the Brythonic tribes were largely aided in their conquests by their iron weapons. Some iron agricultural implements may, indeed, have reached the Goidelic tribes before their conquest by Brythons; but it is hardly likely that the Brythons would strengthen their enemies by selling them iron weapons of war. There is, indeed, no record of the discovery of any prehistoric iron weapon in Breconshire,' but such finds are notoriously rare, as iron so rapidly rusts away in the earth. Whether the "crannog" on Llangors lake was the work of men who were acquainted with iron, there does not seem enough evidence to say.

In Roman times, the men of the south-eastern portion of Wales were known as the Silures, but their precise boundaries cannot be determined with certainty. As to their appearance, the classical passage is that contained in Tacitus, Agricola, c. xi; where he calls attention to the different physical characteristics of the inhabitants of different parts of Britain, and indicates the probability that these differences could be accounted for by a difference in the country of origin of each section. The Caledonians resembled the Germans, the Silures the men of Spain, and the inhabitants of the parts nearest Gaul the men of that country. It should be borne in mind that, in Graeco-Roman times, Spain was thought to be much nearer to western Britain than it really is. It is interesting to note that Tacitus had observed a clear difference in physical appearance between the men of the south-east of England and the Silures: the probable explanation being that, in the case of the latter, the bulk of the population was of the old pre-Celtic race. The oftquoted words of Tacitus are: "Silurum colorati vultus, torti plerumque crines, et posita contra Hispania Iberos

1 Nor any "Late-Celtic" object. The nearest discoveries of such objects are those of the gold ornaments of Cerriggwynion, in Radnorshire, on the one hand, and those of Dolaucothy on the other. A "Late-Celtic" collar was found in 1896 at Llandyssil, and is now in the Bristol Museum.

veteres traiecisse easque sedes occupasse fidem faciunt." If we turn to the Annals, Bk. xII, 31-40, we find that, in their great struggle against the Romans, the Silures were under the leadership of Caratacus (Caradog), whose name was thoroughly Brythonic, and who was evidently himself a Brython. Moreover, the account given by Tacitus clearly implies that Caradog was no alien to the Silures, but was able to address them in a tongue which they understood. In the whole of the account given by Tacitus of the stubborn and courageous resistance of the Silures to the Romans, there is no suggestion that they were linguistically different from the other tribes of Britain; hence we may legitimately conclude that their governing classes, at any rate, were, even at that time, Brythonic in speech. This does not preclude the possibility that, in the hilly country of the Beacon range, for example, and it may be, from there continuously to Gower and Kidwelly, the ancient Goidelic stratum was still dominant, especially as it could then be reinforced from time to time by sea from Ireland. After the departure of the Romans it is not improbable, either, that some of these hill-tribes, with help from Ireland, may have regained possession of the Usk Valley and the neighbouring districts, and that some such movement is indicated in the narrative of Brychan. It will be remembered that the districts of Gower and Kidwelly are expressly mentioned by Nennius as ones in which the sons of Liethan ruled, until they were expelled by Cunedda and his sons.

In discussing the ethnology of Breconshire, the writer has not found it possible, within the limits of this paper, to enter at all fully into the difficult question of the Ogam inscriptions. The discovery of an Ogam inscription so far east as Silchester, in a district which

1 The ancient name "Abone" near Venta Silurum (Caer Went), seems more Brythonic than Goidelic, the old Irish form being 'abann,' river.

