geography and the social sciences but likewise of human physiology and psychology--from the point of view of the individual and his family, at least--and of certain aspects of meteorology and geology. To predict that the islands of Japan will experience innumerable earthquakes is of little value; who will predict the date and location of the next major earthquake?

We know further that the complex interrelations of phenomena that we study in systematic geography cannot be taken into the laboratory where some of the variables may be controlled in experiments, so that we could learn the exact significance of each factor. This handicap, again, geography shares not only with the social sciences but also with human physiology, with most of the branches of geology, and with astronomy. The students in all these fields can work only in the laboratory of reality, can observe only those experiments that reality chooses to perform for them. Actually the same limitation applies to the physicist studying the actions of electrons in the laboratory; he cannot control the individual electrons which he is attempting to study. There is "no hard and fast line between observation and experiment," Cohen concludes [115, 111].

In the performances that reality presents us as substitutes for laboratory experiments, we know that geography is handicapped in two ways. Whereas some fields are presented with thousands or millions of repetitions of nearly similar cases, in geography, as well as in geology and astronomy, and in parts of all the social sciences, there may be only hundreds of similar cases, or only a handful, or often but a single case. Where the number of factors involved in the relationships is relatively small, as in astronomy, and the clearness and exactness of observation may be at least as fine as in the laboratory, a few cases may suffice to provide an adequate basis upon which to develop scientific laws of high degree of certainty; but in geography, geology, and the social sciences, one or both of those conditions are lacking.

One further difficulty remains in any branch of science that must deal with extremely complex functions of a large number of more or less independent variables. If some divine power should present the scientist with complete statements of every one of the interrelations involved, expressed (if that were possible) in mathematical equations of the greatest complexity, and if then, in any particular situation, divine power should also provide complete and accurate knowledge of the individual factors, the complexity of the problem which must then be solved to arrive at certain knowledge of the outcome would be beyond the ability of finite minds (A conclusion suggested by the philosopher, Charles Hartshorne).

In conclusion, therefore, geography is, by its nature, one of the branches of science from which we are to expect relatively little knowledge of the future of such a degree of certainty as to justify the word "prediction." (This is an over-statement. See Supplementary Note 55) Undoubtedly one could postulate many cases where such certainty was possible--even in our present state of development. If large deposits of highgrade iron ore should be discovered in West Virginia we could not only predict the mining development that would result, but we could no doubt predict, in a general way, notable changes in the iron and steel industry in the Pittsburgh and Calumet Districts. The reader will observe that this is not only an extreme case but one in which the relationships are unusually simple, since three variables are of major importance [352]. For the most part, the knowledge of the future that systematic geography can provide is limited to that lesser degree of certainty that we express by such terms as "trends" or "likelihoods," and must further be qualified by many uncertain factors due to the more or less arbitrary action of individual men or groups of men.

In sum, we may justifiably predict that a mature geography developed to the maximum, cannot attain more than a very restricted capacity for prediction [cf. Schmidt, 7, 210-13; Colby, 107, 35 f.; and Finch, 223, 19].

While we may dismiss the question of ability to predict as not fundamentally relevant, the pursuit of universals, generic concepts, and principles, must be regarded as of major importance for the development of any science. In geography, the greatest opportunity to develop generic concepts is in systematic geography. A large part of the work in each section of systematic geography is concerned with phenomena and relations between phenomena that repeat themselves in similar specimens in different parts of the world, so that it is possible to express them in universals and thereby to develop principles. Consequently, for those among the ranks of geographers who by reason of temperament, ability, or training, prefer to study the generic, with the opportunity to develop scientific principles or laws, there is plenty of work to be done in geography. Since such work is not merely an integral part of geography, but forms the necessary base for the studies of regional geography, no geographer need berate these students as deserters from the field. Those who use Fenneman's picture of the field of geography and speak of regional geography as the center, should not overlook his qualification: "There is no intention of assigning more dignity to one part of the field than to another, nor of asking any man to turn aside from that which interests him to something else. There is no more inherent worth in a center than in a border" [206, 10].

Our examination of the character of systematic geography emphasizes the inescapable comprehensiveness of geography, a condition that would not be reduced in the slightest if one were to omit regional geography; on the contrary, the one method by which the diversity of interests is brought into unified study would be lost. Even if one should attempt to reduce systematic geography to the study of natural, non-human, elements, it would still be concerned with a heterogeneity of phenomena as great as that of all the systematic natural sciences put together, and the elimination of human factors would make it impossible to unify this diversity in the study of actual regions. Any attempt to arrive at a unified field by further whittling can change the situation only relatively: if one throws our plant and animal geography, one still has subjects as different as the study of climates and landforms. As these are both physical sciences, and are both concerned with the earth, they can logically be combined, either from the point of view of physics, or the earth. From the point of view of physics, they are widely separated fields that are not brought into logical combination by the incidental fact that they both concern the earth. If the earth forms the unifying framework they are combined only in the earth surface, as broadly conceived, where they are inextricably intermixed with the elements studied in plant, animal, and human geography. Only in the study of all earth surface features in their actual interrelated combinations in regions, can the heterogeneity of systematic geography be unified into one science. "We need not be frightened away by the fullness and breadth of the problems," Richthofen concluded. "The field is great. But the work can be divided among many. No one today can do research in all parts of geography. But he who devotes himself seriously to geography, can master it sufficiently to follow advances in all branches; and he who, through modest limitation, is fortunate enough to investigate productively in one part, should always strive to comprehend the relation of that part to the rest and never to lose sight of the interconnection of the whole" [73, 67-70].

At the same time, Richthofen felt that the individual geographer who wished to contribute research to the advance of geography, "the higher he sets his goal, the more should he concentrate his preparation on one part of (systematic) geography and the particular systematic science that forms its foundation, without neglecting instruction in the other parts." It was natural for Richthofen to emphasize geology as "the surest foundation," since that had been his own, but both Oberhummer and Gradmann are drawing the logical conclusion from his general principle when they state that individual geographers may just as properly select some other science as their principal supplementary field--whether meteorology, botany, economics, or some other [124,11; 251, review, 552].

