Ancient Britain and the Invasions of Julius Caesar

The Configuration of the Coast of Kent in the Time of Caesar

This volume is not a treatise upon the physical geography of Ancient Britain; and I am only concerned with geographical questions in so far as they are essential to a right understanding of the history. It is impossible to understand the narrative of Caesar’s invasions of Britain without considering how far the physical geography of that part of the island which was the theatre of his operations differed from what it is now.

I. Between Ramsgate and Sandown Castle

Thanet, as everybody knows, was an island in Caesar’s time; and Bede2456 says that it was separated from the mainland by an estuary three furlongs broad: but the late George Dowker2457 concluded from ‘an attentive examination of the estuary’ that it was ‘much shallower and narrower than is generally supposed’.

John Lewis,2458 a well-known antiquary of the eighteenth century, and William Boys,2459 the historian of Sandwich, maintained that an estuary, in which was included the harbour of Richborough, known to the Romans as Portus Ritupis, had extended from the cliffs of Ramsgate southward to Walmer, covering the sites of Stonar and Sandwich and indeed the whole of the low ground between Sandwich and Deal, and washing the shore of an island on which stood Richborough Castle. A recent writer, Mr. H. Sharpe,2460 who endorses this opinion, argues that the Roman road from Canterbury to Richborough harbour (ad portum Ritupis2461) terminated at Each End. The road ‘cannot’, he insists, ‘have run to Sandwich in Roman times. Montagu Burrows... Cinque Ports, 1888, p. 30,2462 says—“Sandwich and Stonar are wholly English. No Roman remains have been found at either”... there is good reason to suppose that the land upon which it [Sandwich] stands and the land over which the Sandwich end of the road runs were not formed when the Romans were here.’2463 And again, ‘There is another reason for supposing that Each End was... the place where the boats left the mainland for the island [of Richborough]. [The road running northward from Dover] is marked on the Ordnance map2464 as a Roman road, and if complete would run to Each End, not to Richborough Castle or to Sandwich... the last mile from [Woodnesborough] to Each End, is missing.’2465

Now, in regard to Stonar, Professor Burrows, as we shall presently see, is mistaken; and, granting that the Roman road from Dover would, if complete, run to Each End, how can Mr. Sharpe prove that it did not run further? The late George Dowker stated, in a paper which was published after his death, that he had himself ‘traced the Roman road to Woodnesborough, and thence by Each End to near the Richborough Island’;2466 and the views of Lewis and Boys, which Mr. Sharpe endorses, as to the wide extent of the estuary at the time of the Roman conquest of Britain have been stultified by discoveries to which Mr. Sharpe does not allude. Roach Smith affirms that ‘Roman remains, indicative of habitations, have been discovered in the sand-hills considerably to the north of Sandown Castle’, and that ‘coins have been found at Stonar, opposite to Richborough’; and from these facts he infers that ‘the recession of the sea from the low land between Thanet and Walmer probably commenced at a period much earlier than has been commonly supposed’.2467

That the hill on which Richborough Castle stood was nearly if not quite insulated is generally admitted;2468 but Mr. George E. Fox remarks that it ‘was probably not washed by the open sea, though a broad channel may have flowed close beside it, forming one of the southern mouths of the strait, while a narrow strip of salt-marsh and sand-bank lay between it and the open sea’. It would be more correct to say that the island, on its eastern side, was separated by a channel from Stonar Beach, the southern extremity of which lay east by north of the site of Sandwich: the sand-hills were on the south-eastern side of this beach, from which they were divided by a narrow channel. Mr. Fox goes on to say that ‘a large extent of what is now marshland, lying to the west of the hill, may then have... formed the haven,2469 making of the camp hill an island’. He argues, however, that, on the eastern side, the channel ‘could not have hugged the hill very closely, as at no great distance to the south of the station on this same side, and in the low ground presumably near the shore, fragments of a Roman house were discovered in 1846’.2470

In the year 1876 Dowker affirmed that ‘the low shore and sand hills’ which now extend from the Deal beach to the latitude of Sandwich ‘extended [in the time of Caesar] much less than at present’;2471 and in a map which accompanied his paper2472 he contrasted the low-water line between Walmer and Sandwich, as he believed it to have existed in 55 B.C., with the low-water line as it existed at the time when he wrote. In the latitude of Sandwich the modern low-water line is traced on this map a mile and a half east of the hypothetical ancient line, which distance gradually diminishes to three-quarters of a mile in the latitude of Worth and about one furlong in the latitude of Deal. I find a difficulty in reconciling this map with Dowker’s own statement that ‘Roman pottery, coins and traces of the Roman occupation have been found in the sand-hills—and indeed below the sand-hills considerably northward of Deal, beyond Sandown Castle’;2473 and from the fact which this statement records it follows that, in the time of the Roman occupation of Britain, the shore-line at the place where the discoveries in question were made cannot have been widely different from what it is now.

II. Between Sandown Castle and Walmer Castle

When we endeavour to trace the shore-line, as it existed in Caesar’s time, opposite Deal and Walmer, we find that the writers who have dealt with the question differ widely among themselves; while Dowker again shows himself a most troublesome witness. Unfortunately this meritorious geologist, who laboured hard to elucidate the geographical questions connected with the ancient history of East Kent, was a bad writer, and sometimes failed to make his meaning clear.

Major Rennell, who was in his day ‘the acknowledged head of British geographers’,2474 believed that Caesar landed at Deal. ‘Of course,’ he says, ‘the margin of the ancient beach, on which Caesar landed, must now be very far within land, as well as very considerably raised.’2475 The words ‘of course’ prepare us for the discovery that Rennell quotes no authority and gives no reasons.

Professor Montagu Burrows,2476 also without giving either authority or reason, tells us that Deal ‘probably had once a haven, which was choked up in very early times’. But choked up it was not unless it existed; and observe that its existence is only ‘probable’. As a matter of fact, the so-called probability is unsupported by any evidence.2477 The professor goes on to say that ‘the old town [of Deal] was already separated from the sea by a considerable interval when Henry [the Eighth] built the three castles of Deal, Sandown, and Walmer for the protection of the coast, which had now become a continuous stretch of steep shingly beach’. Now if, in the time of Henry the Eighth, ‘the old town was already separated from the sea by a considerable interval,’ the inference is that it had once been quite close to the sea; and of this there is no evidence. Was the professor thinking of Leland,2478 who describes ‘Deale’ as ‘half a Myle fro the Shore of the Se, a Fisshcher Village iii. Myles or more above Sandwic’? If so, why should he assume that because Deal in the time of Leland, that is to say, of Henry the Eighth, was half a mile from the sea, it had once been on the sea? The only conceivable reply to this question would be that as Upper Deal is now more than half a mile from the sea,2479 and as, according to Leland, it was only half a mile from the sea in the time of Henry, it may once have been actually on the seashore. But Deal Castle was built by Henry; and the sea was therefore at least as far from Upper Deal in his time as it is now. The truth is that Leland’s ‘Myles’ were sometimes very long: he tells us that Sandwich was ‘iii. Myles’ from Deal, and it is really six.

