Britain; a buttressed building of the type most likely to be a storehouse; an officers' residence (?) like those at Housesteads and Gellygaer, and various barracks of the usual oblong shape. Outside the fort to the east is an annexe with a building three hundred feet long, while other structural remains occur outside the west front.

No less striking are discoveries of more or less portable objects in various pits. A huge hole twenty-six feet deep was noted and excavated in the head quarters courtyard. It yielded an altar (twelve feet deep), and below that worked stones from the building above, human bones of one or two men, and (at twenty-two feet) some three hundred scales of bronze scale armour. Outside the fort, other pits yielded a great quantity of armour, helmets in bronze and iron, shoulder and elbow pieces, portions of a cuirass, an iron sword, a chariot wheel almost perfect, metal disks as if for phalerae, and so forth. Several of the objects are unique: all are remarkable, especially an iron helmet with a fixed visor in the form of a human face, of singularly good workmanship. No such finds have been made in Britain for many years. In the way of inscriptions we have an altar—I. O. M. C. Arrius Domitianus c. leg. xx.v.v.v.s.l.m.— "To Juppiter Opt. Max. dedicated by C. Arrius D. centurion of the Twentieth Legion." The name "C. Arrius" passed, in the newspapers of the time when it was found, through a curious set of corruptions, becoming in turn CARRIVS, CARRLVS, and lastly (in The Daily News) CAROLVS. Much of the armour and other objects also bore the names of soldiers who had used them-Senecionis, Dometi Attici, Lucani, and the like. Smaller objects found in these pits or elsewhere in the excavations included bronze and earthenware vessels, "Samian," of the first and second centuries, and about ten dozen coins, many of them Flavian and none later than about A.D. 180.

Further excavation will doubtless explain much that is puzzling in these finds. But their historical value is already apparent. The older ditch and the earlier coins and pottery must date from Agricola, and Melrose thus takes its place

beside Bar Hill, Camelon, and Inchtuthill. As is clear at Bar Hill, the Agricolan fort was probably earth-walled and the occupation brief. The work of this general, it seems, did not last north of the Cheviots, or perhaps the government of Domitian did not care to maintain outlying conquests. Fifty-five years later came a new advance under Pius, and the fort of Melrose was re-occupied, slightly enlarged and fitted with stone walls and buildings. This occupation lasted longer than the first, but shorter than has mostly been supposed. The coins found at Melrose show that it, like the rest of Caledonia north of the Cheviots, was lost or given up about A.D. 180. Whether it was re-occupied later is at present doubtful. The invasion of Severus in A.D. 208 probably brought Roman troops through Melrose. But if his troops occupied the site, we should probably have found his coins in the recent excavations.

Other military sites compete feebly with Melrose, but some have yielded interesting results. Ribchester, in Lancashire, has revealed its head quarters—a building said to measure eighty-one by one hundred and thirty-seven feet, and to contain two courtyards with columns, in a usual fashion. At Melandra, Manchester, the local archaeologists have cleared more of the fort, and made the head quarters more intelligible. They have also shown that much remains to be found here by properly supervised and systematic search. At Manchester, the area of the fort has been more closely defined, and some stone buildings traced. At Holt on the Dee, near Wrexham, stamped tiles suggest either a brickfield or an outpost of the Twentieth Legion, the Chester garrison. In South Wales, at Merthyr, the discovery of a "storehouse" indicates a fort, no doubt intended to coerce the hillmen and to guard a neighbouring road. At Colbren, on the Roman road from Neath (Nidum) to Y Gaer, near Brecon, the site of what may be an early Roman fort has been trenched, though somewhat inconclusively. In sum, our knowledge of the military aspects of Roman Britain has advanced in many details.

Non-military sites have yielded fewer and less noteworthy finds. The systematic uncovering of Silchester and Caerwent has proceeded steadily, and the ground plans of these two Romano-British towns have been further recovered. The long and patient labours of their excavators deserve all praise, and the scientific value of the results, when completed, will be very great. But the work actually done at the two places in recent months offers no features of individual interest. A little more-principally mosaic floors-has been uncovered of the Roman remains of Colchester, of Cirencester, of Dorchester, and even of Winchester; this last a scanty gain of good from the ruin of the cathedral foundations. Villas have been explored or detected at Lippen Wood; in the Meon Valley on Lansdowne, north of Bath; at Welwyn, near Watchfield, in Berkshire; and possibly also at Middleham, near Jervaulx Abbey, in Yorkshire. A hoard of coins found in Grovely Wood, in Wiltshire, supplies a new example of the curious hoards of late fourth-century silver coins which are almost peculiar to Britain, and which I have treated collectively in The Victoria History of Somerset. A hoard of some seven thousand Constantinian copper coins found at Stanley, near Wakefield, is of more usual character. Two inscribed "Aucissa fibulae, rescued from the Ham Hill quarries in Somerset, and two found probably in North Lincolnshire, form additions to an interesting class of early first-century brooches.

In respect of literature I should wish to mention Dr. G. Macdonald's account of the very interesting fort of Bar Hill, near Glasgow (Glasgow, MacLehose), which forms an admirable account of important excavations, and to the Manchester Classical Association's volume on Melandra, edited by Professor Conway (Manchester University Press). I may also be, perhaps, permitted to allude to the progress of "The Victoria County History."

F. HAVERFIELD.