everybody knows, of course, that up and down over the face of england a whole crop of places may be found with such terminations as lancaster, doncaster, manchester, leicester, gloucester, or exeter; and everybody also knows that these words are various corruptions or alterations of the latin castra, or perhaps we ought rather to say of the singular form, castrum. so much we have all been told from our childhood upward; and for the most part we have been quite ready to acquiesce in the statement without any further troublesome inquiry on our own account. but in reality the explanation thus vouchsafed us does not help us much towards explaining the real origin and nature of these ancient names. it is true enough as far as it goes, but it does not go nearly far enough. it reminds one a little of charles kingsley's accomplished pupil-teacher, with his glib derivation of amphibious, 'from two greek words, amphi, the land, and bios, the water.' a detailed history of the root 'chester' in its various british usages may serve to show how far such a rough-and-ready solution as the pupil-teacher's falls short of complete accuracy and comprehensiveness.

in the first place, without troubling ourselves for the time being with the diverse forms of the word as now existing, a difficulty meets us at the very outset as to how it ever got into the english language at all. 'it was left behind by the romans,' says the pupil teacher unhesitatingly. no doubt; but if so, the only language in which it could be left would be welsh; for when the romans quitted britain there were probably as yet no english settlements on any part of the eastern coast. now the welsh form of the word, even as given us in the very ancient latin welsh tract ascribed to nennius, is 'caer' or 'kair;' and there is every reason to believe that the celtic cathir or the latin castrum had been already worn down into this corrupt form at least as early as the days of the first english colonisation of britain. indeed i shall show ground hereafter for believing that that form survives even now in one or two parts of teutonic england. but if this be so, it is quite clear that the earliest english conquerors could not have acquired the use of the word from the vanquished welsh whom they spared as slaves or tributaries. the newcomers could not have learned to speak of a ceaster or chester from welshmen who called it a caer; nor could they have adopted the names of leicester or gloucester from welshmen who knew those towns only as kair legion or kair gloui. it is clear that this easy off-hand theory shirks all the real difficulties of the question, and that we must look a little closer into the matter in order to understand the true history of these interesting philological fossils.

already we have got one clear and distinct principle to begin with, which is too often overlooked by amateur philologists. the latin language, as spoken by romans in britain during their occupation of the island, has left and can have left absolutely no directs marks upon our english tongue, for the simple reason that english (or anglo-saxon as we call it in its earlier stages) did not begin to be spoken in any part of britain for twenty or thirty years after the romans retired. whatever latin words have come down to us in unbroken succession from the roman times—and they are but a few—must have come down from welsh sources. the britons may have learnt them from their italian masters, and may then have imparted them, after the brief period of precarious independence, to their teutonic masters; but of direct intercourse between roman and englishman there was probably little or none.

three ways out of this difficulty might possibly be suggested by any humble imitator of mr. gladstone. first, the early english pirates may have learnt the word castrum (they always used it as a singular) years before they ever came to britain as settlers at all. for during the long decay of the empire, the corsairs of the flat banks and islets of sleswick and friesland made many a light-hearted plundering expedition upon the unlucky coasts of the maritime roman provinces; and it was to repel their dreaded attacks that the count of the saxon shore was appointed to the charge of the long exposed tract from the fenland of the wash to the estuary of the rother in sussex. on one occasion they even sacked london itself, already the chief trading town of the whole island. during some such excursions, the pirates would be certain to pick up a few latin words, especially such as related to new objects, unseen in the rude society of their own native heather-clad wastes; and amongst these we may be sure that the great roman fortresses would rank first and highest in their barbaric eyes. indeed, modern comparative philologists have shown beyond doubt that a few southern forms of speech had already penetrated to the primitive english marshland by the shores of the baltic and the mouth of the elbe before the great exodus of the fifth century; and we know that roman or byzantine coins, and other objects belonging to the mediterranean civilisation, are found abundantly in barrows of the first christian centuries in sleswick—the primitive england of the colonists who conquered britain. but if the word castrum did not get into early english by some such means, then we must fall back either upon our second alternative explanation, that the townspeople of the south-eastern plains in england had become thoroughly latinised in speech during the roman occupation; or upon our third, that they spoke a celtic dialect more akin to gaulish than the modern welsh of wales, which may be descended from the ruder and older tongue of the western aborigines. this last opinion would fit in very well with the views of mr. rhys, the celtic professor at oxford, who thinks that all south-eastern britain was conquered and colonised by the gauls before the roman invasion. if so, it maybe only the western welsh who said caer; the eastern may have said castrum, as the romans did. in either of the latter two cases, we must suppose that the early english learnt the word from the conquered britons of the districts they overran. but i myself have very little doubt that they had borrowed it long before their settlement in our island at all.

