r, an open ditch, or catch-water, is
the most effectual, as well as the most obvious mode. There are many
instances in New England, where lands upon the lowest slopes of hills
are overflowed by water which fell high up upon the hill, and, after
passing downward till arrested by rock formation, is borne out again to
the surface, in such quantity as to produce, just at the foot of the
hill, almost a swamp. This land is usually rich from the wash of the
hills, but full of cold water.

To effect perfect drainage of a portion of this land, which we will
suppose to be a gentle slope, the first object must be to cut off the
flow of water upon or near the surface. An open ditch across the top
would most certainly effect this object, and it may be doubtful whether
any other drain would be sufficient. This would depend upon the quantity
of water flowing down. If the quantity be very great at times, a part of
it would be likely to flow across the top of an under-drain, from not
having time to percolate downward into it.

In all cases, it is advised, where our work stops upon a slope, to
introduce a cross-drain, connecting the tops of all the minor-drains.
This cross-drain is called a _header_. The object of it is to cut off
the water that may be passing along in the subsoil down the slope, and
which would otherwise be likely to pass downward between the system of
drains to a considerable distance before finding them. If we suppose the
ground saturated with water, and our drains running up the slope and
stopping at 4 feet depth, with no header connecting them, they, in
effect, stop against 4 feet head of water, and in order to drain the
land as far up as they go, must not only take their fair proportion of
water which lies between them, but must draw down this 4 feet head
beyond them. This they cannot do, because the water from a higher
source, with the aid of capillary attraction, and the friction or
resistance met with in percolation, will keep up this head of water far
above the drained level.

In railway cuttings, and the like, we often see a slope of this kind cut
through, without drying the land above the cutting; and if the slope be
disposed in alternate layers of sand or gravel, and clay, the water will
continue to flow out high up on the perpendicular bank. Even in porous
soils of homogeneous character, it will be found that the _head_ of
water, if we may use the expression, is affected but a short distance
by a drain a