Essays and studies: by members of the English Association/English place-names

by Henry Bradley

ENGLISH PLACE-NAMES^[1]

The Ordnance Map of England contains many thousands of names—names of towns, villages, districts, rivers, mountains, and the like. From the point of view of the unlearned, these names may be divided into two classes: those which have an obvious meaning, and those which have not. As examples of the first class we may take Newcastle, Highbridge, Redhill, and Blackwater. These, and many more, are at once perceived by everybody to be significant, though topographical or historical knowledge may be needed to show the precise reason for which they were given. On the other hand, names like London, Kent, Thames, Helvellyn—in fact, the vast majority of the names on the map—are, to persons not specially instructed, mere arbitrary sequences of sounds, which might just as fitly have been applied to any other places or natural objects as to those which they actually serve to identify. At the same time, no intelligent person can doubt that even these names cannot have been given without some reason, and that if we had only sufficient knowledge we should find them just as significant as those which need no interpretation.

It is natural that curiosity with regard to the meaning of these enigmatical names should be very widely felt, and the writers who have attempted to satisfy this curiosity are innumerable. Unfortunately nearly everything that has been written on the subject is worthless.^[2] With very few exceptions, the scholars who have possessed the philological knowledge requisite for the scientific treatment of the subject have been so conscious of its difficulties that they have preferred to leave it alone. It has therefore fallen into the hands of unqualified persons, for many of whom it seems to have an unaccountable attraction. Their usual procedure is to ransack the dictionaries of Anglo-Saxon, Danish, Welsh, and other languages, and if they can find words in any of these which bear some resemblance to the syllables of the name to be explained, and which, when joined together without the slightest regard to grammatical rules, can be made to yield something like a plausible sense, they imagine that they have solved the problem of its etymology.

It must be admitted that the explanations arrived at in this haphazard fashion are often much more interesting than those which are the result of methodical research. And no wonder! An etymologist who can operate at will with the words of half a dozen languages, and has no inconvenient grammatical knowledge to hamper him in putting them together, is able to make a name mean almost anything he likes; and if he is a person of taste he will of course choose to find in it some bit of picturesque description, a reference to ancient beliefs or superstitions, or a memorial of some historical event. Fact is usually less entertaining than fiction, and for this reason false etymologies are to most people more attractive than true ones. An opinion which is widely prevalent, and sometimes frankly avowed, is that no certainty is attainable in the interpretation of place-names, and that therefore it is absurd to reject a pretty or amusing explanation merely because philological pedants, for some unintelligible reason, choose to assert that it is untenable.

Now I hope to show some reason for believing that by the use of proper methods, the origin of our place-names may often be determined with a high degree of probability, and even with positive certainty. That the matter is not of the highest importance may be freely admitted; but as false local etymology has in the past been the source of a great deal of historical error, so sound local etymology may sometimes be a valuable help in the discovery of historical truth. At any rate, if the subject is worth studying at all, it is worth studying with an honest desire to arrive at the truth, even though the truth should turn out, as it sometimes will, to be disappointingly commonplace.

Assuming, then, that we wish to know what our English place-names really do mean, and not to be amused with baseless fancies about them, let us consider what are the methods by which their meaning may be discovered.

A large proportion of the names on our maps are of great antiquity, and occur in existing documents, many of which are more than a thousand, and some nearly two thousand years old. When we have a name to interpret, our first step should obviously be to ascertain its oldest known spelling—or, if possible, the two or three earliest spellings, to avoid the risk of being misled by some ancient scribe's blunder. Of course we have to make sure that the name in our early document is that of the right place—a precaution often neglected with unfortunate results. The next step is to discover what pronunciation the old written form represents. Most of the letters of the alphabet had anciently different sounds from those they have now, and even at one and the same period different writers did not always use the letters with the same values. If we misinterpret the spelling—if, for instance, we fancy that the letter z was meant to be sounded as in zeal, when it really stood for ts—we shall be led astray in our etymologizing. It is often difficult to find out the pronunciation expressed by the old spelling of a name, but with adequate scholarship it is usually not impossible.

Suppose now that we have arrived at the pronunciation of our place-name as it was, say, a thousand years ago. We shall very likely find that the name has altered so much that any etymological guess based merely on the modern form must of necessity be wrong. We may find also, if we have the needful linguistic knowledge, that the name now interprets itself; that it is either a simple word or a grammatically-formed compound in the English of the year 900, or in some other language (perhaps Old Danish) which was spoken in the district at the time when the document was written; and that it expresses a meaning which either correctly describes the place or tells something credible about its history. In that case we may reasonably assume that our problem is solved.

On the other hand, it may turn out that the earliest accessible form of the name would have been just as unintelligible at the time when our document was written as it is to-day. Even then, however, our inquiry is not necessarily hopeless. For modern scholars have learned, by laborious comparison, to know the laws according to which English sounds have changed from century to century. Hence, when we know how a name was pronounced at a certain epoch, we can often infer with certainty how it must have been pronounced some centuries earlier. The name, reduced to its prehistoric form, may be found to be obviously significant in the language of the earlier date. If not, we can only conclude that the etymology is with our present means insoluble, and wait for further light.

A certain number of the names still found on the map of England occur, mostly in widely different forms, in documents written during the Roman dominion, from fifteen hundred to two thousand years ago. There is a Roman military road-book of the second century, commonly called the Antonine Itinerary, which gives the distances, along the great highways, from one town to another. There is a work on Geography by Claudius Ptolemy of Alexandria, written in Greek about A.D. 150, which records the latitudes and longitudes of British river-mouths, capes, and sometimes of towns. There are also Roman military documents of somewhat later date that contain many of our names; and others are to be found in various Latin writers from Caesar downwards.

Now these earliest records were written before the English—the Angles, Saxons, and Jutes—came to this island. The names found in them must therefore belong to the language of the ancient Britons, unless indeed they were given by the still older inhabitants, in which case it would be hopeless to think of finding out their meaning. About the ancient British tongue we do know something, and many of our oldest names have been satisfactorily explained by its means.

