Trajan to advance against a dangerous enemy through difficult hills with his troops divided. The plain man would probably say that the army made its two bridges side by side, and would find a place for both at a point near the little town of Bazias, where, for uncounted ages, men have crossed from one side of the Danube to the other by an easy ferry.

We have been occupied in suggesting that, where historians of the present day cannot apply new facts obtained by new methods, they are not much wiser than their predecessors; they are only a good deal more ingenious-and perverse. Mommsen transformed Roman history by introducing new methods and unearthing multitudes of new facts. He was like the War Minister who draws recruits from wider and wider areas, or the Finance Minister who broadens the basis of taxation. The new evidence which Mommsen and indeed the whole nineteenth century brought forward was that to be gained from inscriptions. On the value of these there is no need to waste many words at the present day. They have, of course, taught us well-nigh an infinity of new knowledge; and the study of them is not in the least exhausted. But we wish now to outline rather some of the other sources, and in particular the unwritten sources, from which Roman historians may expect a vast increase of knowledge. The ordinary reader is likely to overlook these things. He gets excited at the prospect of new texts emerging from the tombs of Egypt or the libraries of Greece and Turkey. He forgets that very few of these new texts concern Roman history, and that the contents of those few are chiefly technical; discoveries like the bit of Abridged Livy' which was dug out of Oxyrhynchus in 1904 are not only very rare, but appeal only to specialists. He forgets also the yet more important fact that the earth around him, in England or France or Germany, is full of precious stuff, though not of texts.

It is not too soon to ask such readers to cast a glance at this new stuff. It is true that scholars have only just begun to work seriously at it. The methods of inquiry are not yet fixed. The evidence has been only to a small extent collected. The excavations needed to tell us about the life of the Roman provinces have not been carried very far; and too often they have been badly and

ignorantly conducted. Many of the inquirers, like a doctor with a new drug, expect a great deal too much. Our successors, half a century on, will look back with wonder and regret at the imperfection of our methods, the narrowness of our endeavours, and the large proportion of evidence which we have spoilt in the process of collecting. Nevertheless something has been done, something that deserves description.

We may begin on the rubbish-heap. The study of smashed crockery is perhaps to-day one of the most important and promising studies of a Roman historian. Everyone knows the hard red 'sealing-wax' ware which we in England call Samian and foreign scholars call Sigillata. It was the ordinary good china' of the earlier Roman Empire, and was used throughout its length from Anglesea to Assuan. This ware was mostly made in Gaul or Germany; and there too it has been studied with much industry in the last few years. German enquirers led the way some fifteen years ago. Now a distinguished French archæologist, M. Déchelette, has worked out, laboriously, minutely and penetratingly, the history of its manufacture at certain Gaulish potteries. The details of its production and distribution form a curious though as yet only half-written chapter in the history of Roman trade. It shows us, for example, a provincial industry, fed no doubt by Roman capital, growing strong enough to compete with Italy, and driving Italian products not only out of Gaul, but out of Italy itself; and all this occurring quite early in the Roman Empire, perhaps in the first half of the first century.

Here, however, we are concerned with it rather as a help to the chronology of the Empire. The ware was manufactured in various shapes and with varying ornamentation; and to some extent a gradual evolution of forms and a development of ornament can be traced through nearly two centuries, and can be dated within rough limits. Several difficulties have so far hindered the attainment of great precision. For one thing, the Roman Empire was an age in which material civilisation and artistic fashions changed somewhat slowly. For another, the clues for fixing dates are few. Worcester or Dresden china can be set down to its proper age by documentary evidence that certain colours and designs were employed

at certain times. Samian has no such records and no such variety of colours; and, though many books, besides M. Déchelette's great work, are beginning to appear on the subject of its chronology, and most of these fix precise epochs for separate forms, fresh discoveries are continually upsetting theories. It is obvious that in too many cases the dating is dictated by the enthusiasm of the pioneer. Even the dates advanced by M. Déchelette appear to need occasional reconsideration in detail.

