the iron age shows a very distinct valley-ward movement with a few contour hill forts in suitable places. It is only in the Roman period that the forest land is seriously attacked at all; and, finally, the Anglo-Saxon age shows a preponderance of settlements in the forest, with a comparative desertion both of the open country and the fens.

If this theory could be accepted in its entirety, there would be no doubt about the comparative ages of the pre-historic trackways. The ridgeways would be the oldest, the hillside roads next in order, and the valley roads the most modern of all. This may prove to be the case, but a good deal of detailed investigation will be necessary before Dr. Fleure's brilliant generalisation can be accepted as an historic truth. As has already been observed, Dr. Fox finds that there is evidence in the Cambridge region that the hillside Icknield Way is as old as the ridgeways. The theory postulates a fairly stable population, whereas the influence of many invasions is a dominant factor in our history. The various invaders brought their own habits with them. Some might be men of the uplands, others of the valleys, and others of the forest, but not always in that order of arrival. The builders of the lake villages were pre-eminently valley people and they belonged to the bronze age. On the other hand, there is now an imposing accumulation of evidence that the builders of the hilltop camps were men of the iron age who arrived in the centuries immediately preceding the Roman occupation. Mr. Peake has suggested that they were the speakers of the Cymric tongue, whose place of origin was the steppes, which would explain their preference for the open spaces. If these things can be established, it would follow that the ridgeways were most used in the last centuries before Christ, although it does not follow that they were first used then.

The question can only be settled by an investigation-more detailed than any that has taken place so far of the exact relations between the ridgeways and the hilltop camps. It is quite easy to over-simplify the problem. It may well be that a general valley-ward movement of population was taking place gradually through many ages, and that it was retarded here and accelerated there by the influence of different invaders with diverse modes of life. Dr. Fox's first map showing thick clusters of neolithic finds on the open higher ground presupposes that these people

had some method of getting about the country. The natural inference would be that they first made the ridgeways. Such an inference would not be displaced by the fact that many centuries later an invading people who lived on the uplands as a matter of deliberate choice, erected their camps on the hilltops and also used the ridgeways. The spade alone can settle the matter.

(6) In the chalk counties, where the soil receives impressions easily and retains them indefinitely, the most impressive evidence of the old roads is to be found in the actual cattle tracks. It would be tedious on this occasion to explain the characteristics whereby they can be distinguished from ditches and from wheel tracks, but to the practised eye their signs are unmistakable. These things, with appropriate illustrations, can be found in Mr. Peake's paper and in Messrs. Hubbard's little book on "Neolithic Dew Ponds and Cattle Ways."

The caution must be uttered, however, that not all cattle tracks are pre-historic. Drovers' roads were common in the Middle Ages and later; in fact, the necessity of feeding the ever-growing population of London infested the ways leading to the capital with a never ending procession of flesh and fowl until the coming of the railways.

Where the old cattle tracks do exist in perfection, as upon the Berkshire ridgeway, they bring pre-historic life before the mind's eye in a manner that no theoretical reconstruction can rival. “It is on the steep descents that these trails are most clearly visible, for there the footprints sink deeper into the ground, and much of the surface is thus disturbed and subsequently washed away by the rain. Here the trails sink to considerable depths, often to five feet or more, while on the slopes of Walker's Hill, above Alton Priors, they attain the depth of at least twelve feet.”

As represented on the ground by the cattle tracks, the old roads are quite different from a modern highway. They are sometimes spread broadly, sometimes narrowly confined, according to the nature of the ground; in short, they represent the limits of deviation within which the cattle wandered. Mr. Peake is inclined to consider that on broad and level summits the average width was five chains, but narrowed to less than half that width where the ground compelled it. On a fairly steep incline, where the trails themselves are easily observed, the road tends to spread out like a fan from a narrow base at the bottom to a broad area

VOL. 241. NO. 491.

F

at the top, for man and beast both tend to zig-zag when facing

an ascent.

(7) A general feature of ridge roads is that they avoid the modern centres of population. Oft-times they go for twenty or thirty miles or more with scarcely a sign of habitation near them, and the modern explorer who follows them, like his pre-historic predecessor who made them, has to descend into the valley to get a drink. A road that appears deliberately to avoid even villages or hamlets is undoubtedly ancient. Obviously it did not grow that way, but the civilisation that formed it has passed from memory, the population that it served has found new homes, and it remains as a lonely but unmistakable relic of the times that are gone.

The second well-marked type of pre-historic road is the Harrow Way. This type derives its name-until a better one can be invented-from the track of that name which runs from Farnham to Salisbury Plain. Roads of this type are merely modified ridgeways. Most of the characteristics of the ridgeways apply to them also, but there are differences sufficient to justify their being placed in a separate class. They keep to the high ground, but they do not hug the actual water-parting with the persistency of the ridgeway. They are therefore more direct than the latter, they tend to cut corners, and do not object to crossing small streams. As an intermediate type they are, perhaps, less easy to find than the very distinctive ridgeways or hillside roads, but good examples are to be obtained where the country is suitable for their development. Up to the present they have been little studied and their archæology is an unworked field.