The explanation, however, is not such an easy business as many popular writers imagine. It is a common delusion that the Britons of the second century spoke modern Welsh, and that therefore a Welsh dictionary is the only instrument needed for the translation of the British names that were written down by Romans eighteen hundred years ago. Now it is perfectly true that modern Welsh is, except for the large number of words that it has borrowed from other tongues, the direct descendant of the ancient British language. But modern research has established the fact that the British of the second century was a fully inflected language, which has developed into Welsh much in the same way as Latin has developed into French; namely, by dropping nearly all its final syllables, and changing most of its sounds (so that the identity of the words is no longer recognizable except by those who know what the sound-changes have been), by losing many of its words, and by adopting words from other languages. Not a few of the supposed Welsh words which amateur etymologists are fond of using to explain ancient British names are actually known to be borrowed from English. Moreover, the Welsh dictionaries are very untrustworthy, and contain many words that never existed at all; some of them having been invented for the express purpose of accounting for place-names. Among these spurious coinages are caint and gwent, which the dictionaries tell us mean 'open or champaign country'. These figments are intended to explain the proper names Caint, which is Welsh for Kent, and Gwent, which is the name of the district round Caerwent in Monmouthshire. In innumerable popular books we are informed that the Romans latinized the British name Caint into Cantium. Now it is true that the ending -um is Latin; but the original British form was not Caint, but Cantion. Ptolemy's spelling Κάντιον is by accident quite correct, because Greek and British happened to agree in the form of neuter nouns of the o declension. The British name has become Caint in Welsh by dropping the ending and transferring the i to the first syllable, these being the regular processes of sound-development, which have been gone through by all words of similar form that have survived. In the fifth century, when the south-east corner of Britain was conquered, the name had probably lost its ending and become Canti, and was adopted by the English in that form. According to the well-known phonetic laws of Old English, this name regularly became Cent (C pronounced K); that is to say, the i first modified the a into e, and then dropped off.^[3] The Old English name for 'people of Cantion' was Cantware, whence Cantwaraburh, 'borough of the Cantware,' now Canterbury. The reason why the original a has not here been changed into e is that already in prehistoric times an Old English word ending in i (after a long root-syllable) lost that sound when it became the first element of a compound. If the compound had been formed at a later date it would have been Centware, and the name of the city would now be Kenterbury.

What Cantion really means it is impossible to say with certainty. We have no means of knowing whether it primarily denoted the country, or the South Foreland (Ptolemy's Κάντιον ἄκρον), or whether it is derived from the tribal name of the inhabitants (in Latin Cantii). As the Irish céide, a market, must descend from an Old Celtic cantion, Sir John Rhŷs has conjectured that the district may have been so called because it was the place of resort for merchants from Gaul; but this seems very doubtful.

As I have already stated, the ground on which lexicographers have inferred the existence of a Welsh word gwent, with the sense 'open or champaign country', or 'plain', is that Gwent is, and has been from early times, the name of the region now known in English as Monmouthshire. The alleged meaning of the name is descriptively appropriate only to a part of the district; and there is reason for believing that the district was so called from its capital, now Caerwent (the prefixed word caer meaning 'fortress'). Caerwent is in Roman records called Venta. It is commonly said that this is a latinized form of the British name Gwent. It is not 'latinized' at all. The ancient British name was Wenta; we may retain the spelling Venta, if we remember that the Latin v was pronounced (at least nearly) like our w. I have already said that Welsh has dropped all its original unaccented final vowels; and it has also turned its original initial w into gw, so that the Welsh form of the Ancient British Wenta is necessarily Gwent.^[4] Besides Caerwent, which the Romans called for distinction Venta Silurum (i.e. Venta of the Silures), there were two other places of the same name, Venta Icenorum, the capital of the Iceni in Norfolk, and Venta Belgarum (of the Belgae); the name of this last Wenta was turned by the English into Wintanceaster,^[5] now Winchester. The meaning of Wenta has not yet been discovered.

The Welsh word dwr, water, which makes a great figure in popular books treating of the etymology of place-names, is not exactly spurious, but it is a mere modern colloquial shortening of dwfr, the ancient British form of which was dubron. It is therefore evident that this modern word cannot be used to explain such ancient names as Durovernon (Canterbury), or Durnovaria (whence the Old English Dornwaraceaster, now Dorchester). An ingenious combination of two absurdities appears in the explanation of the name of the river Derwent as 'dwr gwent, the water of the gwent or champaign country', which has found its way into many school-books.

It would be just as reasonable to try to read Virgil by means of a French dictionary and with no grammar, as to try to translate ancient British names by means of a Welsh dictionary. To say this is to condemn nearly everything that has been written on the subject in popular books.

Perhaps it may not be too much to hope that intelligent readers will admit that much of what I have said has at least some prima facie likelihood. No one who knows much about the history of any language will think it probable that Welsh has remained unchanged for some eighteen hundred years. And when I assert that the changes have been as great as those which have transformed Latin into French, and that therefore the Welsh dictionary, in the hands of a person not specially trained to use it, is worthless as a key to the interpretation of British names of the first or second century, it will perhaps be acknowledged that the statement is not intrinsically incredible. But the question may reasonably be asked, what are the means by which philologists claim to be able to interpret some of these ancient names? There is no surviving British literature of the period of Roman rule; how then is it possible to discover what the British language of this early date was like? The question deserves an answer, and I will endeavour to give it, so far as can be done without presupposing more philological knowledge than ordinary educated readers may be expected to possess.

In the first place, the history of the Welsh language for many centuries back can be traced by means of its extensive literary remains. When we compare the forms of Welsh words as they appear in the oldest literature with those in which they are now current, we discover that many of the sounds of the language have undergone certain uniform changes. For instance, if a modern Welsh word contains a b, d, or g between vowels or at the end, we know that those sounds are alterations of an original p, t, or k respectively. Again, if a word in the oldest known Welsh contained an intervocalic or final b, m, d, or g, then we find (if the word has come down into the modern language) that the consonant has not remained unaltered: b and m have become v (written f), d has become dd (pronounced like th in father), and g has disappeared. Several other changes are equally well attested; so that we can say, not as a matter of conjecture but as a matter of historic fact, that most of the words in the modern Welsh dictionaries have undergone noteworthy changes in form. Still, the evidence of Welsh literature does not carry us quite as far back as we require to go. The oldest literary Welsh had already lost the endings indicating case and gender, which ancient Gaulish inscriptions show that the language originally possessed. These, however, have not disappeared without leaving traces. For example, while the dictionary form of the adjective meaning 'white' is gwyn, representing the Old British masculine windos and neuter windon, the Welsh for 'white river' is afon wen. The explanation of this is that the Old British was abonā windā; the final ā of the feminine adjective has had the effect of changing the i into e, and the fact that the feminine noun ended in a vowel has prevented the w of the following adjective from becoming gw, as it did when the word began a sentence or when the preceding word ended in a consonant. Much light on the prehistoric forms of Welsh words is afforded by comparison with the closely related Irish language, in which the nouns and adjectives often retain more distinct traces of the primitive declensions than the corresponding words in Welsh. Further, just as a rare Old French word, not surviving in modern French, may sometimes be interpreted by reference to the equivalent word in Italian or Spanish, it often happens that words contained in Ancient British place-names have been preserved (with the normal change of form) in Irish, although even the earliest known Welsh had already lost them.

