1907 SHOUT NOTICES G13

worth while to reprint the whole of Grote; e.g. part i. of his work clearly had to go, and also his chapters on oriental history : but it is a far more truncated book the editors give us, one practically confined to the sixth and fifth centuries, and marked by great omissions even as to the first of these. Again, it was no doubt impossible to reprint Grote's notes in full, but we should have welcomed more of them, even at the cost of losing some of the editors' comments. They taught many generations of scholars a critical use of texts, and still have their value in that way. Above all we must complain of the tampering with the text of Grote. This surely ought to have been reproduced in full; but omissions and alterations are made in it without warning ; e.g. on p. 321 nearly a page

attributing the vo/io<£u'/Wes to the time of Pericles is omitted, while Downloaded from Grote's view as to the vofioOhai becomes ' possible ' instead of ' probable.' Still worse is the case in the story of the Persian wars, e.g. p. 199, where the text is shorn without any warning of the story of the first Greek retreat from Artemisium and of the corruption of Theinistocles. Hero- dotus's stories may be false, but at any rate Grote accepted them without

question, and it is misleading to make him omit them. However it is http://ehr.oxfordjournals.org/ better to read even a truncated Grote than not to read him at all, and the book deserves and ought to meet with success. J. W.

In The Roman Forts on the Bar Hill, Dumbartonshire, by Dr. George Macdonald and Mr. Alexander Park, with a note on the archi- tectural details, by Mr. Thomas Boss (Glasgow: Maclehose, 1906), we have a business-like account of the excavations undertaken in

1902-5 at one of the most important stations on the vallum of Anto- at Université Laval on May 11, 2015 ninus Pius. The entire cost of these excavations was defrayed by Mr. Alexander Whitelaw of Garshore, to whose liberality Scottish archaeologists already owed much. Mr. Whitelaw's public spirit de- serves the fullest recognition. Such generosity has made it possible in recent years for antiquaries north of the Border to take many steps towards forming a coherent account of the Roman occupation of Scot- land. The chief interest of the excavations lies in the fact that they revealed the existenca, beneath the fort contemporary with the vallum, of an earlier structure, whose outermost was turned into the inner fosse of the later castelku'n. It was no doubt in order to utilise the site already occupied that the Eoman builders departed from the practice elsewhere observed of applying the fort to the southern side of the vallum ; for the Bar Hill fort lies some twenty-five yards to the south of the . Between the two intervenes the ' military way.' Now the earlier fort, represented by its fosse only, was evidently occupied for a period not exceeding half a century, and it follows with practical certainty that we have here one of the praesidia built by Agricola in A.D. 81.l We are thus brought into contact, as at Melrose, with the work of that commander: the choice of the site illustrates his sagacity in selecting points of defence 2 and at the same time the smallness of the force which he employed to hold Northern Britain ; for its extreme measure- ments are 191 x 160 feet. Both the earlier and the later forts are chiefly interesting as showing the ingenuity with which the system of 1 Ta?. Agric. 23. J Ibid. 22. 614 SHORT NOTICES July fosses was adapted to the defence of weak points. Thus the fosse surrounding the Antonine fort is double except on the north side, where the vallum gives additional protection; the east and south gates are further defended by a titulus,3 and the former of these is con- siderably prolonged on account of the ' dead ground ' in front of it, which might give cover to an advancing enemy. Finally, on the west side Agricola's ditch was never filled up, and the we3t gate could only be approached by a plank bridge. It should also be noted that the structure of the turf rampart is even more clear than that of the vallum itself (which was still a matter of argument at the time when the Antonine Report was published); it was made of sods placed grass to grass in regular layers which break joint. Both in this particular and also in Downloaded from the form of the ditch, which is V-shaped with a central flat-bottomed de- pression, the turf wall observed near Birdoswald furnishes a close parallel, which may prove significant. Among the minor finds the most inter- esting are a wooden chariot wheel, doubtless of Celtic manufacture, a large and varied assortment of foot-wear, and some sham denarii of tin, made for devotional use. There is nothing here, as elsewhere on the http://ehr.oxfordjournals.org/ vallum, to indicate that the occupation lasted beyond the time of Commodus. H. S. J.

Count Ugo Balzani's excellent Italian translation of Mr. Bryce's Holy Roman Empire has appeared in a second edition (II Sacro Romano Impero. Milan: Hoepli, 1907), in which the extensive alterations, as well as the supplementary chapters, contained in the English edition of at Université Laval on May 11, 2015 1904 are incorporated. It is sufficient here to notice this fresh testimony to the permanent position which Mr. Bryce's book has won in the literature of Europe. Z.

In the Bibliotheque de VEcole des CMrtes, lxvii. 5, 6, M. Leopold Delisle publishes a paper of considerable importance for the chronology of the documents of the reign of Henry II of England. It has been long recognised that the use of the formula Dei gratia in the king's title is characteristic of the later years of his reign, and M. Delisle seeks to determine more precisely the time at which the insertion began to be made. He arrives at the conclusion that the change was made at some date after May 1172 or at the beginning of 1173. That it was connected with Henry^s absolution from complicity in the murder of Archbishop Thomas at Avranches on 21 May 1172 is suggested, but not pressed. It might have been added that the king's second submission at Avranches on 27 September would furnish an equally appropriate occasion for the adoption of the new formula. But the main point on which M. Delisle lays stress is this : that, by itself, the absence or presence of the phrase Dei gratia at once decides a document to be earlier or later than the beginning of 1173. It will be interesting to see whether this theory can be maintained in its absolute form. If it can, it will be a great assist- ance to the future editor of Eyton's Court, Household, and Itinerary of Henry II. E. L. P. = Hyginus, 'Castr. 43.