of the Thames above its confluence with the Thame, is one of these inventions. This part of the river is mentioned eighteen times in Berkshire charters of Saxon date, and also in various Oxfordshire charters; and it is always called Temese or Tamese. The origin of 'Isis is quite simple. Behold the unadorned beauty of the method! (1) Tamisis (or Tamesis) is the name of the whole river; (2) Thame is the name of part of the river. Subtracting (2) from (1), then (3) Isis is the name of the other part. Again, the Anton at Andover in Hampshire owes its name to East Anton, which suggested to an antiquary who knew his Tacitus the identification of the stream, the real name of which is Ann or Anna, with the river Antona, mentioned by the Roman author in his account of the preparations made to meet the revolt of the Iceni. The numerous Cæsar's Camps of the Ordnance map seem to bear striking testimony to the industry of that great general; but, unfortunately, many of them are not Roman, and the connexion with Cæsar's name of those that are Roman cannot be even remotely established. There are other well-known instances of such antiquarian inventions.

It was especially with the remains and nomenclature of Roman Britain that the furious activity of the 18thcentury antiquary concerned itself. But the Anglo-Saxon and Danish invasions afforded an equally attractive field for the play of fancy. At a later date Celtic antiquities became fashionable, with results the strangeness of which can only be appreciated by those who have read Dr Guest's Origines Celticæ.' Place-names were called into evidence in these fields of enquiry. They were explained to suit some interesting archæological hypothesis; but, worse than that, there is reason to believe that they were sometimes distorted, and, worst of all, that some of the distortions came to stay upon the map. It is not merely the antiquarian writer who has to be reckoned with; there is the antiquarian talker, a far more dangerous person, because far more difficult to detect. His methods were simple; and, human nature being what it is, only too effective. Mr So-and-so, reputed in his own neighbourhood to be a man of profound learning, asserted that the commonly used name of such and such a place was wrong, and ought to be something else. Among the

semi-educated of his neighbourhood the notion was accepted, though the blissful ignorance of the ignorant might perpetuate the old traditional name. Yet it is undoubtedly the case that some of these learned fictions got into some of those most miscellaneous maps which preceded the era of the Ordnance Survey, and from them in some cases made their way into the maps of the Survey itself. Others got into the modern map by way of personal enquiry.

Fortunately, when the Ordnance surveyor collected place-names from local enquiry, it was long odds that he applied for information to one of the ignorant many, who had inherited a tradition uncontaminated by the emendations and corruptions of reputed learning. There are many who criticise severely the information given in our Ordnance maps; but there can be very few indeed who have a thorough working knowledge of the largescale series, for the very good reason that they are too expensive to buy in large quantities. The actual facts with respect to them are, firstly, that the information with regard to antiquities, a subject upon which the compilers consulted the professedly expert, requires considerable and wide revision; secondly, that the forms of place-names, which were not as a rule obtained from experts, are found to be singularly free from error, when tested by the evidence of Tithe Awards and Enclosure Acts. The only error which is at all common is one which is just as common in the Tithe Awards, namely, the tendency to put names into the possessive case in instances in which they are supposed to be, but are not, personal names. This happens especially with names which have no longer a recognisable meaning. It is possible that this tendency goes far back into the past.

Any one who would appreciate the advance which has been made in little more than the last decade in the interpretation of place-names should compare the books which have been published within that time with

The six-inch sheets for Oxfordshire alone cost about 187.; for Berkshire about 197.; for Hampshire about 247.

The present writer speaks from his experience of the correction of between five and six hundred sheets of the six-inch map.

such a work as Canon Isaac Taylor's 'Names and their History,' published twenty-three years ago. That book excited the most lively interest at the time at which it appeared, and it deserved the welcome which was accorded to it, for it was a work of great erudition; and, whatever its defects tested by the knowledge of twenty years later, it was a great advance on previous work of the kind. It was of far wider application than the books which have succeeded and to a certain extent superseded it, because it dealt with place-names all the world over, and not merely with those of England. But it comes within the area of the present review, because more than half of it is devoted to English nomenclature. In that part of it, be it said with regret, it is most misleading. It would take a long time to point out all the mistakes; and all that can be done here is to justify the criticism. Osney, near Oxford, has no connexion with the river-name Ouse; it is Osan-ig, 'Osa's Island.' Penhurst has nothing to do with the Celtic 'pen,' 'mountain,' or 'hill,' but with a commonplace cattle-pen. The ending -don in English place-names does not mean a 'fort,' but a 'down.' 'Combe,' as an element in place-names, does not occur 'especially in those counties in which the Celtic element is strong,' such as Devon and Cornwall. In Hampshire, for instance, where Celtic survivors of the Conquest must have been few in number, it is one of the commonest elements in names. Neither Lichfield in Staffordshire nor Litchfield in Hampshire has any nominal connexion with any word meaning 'corpse.' Hungerford has nothing to do with the Angles, but is a corruption of Hangra-ford, 'Ford of the Hanger' or 'Hanging Wood.'

