instances. In fact, it is the merest negative presumption to assume that, because no metal is found in a grave, it must be prehistoric. It may be of any age, down to yesterday's, in so far as such proof is concerned.
Even the presence of metal, however, does not disturb the faith of some antiquaries. The Baron de Bonstetten, for instance, opened a tumulus not far from Crubelz. At one foot (30 centimetres) below the undisturbed surface the usual deposit of flint implements was found; and two feet (60 centimetres) below them two statuettes of Latona in terra-cotta and a coin of Constantine II. were found, but without this in the least degree shaking his undoubting faith in the prehistoric antiquity of the tomb![1]
Numerous other Roman coins have been found in these French monuments, but their testimony is disregarded. In the Manné er H'roëk, commonly called the Butte de César, about half a mile south from Locmariaker, near the surface, eleven medals of the Roman emperors, from Tiberius to Trajan, were found, together with fragments of bronze, glass, and pottery, but there were no signs of a secondary interment.[2] In like manner, in another monument at Beaumont-sur-Oise, Roman moneys were found, but, as M. Bertrand is careful to explain, in a stratum above the stone and flint implements, which, of course, he believed to mark the true date of the monument.[3] It seems impossible, however, that all these Roman coins can have been accidentally placed there. Those of Valentinian and Theodosius in the mound at New Grange were precisely in the same position as those of Titus, Domitian, and Trajan in the Butte de César or those of Beaumont, and so were those of Constantine found at Uley, in Gloucestershire (ante, p. 165). Those of Valentinian at Minning Lowe were in the tomb itself; so probably might others have been found in the other tombs had they not previously been rifled. It is not easy to assign a motive for placing these coins in the upper part of the mound externally. Their being found in that position at New Grange, Uley, Locmariaker, and Beaumont, is, however, sufficient to prove it was not