could hardly have been Goidelic,' makes one chary of drawing far-reaching ethnological inferences from two or three Ogam inscriptions, found, as they are in Breconshire, in the neighbourhood of an ancient avenue of communication between Ireland and parts of the west of England, such as seems to have run through the Usk Valley. Moreover, as Principal Rhys has pointed out, the Latin forms of the names found on bilingual Ogam inscriptions show clearly that Brythonic was socially the dominant Celtic language, though Goidelic may have existed in a position of inferiority. Nor is it safe to assume that the Ogam script was never used to write Brythonic as well as Goidelic, especially as the use in Ogam of "tt" for "th," and "cc" for ch,", would have been suggested, not by Goidelic, but by Brythonic usage. It seems hardly likely that orthographical ideas would have been borrowed from Brythonic to be used only in Goidelic. There is no reason for thinking, however, that any of the Breconshire Ogams are written in Brythonic. The "Moqvutreni" (Ogam) and the "Maccutreni" (Roman script) of the Trecastell inscription are unmistakably Goidelic. The Trallwng and Glanusk Ogams seem to be themselves Goidelic, but the Latin inscription in each case, in the form of the proper names, suggests a Brythonic influence. Hence, the precise ethnological inference to be drawn from these inscriptions is uncertain.

66

3

A line of enquiry which may lead in course of time to a fuller knowledge of the Celtic invasions of the district, is the careful study of the ancient "British' camps, as compared with similar structures elsewhere. These, when carefully examined, might indicate the

1 See Principal Rhys in Report of the Land Commission, chap. viii. Such examples in Breconshire are Cunoceni, Dunocati.

2 The use of "tt" for "th," and "cc" for "ch," is mentioned by Principal Rhys in the Report of the Land Commission, chap. viii., and by the Hon. Whitley Stokes in his work on The Celtic Declension, in Bezzenberger's Beiträge, vol. xi., p. 144.

relations of the early Celtic tribes of the neighbourhood of Builth, Brecon, and Talgarth, and Crickhowell, all of which appear to have been important military

centres in ancient time.

Doubtless, considerable light would also be thrown on the ethnology of the district, by a careful comparison of the Welsh dialect of Breconshire with those of the neighbouring counties. Similarly, an anthropological study of the physical types of the county, such as was commenced by Dr. Beddoe in his Races of Britain, would no doubt yield important results.

In dealing with a subject such as this, further advance can only be made by following up various clues from different points of view. The clues may often be slight, and from the nature of the subject there is much room for error, but the combined result of these different investigations may lead to an approximately correct dóğa, even if we cannot obtain a clear and certain moτýμn. Let us hope that, sooner or later, this kind of work may be done for the whole country by means of a thorough Archæological and Anthropological Survey.

A SURVEY OF THE LORDSHIP OF HAVERFORD IN 1577.

BY HENRY OWEN, D.C.L. OXON., F.S.A.

THERE is at the Public Record Office (Land Rev. Misc. Book, vol. 238) a survey of the "Castle and Lordship of West Haverford with the Town and County of Haverford, otherwise Haverfordwest, late part of the possessions of Jasper, late Duke of Bedford," taken on the 14th May, 1577, by Robert Davy,' the deputy of John Herberte, Esq., the Queen's Surveyor for South Wales, together with the renewal of divers rents at the discretion of the said Robert and of Maurice Canon,2 gentleman, the deputy of Sir Edward Mansell, the Seneschal of Haverford..

The survey begins with the following memorandum :

Fo. 20a.- "The said Castell and Towne of Haverfordwest are scituate within the Countie of Pembrooke aforesaid adioyning unto a Creeke of Milforde wch floweth into the lande a quarter of a myle above the said Towne and Castell being of such depth as at a spring tyde a Shippe of x1 tonne maie come harde to the Towne: And within iiii myles of the said Castell & Towne viz at Knapwood Roades a Shippe of greatest burthen maie come wch said Castell and Towne are xii myles from the mowth of Milforde aforesaid v myles from the Towne of Pembrooke and ix myles from the Towne of Tinbye.

1 Receiver for South Wales in 1595; see Owen's Pembrokeshire, I, 566.

2 He was the father of Sir Thomas Canon, the antiquary. The family owned Cilgetty, which passed to Picton Castle upon the marriage of Elizabeth Canon with Edward Philipps.

3 Above Langum; it is mentioned by George Owen among the thirteen 'roades' of Milford Haven.

4 The surveyor's mileage is vague, as it generally was until the present statute mile was fixed by 35 Eliz., cap. 6, s. 8.