however this may be—and i confess i have been a little puritanically minute upon the subject—the english settlers learned to use the word from the first moment they landed in britain. in its earliest english dress it appears as ceaster, pronounced like keaster, for the soft sound of the initial in modern english is due to later norman influences. the new comers—anglo-saxons, if you choose to call them so—applied the word to every roman town or ruin they found in britain. indeed, all the latin words of the first crop in english—those used during the heathen age, before augustine and his monks introduced the roman civilisation—belong to such material relics of the older provincial culture as the sleswick pirates had never before known: way from via, wall from vallum, street from strata, and port from portus. in this first crop of foreign words ceaster also must be reckoned, and it was originally employed in english as a common rather than as a proper name. thus we read in the brief chronicle of the west saxon kings, under the year 577, 'cuthwine and ceawlin fought against the welsh, and offslew three kings, conmail and condidan and farinmail, and took three ceasters, gleawan ceaster and ciren ceaster and bathan ceaster.' we might modernise a little, so as to show the real sense, by saying 'glevum city and corinium city and bath city.' here it is noticeable that in two of the cases—gloucester and cirencester—the descriptive termination has become at last part of the name; but in the third case—that of bath—it has never succeeded in doing so. ages after, in the reign of king alfred, we still find the word used as a common noun; for the chronicle mentions that a body of danish freebooters 'fared to a waste ceaster in wirral; it is hight lega ceaster;' that is to say, legionis castra, now chester. the grand old english epic of beowulf, which is perhaps older than the colonisation of britain, speaks of townsfolk as 'the dwellers in ceasters.'

as a rule, each particular roman town retained its full name, in a more or less clipped form, for official uses; but in the ordinary colloquial language of the neighbourhood they all seem to have been described as 'the ceaster' simply, just as we ourselves habitually speak of 'town,' meaning the particular town near which we live, or, in a more general sense, london. thus, in the north, ceaster usually means york, the roman capital of the province; as when the chronicle tells us that 'john succeeded to the bishopric of ceaster'; that 'wilfrith was hallowed as bishop at ceaster'; or that '?thelberht the archbishop died at ceaster.' in the south it is employed to mean winchester, the capital of the west saxon kings and overlords of all britain; as when the chronicle says that 'king edgar drove out the priests at ceaster from the old minster and the new minster, and set them with monks.' so, as late as the days of charles ii., 'to go to town' meant in shropshire to go to shrewsbury, and in norfolk to go to norwich. in only one instance has this colloquial usage survived down to our own days in a large town, and that is at chester, where the short form has quite ousted the full name of lega ceaster. but in the case of small towns or unimportant roman stations, which would seldom need to be mentioned outside their own immediate neighbourhood, the simple form is quite common, as at caistor in norfolk, castor in hunts, and elsewhere. at times, too, we get an added english termination, as at casterton, chesterton, and chesterholme; or a slight distinguishing mark, as at great chesters, little chester, bridge casterton, and chester-le-street. all these have now quite lost their old distinctive names, though they have acquired new ones to distinguish them from the chester, or from one another. for example, chester-le-street was conderco in roman times, and cunega ceaster in the early english period. both names are derived from the little river cone, which flows through the village.

before we pass on to the consideration of those castra which, like manchester and lancaster, have preserved to the present day their original roman or celtic prefixes in more or less altered shapes, we must glance briefly at a general principle running through the modernised forms now in use. the reader, with his usual acuteness, will have noticed that the word ceaster reappears under many separate disguises in the names of different modern towns. sometimes it is caster, sometimes chester, sometimes cester, and sometimes even it gets worn down to a mere fugitive relic, as ceter or eter. but these different corruptions do not occur irregularly up and down the country, one here and one there; they follow a distinct law and are due to certain definite underlying facts of race or language. each set of names lies in a regular stratum; and the different strata succeed one another like waves over the face of england, from north-east to south-westward. in the extreme north and east, where the english or anglian blood is purest, or is mixed only with danes and northmen to any large extent, such forms as lancaster, doncaster, caistor, and casterton abound. in the mixed midlands and the saxon south, the sound softens into chesterfield, chester, winchester, and dorchester. in the inner midlands and the severn vale, where the proportion of celtic blood becomes much stronger, the termination grows still softer in leicester, bicester, cirencester, gloucester, and worcester, while at the same time a marked tendency towards elision occurs; for these words are really pronounced as if written lester, bister, cisseter, gloster, and wooster. finally, on the very borders of wales, and of that damnonian country which was once known to our fathers as west wales, we get the very abbreviated forms wroxeter, uttoxeter, and exeter, of which the second is colloquially still further shortened into uxeter. sometimes these tracts approach very closely to one another, as on the banks of the nene, where the two halves of the roman durobriv? have become castor on one side of the river, and chesterton on the other; but the line can be marked distinctly on the map, with a slight outward bulge, with as great regularity as the geological strata. it will be most convenient here, therefore, to begin with the casters, which have undergone the least amount of rubbing down, and from them to pass on regularly to the successively weaker forms in chester, cester, ceter, and eter.

