It is hard to detect any principle underlying the use of these various forms. No doubt differences of historical origin are ultimately the causes of the mixture. But our present knowledge does not reveal these origins. The evidence is, indeed, contradictory at every point. If the Graeco-Macedonian fashion be quoted as precedent for square or squarish 'insulae', the Terremare show the same. If the theoretical scheme of the earlier Roman camp seemed based on the long narrow oblong, the actual remains of legionary encampments of the second century B.C. at Numantia include many squares. If one part of Pompeii exhibits oblongs, another part is made up of squares. If Piacenza, first founded in north Italy about 183 B.C., and founded again a hundred and fifty years later, is laid out in squares, its coeval neighbour Modena prefers the oblong. If the old Greek city of Naples embodies an extreme type of oblong, so does the later Augustan Carthage (pp. 100, 113). In the historic period, it would seem, no sharp line was drawn, or felt to exist, between the various types of 'insulae'. In the main, the square or squarish-oblong was preferred. Local accidents, such as the convenience of the site at Carthage, led to occasional adoption of the narrower oblong. The Roman land-surveyors, it is true, distinguished the square and the oblong in a very definite way. The square, they alleged, was proper to the Italian land or to such provincial soil as enjoyed the privilege of being taxed--or freed from taxation--on the Italian scale. The oblong they connected with the ordinary tax-paying soil of the provinces. This distinction, however, was not carried out even in the agrarian surveys with which these writers were especially concerned,[63] and it applies still less to the towns. No doubt it is a fiction of the office. It would be only human nature if the surveyors, finding both forms in use, should invent a theory to account for them. [63] Schulten, _Bonner Jahrbücher_, ciii. 23, and references given there. The system sketched in the preceding paragraphs seems, as has been said (p. 73), to have sprung from a fusion of Greek or Graeco-Macedonian with Italian customs. Roman town-planning, like Roman art, was recast under Hellenistic influence and thus gained mathematical precision and symmetry. When this happened is doubtful. Foreign scholars often ascribe it to Augustus and find a special connexion between the first emperor and the chess-board town-plan. But the architect Vitruvius, who dedicated his book to Augustus and who gives some brief notice to town-planning, urges strongly that towns should not be laid out on the chess-board pattern, but rather on an eight-sided or (as we might call it) star-shaped plan.[64] He would hardly have denounced a scheme which had been specially taken up by his patron, nor indeed does his criticism of the chess-board system sound as if he were denouncing a novelty in Italian building. [64] i. 5 (21), 6 (28, 29).