The vocabulary, then, which has to be used for the interpretation of the British place-names of the Roman period, is not that of modern or even of early Welsh, but is a vocabulary reconstructed by means of a comparative study of the Celtic languages. It would, of course, be absurd to suppose that the whole of the language of early Britain can be recovered by the inferential methods that I have described. In all probability the ancient language contained many words that have been lost both in Welsh and Irish; and even of those that have survived, many, we may reasonably assume, had senses not precisely identical with those which they have come to express in historical times. Hence it is very likely that many of the British names found in Roman records will never be explained with certainty. Still, there are a considerable number of which the etymology can be clearly ascertained. It is of some importance to point out that the Roman spelling of British names, wherever we are able to control it, turns out to be remarkably accurate. There are no such strange deformations as are found in the English representations of Indian names in the early days of the East India Company. It is not true, as is popularly supposed, that the Romans euphonized the barbarous names by inserting vowels between the syllables; the phonetics of Latin and Old Celtic were so much alike that such a procedure was unnecessary. It is true that the mediaeval transcribers of Roman records have often blundered grievously, as we see, for instance, from the various readings in our copies of the Antonine Itinerary. But wherever the original reading of a Roman-British name is certain, we may rest secure that we have before us an exact reproduction of the native pronunciation.

One of the names which have been furnished with pretended explanations out of the Welsh dictionary is London, anciently Londinion—or, as Tacitus latinized it, Londinium. Most of the modern histories of London tell us that this name means either 'fort or town by the lake' (from the Welsh llyn, lake, and din, fort or town), or 'fort or town of ships', from the Welsh llong, a ship. Now these Welsh words are, unlike caint and gwent, genuine and ancient; but they have changed somewhat in form during eighteen hundred years.

Philologists who have investigated the history of the British language all agree that in the first century the British form of the compound meaning 'lake-fort' would have been Lindudūnon, and that of the compound meaning 'ship-fort' would have been Longodūnon. Now since, as I have just said, the Romans did not 'corrupt' the Celtic names which they wrote down, it is obvious that neither of these forms will account for 'Londinium'. There is another fatal objection to the interpretation 'ship-fort', viz. that the British word longā, a ship, is known to be an adoption of the Latin navis longa, and would therefore be unlikely to occur in the name of a British city as early as the time of Tacitus.

We have to accept the form Londinion as it stands, without any lawless meddling with its consonants or vowels. What is certain about it is that it is not a compound; that is to say, it contains one root and not two. It is derived from a word londos (or londā, londi, &c.; the declensional syllable is uncertain) by the successive addition of two suffixes, -ino- and -io-, which originally served to form adjectives. M. D'Arbois de Jubainville has conjectured that Londinion, properly the neuter of an adjective, means 'the place belonging to a man named Londinos', and that this personal name is derived from londos, a word that survives in Irish as lonn, savage, wild. This guess seems to be the only one hitherto offered that has the merit of being philologically possible. The meaning given to the assumed personal name is not inconceivable, when we remember that in early times such words as 'bear' and 'wolf' were continually used in the formation of names of men and even of women. Still, we must not suppose that this explanation of the name Londinion is certain, or anything near it. There may have been other words of the form lond- besides the one that survives in Irish, and even that word may have greatly changed its meaning. It may be that Londinion does mean 'the place of Londinos', and yet that the word from which the man's name is derived may have had some other meaning than 'savage'. But as no such name as Londinos has yet been found borne by any Briton or Gaul, the theory of personal derivation remains uncertain. All we really know is that the name of the city is derived, by the addition of suffixes, from a word lond-, the meaning of which is still obscure. However, our inability to find a satisfactory solution of the problem is no reason for acquiescing in guesses which we know to be wrong. It is only a reason for waiting patiently for further light.

London is almost the only instance^[6] in which a British name of a town has remained nearly unaltered for nineteen hundred years. As a rule, when the Angles and Saxons adopted the name of an inhabited place, they appended to it a descriptive word in their own language, such as ceaster (borrowed from the Latin castrum), a fortified city, burh, borough, or wīc, a dwelling-place. Thus Wenta became Wintanceaster, now contracted to Winchester. If the British name was a long one, it lost some of its syllables; for instance, Sorwiodūnon became Searoburh, now Salisbury; and Manduessedon, with the addition of ceaster, remains as Mancetter. This process is exemplified in the history of the name of York. The ancient British name was Eburācon, which is probably derived from a man's name Eburos, though it may possibly mean a place where yew-trees grew. It is often said that Eburacon comes from Ebura, a supposed name of the Ouse, one of whose tributaries is still called the Ure. But this is a mistake. The argument on which it is founded is as follows. The name of the French city Évreux is derived from the tribe called Eburovīces; and as the territory of this tribe bordered on the river Eure, antiquaries have inferred that the name Eure must be a contraction of Ebura. But in fact the ancient name of the river was Autura, which by regular sound-change has become Eure. There is thus no reason for thinking that there ever was such a Celtic river-name as Ebura; if it had existed, its form in modern French would, according to phonetic laws, have been Yèvre or Hièvre. Further, it is a quite baseless assumption that the Ouse at York was ever called by the name that is now borne by its tributary stream; and Ure is a mere modern misspelling of Yore. As we do know that Eburos was a Celtic name actually borne by men, and was also a common noun meaning 'yew-tree', the probability is that Eburācon is derived from the one or the other. In Gaul most of the place-names ending in -ācon (latinized -ācum, in modern French -ay or ac- according to the local dialect) are known to be derived from persons; and that affords some reason for preferring the personal derivation in this instance.

By the time of the Angle conquest, the British language had changed a good deal, and Eburācon had come to be pronounced Evurōc. In accordance with their usual practice the Angles dropped the last syllable and added their word wīc, so that the name became Eoforwīc (the f being pronounced as v). This form is not due, as is commonly said, to a popular etymology connecting the name with the Old English word eofor, a wild boar. In Old English the vowel e ordinarily became eo when there was a u in the next syllable, so that Evur- could not possibly have become anything else than Eofor-. In the ninth century the city fell into the hands of the Danes, who turned the name into Iorvīk, afterwards contracted into York.

Another city that retains part of its old name in a corrupted shape is Lichfield. The common notion is that this name means 'field of corpses', from the Old English līc, a dead body. This cannot be right, because in the eighth century the historian Bæda writes the name not as Licfelth but as Lyccidfelth. The true explanation is as follows. About two miles from Lichfield are the ruins of a Roman fortification, the name of which in the second century was Lētocēton,^[7] meaning 'grey wood'. As the British language changed, the name came to be pronounced Luitcoit. In early Welsh writers Cair Luitcoit (cair, now caer, meaning fortress) is mentioned as an ancient city. When the Angles conquered the country, they seem to have left the old fortified town uninhabited (as they often did), and made their own settlement in the 'Luitcoit-field'—that is to say, in the tract of country adjacent to the old Lētocēton. The alteration of Luitcoit into Lyccid is not very great. A name of similar origin, by the way, is Chesterfield; the 'chester' or Roman station is a mile away from the town that has arisen in its 'field '.

It is worth while to mention that in the twelfth century the historian Henry of Huntingdon,^[8] finding the name Cair Luitcoit in a list of ancient cities, chose to identify it with Lincoln. No doubt Luitcoit does sound something like Lincoln as pronounced by a man with a bad cold; but the names have etymologically nothing in common, and there is sufficient evidence to prove that the real Luitcoit was Lichfield. Nevertheless, the historians of Lincoln go on telling us that an ancient name of the city was Cair Luitcoit; and some of them, to make the story look more plausible, have turned Luitcoit into 'Lindcoed'.

