river, stands the prosperous market town, the road and railway junction upon which all the various glen-villages converge.  A day’s march further down, and at the convergence of several such valleys, stands the larger county-town—­in the region before me as I write, one of added importance, since not only well nigh central to Scotland, but as the tidal limit of a till lately navigable river.  Finally, at the mouth of its estuary, rises the smoke of a great manufacturing city, a central world-market in its way.  Such a river system is, as geographer after geographer has pointed out, the essential unit for the student of cities and civilisations.  Hence this simple geographical method of treatment must here be pled for as fundamental to any really orderly and comparative treatment of our subject.  By descending from source to sea we follow the development of civilisation from its simple origins to its complex resultants; nor can any element of this be omitted.  Were we to begin with the peasant hamlet as our initial unit, and forget the hinterlands of pasture, forest, and chase (an error to which the writer on cities is naturally prone), the anthropologist would soon remind us that in forgetting the hunter, we had omitted the essential germ of active militarism, and hence very largely of aristocratic rule.  Similarly, [Page:  106] in ignoring the pastoral life, we should be losing sight of a main fount of spiritual power, and this not only as regards the historic religions, but all later culture elements also, from the poetic to the educational.  In short, then, it takes the whole region to make the city.  As the river carries down contributions from its whole course, so each complex community, as we descend, is modified by its predecessors.  The converse is no doubt true also, but commonly in less degree.

In this way with the geographer we may rapidly review and extend our knowledge of the grouping of cities.  Such a survey of a series of our own river-basins, say from Dee to Thames, and of a few leading Continental ones, say the Rhine and Meuse, the Seine and Loire, the Rhone, the Po, the Danube—­and, if possible, in America also, at least the Hudson and Mississippi—­will be found the soundest of introductions to the study of cities.  The comparison of corresponding types at once yields the conviction of broad general unity of development, structure, and function.  Thus, with Metschnikoff we recognise the succession of potamic, thalassic, and oceanic civilisations; with Reclus we see the regular distribution of minor and major towns to have been largely influenced not only by geographical position but by convenient journey distances.  Again, we note how the exigencies of defence and of government, the developments of religion, despite all historic diversities, have been fundamentally the same.  It is not, of course, to be forgotten how government, commerce, communications, have concentrated, altered or at least disguised the fundamental geographical