e adjacent ditches.
The plan which I gave last year has now to be corrected so as to show
a triple ditch between the south gate and the south-east corner and a
double ditch from the south-east corner to the east gate. This latter
section of ditch was, however, filled up at some time with clay, and the
bath planted on top of it. At presumably the same time a ditch was run
out from the south-east corner so as to enclose the bath and form an
annexe; in this annexe was found a broken altar-top with a few letters
on it (below, p. 29). Search was also made for rubbish-pits on the north
side of the fort, but without any result.

On other parts of the Wall Dr. Macdonald has gained further successes.
Evidence seems to be coming out as to the hitherto missing forts of
Kirkintilloch and Inveravon. More details have been secured of the fort
at Mumrills--fully 4-1/2 acres in area and walled with earth, not with
the turf or stone employed in the ramparts of the other forts of the
Wall. The line of the Wall from Falkirk to Inveravon, a distance of four
miles, has also been traced; it proved to be built of earth and clay,
not of the turf used in the Wall westwards. Dr. Macdonald suggests that
the eastern section of the Wall lay through heavily wooded country,
where turf was naturally awanting.

(iii) _Traprain Law._ Very interesting, too, are the preliminary
results secured by Mr. A. O. Curie on Traprain Law. This is an isolated
hill in Haddingtonshire, some twenty miles east of Edinburgh, on the
Whittingehame estate of Mr. Arthur Balfour. Legends cluster round it--of
varying antiquity. It itself shows two distinct lines of fortification,
one probably much older than the other, enclosing some 60 acres. The
area excavated in 1914 was a tiny piece, about 30 yards square; the
results were most promising. Five levels of stratification could be
distinguished. The lowest and earliest yielded small objects of native
work and Roman potsherds of the late first century: higher up, Roman
coins and pottery of the second century appeared, and in the top level,
Roman potsherds assigned to the fourth century. One Roman potsherd, from
a second-century level, bore three Roman letters IRI, the meaning of
which is likely to remain obscure. As the inscribed surface came from
the inside of an urn, the writing must have been done after the pot was
broken, and presumably on the hill itself. Among the native finds were
stone and clay moulds for casting metal