The British name of Dorchester was Durnovaria (the v being pronounced as w). Now as durno- means 'fist', and war is the root of the Welsh gware, play, it seems possible that the town got its name because it was adjacent to a place set apart for pugilistic encounters. However this may be, the Saxons at first called the place Dornwara-ceaster (retaining the contemporary British form of the name unchanged), but afterwards shortened this to Dornceaster, whence the modern form. The inhabitants of the region about 'Dornceaster' were called Dorn-sǣite (sǣte being a word for settlers or dwellers), and this has now become, in the form Dorset, the name of the county.

The Roman-British name Regulbium (in Kent) seems to have meant 'promontory', from gulbā, a beak, with a Celtic prefix ru- or ro-, corresponding to the Latin pro-. The name became in Old English times Raculf and Reculf,^[9] and subsequently Raculfesceaster. The modern form, Reculver, shows abnormal contraction.

Another name of the same region, Rutupiæ (whence the 'Rutupine oysters' mentioned by Juvenal), seems also to contain the prefix ru-, but its etymology has not been determined. The name has undergone strange transformations. In Bæda it appears as Reptacæstir; at a later time the affix 'borough' was substituted for 'Chester', and the place is now called Richborough, an intermediate form being Ratesburgh.

The modern Wroxeter, in Old English Wrocenceaster or Wreocenceaster, is the Roman Viroconium or Uriconium. The name has obviously some connexion with that of the Wrekin, the conspicuous hill in the neighbourhood. According to Sir John Rhŷs, Viroconium was so called from a local tribe whose eponymus was an ancestor or chief named Virocū (genitive Viroconas). This personal name is a compound of viros, man, and cū, dog, these words occurring frequently as elements in Celtic personal names. The Wrekin is by the early Welsh bards called Dinlle Urecon. Sir J. Rhŷs has adduced strong reasons for believing that Dinlle means 'the fortress of the god Lugus', and the name of the tribe was appended in order to distinguish this particular fortress from others of the same name. The English settlers in the district preserved only the second part of the Welsh name, so that the hill is now called simply Wrekin. A place in the neighbourhood, now called Wrockwardine, was in the thirteenth century Wrocheumrthin; here we find the Wrocen- of Wroxeter combined with the Old English worðign, an estate or farm.

The name of the god Lugus, mentioned a little before, appears to be the first element in Luguvalium, now Carlisle. The modern name represents, in its second syllable, the later pronunciation of Luguvalium in the Welsh dialect of Cumbria; the first syllable is the word caer, city or fortress, which the Welsh have prefixed to many of the ancient names of cities.

Inasmuch as the Romans in Britain must often have found it necessary to erect towns or military stations in places that were uninhabited and had not native names, we might have expected to find that they often gave Latin names to these new foundations. As a matter of fact, names of Latin etymology are almost entirely absent from the lists of towns and stations given in Roman military documents. What the Romans usually did, when they had to find a name for a station in a place that had no native appellation, was to take the British name of the adjacent river, and use it as a place-name. Thus Dānum, now Doncaster, is from the name of the river Don; Isca, now Exeter, is from the Exe; Dēva, now Chester, is from the Dee; Derventio is the Roman name of two places on rivers still called Derwent; Cunētio was on the river Kennet. In cases of this kind, where the Roman name of an inhabited place is simply identical with the British name of a river, without any native affix such as dunum, fort, or briva, bridge, we may perhaps reasonably infer that the town or station was founded by the Romans on a spot that had previously had no name. Possibly a similar conclusion may be justified with regard to Lindum (Lincoln), for the name is not a compound, but merely the British word lindus (modern Welsh llyn), a lake. If so, the common assumption of antiquaries that a pre-Roman town existed on the site of Lincoln is erroneous.

It is now time to speak of the names that first occur in documents written after the English became masters of Britain. Of these documents the most important for our purpose are the monastic charters from the seventh to the twelfth century, which give us English names in their correct contemporary spelling, and thereby often make their etymology quite transparent. Next in order of importance comes the Domesday Book of William the Conqueror. It is true that the vast number of names contained in this record are often badly spelt; but if we understand the principle of its orthography we can often discover with certainty what the names really were. The other sources from which we can derive information respecting the early forms of place-names are too numerous to specify.

Some of the names that first appear in these records are British, and as they date from a time when the British tongue had come nearer to modern Welsh, they are often easier to interpret than those which are found in the older sources. I have not ventured to say anything of the rivernames mentioned by Roman writers, because their etymology is very obscure. Tamēsis or Tamēsa (Thames), Trisantona (Trent), Sabrina, (Severn), Alauna (Allen and Lune), are examples of names that apparently belong to too early a stage of the language to be interpreted at present with any certainty. I may mention, however, that the name of the Dee, by Roman writers called Dēva, has been with great probability explained by Sir J. Rhŷs as meaning 'goddess', implying that the personified river was an object of worship. One or two of the river-names found in later records are more intelligible. We have several rivers named Avon, which is merely the common Welsh word for 'river'. The streams had no doubt proper names of their own, but the Anglo-Saxons did not get hold of them. The river Ock, which falls into the Thames at Abingdon, is called in Old English charters Eoccene; and this seems to show that the name is derived from the British ehōc (modern Welsh ehawg), a salmon. There are very sufficient reasons why no tributary of the Thames now contains salmon; but it may well have been otherwise fifteen hundred years ago.

Let us now consider the names that were given by the Angles and Saxons. In absolute strictness, it is not correct to speak of these names as having been 'given'. The Anglo-Saxons deliberately gave names to their children, their swords, their houses, and their ships; but they do not seem to have been in the habit of inventing or choosing names for places. How then, it may be asked, did the names come into existence? The question may best be answered by citing a few examples. A thousand years ago or more, a man named Brihthelm lived on the coast of Sussex. When his neighbours spoke of his abode as Brihthelm's tūn—this word, now pronounced 'town', having then the sense of 'farm enclosure'—they were clearly not inventing a name for the place, but merely referring to it in the most obvious way possible. But long after Brihthelm was dead and forgotten, Brihthelmestūn continued to be the name of the farm and of the village that had gathered round it. The village grew into a large town, which till quite lately was called Brighthelmstone, though the name is now contracted to Brighton.

Again, on the Thames some fifty-five miles in a straight line from London there was a ford over which drovers led their cattle, and a few miles higher up there was another ford across which the swine were driven to and from their pasture in the wood. The cluster of houses near the 'oxen-ford' has grown up into the city of Oxford^[10]; and though no village arose at the 'swine-ford', the bridge by which the ford has been superseded is still called Swinford Bridge. Once more, a list of boundaries of an estate in Sussex, written in the ninth century, mentions a 'new building'—in Old English Nȳtimbre. Perhaps it may not have been very new even then; but it had been new once. The new building became old, and disappeared; but still, after a thousand years, a tiny hamlet bears the name of Nightimber. Similarly, most of the places called Newton, Newnham or Nuneham (Nīwan hām, new 'home'), Newbottle or Newbold (bold, botl, a house) have had those names for 800 years, and probably much longer.

