The Britannica article gives a very imperfect record of the early forms of the name. But in all probability it is correct in holding that the usual explanation fort on the Taff must be wrong. No early writer ever calls it Caerdaf (which would be the proper Welsh spelling if this were so), unless we make exception of the English antiquary Leland, in the days of Henry VIII, and he was only writing down his own guess. The earliest spelling now known is of date 1128, Kardi; a little later we find Kardid, whilst in the Pipe Roll for 158-59 we have Cardif. The modern Welsh is Caerdydd, pronounced Caer deethe. These forms suggest the meaning fort, castle of Didius.' Within the last few years it has become certain that Cardiff stands on the site of a Roman fort; and so this Didius will probably be that Roman general who, in A.D. 50, fought against the Silures, the British tribe who inhabited this region. If this conjecture be right, Cardiff will take rank as one of the earliest known Reman stations in the British Isles." Upon this last statement one would like to hear the observations of a competent local antiquary such as Mr. John Ward. We believe that there are arch- aeological reasons for thinking that the Roman fort at Cardiff was of comparatively late construction, and on the other hand there is the curious fact that it was entirely omitted from the Antonine Itinerary. Upon the actual archaeology of Wales the author in some places is a little weak., e.g., we get the astonishing statement that no Roman in- scriptions have yet been found in Wales. The chapter entitled The Keltic Element is chiefly concerned with the extent of that element in the names of England proper. The conclusion is that the deeper and more thorough the investiga- tion, the smaller seems the sure Keltic residuum," and the large claims that have sometimes been made cannot be substantiated. Except for river names there are very few over the whole country that can be definitely established as Keltic, even in such counties as Westmoreland and Durham. In Northumberland the} are numerous, and down the north-west coast from Cumberland to Cheshire, equally so along the so-called Keltic fringe and in Monmouth, which is really Wales, and in Cornwall they are in the majority. On the other hand the river names are predominantly Keltic. About all the rivers of any size there is no doubt at all, and to most of the smaller ones the same remark applies. The probable explanation is that the rivers unlike most other natural features belong to many districts and to avoid confusion the Saxons kept the old names on, and adapted their tongues to them as best they could." Naturally, in passing from England to Wales we find that the solid core of the great majority of the names has changed from English to Welsh (i.e., the Brythonic form of the Keltic speech), but the centre of interest is still the same, viz., to determine how the other elements have distributed themselves round that solid core. The Roman element is practically negligible, and the few Latin names that we know are almost all simply Latinized forms of the Keltic name. This tends to confirm Professor Haverfield's proposition that the Roman occupation of Wales was primarily a military guard for the mining operations, and not a colonization. The Saxon element, i.e., really ancient English names, are fairly frequent along the fluctuating border, but in many cases there are different names for the same place in the two languages. A good instance right in the heart of Welsh Wales is Snowdon, which certainly goes back to 1140, with the local alternative of Y Wyddfa. The Saxon and the later Norman influence is naturally most potent where the occupation was most thorough-in the Vale of Glamorgan, Gower, and the Englishry of Pembroke- shire. But the old names have shown remarkable vitality in many of these districts, e.g., in the South- west of the Gower Peninsula, where no Welsh has been spoken for centuries. The county names form a very interesting group that considerations of space forbid our entering into. Five of the twelve bear non-Welsh names. The other great non-Keltic element is given by the coast names. Almost all of these are English or Norse, the Norse predominating in Pembrokeshire and along the south coast as far as Flat Holm, the English in the remainder. There are many more points of interest discussed in the somewhat discus- sive chapter on Welsh names, that we cannot now mention. The body of the work, of course, contains details of all the Welsh names of importance, and it is certainly indispensible for the proper study of the subject. Transactions of the Carmarthenshire Antiquarian Society, 1914-15." Printed at The Welshman, Carmarthen. In spite of this devastating War, the Carmarthen- shire Antiquarian Society has been able to issue an interesting volume of Transactions, and to carry out valuable antiquarian work. Through its instru- mentality the Carmarthen Castle Gatehouse has been