An individual geographer who specializes in a branch of systematic geography, and is adequately equipped in the corresponding systematic science, will no doubt have occasion to make studies in that other field as well as in systematic geography. Just as it has always been regarded as appropriate for individual geographers who were adequately equipped therefore to do research in geology, it may similarly be appropriate for individual geographers--under the same condition--to do research in anthropology. economics, or political science. Inasmuch, however, as geographers are not capable of judging the research in other fields, it seems logical that such research should be presented, not to geographers, but to the workers in those other fields.

Need it be added that such transfers of point of view may equally well be made in the opposite direction? The student of a systematic science, interested in the geographic aspects of his field, will frequently be able to contribute to the field of geography, and one trusts that in this exchange there need be no grumblings of trespassing on either side.¹⁰⁴

If geography, in America as well as in Europe, may be said to have returned, in a certain sense, to the point of view that was common with Humboldt and Ritter (see Sec. II D), its long period of concentration on systematic studies has enabled it to return far better equipped with generic concepts and principles with which to interpret the findings of regional geography--though unfortunately this equipment is relatively deficient in respect to human or cultural features, both in geographic literature and in the training of most of its students.

Many geographers who have accepted this shift in emphasis evidently have done so under the provisional assumption that regional geography is to be made as "scientific" as systematic geography has been, that somehow it must be raised to the plane on which scientific principles may be constructed. We have noted a number of difficulties into which this ambition has led. In our final consideration of regional geography it is necessary to understand clearly certain limitations imposed upon the student that are not found in systematic geography.

After a number of unsuccessful attempts to express the special nature of regional study in words, I find it can be most clearly presented if we may use mathematical symbols, though we shall not, of course, find it possible to express such complicated problems in any real mathematical formulae or equations.

Any particular geographic feature, z, varying throughout a region, might theoretically be represented as a function, f (x, y), x and y representing coordinates of location. As a function of two variables, any such feature that we are able to measure in a thematically--such as slope, rainfall, or crop yield--can be represented concretely by an irregular surface. Such a surface would then present the actual character of that feature for the whole region; it would, theoretically, be correct for every point, and for every small district. Furthermore, if the function involved were not too complicated, the theory of integral calculus would permit us to integrate the total of that feature for any limited section, as well as for any individual point. In a sense, part of our work in systematic geography corresponds to this form of presentation.

Likewise, the relation of any two or three geographic factors to each other within a region--e.g., the relation of crop yield to rainfall and humus content of soil--might be represented as a functional equation involving that many variables: z₃= f ' (z₁, z₂). The concrete representation of this relation would require again a surface form. More commonly, in systematic geography, we consider only the relation of one factor to but one other, which we may then represent as a curve on a plane surface. Each of these factors, z, is of course a different function, f (x, y), and the more complex equation, z₃= f ' (z₁, z₂) holds true only if z₃, is unaffected by other z factors, or if those which affect it are constant throughout the region under consideration. Neither of these conditions is strictly true: almost any geographic element we may consider is affected by more than two of the natural elements, and may also be affected by incommensurable, or quite unknown, human factors; and all of the factors considered vary to some extent no matter how small the area considered. Consequently, we have introduced a degree of distortion of reality even at this step in systematic geography.

We may introduce a further step by establishing element-complexes, u, each representing functions of many z elements, varying, by more or less regular rules, with the variations in a smaller number of those elements. Thus, given certain conditions of soil, slope, temperature, and rainfall we may presume within a wide margin of both inaccuracy and uncertainty, certain conditions of natural vegetation and wild animal life, and we may express the total of all these z elements by one u element-complex. If it were conceivable that we could express this feature, u, arithmetically, its character over an area would likewise form an irregular surface that would indicate its character for any limited part. From the nature of these element-complexes, however, it is obvious that any such representation would have a high degree of unreliability.

In regional geography, however, we are concerned with a vastly more complicated function of the location co-ordinates. It cannot be expressed as the function of any one element or element-complex, but rather of various semi-independent element-complexes, u, and of additional semi-independent elements, z'. Thus, the total geography, w, at any point, might be expressed by the function, F (u₁, u₂...u_n, z'₁, z'₂...z'_n). If we could have accurate and complete information concerning the form of the function, F, and every one of the element-complexes, u--each as a function of various z elements--and of the semi-independent elements z', the function would be so complicated that we could not hope to represent it by any concrete form, even in terms of n-dimensional space. We would have a function that could be solved only for each point, x, y, in the region, but could not be correctly expressed for any small part larger than a point. In other words, we could study the geography of the area only from the study of the geography of the infinite number of points within it. This task, being infinite, is impossible. The problem of regional geography, as distinct from a geography of points, is how to study and present the geography of finite areas, within each of which the total complex function involved depends on so many complex functions, complexly interrelated, as to permit of no solution by any theory of integration.

Consequently we are forced to consider, not the infinite number of points at each of which w is in some degree different, but a finite number of small, but finite, areal divisions of the region, within each of which we must assume that all the factors are constant. In order, then, to cover an entire region we will need but a finite number of resultants, w, each representing the geography of a small unit of area rather than of a point. This method is legitimate only if one remembers that it inevitably distorts reality. The distortion can be diminished by taking ever smaller unit areas, but it cannot be eliminated entirely; no matter how small the unit, we know that the factors which we assume to be constant within it are in fact variable. In practice, the smallest units that we can commonly take time to consider are sufficiently large to permit of a marked degree of variation, and therefore of a significant distortion of reality in our results.

To express our conclusion in more common terms, in any finite area, however small, the geographer is faced with an interrelated complex of factors, including many semi-independent factors, all of which vary from point to point in the area with variations only partially dependent on each other. He cannot integrate these together except by arbitrarily ignoring variations within small units of area, i.e., by assuming uniform conditions throughout each small, but finite unit. He may then hope to comprehend, by analysis and synthesis, the interrelated phenomena within each particular unit area.

