as possible on one map; but not everything, for that would be simply chaos. This calls for considerable skill in generalization, and often some artistic talent besides, as for example in making topography more legible to untrained persons by using hachuring instead of contours.

Climate and population are so complex that it would probably be impossible to put all the features of either one on one map and still have it intelligible. In mapping crops the principal difficulty is one of generalization. On a map of sufficiently large scale one could show every field with its particular crop, but on the scales ordinarily used that would be out of the question, and one. would have to resort to some such expedient as determining the leading crop in each area, and mapping a corn region, a wheat region, a cotton region, etc. In mapping soils and vegetation one would have the same difficulty, plus the much greater one of deciding just where to draw the line between different types, which might intergrade in all sorts of ways. Detailed soil maps of several hundred counties have been published by the U. S. Bureau of Soils, and one needs only to compare the maps of neighboring or similar counties made several years apart, or by different men, to see how much individual judgment may vary in classifying the soils. The few soil maps of the whole United States that have been published (see for example those in Bulletins 85 and 96 of the Bureau of Soils) are primarily soil province maps, based on geology, topography, etc., and tell very little about the soils themselves. Pretty good vegetation maps have been made of some small areas, but all the existing ones of whole states and larger areas are far from perfect. (The latest of North America is that of Harshberger in his Phytogeographic Survey of North America, 19111, and the latest of the United States is that of Shreve in the Geographical Review for February, 1917.2)

The writer has gradually evolved (but cannot say perfected) a method of mapping soils, vegetation, crops, and some other complex features, which though it involves no radical innovations is probably an improvement on any now in use. It does not fulfill the ideal of putting everything on one map, but it shows at a glance the prevailing type of soil, forest, etc., in the areas mapped, and has the special advantage of leaving very little to individual judgment after the facts are gathered in

'Reviewed in Bull. Am. Geog. Soc., 45:38-42, January, 1913.

*Reviewed in Torreya, 17:103-106, June, 1917.

sufficient detail and the regional boundaries located. It might be called the quantitative regional method.

The first aim of the geographer in studying any considerable part of the earth's surface, such as a state or country, should be to subdivide it into areas as homogeneous as possible, to give a maximum of contrast between neighboring divisions and require the least possible repetition in describing them. Just how finely one should subdivide is largely a matter of judgment and expediency and the scale used. Small areas are generally more homogeneous than large ones, but if we make them too small they will also be too numerous and too much alike. The boundary between any two neighboring regions should correspond with a comparatively abrupt change in one or more geographical features, whether topography, soil, vegetation, or something else. Of course in many cases we can find no sharp boundary between regions that are obviously different on the whole (especially if the principal difference is a climatic one), and must draw the lines more or less arbitrarily. Again there may be differences of opinion as to what sort of differences should be taken into consideration, what are major and minor regions, etc. But these difficulties are nothing new to scientists, being inherent in taxonomy. They have been met thousands of times in classifying geological formations, minerals, soils, plants, animals, etc., and some have been settled while others are still in dispute.

3

There have been many attempts to map the geographical divisions of North America or the United States, some of which suffer from the narrow point of view of the author, who may be primarily a physiographer, climatologist, botanist, or demographer, rather than an all-round geographer. But plants are not classified by their leaves, flowers, or fruits alone, or animals by their skeletons or teeth alone, but all possible characters are taken into consideration, as should be done likewise with geographical divisions. A geographical map based primarily on topography (like Fenneman's) does very well in parts of the country where there is considerable relief, but in the coastal plain we often find regions much alike in topography differing considerably in soil and therefore in vegetation and agricultural features, if not also in density and composition of population.