Dowker, in the paper which he published in 1876,2480 maintained that ‘Deal probably did not exist in Roman times’, and that, when Caesar landed in Britain, ‘the coast was cut back behind Deal’:2481 that is to say, he virtually committed himself to agreement with the view, already stated, of Major Rennell. In the same paper he affirmed that ‘the present town of Deal is situated on a comparatively recent beach’, and went on to say, in proof of his assertion, ‘I have evidence of the beach at the back of Deal containing mediaeval remains.’2482 What the evidence was, he did not say; and what he meant by ‘the beach at the back of Deal’, I do not know. In 1887 another paper2483 was published, containing a report of his views. Herein I find that there is ‘no evidence’ of ‘a shore-line cutting far back beyond the Deal beach’. No evidence in 1887, though in 1876 the evidence was irrefragable.2484

The opinion of Stukeley,2485 who believed that Caesar had landed between Walmer Castle and Deal, was diametrically opposed to that of Rennell. He maintained that Caesar’s camps must have been ‘absorpt by the ocean, which has so long been... wasting the land away’. ‘Even since Henry the VIII^ths time,’ he continued, ‘it has carried off the seaward esplanade of the three castles’ [of Walmer, Deal, and Sandown].2486 But it does not follow that in the interval which separated the time of Caesar from the time of Henry the Eighth the sea in the neighbourhood of Deal had been continuously gaining upon the land. It would appear that in the last four centuries it has alternately advanced a little and receded.2487 In 1615, 1626, and 1627 the waves were wearing away the walls which had been erected for the protection of the castles of Walmer and Deal.2488 During the latter half of the eighteenth century, however, shingle was being rapidly thrown up along the coast between St. Margaret’s Bay and a point which, as Mr. Elvin2489 says, was ‘considerably to the north of Sandown Castle’; and, although during the first thirty years of the nineteenth century the sea was again encroaching, at all events at Walmer, the bank of shingle between the Rifle Range at Kingsdown and Walmer then began again to increase, while northward of Deal as far as Sandown Castle the sea was simultaneously gaining ground. In 1885 shingle was still accumulating at Walmer Castle and also at Deal, although it was recognized that at the latter place its movements were variable. For some years previously, however, the shingle which formerly protected the cliffs between St. Margaret’s and Kingsdown had been travelling northwards past Walmer to Deal; and during the fourteen years that followed 1885 the same process was going on: I daresay it is going on still. At Deal, wrote Dowker in 1899,2490 ‘the shore line has been nearly stationary until we approach the north end of Deal, where the... sea had washed most of the beach away and carried it past the Castle.’ Finally, it must be borne in mind that from various places between Walmer and the North Foreland a great deal of shingle has been abstracted.2491 Still, if The North West View of Walmer Castle, by S. and N. Buck, which was published in 1735, was approximately accurate, the sea was a good deal nearer the castle then than it is now; and the observations that were made between 1741 and 1884 show that while in that period the sea at Sandown Castle gained 200 feet upon the land, off Deal Castle the increase of shingle amounted to 120 feet, and off Walmer Castle to no less than 385.

The Reverend Beale Poste, a well-known antiquary of the nineteenth century, maintained2492 that the bank of beach upon which Deal stands must have existed in the time of Caesar, ‘since numerous Roman coins are found at neap tides at low water on the chalk at the edge of the beach.’ He added that ‘when the piles for the pier were driven into the beach in 1842, it was found in a highly concrete state, almost like rock, denoting great antiquity’. The former statement, if it is correct,2493 would seem to prove that the shore-line has receded, in other words, that the sea has on the whole gained upon the land since the days of Caesar; the argument based upon the condition of the beach into which the piles were driven only tends to show that the lower stratum of the beach was old.

Quite recently a discovery has been made which ought to set the question at rest. Romano-British interments have been unearthed about seven hundred yards north of Walmer Castle, ‘on the low ground... adjoining, and only on a slightly higher level than the Castle meadows.’2494 The spot where they lay is about two hundred and fifty feet west of the high-water mark of ordinary tides. The discovery, as Mr. Cumberland Woodruff remarks,2495 proves that ‘the shore lands [between Walmer and Deal] were protected then as now, though probably [or rather certainly] by a much thinner line of shingle’.2496

The conclusion appears to be this. There is no reason to suppose that the coast-line between Sandown Castle and Walmer Castle was very different in Caesar’s time from that which is depicted on the Ordnance Map; and there is positive proof that between Walmer Castle and Deal Castle, at some period of the Roman occupation, it was nearly the same. On the other hand, it is certain that since Caesar landed a great deal of shingle has accumulated along this part of the coast, especially at Walmer; and it may be inferred that the beach was less steep then than it is now.

III. The Goodwin Sands

Before we attempt to inquire what was the condition of the Goodwin Sands in the time of Caesar, it will be well to state the relevant facts which have been ascertained since exact observations began to be recorded.

‘The north-eastern part of the North Goodwin,’ says the author of the Channel Pilot,2497 ‘dries in places 7 feet at low water; the South Goodwin not more than 4 or 5 feet at any part.’

The form of the sands is altered periodically by the tides. Beale Poste argued in 1857 that the Goodwin Sands were still growing, as ‘Kingsdown Mark, a pile... built in the reign of Elizabeth to show the South Sand head, is... of no use, the sand having now extended itself a mile further to the southward’. Moreover, he says, it was stated in the Report of the Commission of the Harbours of Refuge for 1845 that ‘the Brake Sand, a branch of the Goodwin Sands in the Small Downs, had moved bodily inwards towards the shore seven hundred yards within the last fifty years’. This, he maintains, can only mean that ‘a deposit has taken place on the inward side of the sand... while the outward side has been eroded by the winds and tides’.2498 In 1885 it was found that ‘the former Bunthead shoal’ had ‘entirely disappeared’,2499 and that ‘the whole body of the South Calliper’ had ‘moved about a mile north-eastward’. Again, it was ascertained by ‘a re-survey of the Downs, Goodwin Sands, and adjacent coast’, executed in 1896, that since 1887 considerable changes had taken place. ‘The Goodwin Sand,’ we learn from this source, ‘has continued its general movement towards the coast, and the area of drying sand has largely increased.’2500

The results of borings carried out at various times in the Goodwin Sands have shown that blue clay, resting on chalk, was found at the depths of 7, 15, 57, and 78 feet.2501 From these data Sir Charles Lyell2502 concludes that the Goodwins ‘are a remnant of land, and not “a mere accumulation of sea sand”;’ and, referring to the destructive storm mentioned in the Anglo-Saxon Chronicle2503 as having occurred in 1099, he conjectures that ‘the last remains of an island, consisting, like Sheppey, of clay, may perhaps have been carried away about that time’.

Dr. Guest2504 holds that in Caesar’s time the Goodwin Sands did not exist. He reminds us that, according to Somner,2505 it was the opinion of ‘several men of judgement’ that they had not appeared until after the time of Earl Godwin, and, remarking that this was also the view of Sir Thomas More, he argues that ‘we may infer that such at that period was the opinion of educated men who had local knowledge’. Leland,2506 he goes on to say, ‘attributed the decay of Sandwich to the Goodwin Sands, and as Sandwich was a flourishing port in the fourteenth century, we may infer that it was not till the fifteenth that the sands attained those formidable dimensions which produced so much mischief.’ Immediately north of Sandown Castle there is, he observes, a tract of land covered with low sand-hills, which, in Philipot’s map of Kent, are called the ‘smale downs’,2507 and upon which the sea has long been encroaching. He accounts for the name given to the roadstead by assuming that it once formed part of the ‘smale downs’, and affirms his belief that ‘the flats round Sandwich once projected into the sea as a low ness or foreland,—probably divided into islands, of which Lomea [an island which John Twine asserted to have formerly existed about four miles from Thanet] was the easternmost’. He assumes that as Lomea is not mentioned in Domesday Book, it perished by some natural convulsion before the end of the eleventh century, and goes on to say that ‘After the destruction of this island, the Goodwin Sands may have been gradually accumulated, not necessarily on the site of the island, but near it, and the Downs just as gradually excavated’.2508 Beale Poste2509 also affirms that in 1098 ‘an island named Lomea was overflowed, on which occasion the sands are said to have been formed. This is mentioned by Giraldus Cambrensis, and from him by Twine.... But Earl Goodwin (sic) died in... 1053, and Domesday-book negatives that any extensive tract of land was overflowed and lost, in this direction.’