nothing, indeed, can be more deceptive than the common fashion, of quoting a roman name from the often blundering lists of the itineraries, and then passing on at once to the modern english form, without any hint of the intermediate stages. to say that glevum is now gloucester is to tell only half the truth; until we know that the two were linked together by the gradual steps of glevum castrum, gleawan ceaster, gleawe cester, gloucester, and gloster, we have not really explained the words at all. by beginning with the least corrupt forms we shall best be able to see the slow nature of the change, and we shall also find at the same time that a good deal of incidental light is shed upon the importance and extent of the english settlement.

doncaster is an excellent example of the simplest form of modernisation. it appears in the antonine itinerary and in the notitia imperii as danum. this, with the ordinary termination affixed, becomes at once dona ceaster or doncaster. the name is of course originally derived in either form from the river don, which flows beside it; and the northumbrian invaders must have learnt the names of both river and station from their brigantian british serfs. it shows the fluctuating nature of the early local nomenclature, however, when we find that b?da ('the venerable bede') describes the place in his latinised vocabulary as campodonum—that is to say, the field of don, or, more idiomatically, donfield, a name exactly analogous to those of chesterfield macclesfield, mansfield, sheffield, and huddersfield in the neighbouring region. the comparison of doncaster and chesterfield is thus most interesting: for here we have two roman stations, each of which must once have had two alternative names; but in the one case the old roman name has ultimately prevailed, and in the other case the modern english one.

the second best example of a caster, perhaps, is lancaster. in all probability this is the station which appears in the notitia imperii as longovico, an oblique case which it might be hazardous to put in the nominative, seeing that it seems rather to mean the town on the lune or loan than the long village. here, as in many other cases, the formative element, vicus, is exchanged for ceaster, and we get something like lon-ceaster or finally lancaster. other remarkable casters are brancaster in norfolk, once branadunum (where the british termination dun has been similarly dropped); ancaster in lincolnshire, whose roman name is not certainly known; and caistor, near norwich, once venta icenorum, a case which may best be considered under the head of winchester. on the other hand, tadcaster gives us an instance where the roman prefix has apparently been entirely altered, for it appears in the antonine itinerary (according to the best identification) as calcaria, so that we might reasonably expect it to be modernised as calcaster. even here, however, we might well suspect an earlier alternative title, of which we shall get plenty when we come to examine the chesters; and in fact, in b?da, it still bears its old name in a slightly disguised form as kaelca ceaster.

first among the softer forms, let us examine the interesting group to which chester itself belongs. its roman name was, beyond doubt, diva, the station on the dee—as doncaster is the station on the don, and lancaster the station on the lune. its proper modern form ought, therefore, to be deechester. but it would seem that in certain places the neighbouring rustics knew the great roman town of their district, not by its official title, but as the legion's camp—castra legionis. at least three such cases undoubtedly occur—one at deva or chester; one at rat? or leicester; and one at isca silurum or caerleon-upon-usk. in each case the modernisation has taken a very different form. diva was captured by the heathen english king, ?thelfrith of northumbria, in a battle rendered famous by b?da, who calls the place 'the city of legions.' the latin compilation by some welsh writer, ascribed to nennius, calls it cair legion, which is also its name in the irish annals. in the english chronicle it appears as lege ceaster, l?ge ceaster, and leg ceaster; but after the norman conquest it becomes ceaster alone. on midland lips the sound soon grew into the familiar chester. about the second case, that of leicester, there is a slight difficulty, for it assumes in the chronicle the form of l?gra ceaster, with an apparently intrusive letter; and the later welsh writers seized upon the form to fit in with their own ancient legend of king lear. nennius calls it cair lerion; and that unblushing romancer, geoffrey of monmouth, makes it at once into cair leir, the city of leir. more probably the name is a mixture of legionis and rat?, leg-rat ceaster, the camp of the legion at rat?. this, again, grew into legra ceaster, leg ceaster, and lei ceaster, while the word, though written leicester, is now shortened by south midland voices to lester. the third legionis castra remained always welsh, and so hardened on cymric lips into kair leon or caerleon. nennius applies the very similar name of cair legeion to exeter, still in his time a damnonian or west welsh fortress.