Since the Anglo-Saxon names of places originated in this spontaneous way, it is not surprising to find that a large proportion of them are derived from names of persons. As these persons are absolutely unknown to us, and were no doubt mostly mere farmers or cottagers, it cannot be said that place-names of this type yield us very interesting information. Such interest as they have lies in the curious changes they have undergone since they were first written down. Some of them, if we had not their old spelling to guide us, would seem to have very different meanings from those which they actually have. Alderley, for instance, looks as if it had something to do with alders, and Barrowcote as if it contained the word barrow, a burial-mound; but in fact the former means Ealdred's lēah or meadow, and the latter Beornweard's cottage. In some cases, when the modern rustic pronunciation of a name completely disguises its etymology, an old-fashioned spelling has been retained, which shows some traces of the original form. It is not difficult to see that Sawbridgeworth and Woolfardisworthy are derived from the well-known personal names Sǣbeorht and Wulfheard; but the modern local forms 'Sapser' and 'Oozery' would, if we had no other evidence, be hopelessly obscure.

The syllable -ing, which frequently occurs in place-names of Old English date, has given rise to a great deal of mistaken speculation. Some have imagined it to be identical with the modern dialect word ing, a meadow; this word, however, did not exist in Old English,^[11] but represents the Old Norse eng, a meadow. It is found in some modern place-names, as Ingbirch worth, Ings; but the syllable ing that appears so often in Old English names has a different origin. In Old English this syllable was added to personal names to form patronymics, like -ίδης in Greek. Thus, in the Biblical genealogies, 'Seth was Adaming' stands for 'Seth was the son of Adam'. It was also appended (in the plural form -ingas, genitive -inga, locative -ingum) to names of places or districts, to form designations for the inhabitants of the locality, as in Sodomingas and Gomorringas, the people of Sodom and Gomorrah, Centingas, the people of Kent. It is in the patronymic function that this suffix is chiefly found in English local nomenclature. Sometimes it stands as the ending of a place-name: Godalming, for instance, was the abode of the Godhelmingas or children of Godhelm. More frequently we find a patronymic, formed with this suffix, in the genitive plural, followed by some such word as hām, home, tūn, town (meaning a single farm), as in Buccingahām, Buckingham, which probably means the abode of the descendants of Bucca.^[12] There are many place-names of similar formation in which the a of the genitive plural is wanting, as in Eadwulfingtūn, now Adlington in Cheshire. It was the opinion of Kemble that in the latter class of names the syllable -ing has not its function of forming patronymics, but is practically equivalent to a genitive suffix, so that Eadwulfingtūn would mean simply 'the farm of Eadwulf '. There does not, however, appear to be any evidence that the ending -ing was used in this way, and I am inclined to think that in place-names of this kind the a of the genitive plural was dropped on account of the length of the personal name to which it was appended. As a rule, we find -inga where the patronymic (in the singular) is a dissyllable, and -ing when it is polysyllabic.

Besides those patronymics in -ingas which designated the children or descendants of some person who lived perhaps two or three generations back, there were others that were the appellations of royal or noble families. We read in the life of St. Guthlac that he belonged to the famous Mercian family of the Iclingas, whose name evidently marks them as the reputed descendants of Icel, the great-grandfather, according to the pedigree in the Chronicle, of Creoda, the first king of Mercia. Whether Icel was a real person or not we have no means of knowing; it is possible that he is a mythic eponymus invented to account for the name of the Iclingas, and that this name is not a true patronymic at all; it is possible, again, that Icel was a personage of ancient heroic tradition, from whom the kings of Mercia chose to consider themselves descended, and to whom the genealogists arbitrarily assigned a definite place in the ancestral series. There is the same uncertainty about the historical character of the eponymi of the royal houses of many early Germanic nations. The Amali (as they are called in Latin; the Gothic form would be Amalungōs), from whom the Ostrogoths chose their kings, claimed descent from an ancient king Amala; the Scyldingas of Beowulf (in Old Norse Skiǫldungar), the royal house of the Danes, had as their supposed ancestor a hero named Scyld (Old Norse Skiǫldr); but the real existence of Amala is doubtful, and there is very strong reason for regarding Scyld as mythical.

According to the theory of Kemble, which has been widely accepted, the great majority of those collective names in -ingas which enter into the composition of Old English local names, or are used as place-names without composition, represent gentes or clans which were formed on the continent before the English invasion of Britain. On this view, wherever we find traces of a body of people called, say, Wealingas (as at Wallingford in Berkshire and Wallington in Surrey), we are to infer that the places are settlements of one and the same clan. The Iclingas who gave their name to Iclingharn in Suffolk are, it is assumed, a branch of the same gens that is represented by the royal house of Mercia. It is even supposed that there must be community of origin between the English and the continental clans bearing coincident names: that, for instance, the Scylfingas whose name appears in the English Shilvington were the kindred of the Scylfingas who in Beowulf are mentioned as the royal family of the Swedes. The precarious character of this theory will be evident if it is borne in mind that all the families descended from ancestors that happened to have the same name would necessarily have the same patronymic designation. Any descendant of a man named Billa was a Billing; and Billa might be the contracted form of any one of the names beginning with Bil-, such as Bilfrith, Bilhæth, Bilgils, Bilnōth, and many others. Hence, although it is true, as Kemble points out, that there are places in eleven English counties, ranging from Northumberland to the Isle of Wight, that are named from the Billingas, it is not safe to infer that one great clan of this name had settlements all over the country. Nor need we adopt the amazing conclusions that the Azdingi, the royal race of the Vandals, are represented in the English Ardingley, and that the Wælsingas from whom Walsingham is named are identical with the Wælsingas of Germanic legend.

In the modern forms of many names the syllable -ing has been either lost or altered. Thus Edmonton was originally Eadhelmingtūn, and Alfreton was Ælfrēdingtūn. On the other hand, the number of names containing this formative element was so great that other names which did not originally contain it have had it inserted by analogy. Abbandūn, 'Abba's down,' has become Abingdon; Gislandūn, 'Gisla's down,' is now Islington^[13]; and some of the places originally called Nīwantun (new 'town' or farm) are now Newington.

Some of the Old English names of places tell us what kind of trees grew near, as Acton, from ác, oak, Sephton, from sæppe, fir, and the more intelligible Ashton, Elmton, Thornton, Appleton. Others show what animals were bred at the place, as Swinton from swine, Shipley and Shipton from sheep; others again relate to agriculture, as Waddon, originally Hwǣtedūn, 'wheat-down,' Linacre (Līn-æcer), flax-field. The many Nortons, Suttons, Eastons, Westons, and Middletons, indicate the situation of the several farms forming part of an estate.