The hillside road is another matter. Some of the best work done upon the archæology of old roads has been devoted to the elucidation of this type. The Pilgrims' Way-Mr. Belloc's Old Road-for the greater part of its length belongs to it, and so does the equally well-known Icknield Way. As in the case of the ridgeways, it will tend to clearness to enumerate its habits, characters, and problems. They are mostly those formulated by Mr. Belloc, but revised in the light of later criticism.

(1) The hillside roads are in many instances Mr. Peake would say in most-parallel to a ridgeway. The obvious conclusion to draw from this fact would be that they were successive ; that the hillside road was later in date than the ridgeway, and

that its construction is evidence of a definite movement towards the valleys. The parallelism is clearly marked in the case of the Berkshire ridgeway and the Icknield Way. Mr. Belloc has noted the same fact in certain sections of the Pilgrims' Way, but only to draw an entirely different conclusion. The point has already been referred to above, but the available data are as yet insufficient to afford the basis for a decisive answer. But the fact of the frequent parallelism is clear, and it must be taken into account in any explanation.

(2) The road does not climb higher than it needs to. This is Mr. Belloc's third "habit," but can hardly be accepted as a dogma. The habit in fact varies according to the local conditions of soil and water supply. The hillside roads always keep above the alluvium and the flood-plain of the valleys, but they do not keep low as an invariable character. There are sections of the Pilgrims' Way along the North Downs where the road could have followed a lower contour if it had wished to do so. In Cambridgeshire, Dr. Fox notes that the Icknield Way keeps well above the spring line. The most reasonable conclusion seems to be that the roads choose the ground that provides the best going. Heavy clays will be avoided; in chalk counties the chalk will be followed if it can be reached. In short, bronze age man behaved as a reasonable man might be expected to behave. He would not climb unnecessarily, but he would secure the best foothold, even if it involved a climb to do so.

(3) "The road always keeps to the southern slope, where it clings to the hills, and to the northern bank (i.e., the southern slope) of a stream." This is one of Mr. Belloc's most debatable propositions, and he appears to have drawn an invalid generalisation from a particular instance. The reason that he gives is an excellent one, namely, that the road kept to the south slope because it was sunny and dry. At the same time he notes the exceptions carefully, and these exceptions appear to indicate the true rule. In three instances the Pilgrims' Way goes to the north of the crest to avoid going round a projecting spur with its re-entrant curve; or, in other words, it traverses the base of a triangle instead of the two other sides. Mr. Belloc was so engrossed in his old road that he forgot another old road on the other side of the chalk mass. The Icknield Way, an equally famous hillside road, runs along the escarpment of the chalk from

Thetford, in Norfolk, until it loses itself on Salisbury Plain. For all that long distance it keeps assiduously to the north or northwestern slope of the hills in direct defiance of Mr. Belloc's rule. A comparison between these two famous examples suggests the true inference. It is not a question of a sunny slope, but of a straight unbroken slope, with the fewest number of projecting spurs. A projecting spur means a long way round or a steep way across; an unbroken slope means steady going. If the unbroken slope is also a sunny slope it is all to the good, but if not, the advantage of the steady going far outweighs the absence of sun.

(4) Wherever the road goes right up to the site of a church, it passes upon the southern side of that site." This intensely interesting suggestion calls for full investigation which it has not yet received. It should be considered not only with reference to churches, but also with reference to other sites of possible religious significance. If the argument set out in the last paragraph holds good, the rule would be that the road goes below the sacred site. Mr. Belloc notes that his road usually passed to the north of that is above-the present villages, which is a valuable indication of the valley-ward movement of population.

(5) The habit of the road, when ascending a hill, of taking a straight way to the summit has already been noticed as characteristic of prehistoric roads of all types. It is only when the use of wheeled traffic becomes general that easy gradients become essential. The tendency to seek the saddle of the watershed when passing from one valley to another is a distinctive mark of the hillside roads only. From the nature of the case it cannot apply to the ridgeways.

The fourth type of prehistoric road is the hollow way. It is the least studied of any, but is perhaps the most widespread. Messrs. Hubbard, in the work already cited, have recorded some good observations, but only in the chalk country. This type of road is totally different from any of those already discussed. It has nothing of the nature of a through route about it, and usually consists of a short stretch-sometimes still in use, but often abandoned-leading down from the higher ground to the lower. Everybody has walked along a sunken lane with high banks obscuring the view; everybody has read of the sunken road of Smohain and the part it played in the battle of Waterloo; but few have speculated upon its antiquity. Roads of this class present