Although the studies of all the unit areas added together will constitute an examination of the entire region, this does not complete the regional study. As Penck has emphasized, it is not sufficient to study individual "chores" (approximately homogeneous districts) and to establish types of chores. "Above all geography must consider the manner in which these are fitted together to form larger units, just as the chemist does not limit himself merely to studying the atoms, but investigates also the manner of their situation beside each other in individual combinations. The comprehension of geographic forms (Gestalten) has scarcely been taken into consideration by the new geography." Just as a mosaic cannot be comprehended, Penck continues, by classifying and studying the individual stones of which it is made, but requires also that we see the arrangement and grouping of the individual pieces so the study of the arrangement of the "chores"¹⁰⁶ will present different structural forms of significance [163, 43 f. ; in part also in his address given in Philadelphia and published in English, [159, 640].

Our second step--in a theoretical approach to regional geography--is to relate the unit areas to each other to discover the structural and functional formation of the larger region. Since all the factors concerned and therefore the resultants, have been made arbitrarily constant for each small unit, it may be permissible to speak of functional relations between one factor in one unit and another in another unit, as though these were functional relations between the units themselves--provided that we understand that this is not strictly true. Further, the regional structure produced by this method will have the character of a mosaic of individual pieces, each of which is homogeneous throughout, many of them so nearly alike that in any actual method of presentation, they will appear as repetitions in different parts of the region. But we are not to be deceived into regarding this mosaic which we have made as a correct reproduction of reality. It is simply the device by which finite minds can comprehend the infinitely variable function of many semi-independent variable factors. The fiction involve is threefold: we have arbitrarily assumed each small unit area to be uniform throughout; we have delimited it from its neighbors arbitrarily, as a distinct unit (individual); and we have arbitrarily called very similar units identical in character.

There are certain other fundamental limitations that must be insisted upon if we are to compare the face of the earth even in the more or less distorted form in which the geographer must present it, to a mosaic. We may say that there is a similarity in the detail of the technique but, unless we are to return to some teleological principle, we cannot liken the face of the earth to any work of art, for we cannot assume that it is the organized product of a single mind. On the contrary, if we may transfer Hetner's analogy of a building built by several architects working independently to Huntington's picture of the "The Terrestrial Canvas," we may say that the face of the earth has been produced by the interrelated combination of different color designs each applied by different artists working more or less independently, and each changing his plan as he proceeded. In systematic geography one might say, we attempt to separate each of the individual designs in order to understand its form and its relation to the others and, thereby, to the total picture. Since the total pictures were not produced simply by superimposing different color plates in printing, but are to some extent, causally related to each other, this separation involves the analysis of the causal and functional relations of each design to the others. In regional geography we first reduce the subtle gradations which the different artists of nature have applied and intermixed on the face of the earth, to the stiff and arbitrary form of the mosaic technique. When we then survey the formation of the mosaic pieces, we are not to expect some unified organized pattern such as every work of art must have. On the other hand, neither need we expect more chaos, or a kaleidoscope; for we know, from our studies in systematic geography, that there were principles involved in the individual designs, and if our determination of the unit areas of homogeneity has not been purely arbitrary, but has been based on the combination of careful measurement and good judgement, we may expect the combinations of these designs to show more or less orderly, through complex patterns. Further, whatever the explanation of these patterns may be, their form is significant to each of the parts, since the development in each unit part is affected by that in the others.

The last thought leads us finally to another major respect in which any analogy of the earth surface to a work of art is inadequate, namely, the fact that, while the latter is static, consisting of motionless forms, the face of the earth includes moving objects that are constantly connecting its various parts. (To attempt to introduce the artist's special use of such terms as "lines of force," "movement," "opposite forces," etc., would merely add to confusion here.) In other words, the geographer must consider functional as well as form. In establishing our arbitrary small unit areas we not only assume that each form is uniform throughout in character, but also in function. Likewise in combining these units into larger regional divisions our problem is complicated by the fact that we must consider the functional relations of the units to one another as well as their form. For example, if two neighboring areal units are so similar that we have painted them as much alike as two pieces of mosaic of the same color, but one of them is functionally related to a city center in one region, the other to a city center in another, are we to include them in different regions, or, if in the same region, in which? Any answer to this question can only be more or less intelligent: there can be no one "correct answer."

Just as it is necessary to know the arrangement of unit areas in a region, it is likewise necessary to understand the arrangement of the region to each other. Both Penck and Granö (who follows a similar line of thought [252, 28-31]) would carry the process on to larger units; the size of the areas concerned is immaterial. Regional geography, therefore, studies the manner in which districts are grouped and connected in larger areas, the manner in which these larger areas are related in areas of greater scale, and so on, until one reaches the final unit area, the world.

There is, however, one important difference at the level as of integration. Both Penck and Granö appear to ignore the fact that the small, but fundamental, element of fiction in the assumption of the homogeneity of the smallest units of areas increases progressively as one advances to larger divisions. Consequently, the determination of these larger divisions requires increasingly arbitrary distortions of fact.

Assuming the first step, the establishment of "homogeneous units" of area, we may proceed to the second by enclosing in a continuous area which we call a region, the greatest possible number of "homogeneous units" that we judge to be nearly similar, together with the smallest number of dissimilar units. Our judgement of similarity will involve subjective judgement as to which characteristics of the homogeneous units are of greater importance than others, so that, at best, the determination of the region is in a sense arbitrary.

Furthermore, we seldom find in reality such a simple solution as that described. Though some geographic features vary but gradually from place to place, the irregular and steep variations of others--such as soils, slopes in mountainous areas, urban settlement, and all other features of essentially linear form, rivers, roads, and railroads--will force us to include in any region, "units" of quite different character. It is necessary therefore to determine which kinds of units are, either in actual interrelation or merely in juxtaposition, characteristic of the region as approximately considered, and then so determine it as to include the greatest number of those several kinds of similar units, with the smallest number of units of other kinds.

In considering any large areas in which we have first recognized "homogeneous units" and are attempting to form them into regions, which we can briefly characterize in terms of similarities or relations among some of those units, we may find the task relatively simple in parts of the area, where perhaps the great majority of the units are notably similar. But it may be extremely difficult in parts between these, which may be characterized by units that are, in some respects, similar to units on one side of them, in other respects, to units on another side. Further, we will find areas containing such a variety of different kinds of units that we cannot see where to include them. In some cases, to be sure, we may recognize such areas as transition zones, but that merely postpones the fundamental problem without solving it. Likewise, to call them "characterless areas," or areas of "general" or "mixed" types is simply to dodge the problem entirely (see Sec. IX E).