But taking for granted that our geographical boundaries

Several of the attempts to subdivide North America have been summarized by Joerg in the fourth volume of the Annals of the Association of American Geographers (1915). The latest on the geography of the United States is by Fenneman in the sixth volume of the same (1917).

have been agreed upon, we now come to the main issue, of mapping the complex features of each region. The method is simply to take one group of features at a time, that can be measured in the same units, compute statistics for one region at a time, and put the different features in each class on the map in order of abundance or importance, indicating them by words of symbols, and using percentages if possible and desirable. For example, soils are divided into types and measured by acres or square miles, vegetation is divided into species and measured in feet of timber or tons of hay, and crops may be measured by either acreage or value.

For many regions in the United States we now have enough Government soil surveys to be fairly representative, and in all but a few of the earliest ones (published about the beginning of this century) the area of each type of soil is given, so that we can add up all the stony clays, fine, sandy loams, etc., and determine the percentage of each. The errors of judgment mentioned a few pages back are just as likely to be in one direction as in another, so that they tend to neutralize each other when several reports on different parts of the same region are available.

For vegetation we have very few reliable quantitative data as yet, except for a few small areas; but the writer by traveling through every known natural division of the southeastern states, by rail or on foot, or both, and jotting down the name of every plant recognized as often as possible, and consolidating the returns afterward, has determined the approximate relative abundance of all the commoner and more conspicuous plants, particularly trees, in that part of the country, and has made a beginning in a few of the northern states.

For crop statistics we have to depend chiefly on Government census reports, which are not altogether satisfactory. Considerable information is given about the yields of different crops in every county, but they are not reduced to a common unit; cotton, for example, being returned in acres and bales, and apples in trees and bushels. Again, pumpkins, watermelons, cabbages, tomatoes, onions, lettuce, and many other things are lumped together under the head of "vegetables." For the whole states the crops are subdivided more minutely in the general volume on agriculture, of the last census, but not in the state supplements; and state statistics are of little use to geographers, for every state in the Union has at least two natural divisions, and

some as many as twenty. There is, however, a way of circumventing some of these difficulties. In each state supplement the total value of each kind of crop (except where different things are lumped together as above indicated) is given, and by dividing this by the number of acres, bales, pounds, or bushels we can get the average value per unit for that state, which we may assume without serious error to be the same for every subdivision of the state. This gives us a means of reducing the county crop statistics to a common unit. Although value varies considerably at different times and places, it is a better criterion for comparison than acreage, for an acre of strawberries or lettuce, for instance, is worth far more than an acre of corn or wheat, because it requires so much more labor for planting, harvesting, etc.

Some of the states gather agricultural statistics of their own which are more complete than the Government statistics, but this does not happen to be the case with Georgia, the one selected for illustration. The crop statistics published biennially by the Florida Agricultural Department give the value of every important crop (separating all the different vegetables) in every county, and these were used by the writer a few years ago in determining the principal crops in all parts of northern Florida.

In putting these features on maps one map should be used for each kind, and the names of as many things as there is room for printed right in the proper spaces, in order to abundance, or area, or value, as the case may be. The differences between any two neighboring regions in that particular are then very evident. Abbreviations or symbols can be used to save space when necessary, and regions too small to put anything in can either be left blank (being of little importance with reference to the whole area mapped), or else their features can be printed out in the margin somewhere and connected with the proper region by a guiding line. As a region which extends through several states but is commonly regarded as a unit (like the Piedmont region and the Appalachian valley) may vary somewhat in different parts, on account of gradations of elevation, climate, or something else, it is well to work out the statistics for each state separately and map them accordingly, to show just what the variations are. Of course the method here described is just as applicable to political divisions as to natural geographi

'See 6th Ann. Rep. Fla. Geol. Surv., pp. 188-189, etc. 1914.

cal divisions, but to use political divisions alone would be unscientific, and would sacrifice the fundamental principle of maximum contrast to expediency. However, where we are not sure of the natural boundaries we may take certain crops of counties for trial, and after computing statistics for them find that shifting a certain county from one region to another gives more contrast, and thus gradually make our regions more nearly natural.

If we had county statistics of manufacturers they could be mapped in the same way as crops, but unfortuntately the Government publishes manufacturing statistics only for states and