equally interesting have been the fortunes of the three towns of which winchester is the type. in the old welsh tongue, gwent means a champaign country, or level alluvial plain. the romans borrowed the word as venta, and applied it to the three local centres of venta icenorum in norfolk, venta belgarum in hampshire, and venta silurum in monmouth. when the first west saxon pirates, under their real or mythical leader, cerdic, swarmed up southampton water and occupied the gwent of the belg?, they called their new conquest wintan ceaster, though the still closer form w?ntan once occurs. thence to winte ceaster and winchester is no far cry. gwent of the iceni had a different history. no doubt it also was known at first as wintan ceaster; but, as at winchester, the shorter form ceaster would naturally be employed in local colloquial usage; and when the chief centre of east anglian population was removed a few miles north to norwich, the north wick—then a port on the navigable estuary of the yare—the older station sank into insignificance, and was only locally remembered as caistor. lastly, gwent of the silurians has left its name alone to caer-went in monmouthshire, where hardly any relics now remain of the roman occupation.

manchester belongs to exactly the same class as winchester. its roman name was mancunium, which would easily glide into mancunceaster. in the english chronicle it is only once mentioned, and then as mameceaster—a form explained by the alternative mamucium in the itinerary, which would naturally become mamue ceaster. colchester of course represents colonia, corrupted first into coln ceaster, and so through col ceaster into its present form. porchester in hants is portus magnus; dorchester is durnovaria, and then dorn ceaster. grantchester, godmanchester, chesterfield, woodchester, and many others help us to trace the line across the map of england, to the most western limit of all at ilchester, anciently ischalis, though the intermediate form of givel ceaster is certainly an odd one.

besides these chesters of the regular order, there are several curious outlying instances in durham and northumberland, and along the roman wall, islanded, as it were, beyond the intermediate belt of casters. such are lanchester in durham, which maybe compared with the more familiar lancaster; great chesters in northumberland, ebchester on the northern watling street, and a dozen more. how to account for these is rather a puzzle. perhaps the casters may be mainly due to danish influence (which is the common explanation), and it is known that the danes spread but sparingly to the north of the tees. however, this rough solution of the problem proves too much: for how then can we have a still softer form in danish leicester itself? probably we shall be nearer the truth if we say that these are late names; for northumberland was a desert long after the great harrying by william the conqueror; and by the time it was repeopled, chester had become the recognised english form, so that it would naturally be employed by the new occupants of the districts about the wall.

no name in britain, however, is more interesting than that of rochester, which admirably shows us how so many other roman names have acquired a delusively english form, or have been mistaken for memorials of the english conquest. the roman town was known as durobriv?, which does not in the least resemble rochester; and what is more, b?da distinctly tells us that justus, the first bishop of the west kentish see, was consecrated 'in the city of dorubrevi, which the english call hrof?s ceaster, from one of its former masters, by name hrof.' if this were all we knew about it, we should be told that b?da clearly described the town as being called hrof's chester, from an english conqueror hrof, and that to contradict this clear statement of an early writer was presumptuous or absurd. fortunately, however, we have the clearest possible proof that hrof never existed, and that he was a pure creation of b?da's own simple etymological guesswork. king alfred clearly knew better, for he omitted this wild derivation from his english translation. the valuable fragment of a map of roman britain preserved for us in the medi?val transcript known as the peutinger tables, sets down rochester as rotibis. hence it is pretty certain that it must have had two alternative names, of which the other was durobriv?. rotibis would easily pass (on the regular analogies) into rotifi ceaster, and that again into hrofi ceaster and rochester; just as rhutupi? or ritup? passed into rituf burh, and so finally into richborough. moreover, in a charter of king ethelberht of kent, older a good deal than b?da's time, we find the town described under the mixed form of hrofi-brevi. after such a certain instance of philological blundering as this, i for one am not inclined to place great faith in such statements as that made by the english chronicle about chichester, which it attributes to the mythical south saxon king cissa. whatever cissanceaster may mean, it seems to me much more likely that it represents another case of double naming; for though the roman town was commonly known as regnum, that is clearly a mere administrative form, derived from the tribal name of the regni. considering that the same veracious chronicle derives portsmouth, the roman portus, from an imaginary teutonic invader, port, and commits itself to other wild statements of the same sort, i don't think we need greatly hesitate about rejecting its authority in these earlier and conjectural portions.

silchester is another much disputed name. as a rule, the site has been identified with that of calleva atrebatum; but the proofs are scanty, and the identification must be regarded as a doubtful one. i have already ventured to suggest that the word may contain the root silva, as the town is situated close upon the ancient borders of pamber forest. the absence of early forms, however, makes this somewhat of a random shot. indeed, it is difficult to arrive at any definite conclusions in these cases, except by patiently following up the name from first to last, through all its variations, corruptions, and mis-spellings.