The individual student, no doubt, would gladly wipe such troublesome areas off the map, but he is not granted that privilege. Neither is a science which seeks to know what the world is like permitted to ignore more difficult areas and confine itself to those easier to organize into its body of knowledge. Since these doubtful areas are commonly not merely narrow borders of transition, but areas of wide extent, perhaps as great or greater than those more clearly classified, there is no basis for assuming that they are of less importance in the total picture of the larger areas, or of the world, than the areas whose character we can more readily describe. Fenneman's statement with reference to the different parts of geography applies even more literally to parts of an area--"there is no more inherent in a center than in a border."

Consequently, when we divide any given area into parts which we call regions, so determined that those characteristics that we have judged to be most important may be most economically stated for each region, we cannot avoid many decisions based on judgement rather than on measurement. We must, therefore, acknowledge that our regions are merely "fragments of land" whose determination involves a considerable degree of arbitrary judgement. On the other hand, if all possible objective measures have been used, and the arbitrary decisions are based on the student's best judgement, we may properly regard his regions as having more validity than is expressed by the bare phrase "arbitrarily selected." On the other hand, the view of various writers previously noted, that geographers could be expected to come to approximate agreement on the specific limits of regions--or even on their central cores--appears, in view of all difficulties listed, overly optimistic.

It hardly needs to be added that the conclusion that geography cannot establish any precise objective basis for regional division does not permit it to shirk the task of organizing regional knowledge into areal divisions determined by the best judgement possible. In order to utilize the generic concepts and principles developed in systematic geography to interpret the findings of regional geography, the latter must be organized into parts that are as significant as is possible. In the present state of development of the field--if not indefinitely--we do not have what would be the simplest solution, namely, a single standardized and universally accepted division and subdivisions of the world into regions. Therefore, each student of regional geography has imposed upon him the task of standardizing his own system of regional division--unless he can utilize that of some colleague. "Standardized" is used here to indicate that the regional system is based on certain standards specifically stated, so that other students may know precisely what the organization is.

The complete organization of regional knowledge in geography requires--whether as a final or as a primary step--the division of the whole world. In whichever direction the process is carried on--and we noted that it requires consideration in both directions (Sec. X A)--the completed system must provide a regional division of the world in which our knowledge of each small part may be logically placed. For this extremely difficult problem we found two different methods of solution. Geographical knowledge may be logically arranged in systems of areas classified according to certain characteristics of the areas. Though this method had distinct utility for comparative purposes, it does not permit organizing all regional knowledge into one system, but requires several independent systems. Furthermore, it does not present the actual relations of areas as parts of larger areas. These relations can be included only in a realistic division of the world into a system of specific regions, in which all regional knowledge may be incorporated in a single logical system. Such a system unfortunately is not provided the geographer by any natural division present in reality, nor by anything corresponding to the simple division of organic forms. It must be developed and constantly modified by geographers as a result of research, at the same time that is being used , always in tentative form, as the organizing structure of regional research.

We have suggested, in very general terms, the manner in which the problem of delimiting regions may be met, in order that geographic knowledge may be organized intelligently in regional units. What kind of knowledge is to be included within the regional study itself? So far as the nature of the material is concerned, we have previously indicated that a complete geography of a region includes all the phenomena that are included in systematic geography--insofar as they may be present in the particular region. The only field of geography that is not included in regional geography, a well as in systematic geography, is historical geography. As there was a different geography in every past period, there may be any number of independent historical geographies, each including its own systematic and regional divisions.

The kinds of phenomena present in regions, the particular manner in which they are present, and the nature of their interrelations, both within each unit area and across unit divisions, determine the particular forms and the functions of the area. Though most students agree in theory that these are of coordinate importance, much of the work recent decades tend to emphasize the study of forms to the neglect of functions. We found this to be particularly pronounced in the work of the "landscape purists" (Sec. VII E). On the other hand, Granö finds that many students, like Spethmann in particular, conceive of an area as "the file of forces, as a dynamic complex." Geography, Granö insists, is not the study of forces, of interrelations, but study of things of interrelation in areas. Judging by the major example which he has presented in German, Granö himself tends to emphasize physiognomy and gives little attention to the functions of areas [252, 114 f.; cf. Waibal, 266, 204].

When we speak of the functions of areas, we are not to forget that in reality the area is not a thing that functions, it is only certain things within it that have functional relations to the things in other areas. If our fiction of the small homogeneous areal unit, uniform in both form and function, permits us to speak figuratively of the unit area as having a functional relation to other unit areas, we are not to ignore the fictive character of this concept by attempting to consider areas as having, in themselves, functional relationships.

In particular, it is necessary to note that the concept of the small areal unit breaks down when we attempt to study "the genesis of an area." When we study the previous historical stages in the geography of an area, we find that any one of our small unit areas of homogeneity may not have had in the past even that incomplete degree of validity that we may grant it today. That is to say , since areas, no matter how small, do not grow as units, but change only as a result of the differential change of different things within them, the unit areas of today was probably not a unit area in an earlier stage, and will probably not be in a future stage. The very concept of mosaic is incompatible with the concept of gradual and differential change. Consequently, the study of genesis in geography can only be undertaken in the form of systematic studies : the study of "the genesis of an area" can only be broken down into studies of the genesis of each of the various objects contained within it. These are therefore studies in systematic geography ; to what extent they may be desirable for an understanding of the geography of any region is a controversial question which we touched on earlier, and need not consider here (Sec. VI B).

We should now be in a position to answer the question that is of greatest importance in contemplating the possible development of regional geography: may we hope to progress in this branch of our field to the construction of universals, of generic concepts, and scientific laws or principles?

