nearest the centre, while in the Fondaco it is
farthest from the centre. The first loss here is of eleven inches, the
second of five, the third of five, and then there is a most subtle
increase of two inches in the extreme arches, as if to contradict the
principle of diminution, and stop the falling away of the building by
firm resistance at its flanks.

I could not get the measures of the upper story accurately, the palace
having been closed all the time I was in Venice; but it has seven
central arches above the five below, and three at the flanks above the
two below, the groups being separated by double shafts.

Sec. VIII. Again, in the Casa Farsetti, the lower story has a centre of
five arches, and wings of two. Referring, therefore, to the last figure,
which will answer for this palace also, the measures of the intervals
are:

Ft. In.
_a_ 8 0
_b_ 5 10
_c_ 5 4
_d_ and _e_ 5 3

It is, however, possible that the interval _c_ and the wing arches may
have been intended to be similar; for one of the wing arches measures 5
ft. 4 in. We have thus a simpler proportion than any we have hitherto
met with; only two losses taking place, the first of 2 ft. 2 in., the
second of 6 inches.

The upper story has a central group of seven arches, whose widths are 4
ft. 1 in.

Ft. In.
The next arch on each side 3 5
The three arches of each wing 3 6

Here again we have a most curious instance of the subtlety of eye which
was not satisfied without a third dimension, but _could_ be satisfied
with a difference of an inch on three feet and a half.

Sec. IX. In the Terraced House, the ground floor is modernized, but the
first story is composed of a centre of five arches, with wings of two,
measuring as follows:

Ft. In.
Three midmost arches of the central group 4 0
Outermost arch of the central group 4 6
Innermost arch of the wing 4 10
Outermost arch of the wing[43] 5 0

Here the greatest step is towards the centre; but the increase, which is
unusual, is towards the outside, the gain being successively six, four,
and two inches.

I could not obtain the measures of the second story, in which only the
central gro