More interesting than these names are those—though they are very few—that contain traces of ancient beliefs or superstitions. A small hamlet in South Yorkshire has the oddsounding name of Dwaraden, which is clearly the Old English dweorga denu, valley of the dwarfs. 'The voice of the dwarfs' (dvergmȧl) is a Scandinavian name for the echo, and at Dwaraden a remarkable echo may still be heard. Very few genuine references to heathenism are found in English place-names. Fanciful etymologists have found the name of the god Woden in many names where it does not exist; but it really does occur in Wansdyke and Wednesbury; and the name Harrow is the Old English heargas, heathen temples.

Many names that now belong to towns or villages were originally not names of inhabited places, but of some natural or artificial landmark. The names ending in tree are examples in point. A man named Oswald planted a tree, perhaps for some commemorative purpose; or perhaps the tree was planted by somebody else in commemoration of Oswald. When people came to live near this tree, they said that they lived 'at Oswald's tree'; and this is the origin of the name of Oswestry. In the same way, Folkestone was anciently Folcan stān, meaning a memorial stone erected by or after a man named Folca.

One curious group of these landmark names is formed by those ending in head: Hartshead, Sheepshead, Swineshead, Farcet (anciently Fearreshēafod, bull's head), Gateshead (meaning goat's head), Oxnead (Oxanhēafod, ox's head), Thickhead (Ticcenhēafod, kid's head), and Manshead. I suspect that these names point to a custom of setting up the head of an animal, or a representation of it, on a pole, to mark the place for public open-air meetings. Some of them are names of hundreds. Now the hundreds into which our counties are divided are often called, not like the counties themselves, from their chief town, but after some place that has always been quite unimportant. The explanation of this curious fact is that it was customary to name a hundred from the spot, most likely in the middle of an uninhabited moor, where the men of the hundred assembled for deliberation. It was necessary to have some landmark to point out the place of meeting; and this is why so many hundreds have names ending in tree, such as Longtree, Edwinstree, Becontree. A Berkshire hundred anciently bore the name of Nakedthorn; a Gloucestershire hundred is called Brightwell's barrow, showing that the rendezvous for the hundred-meeting was at the mound containing the remains of a man named Brihtwold; and a Bedfordshire hundred is called Manshead, though no inhabited place of that name is known.

Many names of towns or villages, though of Anglo-Saxon formation, are derived from the British names of the rivers or brooks on which the places stand, as Cheltenham from the Chelt, Charwelton from the Cherwell. Sometimes, though both the river-name and the name of the place still survive, the connexion of the two is disguised by change of pronunciation. The river Pedrida is now called Parret, but the village-name Pedridan-tūm has become Petherton. Occasionally the place built on the bank of a river is called by the name of the river without any addition, as Thame in Oxfordshire. The Anglo-Saxons were so careful in recording their boundaries that they found it worth while to ascertain the British names of quite insignificant little brooks. Many of these bear no name at all on our modern maps; but their names are retained by villages on their banks. Although Dovercourt looks like a modern name containing the word 'court', it is found in records of the tenth century, and is a British compound meaning 'little water'. And Winfrith, in Dorset, is in Domesday Book written Wenfrode, which shows that it is identical with the common Welsh name Gwenffrwd, 'white brook.'

But it is not always the case, when a town or village has a name similar to that of the river on which it stands, that the name of the town is derived from that of the river. For our map-makers have had an evil trick of inventing names for small streams which they found nameless; and their usual way of doing this has been to take a syllable out of the name of some place on the bank of the river. Thus Kimbolton, in Huntingdonshire, is derived from the personal name Cynebald; but the river on which the place stands has been provided by the map-makers with the name Kim. Similarly, a river-name Hextild has been evolved from Hextildesham, a mediaeval form of the Old Northumbrian Hagustaldesham, now Hexham, the 'home' of a hagosteald or unmarried warrior. The name of the river Brain is a figment invented to account for Braintree (in Domesday Branchetreu). The river Penk, in Staffordshire, owes its name to a false analysis of Penkridge into 'Penk' and 'ridge'; but Penkridge is an altered pronunciation of Pencrich, the original form of which appears in the name of the neighbouring Roman station Pennocrucium. It is a compound of the words which in Welsh are pen, head, and crûg, mound. The river at Hitchin has been called Hiz. This is an antiquarian revival of the Norman spelling of Hitchin; the letter z, pronounced ts, having been adopted as an approximate rendering of the English sound tch. The Latin name of St. Albans, Verulamium, was familiar to antiquaries from being mentioned by Bæda, and in the sixteenth century was sometimes used in the anglicized form Verulam. From this was inferred the river-name Ver, which still keeps its place on modern maps. Curiously enough, the same process had been gone through hundreds of years before, for in a tract of the eleventh century on the resting-places of the saints of England, Wærlameceaster (i. e. Verulamium) is said to be on the river Wærlame.

A great deal of popular mis-knowledge about the names of English rivers may be traced to the perverse ingenuity of John Leland, the laborious but not judicious antiquary of the time of Henry VIII. It was he that invented for the river Witham (flowing through the district of Lindisse or Lindsey) the name of Lindis, now familiar to many from its occurrence in Jean Ingelow's poem 'The High Tide on the Lincolnshire Coast.' He also hit upon the clever but preposterous notion that the name of the Nene originated in a scribal error for Avene; and accordingly he calls this river constantly the Avon, and explains Northampton as 'North-Avon-dune'. Leland may also be said to be the author of the name 'Isis' for the Thames at Oxford, though in this instance he merely improved upon a fiction of much earlier date. A wellknown thirteenth-century map of England contains the remark that 'as the Jordan is formed by the union of two rivers called Jor and Dan'—a myth of early travellers in Palestine—'so the Tamyse is formed by the two rivers Tame and Yse.' The name of the Thame being genuine, the supposed analogy from the Holy Land naturally suggested 'Yse' as the name of the Thames above the point at which it receives that river. Leland seems to have been the first person who gave to the name the quasi-classical form in which it is now current.

Although, as has already been mentioned, the Angles and Saxons nearly always adopted the names for rivers, and even small brooks, which they found used by the native population, there are one or two of our river-names that are of English and not of British etymology. The names Manifold and Blackwater tell their own story. Less transparent is the name of the Wensum at Norwich, which seems to be identical with Wantsumu, applied by Bæda to the Kentish Stour. This is probably an otherwise unrecorded feminine adjective wændsumu (agreeing with ēa, river), which would mean 'winding', from the verb wændan (later wendan) to turn.

Some Old English names which are compounds of two words still in common use are nevertheless unintelligible in their modern form, because the changes that have taken place in English pronunciation have operated differently in the compound and in the simple words. When a word is part of a compound (especially of a proper name, where the literal meaning is of no practical importance), it is spoken more quickly than when it stands alone, and therefore the vowels are shortened. The Old English ác survives as 'oak', and the Old English tūn as town; but the place that was called Āctūn is now not Oaktown but Acton. Similarly, the many places with names containing the word strǣt, now 'street', which indicates that they stood on a Roman road, are now called Stratford, Stratton or Stretton, and Streatham. So, too, we say Bradford instead of Broadford, and Dunham instead of Down-home (though some places have the name as Downham).