One form of generalization used in regional geography we have already described: the construction of regions from small unit areas. The philosopher, Kries, has distinguished such generalizations of heterogeneous and semi-independent parts as a third type of scientific description, together with type concepts and the description of the unique [according to Graf, 156, 57-62, 105]. The importance of the distinction lies in the fact that this form of generalization offers no basis for establishing general principles; for that we must have type concepts.

It is obvious that any universal principles that we might attempt to construct on the basis of the fictive areal units set up for the purposes of description, could have no more validity than the units themselves. Unless these are taken as extremely small units, the margin of error introduced by our personal judgment would lead, in any principles we might set up, to a degree of error so great as to render them of very doubtful value.

Regardless of that essential difficulty, however, we found that even these arbitrary units, each involving a complex combination of associated forms, cannot be classified into a system of types based on the sum totals of its varied and semi-independent factors. Though in any one region we find unit areas so similar that we may, with but a minor degree of error, call them alike, we do not find unit areas of that kind of similarity in other regions of the world. A small district somewhere in the Upper Rhine Plain may be very much like many other such districts in the same region, but no matter how small a district we take, it is fundamentally different from any unit area in any other world region (see Sec. XI D).

We arrive, therefore, at a conclusion similar to that which Kroeber has stated for history: "the uniqueness of all historical phenomena (meaning, I take it, the particular combination of phenomena at a particular time) is both taken for granted and vindicated. No laws or near-laws are discovered" [116, 542]. The same conclusion applies to the particular combination of phenomena at a particular place.

One is not to suppose, however, that regional geography is studied without the use of generic concepts and principles. On the contrary, the interpretation of the interrelations of phenomena within each region depends upon the type concepts and principles developed in systematic geography [cf. Schmidt, 7, 194]. In other words, for the individual items included in regional geography, and the simpler relations between them, we depend constantly on universal concepts supplied from the systematic studies, but the total interrelated combination of each areal unit represents an essentially unique case for which we can have no universals.

An objection may be made that one form of study used by many geographers in the consideration of regions represents an approach to the construction of scientific laws--namely what has been called "comparative regional geography," the comparison of regions of notable similarity. As a current example we may cite Maull's effective comparison of the Amazon, Congo, and Insulindia areas [179, 184-6]. The fact that in other sciences "comparative studies" have marked an adolescent period preceding the flowering of a nomothetic science, has led many to suppose that regional geography might be expected to grow out of its youth by progressing from comparisons to scientific principles.

The essential idea involved is nothing new in geography. Introduced by Humboldt--if not by earlier writers--it was used, according to Hettner, by Brehm, Nehring, and particularly by Richthofen [161, 403 f.]. (See also Jessen's paper on the use of comparisons [440]) Plewe found, however, that these represented merely occasional examples, that our literature contained no comparative regional geography as a branch of the field [8, 46-55, 77]. Such occasional comparisons, he noted, are used in all sciences, citing as an example, Th. Litt's comparative study of Kant and Herder (Berlin, 1931). Historians, we may add, frequently find it valuable to compare the developments of any two or more periods that are significantly similar in certain respects. These examples should make us sceptical of the likelihood of our discovering anything that could be called laws, or near-laws, of regional geography.

Passarge recognized the limitations that prevent a comparative Länderkunde from developing universal concepts, but still (in 1936) believes that these can be avoided or overcome in a comparative Landschaftskunde [272, 61]. In order to discover the laws of regions it is necessary, he says, to have a tertium comparationis and this, he believes, is provided by his system of abstract types. As we saw in our previous discussion (Sec. X E) he has, in part, merely reduced the difficulties, by reducing the size of areas concerned, and for the rest he has simply dodged the limitations by setting up types that are not even in outline complete abstractions of real areas. The difference between the real Land and the real Landschaft (as area) is only a difference in size; a tertium comparationis is equally impossible in both cases. We may go on comparing areas of whatever size forever with no hope of discovering regional laws.

Plewe concludes, therefore, that the comparative study of regions is neither a preparatory step to a nomothetic regional geography nor an independent branch of geography. Ritter's introduction of the concept, over a century ago, represented a transfer from a quite different kind of science; he never clearly defined his concept, and others who have taken it up have used it in many different ways but without leading to any important development [8 , 82 f.].

Nevertheless the use of this method, as a supplementary device, appears to offer certain distinct advantages. If widely separated regions are in many respects similar, so that, in respect to certain elements or element-complexes, they may be classified as of the same type, the comparison of their similarities, and particularly of their differences, may well serve as a check on the interpretations we place upon the relation of phenomena within each one of them.

Even more useful is the employment of this method in the comparison of localities within a major region, where there may be a much larger number of element-complexes of the same type. By selecting those localities that are alike with respect to the greatest number of features, and comparing them with those that are like them in many, but not all, of these features, we may have a key to the significance of specific features for the area as a whole.

To take a well-known example: the consideration of the major characteristics of the Cotton Belt as a whole might lead one to suppose that--taking certain cultural conditions for granted--the importance of cotton in the area was to be explained simply in terms of climatic conditions. We have learned, however, by contrasting the localities in which cotton is the all-important crop, with those where cotton is of minor importance though the climatic conditions are the same, that the cotton crop of the South as a whole is not to be understood without considering the character of the soil.

Likewise, American geographers, at least, have long realized what is not so clearly recognized in popular thought--or even by many European geographers--that the climatic conditions of the South do not directly explain that feature which is of greatest importance in the contrast between North and South--namely the high proportion of Negro population. By the same method of comparison of localities one finds that this element--and all the cultural elements associated with it--cannot be understood without considering the combination of climatic and soil conditions that are necessary for cotton. This conclusion, however, is incomplete: in the cotton district of greatest importance today, in central Texas, the proportion of Negro population is low. The complete explanation can be reached only when one also compares the localities which were developed for plantation crops--including tobacco as well as cotton--before the end of the slavery period, with those localities developed for the same crops since that time [359].

This method of comparing localities within the same larger region might appear to lead to generic principles. But it can lead to conclusions that are applicable only to the single larger region concerned. If we should add to the districts of the Cotton Belt a district in the Yangtse Valley and a district in the Bombay province, we could not include them all under any generic concepts of districts. In the comparison limited to districts in the Cotton Belt, we are not, as we noted in an earlier connection, comparing separate units, but only similar parts of a single larger region, parts whose similarity is simply a result of the fact that they are parts of the same region. Valuable as the device may be for checking our interpretations, it leads to no universal concepts or principles.