the cesters are even more degraded (philologically speaking) than the chesters, but are not less interesting and illustrative in their way. their farthest northeasterly extension, i believe, is to be found at leicester and towcester. the former we have already considered: the latter appears in the chronicle as tofe ceaster, and derives its name from the little river towe, on which it is situated. anciently, no doubt, the river was called tofe or tofi, like the tavy in devonshire; for all these river-words recur over and over again, both in england and on the continent. in this case, there seems no immediate connection with the roman name, if the site be rightly identified with that of lactodorum; but at any rate the river name is celtic, so that towcester cannot be claimed as a teutonic settlement.

cirencester, the meeting-place of all the great roman roads, is the latin corinium, sometimes given as durocornovium, which well illustrates the fluctuating state of roman nomenclature in britain. as this great strategical centre—the key of the west—had formerly been the capital of the dobuni, whose name it sometimes bears, it might easily have come down to us as durchester, or dobchester, instead of under its existing guise. the city was captured by the west saxons in 577, and is then called ciren ceaster in the brief record of the conquerors. a few years later, the chronicle gives it as cirn ceaster; and since the river is called chirn, this is the form it might fairly have been expected to retain, as in the case of cerney close by. but the city was too far west not to have its name largely rubbed down in use; so it softened both its initials into cirencester, while cissan ceaster only got (through cisse ceaster) as far as chichester. at that point the spelling of the western town has stopped short, but the tongues of the natives have run on till nothing now remains but cisseter. if we had only that written form on the one hand, and durocornovium on the other, even the boldest etymologist would hardly venture to suggest that they had any connection with one another. of course the common prefix duro, is only the welsh dwr, water, and its occurrence in a name merely implies a ford or river. the alternative forms may be anglicised as churn, and churnwater, just like grasmere, and grasmere lake.

i wish i could avoid saying anything about worcester, for it is an obscure and difficult subject; but i fear the attempt to shirk it would be useless in the long run. i know from sad experience that if i omit it every inhabitant of worcestershire who reads this article will hunt me out somehow, and run me to earth at last, with a letter demanding a full and explicit explanation of this silent insult to his native county. so i must try to put the best possible face upon a troublesome matter. the earliest existing form of the name, after the english conquest, seems to be that given in a latin charter of the eighth century as weogorna civitas. (here it is difficult to disentangle the english from its latin dress.) a little later it appears in a vernacular shape (also in a charter) as wigran ceaster. in the later part of the english chronicle it becomes wigera ceaster, and wigra ceaster; but by the twelfth century it has grown into wigor ceaster, from which the change to wire ceaster and worcester (fully pronounced) is not violent. this is all plain sailing enough. but what is the meaning of wigorna ceaster or wigran ceaster? and what roman or english name does it represent? the old english settlers of the neighbourhood formed a little independent principality of hwiccas (afterwards subdued by the mercians), and some have accordingly suggested that the original word may have been hwiccwara ceaster, the chester of the hwicca men, which would be analogous to cant-wara burh (canterbury), the bury of the kent men, or to wiht-gara burh (carisbrooke), the bury of the wight men. others, again, connect it with the braunogenium of the ravenna geographer, and the cair guoranegon or guiragon of nennius, which latter is probably itself a corrupted version of the english name. altogether, it must be allowed that worcester presents a genuine difficulty, and that the facts about its early forms are themselves decidedly confused, if not contradictory. the only other notable ceasters, are alcester, once alneceaster, in worcestershire, the roman alauna; gloucester or glevum, already sufficiently explained; and mancester in staffordshire, supposed to occupy the site of manduessedum.

among the most corrupted forms of all, exeter may rank first. its latin equivalent was isca damnoniorum, usk of the devonians; isca being the latinised form of that prevalent celtic river name which crops up again in the usk, esk, exe, and axe, besides forming the first element of uxbridge and oxford; while the tribal qualification was added to distinguish it from its namesake, isca silurum, usk of the silurians, now caerleon-upon-usk. in the west country, to this day, ask always becomes ax, or rather remains so, for that provincial form was the king's english at the court of alfred; and so isca became on devonian lips exan ceaster, after the west saxon conquest. thence it passed rapidly through the stages of exe ceaster and exe cester till it finally settled down into exeter. at the same time, the river itself became the exe; and the exan-mutha of the chronicle dropped into exmouth. we must never forget, however, that exeter, was a welsh town up to the reign of athelstan, and that cornish welsh was still spoken in parts of devonshire till the days of queen elizabeth.