Within ten years of the publication of Names and their History' Prof. Skeat published the beginnings of his work; and in the last fifteen years others have worked simultaneously on the place-names of various counties. Prof. Skeat's books deal with little more than the town and village names; and the same is more or less true of Mr Alexander's book on Oxfordshire, and Mr Roberts' on Sussex. The other authors † include a large number

* Published respectively by the Oxford University Press, 1912, and the Cambridge University Press, 1914.

E.g. Mr W. St Clair Baddeley, 'Place-Names of Gloucestershire' (Bellows, 1913); H. Mutschmann, Place-Names of Nottinghamshire'

of local names, which will be very valuable when sufficient data have been published to make it possible to draw general conclusions with regard to the distribution of names. Mr Duignan and Mr Baddeley have added archæological information of great interest-a very attractive feature in a subject which tends otherwise to become somewhat dry. It is impossible to foresee all the interesting results which would follow from a complete collection of the town, village, hamlet, local, and field names of this country, but two at least might be confidently expected; in the first place, the discovery of new regions of different dialect and usage in language; and, secondly, judging from the case of Hampshire and the Isle of Wight, the discovery of various regions in Saxon England which differed in some degree from their neighbours in respect to language.

The books under review contain matter which is contentious, because the investigators have been confronted again and again with very obscure and difficult problems. The demonstrable errors are very few, and, for the most part, unimportant. Prof. Sedgefield will probably not repeat in a second edition his remarks on Cunning Garth, which he asserts to mean 'King's Garth or Enclosure,' whereas it is only a form of Coneygarth, 'Rabbit Warren,' a name which, even in the field-names of Hampshire alone, is spelt in six different waysConeygear, Coneger, Coneygre, Coneygar, Coneygeer, and Cunningher. His somewhat widespread belief in the combination of Norse and Anglian elements in the same name is possibly right; but the phenomenon does not appear to occur frequently in other regions of Norse settlement in this country. Mr Roberts, too, might 'edit' his remarks on Arundel. But, after all, the truest criticism of these books on county names is that they show the wonderful progress which has been made in this still young twentieth century in this subject.

Of Mr Johnson's book on the place-names of England it may be said that it is as one born out of due timenot indeed too late, but too early. Enough has been said

(Camb. Univ. Press, 1913); A. Goodall, 'Place-Names of S.W. Yorkshire' (Camb. Univ. Press, 1914); W. J. Sedgefield, 'Place-Names of Cumberland and Westmoreland' (Longmans, 1915).

to show that the whole subject is at this moment undergoing a revolution which is far from complete; and that fact alone is sufficient to stamp such a work as ill-timed. The author has followed such recent writers as had published their work before his was finished; and there he is on safe ground. But he seems to have had equal faith in much of the work of the darkness which preceded the present dawn; he has at any rate reproduced many of the most notorious errors of the past.

So far, modern investigation has been confined to but a few of all the counties of England, and therefore does not provide material for final generalisations on the distribution of Celtic, Anglo-Saxon, Danish, and Norse elements in place-names. But still some interesting and unexpected results have been obtained. It was a favourite theory in the past, a theory very conspicuous in Canon Taylor's work, that later immigrants into this country had frequently attached topographical terms of their own to pre-existing Celtic names. Torpenhow, the name of a village in Cumberland, was what might be called the supreme instance' of this theory. How' was originally the simple name of the local hill. New settlers of different race came; and, not knowing that 'how' meant 'hill,' called it 'Penhow,' 'Hill-hill.' A third stratum of settlers, ignorant, as indeed they might well be, of the meaning of 'Penhow,' called the hill Torpenhow,' 'Hill-hill-hill.' The theory was very interesting and very fallacious. The whole name is pure Norse, Torfin-haugr, 'Torfin's Grave-mound.'

Save in those parts of England which remained in Celtic occupation after the first period of the AngloSaxon conquest, Celtic names or elements in names are very rare, and are mostly those of rivers. It was still more rare for the Saxons to add any element of their own to a Celtic river-name they had adopted. Micheldever in Hampshire is one of these rare cases, the name being compounded of the Saxon 'micel,' 'great,' and the Celtic defr' or 'dofr,' 'river.' But it is still commonly supposed that only the larger rivers and streams, such as Avon, Ouse, Stour, and so forth, retained their Celtic names, which are said to be all different words meaning 'water.' That this supposition is wrong is shown by the fact that in certain Saxon charters of Berkshire