Regional geography, we conclude, is literally what its title expresses: the description of the earth by portions of its surface. Like history, in the more common sense of periodic history, it is essentially a descriptive science concerned with the description and interpretation of unique cases, from which no scientific laws can be evolved. Though this is undoubtedly a disadvantage, making the interpretation of findings far more difficult than in those fields that are able to develop general laws to explain individual cases, it does not mean that regional geography lacks any scientific goal. As previously noted, the construction of scientific laws is not the purpose of science, but a means toward its purpose, the understanding of reality. To any "who find the title 'earth description' (Erdbeschreibung, or geographia) insufficiently learned and scientific," Heiderich has answered, "description is the last and highest goal of scientific work, to be sure, not a mere outward external description that remains on the surface of the object, but a description that aims...to comprehend synthetically all that has been learned analytically from the characteristics of the object" [153, 213]. All that science requires is that, in order that the interpretive description may have a maximum degree of accuracy and certainty, universals shall be constructed and used wherever possible. Regional geography utilizes all the appropriate generic concepts and principles developed both in the systematic sciences that study particular kinds of phenomena, and in systematic geography which studies their relations to each other over the earth.

The conclusion to which we have arrived concerning the nature of regional geography may enable us to answer one or two questions that have been raised by a number of students in very recent years. The course of thought among American geographers concerning regional studies has been discussed in two articles published in Germany during the past year, by Broek and by Pfeifer [108; 109].From these surveys, and the critical articles to which they refer, the reader might suppose that after a period of enthusiastic concentration on regional studies that was introduced by the methodological papers of Barrows and, more particularly, of Sauer, American geographers had now begun to doubt whether much was to be expected from regional geography after all. It may be that the testimony has been exaggerated in the echoes back and forth across the Atlantic; possibly we are presented with a revolt within a single university department that has reverberated among its present and former members here and in Germany.¹⁰⁷ Undoubtedly, however, other American geographers in oral discussions have expressed a note of scepticism concerning the results to be expected from regional studies.

In a number of cases, the sceptics have spoken, or written, as though after a long and earnest attempt to advance geography by regional studies, we had discovered that the works produced did not add up to, or yield, significant general results. It is difficult to believe that this argument is meant seriously. American geography has concentrated its efforts on regional studies for scarcely twenty years and never completely. During that time, perhaps a score or so of research students have each made one, two, or three regional studies in areas scattered from the Peace River Country to São Paulo, from Europe to China. Since neither of the two principal American promoters of the regional concept in theory has presented a concrete example of a full study in (present) regional geography, each of the individual research students has had to work out more or less independently his own methods of determining his region, of selecting the phenomena for consideration in it, and of presenting his results. Would anyone seriously consider that we have had a fair test of the possibilities of developing general results from regional studies? Even if all the work had been carried out under standardized procedures, such a small number of cases scattered over more than half the world could hardly be expected "to add up" to any general results, or to provide the basis for generalizations.

It seems more likely that many students have begun to suspect, for other reasons, that no matter how many regions are studied, no matter by what methods, no scientific laws will be forthcoming. This conclusion we have found can be thoroughly demonstrated in theory, so that we can agree that any who have made regional studies with that ultimate purpose in mind have been following a will-o'-the-wisp; the sooner it is abandoned the better for all concerned.

If, however, the purpose of geography is to gain a knowledge of the world in terms of the differential development of its different areas, the task of studying regions as areal divisions of the world, is not subject to question in geography. Neither need the workers in any science feel discouraged if the efforts of a relatively small number of workers over a period of less than twenty years have produced less than enthusiasts led them to expect. Though the object of geography, the world, is large, it is limited in size, and we must assume that geography has a long life ahead of it. No doubt the group efforts of American geographers would show more productive results if they could concentrate the attention of all or most of their members on some limited part of the world--as French geographers have done on their own country. But the many factors that persuade students to travel far afield are not to be restrained, even if it were desirable. One can only hope for a larger total number of workers, and possibly, for increasing concentration within this country of the work of particular groups--as at Wisconsin--on the regions of a relatively limited area. In particular, as Finch notes, we should not expect results of far-reaching scientific value from the practice of "skimming the cream of the more clearly given from a region and its abandonment for another area" [223, 26]. The value that studies of this kind may have for teaching purposes may justify the time and effort spent on them--provided that the areas concerned are significant for class instruction. Lasting progress in research in regional geography will require a much greater amount of concentration of the individual's time--whether or not one would go as far as Finch and consider one region as sufficient for one student's life work.

On the other hand, these considerations do inevitably raise the question of what size of area should be considered worthy of research in regional geography. The regional studies launched under Vidal's leadership in France formerly examined areas the size of a province, but increasingly smaller areas have been selected. Demangeon feels that the extreme limit of "microscopic" study has been reached by Allix, whose examination of "L'Oisans," a part of an Alpine valley in the Dauphine smaller than an arrondissment, requires 915 pages, with a bibliography of 861 works [329]. This averages, Demangeon reckons, a little over a page per square kilometer or for 12 inhabitants. American geographers, by comparison, hardly seem justified in applying to their work the word "microscopic."

The question raised admits of no simple answer. Historians welcome extremely detailed studies of very short periods, in addition to less intensive studies of an extensive series of periods. The criterion, in either field, is the same--namely the significance of the study--but that is a criterion for which we have no objective measure. We have previously ("previously"; on pages 288 f.) suggested two major considerations--namely, the significance of the area in itself,¹⁰⁸ and its possible significance as representative of a large area, or a large number of similar small areas. Outside of the proper interest of the citizens of L'Oisans itself in the geography of their own district, we may assume that the world of knowledge in general has little need for such an exhaustive study of this small and unimportant district. On the other hand, if we had but little knowledge of the valleys of the French Alps, and reconnaissance had shown that this particular district was in large degree representative of hundreds of others, such a study might provide us with an approximate view of the regional geography of the entire area--or of a large part of it. Presumably, however, such a study would be limited by the desire to express primarily those characteristics that were representative, and one might question whether that would require a thousand pages. Demangeon finds much of the study superfluous because it merely duplicates findings that Blanchard and others have presented in works on similar districts. Insofar as Allix's work has served to corroborate that of his predecessors, that fact might have been more briefly presented. On the other hand, another competent critic¹⁰⁹ finds that Allix has contributed a much more thorough treatment of the problems representative of the French Alps than have any of his predecessors.

