148 SECTIONAL ADDRESSES. 
was provided in order to allow of cleaning before the beginning of the 
long tunnel, in which it would naturally have been exceptionally difficult. 
The tunnel must be about 2} kilometres long, and the fall is 5°90 m. to 
the tank where the branch rejoins the main aqueduct, or 1 in 381, or 
2°62 per 1,000. 
We must now return to the main line, which has a fine bridge, the 
so-called Ponte degli Arci, over a tributary of the Anio. The original 
bridge was a massive structure in opus quadratum, most of which has 
disappeared, though the impressions of the blocks are visible on the pier 
of the great brick arch on the south-west bank, and some of the masonry 
itself in the base of the last pier on the north-east bank. The brickwork 
with which the concrete of the greater part of the bridge is faced is, once 
more, Severan in character. When the aqueducts emerge on the hillside 
above Tivoli we find the four specus distinct from one another once more. 
There is a very interesting point where from a reservoir of the Anio 
Novus a branch channel runs off, falling sharply (about 1 in 10), and sup- 
plying when required, by means of vertical shafts, the channels of the 
three lower aqueducts. 
After passing the point of junction of the tunnel built by Paquedius 
Festus, the next feature of interest is the fine bridge known as the Ponte 
§. Antonio, which served to carry the Anio Novus across a deep and 
narrow valley. We may note here a right-angled turn, which often occurs, 
to break the speed of the water immediately before reaching the bridge. 
The channel is surprisingly narrow, being only 80 centimetres wide and 
about 3°12 metres high. The bridge was originally a massive structure 
in ashlar masonry of volcanic tufa, and the fine central arch, 32°30 metres 
in height and 10°40 in span, is still visible on the westside. The width was 
originally only 2°60 metres and the total length is about 120 metres. 
The channel was probably in concrete faced with opus reticulatum, at any 
- rate at the ends of the bridge, where it is still visible. The whole structure 
was enclosed, in post-Severan times, with brick-faced concrete, with 
smaller arched openings. In the centre there were four of these, one 
above the other, flanked by huge buttresses. Reinforcements of concrete 
faced with pieces of aqueduct deposit from the channel were added still 
later. . 
In the next valley, that of the Mola di S. Gregorio, there is a long 
bridge of the Anio Vetus, which, however, is a construction of the time 
of Hadrian, itself restored later—the change of period occurs in the arch 
over the stream. The original channel ran underground up the north 
bank of the valley until it could pass under the stream, and then returned 
on the south bank, where its channel may still be seen. It is, after all, 
unlikely that in 270 B.c. the Romans would have constructed an aqueduct 
above-ground which could so easily have been cut by an enemy, and 
Augustus followed the older line in his reconstruction. The bridge has a 
rapid descent of 2°92 in 25°30 metres, or 1 in 8°66, or 116°5 per 1,000, 
at the end (the only case known to me at the end of a bridge) into the 
newer channel, which is some six metres higher than the older channel, 
with which it seems to have no communication. It continues to run for 
some way along the valley before it turns at right angles to tunnel through ~ 
the ridge separating it from the next one, that of Ponte Lupo, with which ~ 
we shall presently deal. One of the bricks in the cornice of this descent 
4 
* 
2 