wroxeter is another immensely interesting fossil word. it lies just at the foot of the wrekin, and the hill which takes that name in english must have been pronounced by the old celtic inhabitants much like uricon: for of course the awkward initial letter has only become silent in these later lazy centuries. the romans turned it into uriconium; but after their departure, it was captured and burnt to the ground by a party of raiding west saxons, and its fall is graphically described in the wild old welsh elegy of llywarch the aged. the ruins are still charred and blackened by the west saxon fires. the english colonists of the neighbourhood called themselves the wroken-s?tas, or settlers by the wrekin—a word analogous to that of wils?tas, or settlers by the wyly; dors?tas, or settlers among the durotriges; and sumors?tas, or settlers among the sumor-folk,—which survive in the modern counties of wilts, dorset, and somerset. similar forms elsewhere are the pecs?tas of the derbyshire peak, the elmeds?tas in the forest of elmet, and the cilterns?tas in the chiltern hills. no doubt the wroken-s?tas called the ruined roman fort by the analogous name of wroken ceaster; and this would slowly become wrok ceaster, wrok-cester, and wroxeter, by the ordinary abbreviating tendency of the welsh borderlands. wrexham doubtless preserves the same original root.

having thus carried the castra to the very confines of wales, it would be unkind to a generous and amiable people not to carry them across the border and on to the western sea. the welsh corruption, whether of the latin word or of a native equivalent cathir, assumes the guise of caer. thus the old roman station of segontium, near the menai straits, is now called caer seiont; but the neighbouring modern town which has gathered around edward's new castle on the actual shore, the later metropolis of the land of arfon, became known to welshmen as caer-yn-arfon, now corrupted into caernarvon or even into carnarvon. gray's familiar line about the murdered bards—'on arvon's dreary shore they lie'—keeps up in some dim fashion the memory of the true etymology. caermarthen is in like manner the roman muridunum or moridunum—the fort by the sea—though a duplicate moridunum in south devon has been simply translated into english as seaton. innumerable other caers, mostly representing roman sites, may be found scattered up and down over the face of wales, such as caersws, caerleon, caergwrle, caerhun, and caerwys, all of which still contain traces of roman occupation. on the other hand, cardigan, which looks delusively like a shortened caer, has really nothing to do with this group of ancient names, being a mere corruption of ceredigion.

but outside wales itself, in the more celtic parts of england proper, a good many relics of the old welsh caers still bespeak the incompleteness of the early teutonic conquest. if we might trust the mendacious nennius, indeed, all our casters and chesters were once good cymric caers; for he gives a doubtful list of the chief towns in britain, where gloucester appears as cair gloui, colchester as cair colun, and york as cair ebrauc. these, if true, would be invaluable forms; but unfortunately there is every reason to believe that nennius invented them himself, by a simple transposition of the english names. henry of huntingdon is nearly as bad, if not worse; for when he calls dorchester 'kair dauri,' and chichester 'kair kei,' he was almost certainly evolving what he supposed to be appropriate old british names from the depths of his own consciousness. his guesswork was on a par with that of the schoolboys who introduce 'stirlingia' or 'liverpolia' into their ovidian elegiacs. that abandoned story-teller, geoffrey of monmouth, goes a step further, and concocts a caer lud for london and a caer osc for exeter, whenever the fancy seizes him. the only examples amongst these pretended old welsh forms which seem to me to have any real historical value are an unknown kair eden, mentioned by gildas, and a cair wise, mentioned by simeon of durham, undoubtedly the true native name of exeter.

still we have a few indubitable caers in england itself surviving to our own day. most of them are not far from the welsh border, as in the case of the two caer caradocs, in shropshire, crowned by ancient british fortifications. others, however, lie further within the true english pale, though always in districts which long preserved the welsh speech, at least among the lower classes of the population. the earthwork overhanging bath bears to this day its ancient british title of caer badon. an old history written in the monastery of malmesbury describes that town as caer bladon, and speaks of a caer dur in the immediate neighbourhood. there still remains a caer riden on the line of the roman wall in the lothians. near aspatria, in cumberland, stands a mouldering roman camp known even now as caer moto. in carvoran, northumberland, the first syllable has undergone a slight contraction, but may still be readily recognised. the carr-dyke in norfolk seems to me to be referable to a similar origin.