It is particularly against such "microgeographic" studies--to use Platt's term--that criticism of regional geography in America has been directed. Recognizing that it would be impracticable, in any reasonable length of time, to cover the whole land area of the world by the total addition of such small studies--and that the total might be indigestible if it could be attained--critics have feared that we would have but a miscellaneous collection of scattered pieces selected at random [cf. Leighly, 220]. More particularly, however, the critics have asked what general principles can eve expect to derive from such minute and scattered studies. Even some who have made such rnicrogeographic studies, like James, have given expression to a later feeling that "the more detailed and specific is the study the more insignificant are the results" [286, 84].¹¹⁰

To these attacks, Platt, in particular has replied vigorously, both in two published papers [221; 224] and in unpublished statements read before this Association. Microscopic geography, be observed in one of the latter, developed "as a rational and timely drive against the limitations of armchair compilation from promiscuous data, of subjective impressions from casual travel, and of environmental theory not founded on data." To attain these purposes, geographers took to the field and "in the field all geographers are microscopic." There "they face the geographer's dilemma in trying to comprehend large regions while seeing at once only a small area." They do not, he insists, plunge into detailed studies of minute areas because the methodological conclusions of others have led them to believe that thereby something will ultimately be gained for geography. On the contrary, their own efforts to comprehend areas of larger extent has led them to the reasoned conclusion that, in addition to general reconnaissance studies and detailed systematic studies covering large areas, accurate generalizations for larger regions require an examination of the total fundamental complex of interrelated features that can be examined, in detail, only in the small area."¹¹¹

Platt's defence of microgeography, however, is based less on theoretical discussion than on the actual work that he has been carrying on for some years in Hispanic America, which forms the most significant series of microgeographic regional studies in American geography [listed in 221, 13; to that list should be added 224]. Of the skeptical questions that have been raised concerning the value of such a series of studies, many appear irrelevant to its purpose. This, I take it, is simply to increase our organized, objective and reliable knowledge of the lands south of the Rio Grande. That such knowledge of the different parts of the world is desirable and requires the research of trained workers is, we repeat, the fundamental justification for the field of geography. That our present knowledge of the Hispanic American area is inadequate is obvious to anyone who has attempted to gather the materials necessary even for an elementary course concerned with that part of the world. Consequently, we are not to test the value of such a series of detailed studies of scattered districts by asking whether they can yield us any "scientific principles," or whether they will aid us in drawing conclusions concerning "the larger relationships" of which Pfeifer speaks [109]. So long as Play does not claim that all geography should consist of such "microgeographic" studies, or of regional studies in general, these questions are irrelevant. The relevant question is, granted that we want more adequate knowledge of the geography of South America, is his method of study appropriate to produce such knowledge?

Few will question the inadequacy of the general surveys of South Ametica now available. In his most recent study, on coastal plantations in British Guiana, Platt has noted that the best available, generalized maps of the continent give erroneous impressions of the soils, vegetation, and population density of the specific districts he studied. Even if we had accurate detailed information on the climates, land forms, soils, crops, races, and commerce of South America, these would not add up to the geography--the areal differentiation--of the different parts of that continent. In studies limited even to provincial scale, the American student is frequently baffled because he lacks the detailed knowledge of the cultural element-complexes that are basic to the cultural geography of the region. For areas in United States or Europe, he may have acquired that knowledge unconsciously, whether as a by-product of fieldwork, or merely from his general knowledge. These essential features must be studied first in relatively small areas--particularly in a world area where there is lack of cultural homogeneity. If, then, one has acquired an understanding of a particular ranch in Panama and may be permitted to assume somewhat similar features scattered through a large area, one has a more correct picture of the geography of the larger area concerned than can be acquired by any small-scale measures [see Platt, 221].

The essential assumption in this proposition, of course, is that the minute district studied is in fact representative of others; as Finch notes, it can hardly be typical in any full sense [223, 24] (See Supplementary Note 57). If it is representative, however, it presumably will be typical in certain limited respects, and it is important that we know in what respects it is approximately typical. In areas that are adequately covered by census and climatological data, geological, topographic, and soil surveys, it may be possible to give approximate answers to these questions from the study of such data. The utility of element-ratios and isopleth maps in this connection has been suggested previously. In other areas, one can only depend on the student's judgment formed from reconnaissance. Though such judgment can only give answers that are far removed from scientific certainty, they are better than no answers at all, and should therefore be provided--even at the risk of being shown erroneous by later work by the same or other students.

Perhaps only in his most recent study has Platt clearly demonstrated the relation of these detailed studies of small districts to the reconnaissance study of large areas. Though the microgeographic area which he studied in detail is not, in this case, "typical of broad regional types," it is shown to be "a normal feature of a coherent plantation district, which in turn has a consistent place in the intricate geographic pattern of South America" [224, 123 if.]. No doubt the significance, to broader regional knowledge, of his previous studies in small and widely separated districts, apparently chosen at random, will be made clear in the ultimate publication of his "reconnaissance study of Hispanic America," of which these detailed unit studies are to form integral parts. (Published in 1942, Latin America: Countrysides and United Regions.)

In sum, the student who presents a study of a small area of no special importance in itself, needs to keep in mind that the purpose is not to present the area in itself, but to provide an accurate illustration of the representative character of a larger region, too large to permit of such intensive study. So long as he keeps this broader purpose in mind, there are no grounds apparent on which we can prescribe the minimum size of area that may be studied.