Before passing from the Angle and Saxon names to those which are of Scandinavian origin, it may be useful to call attention to one or two grammatical points. Old English place-names were used more frequently in the dative than in any other case; indeed certain prepositions governing this case were often treated as a part of the name, as when a charter of the eighth century speaks of land 'in loco quern dicunt æt Eastune', or when Bæda refers to a place called 'in Getlingum' (literally 'among the Gythlingas'), now Gilling. Hence the modern form of a name very commonly represents the Old English dative. The ending -bury, from the dative byrig, is more usual than -borough from the nominative burh. (The Scandinavian names containing the same word have it only in the form -borough.) The dative plural of all nouns ended in -um, which has sometimes been corrupted into -ham or -holm, but sometimes remains in other forms, as in Moorsome from Mōrhūsum, 'at the moorhouses.' Hallam, from which is derived Hallamshire, the name of the district including Sheffield, is probably æt Healum, from the dative plural of healh, haugh. When the first element of a compound is an adjective, it usually appears in the dative of what is called the weak declension. In the Southern and Midland dialects the ending of this form was -an, so that we meet with names like æt Hēan lēage, 'at the high lea,' now Hanley (hēan being the dative of hēah, high). In other names that have a first element ending in -an, this syllable is the inflexion of the genitive either of a personal name, as in Badecan wiellon, 'Badeca's wells,' now Bakewell, or of a river-name (the nominative of which ended in -a or -e), as in Exan mūtha, Exmouth. As may be seen from these instances, the n often dropped out in the later forms, but in many names it still remains, especially before a vowel or h, as in Tuican hom, now Twickenham. The northern dialect, which extended to the southern border of Yorkshire, was characterized by having -a (or some other vowel) wherever the other dialects had -an as a grammatical termination. It is noteworthy that this difference of dialect is traceable in the modern forms of local names, those which contain an inflexional n being found quite close to the Yorkshire border, but not within it. There are two villages a few miles apart, one in Yorkshire and the other in Derbyshire, which have names etymologically identical, meaning 'the high lea'; but the name of the Derbyshire village is Handley (formerly Hanley, æt hēan lēage), while the Yorkshire village is called Heeley. Another point of difference between Old English dialects is that in the north the ending -es of the genitive had, at least as early as the tenth century, come to be often extended from the masculine and neuter nouns to those of the feminine gender. Hence, while the word gāt, she-goat, had in the south gāte as its genitive form, the modern name of the place called by Bæda Ad Caprae Caput is Gateshead.

The names left us by the Danes and Norsemen are of great historical importance, because it is from their evidence, aided by that of local dialects, that we can determine what parts of the country were most abundantly settled by those peoples. The ending -by, as in Derby, Whitby, is always the mark of a place once inhabited by Scandinavians; and so is the ending thwaite, which is the Old Norse thveit, a paddock. The word thorp, a village, was Old English as well as Scandinavian; but the Norsemen made more frequent use of it than did the English.

The Danish names of places are formed on much the same principles as those of the native English. Many of these are from names of persons, and some of these have become curiously altered in pronunciation. Geirwind's 'by' or farm is now Garrowby; Geirmund's thorp is now Granthorpe; and Thormod's tūn is now Thrumpton. Then we find villages named from the trees that grew there. There are Danish Applebys as well as Old English Appletons; Askwith is Danish for ash wood, and Ayscough represents the Old Norse eikiskógr, oak wood.

In some of the districts settled by the Danes the counties are divided, not into hundreds, but into wapentakes. This is a Scandinavian word, meaning 'show of weapons', or military muster. In some places, however, the Danes retained the English term hundred. Several hundreds and wapentakes have Scandinavian 'landmark' names indicating the place of rendezvous for meetings, such as Osgoldcross (Asgaut's cross) in Yorkshire, Normancross in Huntingdonshire, and Aslacoe (Aslak's haugr or burial-mound) in Lincolnshire.

The parts of England in which names of Scandinavian origin are most abundant are Cumberland and Westmoreland, North Lancashire, the northern and especially the northeastern part of Yorkshire, and Lincolnshire. East Anglia, although it was long under Danish rule and has a great number of Scandinavian words in its dialect, has a larger proportion of native Anglian names than the districts just mentioned. In South Yorkshire, South Lancashire, Cheshire, Derbyshire, Nottinghamshire, Leicestershire, Rutland, Huntingdonshire, Cambridgeshire, and Northamptonshire, Scandinavian names are numerous, and in some particular neighbourhoods are in the majority. It is only in the north-west that river-names of this origin are found. Greta and Rotha correspond to the Grjotá and Rauðá ('rock river' and 'red river') of Iceland; Brathay appears to be Breiðá, 'broad river.'

It is from the Danes that we have learnt to speak of the three divisions of Yorkshire as the North, West, and East Ridings. The word is properly thrithing (its original shape was thrithjungr), which means a third part, just as farthing means a fourth part. The Normans had a difficulty in pronouncing the th, and said Nort Triding, Est Triding, and West Triding. If these combinations are spoken quickly, it is not easy to sound the second t, and hence it is that the original thrithing has become Riding. The word had a short i, but as it was an official term more frequently read than heard, the modern pronunciation is a rendering of the written form.

The instance just given is not the only one in which the modern forms of names are due to the inability of Normans to utter English sounds correctly. Many people must have wondered why the county of Shropshire has an alternative name Salop. It came about in this way. The original English name was Scrobbesbyrigscīr (the sc being pronounced sh), from the name of the town Scrobbesbyrig, now Shrewsbury. The combination shr, and indeed the simple sound sh, did not exist in French of the twelfth century, and when the Normans tried to say shrob they could get no nearer than Salop. The same cause accounts for the modern form of the name Salisbury, which has had a curious history. The ancient British name was Sorwiodūnon, which perhaps meant the fortress beside the river Sorwios or Sorwia,^[14] now called the Avon. The Saxons, after the fashion I have already described, shortened the long British name, and added their word burh or byrig, a fortress. The name thus became Searobyrig; and as most of the names with byrig began with a genitive ending in s, analogy caused the corruption of Searobyrig into Searesbyrig. Last of all came the Normans, who changed the r into l, just as they did in the case of Salop.

The Normans also had difficulties with the pronunciation of the names ending in chester, and when the places were garrison towns, where the officials probably spoke nothing but French, their mispronunciations have persisted. It is probably owing to this cause that we say Exeter instead of Exchester, and that we write Leicester, Worcester, and Gloucester, and pronounce Lester, Wooster, and Gloster. The old name of Cambridge was Grantebrycg, meaning the bridge over the river Grante.^[15] The Normans could have pronounced this right if they had tried; but they were careless, and the name became Cantebrugge (as it is written by Chaucer). This was afterwards softened into Cambridge, and the map-makers turned the name of the river into Cam in order to make it fit the modern name of the town. The change of the Old English Snotingahām into Nottingham is probably due to the inability of the Normans to pronounce sn. It may be surmised that the people of Nottingham will bear them no ill-will on this account. The Normans also turned the name of Dūnholm into Duresme, which we now write Durham. There were some other English place-names which the Normans mispronounced, but which still remain uncorrupted. We have not adopted their Nicole for Lincoln, and their Londres for London survives only in France.

