37-t ROMAX ARCHITECTURE. Part I. about 100 ft. above the level of their foundation in the centre. That of Segovia is raised on light piers, the effect of which is perliaps somewhat spoiled by numerous offsets, and the upper tier is if anything too light for the lower. These defects are avoided at Tarragona, the central arches of which are shown in Woodcut No. 250. In this example the proportion of the upper to the lower arcade is more perfect, and the wliole bears a character of lightness combined with constructive ' solidity and elegance unrivalled, so far as I know, in any other work of its class. It wants, however, the grandeur of the Pont du Gard ; for though its length is about the same, exceeding 800 ft., it has neither its height nor the impression of power given by the great arches of that building, especially when contrasted with those that are smaller. ■■ ^ - tj. ' ^ ' raj 249, Aqueduct of Segovia. Scale 50 ft. to 1 in. Aqueduct of Tarragona. Scale 50 ft. to 1 in. The Roman bridges were designed on the same grand scale as their aqueducts, though from their nature they of coui'se could not possess the same grace and lightness. This was, however, more than com- pensated by their inherent solidity and by the manifestation of strength im])arted by the Romans to all these structures. They seem to have been designed to last for ever ; and but for the violence of man, it would be hardly possible to set limits to their durability. Many still remain in almost every corner of the Roman Empire ; and wherever found are easily recognized by the unmistakable impress of Roman grandeur which is stamped upon them. One of the most remarkable of these is that which Trajan erected at Alcantara, in Spain, represented in the annexed woodcut. The roadway is perfectly level, as is generally the case in Roman bridges, though the mode by which this is obtained, of springing the arches from different levels, is perhaps not the most pleasing. To us at least it is unfamiliar, and has never, I think, been adopted in modern times. In such a case we should either have made the arches all equal — a