Yet, even so, the study of Samian has taught us much. A good instance may be found in the history of our own island during the Roman period, and, indeed, during the latter part of the first century. It was well known, long before anyone thought of dating Samian, that Agricola, one of Vespasian's best administrators and for several years governor of Britain, invaded Scotland, planted there a number of forts, and fought a final battle at the foot of the Graupian Hill. That is told us by Agricola's son-in-law, the historian Tacitus. Tacitus, however, had no maps to illustrate his pages after the manner of a modern school-book, and he probably had no particular interest in the precise positions of Agricola's fights or fortifications. The world was therefore left to trace Agricola's footsteps along a very scanty scent; and, though the fertile imagination of the Scottish mind has done its wonders, nobody recovered much solid historical fact. Now, at last, Samian potsherds have come to our aid; the conclusions worked out by M. Déchelette and others, notably Prof. Dragendorff, of Berlin, give real help. We find that pieces which can be dated to Agricola's time have been dug up on various sites in northern Britain; and thence historical deductions follow. We know with fair certainty that Agricola had a post at Carlisle. We have some reason to think he besieged the great native hillfort of Birrenswark, which overlooks the moors of northern Cumberland and southern Scotland. We know, too, that on the east coast-doubtless his principal line of advance-he built a road from the Tyne at Corbridge across Cheviot to a large fort overhanging the Tweed beside Melrose, and thence northwards to the Forth and beyond it. We can trace some of the blockhouses which (as Tacitus tells us) he planted across the narrow isthmus between the Forth and the Clyde, and we can track his

steps yet further on, past Stirling and Perth. At the point where the valleys of the Tay and the Isla join-the point, as it happens, where the Highland and Caledonian railways also meet-his army lay for months in a large encampment still visible among the policies of Delvine. At all these spots the spade has revealed bits of Samian, and occasionally other things, belonging to the age of Agricola, which serve as proofs of his or his soldiers' presence, and which enable archæologists to assign to them various earthworks and fortifications.

At Delvine the story stops for the present. Other sites further north-east towards Forfar and Aberdeen, and even beyond, show other earthworks which bear every mark of Roman engineering. In due time these in their turn will be trenched, and fresh potsherds will emerge to tell us whether they date from Agricola or from some later Roman. As Prof. Dragendorff truly observes in a contribution to the Journal of Roman Studies,' much of the history of Roman Scotland depends practically on potsherds. And potsherds in such a matter are excellent guides. They are almost imperishable; they are exceedingly numerous; nearly all of them have obviously been broken or thrown away at no long period after their manufacture, and they provide what we may call really contemporary evidence. Here and there antiquated pieces may have lingered on; but Roman generals did not march through Scotland with cabinets of rare old china in their baggage. If they were foolish enough to take antiques on distant frontier campaigns, they would have taken pieces of silver or of bronze.

A yet longer series of facts in Roman frontier history has been recovered from Roman potsherds in southwestern Germany. The Roman boundary, which stretched from the Rhine near Bonn to the Danube near Regensburg, was watched by an elaborate system of forts and marked by a continuous frontier rampart of earth or stone. It was It was a less elaborate, but more extensive, work of defence than our walls in northern Britain; and it clearly was altered and moved forward at various dates when the Empire was strong or when danger threatened. The beginning of the occupation we know from three or four chance references in literature and inscriptions. Just before Agricola invaded Scotland,

another officer of Vespasian was moving east from the Roman fortress at Strassburg, conquering the Black Forest and the hills to the north of it, and joining hands with Roman forces in the upper valley of the Danube. But the actual remains of Roman rule east of the Rhine extend far beyond this narrow space; and the history of the further conquests can best be learnt by the aid of potsherds.

6

That history has not yet been fully worked out, but much is already known. An abundant literature'-if a mass of learned pamphlets can rightly be called literaturehas been appearing on this subject in the country where it is naturally most popular-western Germany; and an excellent English summary was read to the Royal Historical Society by the late Prof. Pelham, which is reprinted with a careful and detailed map in his collected 'Essays.' The result, so far, may be briefly indicated here. Potsherds and a few other pieces of chance evidence show that the Emperor Domitian followed up the advance of his father Vespasian, and—incapable soldier as he is generally deemed, even by Prof, von Domaszewski-did really add to the size and strength of the Empire in Germany. He seized and fortified on a definite system the range of Taunus and the fertile land which lies to the south of it round Frankfurt, and very possibly a good deal else to the south of that. Then, forty years later, Hadrian organised the whole line with a barrier of wooden palisading stout enough to stop the freebooter and the raider; and after his death a further advance was made along parts of the line, and fresh defences were built. The great military achievement of the Roman Empire was the safeguarding of the civilised lands within it against barbarian assaults through two centuries and a half. The German Limes is no small part of that work. That we can trace it so well as we can, is mainly due to the smashed Samian of the rubbish pits.

Potsherds are not the only ancient trifles which matter to the Roman historian of to-day. Brooches, rings, lamps and the like can also be dated more or less closely, though they may perhaps have been more often handed on from one owner to another than the plate which breaks so easily. But we prefer to turn to another kindred class of evidence, not for strict chronology, but for the history of