Now John Twine2510 (or Twyne) merely says that he has read about Lomea in the works of ‘certain writers’. It was once, he says, a low fertile island, which was submerged in consequence of a great storm, and covered with sand, and it is now the Goodwin Sands. As for Giraldus Cambrensis, I have searched his writings diligently, and I can find no mention whatever therein either of Lomea2511 or of the Goodwin Sands. The name ‘Downs’ is easily accounted for. ‘The DOWNS,’ says the author of the Channel Pilot,2512 ‘in a general sense, implies the numerous banks lying immediately off the coast between the South and North Forelands... that [anchorage] which is commonly... known as the Downs is off the town of Deal between Walmer Castle and the northern part of the town,’ &c. I see no reason to doubt that the name of the roadstead is derived from the aforesaid banks and from the sand-dunes on the shore.

Somner,2513 remarking that, according to the common opinion, Lomea was submerged in 1097, observes that there is no notice of such an island either in Domesday Book or in ‘any Author whether foreign or domestick, of any antiquity, that ever I could meet with’.

The late C. H. Pearson2514 inferred from ‘the legend of their formation’2515 that the sands were ‘first remarked about the end of the eleventh century’, and that they were ‘probably formed by bank-currents gradually depositing sand about a shoal’.

On the other hand, S. Pritchard,2516 the historian, so called, of Deal, argues that the sands must have existed ‘from all time’ as otherwise Deal and the adjoining country would inevitably have been inundated. Why? The island, the former existence of which is assumed by Sir Charles Lyell, would have been as good a protection as the sands; and in the time, which was certainly anterior to the Roman invasion of Britain, when the shingle bank had not accumulated to a sufficient height,2517 the very small area in the neighbourhood of Deal which is below high-water mark may have been inundated, unless, as Dowker2518 and Mr. Spurrell2519 believe, the level of the land has been depressed since the Roman occupation.

The reader has doubtless already concluded that it is impossible to affirm either that the Goodwin Sands existed in the time of Caesar, or that they had not then accumulated to such a degree as to attract attention, or that their place was occupied by an island. If the silence of Domesday Book and, as it should seem, the absence of any other positive testimony constitutes an argument against the hypothesis of Sir Charles Lyell,2520 the same argument may be advanced to show that before the Norman Conquest the sands had not begun to appear. Yet, as we shall see in a subsequent article, there is some reason to believe that either sands or an island were there when Caesar invaded Britain.2521 Tradition, vague as it is, combined with Lyell’s authority, disposes me to accept tentatively the latter alternative.

IV. The South Foreland and the Dover Cliffs

Professor Montagu Burrows2522 affirms that ‘the space over which the tides travel [in the Straits of Dover] must be at least two miles wider than it was some 2,000 years ago’. This is one of the ex cathedra statements in which the professor’s work abounds, and for proof of which his amazed readers search his pages in vain. Dowker’s estimate is more moderate: he only bids us ‘assume the Straits are now one mile wider than when Caesar visited our shores’;2523 but, like Professor Burrows, he requires us to make this assumption in the dark.

In M. Vivien de St.-Martin’s great work it is stated that Cape Grisnez ‘perd en moyenne 25 centim. par an; autrement dit, il recule 25 m. par siècle’.2524 Assuming the accuracy of this statement, and assuming, further, that the rate of erosion has been constant since the invasion of Caesar, Cape Grisnez then projected seaward 489 metres, or about 534 yards further than it does now. I take for granted that the statement is based upon exact and prolonged observation; but when did that observation begin?2525

As for the South Foreland, it is certain that, as Dowker says,2526 it is (or at all events was in 1885 and for some years previously) ‘being gradually undermined by the sea’; but it would be a great mistake to leap to the conclusion that this erosion has been going on continuously since the time of Caesar. In 1850 Captain K. B. Martin, who was harbour-master of Ramsgate, affirmed that the cliff between Dover and the South Foreland, being protected by ‘an inclined plane of shingle’ from the sea, had ‘preserved its contour from time immemorial’.2527 The phrase is somewhat vague: but the captain was a careful observer; and we may believe him when he tells us that since his boyhood, fifty years before the time when he wrote, there had been no change.2528 Why, then, were the Dover cliffs and the South Foreland being gradually eaten away in 1876, when Dowker wrote, and in 1884? Simply because the supply of shingle had, from various causes, been cut off.2529 The erosion, said Mr. E. R. N. Druce, Engineer to the Government pier at Dover, takes place ‘at no particular rate, but falls of cliff at the points above named have taken place at intervals for some years past... since they have lost the protection of the shingle at their base’. He added that the loss was ‘confined to areas bare of shingle’, and that, so far as he could ascertain, there existed no ‘data for determining the rate of erosion from early maps or other documents’.2530 It would appear, then, that Professor Burrows’s assertion is based upon pure imagination.

When the subsidence which had taken place in the Neolithic Age was virtually complete the sea was bordered by a narrow plain, to which the high ground descended gradually. Erosion was at first rapid while the waters were devouring loose talus; but when beaches had had time to form it was of course retarded.2531 How slow it is where the rocks are hard is proved by the fact that the contour of a prehistoric camp near Hastings shows that the seaward defence was formed not by an artificial rampart but by the East Cliff.2532 Yet Professor Burrows asks us to believe that erosion has been as rapid in the chalk of the South Foreland as in the soft cliffs between Flamborough Head and the Thames.2533 Generally speaking, as erosion proceeds, cliffs become higher;2534 and it is obvious that if the Channel had been two miles wider in Caesar’s time, the Dover cliffs, if they had existed, would have been insignificant. But since Caesar described them as ‘precipitous heights’,2535 and Cicero as ‘astonishing masses of cliff’,2536 they were evidently little lower then than now. Let the reader ponder these things, and he will realize how monstrously exaggerated is the estimate which assigns to the Straits of Caesar’s time a breadth two miles less than our modern maps show.2537

V. Dover Harbour

That a natural harbour existed at Dover in the time of Caesar is beyond dispute. It is mentioned under the name of Portus Dubris in the Itinerary of Antonine;2538 and it was connected by a Roman road with Canterbury and London, and also with Richborough. Napoleon the Third2539 affirms that it was entirely choked up about 950 A.D.; but this is a blunder, for the harbour is mentioned in Domesday Book.2540 Even as late as 1582 it was stated by an engineer, named Thomas Digges, that ‘Before the peere was builte out, there are men alyue can remember that was no banckes or shelues of beache to be seene before Douer,2541 but all cleane sea betwene Arteclif [Archcliff] tower and the castle clyffe’.2542 Captain Martin2543 holds that the remains of anchors which have been dug up out of meadows in the valley prove that the estuary was navigable as far as Crabble;2544 and he believes that it actually extended to Water’s End,2545 and covered the sites of the villages of Charlton and Buckland. Canon Puckle, however, argues that ‘the primitive haven’ covered a space which extended barely a quarter of a mile inland, ‘bounded by the lower half of St. James’ Street, Dolphin Lane, and Russell Street, and the east end of Dolphin Lane,’2546 and he states that when this area was ‘partly uncovered in excavating for the new Russell Street gas works, quays and hawser-rings were brought to light’. Captain Martin’s estimate, which is based upon very uncertain data, must be regarded as an exaggeration: the estuary may possibly have extended up to Crabble, but was certainly not navigable so far except perhaps by coracles. Many years ago the remains of a Roman bath were discovered on the site of St. Mary’s church,2547 and in 1887 a statue belonging to the period of the Roman occupation was found ‘during excavations for the foundation of the Carlton Club, in the Market Place’.2548 These discoveries help to define approximately the western limit of the harbour; and I believe that Planche 17 of the Atlas accompanying Napoleon’s Histoire de Jules César2549 represents it with tolerable accuracy.

Map of Romney Marsh Proper and the Parts Adjacent

Reproduced from the map facing page liii of T. Lewin’s “Invasion of Britain by Julius Cæsar,” 2nd. Edit. Showing what lands would have been covered by the sea at high water (medium spring tides) before the construction of the Rhee Wall. The figures denote the depth in feet, according to levels taken by J. Elliott, of the present surface below the high-water mark of Spring Tides.