most curious of all the english caers, however, is carlisle. the antonine itinerary gives the town as luguvallium. b?da, in his barbarised latin fashion calls it lugubalia. 'the saxons,' says murray's guide, with charming na?veté, 'abbreviated the name into luel, and afterwards called it caer luel.' this astounding hotchpotch forms an admirable example of the way in which local etymology is still generally treated in highly respectable publications. so far as we know, there never was at any time a single saxon in cumberland; and why the saxons, or any other tribe of englishmen, should have called a town by a purely welsh name, it would be difficult to decide. if they had given it any name at all, that name would probably have been lul ceaster, which might have been modernised into lulcaster or lulchester. the real facts are these. cumberland, as its name imports, was long a land of the cymry—a northern welsh principality, dependent upon the great kingdom of strathclyde, which held out for ages against the northumbrian english invaders among the braes and fells of ayrshire and the lake district. these cumbrian welshmen called their chief town caer luel, or something of the sort; and there is some reason for believing that it was the capital of the historical arthur, if any arthur ever existed, though later ages transferred the legend of the british hero to caerleon-upon-usk, after men had begun to forget that the region between the clyde and the mersey had once been true welsh soil. the english overran cumberland very slowly; and when they did finally conquer it, they probably left the original inhabitants in possession of the country, and only imposed their own overlordship upon the conquered race. the story is too long a one to repeat in full here: it must suffice to say that, though the northumbrian kings had made the 'strathclyde welsh' their tributaries, the district was never thoroughly subdued till the days of edmund the west saxon, who harried the land, and handed it over to the king of scots. thus it happens that carlisle, alone among large english towns, still keeps unchanged its cymric name, instead of having sunk into an anglicised chester. the present spelling is a mere etymological blunder, exactly similar to that which has turned the old english word igland into island, through the false analogy of isle, which of course comes from the old french isle, derived through some form akin to the italian isola, from the original latin insula. kair leil is the spelling in geoffrey; cardeol (by a clerical error for carleol, i suspect) that in the english chronicle, which only once mentions the town; and carleol that of the ordinary medi?val historians. the surnames carlyle and carlile still preserve the better orthography.

to complete the subject, it will be well to say a few words about those towns which were once ceasters, but which have never become casters or chesters. numerous as are the places now so called, a number more may be reckoned in the illimitable chapter of the might-have-beens; and it is interesting to speculate on the forms which they would have taken, 'si qua fata aspera rupissent.' among these still-born chesters, newcastle-upon-tyne may fairly rank first. it stands on the roman site, called, from its bridge across the tyne, pons aelii, and known later on, from its position on the great wall, as ad murum. under the early english, after their conversion to christianity, the monks became the accepted inheritors of roman ruins; and the small monastery which was established here procured it the english name of muneca-ceaster, or, as we should now say, monk-chester, though no doubt the local modernisation would have taken the form of muncaster. william of normandy utterly destroyed the town during his great harrying of northumberland; and when his son, robert curthose, built a fortress on the site, the place came to be called newcastle—a word whose very form shows its comparatively modern origin. castra and ceasters were now out of date, and castles had taken their place. still, we stick even here to the old root: for of course castle is only the diminutive castellum—a scion of the same roman stock, which, like so many other members of aristocratic families, 'came over with william the conqueror.' the word castel is never used, i believe, in any english document before the conquest; but in the very year of william's invasion, the chronicle tells us, 'willelm earl came from normandy into pevensey, and wrought a castel at hastings port.' so, while in france itself the word has declined through chastel into chateau, we in england have kept it in comparative purity as castle.

york is another town which had a narrow escape of becoming yorchester. its roman name was eburacum, which the english queerly rendered as eoforwic, by a very interesting piece of folks-etymology. eofor is old english for a boar, and wic for a town; so our rude ancestors metamorphosed the latinised celtic name into this familiar and significant form, much as our own sailors turn the bellerophon into the billy ruffun, and the anse des cousins into the nancy cozens. in the same way, i have known an illiterate englishman speak of aix-la-chapelle as hexley chapel. to the name, thus distorted, our forefathers of course added the generic word for a roman town, and so made the cumbrous title of eoforwic-ceaster, which is the almost universal form in the earlier parts of the english chronicle. this was too much of a mouthful even for the hardy anglo-saxon, so we soon find a disposition to shorten it into ceaster on the one hand, or eoforwic on the other. should the final name be chester or york?—that was the question. usage declared in favour of the more distinctive title. the town became eoforwic alone, and thence gradually declined through evorwic, euorwic, eurewic, and yorick into the modern york. it is curious to note that some of these intermediate forms very closely approach the original eburac, which must have been the root of the roman name. was the change partly due to the preservation of the older sound on the lips of celtic serfs? it is not impossible, for marks of british blood are strong in yorkshire; and nennius confirms the idea by calling the town kair ebrauc.

