138 ROMAN FORTIFICATION. 



encircle a camp is this on calculating the area within the ramparts 

 it will be found that a camp of this size would accommodate more 

 than four legions with their auxiliaries (1) and Vespasian only had 

 one. On the other hand the plan of a Roman town was rarely 

 rectangular. Yitruvius recommends that it shall not he square, 

 nor formed with projecting angles, but polygonal (circumitionibus) 

 that the enemy may be seen from more places j for a part in which 

 angles project is not easily defended because the angle protects the 

 enemy more than the citizen (" Architectura" Lib. I. cap. 5.) In 

 passing we may remark that the plan of the interesting Roman town 

 of Silchester, which is now being excavated, was polygonal as 

 recommended by Yitruvius. Yegetius bears similar testimony. 



Let us see what was the nature of the fortifications which 

 surrounded Dorchester. We must, however, first understand how 

 cities were built at the time when Dorchester became a Roman 

 possession that we may have some standard with which we can 

 compare the works we shall find here. 



The first consideration when a new town was to be laid out was 

 the situation, and next the convenience and healthiness of the spot 

 proposed ; in these respects the Romans were most careful in the 

 selection of their sites both for camps and town. Yitruvius devotes 

 a chapter (Lib. II. cap. 4) to these points alone, whilst Yegetius lays 

 additional stress on the importance of choosing a site naturally strong 

 (Lib. IY. cap. 1). In these respects Dorchester conforms to the old 

 Roman traditions ; no stronger position than the town occupies 

 could have been found in the neighbourhood ; its healthiness is 

 proverbial, and it is conveniently situated for water just above the 

 Frome. 



When the site was determined, the next point to be settled was 

 the size of the proposed town and the plan of the walls. The 

 buildings and streets were an after consideration. It is a misfortune 

 that no exact plan of the Roman walls exists. It is said to have 



(1) According to the system of castramentation which was practised 

 about this time it has been computed that one thousand men would occupy 

 13,027 superficial feet (English). 



