mediæval times, which also intrudes itself to intensify the confusion, is probably
likewise straight-sided. the prehistoric fort is, on the other hand, as an invariable
rule, curvilinear, either oval or circular or of some irregularly round figure such as the
eccentricities of the site demanded. The curvilinear forts, further, may be constructed
of stone, with or without entrenchment, or they may be pure earthworks with ram-
parts and ditch, or a combination of stone and earthwork; and lastly, their ditches
may be entirely excavated in soil, or their construction may have necessitated the
cutting of rock. As the rectangular oblong fort, by its shape and certain considera-
tions of situation, may, with some degree of certainty be recognised as Roman, if
the mediæval element be disregarded, so the curvilinear fort, the trench of which
is cut through rock, may be presumed to belong to the Iron Age. We are thus
left with stoneworks and simple earthworks to assign to their proper periods.
This is probably not possible by the aid of superficial observation, as the builder
of an Iron Age fort might be so fortunate as to find no rock to interfere with his
entrenchment, or the erection of a stone fortress might be his simplest and most
effective contrivance on a particular site which he deemed it necessary to occupy.
To lift the veil and indicate the features which may be peculiar to any given period,
spade-work is necessary, and until that is forthcoming the chronology of our numerous
forts must in large measure remain in doubt. Environment, comparison with
excavated examples, and occasional discoveries of associated relics are all factors
which may be called in to help, and the cautious use of these may enable us to
determine the period to which certain of the Dumfriesshire forts belong.
First, as being least open to doubt, let us take the Roman forts into considera-
tion. Of these there are four which by plan as well as by the positive evidence
afforded by excavation are assuredly of this origin: Birrens (No. 462), two at Birrens-
wark (No. 272), and Raeburnfoot (No. 172). Another, Gilnockie (No. 45), by the
details of its plan and by analogy with ascertained examples elsewhere, there can be
little doubt merits a similar attribution. But, after accounting for these five, we
have still a number of rectilinear forts in the county concerning which only the ap-
plication of the spade will suffice to determine whether they are mediæval or Roman.
Certain of their characteristics may be pointed out. One only of them, the fort at
Kirkmahoe Manse (No. 340), is large, and it is in a very fragmentary condition. The
site, a plateau flanked on one side by low marshy ground and on the other by a steep
bank overlooking haugh-land that stretches to the Nith, is such as a Roman general
might well have selected. But when we say that the site alone presents no inherent
impossibility to such an attribution, it is as much as we are justified in asserting.
Proceeding up Nithsdale for some miles to Durisdeer, in the glen of the Kirkburn,
at the entrance to the Well-path, a pass which leads through wild hill country into
Clydesdale, we find another small oblong rectangular earthwork of uncertain origin
(No. 163). The trench which surrounds it is boldly cut, and for some distance its
course has been hewn through rock, which does not suggest that we have here a mere
temporary encampment. The entrance is in the middle of one end, and some 24
feet in front of it there has been dug an outer ditch or traverse, a feature quite
consonant with Roman methods. Here again the spade alone can decide, but, as
in the case of the fort at Kirkmahoe, there is no inherent impossibility of a
Roman origin.
Passing into Annandale, we find in the parish of Middlebie, at no great distance
from Birrens, two constructions which deserve some notice.One at Purdomstown
(No.466), adjoining, and partly covered by, the Annan Waterworks, is a quasi-

-- li

  Transcribers who have contributed to this page.

valrsl- Moderator

  Location information for this page.