among the other ceasters which have never developed into full-blown chesters, i may mention bath, given as akemannes ceaster and bathan ceaster in our old documents, so that it might have become achemanchester or bathceter in the course of ordinary changes. canterbury, again, the roman durovernum, dropped through dorobernia into dorwit ceaster, which would no doubt have turned into a third dorchester, to puzzle our heads by its likeness to dorne ceaster in dorsetshire, and to dorce ceaster near oxford; while chesterton in huntingdonshire, which was once dorme ceaster, narrowly escaped burdening a distracted world with a fourth. happily, the colloquial form cantwara burh, or kentmen's bury, gained the day, and so every trace of durovernum is now quite lost in canterbury. north shields was once scythles-ceaster, but here the chester has simply dropped out. verulam, or st. albans, is another curious case. its romano-british name was verulamium, and b?da calls it verlama ceaster. but the early english in sleswick believed in a race of mythical giants, the w?tlingas or watlings, from whom they called the milky way 'watling street.' when the rude pirates from those trackless marshes came over to britain and first beheld the great roman paved causeway which ran across the face of the country from london to caernarvon, they seemed to have imagined that such a mighty work could not have been the handicraft of men; and just as the arabs ascribe the rock-hewn houses of petra to the architectural fancy of the devil, so our old english ancestors ascribed the roman road to the titanic watlings. even in our own day, it is known along its whole course as watling street. verulam stands right in its track, and long contained some of the greatest roman remains in england; so the town, too, came to be considered as another example of the work of the watlings. b?da, in his latinised northumbrian, calls it v?tlinga ceaster, as an alternative title with verlama ceaster; so that it might nowadays have been familiar to us all either as watlingchester or verlamchester. this is one of the numerous cases where a roman and english name lived on during the dark period side by side. in some of mr. kemble's charters it appears as walinga ceaster. but when offa of mercia founded his great abbey on the very spot where the welsh martyr alban had suffered during the persecution of diocletian, roman and english names were alike forgotten, and the place was remembered only after the british christian as st. albans.

there are other instances where the very memory of a roman city seems now to have failed altogether. for example, b?da mentions a certain town called tiowulfinga ceaster—that is to say, the chester of the tiowulfings, or sons of tiowulf. here an english clan would seem to have taken up its abode in a ruined roman station, and to have called the place by the clan-name—a rare or almost unparalleled case. but its precise site is now unknown. however, b?da's description clearly points to some town in nottinghamshire, situated on the trent; for st. paulinus of york baptized large numbers of converts in that river at tiowulfinga ceaster; and the site may therefore be confidently identified with southwell, where st. mary's minster has always traditionally claimed paulinus as its founder. b?da also mentions a place called tunna ceaster, so named from an abbot tunna, who exists merely for the sake of a legend, and is clearly as unhistorical as his piratical compeer hrof—a wild guess of the eponymic sort with which we are all so familiar in greek literature. simeon of durham speaks of an equally unknown delvercester. syddena ceaster or sidna cester—the earliest see of the lincolnshire diocese—has likewise dropped out of human memory; though mr. pearson suggests that it may be identical with ancaster—a notion which appears to me extremely unlikely. wude cester is no doubt outchester, and other doubtful instances might easily be recognised by local antiquaries, though they may readily escape the general arch?ologist. in one case at least—that of othon? in essex—town, site, and name have all disappeared together. b?da calls it ythan ceaster, and in his time it was the seat of a monastery founded by st. cedd; but the whole place has long since been swept away by an inundation of the blackwater. anderida, which is called andredes-ceaster in the chronicle, becomes pefenesea, or pevensey, before the date of the norman conquest.

it must not be supposed that the list given here is by any means exhaustive of all the casters and chesters, past and present, throughout the whole length and breadth of britain. on the contrary, many more might easily be added, such as ribbel ceaster, now ribchester; berne ceaster, now bicester; and bl?dbyrig ceaster, now simply bladbury. in northumberland alone there are a large number of instances which i might have quoted, such as rutchester, halton chesters, and little chesters on the roman wall, together with hetchester, holy chesters, and rochester elsewhere—the county containing no less than four places of the last name. indeed, one can track the roman roads across england by the chesters which accompany their route. but enough instances have probably been adduced to exemplify fully the general principles at issue. i think it will be clear that the english conquerors did not usually change the names of roman or welsh towns, but simply mispronounced them about as much as we habitually mispronounce llangollen or llandudno. sometimes they called the place by its romanised title alone, with the addition of ceaster; sometimes they employed the servile british form; sometimes they even invented an english alternative; but in no case can it be shown that they at once disused the original name, and introduced a totally new one of their own manufacture. in this, as in all other matters, the continuity between romano-british and english times is far greater than it is generally represented to be. the english invasion was a cruel and a desolating one, no doubt; but it could not and it did not sweep away wholly the old order of things, or blot out all the past annals of britain, so as to prepare a tabula rasa on which mr. green might begin his history of the english people with the landing of hengest and horsa in the isle of thanet. the english people of to-day is far more deeply rooted in the soil than that: our ancestors have lived here, not for a thousand years alone, but for ten thousand or a hundred thousand, in certain lines at least. and the very names of our towns, our rivers, and our hills, go back in many cases, not merely to the roman corruptions, but to the aboriginal celtic, and the still more aboriginal euskarian tongue.