ROMNEYMARSH
as (according to T. Lewin’s final view)
It was certainly in the
TIME OF THE SAXONS
probably in the
TIME OF THE ROMANS
and perhaps in the
TIME OF THE BRITONS

VI. Between Dover and Sandgate

During the last three centuries, at all events, the coast between Sandgate and Dover has undergone considerable changes. Large quantities of stone have been removed from the Folkestone cliffs; and landslips have occurred at Shakespeare’s Cliff, between Folkestone and Sandgate, and behind East Wear Bay.2550 It would be useless, however, for our purpose, to describe these changes in detail; for they do not affect the topographical questions that belong to the history of Caesar’s invasions of Britain. Excepting the disappearance of the little haven that once existed at Folkestone, the general character of this section of the coast was much the same in 55 B.C. as to-day. It may be, however, that the aspect of the high ground above East Wear Bay was different. Between the cliffs and the heights which rise about a quarter of a mile to the north of them there is a wild and broken plateau, called the Warren, through which the railway runs. Referring to this, William Phillips, a geologist of some repute, wrote in 1821, ‘The cliff, bounding this ruin towards the sea, is, from its position, not in situ; and it is equally clear that the enormous masses of which it is composed, have fallen forward [probably by ‘repeated falls’] from near the summit of the cliff in situ.’2551 When these convulsions began to transform the landscape cannot, as far as I know, be ascertained.

VII. Romney Marsh

Between Hythe and Dungeness, on the other hand, there has been complete transformation. There, within the brief span of historical time, wind, tide, and river, and finally the labour of man, have wrought changes as remarkable as those that in other regions required the lapse of ages which the imagination fails to conceive. The antiquary who walks from Westenhanger Station to the brow of Lympne Hill, and looks out over the vast field of shingle that extends seaward, and, on his left, towards Hythe, and then over the broad level of the marsh that stretches away on his right between the Wealden upland and Dymchurch Wall, will easily picture to himself the scene that once was there.

1. Before we attempt to construct a map which may represent the coast-line between Sandgate and Dungeness, as it was in the time of Caesar, it will be well to state those relevant facts which are accepted by all geographers. There was a time when the area of Romney Marsh was covered by a bay. At a later epoch the marsh was fringed by a bar of shingle, which extended from Winchelsea to a point nearly opposite Shorncliffe. Between West Hythe and Shorncliffe streams flowed down from the hills, gradually forced an opening in the shingle opposite Hythe, through which the sea entered, and thus formed Hythe harbour, which, after remaining open for many centuries, was finally choked up about 300 years ago. For some time after the marsh became habitable the shingle protected it from the sea on the south, but gradually was so diminished that it became necessary to construct a sea wall. The river Rother debouched at some point within the area of Romney Marsh. During the Roman occupation of Britain there was a harbour called the Portus Lemanis, which has been located by one writer at Romney and by others at Lympne, while some have identified it with Hythe Haven. West of West Hythe Oaks, the marsh ‘is a rich mould... while all to the east, as far as Sandgate, is (with the exception of a narrow strip to the south and east of Hythe, between the sea-beach and the hills) one vast bed of shingle’.2552

2. The whole of Romney Marsh, properly so called,2553 is even now below the level of high water at spring tides. The hills which form its northern boundary have themselves changed since the time when the waves broke against their base. In the course of ages they have lost their original sharpness of outline, and, as we learn from the geologist who has described the formation of the Weald, have been ‘worn down into undulating ground’;2554 and nearly 200 years ago a local observer described how, after an unusually wet season, Lympne Hill had been completely transformed, in a single night, by a landslip.2555 But these changes are insignificant in comparison with that by which the old Bay of Appledore has become a fertile pasture. Of what material is this land composed? According to the late Thomas Lewin, it is ‘absolutely and exclusively a sea deposit’; and, in proof of this assertion, he pointed to ‘the marine shells which pervade the whole mass’.2556 But it needs little acumen to see that the presence of marine shells in the marsh does not justify Lewin in using the words ‘absolutely and exclusively’; and the late Colonel George Greenwood maintained that the marsh had been formed by material brought down from the Weald by ‘the aqueous erosion of the Rother’.2557 As a matter of fact, it was formed by the combined action of river and sea.2558 But unless and until a series of borings are systematically made, it will be impossible to describe the recent strata with precision.2559

According to Topley, ‘The cause of the original formation of Romney Marsh is altogether unknown. It is usually attributed to “the meeting of the tides”; but as this takes place over a rather wide area, and as shingle beaches and alluvial flats occur where no tides meet, the explanation is not altogether satisfactory.’2560 The well-known geologist, F. Drew, explains that as soon as the bay had become so shallow from the accumulation of silt that its bed was exposed at low water, the sediment carried down by the Rother began to be deposited on the surface. Like Topley, he confesses that how the silt had accumulated is ‘not quite clear’; and he thinks that ‘the newly formed surface’ may have been ‘actually upheaved by oscillation of level, forming a plain well raised above the level of the sea’,2561 which, however, before the historic period, must have suffered a subsidence.2562 This supposition was based upon the fact that trees are found near Appledore a few feet below the surface, which, if they are in situ, must have grown at a time when the marsh was above the level of the sea, and were perhaps contemporaneous with the submerged forests of Devonshire and Cornwall.2563 Some authorities, however, as we shall presently see, hold2564 that they were drifted into their present position.

The late James Elliott, who in the middle of the nineteenth century was engineer of Dymchurch Wall, diligently investigated the history of the marsh, and added much to our knowledge. While the marsh was being formed it was gradually closed by a bar of shingle, composed of pebbles which had been partly broken off from the cliffs on the south-west, partly carried down by rivers,2565 and had been driven up the Channel by the prevailing winds.2566 Elliott remarks that ‘the result of such a protection from the open sea would be, that all matter brought down by the hills would rest nearly where it was first deposited, and, in process of time, dry land, at certain states of the tide, would appear’; and that, on the ebb of every tide, ‘all the water in the bay gradually receded towards the hills, and... made its exit at the eastern end of the shingle bank.’2567 He concludes that the shingle extended rapidly until it reached the eastern end of what is now Dymchurch Wall, but that its progress thenceforward was extremely slow. Meanwhile the sediment deposited by the sea was gradually raising the surface of the marsh.2568 Elliott, whose statements and opinions were incorporated by Lewin in his book on the invasions of Caesar, affirms that the advancing shingle spit was ‘intersected only by a channel between Lydd and Romney’, which was ‘the mouth of the estuary which lay behind the shingle’;2569 but Lewin, in a later article on the Portus Lemanis,2570 appears to have abandoned this view, for he there implies that the spit was continuous. At some period which preceded the erection of the Rhee Wall, that is to say, the first enclosure or ‘inning’ of the marsh, it would appear to have reached the foot of the hills at West Hythe Oaks.2571 The result, according to Lewin, was that the marsh was temporarily enclosed. But, he says, ‘this bar to the exit of waters from the marsh could not long continue, for, though the sea was excluded, the Limen [that is to say, the Rother]... and twenty smaller streams were continually increasing the volume of water within the marsh, and... the shingle spit was burst asunder between Romney and Lydd.’ Thus, if Lewin’s final view is correct, the sea again found an entrance on the west of Romney, and continued to overflow the marsh at high tide until it was finally shut out by the erection of the Rhee Wall. West Hythe Oaks was not the final goal of the shingle spit. For a long period, as Lewin remarks, ‘the shingle from the west continued to advance... and for a time without again touching the hills;’ but at length the advancing spit ‘was again wrested aside and dashed against the hills at Hythe, between the present barracks and the more eastern of the two Hythe bridges over the canal’. According to Elliott, however, whose view was adopted by Lewin in the Appendix to his book on the invasion of Britain by Caesar, the shingle was not ‘dashed against the hills at Hythe’, but opposite Shorncliffe. Anyhow the final result was that from the eastern end of what is now Dymchurch Wall to a point nearly opposite Shorncliffe there extended an irregular tract of shingle, broken only opposite Hythe by an opening, which led to a narrow harbour extending along the foot of the hills. This opening was due to the streams which flowed down from the hills and found a vent by bursting the barrier of shingle, and the scour of which kept the harbour open until, about three hundred years ago, it was finally choked up. According to Elliott, the western extremity of this harbour was at West Hythe Oaks; according to Lewin’s final view at Hythe itself. Between Dymchurch and Hythe the shingle formed a broad field; but the section between Hythe and Shorncliffe, which formed the southern boundary of Hythe harbour, was long and narrow. The whole tract was ‘perfectly flat and above high-water mark’; and Elliott argues that it extended much further seaward in Caesar’s time than it does now, because, while the supply of shingle drifted from the south-west was cut off by the gradual elongation of Dungeness, the eastward movement of the shingle along the fringe of the marsh still went on.2572 This argument he supports by a comparison of the Ordnance Survey map executed in 1817 with an old map of the marsh, probably made about the year 1550, which is in the Cottonian MSS.2573 at the British Museum. Assuming the accuracy of the old map, it would appear that in the 267 years the shingle had receded about two furlongs; and Elliott concluded that in Caesar’s time the coast line at Hythe must have been nearly a mile from the hills. Having had considerable experience in the handling of old maps, I so far differ from Elliott that I am rather disposed to assume the inaccuracy of the one on which he relies; but he is quite justified in concluding that the coast line was much further from the Hythe hills in 55 B.C. than now.2574