While the chief result of Norman influence on the local nomenclature of England consisted in the alteration of the existing place-names, the Normans also introduced some new names of their own. Many of them belong to monasteries, and it is interesting to note how often they have meanings referring to the beauty of the site, as in Beauvale, Beaulieu, Belvoir, Beauchief—a notion which seldom or never appears in names of earlier origin. Other French names of monasteries are Roche (Abbatia de Rupe), Jorvaulx ('the vale of the river Yore'), Rievaulx ('the vale of the Rie'), Grosmont ('great mount'). Richmond ('rich mount'), the name of Earl Alan's castle in Yorkshire, and afterwards of the town that grew up around it, gave, many centuries later, the title of an earldom to Henry Tudor, who as King Henry VII changed the name of his palace of Shene, in Surrey, to Richmond. Other noteworthy examples of French place-names in England are Grampound (Old French grant pont, great bridge), and Pomfret (pont freit, broken bridge), which is usually written in the latinized form Pontefract (ad pontem fractum), owing to the influence of official documents written in the learned language.

The few local names of French formation had their origin, of course, in the period, extending from the Conquest to the fourteenth century, in which the French language was generally current among certain classes of English society. The rise of new names of native etymology did not cease at the Conquest; it has indeed continued down to the present time. Of the names that were formed in the Middle English and the earlier modern English period, many are no longer intelligible without special study, either because they have undergone contraction or corruption, or because they contain obsolete words or obsolete forms of words still in use. For the interpretation of these names recourse must be had to documentary and linguistic research, according to the methods previously explained in their application to the nomenclature of the earlier periods. I do not propose here to treat of the English names of post-Conquest origin, but it may be worth while, in conclusion, to refer to the 'double-barrelled' names, as they may be called, that abound in our modern gazetteers. Very many of the ancient names of villages, of Old English or Scandinavian origin, happen to be common to a great number of places in different parts of the country. There are, for instance, something like eighty Suttons. To prevent confusion between these namesakes, there has in many instances been added the surname of a family who once held the manor to which the village belonged, as in Sutton Mandeville, Kingston Bagpuze, Stoke D'Abernon, Ashby-de-la-Zouch. Sometimes the appended name is that of a river, as Brixton Deverill, Appleton Wiske; or, with the preposition expressed, Kingston-on-Hull (now shortened to Hull), Newcastle-on-Tyne, Bradford-on-Avon. In the last instance the distinctive affix was imposed by the Post Office in the reign of Queen Victoria. Where the preposition under is used, the following word is usually that of a forest, as in Shipton-under-Wychwood, Newcastle-under-Lyme; but there are exceptions, such as Ashton-under-Hill, Wootton-under-Edge. In some instances the appended name is that of a district, with or without a preposition, as Sutton-in-Ashfield, Newton-in-Makerfield, Sutton Coldfield. The form of compound names is sometimes taken from documents in Law Latin or Law French, as in Aston-sub-Edge, Chapel-en-le-Frith, Laughton-en-le-Morthen; in Chester-le-Street, Bolton-le-Moors, Newton-le-Willows, the le is not the article, but the Old French preposition lez, beside. There are also names which contain Latin designations of former ownerships, as Aston Episcopi, Buckland Monachorum, Lyme Regis. Church dedications often appear as distinctive affixes, as in Ottery St. Mary, Middleton St. George. Such indications of situation as Kirby Moorside and Chewton Mendip require no explanation. In general, the various kinds of identifying appendages to place-names, though less interesting to philologists, are much more instructive to the student of local history than the original names themselves, and deserve a larger share of attention than they have hitherto received.

Henry Bradley.

1. This paper is an expansion of a lecture delivered at the London Institution in January, 1907.
2. Among the few books that can be mentioned as exceptions is the late Canon Isaac Taylor's Names and their Histories. The appendix on English village names in this work is in fact the only good general survey of this part of the subject that has yet been published. The earlier book by the same author, Words and Places, is admirably written, but its statements respecting English names are almost wholly obsolete. Professor Skeat has published some valuable pamphlets on the place-names of several counties. The two books by Mr. W. H. Duignan, on the place-names of Staffordshire and of Worcestershire, are greatly superior to the ordinary run of such works. The author makes no pretence of first-hand scholarship, but he has consulted scholars. Many of his interpretations are erroneous, but his industrious collection of documentary evidence has great value.
3. The Old English Cent is feminine, corresponding to Bæda's Latin Cantia, which perhaps represents a real British form.
4. It may be remarked that in the Central dialect of Old French a g was prefixed to an initial w in words adopted from the Germanic languages; thus guarde. (modern French garde, English guard) is from the Germanic word represented by the English ward. If this is due to the influence of Gaulish (which must have been nearly identical with Ancient British), it is not unlikely that the British initial w may even in Roman times have had a faint g sound before it. But this sound must at any rate have been much weaker than in modern Welsh, as otherwise it would have been preserved in the Latin and Old English forms of British place-names. In modern Welsh initial gw becomes w when the word is syntactically connected with a preceding word which in Ancient British ended in a vowel.
5. The vowel-change is regular; in the earliest Old English there was no such combination of sounds as ent.
6. Another example is Dover, in the Antonine Itinerary Dubris (ablative), which can hardly be anything else than the plural of dubron, water.
7. This is, in the Roman documents, recorded in two forms which are due to scribal errors. In the Antonine Itinerary it is Etoceto (ablative), and in the Ravenna Geographer Lectocetum.
8. The mistaken identification is found also, a few years later, in Geoffrey of Monmouth.
9. The b of the British name seems to have already changed to v by the time of the Jutish conquest of Kent, and in Old English spelling this sound was represented by f. The sound of g (as in go) did not exist in early Old English (the letter g having a different value), and the nearest approach to it was k (written c). Compare the Old English Crēacas for Graeci, Greeks. The change by which an original g between vowels has disappeared in Welsh had, we may infer, not taken place (or, at least, did not yet affect the initial of the second element of a compound) when Regulbium fell into English hands.
10. The many persons who maintain that Oxford has any other etymology than this, merely show that they have not made the preliminary studies that would entitle them to express an independent opinion on the origin of place-names.
11. A curious mistake has been made by some writers, who have asserted that the word occurs in a list of boundaries in a charter of A.D. 938. This is not the fact; the transcriber of the charter wrote ing on by mistake for inn on, meaning 'into '.
12. This name is known to have been borne by several persons.
13. The Old English g in this name was pronounced y, and in the modern form has disappeared.
14. There is no actual proof that this was the name, but the Ravenna Geographer of the eighth century mentions a British river Sarva, which perhaps may be a later form of Sorwia, and would account for the Saxon form Searobyrig.
15. This is a British name, perhaps identical with that of the Gaulish river Gerontona. Compare the Old English Treante, Trent, from the ancient British Trisantona.