had well-built sewers beneath them.

Trajan's Timgad was plainly small. On any estimate of the number of
houses, the original draft of veterans sent there in A.D. 100 can
hardly have exceeded 400, and the first population, apart from slaves,
must have been under 2,000. This agrees with the figures of Aosta (p.
89). There, 100 acres took 3,000 veterans and their families; here the
area is about one-third of 100 acres and the ground available for
dwellings may perhaps have been one-sixth. In neither case was space
wasted. There was not probably at Aosta, there certainly was not at
Timgad, any provision of open squares, of handsome facades, of temples
seen down the vista of stately avenues; there were not even private
gardens. The one large unroofed space in Timgad was the half-acre shut
within the Forum cloister. This economy of room is no doubt due to the
fact that the 'colonia' was not only a home for time-expired soldiers,
but, as Prof. Cagnat has justly observed, a quasi-fortress watching
the slopes of Mount Aures south of it, just as Aosta watched its
Alpine valley. As Machiavelli thought it worth while to observe, the
shorter the line of a town's defence, the fewer the men who can hold
it. The town-planning of Timgad was designed on other than purely
architectural or municipal principles. For this reason, too, we should
probably seek in vain any marked distinction between richer and poorer
quarters and larger or smaller houses.[95] The centurions and other
officers may have formed the first municipal aristocracy of Timgad, as
retired officers did in many Roman towns, but there can have been no
definite element of poor among the common soldiers.

[95] Ballu detects a 'quartier industriel' in the outer town, but
the evidence does not seem to warrant so grand a term.

Such was Trajan's Timgad, as revealed by excavations now about
two-thirds complete. The town soon burst its narrow bounds. A Capitol,
Baths, a large Meat-market, and much else sprang up outside the walls.
Soon the walls themselves, like those of many mediaeval towns--for
example, the north and west town-walls of Oxford--were built over and
hidden by later structures. The town grew from one of 360 to a breadth
of over 800 yds. And as it expanded, it broke loose from the
chess-board pattern. The builders of later Timgad did not resemble
those of later Turin. Even the _decumanus_, the main 'east and west'
street, wandered away north-west in an unc