Elliott’s account of the formation of the Marsh has, however, been recently disputed in a paper by George Dowker,2575 which, although it swarms with bibliographical and historical mistakes,2576 cannot safely be ignored. The author begins by endeavouring to show that the Rother originally entered the sea at Romney; that it gradually raised both its bed and its banks by depositing sediment; and that ‘the Rhee Wall was, in the first place, a natural river-bank’—the bank of the Rother—‘subsequently raised and altered by the Barons of the Cinque Port of Romney’,2577 but (if I have grasped his meaning, which is often obscure) only between Snargate and Warehorn.2578 He tells us that ‘The sequence of changes in the Marsh may be summarized as follows:—Firstly, a shallow bay existed in a depression in the underlying rocks. Into this bay the waters of the Rother, Tillingham, and Brede, on their way to their outlet near Romney, deposited their silt, so that the northern half of the Marsh had become dry land previous to the time of the Romans. Around this bay were formed sand-hills. In time of flood the waters of the river that ran out at Romney overflowed, and, depositing silt, raised the banks on either side. A slight depression of the land commenced, and has continued. Beaches accumulated, especially between Romney and Hythe, and between Romney and Winchelsea. Romney probably formed a promontory near Dymchurch, near where the ancient river, then called the Limen, discharged its waters.’2579 He explains that originally the sea was excluded from the marsh by sand-hills, and that ‘the sand-hills appear to have been formed at a period before the accumulation of the beaches had commenced, since the beach effectually stops the formation of sand-hills’.2580 No sand-hills now exist in the marsh, except between Rye and Lydd, near New Romney, and near West Hythe; but, says Dowker, ‘We may connect these sand-hills by a hypothetical line extending from Rye to Hythe.’2581 The reason which he gives for believing that there has been a depression of the land since the time of the Romans is that he has found evidences of post-Roman subsidence in ‘the neighbourhood of Richborough, Reculvers, and the Swale marshes of Sittingbourne’.2582

Now Dowker gives no sufficient reason for refusing to accept Elliott’s view (which he travesties) that the sea once found its way over the marsh through a gap between the advancing shingle and the hills, and also through a break in the shingle spit,—in other words, for maintaining that the marsh had become dry land before the shingle beach was formed. The notion that the Rhee Wall was, ‘in the first place, a natural river-bank’ is simply fantastic. To begin with, its direction is almost a straight line, whereas it is well known that in open plains, where the slope is slight, rivers invariably pursue tortuous courses.2583 Along what is now called the Rhee Wall runs the high road from Appledore to New Romney. It occupies what was formerly a channel embanked on either side; and this channel provided an outlet for the waters of the Rother, whose actual mouth was at Appledore.2584 As Elliott says, ‘In erecting this wall it became necessary to provide some exit for the waters from the hills as well as for the drainage of the land enclosed. This was done by cutting a channel parallel with the wall from the pool or lake at the embouchure of the river Limene at Appledore to the sea at Romney... the wall was necessary to be continued across this lake until it met the high land at Appledore.’2585 Again, I cannot understand why, if Romney Marsh Proper became dry land before the time of the Romans without being artificially enclosed, Walland Marsh and Guildford Marsh, which lie west and south of the Rhee Wall, should still have been periodically overflowed by the sea; nor is it clear how in that case the Rother could have excavated its hypothetical channel along the line of the Rhee Wall. Lastly, it is impossible, on Dowker’s theory, to locate the Portus Lemanis. He denies that it was at Lympne: it could not, on his theory, have been at Hythe or at West Hythe, for he implies that the shingle beach, behind which lay the historic Hythe Haven, did not yet exist;2586 and Romney—the only other possible site—is, as I shall afterwards show, out of the question.

I am not concerned to dispute Dowker’s theory that the sea was excluded from the marsh on the south by sand-hills before the shingle beach was formed, though the mere presence of patches of blown sand near West Hythe and near Romney does not justify him in connecting them by ‘a hypothetical line extending from Rye to Hythe’; nor does he offer any theory to account for the disappearance of this hypothetical line after it began to be protected by a barrier of shingle. The important point is that the fact of the erection of the Rhee Wall proves that before it existed Romney Marsh Proper was liable to be flooded by high tides.

3. It has long been a vexed question where, in the time of Caesar, and during the Roman occupation of Britain, the Rother discharged itself. Hasted2587 affirms that the bed of the river ‘may yet very easily be traced... under the hills from West Hythe to Appledore’. Beale Poste,2588 who agrees with him, says that, according to the Itinerary of Antonine, the port of the river Lemanis, which he identifies with the Rother, was the Portus Lemanis; that, according to Somner, ancient records mention ‘the Lymne branch of the Rother as still in existence in... 820 at... Warehorne, at about... three miles from the bend of our river towards Lymne’; and that ‘we find the name Portus Limneus in Ethelwerd’s Chronicle, iv, 3, in his annals of... 893, which seems to imply the “Port of the river Lemanis”.’ Holloway,2589 the historian of Romney Marsh, after saying, like Hasted, that ‘traces of the ancient bed of a river are still visible under the foot of the Kentish cliffs’, adds that ‘our ancient chroniclers, according to Lambarde, called this same place “Limene Mouthe”, and which is interpreted by Leland to betoken the mouth of the river Rother’. Drew2590 holds that the river Limen, or, as it is called by the anonymous geographer of Ravenna, Lemana,2591 must in the ninth century have flowed past Sandtun, ‘the patch of Blown Sand between West Hythe and Butter’s (or Botolph’s) Bridge,’ because in a charter of the year 833 allusion is made to ‘a piece of land at Sandtun that was bounded on the south by the river Limen’. Finally, Mr. F. P. Gulliver thinks it probable that the Rother had, a thousand years ago, two ‘main distributaries’, one of which flowed out ‘through an inlet in the bar south-west of Hythe’.2592

