THE DEFINITION OF THE CITY 537
with the amount of its territorial element. Thus administrative statistics are most frequently based upon this assumption. The French censuses since 1846 and L'institut international de statis- tique since 1887, have applied the term city to every aggregation of more than 2,000 inhabitants. Such a definition cannot serve as the basis of a scientific study, and it has long been denounced as arbitrary. The space occupied by the establishment is too external a characteristic and varies too much according to the circumstances. Several historians have defined the city of the Middle Ages by another morphological characteristic, the pres- ence of a fortification.® But apart from the fact that this char- acteristic, which does not obtain in the case of modern cities, lacks universality, it still could not be used to define and to specify the mediaeval city, for many villages and even farms were also fortified.*^
• The German writers of the eleventh century distinguished two kinds of places: unfortified places (villages) and fortified places (cities). Thus they con- trast the urbs, castellum or civitas with the villa or the vicus. See Keutgen, Untersuchungen iiber deutschen Stadtverfassung, p. 46. Mauer (Geschichte der St'ddteverfassung, I, 31 ff.) says the same: "Cities are villages surrounded by walls." See also Babeau, La ville sous I'ancien regime, p. 239, who de- tects in the rampart the essential quality of the city.
^ In all primitive societies villages are fortified. Africa : Masqueray, For- mation des cites chez les populations sedentaires de fAlgerie, p. 86 ; Cyr. van Overberg, Les Mayombe, p. 160 (Belgian Congo); Asia: Cabaton, Les peuplades demicivilisees de I'Indochine {Conferences Scale coloniale, 1907-1908, p. 94) ; The Book of Ser Marco Polo, ed. Yule, II, 131 (China) ; Hunter, Imperial Gazetteer of India, I, 458 ; XIII, 88 ff. ; Codrington, The Melanesians, p. 302 (the special necessity for protection against the head-hunters) ; America : Tylor, Primitive Culture [French translation, I, 54] (Sioux, Iroquois) ; Dor- sey, Omaha Sociology, Third Rept. Bur. Amer. Ethn., pp. 313, 314 (description and plan of Fort Ponka) ; Diehl, L'Afrique byzantine, pp. 224, 292 ; Jullian, Histoire de la Gaule, II, 214, 215 (existence of open and fortified villages for refuge in war time) ; Flach, Origine historique de I'habitation, pp. 45 S., and Origines de I'ancienne France, II, 301 ff . ; Thierry, Monuments pour servir d, fhistoire du Tiers-£tat, IV, 785 (the villages of Ponthieu were almost all fortified) ; Stouff, "La description de plusieurs fortresses et seigneuries de Charles le Temeraire," Revue bourg. Ens. Sup., XII, 14 (a village still fortified in 1473). Here is apparent a further reason why aggregations of 22 households, of 50 households, etc., are designated in the texts by the title castrum (Molinier, "La senechaussee de Rouergue," Bib. £cole Chartes, 1883, pp. 468, 470 ff., numerous examples). See also K. Hegel, Entstehung des