Our historical survey showed that, while modern geography from its beginnings has included both of these points of view--in theory even in Varenius's outline--it has experienced notable shifts in emphasis from one to the other. Whereas the work of Humboldt combined both points of view, under the influence of Ritter, systematic studies were placed in a subordinate position and easily lost sight of. Though the protests of Bucher and Fröbel were of no avail at the time, a later generation following Peschel, and motivated by scientific standards developed in such fields as geology, swung the center of interest the other way. As late as l9l9 Hettner found that in Germany systematic geography was generally regarded as "something higher, more distinguished" than regional geography. He therefore repeated the arguments that he had presented at various times during nearly a quarter of a century to show that the two parts of the field were scientifically on the same level [142, 22 f.]. Less than a decade later, however, he found it necessary to presentopposite arguments to urge the same conclusion; for "youth, which is given to exaggeration, has turned Or too much away from systematic geography" [161, 401];(we can hardly suppose that Hettner was unaware of the fact that some of those concerned were not much younger than he). The reaction that had taken place earlier in France, under Vidal, swept German geography in the post-War years toward an increasing emphasis on regional geography, as the real goal of geographic work. Thus Obst, believing that one could develop a science of "Länderkundliche Typologie" wished to shift the time-honored term of "general geography" to the study of regional types as the goal of geography, and placed systematic geography (as allgemeine Erdkunde) in the subordinate position of a necessary propaedeutic [178, 6-9]; somewhat similar views have been expressed by Braun [155, 5],Volz [151, 247], Ule [170, 486], and Gradmann [quoted in 166, 13].

Likewise in this country, the emphasis that Barrows, and particularly Sauer, had placed upon the study of regions (in the latter case, "landscapes") led some to regard systematic studies as necessary only for instructional purposes but inappropriate for geographical research.

On the other hand, such veterans as Hettner and Penck have never wavered in their insistence that both points of view were of equal importance in geography [Hettner has said as much in almost every methodological treatment he has written; for Penck, see 129, 639; 137, 173-76; 163, 44]. The very fact that geography has experienced these successive shifts in emphasis from one side to the other is in itself indirect evidence that both are of coordinate importance in the field [cf. Hettner, 2, 306].

In his critical investigation of geography as a single, unified field of science, Kraft finds that, while one could dismiss the charge of dualism of content--natural and human features--as invalid, the inclusion of the systematic and the regional points of view was an unquestionable form of dualism. He agrees with Hettner, however, that this dualism cannot be expressed simply as the combination of a nomothetic and an idiographic science; systematic geography must include the study of unique cases, and regional geography must use generic concepts and principles. In any case, neither construction of laws nor the description of the unique represents the purpose of geography, or of any other science. The purpose of geography is the same in both branches, the comprehension of the areal differentiation of the earth, and this purpose cannot be solved either by systematic studies alone nor by regional studies alone, but requires both approaches. Consequently, he concludes, this dualism in approach is justified as necessary for the single aim which makes geography a unified science [166, 11-13].

This view, we may add, is further supported by the fact, stressed by Hettner, that it is frequently difficult to classify particular studies under one heading or the other. The difference is not in the substance, but in the point of view, and in certain kinds of studies these may be combined. For example, the systems of land-use classification previously discussed (Sec. X F, G) are intended to provide backgrounds for agricultural regional geography and they involve, in outline, a major part of the regional study of any area. At the same time, however, they represent systematic studies of particular element-complexes in their world distribution, so that it is by no means clear whether they belong more in the one or the other of our two major divisions.

Finally, if one agrees that both regional and systematic studies are included as essential parts of geography, we may perhaps dismiss any question of relative importance as irrelevant. For systematic geography, regional studies provide, not merely a source of detailed factual information that otherwise would hardly be available, but they also indicate problems of relationships that might easily be overlooked in systematic geography, and they provide the final testing ground for the generic concepts and principles of systematic geography. On the other hand, it is even more obvious that progress in interpretation of the interrelated phenomena of regional geography is constantly dependent on the development of such universals by systematic studies. Any assumption that these studies can be left to the systematic sciences concerned with each particular category of phenomena has been shown by experience to be unwarranted. The aspects of these phenomena with which geography is concerned--their relation to other earth phenomena in different parts of the world--are not of direct concern to those systematic sciences and are more commonly left unstudied, unless geographers study them, as Lehmann has shown. Systematic geography, he therefore concludes, is not to be thought of as a border area of geography, or merely as a propaedeutic, but represents "organs vital to the growth of geography, without which its regional crowning can as little exist as a real tree without its roots" [113,236 f.].

Further, Lehmann suggests, the point of view developed in systematic geography is different from the general point of view in regional geography but at the same time of such value to it that every regional geographer should work productively in some systematic branch or branches (he recommends two or more). On the other hand, Penck, whose most notable contribution has no doubt been the systematic study of landforms, urges that "the cultivation of regional studies is indispensable for the geographer; they form for him the touchstone of his whole concept of geography, of his geographic system" [129, 639; cf. also Graf, 156, 82].

The mutual dependence of the two interconnected points of view in geography has been consistently maintained by Hettner from his earliest methodological treatment of more than forty years ago to the present time. The development of sound universal concepts in systematic geography is the essential basis for progress in regional geography, but since systematic geography is in method similar to the systematic sciences, "the geographer who works only in it and does not cultivate regional geography runs the risk of leaving the ground of geography entirely. He who does not understand regional geography is no true geographer. While regional geography alone, without systematic geography, is incomplete, it remains geographic; systematic geography without regional geography cannot fulfill the full function of geography and easily falls out of geography" [142, 22f.].

We may assume, therefore, that there is plenty of work to be done in the field of geography by both methods of approach. It is not for any student specializing in either approach to speak with scorn or condescension of those who are working in the other. "Differences of approach," as Kroeber suggests, "are probably at bottom largely dependent on differences of interest in individuals" [116, 569]. Paraphrasing his statement further, we may conclude that it is perfectly legitimate to confine one's interest to the specific approach of systematic geography, or to the integrating approach of regional geography, or to use alternately one or the other according to occasion. But sympathetic tolerance is intrinsically desirable and certainly advantageous to understanding: to scientia.