Hasted’s statement is quite incorrect. Elliott, who knew every inch of Romney Marsh, positively affirms that ‘between Lymne and Appledore... not the slightest trace of any river remains’;2593 and his statement is confirmed by Topley.2594 Dowker2595 also observes that if the Rother had ever flowed out near Hythe, ‘it must have occupied the space where the Military Canal exists, in which case it has left no historical or other trace behind, and against such a river the Ree Wall could have been no protection.’ Moreover, if there is any force in the argument of Drew, the river flowed south of the blown sand near Butter’s Bridge, that is to say, a good mile from the hills.2596 Elliott accounts for the belief that the river entered the sea near Lympne by the fact that a depression exists along the foot of the hills, ‘many taking that to be the river which in truth was only an estuary... and which would only assume something of the character of a river at low water.’2597 In reply to Beale Poste, it is sufficient to remark that the Itinerary does not say that the port of the river Lemanis (or rather Lemana) was the Portus Lemanis, nor does it even mention the river: it simply gives the distance of the Portus Lemanis from Durovernum, or Canterbury.2598 Beale Poste misquotes Somner, who does not say a single word about ‘the Lymne branch of the Rother’.2599 It is quite true that we find the words portu Limneo in the Chronicle of Ethelwerd;2600 but it is not easy to see how these words convey any more information about the geographical position of the port than the words portus Lemanis. As to Holloway’s argument, all that Lambarde2601 says is that Robert Talbot,2602 ‘a man of our time,’ was of opinion that Shipway, near West Hythe, was so called ‘because it lay in the way to the Haven where the ships were woont to ride.2603 And that haven,’ adds Lambarde, ‘taketh hee to be the same which... is called... of Antoninus Limanis, of our chroniclers Limene Mouth, and interpreted by Leland to betoken the mouth of the river of Rother.’ The Anglo-Saxon Chronicle states distinctly that the mouth of the Limen was at Appledore;2604 and Leland was far too acute to be duped by the notion that it had ever been at Lympne: ‘where the Ryver Limene should be,’ he says, ‘I can not tel, except yt should be that that cummeth above Appledor... and that ys Cowrs ys now changed.’2605 With regard to Drew’s argument, allusion is made in two charters2606 to ‘a piece of land at Sandtun, that was bounded on the south by the river Limen’, namely, a charter of King Aethilberht of Kent, dated February 20, 732, and a charter of King Ecgberht of Kent, dated 833. In the latter it is stated that there were salt-pans ‘in the same place’, namely at Sandtun;2607 and in both the boundaries of the land are defined in almost identical terms,—‘the boundaries of this piece of land are, on the east the King’s land; on the south the river called the Limen; on the west and on the north the Hudan Fleot.’2608 That Sandtun was the patch of blown sand between West Hythe and Botolph’s Bridge is a pure assumption on the part of Drew. Furthermore, he would have found it difficult to indicate the position of ‘the King’s land’ on the east, seeing that on the east, if the Limen debouched opposite Lympne, there was only shingle or sea. Finally, it is certain that before 833 Romney Marsh Proper had been enclosed; and how a river could have flowed along the north of the marsh across the Rhee Wall, or how, if it had worked this miracle, it should have subsequently disappeared without leaving any trace of its existence, is more than I can understand.2609 At all events the level of the marsh, which is 6 feet 6 inches lower at Appledore Dowles than at West Hythe Oaks, proves that, even assuming the former existence of such a river, centuries must have elapsed from the time when it ceased to flow beneath the hills to the time when the shingle closed the marsh at West Hythe Oaks.2610

Elliott2611 concluded, ‘from several careful surveys of the whole district,’ that the mouth of the Limen was at Appledore, where it entered the estuary; and, as Roach Smith2612 truly remarks, this conclusion is confirmed by the Anglo-Saxon Chronicle. ‘We now,’ says Elliott,2613 ‘find the whole country about the mouth of the Limene, at Appledore, in a circuit of about a mile (and at no other part), at a few feet under the present surface, covered with trees of the oak, alder, and birch... evidently, from their position, having been drifted from a distance, and deposited where now found.’ Lewin2614 points out that this ‘is the very lowest part of the marsh’; and he holds that ‘the presence of oak trees... decides that the trees are not in situ,2615 for... there is something in the Marsh mould uncongenial to the oak’. The course of the river, Elliott tells us, is ‘still traceable between Appledore and the Isle of Oxney, and thence into the estuary, about half a mile south of Appledore’. Once, as we have seen, according to Elliott, the estuary found an exit opposite Lympne: when this was closed, there remained only the channel between Romney and Lydd.2616

4. It is now necessary to inquire what was the geographical position of the Portus Lemanis. The reader will, of course, see that this question is quite distinct from that which he has just been considering. Whether the Rother ever flowed along the north of the marsh or not, everybody admits that the sea once had access there even at low tide; and the question is whether the Portus Lemanis was this estuary, or rather that part of it which lay below Lympne Hill. This is the generally accepted view.2617 In support of it Appach2618 argues as follows:—First, the name ‘Lympne’ is obviously a corruption of Lemanis, and Leland found a tradition existing that Lympne had once been a port. Secondly, at Lympne, Stone Street, the Roman road from Canterbury, ‘terminates abruptly,’ and ‘no trace whatever of its continuance southward into the marsh can be discovered’. ‘For what reason,’ asks Appach, ‘could this road have been made if Lympne was not then a port?’ He goes on to observe that, according to the Itinerary of Antonine, ‘Portus Lemanis was one stage distant from Canterbury;’ that, besides Stone Street, the only Roman roads which converged at Canterbury were those which led to Reculver, Richborough, and Dover; and therefore that the Portus Lemanis must have been situated on Stone Street, and obviously at its termination. Thirdly, according to the Itinerary, the distance from Canterbury to the Portus Lemanis was 16 Roman miles, or about 25,872 yards;2619 and the actual distance from ‘the margin of the marsh below Lympne measured along the Stone Street to the point where all the Roman roads at Canterbury would converge, if produced, is fifteen statute miles’, or 26,400 yards.2620 Fourthly, the existence of Stutfall Castle proves that the Portus Lemanis was at Lympne; and, moreover, the castle ‘had no southern wall because the sea came up to the foot of the fortifications’. Fifthly, in the Table of Peutinger, Lemanis is ‘marked with a castle, like Richborough and Dover’.

These arguments may, at first sight, appear conclusive: in reality they are worthless. (1) Leland2621 does not mention any tradition about the port: he simply asserts that ‘Lymme Hill or Lyme was sumtyme a famose Haven, and good for Shyppes that might cum to the Foote of the Hille’. Lambarde,2622 it is true, says that there was in his time a tradition that Shipway was so called because ‘it lay in the way to the Haven where the ships were woont to ride’; and he calls this tradition ‘the report of the countrie people, who hold faste the same opinion which they have by tradition receaved from their Elders’. Also he himself asserts that ‘at the first, ships were accustomed to discharge at Lymme’. But Shipway ‘lay in the way’ to West Hythe, not to Lympne. As for the alleged tradition, everything depends upon the date of its origin; and this cannot be ascertained. The name ‘Lympne’ may be connected with Lemanis; but this does not prove that the Portus Lemanis was at the foot of the heights on which Lympne stands: if it had been east of Stutfall Castle, and the nearest town in Roman times or later had been on the site of Lympne, the origin of the name would be perfectly clear. (2) Appach insists that Stone Street ‘terminates abruptly’ at Lympne; but, as a matter of fact, a road diverges to the right from the straight course of Stone Street at New Inn Green, and terminates just north of Stutfall Castle.2623 Mr. Thurston of Ashford points out that if the course of Stone Street were continued in a straight line from New Inn Green, it ‘would point to the Shipway [or Shepway] Cross, and continue down the present roadway which descends the hill to West Hythe; and’, he adds, ‘this is the only place along the hill where a roadway could possibly descend it in a straight line, and I believe it was naturally selected as the road to the ships or port.’2624 (3) As for the argument based upon the distance given in the Itinerary from Canterbury to the Portus Lemanis, a moment’s reflection will convince any reader who uses his map that it holds good for the theory that the Portus Lemanis was at West Hythe as well as for the view which Appach defends. (4) The situation of Stutfall Castle may no doubt be used as an argument to prove that the Portus Lemanis was at Lympne: but the castle is barely a mile and a half from West Hythe Oaks, which, as we shall presently see, was in all probability the western end of the port; and, although it was believed when Appach wrote that the castle had no southern wall, excavation has since proved that it had.2625 Appach’s last argument depends, like the one which precedes it, upon the assumption that Stutfall Castle would have been useless unless it had stood in immediate proximity to the Portus Lemanis. What if Lemanis was ‘marked with a castle’? Why should not the castle have protected the neighbouring part of ‘the Saxon shore’ and a harbour at West Hythe?

The late antiquary, W. H. Black,2626 remarked further, that the discovery of a Roman altar in Stutfall Castle, erected by the ‘admiral of the British fleet’ (praefectus classis Britannicae), proves that the Portus Lemanis was at Lympne; and, observing that ‘the Saxon Chronicle tells us of the arrival of a fleet of Danes at “Limene mouth”’, he argues that ‘it is impossible to deny the identity of Lymne with that name’. But, whatever may be the etymological connexion between Lympne and Limene, it has been shown already that according to the very chronicle which Black cites, the mouth of the Limen was at Appledore;2627 and the discovery of the Roman altar is perfectly consistent with the view that the harbour which was the admiral’s naval base was near West Hythe.

Elliott originally held that the Portus Lemanis was the estuary at Lympne;2628 and his opinion was quoted by superficial writers in support of this view several years after he had himself discarded it. For he finally came to the conclusion that, even as early as Caesar’s time, there was no harbour at Lympne.2629 He tells us that ‘recent investigations in taking a series of levels over the whole of Romney Marsh have established the fact that the estuary must have been closed at the eastern extremity (where the Portus Lemanis is commonly looked for) many centuries before the sea was shut out from... Romney Marsh Proper; for at the extreme eastern end of Romney Marsh, by Hythe Oaks, the surface of the land is 18 inches higher than it is a mile westward, a state of things that could not have existed had there been any outlet towards the east after the closing of the Marsh westward. The inset and outset of the tides twice a day to and from the estuary would have counteracted the silting, and produced not an elevation, but a depression of the surface. There is... a regular and continuous fall of the land next the hills, from Hythe Oaks into Appledore Dowles... the lowest part of the Marsh being 6 feet 6 inches lower than the land at Hythe Oaks. There could have been no silting after the inclosure of the Marsh, and the present level is such as it was when the Marsh was reclaimed.... The barrier which sealed up the eastern mouth of the estuary was the accumulation of shingle from the west, and (sic) which long before the historic period had reached the hills at Hythe Oaks. If Romney Marsh, at the foot of the castrum [Stutfall Castle], was dry land at that time [A.D. 368-9, when Theodosius2630 was in Britain] and occupied by the Romans (as we know to have been the case), Stutfall could not have been the “Portus Lemanis”... as it was not accessible from the sea, and lay a mile and a half at least from it. The sea could not have flowed there without putting the whole of Romney Marsh Proper under water to the depth of eight or ten feet every springtide.’ Similarly, Lewin2631 states, on the authority of Elliott, that ‘the greater elevation of the soil towards the east of Romney Marsh Proper can be only accounted for by the fact that when the shingle “full” had been thrown quite across the Marsh at West Oaks... the sea still entered from the west, and that, thenceforth, the process of silting went on for many centuries... most rapidly towards the east, where the water was tranquil, and less rapidly towards the [site of the subsequently erected] Rhee Wall, in which direction was the scour of the current’.

‘Many centuries’ is a vague expression; but for ‘many’ substitute ‘three’, and, even for the time of Caesar, the argument still holds good,—unless Elliott’s theory of the formation of the marsh is to be rejected.

But there are writers whom Elliott’s reasoning (if indeed they have considered it) leaves unconvinced. According to Mr. George E. Fox, it has been proved by excavation that the existing castellum at Stutfall is not earlier than the time of Constantine;2632 but Sir Victor Horsley, while confirming this statement, tells us that he has himself found ‘in the foundation of the chief gate an altar... marked with barnacles, having been clearly at one time under the sea’; and from this he infers that an earlier fort was ‘overwhelmed by an incursion of the sea over Romney level’. Sir Victor also tells us that he has found ‘in the concrete boulder formation of the south wall... a coin of Maximinus, who flourished 237 A.D.’, and ‘at the foot of the wall on the inner side, a Gaulish coin of Tetricus the elder, of a date about 260, and finally in the black soil of the camp, i.e. in the most recent and superficial layers, numerous coins of the Constantine family’.2633

I do not know whether Sir Victor Horsley concludes from these discoveries that there was a harbour at Lympne when the earlier hypothetical castellum at Stutfall was destroyed; but at all events that is the opinion of Mr. Fox. But the ‘incursion of the sea’ which Sir Victor Horsley believes to have overwhelmed the original fort, if it was not caused by an abnormally high tide rushing in between Romney and Lydd before the erection of the Rhee Wall, may have been due to a similar tide which burst the bar of shingle between Dymchurch and West Hythe. Even after the marsh had been artificially enclosed, such floods occurred. Stukeley2634 tells us that ‘George Hunt, an old man, living in the farm-house... says, once the sea-bank broke, and his house with all the adjacent marshes was floated’,2635 &c.

Lewin maintained that the Portus Lemanis was neither at Lympne nor at West Hythe, but at Hythe. This, it should be noted, was the conclusion at which he finally arrived:2636 when he wrote his book on the invasion of Julius Caesar, he held that in 55 B.C. there was a port at Lympne, although in the Appendix to that book he discarded this view, and argued that the only port was a pool harbour extending behind a shingle spit from West Hythe Oaks to a point opposite Shorncliffe. His final view, as we have already seen,2637 was that this harbour extended no further westward than Hythe itself: but in giving utterance to this opinion he did not explain why he had abandoned the one which preceded it, and indeed made no allusion to it at all.

He states that ‘in the course of ages’, after the shingle had reached West Hythe Oaks, it ‘was again wrested aside and dashed against the hills at Hythe, between the present barracks and the more eastern of the two Hythe bridges over the canal’. He goes on to say that ‘the part between Hythe Oaks and Hythe (now Duck Marsh) was thus barred from the sea, and became a lake into which flowed the rivulet called Slabrook and other springs, and these waters accumulating forced their way back at Hythe Oaks, and there opened a way for themselves... into the estuary in the west; but, as the flood was not considerable, the outlet was of no great breadth. The shingle spit... was again carried along eastward until it reached Shorncliffe.... Between Hythe and Shorncliffe, however, was left behind (i.e. north of) the spit, a triangular space, into which flowed two streams... one from Saltwood and the other called Seabrook, and the waters within this spit were gradually swollen, until they forced a passage through the shingle, at a point near the end of the elm avenue at Hythe.’ The change which his opinion underwent will be at once apparent to any one who compares the map which Elliott constructed for The Invasion of Britain by Julius Caesar (facing page liii) with that which accompanies the article in the fortieth volume of Archaeologia2638 (facing page 369). Lewin argues that it was so easy to exclude the sea from Duck Marsh that ‘probably the inclosure was made by the Britons before the arrival of the Romans. On the south-east,’ he explains, ‘the shingle bank was continuous up to the hills... on the west the sea entered only from the marsh at the foot of the hills by a narrow channel; and all that was required was a short dam at this point between the shingle bed and the hills.’ The remains of this dam, Lewin observes, are ‘still distinguishable... at Hythe Oaks, but the part next the hills has been swept away by the military canal. This partial inclosure, prior to the inclosure of Romney Marsh, accounts for a fact otherwise inexplicable, viz. that Duck Marsh is not within the jurisdiction of Romney Marsh.’2639

Perhaps. But the date of the construction of the dam is not known. May it not have been made after, or simultaneously with, the erection of the Rhee Wall, to secure Romney Marsh against all possibility of inundation, not to protect Duck Marsh, which, according to Lewin’s earlier view, was originally overflowed by Hythe harbour? In other words, is it not possible that when the dam was made Hythe harbour extended westward as far as West Hythe Oaks? This, as I have already said, was not merely Lewin’s original view: it was also the view which Elliott, his friend and adviser, retained after the publication of the article in Archaeologia. At all events this view finds expression in a map which Elliott prepared for Furley’s History of the Weald of Kent, which was not published until 1871, five years after the appearance of Lewin’s article. That being the case, and considering that Lewin did not explain the reasons which led him to change his opinion, I am unable to follow him.

In support of the theory that the Portus Lemanis was at Hythe Lewin argues, first, that Stone Street terminated at West Hythe; secondly, that the port could not have been at West Hythe; otherwise ‘the whole of West Hythe... would have been deluged’. ‘The very name,’ he adds, ‘shows that Hythe was the principal town, and West Hythe an accretion to it.’ Thirdly, he affirms that Roman remains have been found at Hythe; and, fourthly, that a branch from Stone Street led to Hythe. He also bases an argument upon the itinerary of Richard of Cirencester, which, as every scholar now knows, is a forgery.2640

Stone Street does terminate, as Lewin says, at West Hythe; but the fact goes to prove that it gave access to a harbour which was at West Hythe.2641 Granting that West Hythe would have been ‘deluged’ if the port had been there, what then? Why should it not have been? Lewin does not explain what he means by ‘the whole of West Hythe’; and, in default of this explanation, it is impossible to understand his argument.2642 He himself, as we have seen, in his book on the invasion of Britain by Julius Caesar makes the port extend westward as far as West Hythe Oaks; and Black shows that, so far from its being true that West Hythe is merely an ‘accretion’ of Hythe, Hythe is merely East Hythe, and that it is so called in Ogilby’s Britannia.2643 The discovery of Roman remains at Hythe does not prove that Hythe was the Portus Lemanis any more than the discovery of Roman remains at Dymchurch proves that the Portus Lemanis was there. Or rather, the discovery does not prove that the Portus Lemanis extended no further westward than Hythe; for I freely admit that it extended in front of and to the east of it. It is not proved that a branch from Stone Street led to Hythe;2644 and if there was such a branch, the fact does not prove that the harbour did not extend as far as West Hythe Oaks. Finally, Black points out that, whereas the distance of Lympne (and, he might have added, of West Hythe) from Canterbury corresponds with that of the Portus Lemanis from Durovernum, as given in the Itinerary of Antonine, the distance of Hythe by road from the same place is two miles further.2645

5. The first step taken for the enclosure of Romney Marsh was the erection of the Rhee Wall. By whom and at what date this work was executed is not certainly known. It is generally attributed to the Romans; but Lewin2646 assures us that Mr. Smiles, in his Lives of the Engineers, ‘expresses an opinion that the Marsh was reclaimed by the Belgae.’ What Mr. Smiles2647 really says is that ‘the reclamation of this tract is supposed to be due to the Frisians’; and he does not tell us by whom the supposition is entertained, or on what grounds it is based. Lewin himself, asking whether [Appledore] ‘Dowles’ is not derived from the Celtic word dol, says that ‘if a part of Romney Marsh was named by the Ancient Britons, the marsh itself must have been reclaimed by them’.2648 From the same word Appach2649 draws precisely the opposite inference. ‘Apuldore Dowles,’ he says, ‘appears to be allied to the Welsh dol, a bend. If so, it would mean a bend or curve, and so a recess or bay; and Apuldore Dowles would mean the bay of Apuldore.’ Whatever may be the value of this argument, the name ‘Apuldore Dowles,’ does not go to prove that Romney Marsh was ‘inned’ by the Britons; for, as Appach2650 truly remarks, there is no other local name in Romney Marsh Proper which shows any trace of a Celtic derivation.

Mr. W. A. S. Robertson,2651 on the other hand, states, on the authority of Professor Skeat, that ‘Rumenea’, the name by which, according to Lambarde,2652 Romney was known to the Saxons, is compounded of the Gaelic word ruimen (marsh) and the Saxon affix ea (river); and he concludes that ‘before the Roman occupation there was in this great estuary sufficient land, uncovered by water, to be denominated... Rum or Ruimen’. Again, arguing that the καινὸς λιμήν, or ‘new harbour’, mentioned by Ptolemy,2653 was at Romney, he says that ‘if it was called into existence by... the Rhee Wall, it follows that the Rhee Wall’ was ‘probably formed at least as early as the first century of the Christian era’.

If the ‘new harbour’ was at Romney! There is not the slightest evidence that it was there.2654 As for the word ruimen, how can Mr. Robertson prove that it was applied to Romney Marsh ‘before the Roman occupation’? Moreover, supposing that the marsh was not embanked by the Britons, there was ‘sufficient land uncovered by water to be denominated Ruimen’ twice every day, when the tide was low, before the Rhee Wall was made; and the name lends no support to Mr. Robertson’s theory.

I do not attach much importance to the argument, first propounded by Sir W. Dugdale2655 and often repeated since, that because the Britons, according to Tacitus2656—or rather, according to a speech put by Tacitus into the mouth of a British chief—were employed by the Romans in draining and embanking marshes, therefore the Romans enclosed this particular marsh. But, considering that Roman remains have frequently been discovered in that part of the marsh which lies on the east of the Rhee Wall,2657 it is surely inexplicable that if the wall was built by the Britons, no Celtic remains have ever been found there.

Appach2658 not only rejects the theory that the Britons built the Rhee Wall, but denies that Romney Marsh Proper was enclosed during the Roman occupation. He maintains that, in Caesar’s time, ‘the northern portion, at all events, and possibly the whole of the interval between the island of Romney and the high ground of Kent was open sea.’ For, he argues, ‘Lympne was the ancient Portus Lemanis... that place could not have been a port unless there had been free access to it from the Channel, and it is clear from the manner in which the marsh and shingle were deposited, that there was always open sea between Lympne and the Channel until the interval between the ancient island at Romney and the high ground of Kent had been closed by the gradual growth of the marsh and shingle.’

The assumption upon which this argument rests has been already disproved: the Portus Lemanis was not at Lympne. Appach’s theory forces him to assume that the sediment which formed the marsh was deposited at an incredibly rapid rate. He maintains2659 that ‘the upper portion of Romney Marsh, for a depth of thirty feet... below its present surface (which would give sufficient water for the heaviest of Caesar’s ships at the lowest Spring tides) might very well have been deposited’ in ‘about five hundred years’. But, according to Elliott,2660 the average rate at which the silt was deposited was not more than about one-eighth of an inch per annum.

Dowker, on the other hand, although he once regarded it as ‘evident that at the period of Caesar’s invasion the marsh was little better than a swamp, great part being under water at high tide’, maintained that the discovery of Roman pottery on the west of Dymchurch disproved Appach’s theory.2661 But he did not take account of dates. Appach himself2662 noted the discoveries which had been made near Dymchurch; but he observed that while some of the objects discovered had been pronounced by the Society of Antiquaries to be ‘decidedly Roman’, others had been attributed by the same body to subsequent periods; and he concluded that the marsh had not been enclosed before the middle of the fifth century.

This theory is pulverized by one fact which Appach ignores. Dymchurch is not the only place in Romney Marsh Proper where Roman remains have been found: they have been discovered in Eastbridge, at Newchurch, at Ivychurch, and indeed over the whole area.2663 On the other hand, Welland Marsh, Guildford Marsh, and Denge Marsh—those parts of Romney Marsh, popularly so called, which extend westward of the Rhee Wall—have yielded none.2664 The inference is certain: Romney Marsh Proper was enclosed during the Roman occupation of Britain.

6. The conclusions which we have now reached are, first, that the Rother did not, in the time of Caesar, enter the sea at Lympne, but debouched into the estuary near Appledore; secondly, that the marsh was then closed at West Hythe Oaks, and therefore that there was no harbour at Lympne; thirdly, that the Rhee Wall had not then been built, and therefore that the marsh was still flooded at spring tides by the inrush of the sea between Romney and Lydd; fourthly, that the Portus Lemanis was a pool harbour extending from West Hythe to a point nearly opposite Shorncliffe; and, lastly, that the Rhee Wall was built in Roman times.

But, as the reader will hereafter see, if these conclusions are erroneous, the error will not lead us astray when we have to determine the place where Caesar landed in Britain.